Raman Spectra of Amino Acids and Peptides from Machine Learning Polarizabilities.

Raman spectroscopy is an important tool in the study of vibrational properties and composition of molecules, peptides, and even proteins. Raman spectra can be simulated based on the change of the electronic polarizability with vibrations, which can nowadays be efficiently obtained via machine learning models trained on first-principles data. However, the transferability of the models trained on small molecules to larger structures is unclear, and direct training on large structures is prohibitively expensive. In this work, we first train two machine learning models to predict the polarizabilities of all 20 amino acids. Both models are carefully benchmarked and compared to density functional theory (DFT) calculations, with the neural network method being found to offer better transferability. By combination of machine learning models with classical force field molecular dynamics, Raman spectra of all amino acids are also obtained and investigated, showing good agreement with experiments. The models are further extended to small peptides. We find that adding structures containing peptide bonds to the training set greatly improves predictions, even for peptides not included in training sets.

Wnt5a Deficiency Leads to Anomalies in Ureteric Tree Development, Tubular Epithelial Cell Organization and Basement Membrane Integrity Pointing to a Role in Kidney Collecting Duct Patterning.

The Wnts can be considered as candidates for the Congenital Anomaly of Kidney and Urinary Tract, CAKUT diseases since they take part in the control of kidney organogenesis. Of them Wnt5a is expressed in ureteric bud (UB) and its deficiency leads to duplex collecting system (13/90) uni- or bilateral kidney agenesis (10/90), hypoplasia with altered pattern of ureteric tree organization (42/90) and lobularization defects with partly fused ureter trunks (25/90) unlike in controls. The UB had also notably less tips due to Wnt5a deficiency being at E15.5 306 and at E16.5 765 corresponding to 428 and 1022 in control (p<0.02; p<0.03) respectively. These changes due to Wnt5a knock out associated with anomalies in the ultrastructure of the UB daughter epithelial cells. The basement membrane (BM) was malformed so that the BM thickness increased from 46.3 nm to 71.2 nm (p<0.01) at E16.5 in the Wnt5a knock out when compared to control. Expression of a panel of BM components such as laminin and of type IV collagen was also reduced due to the Wnt5a knock out. The P4ha1 gene that encodes a catalytic subunit of collagen prolyl 4-hydroxylase I (C-P4H-I) in collagen synthesis expression and the overall C-P4H enzyme activity were elevated by around 26% due to impairment in Wnt5a function from control. The compound Wnt5a+/-;P4ha1+/- embryos demonstrated Wnt5a-/- related defects, for example local hyperplasia in the UB tree. A R260H WNT5A variant was identified from renal human disease cohort. Functional studies of the consequence of the corresponding mouse variant in comparison to normal ligand reduced Wnt5a-signalling in vitro. Together Wnt5a has a novel function in kidney organogenesis by contributing to patterning of UB derived collecting duct development contributing putatively to congenital disease.

GSK3ß-dependent phosphorylation alters DNA binding, transactivity and half-life of the transcription factor USF2.

The upstream stimulatory factor 2 (USF2) is a regulator of important cellular processes and is supposed to have also a role during tumor development. However, the knowledge about the mechanisms that control the function of USF2 is limited. The data of the current study show that USF2 function is regulated by phosphorylation and identified GSK3ß as an USF2-phosphorylating kinase. The phosphorylation sites within USF2 could be mapped to serine 155 and threonine 230. In silico analyses of the 3-dimensional structure revealed that phosphorylation of USF2 by GSK3ß converts it to a more open conformation which may influence transactivity, DNA binding and target gene expression. Indeed, experiments with GSK-3ß-deficient cells revealed that USF2 transactivity, DNA binding and target gene expression were reduced upon lack of GSK3ß. Further, experiments with USF2 variants mimicking GSK3ß phosphorylated USF2 in GSK3ß-deficient cells showed that phosphorylation of USF2 by GSK3ß did not affect cell proliferation but increased cell migration. Together, this study reports a new mechanism by which USF2 may contribute to cancerogenesis.

The structure of the CD3ζζ transmembrane dimer in lipid bilayers.

Virtually every aspect of the human adaptive immune response is controlled by T cells. The T cell receptor (TCR) complex is responsible for the recognition of foreign peptide sequences, forming the initial step in the elimination of germ-infected cells. The recognition leads to an extracellular conformational change that is transmitted intracellularly through the Cluster of Differentiation 3 (CD3) subunits of the TCR-CD3 complex. Here we address the interplay between the disulfide-linked CD3ζζ dimer, an essential signaling component of the TCR-CD3 complex, and its lipidic environment. The disulfide bond formation requires the absolute presence of a nearby conserved aspartic acid, a fact that has mystified the scientific community. We use atomistic simulation methods to demonstrate that the conserved aspartic acid pair of the CD3ζζ dimer leads to a deformation of the membrane. This deformation changes the local environment of the cysteines and promotes disulfide bond formation. We also investigate the role of a conserved Tyr, highlighting its possible role in the interaction with other transmembrane components of the TCR-CD3 complex.

A class V chitinase from Arabidopsis thaliana: gene responses, enzymatic properties, and crystallographic analysis.

Expression of a class V chitinase gene (At4g19810, AtChiC) in Arabidopsis thaliana was examined by quantitative real-time PCR and by analyzing microarray data available at Genevestigator. The gene expression was induced by the plant stress-related hormones abscisic acid (ABA) and jasmonic acid (JA) and by the stress resulting from the elicitor flagellin, NaCl, and osmosis. The recombinant AtChiC protein was produced in E. coli, purified, and characterized with respect to the structure and function. The recombinant AtChiC hydrolyzed N-acetylglucosamine oligomers producing dimers from the non-reducing end of the substrates. The crystal structure of AtChiC was determined by the molecular replacement method at 2.0 Å resolution. AtChiC was found to adopt an (ß/α)(8) fold with a small insertion domain composed of an α-helix and a five-stranded ß-sheet. From docking simulation of AtChiC with pentameric substrate, the amino acid residues responsible for substrate binding were found to be well conserved when compared with those of the class V chitinase from Nicotiana tabacum (NtChiV). All of the structural and functional properties of AtChiC are quite similar to those obtained for NtChiV, and seem to be common to class V chitinases from higher plants.

Dimerization of human lysyl hydroxylase 3 (LH3) is mediated by the amino acids 541-547.

Lysyl hydroxylases (LH), which catalyze the post-translational modifications of lysines in collagen and collagen-like proteins, function as dimers. However, the amino acids responsible for dimerization and the role of dimer formation in the enzymatic activities of LH have not yet been identified. We have localized the region responsible for the dimerization of lysyl hydroxylase 3 (LH3), a multifunctional enzyme of collagen biosynthesis, to a sequence of amino acids between the glycosyltransferase activity and the lysyl hydroxylase activity domains. This area is covered by amino acids 541-547 in human LH3, but contains no cysteine residues. The region is highly conserved among LH isoforms, and is also involved in the dimerization of LH1 subunits. Dimerization is required for the LH activity of LH3, whereas it is not obligatory for the glycosyltransferase activities. In order to determine whether complex formation can occur between LH molecules originating from different species, and between different LH isoforms, double expressions were generated in a baculovirus system. Heterocomplex formation between mouse and human LH3, between human LH1 and LH3 and between human LH2 and LH3 was detected by western blot analyses. However, due to the low amount of complexes formed, the in vivo function of heterocomplexes remains unclear.

Computer modeling of brain tumor growth.

An important objective of brain tumor modeling is to predict the progression of tumors so as to provide guidance about the best possible medical treatment to halt or slow the tumor's growth. Such computer models also provide a deeper insight into the physiology of tumors. In addition, one can study various what-if scenarios, for instance, investigating the response of tumors following the administration of a drug or a variety of drugs. Abrupt changes in growth rate can also be important for surgical decision-making. Despite increased interest in modeling techniques, relatively little progress has been made in improving such technologies. One problem is the limited data available from patients, typically 1 to 3 MRI (magnetic resonance imaging) sessions, from which one has to extrapolate the type of tumor so as to successfully predict its evolution over time. Here, the biological and clinical aspects of tumor growth and treatment with surgery, radiotherapy and drugs are discussed in the light of a patient with a brain tumor showing accelerated growth over time. Then, the contributions of mathematical modeling of tumor growth and effects of treatment are presented. Current tumor growth models can be roughly divided in three main categories, (i) cellular and microscopic models that emphasize isolated cell behavior, (ii) macroscopic models that concentrate on the development of cell density over time, and (iii) hybrid approaches that contain elements of both microscopic and macroscopic models. The mathematical theory that underlies these simulation methods is remarkably similar to the physical theory that forms the basis of protein modeling and molecular mechanics tools. A severe limitation of current models is that they are in fact not patient-specific at all.

Crystal structure of yeast peroxisomal multifunctional enzyme: structural basis for substrate specificity of (3R)-hydroxyacyl-CoA dehydrogenase units.

(3R)-hydroxyacyl-CoA dehydrogenase is part of multifunctional enzyme type 2 (MFE-2) of peroxisomal fatty acid beta-oxidation. The MFE-2 protein from yeasts contains in the same polypeptide chain two dehydrogenases (A and B), which possess difference in substrate specificity. The crystal structure of Candida tropicalis (3R)-hydroxyacyl-CoA dehydrogenase AB heterodimer, consisting of dehydrogenase A and B, determined at the resolution of 2.2A, shows overall similarity with the prototypic counterpart from rat, but also important differences that explain the substrate specificity differences observed. Docking studies suggest that dehydrogenase A binds the hydrophobic fatty acyl chain of a medium-chain-length ((3R)-OH-C10) substrate as bent into the binding pocket, whereas the short-chain substrates are dislocated by two mechanisms: (i) a short-chain-length 3-hydroxyacyl group ((3R)-OH-C4) does not reach the hydrophobic contacts needed for anchoring the substrate into the active site; and (ii) Leu44 in the loop above the NAD(+) cofactor attracts short-chain-length substrates away from the active site. Dehydrogenase B, which can use a (3R)-OH-C4 substrate, has a more shallow binding pocket and the substrate is correctly placed for catalysis. Based on the current structure, and together with the structure of the 2-enoyl-CoA hydratase 2 unit of yeast MFE-2 it becomes obvious that in yeast and mammalian MFE-2s, despite basically identical functional domains, the assembly of these domains into a mature, dimeric multifunctional enzyme is very different.

Theoretical investigations of prostatic acid phosphatase.

The phosphotyrosyl protein phosphatase activity of prostatic acid phosphatase (PAP) has been well established. It has also been suggested that PAP partly regulates the activity of growth factor receptors by dephosphorylating the autophosphorylysable tyrosines in them. We studied the binding of the peptides from epidermal growth factor receptor (EGFR) and its homolog (ErbB-2), corresponding to their autophosphorylation sites, to PAP using theoretical modeling and molecular dynamics (MD) simulation methods. Nine different peptides, each with a phosphotyrosine residue, were docked on human PAP. The binding energies of these peptide-PAP complexes were calculated theoretically and compared to experimentally obtained affinities. The peptide Ace--DNLpYYWD--NH2 from ErbB-2(1197-1203) showed the most favorable free energy of binding when estimated theoretically. The results demonstrate that the presence of another tyrosine residue proximate to C-terminal of autophosphorylysable Tyr enhances the binding affinity considerably. The presence of a bulky group instead prevents the binding, as is observed in case of peptide Ace--NLYpYWDQ--NH2 which failed to bind, both in theoretical calculations and experiments. Thus we demonstarted that PAP could potentially bind to EGFR and Erbb-2 and dephosphorylate them. Thus it could be involved in the regulation of the function of such receptors. In addition, complexes of a peptide from AngiotensinII and phosphotyrosine(pY) with human PAP were also modeled. The effects of different protonation states of the titratable active site residues on ligand (pY) binding have also been investigated. For a favorable binding His12 and Asp258 should be neutral, His257 should be positively charged and the phosphate group of the ligand should be in PO(4) (3-) state. Furthermore, the analysis of protein motion as observed during simulations suggests the loop-loop contact in the PAP dimer to be of importance in cooperativity.

Characterization of collagenous peptides bound to lysyl hydroxylase isoforms.

Lysyl hydroxylase (LH, EC 1.14.11.4) is the enzyme catalyzing the formation of hydroxylysyl residues in collagens and other proteins with collagenous domains. Although lower species, such as Caenorhabditis elegans, have only one LH orthologue, LH activity in higher species, such as human, rat, and mouse, is present in three molecules, LH1, LH2, and LH3, encoded by three different genes. In addition, LH2 is present in two alternatively spliced forms (LH2a, LH2b). To understand the functions of the four molecular forms of LH in vertebrates, we analyzed differences in the binding and hydroxylation of various collagenous peptides by the LH isoforms. Nine-amino acid-long synthetic peptides on Pepspot were used for the binding analysis and an activity assay to measure hydroxylation. Our data with 727 collagenous peptides indicated that a positive charge on the peptide and specific amino acid residues in close proximity to the lysyl residues in the collagenous sequences are the key factors promoting peptide binding to the LH isoforms. The data suggest that the LH binding site is not a deep hydrophobic pocket but is open and hydrophilic where acidic amino acids play an important role in the binding. The data do not indicate strict sequence specificity for the LH isoforms, but the data indicated that there was a clear preference for some sequences to be bound and hydroxylated by a certain isoform.

Calculation of affinities of peptides for proteins.

Several methodologies were employed to calculate the Gibbs standard free energy of binding for a collection of protein-ligand complexes, where the ligand is a peptide and the protein is representative for various protein families. Almost 40 protein-ligand complexes were employed for a continuum approach, which considers the protein and the peptide at the atomic level, but includes solvent as a polarizable continuum. Five protein-ligand complexes were employed for an all-atom approach that relies on a combination of the double decoupling method with thermodynamic integration and molecular dynamics. These affinities were also computed by means of the linear interaction energy method. Although it generally proved rather difficult to predict the absolute free energies correctly, for some protein families the experimental ranking order was correctly reproduced by the continuum and all-atom approach. Considerable attention has also been given to correctly analyze the affinities of charged peptides, where it is required to judge the effect of one or more ions that are being decoupled in an all-atom approach to preserve electroneutrality. The various methods are further judged upon their merits.

Retrieval-independent localization of lysyl hydroxylase in the endoplasmic reticulum via a peptide fold in its iron-binding domain.

Lysyl hydroxylase (LH) is a peripheral membrane protein in the lumen of the endoplasmic reticulum (ER) that catalyses hydroxylation of lysine residues in collagenous sequences. Previously, we have mapped its primary ER localization motif within a 40-amino acid segment at its C-terminus. Here, we have characterized this localization mechanism in more detail, and our results indicate that this segment confers ER residency in a KDEL-receptor-independent manner, and without any apparent recycling of the enzyme between the Golgi apparatus and the ER. In addition, we show that a rather long peptide region, rather than a specific peptide sequence per se, is required for efficient retention of a reporter protein in the ER. Accordingly, the minimal retention motif was found to require the last 32 C-terminal amino acids, and sequential substitution of all five charged residues within this critical segment interfered only marginally with the retention or association of the enzyme with the ER membranes. Moreover, our fold-recognition and structure-prediction analyses suggested that this critical peptide segment forms an extended loop within LH's iron-binding domain, and that this loop is exposed and readily accessible for binding. Collectively, our results define a novel retrieval-independent retention mechanism in the ER.