Activating ERBB4 mutations in non-small cell lung cancer.

Recent efforts to comprehensively characterize the mutational landscape of non-small cell lung cancer have identified frequent mutations in the receptor tyrosine kinase ERBB4. However, the significance of mutated ERBB4 in non-small cell lung cancer remains elusive. Here, we have functionally characterized nine ERBB4 mutations previously identified in lung adenocarcinoma. Four out of the nine mutations, Y285C, D595V, D931Y and K935I, were found to be activating, increasing both basal and ligand-induced ErbB4 phosphorylation. According to structural analysis, the four activating mutations were located at critical positions at the dimerization interfaces of the ErbB4 extracellular (Y285C and D595V) and kinase (D931Y and K935I) domains. Consistently, the mutations enhanced ErbB4 dimerization and increased the trans activation in ErbB4 homodimers and ErbB4-ErbB2 heterodimers. The expression of the activating ERBB4 mutants promoted survival of NIH 3T3 cells in the absence of serum. Interestingly, serum starvation of NIH 3T3 cells expressing the ERBB4 mutants only moderately increased the phosphorylation of canonical ErbB signaling pathway effectors Erk1/2 and Akt as compared with wild-type ERBB4. In contrast, the mutations clearly enhanced the proteolytic release of signaling-competent ErbB4 intracellular domain. These results suggest the presence of activating driver mutations of ERBB4 in non-small cell lung cancer.

Identification and characterization of a novel luciferase-like protein in the human female reproductive tract.

A novel cDNA was cloned from human endometrium, matching a human gene with the interim name KIAA1463. An mRNA identified by 5'-rapid amplification of cDNA ends was found to be 3349 nt in length. PCR analysis also identified another transcript of 6626 nt, with an open reading frame encoding a 900 amino acid protein. A fold recognition program identified similarity to firefly luciferase containing an AMP-binding motif; hence, we refer to the predicted protein as the AMP binding/luciferase-like protein (ALLP). ALLP mRNA and protein were expressed throughout the female reproductive tract with the highest levels found in the ovary and uterus. In situ hybridization and immunohistochemistry showed predominant localization of the ALLP mRNA/protein in endometrial glandular epithelium and within the theca and granulosa cells in the ovary. In the endometrium expression of ALLP, mRNA and protein were higher during d 16-21 of the secretory phase of the cycle. Western blot analysis showed decreased expression of ALLP in the postmenopausal endometrium, and hormone replacement therapy increased the expression of ALLP. Endometrial adenocarcinoma cell lines expressed more ALLP, compared with cultured primary endometrial cells or normal endometrial tissue. The ubiquitous expression of ALLP in reproductive and nonreproductive tissues suggests that this protein, which is probably regulated by ovarian steroids, plays an important metabolic role and may be involved in such processes as implantation and tumorigenesis.

A fragment library based on Gaussian mixtures predicting favorable molecular interactions.

Here, a protein atom-ligand fragment interaction library is described. The library is based on experimentally solved structures of protein-ligand and protein-protein complexes deposited in the Protein Data Bank (PDB) and it is able to characterize binding sites given a ligand structure suitable for a protein. A set of 30 ligand fragment types were defined to include three or more atoms in order to unambiguously define a frame of reference for interactions of ligand atoms with their receptor proteins. Interactions between ligand fragments and 24 classes of protein target atoms plus a water oxygen atom were collected and segregated according to type. The spatial distributions of individual fragment - target atom pairs were visually inspected in order to obtain rough-grained constraints on the interaction volumes. Data fulfilling these constraints were given as input to an iterative expectation-maximization algorithm that produces as output maximum likelihood estimates of the parameters of the finite Gaussian mixture models. Concepts of statistical pattern recognition and the resulting mixture model densities are used (i) to predict the detailed interactions between Chlorella virus DNA ligase and the adenine ring of its ligand and (ii) to evaluate the "error" in prediction for both the training and validation sets of protein-ligand interaction found in the PDB. These analyses demonstrate that this approach can successfully narrow down the possibilities for both the interacting protein atom type and its location relative to a ligand fragment.

Adenine recognition: a motif present in ATP-, CoA-, NAD-, NADP-, and FAD-dependent proteins.

Adenosine triphosphate (ATP) plays an essential role in energy transfer within the cell. In the form of NAD, adenine participates in multiple redox reactions. Phosphorylation and ATP-hydrolysis reactions have key roles in signal transduction and regulation of many proteins, especially enzymes. In each cell, proteins with many different functions use adenine and its derivatives as ligands; adenine, of course, is present in DNA and RNA. We show that an adenine binding motif, which differs according to the backbone chain direction of a loop that binds adenine (and in one variant by the participation of an aspartate side-chain), is common to many proteins; it was found from an analysis of all adenylate-containing protein structures from the Protein Data Bank. Indeed, 224 protein-ligand complexes (86 different proteins) from a total of 645 protein structure files bind ATP, CoA, NAD, NADP, FAD, or other adenine-containing ligands, and use the same structural elements to recognize adenine, regardless of whether the ligand is a coenzyme, cofactor, substrate, or an allosteric effector. The common adenine-binding motif shown in this study is simple to construct. It uses only (1) backbone polar interactions that are not dependent on the protein sequence or particular properties of amino acid side-chains, and (2) nonspecific hydrophobic interactions. This is probably why so many different proteins with different functions use this motif to bind an adenylate-containing ligand. The adenylate-binding motif reported is present in "ancient proteins" common to all living organisms, suggesting that adenine-containing ligands and the common motif for binding them were exploited very early in evolution. The geometry of adenine binding by this motif mimics almost exactly the geometry of adenine base-pairing seen in DNA and RNA.

When fold is not important: a common structural framework for adenine and AMP binding in 12 unrelated protein families.

ATP is a ligand common to many proteins, yet it is unclear whether common recognition patterns do exist among the many different folds that bind ATP. Previously, it was shown that cAMP-dependent protein kinase, D-Ala:D-Ala ligase and the alpha-subunit of the alpha 2 beta 2 ribonucleotide reductase do share extensive common structural elements for ATP recognition although their folds are different. Here, we have made a survey of structures that bind ATP and compared them with the key features seen in these three proteins. Our survey shows that 12 different fold types share a specific recognition pattern for the adenine moiety, and 8 of these folds have a common structural framework for recognition of the AMP moiety of the ligand. The common framework consists of a tripeptide segment plus three additional residues, which provides similar polar and hydrophobic interactions between the protein and mononucleotide. Consensus interactions are represented by four key hydrogen bonds present in each fold type. Two of these four hydrogen bonds, together with three aliphatic residues, form a specific recognition pattern for the adenine moiety in all 12 folds. These similarities point to a structural-functional requirement shared by these different mononucleotide-binding proteins that represent at this time 28% of the adenine mononucleotide complexes found in the Brookhaven Protein Data Bank.

Common structural elements in the architecture of the cofactor-binding domains in unrelated families of pyridoxal phosphate-dependent enzymes.

A detailed comparison of the structures of aspartate aminotransferase, alanine race-mase, the beta subunit of tryptophan synthase, D-amino acid aminotransferase and glycogen phosphorylase has revealed more extensive structural similarities among pyridoxal phosphate (PLP)-binding domains in these enzymes than was observed previously. These similarities consist of seven common structural segments of the polypeptide chain, which form an extensive common structural organization of the backbone chain responsible for the appropriate disposition of key residues, some from the aligned fragments and some from variable loops joined to these fragments, interacting with PLPs in these enzymes. This common structural organization contains an analogous hydrophobic minicore formed from four amino acid side chains present in the two most conserved structural elements. In addition, equivalent alpha-beta-alpha-beta supersecondary structures are formed by these seven fragments in three of the five structures: alanine racemase, tryptophan synthase and glycogen phosphorylase. Despite these similarities, it is generally accepted that these proteins do not share a common heritage, but have arisen on five separate occasions. The common and contiguous alpha-beta-alpha-beta structure accounts for only 28 residues and all five enzymes differ greatly in both the orientation of the PLP pyridoxal rings and their contacts with residues close to the common structural elements.

Finding local structural similarities among families of unrelated protein structures: a generic non-linear alignment algorithm.

We have developed a generic tool for the automatic identification of regions of local structural similarity in unrelated proteins having different folds, as well as for defining more global similarities that result from homologous protein structures. The computer program GENFIT has evolved from the genetic algorithm-based three-dimensional protein structure comparison program GA_FIT. GENFIT, however, can locate and superimpose regions of local structural homology regardless of their position in a pair of structures, the fold topology, or the chain direction. Furthermore, it is possible to restrict the search to a volume centered about a region of interest (e.g., catalytic site, ligand-binding site) in two protein structures. We present a number of examples to illustrate the function of the program, which is a parallel processing implementation designed for distribution to multiple machines over a local network or to run on a single multiprocessor computer.

Analogous conformations of both binding and effector regions in cyclosporin A, FK506 and rapamycin.

Three immunosuppressant drugs, cyclosporin A, FK506 and rapamycin were compared in their three-dimensional structures by computer modelling. The pairwise comparisons of cyclosporin A, FK506 and rapamycin show two structurally common fragments. One fragment is Mle9-Bmt1 region in cyclosporin A, C22-O5 region in FK506 and C29-O5 region in rapamycin. Another fragment is Mle4-Mle6 region in cyclosporin A and C14-C21 regions in FK506 and rapamycin. The correspondence of the structurally analogous regions with the regions which are involved in the interactions with peptidyl-prolyl cis/trans isomerases and calcineurin or FKBP-rapamycin-associated protein is discussed.

Two "unrelated" families of ATP-dependent enzymes share extensive structural similarities about their cofactor binding sites.

Two proteins, D-alanine:D-alanine ligase and cAMP-dependent protein kinase, share a remarkable degree of structural convergence despite having different three-dimensional folds and different enzymatic functions. Here we report that as many as 103 residues from 10 segments form two identical super-secondary structures between which the cofactor ATP is bound. The cofactor, two bound metal cations, and several water molecules form a large network of electrostatic and hydrophobic interactions common to both enzymes, and these are mediated by the similar placement of equivalent amino acids within the common supersecondary structures.

Molecular models of two competitive inhibitors, IL-2delta2 and IL-2delta3, generated by alternative splicing of human interleukin-2.

Molecular models of IL-2delta2 and IL-2delta3, two alternative splice variants of human IL-2 without exon 2 and 3, respectively, are described. These alternative splice variants attract particular interest as potential competitive inhibitors of the cytokine. Tertiary structure of IL-2 consists of four-helix bundle including helices A, B, C and D and a beta-pleated sheet. Exon 2 encodes the A-B loop (Asn30-Lys49 residues) linking helices A and B running in one direction. Rotation of the helix A around putative centre during the construction of IL-2delta2 model have not produced any significant changes in the hydrophobic core of IL-2 molecule. However, a large hole was formed on the surface of IL-2delta2 molecule instead of A-B loop in IL-2 fold. A high affinity IL-2 receptor is formed by combination of alpha, beta, and gamma(c) chains. Comparison of the model of the receptor bound IL-2 with the model of IL-2delta2 has shown that their beta-chain binding sites have minimum differences as distinct from alpha and gamma(c) chain-binding sites. Exon 3 encodes Ala50-Lys97 fragment which forms helices B and C with their short connecting loop. Model IL-2delta3 consists of helices A and D and long linking loop. This loop was composed of A-B and C-D loops which run in opposite directions in IL-2 structure and contain beta-strands making a beta-pleated sheet. Conformation of the linking loop relatively to helices A and D was stabilized by creation of a disulphide bond between cysteines 105 and 125. In addition, the hydrophobic residues of beta-sheet interact with the hydrophobic surface of A-D helical complex and close the latter from contacts with solution. Comparison of the model of IL-2 bound to receptor with IL-2delta3 model has shown that absence of helices B and C in IL-2delta3 model results in insignificant conformational changes only in residues interacting with gamma(c) chain of the receptor. The beta/gamma(c) heterodimer is an intermediate affinity receptor of IL-2. Most likely, both IL-2delta2 and IL-2delta3 are naturally occurring IL-2 antagonists since they keep the ability of binding with an intermediate affinity receptor of this cytokine and fail to engage the alpha chain of its high affinity receptor.

Enzyme-mononucleotide interactions: three different folds share common structural elements for ATP recognition.

Three ATP-dependent enzymes with different folds, cAMP-dependent protein kinase, D-Ala:D-Ala ligase and the alpha-subunit of the alpha2beta2 ribonucleotide reductase, have a similar organization of their ATP-binding sites. The most meaningful similarity was found over 23 structurally equivalent residues in each protein and includes three strands each from their beta-sheets, in addition to a connecting loop. The equivalent secondary structure elements in each of these enzymes donate four amino acids forming key hydrogen bonds responsible for the common orientation of the "AMP" moieties of their ATP-ligands. One lysine residue conserved throughout the three families binds the alpha-phosphate in each protein. The common fragments of structure also position some, but not all, of the equivalent residues involved in hydrophobic contacts with the adenine ring. These examples of convergent evolution reinforce the view that different proteins can fold in different ways to produce similar structures locally, and nature can take advantage of these features when structure and function demand it, as shown here for the common mode of ATP-binding by three unrelated proteins.