Shallow Dialogue Processing Using Machine Learning Algorithms (or not)

This paper presents a shallow dialogue analysis model, aimed at human-human dialogues in the context of staff or business meetings. Four components of the model are defined, and several machine learning techniques are used to extract features from dialogue transcripts: maximum entropy classifiers for dialogue acts, latent semantic analysis for topic segmentation, or decision tree classifiers for discourse markers. A rule-based approach is proposed for solving cross-modal references to meeting documents. The methods are trained and evaluated thanks to a common data set and annotation format. The integration of the components into an automated shallow dialogue parser opens the way to multimodal meeting processing and retrieval applications.

Macromolecular interaction studies on FF1-3 of human CA150 and STAT1-importin alpha5 using biophysical and biochemical methods

Macromolecular interactions, such as protein-protein interactions and protein-DNA interactions, play important roles in executing biological functions in cells. However the complexity of such interactions often makes it very challenging to elucidate the structural details of these subjects. In this thesis, two different research strategies were applied on two different two macromolecular systems: X-ray crystallography on three tandem FF domains of transcription regulator CA150 and electron microscopy on STAT1-importin α5 complex. The results from these studies provide novel insights into the function-structure relationships of transcription coupled RNA splicing mediated by CA150 and the nuclear import process of the JAK-STAT signaling pathway. ^ The first project aimed at the protein-protein interaction module FF domain, which often occurs as tandem repeats. Crystallographic structure of the first three FF domains of human CA150 was determined to 2.7 Å resolution. This is the only crystal structure of an FF domain and the only structure on tandem FF domains to date. It revealed a striking connectivity between an FF domain and the next. Peptide binding assay with the potential binding ligand of FF domains was performed using fluorescence polarization. Furthermore, for the first time, FF domains were found to potentially interact with DNA. DNA binding assays were also performed and the results were supportive to this newly proposed functionality of an FF domain. ^ The second project aimed at understanding the molecular mechanism of the nuclear import process of transcription factor STAT1. The first structural model of pSTAT1-importin α5 complex in solution was built from the images of negative staining electron microscopy. Two STAT1 molecules were observed to interact with one molecule of importin α5 in an asymmetric manner. This seems to imply that STAT1 interacts with importin α5 with a novel mechanism that is different from canonical importin α-cargo interactions. Further in vitro binding assays were performed to obtain more details on the pSTAT1-importin α5 interaction. ^

Generation of human computational models with machine learning

Services in smart environments pursue to increase the quality of people?s lives. The most important issues when developing this kind of environments is testing and validating such services. These tasks usually imply high costs and annoying or unfeasible real-world testing. In such cases, artificial societies may be used to simulate the smart environment (i.e. physical environment, equipment and humans). With this aim, the CHROMUBE methodology guides test engineers when modeling human beings. Such models reproduce behaviors which are highly similar to the real ones. Originally, these models are based on automata whose transitions are governed by random variables. Automaton?s structure and the probability distribution functions of each random variable are determined by a manual test and error process. In this paper, it is presented an alternative extension of this methodology which avoids the said manual process. It is based on learning human behavior patterns automatically from sensor data by using machine learning techniques. The presented approach has been tested on a real scenario, where this extension has given highly accurate human behavior models,

Experimental tests in human-robot collision evaluation and characterization of a new safety index for robot operation

This paper describes an experimental procedure consisting of impact tests that simulate a collision of a human head with an industrial robot with the aim to validate a safety index named as New Index for Robots (NIR) and its outputs. The experiments in this paper are based on lab tests. It is an attempt to characterize the NIR index underlying the main parameters that are involved in crash interaction and to highlight limitations and weakness of suggested impact tests.

Interaction of the human androgen receptor transactivation function with the general transcription factor TFIIF

The human androgen receptor (AR) is a ligand-activated transcription factor that regulates genes important for male sexual differentiation and development. To better understand the role of the receptor as a transcription factor we have studied the mechanism of action of the N-terminal transactivation function. In a protein–protein interaction assay the AR N terminus (amino acids 142–485) selectively bound to the basal transcription factors TFIIF and the TATA-box-binding protein (TBP). Reconstitution of the transactivation activity in vitro revealed that AR142–485 fused to the LexA protein DNA-binding domain was competent to activate a reporter gene in the presence of a competing DNA template lacking LexA binding sites. Furthermore, consistent with direct interaction with basal transcription factors, addition of recombinant TFIIF relieved squelching of basal transcription by AR142–485. Taken together these results suggest that one mechanism of transcriptional activation by the AR involves binding to TFIIF and recruitment of the transcriptional machinery.

Evidence for two nonidentical drug-interaction sites in the human P-glycoprotein

Human P-glycoprotein (Pgp) confers multidrug resistance to cancer cells by ATP-dependent extrusion of a great many structurally dissimilar hydrophobic compounds. The manner in which Pgp recognizes these different substrates is unknown. The protein shows internal homology between its N- and C-terminal halves, each comprised of six putative transmembrane helices and a consensus ATP binding/utilization site. Photoactive derivatives of certain Pgp substrates specifically label two regions, one on each half of the protein. In this study, using [125I]iodoarylazidoprazosin ([125I]IAAP), a photoactive analog of prazosin, we have demonstrated the presence of two nonidentical drug-interaction sites within Pgp. Taking advantage of a highly susceptible trypsin cleavage site in the linker region of Pgp, we characterized the [125I]IAAP binding to the N- and C-terminal halves. cis(Z)-Flupentixol, a modulator of Pgp function, preferentially increased the affinity of [125I]IAAP for the C-terminal half of the protein (C-site) by reducing the Kd from 20 to 6 nM without changing the labeling or affinity (Kd = 42–46 nM) of the N-terminal half (N-site). Also, the concentration of vinblastine (Pgp substrate) and cyclosporin A (Pgp modulator) required for 50% inhibition of [125I]IAAP binding to the C-site was increased 5- to 6-fold by cis(Z)-flupentixol without any effect on the N-site. In addition, [125I]IAAP binding to the N-site was less susceptible than to C-site to inhibition by vanadate which blocks ATP hydrolysis and drug transport. These data demonstrate the presence of at least two nonidentical substrate interaction sites in Pgp.

Interaction of human apurinic endonuclease and DNA polymerase β in the base excision repair pathway

Mutagenic abasic (AP) sites are generated directly by DNA-damaging agents or by DNA glycosylases acting in base excision repair. AP sites are corrected via incision by AP endonucleases, removal of deoxyribose 5-phosphate, repair synthesis, and ligation. Mammalian DNA polymerase β (Polβ) carries out most base excision repair synthesis and also can excise deoxyribose 5-phosphate after AP endonuclease incision. Yeast two-hybrid analysis now indicates protein–protein contact between Polβ and human AP endonuclease (Ape protein). In vitro, binding of Ape protein to uncleaved AP sites loads Polβ into a ternary complex with Ape and the AP-DNA. After incision by Ape, only Polβ exhibits stable DNA binding. Kinetic experiments indicated that Ape accelerates the excision of 5′-terminal deoxyribose 5-phosphate by Polβ. Thus, the two central players of the base excision repair pathway are coordinated in sequential reactions.

Interaction of Class I Human Leukocyte Antigen (HLA-I) Molecules with Insulin Receptors and Its Effect on the Insulin-Signaling Cascade

Insulin receptor (IR) and class I major histocompatibility complex molecules associate with one another in cell membranes, but the functional consequences of this association are not defined. We found that IR and human class I molecules (HLA-I) associate in liposome membranes and that the affinity of IR for insulin and its tyrosine kinase activity increase as the HLA:IR ratio increases over the range 1:1 to 20:1. The same relationship between HLA:IR and IR function was found in a series of B-LCL cell lines. The association of HLA-I and IR depends upon the presence of free HLA heavy chains. All of the effects noted were reduced or abrogated if liposomes or cells were incubated with excess HLA-I light chain, β2-microglobulin. Increasing HLA:IR also enhanced phosphorylation of insulin receptor substrate-1 and the activation of phosphoinositide 3-kinase. HLA-I molecules themselves were phosphorylated on tyrosine and associated with phosphoinositide 3-kinase when B-LCL were stimulated with insulin.

Functional interaction between c-Jun and promoter factor Sp1 in epidermal growth factor-induced gene expression of human 12(S)-lipoxygenase

The functional role of the interaction between c-Jun and simian virus 40 promoter factor 1 (Sp1) in epidermal growth factor (EGF)-induced expression of 12(S)-lipoxygenase gene in human epidermoid carcinoma A431 cells was studied. Coimmunoprecipitation experiments indicated that EGF stimulated interaction between c-Jun and Sp1 in a time-dependent manner. Overexpression of Ha-ras and c-Jun also enhanced the amount of c-Jun binding to Sp1. In addition, the c-Jun dominant negative mutant TAM-67 not only inhibited the coimmunoprecipitated c-Jun binding to Sp1 in a dose-dependent manner in cells overexpressing c-Jun but also reduced promoter activity of the 12(S)-lipoxygenase gene induced by c-Jun overexpression. Treatment of cells with EGF increased the interaction between the Sp1 oligonucleotide and nuclear c-Jun/Sp1 in a time-dependent manner. Furthermore, EGF activated the chimeric promoter consisting of 10 tandem GAL4-binding sites, which replaced the three Sp1-binding sites in the 12(S)lipoxygenase promoter only when coexpressed with GAL4-c-Jun () fusion proteins. These results indicate that the direct interaction between c-Jun and Sp1 induced by EGF cooperatively activated expression of the 12(S)-lipoxygenase gene, and that Sp1 may serve at least in part as a carrier bringing c-Jun to the promoter, thus transactivating the transcriptional activity of 12(S)-lipoxygenase gene.

ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human N-CoR/mSin3/HDAC1 complex

The t(8;21) translocation between two genes known as AML1 and ETO is seen in approximately 12–15% of all acute myeloid leukemia (AML) and is the second-most-frequently observed nonrandom genetic alteration associated with AML. AML1 up-regulates a number of target genes critical to normal hematopoiesis, whereas the AML1/ETO fusion interferes with this trans-activation. We discovered that the fusion partner ETO binds to the human homolog of the murine nuclear receptor corepressor (N-CoR). The interaction is mediated by two unusual zinc finger motifs present at the carboxyl terminus of ETO. Human N-CoR (HuN-CoR), which we cloned and sequenced in its entirety, encodes a 2,440-amino acid polypeptide and has a central domain that binds ETO. N-CoR, mammalian Sin3 (mSin3A and B), and histone deacetylase 1 (HDAC1) form a complex that alters chromatin structure and mediates transcriptional repression by nuclear receptors and by a number of oncoregulatory proteins. We found that ETO, through its interaction with the N-CoR/mSin3/HDAC1 complex, is also a potent repressor of transcription. This observation provides a mechanism for how the AML1/ETO fusion may inhibit expression of AML1-responsive target genes and disturb normal hematopoiesis.

Translational control of human p53 expression in yeast mediated by 5′-UTR–ORF structural interaction

We have expressed human p53 cDNA in the yeast Saccharomyces cerevisiae and shown that the level of production and the length of the p53 protein depends on the presence of untranslated mRNA regions (UTRs). The expression of the ORF alone leads to a p53 protein of correct size (53 kDa) that accumulates to high levels, concomitantly with the presence of a small amount of a p40 protein (40 kDa). However, when either the entire 5′-UTR and a part of the 3′- or 5′-UTR alone is used, this leads to the production of small amounts of the 40 kDa truncated form only. The p40 protein corresponds to a truncated form of p53 at the C-terminal extremity since it reacts only with a monoclonal antibody recognising the N-terminal epitope. This effect on the amount and length of p53 protein had no correlation at the mRNA level, suggesting that translational control probably occurs through the 5′-UTR. We propose a model of structural interaction between this UTR and a part of the ORF mRNA for the regulation of p53 expression in this heterologous context.

Interaction between human tRNA synthetases involves repeated sequence elements.

Aminoacyl-tRNA synthetases (tRNA synthetases) of higher eukaryotes form a multiprotein complex. Sequence elements that are responsible for the protein assembly were searched by using a yeast two-hybrid system. Human cytoplasmic isoleucyl-tRNA synthetase is a component of the multi-tRNA synthetase complex and it contains a unique C-terminal appendix. This part of the protein was used as bait to identify an interacting protein from a HeLa cDNA library. The selected sequence represented the internal 317 amino acids of human bifunctional (glutamyl- and prolyl-) tRNA synthetase, which is also known to be a component of the complex. Both the C-terminal appendix of the isoleucyl-tRNA synthetase and the internal region of bifunctional tRNA synthetase comprise repeating sequence units, two repeats of about 90 amino acids, and three repeats of 57 amino acids, respectively. Each repeated motif of the two proteins was responsible for the interaction, but the stronger interaction was shown by the native structures containing multiple motifs. Interestingly, the N-terminal extension of human glycyl-tRNA synthetase containing a single motif homologous to those in the bifunctional tRNA synthetase also interacted with the C-terminal motif of the isoleucyl-tRNA synthetase although the enzyme is not a component of the complex. The data indicate that the multiplicity of the binding motif in the tRNA synthetases is necessary for enhancing the interaction strength and may be one of the determining factors for the tRNA synthetases to be involved in the formation of the multi-tRNA synthetase complex.

Human protein Sam68 relocalization and interaction with poliovirus RNA polymerase in infected cells.

A HeLa cDNA expression library was screened for human polypeptides that interacted with the poliovirus RNA-dependent RNA polymerase, 3D, using the two-hybrid system in the yeast Saccharomyces cerevisiae. Sam68 (Src-associated in mitosis, 68 kDa) emerged as the human cDNA that, when fused to a transcriptional activation domain, gave the strongest 3D interaction signal with a LexA-3D hybrid protein. 3D polymerase and Sam68 coimmunoprecipitated from infected human cell lysates with antibodies that recognized either protein. Upon poliovirus infection, Sam68 relocalized from the nucleus to the cytoplasm, where poliovirus replication occurs. Sam68 was isolated from infected cell lysates with an antibody that recognizes poliovirus protein 2C, suggesting that it is found on poliovirus-induced membranes upon which viral RNA synthesis occurs. These data, in combination with the known RNA- and protein-binding properties of Sam68, make Sam68 a strong candidate for a host protein with a functional role in poliovirus replication.

Tumor dormancy and cell signaling: anti-mu-induced apoptosis in human B-lymphoma cells is not caused by an APO-1-APO-1 ligand interaction.

Signal transduction initiated by crosslinking of antigen-specific receptors on T- and B-lymphoma cells induces apoptosis. In T-lymphoma cells, such crosslinking results in upregulation of the APO-1 ligand, which then interacts with induced or constitutively expressed APO-1, thereby triggering apoptosis. Here we show that crosslinking the membrane immunoglobulin on human lymphoma cells (Daudi) (that constitutively express APO-1) does not induce synthesis of APO-1 ligand. Further, a noncytotoxic fragment of anti-APO-1 antibody that blocks T-cell-receptor-mediated apoptosis in T-lymphoma cells does not block anti-mu-induced apoptosis. Hence, in B-lymphoma cells, apoptosis induced by signaling via membrane IgM is not mediated by the APO-1 ligand.

Production of avian leukosis virus particles in mammalian cells can be mediated by the interaction of the human immunodeficiency virus protein Rev and the Rev-responsive element.

In human immunodeficiency virus type 1-infected cells, the efficient expression of viral proteins from unspliced and singly spliced RNAs is dependent on two factors: the presence in the cell of the viral protein Rev and the presence in the viral RNA of the Rev-responsive element (RRE). We show here that the HIV-1 Rev/RRE system can increase the expression of avian leukosis virus (ALV) structural proteins in mammalian cells (D-17 canine osteosarcoma) and promote the release of mature ALV virions from these cells. In this system, the Rev/RRE interaction appears to facilitate the export of full-length unspliced ALV RNA from the nucleus to the cytoplasm, allowing increased production of the ALV structural proteins. Gag protein is produced in the cytoplasm of the ALV-transfected cells even in the absence of a Rev/RRE interaction. However, a functional Rev/RRE interaction increases the amount of Gag present intracellularly and, more strikingly, results in the release of mature ALV particles into the supernatant. RCAS virus containing an RRE is replication-competent in chicken embryo fibroblasts; however, we have been unable to determine whether the particles produced in D-17 cells are as infectious as the particles produced in chicken embryo fibroblasts.