Search

Gene expression is modulated by transcription factors (TFs), which are proteins that generally bind to DNA adjacent to coding regions and initiate transcription. Each target gene can be regulated by more than one TF, and each TF can regulate many targets. For a complete molecular understanding of transcriptional regulation, researchers must first associate each TF with the set of genes that it regulates. Here we present a summary of completed work on the ability to associate 104 TFs with their binding sites using support vector machines (SVMs), which are classification algorithms based in statistical learning theory. We use several types of genomic datasets to train classifiers in order to predict TF binding in the yeast genome. We consider motif matches, subsequence counts, motif conservation, functional annotation, and expression profiles. A simple weighting scheme varies the contribution of each type of genomic data when building a final SVM classifier, which we evaluate using known binding sites published in the literature and in online databases. The SVM algorithm works best when all datasets are combined, producing 73% coverage of known interactions, with a prediction accuracy of almost 0.9. We discuss new ideas and preliminary work for improving SVM classification of biological data.

Because cytotoxic T lymphocytes (CTL) may be important for preventing direct cell-to-cell transmission of human immunodeficiency virus (HIV), the agent responsible for acquired immunodeficiency syndrome, we have begun to investigate the epitope specificity and immune response (Ir) gene control of anti-HIV CTL responses in experimental animals. Mice were infected with a recombinant vaccinia virus expressing the HIV gp160 envelope gene, and the primed lymphocytes were restimulated in vitro with a transfected histocompatible cell line expressing the same gene. Our results show that H-2d mice are CTL high responders and H-2k mice are low responders to the HIV gp160 envelope protein under these conditions. Moreover, the H-2d mice respond predominantly to a single immunodominant site represented by a 15-residue synthetic peptide conforming to the amphipathic alpha-helix model of T-cell epitopes and seen by CD4- CD8+ CTL in association with the Dd class I major histocompatibility complex (MHC) molecules. The facts that CTL responses were detected in the context of only one of four class I MHC molecules tested and that the response was limited predominantly to a single epitope indicate that the CTL repertoire elicited by the HIV envelope protein in association with murine class I MHC molecules may be very limited. In addition, this epitope occurs in a highly variable segment of the envelope protein. This puts constraints on the use of a single peptide sequence from this region in a vaccine, as such a vaccine would have to be polyvalent. Nevertheless, this same variability suggests that this region may be under selective pressure from human CTL, and therefore that this site may be immunodominant in humans as well as mice and so of clinical importance in vaccine development.

We present an analysis that synthesizes information on the sequence, structure, and motifs of antigenic peptides, which previously appeared to be in conflict. Fourier analysis of T-cell antigenic peptides indicates a periodic variation in amino acid polarities of 3-3.6 residues per period, suggesting an amphipathic alpha-helical structure. However, the diffraction patterns of major histocompatibility complex (MHC) molecules indicate that their ligands are in an extended non-alpha-helical conformation. We present two mutually consistent structural explanations for the source of the alpha-helical periodicity, based on an observation that the side chains of MHC-bound peptides generally partition with hydrophobic (hydrophilic) side chains pointing into (out of) the cleft. First, an analysis of haplotype-dependent peptide motifs indicates that the locations of their defining residues tend to force a period 3-4 variation in hydrophobicity along the peptide sequence, in a manner consistent with the spacing of pockets in the MHC. Second, recent crystallographic determination of the structure of a peptide bound to a class II MHC molecule reveals an extended but regularly twisted peptide with a rotation angle of about 130 degrees. We show that similar structures with rotation angles of 100-130 degrees are energetically acceptable and also span the length of the MHC cleft. These results provide a sound physical chemical and structural basis for the existence of a haplotype-independent antigenic motif which can be particularly important in limiting the search time for antigenic peptides.

We propose, on the basis of physical chemical and biological requirements for T-cell activation by antigen, that sites on a protein that can stimulate T lymphocytes will be capable of forming a stable amphipathic structure (i.e., one with separated hydrophobic and hydrophilic surfaces), displaying periodicity in hydrophobic residues. A spectral analysis of the 12 antigenic sites to which the method could be applied indicates that the amphipathic periodicity hypothesis is valid for 10 of them, generally with reliabilities that are well above 98%, with periodicities compatible with an alpha-helical structure. An 11th case manifests a different type of amphipathicity. The analyses require only a knowledge of amino acid sequence. The finding that T-cell antigenic sites show a high correlation with amphipathicity greatly simplifies the search for such sites and is potentially important for vaccine development.

Much effort has been devoted to the analysis of antibodies to acquired immunodeficiency syndrome virus antigens, but no studies, to our knowledge, have defined antigenic sites of this virus that elicit T-cell immunity, even though such immunity is important in protection against many other viruses. T cells tend to recognize only a limited number of discrete sites on a protein antigen. Analysis of immunodominant helper T-cell sites has suggested that such sites tend to form amphipathic helices. An algorithm based on this model was used to identify two candidate T-cell sites, env T1 and env T2, in the envelope protein of human T-lymphotropic virus type IIIB that were conserved in other human immunodeficiency virus isolates. Corresponding peptides were synthesized and studied in genetically defined inbred and F1 mice for induction of lymph node proliferation. After immunization with a 426-residue recombinant envelope protein fragment, significant responses to native gp 120, as well as to each peptide, were observed in both F1 combinations studied. Conversely, immunization with env T1 peptide induced T-cell immunity to the native gp 120 envelope protein. The genetics of the response to env T1 peptide were further examined and revealed a significant response in three of four independent major histocompatibility haplotypes tested, an indication of high frequency responsiveness in the population. Identification of helper T-cell sites should facilitate development of a highly immunogenic, carrier-free vaccine that induces T-cell and B-cell immunity. The ability to elicit T-cell immunity to the native viral protein by immunization with a 16-residue peptide suggests that such sites represent potentially important components of an effective vaccine for acquired immunodeficiency syndrome.

We present a conceptual framework for evaluating the effect on the self-association of proteins in membranes due to the presence of other proteins at high concentrations (excluded volume effect) and the high concentration and preoriented state of the reactive species. We have calculated the magnitude of such effects using plausible values for the concentrations of proteins in membranes, for the degree to which proteins may tilt and move vertically, and for their dimensions. Compared to the association of monomers tumbling freely in an experimentally realistic volume, we calculate that these factors alone can increase the likelihood of forming dimers 10(6)-fold and of forming trimers and higher oligomers many orders of magnitude greater. We discuss the implications of our calculations for experimental manipulations of membrane proteins, for biosynthetic assembly of multisubunit membrane proteins and formation of membrane lesions by assemblies of exogenous proteins, and for the activation of cellular events induced by the interaction of membrane receptors with themselves or with other membrane proteins.

We develop a theory of ligand diffusion in the presence of a central potential toward or away from receptor patches that are uniformly distributed over a spherical cell. The current onto the receptors is reduced to less than that onto the sphere by a factor that is a nonlinear function of the number of free receptors, their size, and their potential energy. Similarly, under conditions defined by the theory, the dissociation rate from a receptor is reduced by the probability of rebinding to some other receptor on the same cell. This complicates the kinetic analysis, leading to the possibility of an occupancy-dependent dissociation rate, but has no effect on the interpretation of thermodynamic data.