Mixture models with multiple levels, with application to the analysis of multifactor gene expression data

Model-based clustering is a popular tool for summarizing high-dimensional data. With the number of high-throughput large-scale gene expression studies still on the rise, the need for effective data- summarizing tools has never been greater. By grouping genes according to a common experimental expression profile, we may gain new insight into the biological pathways that steer biological processes of interest. Clustering of gene profiles can also assist in assigning functions to genes that have not yet been functionally annotated. In this paper, we propose 2 model selection procedures for model-based clustering. Model selection in model-based clustering has to date focused on the identification of data dimensions that are relevant for clustering. However, in more complex data structures, with multiple experimental factors, such an approach does not provide easily interpreted clustering outcomes. We propose a mixture model with multiple levels, , that provides sparse representations both "within" and "between" cluster profiles. We explore various flexible "within-cluster" parameterizations and discuss how efficient parameterizations can greatly enhance the objective interpretability of the generated clusters. Moreover, we allow for a sparse "between-cluster" representation with a different number of clusters at different levels of an experimental factor of interest. This enhances interpretability of clusters generated in multiple-factor contexts. Interpretable cluster profiles can assist in detecting biologically relevant groups of genes that may be missed with less efficient parameterizations. We use our multilevel mixture model to mine a proliferating cell line expression data set for annotational context and regulatory motifs. We also investigate the performance of the multilevel clustering approach on several simulated data sets.     

A human monoclonal antibody to a complex epitope in the V3 region of gp120 of human immunodeficiency virus type 1 has broad reactivity within and outside clade B.

We have used virus neutralization and antibody-binding techniques to define the epitope for a human monoclonal antibody, designated 19b, within the V3 region of the gp120 surface glycoprotein of human immunodeficiency virus type 1. Unusually, the 19b epitope encompasses residues on both flanks of the V3 loop. However, 19b binding to gp120 is independent of sequences at the crown of the V3 loop, provided that they are compatible with the formation of a type II beta turn that is presumably necessary to juxtapose the antigenic residues on the V3 flanks. By comparing the V3 sequences of virus gp120s able and unable to bind 19b, we were able to define the canonical 19b epitope as -I----G--FY-T, where residues at the positions indicated by the gaps do not contribute directly to the 19b-binding site. A few conservative substitutions at the more critical residues are also compatible with 19b binding. Inspection of V3 sequences in the human immunodeficiency virus database indicated that the canonical 19b epitope is well conserved among isolates from the North American-European clade B and also among clade E isolates from Thailand and clade F isolates from Brazil. A minority of gp120s from clades A and C also possess the 19b epitope. Consistent with the theoretical predictions of its cross-clade reactivity, 19b was found to bind to gp120s from clades A, B, C, E, and F in immunoassays. However, 19b was not able to reduce the infectivity of primary viruses from clades A, E, and F that were predicted to possess the 19b epitope and only modestly reduced the infectivity of a clade C virus at low input virus concentrations. Cross-clade neutralization via V3-directed antibodies may, therefore, be difficult, even if the antibodies show broad reactivities in binding assays and the viruses theoretically possess the relevant binding site.     

Optimization of bovine serum albumin sorption and recovery by hydrogels

Aqueous two-phase systems are composed of aqueous solutions of either two water-soluble polymers, usually polyethylene glycol (PEG) and dextran (Dx), or a polymer and a salt, usually PEG and phosphate or sulfate. Partitioning of proteins in such systems provides a powerful method for separating and purifying mixtures of biomolecules by extraction. If one of the phase forming polymers is a crosslinked gel, then the solution-controlled gel sorption may be considered as a modification of aqueous two-phase extraction. Since PEG/dextran systems are widely used in aqueous two-phase extraction and dextran gels (Sephadex) are common chromatographic media, we choose a PEG/dextran gel system as a model system in this study. The partitioning behavior of pure bovine serum albumin (BSA) in PEG/dextran gel systems is investigated to see the effects of variations in PEG and NaCl concentrations on the partition coefficient K. By making use of the Box-Wilson experimental design, K is shown to be maximized at 9.8 (%, w/w) PEG and 0.2 M NaCl concentrations, respectively, as 182.     

Inter‐annual and body topographic consistency in the plumage bacterial load of Great Tits

ABSTRACT Plumage bacteria may play an important role in shaping the life histories of birds. However, to design suitable experiments to examine causal relationships between plumage bacteria and the fitness of host birds, natural variation in plumage bacterial communities needs to be better understood. We examined within‐individual consistency of plumage bacterial contamination in Great Tits (Parus major), comparing different body regions (ventral vs. dorsal) and comparing contamination between years. Numbers of free‐living and attached bacteria and the species richness of feather‐degrading bacterial assemblages were studied using flow cytometry and ribosomal intergenic spacer analysis (RISA). Numbers of both types of bacteria were higher on dorsal than on ventral feathers. Numbers of free‐living, but not attached, bacteria on the two body regions were highly positively correlated. There was also a strong within‐individual correlation between numbers of attached bacteria during the same breeding stages in different years. These results suggest that, despite variation in absolute levels of feather bacterial loads between years and different body regions, sampling individual birds can provide reliable estimates of relative levels of bacterial contamination, as long as sampling time and body region are carefully standardized.     

Determination of the basicity of Mannich ketones by capillary electrophoresis

Capillary electrophoretic (CE) method to characterise Mannich ketones (MKs) containing morpholine moiety was developed. Basicity (pKa,exp) of the MKs was characterised by measured data derived from the electrophoretic mobility values obtained in the CE separation. The MKs were found to be weaker bases than the parent morpholine molecule itself and the experimentally determined basicity values proved to be dependent on the chemical structure of the MKs. Since the basicity of the MKs has an influence on their reactivity and biological activity it seems to be useful to determine experimentally the pKa,exp values of the newly synthesised compounds to support rational drug design or screening of the molecule libraries.     

Size characteristics of human leucocyte interferon under reducing and non-reducing conditions.

Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis was used to characterize human leucocyte interferon (HuIFN-alpha) under reducing and non-reducing conditions. Under non-reducing conditions HuIFN-alpha possesses two size forms, but under reducing conditions (r-HuIFN-alpha) only one is observed. The apparent molecular weight of this one form varies with the concentration of 2-mercaptoethanol used. When r-HuIFN-alpha is permitted to reoxidize the bimodal configuration of HuIFN-alpha is restored. The size heterogeneity of native HuIFN-alpha can be eliminated by mild treatment with NaIO4 [HuIFN-alpha/IO4; Stewart II, Lin, Wiranowska-Stewart & Cantell (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 4200-4204]. The size of the HuIFN-alpha/IO4 increases after treatment with 2-mercaptoethanol (r-HuIFN-alpha/IO4) and the apparent molecular weight of this component also varies with the concentration of 2-mercaptoethanol used. In the case of r-HuIFN-alpha the single peak observed apparently originates from both the higher- and lower-molecular-weight components.     

Cerebellar granule cells show age-dependent migratory differences in vitro

Developmental differences between cerebellar granule cells during their migratory period were revealed using dissociated granule cell cultures isolated from 4, 7, or 10 days old (P4, P7, P10) mice. Under all culture conditions, the great majority of cultivated cell populations consisted of those granule cells that had not reach their final destination in the internal granule cell layer (IGL) by the age of isolation. In vitro morphological development and the expression of migratory markers (TAG-1, astrotactin, or EphB2) showed similar characteristics between the cultures. The migration of 1008 granule cells isolated from P4, P7, and P10 cerebella and cultivated under identical conditions were analyzed using statistical methods. In vitro time-lapse videomicroscopy revealed that P4 cells possessed the fastest migratory speed while P10 granule cells retained their migratory activity for the longest time in culture. Cultures obtained from younger postnatal ages showed more random migratory trajectories than P10 cultures. Our observations indicate that despite similar morphological and molecular properties, migratory differences exist in granule cell cultures isolated from different postnatal ages. Therefore, the age of investigation can substantially influence experimental results on the regulation of cell migration.     

The protonation state of a heme propionate controls electron transfer in cytochrome c oxidase

In cytochrome c oxidase (CcO), exergonic electron transfer reactions from cytochrome c to oxygen drive proton pumping across the membrane. Elucidation of the proton pumping mechanism requires identification of the molecular components involved in the proton transfer reactions and investigation of the coupling between internal electron and proton transfer reactions in CcO. While the proton-input trajectory in CcO is relatively well characterized, the components of the output pathway have not been identified in detail. In this study, we have investigated the pH dependence of electron transfer reactions that are linked to proton translocation in a structural variant of CcO in which Arg481, which interacts with the heme D-ring propionates in a proposed proton output pathway, was replaced with Lys (RK481 CcO). The results show that in RK481 CcO the midpoint potentials of hemes a and a(3) were lowered by approximately 40 and approximately 15 mV, respectively, which stabilizes the reduced state of Cu(A) during reaction of the reduced CcO with O(2). In addition, while the pH dependence of the F --> O rate in wild-type CcO is determined by the protonation state of two protonatable groups with pK(a) values of 6.3 and 9.4, only the high-pK(a) group influences this rate in RK481 CcO. The results indicate that the protonation state of the Arg481 heme a(3) D-ring propionate cluster having a pK(a) of approximately 6.3 modulates the rate of internal electron transfer and may act as an acceptor of pumped protons.     

Reactions of 3-isochromanone with aromatic aldehydes--microwave assisted condensations performed on solid basic inorganic supports

An improved Knoevenagel condensation of 3-isochromanone and aromatic aldehydes can be achieved by microwave irradiation on solid supports in the presence of various catalysts. This synthetic method offers some major advantages, especially the possibility to change the ratio of E/Z isomers.