Salk Institute for Biological Studies

We derive a new self-organizing learning algorithm that maximizes the information transferred in a network of nonlinear units. The algorithm does not assume any knowledge of the input distributions, and is defined here for the zero-noise limit. Under these conditions, information maximization has extra properties not found in the linear case (Linsker 1989). The nonlinearities in the transfer function are able to pick up higher-order moments of the input distributions and perform something akin to true redundancy reduction between units in the output representation. This enables the network to separate statistically independent components in the inputs: a higher-order generalization of principal components analysis. We apply the network to the source separation (or cocktail party) problem, successfully separating unknown mixtures of up to 10 speakers. We also show that a variant on the network architecture is able to perform blind deconvolution (cancellation of unknown echoes and reverberation in a speech signal). Finally, we derive dependencies of information transfer on time delays. We suggest that information maximization provides a unifying framework for problems in "blind" signal processing.

A fluid mosaic model is presented for the gross organization and structure of the proteins and lipids of biological membranes. The model is consistent with the restrictions imposed by thermodynamics. In this model, the proteins that are integral to the membrane are a heterogeneous set of globular molecules, each arranged in an amphipathic structure, that is, with the ionic and highly polar groups protruding from the membrane into the aqueous phase, and the nonpolar groups largely buried in the hydrophobic interior of the membrane. These globular molecules are partially embedded in a matrix of phospholipid. The bulk of the phospholipid is organized as a discontinuous, fluid bilayer, although a small fraction of the lipid may interact specifically with the membrane proteins. The fluid mosaic structure is therefore formally analogous to a two-dimensional oriented solution of integral proteins (or lipoproteins) in the viscous phospholipid bilayer solvent. Recent experiments with a wide variety of techniqes and several different membrane systems are described, all of which abet consistent with, and add much detail to, the fluid mosaic model. It therefore seems appropriate to suggest possible mechanisms for various membrane functions and membrane-mediated phenomena in the light of the model. As examples, experimentally testable mechanisms are suggested for cell surface changes in malignant transformation, and for cooperative effects exhibited in the interactions of membranes with some specific ligands. Note added in proof: Since this article was written, we have obtained electron microscopic evidence (69) that the concanavalin A binding sites on the membranes of SV40 virus-transformed mouse fibroblasts (3T3 cells) are more clustered than the sites on the membranes of normal cells, as predicted by the hypothesis represented in Fig. 7B. T-here has also appeared a study by Taylor et al. (70) showing the remarkable effects produced on lymphocytes by the addition of antibodies directed to their surface immunoglobulin molecules. The antibodies induce a redistribution and pinocytosis of these surface immunoglobulins, so that within about 30 minutes at 37 degrees C the surface immunoglobulins are completely swept out of the membrane. These effects do not occur, however, if the bivalent antibodies are replaced by their univalent Fab fragments or if the antibody experiments are carried out at 0 degrees C instead of 37 degrees C. These and related results strongly indicate that the bivalent antibodies produce an aggregation of the surface immunoglobulin molecules in the plane of the membrane, which can occur only if the immunoglobulin molecules are free to diffuse in the membrane. This aggregation then appears to trigger off the pinocytosis of the membrane components by some unknown mechanism. Such membrane transformations may be of crucial importance in the induction of an antibody response to an antigen, as well as iv other processes of cell differentiation.

Analyses of steroid receptors are important for understanding molecular details of transcriptional control, as well as providing insight as to how an individual transacting factor contributes to cell identity and function. These studies have led to the identification of a superfamily of regulatory proteins that include receptors for thyroid hormone and the vertebrate morphogen retinoic acid. Although animals employ complex and often distinct ways to control their physiology and development, the discovery of receptor-related molecules in a wide range of species suggests that mechanisms underlying morphogenesis and homeostasis may be more ubiquitous than previously expected.

The reference human genome sequence set the stage for studies of genetic variation and its association with human disease, but epigenomic studies lack a similar reference. To address this need, the NIH Roadmap Epigenomics Consortium generated the largest collection so far of human epigenomes for primary cells and tissues. Here we describe the integrative analysis of 111 reference human epigenomes generated as part of the programme, profiled for histone modification patterns, DNA accessibility, DNA methylation and RNA expression. We establish global maps of regulatory elements, define regulatory modules of coordinated activity, and their likely activators and repressors. We show that disease- and trait-associated genetic variants are enriched in tissue-specific epigenomic marks, revealing biologically relevant cell types for diverse human traits, and providing a resource for interpreting the molecular basis of human disease. Our results demonstrate the central role of epigenomic information for understanding gene regulation, cellular differentiation and human disease.

The fly Drosophila melanogaster is one of the most intensively studied organisms in biology and serves as a model system for the investigation of many developmental and cellular processes common to higher eukaryotes, including humans. We have determined the nucleotide sequence of nearly all of the approximately 120-megabase euchromatic portion of the Drosophila genome using a whole-genome shotgun sequencing strategy supported by extensive clone-based sequence and a high-quality bacterial artificial chromosome physical map. Efforts are under way to close the remaining gaps; however, the sequence is of sufficient accuracy and contiguity to be declared substantially complete and to support an initial analysis of genome structure and preliminary gene annotation and interpretation. The genome encodes approximately 13,600 genes, somewhat fewer than the smaller Caenorhabditis elegans genome, but with comparable functional diversity.

are integrated upstream of GUN1 within plastids, which leads to ABI4-mediated repression of nuclear-encoded genes.

In recent years, members of the protein kinase family have been discovered at an accelerated pace. Most were first described, not through the traditional biochemical approach of protein purification and enzyme assay, but as putative protein kinase amino acid sequences deduced from the nucleotide sequences of molecularly cloned genes or complementary DNAs. Phylogenetic mapping of the conserved protein kinase catalytic domains can serve as a useful first step in the functional characterization of these newly identified family members.

Over 225,000 independent Agrobacterium transferred DNA (T-DNA) insertion events in the genome of the reference plant Arabidopsis thaliana have been created that represent near saturation of the gene space. The precise locations were determined for more than 88,000 T-DNA insertions, which resulted in the identification of mutations in more than 21,700 of the approximately 29,454 predicted Arabidopsis genes. Genome-wide analysis of the distribution of integration events revealed the existence of a large integration site bias at both the chromosome and gene levels. Insertion mutations were identified in genes that are regulated in response to the plant hormone ethylene.

Neural stem cells exist not only in the developing mammalian nervous system but also in the adult nervous system of all mammalian organisms, including humans. Neural stem cells can also be derived from more primitive embryonic stem cells. The location of the adult stem cells and the brain regions to which their progeny migrate in order to differentiate remain unresolved, although the number of viable locations is limited in the adult. The mechanisms that regulate endogenous stem cells are poorly understood. Potential uses of stem cells in repair include transplantation to repair missing cells and the activation of endogenous cells to provide "self-repair. " Before the full potential of neural stem cells can be realized, we need to learn what controls their proliferation, as well as the various pathways of differentiation available to their daughter cells.

The potential of the diverse chemistries present in natural products (NP) for biotechnology and medicine remains untapped because NP databases are not searchable with raw data and the NP community has no way to share data other than in published papers. Although mass spectrometry (MS) techniques are well-suited to high-throughput characterization of NP, there is a pressing need for an infrastructure to enable sharing and curation of data. We present Global Natural Products Social Molecular Networking (GNPS; http://gnps.ucsd.edu), an open-access knowledge base for community-wide organization and sharing of raw, processed or identified tandem mass (MS/MS) spectrometry data. In GNPS, crowdsourced curation of freely available community-wide reference MS libraries will underpin improved annotations. Data-driven social-networking should facilitate identification of spectra and foster collaborations. We also introduce the concept of 'living data' through continuous reanalysis of deposited data.

The application of molecular cloning technology to the study of the glutamate receptor system has led to an explosion of knowledge about the structure, expression, and function of this most important fast excitatory transmitter system in the mammalian brain. The first functional ionotropic glutamate receptor was cloned in 1989 (Hollmann et al 1989) , and the results of this molecular-based approach over the past three years are the focus of this review. We discuss the implications of and the new questions raised by this work-which is probably only a glance at this fascinating and complex signaling system found in brains from the snails to man. Glutamate receptors are found throughout the mammalian brain, where they constitute the major excitatory transmitter system. The longest-known and best-studied glutamate receptors are ligand-gated ion channels, also called ionotropic glutamate receptors , which are permeable to cations. They have traditionally been classified into three broad subtypes based upon pharmaco­ logical and electrophysiological data: a-amino-3-hydroxy-5-methyl-4isoxazole propionate (AMPA) receptors, kainate (KA) receptors , and N-methyl-D-aspartate (NMDA) receptors. Recently, however, a family of G protein-coupled glutamate receptors , which are also called metabotropic glutamate or transl -aminocyclopentanel ,3-dicarboxylate (tACPD) recep­ tors, was identified (Sugiyama et al 1987) . (For reviews of the classification and the pharmacological and electrophysiological properties of glutamate receptors see Mayer & Westbrook 1987, Collingridge & Lester 1989, Honore 1989, Monaghan et al 1989, Wroblewski & Danysz 1 989, Hansen &

A peptide with high potency and intrinsic activity for stimulating the secretion of corticotropin-like and β-endorphin-like immunoactivities by cultured anterior pituitary cells has been purified from ovine hypothalamic extracts. The primary structure of this 41-residue corticotropin- and β-endorphin-releasing factor has been determined to be: H-Ser-Gln-Glu-Pro-Pro-Ile-Ser-Leu-Asp-Leu-Thr-Phe-His-Leu-Leu-Arg-Glu-Val-Leu-Glu-Met-Thr-Lys-Ala-Asp-Gln-Leu-Ala-Gln-Gln-Ala-His-Ser-Asn-Arg-Lys-Leu-Leu-Asp-Ile-Ala-NH 2 . The synthetic peptide is active in vitro and in vivo.

Cancer-associated fibroblasts (CAFs) are a key component of the tumour microenvironment with diverse functions, including matrix deposition and remodelling, extensive reciprocal signalling interactions with cancer cells and crosstalk with infiltrating leukocytes. As such, they are a potential target for optimizing therapeutic strategies against cancer. However, many challenges are present in ongoing attempts to modulate CAFs for therapeutic benefit. These include limitations in our understanding of the origin of CAFs and heterogeneity in CAF function, with it being desirable to retain some antitumorigenic functions. On the basis of a meeting of experts in the field of CAF biology, we summarize in this Consensus Statement our current knowledge and present a framework for advancing our understanding of this critical cell type within the tumour microenvironment.

A peptide has been isolated from ovine hypothalamus which, at 1 x 10(-9)M, inhibits secretion in vitro of immunoreactive rat or human growth hormones and is similarly active in vivo in rats. Its structure is H-Ala-Gly-Cys-Lys-Asn-Phe-Phe-Trp-Lys-Thr-Phe-Thr-Ser-Cys-OH The synthetic replicate is biologically active.

Rapid generation of superoxide and accumulation of H2O2 is a characteristic early feature of the hypersensitive response following perception of pathogen avirulence signals. Emerging data indicate that the oxidative burst reflects activation of a membrane-bound NADPH oxidase closely resembling that operating in activated neutrophils. The oxidants are not only direct protective agents, but H2O2 also functions as a substrate for oxidative cross-linking in the cell wall, as a threshold trigger for hypersensitive cell death, and as a diffusible signal for induction of cellular protectant genes in surrounding cells. Activation of the oxidative burst is a central component of a highly amplified and integrated signal system, also involving salicylic acid and perturbations of cytosolic Ca2+, which underlies the expression of disease-resistance mechanisms.

Eye movements, eye blinks, cardiac signals, muscle noise, and line noise present serious problems for electroencephalographic (EEG) interpretation and analysis when rejecting contaminated EEG segments results in an unacceptable data loss. Many methods have been proposed to remove artifacts from EEG recordings, especially those arising from eye movements and blinks. Often regression in the time or frequency domain is performed on parallel EEG and electrooculographic (EOG) recordings to derive parameters characterizing the appearance and spread of EOG artifacts in the EEG channels. Because EEG and ocular activity mix bidirectionally, regressing out eye artifacts inevitably involves subtracting relevant EEG signals from each record as well. Regression methods become even more problematic when a good regressing channel is not available for each artifact source, as in the case of muscle artifacts. Use of principal component analysis (PCA) has been proposed to remove eye artifacts from multichannel EEG. However, PCA cannot completely separate eye artifacts from brain signals, especially when they have comparable amplitudes. Here, we propose a new and generally applicable method for removing a wide variety of artifacts from EEG records based on blind source separation by independent component analysis (ICA). Our results on EEG data collected from normal and autistic subjects show that ICA can effectively detect, separate, and remove contamination from a wide variety of artifactual sources in EEG records with results comparing favorably with those obtained using regression and PCA methods. ICA can also be used to analyze blink-related brain activity.

The hippocampus is one of the few areas of the rodent brain that continues to produce neurons postnatally. Neurogenesis reportedly persists in rats up to 11 months of age. Using bromodeoxyuridine (BrdU) labeling, the present study confirms that in the adult rat brain, neuronal progenitor cells divide at the border between the hilus and the granule cell layer (GCL). In adult rats, the progeny of these cells migrate into the GCL and express the neuronal markers NeuN and calbindin-D28k. However, neurogenesis was drastically reduced in aged rats. Six-to 27-month-old Fischer rats were injected intraperitoneally with BrdU to detect newborn cells in vivo and to follow their fate in the dentate gyrus. When killed 4-6 weeks after BrdU labeling, 12- to 27-month-old rats exhibited a significant decline in the density of BrdU-positive cells in the granule cell layer compared with 6-month-old controls. Decreased neurogenesis in aging rats was accompanied by reduced immunoreactivity for poly-sialylated neural cell adhesion molecule, a molecule that is involved in migration and process elongation of developing neurons. When animals were killed immediately (12 hr) after BrdU injection, significantly fewer labeled cells were observed in the GCL and adjacent subgranular zone of aged rats, indicative of a decrease in mitotic activity of neuronal precursor cells. The reduced proliferation was not attributable to a general aged-related metabolic impairment, because the density of BrdU-positive cells was not altered in other brain regions with known mitotic activity (e.g., hilus and lateral ventricle wall). The decline in neurogenesis that occurs throughout the lifespan of an animal can thus be related to a decreasing proliferation of granule cell precursors.

A workshop was convened by the AACR to discuss the rapidly emerging cancer stem cell model for tumor development and progression. The meeting participants were charged with evaluating data suggesting that cancers develop from a small subset of cells with self-renewal properties analogous to organ

Running increases neurogenesis in the dentate gyrus of the hippocampus, a brain structure that is important for memory function. Consequently, spatial learning and long-term potentiation (LTP) were tested in groups of mice housed either with a running wheel (runners) or under standard conditions (controls). Mice were injected with bromodeoxyuridine to label dividing cells and trained in the Morris water maze. LTP was studied in the dentate gyrus and area CA1 in hippocampal slices from these mice. Running improved water maze performance, increased bromodeoxyuridine-positive cell numbers, and selectively enhanced dentate gyrus LTP. Our results indicate that physical activity can regulate hippocampal neurogenesis, synaptic plasticity, and learning.

We present a system for recognizing human faces from single images out of a large database containing one image per person. Faces are represented by labeled graphs, based on a Gabor wavelet transform. Image graphs of new faces are extracted by an elastic graph matching process and can be compared by a simple similarity function. The system differs from the preceding one (Lades et al., 1993) in three respects. Phase information is used for accurate node positioning. Object-adapted graphs are used to handle large rotations in depth. Image graph extraction is based on a novel data structure, the bunch graph, which is constructed from a small get of sample image graphs.