Bifidobacteria are Gram-positive prokaryotes that naturally colonize the human gastrointestinal tract (GIT) and vagina. Although not numerically dominant in the complex intestinal microflora, they are considered as key commensals that promote a healthy GIT. We determined the 2.26-Mb genome sequence of an infant-derived strain of Bifidobacterium longum, and identified 1,730 possible coding sequences organized in a 60%-GC circular chromosome. Bioinformatic analysis revealed several physiological traits that could partially explain the successful adaptation of this bacteria to the colon. An unexpectedly large number of the predicted proteins appeared to be specialized for catabolism of a variety of oligosaccharides, some possibly released by rare or novel glycosyl hydrolases acting on ''nondigestible'' plant polymers or host-derived glycoproteins and glycoconjugates. This ability to scavenge from a large variety of nutrients likely contributes to the competitiveness and persistence of bifidobacteria in the colon. Many genes for oligosaccharide metabolism were found in self-regulated modules that appear to have arisen in part from gene duplication or horizontal acquisition. Complete pathways for all amino acids, nucleotides, and some key vitamins were identified; however, routes for Asp and Cys were atypical. More importantly, genome analysis provided insights into the reciprocal interactions of bifidobacteria with their hosts. We identified polypeptides that showed homology to most major proteins needed for production of glycoprotein-binding fimbriae, structures that could possibly be important for adhesion and persistence in the GIT. We also found a eukaryotic-type serine protease inhibitor (serpin) possibly involved in the reported immunomodulatory activity of bifidobacteria.

The availability of structural classification data has enabled this novel overview analysis. We conclude that function at the top level of the EC number enzyme classification is not related to fold, as only a very few specific residues are actually responsible for enzyme activity. Conversely, the fold is much more closely related to ligand type.

Using structure based genome mining targeting vascular endothelial and platelet derived growth factor immunoglobulin (Ig) like folds, we have identified a sequence corresponding to a single transmembrane protein with two Ig domains, which we cloned from a human brain cDNA library. The cDNA is identical to hepatocyte cell adhesion molecule (hepaCAM), which was originally described as a tumor suppressor gene in liver. Here, we show that the protein is predominantly expressed in the mouse and human nervous system. In liver, the expression is very low in humans, and is not detected in mice. To identify the central nervous system (CNS) regions and cell types expressing the protein, we performed a LacZ reporter gene assay on heterozygous mice in which one copy of the gene encoding the novel protein had been replaced with beta-galactosidase. beta-galactosidase expression was prominent in white matter tracts of the CNS. Furthermore, expression was detected in ependymal cells of the brain ventricular zones and the central canal of the spinal cord. Double labeling experiments showed expression mainly in CNPase positive oligodendrocytes (OL). Since the protein is predominantly expressed in the CNS glial cells, we named the molecule glial cell adhesion molecule (GlialCAM). A potential role for GlialCAM in myelination was supported by its up-regulation during postnatal mouse brain development, where it was concomitantly expressed with myelin basic protein (MBP). In addition, in vitro, GlialCAM was observed in various developmental stages of OL and in astrocytes in processes and at cell contact sites. In A2B5 positive OL, GlialCAM colocalizes with GAP43 in OL growth cone like structures. Overall, the data presented here indicate a potential function for GlialCAM in glial cell biology.

The determination of protein structures has furthered our understanding of how various proteins perform their functions. With the large number of structures currently available in the PDB, it is necessary to be able to easily study these proteins in detail. Here new software tools are presented which aim to facilitate this analysis; these include the PDBsum WWW site which provides a summary description of all PDB entries, the programs TOPS and NUCPLOT to plot schematic diagrams representing protein topology and DNA-binding interactions, SAS a WWW-based sequence-analysis tool incorporating structural data, and WWW servers for the analysis of protein±protein interfaces and analyses of over 300 haem-binding proteins.