Although brain-derived neurotrophic factor (BDNF) plays a central role in recovery after cerebral ischemia, little is known about cells involved in BDNF production after stroke.The present study testes the hypothesis that neurons are not the unique source of neosynthesized BDNF after stroke and that non neuronal-BDNF producing cells differ according to the delay after stroke induction. For this purpose, cellular localization of BDNF and BDNF content of each hemisphere were analysed in parallel before and after (4h, 24h and 8d) ischemic stroke in rats. Stroke of different severities was induced by embolization of the brain with variable number of calibrated microspheres allowing us to explore the association between BDNF production and neuronal death severity. The main results are that a) unilateral stroke increased BDNF production in both hemispheres with a more intense and long-lasting effect in the lesioned hemisphere, b) BDNF levels either of the lesioned or unlesioned hemispheres were not inversely correlated to neuronal death severity whatever the delay after stroke onset, c) in the unlesioned hemisphere, stroke resulted in increased BDNF staining in neurons and ependymal cells (at 4h and 24h), d) in the lesioned hemisphere, beside neurons and ependymal cells, microglial cells (at 24h), endothelial cells of cerebral arterioles (at 4h and 24h) and astrocytes (at 8d) exhibited a robust BDNF staining as well. Taken together, overall data suggest that non neuronal cells are able to produce substantial amount of BDNF after ischemic stroke and that more attention should be given to these cells in the design of strategies aimed at improving stroke recovery through BDNFrelated mechanisms.

Escherichia coli ribosomal protein S8 was previously shown to bind a 16s rRNA fragment (nucleotides 584-756) with the same affinity as the complete 16s rRNA, and to shield an irregular helical region (region C) [Mougel, M., Eyermann, F., Westhof, E., Romby, P., Expert-Bezanqon, Ebel, J. P., Ehresmann, B. & Ehresmann, C. (1987). J. Mol. Biol. 198, 91-1071. Region C was postulated to display characteristic features : three bulged adenines (A595, A640 and A642), a noncanonical U598-U641 pair surrounded by two G . C pairs. In order to delineate the minimal RNA binding site, deletions were introduced by site-directed mutagenesis and short RNA fragments were synthesized. Their ability to bind S8 was assayed by filter binding. Our results show that the RNA binding site can be restricted to a short helical stem (588-609633-6.51) containing region C . The second part of the work focused on region C and on the role of conserved nucleotides as potential determinants of S8 recognition. Single and double mutations were introduced by site-directed mutagenesis in fragment 584-756, and their effect on S8 binding was measured. It was found that the three bulged positions are essential and that adenines are required at positions 640 and 642. U598 is also crucial and the highly conserved G597 . C643 pair cannot be inverted. These conserved nucleotides are either directly involved in the recognition process as direct contacts or required to maintain a specific conformation. The strong evolutionary pressure and the small number of positive mutants stress the high stringency of the recognition process.RNA-binding proteins recognize defined parts of RNA molecules in a highly specific manner, but in most cases the basic molecular mechanism of this recognition process remains not well known. Ribosomal components offer a valuable system for the possible dissection of the underlying principles of specific protein-RNA interactions. The Escherichia coli 30s subunit is composed of 21 proteins and some of them, named primary proteins, interact independently and specifically with selected regions of ribosomal 16s rRNA. Among them, protein S8 plays a central role both in 30s ribosomal subunit assembly (Held et al., 1974) and in the regulation of the Spc ribosomal protein operon (Yates et al., 1980: Dean et al., 1981. During 30s assembly, S8 binds to the central domain of 16s rRNA in the early stage of the process (Held et al., 1974) and may interact cooperatively with two other small subunit proteins: S6 and S38 Svensson et al., 1988). It was suggested that the region near S8, in the body of 30s particle, represents one of the two nucleation sites during assembly (Moore, 1987). A variety of methods, including partial nuclease digestion (Zimmermann et al., 1975;Ungewickell et al., 1975; Zim- (Mougel et al., 1987;Svensson et al., 1988) and sitedirected mutagenesis Gregory et al., 1988) was used to localize the region within 16s rRNA that mediates S8 recognition. In a previous work, we characterized the interactions between protein S8 and 16s rRN...

The membrane surface of polarized renal epithelial cells (MDCK cells) grown as a monolayer was imaged with the atomic force microscope. The surface topography of dried cells determined by this approach was consistent with electron microscopy images previously reported. Fixed and living cells in aqueous medium gave more fuzzy images, likely because of the presence of the cell glycocalix. Treatment of living cells with neuraminidase, an enzyme that partly degrades the glycocalix, allowed sub-micrometer imaging. Protruding particles, 10 to 60 nm xy size, occupy most of the membrane surface. Protease treatment markedly reduced the size of these particles, indicating that they corresponded to proteins. Tip structure effects were probably involved in the exaggerated size of imaged membrane proteins. Although further improvements in the imaging conditions, including tip sharpness, are required, atomic force microscope already offers the unique possibility to image proteins at the membrane surface of living cells.

Previous experiments showed that S15 inhibits its own translation by binding to its mRNA in a region overlapping the ribosome loading site. This binding was postulated to stabilize a pseudoknot structure that exists in equilibrium with two stem-loops and to trap the ribosome on its mRNA loading site in a transitory state. In this study, we investigated the effect of mutations in the translational operator on: the binding of protein S15, the formation of the 30S/mRNA/tRNA(fMet) ternary initiation complex, the ability of S15 to inhibit the formation of this ternary complex. The results were compared to in vivo expression and repression rates. The results show that (1) the pseudoknot is required for S15 recognition and translational control; (2) mRNA and 16S rRNA efficiently compete for S15 binding and 16S rRNA suppresses the ability of S15 to inhibit the formation of the active ternary complex; (3) the ribosome binds more efficiently to the pseudoknot than to the stem-loop; (4) sequences located between nucleotides 12 to 47 of the S15 coding phase enhances the efficiency of ribosome binding in vitro; this is correlated with enhanced in vivo expression and regulation rates.

No abstract