Créancier et al. (page 275) examined translational regulation mediated by the internal ribosome entry site (IRES) of FGF-2, and found that in contrast to a picornaviral IRES, the FGF-2 IRES is subject to stringent tissue-specific regulation. The finding has broad implications for basic studies on gene regulation in development, and could also have applications in gene therapy.

Using a bicistronic luciferase reporter vector, the authors analyzed translation from the IRES of FGF-2 or encephalomyocarditis virus (EMCV) in tissue culture and transgenic mice. The FGF-2 IRES is active in a variety of human and non-human cell types, suggesting that its activity is conserved among mammals. In transgenic mice, FGF-2 IRES activity is spatiotemporally regulated during development and is later restricted to the adult brain, in contrast to the nearly ubiquitous activity of the EMCV IRES. In addition to demonstrating for the first time that a mammalian IRES is subject to tissue-specific regulation in vivo, the new data are strikingly parallel to the proposed functions of FGF-2 in development and in the adult brain, suggesting a central role for translational regulation during embryogenesis. The stringent tissue specificity of the FGF-2 IRES in the adult suggests that it could also be a useful tool for expressing neurotrophic factors or other therapeutic gene products in the central nervous system.

The global repression of nuclear RNA synthesis that accompanies mitosis does not seem to require nucleosomal chromatin condensation, according to the results of Spencer et al. (page 13). An important but poorly understood process, mitotic transcriptional repression has been hypothesized to occur either by mechanisms involving chromatin condensation or by transcription factor inactivation. In vitro data have been consistent with the latter model, but the process has been difficult to analyze in vivo.

To address this issue, Spencer et al. analyzed transcription and RNA polymerase II localization in mitotic cells infected with herpes simplex virus type 1. The authors found that viral DNA remains noncondensed and nucleosome-free during mitosis, but mitotic transcription is repressed on both condensed chromosomes and noncondensed viral DNA. In agreement with earlier in vitro data, the results are consistent with a model in which mitosis-specific modifications of transcriptional proteins are sufficient for mitotic transcription repression. The data do not exclude the possibility that interactions between mitotic chromatin and condensation proteins could be involved in transcription repression, but the availability of a new in vivo model should make more detailed mechanistic studies possible.

Although ceramide formation appears to be a universal phenomenon in the late stages of apoptosis, the significance of this observation has remained poorly understood. Tepper et al. (page 155) now show that ceramide is derived from sphingomyelin originally located in the outer leaflet of the plasma membrane, and suggest that the breakdown of sphingomyelin, rather than the production of ceramide, is important for the execution phase of apoptosis.

Using a fluorescent sphingomyelin analogue, Tepper et al. found that sphingomyelin on the outer leaflet of the membrane flips during apoptosis, bringing it into contact with a cytosolic sphingomyelinase, which produces ceramide by breaking down sphingomyelin. Cells deficient in lipid scrambling exhibit normal nuclear and mitochondrial apoptotic features, but show aberrant membrane morphology during apoptosis and do not produce ceramide.

Based on their results, the authors propose a model in which the loss of plasma membrane phospholipid asymmetry causes the breakdown of sphingomyelin and production of ceramide. Sphingomyelin breakdown then causes an efflux of cholesterol, leading to biophysical changes that permit membrane blebbing. The new data, and the authors' model, are also discussed in an accompanying Comment article by Green (page F5).

Using two-dimensional gel chromatography and scanning transmission electron microscopy (STEM), Hölzl et al. (page 119) have determined the subunit composition of the Drosophila 26S proteasome regulatory complex (RC) and localized a deubiquitylating enzyme within the complex. The new findings are an important step towards developing a detailed structural model of the RC, a key component in the ubiquitin-mediated protein degradation pathway.

Whereas the proteolytic core complex of the 26S proteasome has been studied extensively, the instability of the purified RC has hindered structural studies on the regulatory machinery. After purifying 26S proteasomes from Drosophila embryos, which are a rich source of the complexes, Hölzl et al. identified 18 RC subunits by 2-D gel electrophoresis and amino acid analysis. Using STEM, the team determined that the molecular mass of a single RC agrees closely with the sum of the molecular masses of the 18 subunits, suggesting that all of the subunits have now been identified. 17 of the subunits have homologues among known yeast and mammalian RC subunits, and one subunit, p37A, is a member of the ubiquitin COOH-terminal hydrolase family, which had not previously been found in RCs. Structural studies using a nonhydrolyzable substrate analogue identified the location of p37A within the RC. Future structural studies can now focus on mapping the locations of the remaining subunits within the complex.

Beginning on page 165, Ackerley et al. demonstrate that in cultured neurons, glutamate slows the axonal transport of neurofilaments and activates members of the mitogen activated protein kinase family, which can phosphorylate neurofilament side-arm domains. The results are the first to link a slowing of neurofilament transport, which is a pathological feature of several neurodegenerative diseases, with glutamate excitotoxicity, which has been hypothesized to have a role in the pathogenesis of these diseases. The authors also describe a new technique for monitoring neurofilament transport, involving transfecting a gene for green fluorescent protein-tagged neurofilament middle chain into cultured neurons.

When the transfected cells are treated with glutamate, neurofilament transport is significantly slowed, and phosphorylation of the neurofilament side-arm domains increases, concurrent with the activation of members of the mitogen-activated protein kinase family. The new data suggest a molecular mechanism in which glutamate excitotoxicity in diseases such as ALS and Alzheimer's disease activates kinases that phosphorylate neurofilament side-arm domains, slowing neurofilament transport and leading to the characteristic neurofilament accumulation seen in these conditions. Although the molecular details of this process remain to be determined, the technique developed by the authors for this work should facilitate future studies in this area.

By Alan W. Dove, 350 E. Willow Grove #406, Philadelphia, PA 19118. E-mail: [email protected]