Analyzing mRNA-protein complexes using a yeast three-hybrid system

RNA-protein interactions are essential for the proper execution and regulation of every step in the life of a eukaryotic mRNA. Here we describe a three-hybrid system in which RNA-protein interactions can be analyzed using simple phenotypic or enzymatic assays in Saccharomyces cerevisiae. The system can be used to detect or confirm an RNA-protein interaction, to analyze RNA-protein interactions genetically, and to discover new protein or RNA partners when only one is known. Multicomponent complexes containing more than one protein can be detected, identified, and analyzed. We describe the method and how to use it, and discuss applications that bear particularly on eukaryotic mRNAs.     

Upregulation of BiP and CHOP by the unfolded-protein response is independent of presenilin expression

Presenilin 1 (PS1), a polytopic membrane protein, has a critical role in the trafficking and proteolysis of a selected set of transmembrane proteins. The vast majority of individuals affected with early onset familial Alzheimer's disease (FAD) carry missense mutations in PS1. Two studies have suggested that loss of PS1 function, or expression of FAD-linked PS1 variants, compromises the mammalian unfolded-protein response (UPR), and we sought to evaluate the potential role of PS1 in the mammalian UPR. Here we show that that neither the endoplasmic reticulum (ER) stress-induced accumulation of BiP and CHOP messenger RNA, nor the activation of ER stress kinases IRE1alpha and PERK, is compromised in cells lacking both PS1 and PS2 or in cells expressing FAD-linked PS1 variants. We also show that the levels of BiP are not significantly different in the brains of individuals with sporadic Alzheimer's disease or PS1-mediated FAD to levels in control brains. Our findings provide evidence that neither loss of PS1 and PS2 function, nor expression of PS1 variants, has a discernable impact on ER stress-mediated induction of the several established 'readouts' of the UPR pathway.     

RGS4 binds to membranes through an amphipathic alpha -helix

RGS4, a mammalian GTPase-activating protein for G protein alpha subunits, requires its N-terminal 33 amino acids for plasma membrane localization and biological activity (Srinivasa, S. P., Bernstein, L. S., Blumer, K. J., and Linder, M. E. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 5584-5589). In this study, we tested the hypothesis that the N-terminal domain mediates membrane binding by forming an amphipathic alpha-helix. RGS4 bound to liposomes containing anionic phospholipids in a manner dependent on the first 33 amino acids. Circular dichroism spectroscopy of a peptide corresponding to amino acids 1-31 of RGS4 revealed that the peptide adopted an alpha-helical conformation in the presence of anionic phospholipids. Point mutations that either neutralized positive charges on the hydrophilic face or substituted polar residues on the hydrophobic face of the model helix disrupted plasma membrane targeting and biological activity of RGS4 expressed in yeast. Recombinant mutant proteins were active as GTPase-activating proteins in solution but exhibited diminished binding to anionic liposomes. Peptides corresponding to mutants with the most pronounced phenotypes were also defective in forming an alpha-helix as measured by circular dichroism spectroscopy. These results support a model for direct interaction of RGS4 with membranes through hydrophobic and electrostatic interactions of an N-terminal alpha-helix.     

Intracellular pH modulation of ADF/cofilin proteins

The ADF/cofilin (AC) proteins are necessary for the high rates of actin filament turnover seen in vivo. Their regulation is complex enough to underlie the precision in filament dynamics needed by stimulated cells. Disassembly of actin by AC proteins is inhibited in vitro by phosphorylation of ser3 and pH<7.1. This study of Swiss 3T3 cells demonstrates that pH also affects AC behavior in vivo: (1) Wounded cells show pH-dependent AC translocation to alkaline-induced ruffling membrane; (2) The Triton extractable (soluble) ADF from Swiss 3T3 cells decreases from 42+/-4% to 23+/-4% when the intracellular pH (pH(i)) is reduced from 7.4 to 6.6; (3) Covariance and colocalization analyses of immunostained endogenous proteins show that ADF partitions more with monomeric actin and less with polymeric actin when pH(i) increases. However, the distribution of cofilin, a less pH-sensitive AC in vitro, does not change with pH; (4) Only the unphosphorylatable AC mutant (A3), when overexpressed as a GFP chimera, uniquely produces aberrant cellular phenotypes and only if the pH is shifted from 7.1 to 6.6 or 7.4. A mechanism is proposed that explains why AC(A3)-GFP and AC(wt)-GFP chimeras generate different phenotypes in response to pH changes. Phospho-AC levels increase with cell density, and in motile cells, phospho-AC increases with alkalization, suggesting a homeostatic mechanism that compensates for increased AC activity and filament turnover. These results show that the behavior of AC proteins with pH-sensitivity in vitro is affected by pH in vivo.     

Human U1 small nuclear RNA genes: extensive conservation of flanking sequences suggests cycles of gene amplification and transposition.

The DNA immediately flanking the 164-base-pair U1 RNA coding region is highly conserved among the approximately 30 human U1 genes. The U1 multigene family also contains many U1 pseudogenes (designated class I) with striking although imperfect flanking homology to the true U1 genes. Using cosmid vectors, we now have cloned, characterized, and partially sequenced three 35-kilobase (kb) regions of the human genome spanning U1 homologies. Two clones contain one true U1 gene each, and the third bears two class I pseudogenes 9 kb apart in the opposite orientation. We show by genomic blotting and by direct DNA sequence determination that the conserved sequences surrounding U1 genes are much more extensive than previously estimated: nearly perfect sequence homology between many true U1 genes extends for at least 24 kb upstream and at least 20 kb downstream from the U1 coding region. In addition, the sequences of the two new pseudogenes provide evidence that class I U1 pseudogenes are more closely related to each other than to true genes. Finally, it is demonstrated elsewhere (Lindgren et al., Mol. Cell. Biol. 5:2190-2196, 1985) that both true U1 genes and class I U1 pseudogenes map to chromosome 1, but in separate clusters located far apart on opposite sides of the centromere. Taken together, these results suggest a model for the evolution of the U1 multigene family. We speculate that the contemporary family of true U1 genes was derived from a more ancient family of U1 genes (now class I U1 pseudogenes) by gene amplification and transposition. Gene amplification provides the simplest explanation for the clustering of both U1 genes and class I pseudogenes and for the conservation of at least 44 kb of DNA flanking the U1 coding region in a large fraction of the 30 true U1 genes.     

Immobilized heparinase: In vitro reactor model

Heparinase immobilized to agarose has previously been shown to be useful in degrading heparin and thereby preventing thromboembolytic complications when this anticoagulant has been used in extracorporeal perfusions. The current study examined the kinetics of this immobilized enzyme. When heparinase is covalently bound to 8% agarose, the partition coefficient of heparin in the catalytic particle is 0.36 +/- 0.048 (N = 10). The immobilized enzyme has a K(m) of 0.15 +/- 0.03 mg/mL and an activation energy of 10.3 +/- 0.57 kcal/gmol (N = 5). These values are statistically indistinguishable from the values for the free enzyme. The immobilized enzyme showed a pH activity optimum between 7.0 and 7.4, compared to the optimum pH of 6.5 for the soluble enzyme. The activity optimum of immobilized heparinase with respect to salt concentration was between 0 and 0.1M. A reactor containing immobilized heparinase recirculating internally at 1300 mL/min behaved as a continuously stirred tank reactor (CSTR) when solutions at a flow rate of 120 mL/min were passed through the device. The residence time distribution was determined using blue dextran (molecular weight 2 x 10(6) daltons), which is sterically excluded from the agarose catalyst. A model of the heparinase reactor based on ideal CSTR behavior and the immobilized enzyme kinetic parameters was developed. It accurately predicted experimental conversions over a range of catalyst volumes, enzyme loadings, and substrate concentrations to within 7% in most cases and with a maximum deviation of 13%.