The Yeast RNA-binding Protein Rbp1p Modifies the Stability of Mitochondrial Porin mRNA*

The Saccharomyces cerevisiae RNA-binding protein Rbp1p was initially identified as a negative growth regulator; however, its function is still obscure. Here, we show that Rbp1p in cells is associated with structures that sediment at 10,000 as well as 100,000 × g. It appears microscopically as punctate signals partially localized to the perinuclear region. Over-expression of Rbp1p in yeast resulted in growth defects on nonfermentable carbon sources, suggesting a function for Rbp1p in mitochondrial biogenesis. Absence of Rbp1p increased the level of mitochondrial porin, whereas over-expression of Rbp1p, but not an N-terminally truncated form, decreased porin levels. Over-expression of Rbp1p also decreased the level of mitochondrial porin mRNA by enhancing its degradation, an effect that was dependent on all three of the Rbp1p RNA recognition motifs. In cells, the porin mRNA is associated with Rbp1p·RNP (ribonucleoprotein) complexes. In vitro binding assays showed that Rbp1p most likely interacts with a (C/G)U-rich element in the porin mRNA 3′-UTR. Based on these observations, we infer that Rbp1p has a role in negatively regulating mitochondrial porin expression post-transcriptionally.

Eukaryotic messenger RNAs (mRNAs) are produced in the nucleus and undergo sequential processing reactions that typically include capping, pre-mRNA splicing, and polyadenylation. Cellular mRNA is associated with protein from the time of its synthesis to its degradation. With growing recognition of the complexity of gene regulation, interest in the mechanism of mRNA degradation has increased rapidly during the past decade. The mRNA turnover is a means to permit timely adjustments to changes in growth conditions or to genetically controlled programs of expression (reviewed in Refs. 1 and 2). In eukaryotic cells, production of a particular gene product can be temporally and spatially regulated, allowing different cell types or different developmental stages of cells to fine-tune their patterns of protein expression. In fact, a number of developmental events, such as pattern formation and terminal differentiation, can be regulated post-transcriptionally by controlling mRNA stability, localization, and translation (reviewed in Refs. 3 and 4).
RNA-binding proteins play key roles in the post-transcriptional regulation of gene expression; many of them contain an RNA recognition motif (RRM 1 ; also called RBD (RNA-binding domain)) (reviewed in Refs. 5 and 6). RRM-containing proteins include small nuclear ribonucleoproteins (snRNP), polyadenylate-binding proteins (PABPs), nucleolin, and heterogeneous nuclear ribonucleoprotein particle (hnRNP) protein. Although the ubiquity of these RNA-binding proteins suggests a general housekeeping role, some of them may have specific functions. For example, the Drosophila ELAV protein plays a role in neuron-specific RNA splicing (7,8) and vertebrate ELAV-like ribonucleoproteins (elr A, B, C, and D) have been implicated in early development and neuronal differentiation (9).
RNA-binding proteins also play a role in mRNA stability. The human Hu protein inhibits AU-rich element (ARE)-mediated RNA decay in vivo and stabilizes deadenylated RNA intermediates in vitro (10 -15). The AU-rich binding protein (AUF1) stabilized the parathyroid hormone mRNA in an in vitro degradation assay and is also involved in regulation of cyclooxygenase-2 mRNA stability (16,17). In Xenopus, estrogen induces the stabilization of the transcripts of the yolk precursor protein vitellogenin, which is a result of the different affinities of the 155-kDa RNA-binding protein, vigilin, for cisacting stability determinants (18). Furthermore, RNA-binding proteins may have functions in translational processes. For example, PABP is essential to stabilize the 3Ј-end of mRNA; however, PABP interacts with the cap-associated eukaryotic initiation factors eIF4G (in yeast and plants) and eIF4B (in plants). In mammals PABP interacts with a novel PABP-interacting protein that also binds eIF4A (19,20). The Fragile X mental retardation syndrome protein, FMR1, and its homologous FXR proteins are associated with ribosomes, predominantly with 60 S large ribosomal subunits (21). These observations demonstrate that RNA-binding proteins serve as regulatory molecules.
A putative RNA-binding protein gene (RBP1, also called NGR1) was identified in Saccharomyces cerevisiae (22). Rbp1p contains three RRMs, two glutamine-rich sequences, and a C-terminal asparagine-methionine-proline-rich region. Although RBP1 is not an essential gene, over-expression of Rbp1p in yeast yields a slow-growth phenotype, suggesting that Rbp1p is functionally critical in certain biochemical processes. We undertook an analysis of the yeast Rbp1p with the aim of elucidating its cellular role. We report here that the level of mitochondrial outer membrane porin was lower in the Rbp1p yeast strains than it was in an rbp1 mutant. We further dem-onstrate that Rbp1p can accelerate porin mRNA turnover, possibly through binding to the 3Ј-UTR of porin mRNA. These data provide the first evidence that Rbp1p might be involved in post-transcriptional regulation of porin expression in intact cells.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Plasmid Construction-The S. cerevisiae strain YPH499 (MATa ura3-52 lys2-801 ade2-101 trp1-⌬63 his3-⌬200 leu2-⌬1) was the parental strain used for manipulation and gene disruption. The S. cerevisiae strain YTC345 (ϭ YRP582, MATa rpb1-1 ura3 leu2) was a kind gift from Dr. Tien-Hsien Chang (Ohio State University) and was used for mRNA turnover studies. Rich medium contained 1% yeast extract, 2% peptone, and 2% glucose or 2% galactose. Synthetic minimal medium contained 0.7% yeast nitrogen base (without amino acids) and 2% glucose or 2% galactose and was supplemented with the appropriate nutrients for auxotrophs (23). The Escherichia coli vectors pT7Blue and pET-15b (Novagen) were used for gene cloning and manipulation of protein expression. Yeasts were transformed by the lithium acetate method (24). Plasmids were constructed by standard protocols (25). For creating the yeast expression vector encoding N-terminal HA-tagged Rbp1p, the sequence encoding yeast RBP1 was amplified by PCR and inserted into the XhoI-XbaI site of the yeast expression vector pVT101U (26). RBP1 truncated at the N terminus (RBP-1dN, amino acids 1-108 deleted) and RBP1 mutated in each RRM (Rbp1-rrm1, Rbp1-rrm2, and Rbp1-rrm3) were generated by PCRbased mutagenesis. The sequences of the resulting constructs were verified. A GeneAmp RNA PCR Core Kit (PerkinElmer Life Sciences) was used for RT-PCR analyses.
Expression and Purification of Recombinant Rbp1p-For construction of a His-tagged Rbp1p fusion protein, a DNA fragment containing the full-length yeast RBP1 coding region was generated by PCR amplification of yeast genomic DNA with sequence-specific primers that incorporated unique NdeI and BamHI sites at the initiating methionine and 6 bp downstream of the stop codon, respectively. The PCR product was ligated to the pT7Blue vector and subcloned into the E. coli expression vector pET15b, yielding pET15b-RBP1. For construction of GST fusion proteins of Rbp1p and its mutants, DNA fragments containing the RBP1 or its mutants (GST-RBP1, GST-RBP1-rrm1, GST-RBP1-rrm2, GST-RBP1-rrm3) were generated by PCR amplification with sequence-specific primers and subcloned into the pGEX6p-1 vector. Procedures for large-scale production and purification of recombinant proteins have been published (27).
Antibody Preparation and Purification-To prepare antibodies against Rbp1p, purified recombinant Rbp1p was isolated by SDS-PAGE. The gel containing Rbp1p was smashed, thoroughly mixed with Freund's adjuvant, and used to immunize rabbits. The anti-Rbp1p anti-serum was diluted 1:5000 for Western blot analysis. The polyclonal anti-Rbp1p antibodies were affinity-purified for immunofluorescence staining (28).
Western Blot Analysis and Immunofluorescence Microscopy-Yeast total proteins were prepared and subjected to Western blot analysis as described (27). Cells were prepared for immunofluorescence staining as described by Huang et al. (29). Alexa 594-or Alexa 488-conjugated anti-IgG antibodies (Molecular Probes) were used as secondary antibodies. H33258 was diluted in mounting solution for nucleic acid staining. Fluorescence microscopy was performed with a Nikon Microphot SA microscope.
Monoclonal anti-yeast mitochondria porin, anti-yeast cytochrome oxidase subunit III, and subunit IV antibodies were purchased from Molecular Probes, Inc. (Eugene, OR). The monoclonal anti-HA antibody was from Berkeley Antibody Co. (BabCO, Richmond, CA). Monoclonal anti-yeast ribosomal protein Rpl3p antibody (also called anti-TCM1 antibody) was a gift from Dr. J. Warner (Albert Einstein College of Medicine, New York), and the polyclonal anti-Kar2p antibody was a gift from Dr. M. Rose (Princeton University, NJ). Polyclonal anti-porin antibody was diluted 1:20000 for Western blot analysis and 1:5000 for immunofluorescence staining.
Sedimentation Centrifugation Analysis-Yeast strains were cultured in 50 ml of synthetic medium containing 2% galactose or 2% glucose with shaking at 30°C overnight. Cells were pelleted and brought into spheroplasts by lyticase treatment. Spheroplasts were disrupted on ice with a Dounce homogenizer as described (29). As a second method, cells were suspended in lysis buffer as described above and broken by vortexing with glass beads (30). Homogenates were incubated at 4°C with gentle rocking for 20 min and then centrifuged for 5 min at 1,600 ϫ g (P4). The supernatants were centrifuged for 10 min at 10,000 rpm to yield mitochondria-rich pellets (P10). The supernatants were further centrifuged at 100,000 ϫ g for 1 h to produce microsome-rich pellets (P100) and supernatants (S100) containing soluble proteins.
Growth Phenotype Analysis-Yeast over-expressing Rbp1p or its truncated forms was streaked on synthetic medium containing 2% glucose. After incubation at 30°C for 3 days, cells were harvested and suspended in sterile double distilled H 2 O. After adjustment of the cell suspensions to A 600 ϭ 1, they were diluted serially 10-fold to 1 ϫ 10 Ϫ4 . Samples (5 l) of each cell suspension were spotted on synthetic medium containing 2% glucose or 2% glycerol as the carbon source, incubated at 30°C for 3 days, and photographed.
Ribosome-Polysome Profile Analysis-Yeast strains were grown with shaking at 30°C in 50 ml of SC medium containing 2% glucose or 2% galactose to early or middle logarithmic phase. Cycloheximide was added to cultures to a final concentration of 50 g/ml followed by incubation at 30°C for 10 min and cooling on ice. Cells were harvested, washed once with TNM buffer (Tris-HCl, pH 7.5, 10 mM, NaCl, 100 mM, MgCl 2 , 30 mM, heparin, 200 g/ml, and cycloheximide, 100 g/ml, dissolved in diethyl pyrocarbonate-ddH 2 O), suspended in 0.4 ml of TNM buffer, and broken by vortex mixing with an equal volume of glass beads for 3 min. After the addition of 0.6 ml of TNM buffer, the lysate was clarified by centrifugation (5,000 ϫ g for 5 min at 4°C), and the supernatant was centrifuged at 12,000 ϫ g for 10 min at 4°C. Supernatant representing about 8 OD 260 units were applied to a continuous 4.5 ml of 7-47% sucrose gradient (containing, 50 mM Tris acetate, pH 7.0, 50 mM NH 4 Cl, and 12 mM MgCl 2 ) and fractionated by centrifugation in a Hitachi P55ST2-403 rotor at 29,000 rpm for 3.5 h at 4°C. After centrifugation, the gradients were analyzed by measurement of absorbance at 254 nm in an ISCO UA-6 absorbent monitor.
RNA Blot Analysis and RNA Turnover Analysis-Yeast total RNA was isolated by the hot acid phenol method (31). RNA blot analysis was carried out as described previously (22) We used yeast strain YTC345 carrying a temperature-sensitive RNA polymerase II allele (rpb1-1) (32) to study the RNA turnover rate. The RBP1 gene of YTC345 was disrupted by hisG-URA3-hisG construction as described previously (22). YTC345 and YTC345 rbp1::URA3 were grown in synthetic medium containing 2% galactose and 0.1% glucose with shaking overnight at 25°C. An equal volume of 50°C sterile double distilled H 2 O was then added, and the culture was placed in a 37°C water bath. Samples of cells were withdrawn at the indicated times for RNA isolation and Northern blot analysis. The mRNA decay was determined by densitometry of Northern blots, using an AlphaImager TM 1220 documentation and analysis system (AlphaImager 1220, version 5.5). The data were processed with SigmaPlot 2000 software for the slope plot.
Gene Construction for RNA Electrophoretic Mobility Shift Assay-The full-length porin ORF and porin ORF additions 200 bp upstream of the start codon and 200 bp downstream of the stop codon were obtained by PCR and cloned into pT7blue plasmid to generate pT7por1. A 192-bp fragment of the porin 3Ј-end ORF (por3Ј), a 345-bp fragment (por3ЈU) containing the 192-bp por3Ј plus 153 bp of the porin mRNA 3Ј-UTR, and a 177-bp fragment (por3Ј-1) containing 24-bp por3Ј and 153 bp of the porin mRNA 3Ј-UTR were obtained by PCR and ligated to pBluescript II KS(ϩ) plasmid to generate pBKSpor3Ј, pBKSpor3ЈU, and pBK-Spor3Ј-1. The dissected fragments of the porin mRNA 3Ј-UTR were performed by PCR and subcloned into the pBluescript II KS(ϩ) plasmid.
In Vitro Transcription-Labeled and unlabeled RNA probes were generated by in vitro transcription. Samples (3 g) of plasmids were linearized by restriction enzyme digestion and extracted with phenol/ chloroform. After precipitation by salt and isopropanol, the linearized plasmids were suspended in 20 l of TE (Tris-EDTA) buffer. For in vitro transcription, the 50-l reaction mixture contained 1ϫ reaction buffer, 7.5 mM dithiothreitol, 5 g of bovine serum albumin, 0.2 mM rATP, 0.2 mM rGTP, 0.2 mM rCTP, 40 units of RNasin (Promega), 0.3 g of template DNA, 50 units of T7 RNA polymerase, (Stratagene), and 75 Ci of [ 32 P]UTP. After incubation at 37°C for 2 h, the template DNA was digested with DNase I. The uncoupled [ 32 P]UTP was removed on a Microspin G-50 column (Amersham Biosciences). The specific activity of RNA products was 1.5ϳ2.0 ϫ 10 6 cpm/g. For in vitro transcription, without [ 32 P]UTP, 0.2 mM rUTP was added to the reaction mixture. Unlabeled RNA fragments were extracted with phenol/chloroform, precipitated by salt and ethanol, suspended in 50 l of diethyl pyrocarbonate-ddH 2 O, and stored at Ϫ20°C.
RNA Electrophoretic Mobility Shift Assays (REMSA)-For RNA mobility shift assays, recombinant GST-tagged RBP1 (GST-RBP1, GST-RBP1-rrm1, GST-RBP1-rrm2, GST-RBP1-rrm3), and GST proteins were synthesized in E. coli(BL21) cells and purified by binding to glutathione-Sepharose beads with elution by reduced glutathione. The 10-l reaction mixture contained of 10 mM Hepes buffer, pH 7.4, 3 mM MgCl 2 , 40 mM KCl, 5% glycerol, 1 mM dithiothreitol, 100 ng of recombinant GST-RBP1 or recombinant GST, 1 ϫ 10 4 cpm of RNA probe (ϳ 10 ng), 2 g of poly(I-C), and 50 g of heparin. After 30 min at room temperature, 1 unit of RNase T1 was added to the mixture, and incubation was continued for 15 min. Loading dye was added, and the reaction components were resolved in 5% acrylamide-bis-acrylamide (39:1) mini-gel at 90 V for about 2 h in Tris-glycine buffer. After drying, the gel was used to expose the gel on an x-ray film. For the competition assays, an excess of the indicated unlabeled competitor RNAs was added to the labeled probe before the reaction.
In Vivo Pull-down Assay and RNA Detection by RT-PCR-For pulldown of the Rbp1p⅐RNA complex cross-linked in intact cells, we employed the tandem affinity purification (TAP) tag method (33). A DNA cassette encoding the TAP tag was integrated into the genome of a haploid cell in-frame with the C-terminal region of the Rbp1p without altering its natural level of expression. Yeast cells expressing Rbp1p-TAP tag were grown overnight in synthetic medium containing 0.2% glucose and 2% galactose, harvested, and washed twice with double distilled H 2 O. For intracellular in vivo cross-linking of the protein-RNA complex (34), cells were suspended in fixing solution (50 mM Hepes, pH 7.4, and 100 mM NaCl) with 0.3% (v/v) formaldehyde and incubated at room temperature for 20 min with gentle rocking. Cross-linking reaction was quenched by the addition of glycine (pH 7.0) to a final concentration of 0.25 M followed by incubation at room temperature for 10 min. Cells were harvested by centrifugation, washed twice with fixing solution, and dispersed in TST buffer (50 mM, Tris-Cl, pH 7.5, 150 mM, NaCl, 0.05%, Tween 20) containing protease inhibitors. Cells were broken by vortex mixing with glass beads, debris was removed by centrifugation (5,000 rpm for 10 min), and the supernatant was incubated with IgG beads (rocking, 4°C for 2 h). The IgG beads were pelleted, washed five times with TST buffer, and suspended in 50 mM Tris-Cl, pH 7.5, 100 mM NaCl containing 2 units of DNase I. After rocking at room temperature for 20 min, the IgG beads were pelleted, washed twice with TST solution, and incubated at 70°C for 45 min in TEDS solution (50 mM Tris-Cl, pH 7.5, 5 mM EDTA, 10 mM dithiothreitol, and 1% SDS) for reversal of cross-linking. The IgG beads were removed, the supernatant was extracted with phenol/chloroform, and RNAs were precipitated with salt and ethanol at Ϫ20°C. Porin and control mRNAs were identified by RT-PCR using specific primers.

Rbp1p Is Associated with Heavily Sedimented Structures and
Is Localized in Part to the Perinuclear Region-To investigate the function of Rbp1p, we raised polyclonal antibodies against recombinant His-tagged Rbp1p (see "Experimental Procedures"). Among total cellular proteins from wild-type yeast and a strain over-expressing Rbp1p, anti-Rbp1p antibodies mainly reacted with an ϳ 90-kDa protein, the size expected for Rbp1p (Fig. 1A, lanes 1 and 2). This protein was not detected in an rbp1 mutant (Fig. 1A, lane 3) or by reaction with preimmune serum (data not shown). Two nonspecific signals were eliminated after affinity purification of anti-Rbp1p antibodies (Fig.  1A, lane 4). Affinity-purified antibodies of Rbp1p were used in all of the later experiments. To determine the subcellular distribution of endogenous Rbp1p, we fractionated spheroplasthomogenized lysates into three fractions, mitochondria-rich (P10), microsome-rich (P100), and soluble fractions (S100), by sedimentation centrifugation. Rbp1p was equally represented in both the P10 and P100 fractions but was barely detectable in the soluble fraction (Fig. 1B), indicating that Rbp1p was cofractionated with heavily sedimented materials.
The subcellular localization of Rbp1p was also assessed by indirect immunofluorescence. The signals were too weak to allow localization of Rbp1p in wild-type strains, which may be because of the low affinity of the Rbp1p antibody for native Rbp1p. For this reason, cells over-expressing epitope-tagged Rbp1p, HA-Rbp1p, were used for subsequent immunofluorescence experiments. As shown in Fig. 1C, most of the Rbp1p immunoreactivity was, like that of Kar2p, concentrated in the perinuclear region. In addition, some punctate signals were distributed in the cytoplasm. In other experiments, HA-Rbp1p appeared to be colocalized with mitochondrial porin in some Lysates of yeast 499 wild-type cells were separated into mitochondria-rich (P10), microsome-rich (P100), and soluble (S100) fractions. Proteins from each fraction were separated by SDS-PAGE followed by immunoblot analysis using antibodies against proteins indicated on the left. Kar2p, ER protein; porin, mitochondrial protein; Rpl3p, ribosomal protein; Pgk1p, cytoplasmic protein. C, localization of Rbp1p by immunofluorescence. rbp1/HA-Rbp1p cells were grown in selection medium and fixed with formaldehyde. Spheroplasts were generated; monoclonal anti-HA antibody and polyclonal anti-Kar2p antibodies were used for Rbp1p and endoplasmic reticulum staining, respectively. Nuclear DNA was stained with H33258. Phase-contrast images are also shown. Arrowheads indicate perinuclear region. cells (data not shown). Our efforts to demonstrate Rbp1p in the vicinity of the mitochondria in most cells were, however, unsuccessful.
Overexpression of Rbp1p Affects Mitochondrial Porin Expression and Impairs Mitochondrial Function-While performing the immunostaining studies with an anti-porin antibody to identify mitochondria, we observed that porin staining was significantly less noticeable in cells over-expressing Rbp1p than in those of rbp1 mutants ( Fig. 2A, arrows). To determine whether Rbp1p suppressed porin expression, we quantified porin protein levels. Cells were grown overnight in synthetic medium containing 2% galactose and 0.2% glucose. The cell lysates were prepared and subjected to Western blot analysis. As shown in Fig. 2B, the amount of porin was lower in cells expressing Rbp1p (wild-type, low copy-expressing or high copyover-expressing Rbp1p) than in the rbp1 mutant. However, over-expression of the mutant Rbp1-dN apparently did not affect the porin level. The amount of ribosomal 60 S protein Rpl3p and cytoplasmic protein Pgk1p was similar in all strains examined. It appeared that over-expression of Rbp1p, but not Rbp1-dN, reduced the steady-state levels of porin.
Next, we investigated whether over-expression of Rbp1p had an effect on mitochondrial function. Yeast cells, when grown in glycerol medium, require mature active mitochondria for oxidative metabolism and growth. Yeast grown in glucose initially does not need active, mature mitochondria, and the mitochondria are not well developed until cells switch from anaerobic glucose fermentation to aerobic ethanol oxidation. Wild type, rbp1 mutant, and rbp1 strains over-expressing Rbp1p or Rbp1-dN were grown overnight in synthetic medium contain-ing glucose. Cells were harvested, suspended in double distilled H 2 O, serially diluted, and plated on glucose or glycerol medium. As shown in Fig. 2C, all strains grew on glucose plates; however, cells over-expressing Rbp1p grew much more slowly. Moreover, with the exception of those cells, all strains grew on glycerol plates (Fig. 2C). These results suggest that over-expressing Rbp1p interferes with events in mitochondrial biogenesis that are required to enable cells to grow on nonfermentable carbon sources.
Rbp1p Does Not Affect General Polysome Profiles-To test whether Rbp1p suppresses the expression of porin at the level of translation, we analyzed the ribosome-polysome profiles in rbp1 mutant or HA-Rbp1p-over-expressing strains. Sucrose gradient sedimentation showed no significant ribosome-polysome defects in the rbp1 mutant and HA-Rbp1p over-expressing strains, although the A 254 of 40 S ribosomal subunits was increased slightly from 0.14 and 0.16 to 0.24 and 0.2, respectively, when Rbp1p was over-expressed in rbp1 and wild type strains (Fig. 3, upper panel). Western blot analysis of 12 gradient fractions (from top to bottom fractions) (Fig. 3, lower  panel) showed that most of the Rbp1 was in the top fractions (fractions 1-3) with a small portion in the 40 S subunit-rich fraction (fraction 4). Rbp1p may not be directly associated with 40 S ribosomal subunits but most likely interacts with mRNPs that co-fractionates with free 40 S ribosomal subunits.
Rbp1p Affects Steady-state Levels of Mitochondrial Porin mRNA-To determine whether Rbp1p altered the steady-state levels of porin mRNA, we compared different yeast strains expressing Rbp1p. Fig. 4 shows that the levels of porin mRNA in rbp1 and RBP1-dN-overexpressing cells were considerably Phase-contrast image is also shown. Arrows indicate cells over-expressing HA-Rbp1p. B, mitochondrial porin in Rbp1p strains (wild-type, rbp1, rbp1/high copy over-expressing HA-Rbp1p-dN, rbp1/high copy over-expressing HA-Rbp1p, rbp1/low copy (CEN-RBP1) expressing Rbp1p (24)). Cells were grown in synthetic galactose medium and disrupted, and protein extracts were prepared as described under "Experimental Procedures." The distribution of Rbp1p and other marker proteins was evaluated by Western blot analysis with specific antibodies (lower panel). Upper panel, SDS-PAGE stained with Coomassie Blue. C, over-expression of Rbp1p inhibited cell growth. Yeast strains were grown and prepared as described under "Experimental Procedures." 10-Fold serial dilutions of each cell suspension were spotted on synthetic medium containing glucose or glycerol as the carbon source and incubated at 30°C for 3 days before being photographed.
higher than in the wild-type and Rbp1p-overexpressing strains, regardless of the carbon source. Amounts of the actin mRNA, however, were similar in all cells grown in galactosecontaining medium. These results indicate that Rbp1p did affect levels of mRNA for the nuclear-encoded mitochondrial protein porin.
Rbp1p Is Involved in the Regulation of Porin mRNA Stability-To determine whether Rbp1p regulates porin mRNA levels post-transcriptionally, we used yeast strain YTC345, which carries an RNA polymerase II temperature-sensitive (37°C) mutation, to evaluate the mRNA degradation rates. YTC345 and YTC345rbp1 strains were cultured at 25°C overnight, and samples were taken at several time points after raising the temperature to 37°C. Isolated RNAs were subjected to Northern blotting. As shown in Fig. 5A, the turnover of porin mRNA, but not of actin, COX3, CTY1, PGK1, or STE2, was slower in the rbp1 strain (half-life ϭ 57 min) than in wild-type cells (half-life ϭ 37 min), suggesting that Rbp1p is involved in the regulation of porin mRNA stability.
To determine whether over-expression of Rbp1p destabilized porin mRNA, YTC345rbp1 yeasts carrying either pVT101U/ HA-RBP1 or pVT101U were cultured overnight at 25°C and transferred to 37°C before harvesting at the indicated times. Degradation of porin mRNA was faster in cells over-expressing Rbp1p (half-life ϭ 23 min) than in rbp1 cells (half-life ϭ 55 min) (Fig. 5B). At least three genes (POR1, POR2, and TOM40) encode mitochondrial porins (35); however, the stability of POR2 and TOM40 mRNA was not affected by the amount of Rbp1p (data not shown). In addition, actin mRNA, but not COX3, CYT1, or PGK1, appeared more unstable in the strain over-expressing Rbp1p than in rbp1 strain, suggesting that Rbp1p may influence decay of more than one mRNA.
Rbp1p contains three copies of the RRM. To determine which of the RRMs is involved in destabilizing porin mRNA, two conserved phenylalanine (F) residues in the RNP1 domain of each RRM were replaced with aspartic acid (D). Mutants of RRM1, RRM2, and RRM3 were designated as Rbp1-rrm1, Rbp1-rrm2, and Rbp1-rrm3, respectively (Fig. 6A). Expression of these Rbp1-rrm mutants in YTC345rbp1 was confirmed by Western blotting (Fig. 6B). mRNA turnover was measured in these mutants (Fig. 6C) as it was in Fig. 5. Over-expression of each of the RRM mutants in YTC345rbp1 was without significant effect on the half-life of porin mRNA (55-60 min), in contrast to the effect of wild-type Rbp1p (half-life ϭ 23 min in Fig 5B). This result indicates that each of the three RRMs in Rbp1p is essential for the regulation of porin mRNA stability.
Rbp1p Interacted with Porin mRNA in Intact Cells-To determine whether Rbp1p interacts with porin mRNA in cells, we employed the TAP tag to purify a Rbp1p complex and associated mRNA, as described under "Experimental Procedures." RNP complexes containing Rbp1p-TAP were first isolated by binding to IgG beads (Fig. 7A). As shown in Fig. 7B, porin   FIG. 3. Effects of ribosome-polysome profiles by rbp1 mutation or over-expressed Rbp1p. Yeasts were grown in synthetic medium containing 2% galactose (initial A 600 ϭ 0.2). In early log phase (ϳA 600 ϭ 0.5-0.6), cycloheximide was added, and cells were harvested as described under "Experimental Procedures." Proteins were extracted; ribosomal subunits and polyribosomes were separated by centrifugation in 7-47% sucrose gradients, and absorbance at 254 nm was recorded with a UA-6 monitor (ISCO). A parallel gradient for each strain was divided into 12 portions (from light to heavy) for analysis by Western blotting to determine the distribution of RBP1, ribosomal 40 S subunit-associated protein PRT1, and ribosomal 60 S subunit protein L3. mRNA was detected in association with Rbp1p-TAP but not with Rbp1p by RT-PCR with specific primers. The control CYT1 was not detected in either precipitate. Detection of the porin mRNA fragment in the Rbp1⅐TAP pull-down complexes suggests that Rbp1 interacts with porin mRNA in vivo.
Rbp1p Interacts with Porin mRNA 3Ј-UTR in Vitro-We employed a REMSA (Fig. 8) to determine whether Rbp1p could interact directly with porin mRNA in vitro. A number of well defined cis-elements appear to mediate the stability of specific mRNAs by interaction with specific trans-acting factors. With few exceptions, cis-elements are positioned within the 3Ј-UTR (36). To test whether Rbp1p could interact with the 3Ј-UTR of porin mRNA, we constructed a 192-bp fragment of the porin 3Ј-end ORF (por3Ј), a 345-bp fragment (por3ЈU) containing the 192-bp por3Ј plus 153 bp of the porin mRNA 3Ј-UTR, and a 177-bp fragment (por3Ј-1) containing the last 24 bp of por3Ј plus the 153 bp of the porin mRNA 3Ј-UTR (Fig. 8A). Purified recombinant GST-tagged Rbp1p (GST-Rbp1p) and GST proteins were used in the REMSA. Fig. 8B shows that the migration of por3Ј-1 and por3ЈU, but not por3Ј, fragments was substantially retarded when incubated with GST-Rbp1p (Fig 8B). The intensity of these shifted bands increased with increasing Rbp1p concentration (data not shown).
The binding specificity was further examined in a competition experiment. The 32 P-por3ЈU fragments were incubated with GST-Rbp1p in the presence of excess unlabeled por3Ј-1 or porin 3Ј fragments. The retarded migration signals of 32 P-por3ЈU probes were efficiently competed by an excess of unlabeled porin 3Ј-1 but not by 100-fold excess porin 3Ј (data not shown). These results indicate that Rbp1p can interact directly with the 3Ј-UTR of porin mRNA in vitro.

Rbp1p Preferentially Interacts with (C/G)U-rich Sequences-
To identify the region of the porin mRNA 3Ј-UTR involved in Rbp1p binding, we constructed four fragments of the porin 3Ј-UTR as illustrated in Fig 9A. Unlabeled RNA fragments were generated by in vitro transcription and quantified. In competition assays with purified recombinant GST-Rbp1p and 32 P-labeled por3Ј-1, virtually no Rbp1p⅐ 32 P-por3Ј-1 retarded complex was seen with a 10-fold excess of unlabeled por3Ј-3 fragment but not with 50-fold concentrations of unlabeled por3Ј-2, -4, or -5 (Fig. 9A, and data not shown). The por3Ј-3 element contains several repeats of C/G(U) 2ϳ3 sequence near its 5Ј end, and por3Ј-5 contains a UUAUUUAUA sequence, which is a typical AU element. Hence, the 3Ј-UTR of porin mRNA contains a target sequence (ϩ48 to ϩ70) with which Rbp1p could interact.
To confirm the existence of that binding sequence in Rbp1p, 20 nucleotides (ϩ50 to ϩ 69) near the 5Ј-end of por3Ј-1 were deleted to generate the por3Ј-1(d3) as illustrated in Fig. 9B. In competition assays (Fig. 9C), the Rbp1p⅐ 32 P-por3Ј-1 retarded complex was essentially abolished by a 10-fold concentration of unlabeled por3Ј-1 but not by a 50-fold concentration of unlabeled por3Ј-1(d3). These results suggested that the Rbp1p preferentially interacts with the (C/G)U-rich sequence rather than the typical AU element. DISCUSSION We serendipitously found that over-expression of Rbp1p impaired mitochondrial function and affected the expression of mitochondrial porin. The half-life of porin mRNA was greater in rbp1 mutants than in wild-type cells, whereas in cells overexpressing Rbp1p from a multicopy plasmid, the stability of A, wildtype YTC345 and YTC345 rbp1 strains were cultured overnight in synthetic medium containing 2% glucose at 25°C. After cultures were transferred to a 37°C water bath, and cells were harvested at 15-min intervals for 60 min. Samples (10 g) of total RNAs were subjected to electrophoresis, blotted to polyvinylidene difluoride membrane, and hybridized with [ 32 P]dCTP-labeled DNA probes as indicated on the left. 25 S and 18 S rRNA are shown in the ethidium bromide-stained gel. STE2 was used as internal mRNA turnover control. B, over-expressing Rbp1p accelerated the turnover of porin and actin mRNAs. RNA from both YTC345 rbp1/pVT101U and YTC345 rbp1/pVT101U-HA-RBP1 strains was prepared and analyzed as described in A.
porin mRNA was decreased. We have presented several pieces of evidence suggesting that Rbp1p binds to the 3Ј-UTR of porin mRNA in vitro and that the three RRMs of Rbp1p are each involved in destabilizing porin mRNA. To our knowledge, this is the first demonstration of specific regulation of a nuclearencoded mitochondrial gene transcript in intact cells, and it also identifies the trans-acting factor involved in regulating porin mRNA stability.
The biochemical characterization of Rbp1p revealed that it was associated in cells with structures sedimenting at P10 and P100 fractions. Rbp1p, like Rpl3p, was present in the P100 fraction, which perhaps is due to its association with ribosomebound mRNA. Analysis of the ribosome-polysome profiles of Rbp1p, however, revealed that most of the Rbp1p was not associated with ribosome-bound mRNA. An alternative possibility is that Rbp1p may sediment when it is present in mito-chondrial RNP granules, which contain mRNA and RNA-binding proteins. Consistent with this notion, we found that on fluorescence microscopy Rbp1p displayed punctate or granule signals, which were concentrated in part in the perinuclear region and also scattered through the cytosol. Moreover, our preliminary two-hybrid screening identified several Rbp1p-interacting molecules, including an RNA-binding protein involved in RNA degradation. 2 Elucidation of the functional relationship of Rbp1p with its interacting proteins may reveal a novel selective RNA regulation.
Mitochondrial biogenesis is a complex process involving the concerted regulation of expression of the two genomes that encode organellar proteins.  (Q1 and Q2), and a C-terminal asparagine-methionine-proline (NMP)rich region. In each RRM, two conserved phenylalanine (F) residues were replaced with aspartic acid (D) by site-directed mutagenesis. The mutants of RRM1, RRM2, and RRM3 are designated Rbp1-rrm1, Rbp1-rrm2, and Rbp1-rrm3, respectively. B, expression of Rbp1-rrm mutants in YTC345rbp1 demonstrated by Western blotting with anti-HA antibody. C, overexpression of Rbp1p-rrm mutants did not accelerate the mRNA turnover of porin. Experiments were carried out as described in the legend for Fig. 5A.

FIG. 7. Rbp1p in cells interacts with porin mRNA.
A, wild-type and TAP-Rbp1 cells were grown and treated with 0.3% formaldehyde for protein-RNA cross-linking as described under "Experimental Procedures." The TAP⅐Rbp1⅐RNP complex was purified by binding to IgG beads. The protein-RNA cross-link was reversed by heating at 70°C for 45 min. The reversed supernatant and IgG beads were subjected to Western blot analysis using affinity-purified Rbp1p antibody. The arrow indicates TAP⅐Rbp1p. B, mRNA in the TAP⅐Rbp1⅐RNP reversed supernatant was extracted and precipitated before amplification. The 114-bp fragment of the porin mRNA 3Ј-UTR was detected by RT-PCR with specific primers. CYT1 3Ј-UTR was used as a negative control.
proteins are encoded in nuclear DNA and synthesized on cytoplasmic polysomes as precursor molecules with short N-terminal extensions. Importation of mitochondrial proteins occurs, for the most part, post-translationally, although some precursors may also be imported co-translationally (37)(38)(39). Mitochondrial proteins that are encoded in nuclear DNA must be reliably delivered to the right place. This could be accomplished by transport of the appropriate mRNAs from the nucleus to a site of protein synthesis near the mitochondria. It was suggested that organelle-specific mRNA-binding factors might serve to cluster and anchor specific mRNAs in the vicinity of mitochondria (40). More recently, a subset of nuclear-encoded mRNAs was reported present adjacent to mitochondria (41). To date, no organelle-specific RNA-binding factors have been identified, although we found that a truncated Rbp1p lacking two glutamine-rich regions was localized to mitochondria (unpublished observation). It will be important to determine whether Rbp1p can shuttle between the endoplasmic reticulum-perinuclear region and the mitochondria and whether this transit is an obligatory aspect of its function in selective nuclear-encoded mitochondrial RNA metabolism.
The level of porin mRNA was decreased when Rbp1p was expressed, and its turnover rate was higher in the Rbp1pexpressing strain. The destabilization of porin mRNA by overexpression of Rbp1p that we observed was not a nonspecific effect. The mRNAs of the nuclear-encoded mitochondrial CYT1, cytosolic PGK1, or mitochondrial DNA-encoded COX3, however, were not destabilized by over-expression of Rbp1p. Surprisingly, the half-life of actin mRNA appeared to be shorter in the Rbp1p-overexpressing strain than in rbp1 cells, suggesting that over-expression of Rbp1p may deplete or sequester a factor(s) involved in degradation of additional mRNA. Rbp1p is a similar in domain organization and amino acid sequence to two cytolytic lymphocyte proteins, TIA-1 and the closely related TIAR (42,43). TIA-1 and TIAR have three RRM domains near their N termini that confer high-affinity binding to uridine-rich motifs (44). TIAR was found as a component of the ARE-associated complex that assembles on the 3Ј-UTR of tumor necrosis factor-␣ transcripts (45,46), suggesting that TIA-1 and TIAR might specifically regulate the expression of tumor necrosis factor-␣. In addition, TIA-1 was reported to be a regulator of alternative pre-mRNA splicing (47,48), and both TIA1 and TIAR can cause the general translational arrest upon environmental stress. Following stress-induced phosphorylation of translation initiation factor eIF-2, TIA-1 and TIAR recruit selected cytoplasmic poly(A) ϩ RNAs to discrete foci known as stress granules (49), suggesting that the TIA-1/TIARdependent sequestration of these mRNAs prevents initiation of their translation. David and Parker (50) suggested that the relative efficiency of translation of a transcript will be a major determinant of mRNA half-life. In other words, any event that results in inefficient translation initiation may promote mRNA degradation. Our data showed that over-expressing Rbp1p resulted in only a slight increase in the amount of 40 S ribosomal subunits, and a portion of Rbp1p was distributed with 40 S ribosomes. Although we found that the stability of yeast porin mRNA was affected by Rbp1p, we cannot rule out a function for Rbp1p as a translational silencer, similar to TIA-1/TIAR, which therefore interferes with or causes a lag in the association of ribosomal subunits with selective mRNAs, resulting in inefficient translation and acceleration of mRNA decay.
The RRM domain of RNA-binding proteins is generally believed to have a functional role in direct RNA binding. Hu proteins are human neuron-specific RNA-binding proteins that contain three RRMs. The first two RRM domains of Hu proteins are necessary and sufficient for binding to an ARE sequence, but the function of the third RRM domain is uncertain (51)(52)(53)(54)(55)(56)(57). Rbp1p as in Hu RRM1 and RRM2, but not RRM3, were necessary for binding the 3Ј-UTR of porin mRNA in vitro. Using TAP affinity purification followed by RT-PCR analysis, we demonstrated that the Rbp1p interacted with porin mRNA in cells. Nevertheless, the RNA turnover assay revealed that all three RRMs were required for destabilization of porin mRNA. A study of the interaction between HuD and a proto- were cloned in pBluescript II KS(ϩ) plasmid. 32 P-labeled por3Ј, por3ЈU, and por3Ј-1 RNA fragments were generated by in vitro transcription. 100 ng of purified recombinant GST-RBP1 or GST protein were incubated with the indicated 32 P-labeled RNA probe (ϳ10 ng, 1 ϫ 10 4 cpm), and REMSA was performed as described under "Experimental Procedures." C, the recombinant GST-tagged Rbp1-rrm-mutated proteins (GST-Rbp1-rrm1, GST-Rbp1-rrm2, and GST-Rbp1-rrm3) were synthesized in E. coli and purified for REMSA. D, 100 ng of purified recombinant GST-RBP1, GST-Rbp1-rrm1, GST-Rbp1-rrm2, or GST-Rbp1-rrm3 protein was incubated with 32 P-labeled por3Ј-1 RNA, and REMSA was performed. Data are representative of at least three experiments. type mRNA instability element (58) has revealed that the individual RRMs bind poorly to the RNA; however, the combination of RRM1 and either RRM2 or RRM3 in the context of the protein increases the binding affinity. Thus, the three RRMs appear to cooperate not only to increase the affinity of interaction but also to stabilize the complex. Moreover, the slower binding and stabilization of the RNA-Hu protein interaction were observed in the presence of all three RRMs, suggesting that a change in the complex structure may occur with time (57). It is possible that the RRM3 of Rbp1p is not involved directly in RNA binding but functions in RNA metabolism by interaction with other proteins. Alternatively, RRM3 may have an auxiliary function to assist in stabilizing the protein-RNA complex in a tertiary structure. Therefore, the biologic activity of Rbp1p not only is dependent on its RNA binding capacity but also is involved in the interaction between molecules.
Regulatory sites in the 3Ј-untranslated region of mRNAs are critical for control of translation, and for RNA stability and localization (41, 59 -62). The ARE is the most extensively studied instability element in mammalian mRNAs. An ARE is invariably present in the 3Ј-UTR of a subset of rapidly degraded mRNAs including those encoding cytokines, protooncogenes, transcription factors, cell-cycle regulators, and cell surface receptors (63). The ARE is uridine-rich, with interspersed adenosine moieties. Chen and Shyu (64) have categorized the AREs into three functional classes based on sequences and mechanisms of decay. The 3Ј-UTR sequence of porin mRNA contains a 54-nucleotide U-rich stretch with interspersed G, C, and A residues (Fig. 9B, ϩ50 to ϩ103 nucleotides). A single nonamer sequence, UUAUUUAUA, in this U-rich region resembles a class I ARE. However, based on the competition experiment, the element does not seem likely to bind Rbp1p.
Several non-ARE cis-elements have been described, including the U-rich polypyrimidine tract. A C(U)nC motif of c-Myc mRNA is recognized by HuD (65). A C(U)nC motif and a CCCUCCC element present in a highly conserved UC-rich region of an androgen receptor mRNA 3Ј-UTR can interact with a group of RNA-binding proteins, including HuR, CP1, and CP2 (66). A C-rich cis-element with a consensus CCUCC or CCCUCCC sequence has been identified in a variety of mRNAs including those for ␣-globin and erythropoietin (67)(68)(69). The 3Ј-UTR of insulin mRNA contains a pyrimidine-rich sequence, which specifically interacts with polypyrimidine tract-binding protein and controls insulin mRNA levels (70). Our results showed that Rbp1p interacts with porin mRNA via its binding to the U-rich region of 3Ј-UTR, which contains a C/G(U)n repeat sequence. Furthermore, the 3Ј-UTR of porin mRNA contains a UA-repeated motif (Fig. 9B, ϩ17 to ϩ24 nucleotides), which is typical of the RNA-binding protein, Hrp1p. Consistent with this observation, our unpublished data 2 from a two-hybrid screening showed that Rbp1p could interact with Hrp1p. Thus, it is possible that the binding specificity of Rbp1p may be attained by interaction with other molecules. In the structure of the full-length 3Ј-UTR of porin mRNA as predicted by RNAdraw (version 1.1b2), the Rbp1p binding sequence (ϩ50 of at least three experiments. B, the por3Ј-1(d3) fragment was generated by deleting 20 nucleotides (ϩ50 to ϩ69) from por3Ј-1. An AU element (ϩ93 nt to 103 nt) in the 3Ј-UTR of porin mRNA is indicated. C, 100 ng of purified recombinant GST-RBP1 protein was incubated with 32 P-labeled por3Ј-1 RNA, and REMSA was performed. For the competition assay, the indicated excess concentration of unlabeled competitor por3Ј-1 or por3Ј-1(d3) was mixed with labeled RNA before protein was added. Data are representative of at least three experiments. D, structure of full-length 3Ј-UTR of porin mRNA calculated and predicted by RNAdraw (version 1.1b2). In the full-length 3Ј-UTR, the defined Rbp1p-binding sequence (panel B, ϩ50 to ϩ69 nucleotides) occupies two-thirds of an open loop.
FIG. 9. Rbp1p preferentially interacts with a (G/C)U-rich element in 3-UTR of porin mRNA. A, competition assays defined the Rbp1p binding sequence in the por3Ј-1 mRNA. A diagram of 3Ј-UTR constructs of porin mRNA is shown. 100 ng of purified recombinant GST-RBP1 protein was incubated with 32 P-labeled por3Ј-1 RNA, and REMSA was performed. For the competition assay, a 20-or 50-fold higher concentration of unlabeled por3Ј-1, -2, -3, -4, or -5 was mixed with the labeled RNA before addition of protein. The competition efficiency of each construct is shown on the right. Data are representative to ϩ69 nucleotides) appears to occupy two-thirds of an open loop (Fig. 9D). It is possible that the formation of a stem-loop structure surrounding the Rbp1p recognition sequence facilitates the Rbp1p-RNA interaction.
In summary, we have shown that Rbp1p could promote porin mRNA decay by binding to its 3Ј-UTR, or, alternatively, modulate the conformation of the porin mitochondrial RNP to allow degradation. The molecular mechanisms by which Rbp1p regulates the degradation of porin mRNA are still elusive. The possibility that Rbp1p recruits other RNA-binding proteins that, in turn, regulate porin expression may exist. Further studies on in vivo binding analysis would help us to unveil how regulatory proteins and mRNA binding regulate the production of nuclear-encoded mitochondrial proteins. Is the activity of Rbp1p limited to mRNA decay, or does it have other functions? Further exploration into the mechanism details of Rbp1p-RNP granule formation and function should provide insights into the choreography of mRNA translation and decay.