Sequence-specific RNA recognition by the Xenopus Y-box proteins. An essential role for the cold shock domain.

The Xenopus Y-box protein FRGY2 has a role in the translational silencing of masked maternal mRNA. Here, we determine that FRGY2 will recognize specific RNA sequences. The evolutionarily conserved nucleic acid-binding cold shock domain is required for sequence-specific interactions with RNA. However, RNA binding by FRGY2 is facilitated by N- and C-terminal regions flanking the cold shock domain. The hydrophilic C-terminal tail domain of FRGY2 interacts with RNA independent of the cold shock domain but does not determine sequence specificity. Thus, both sequence-specific and nonspecific RNA recognition domains are contained within the FRGY2 protein.

The Xenopus laevis Y-box protein FRGY2 is a major component of ribonucleoprotein storage particles containing maternal mRNA within Xenopus oocytes (Murray et al., 1992;Deschamps et al., 1992;Wolffe, 1993a, 1993b). FRGY2 has an active role in facilitating the translational silencing or masking of maternal mRNA (Richter and Smith, 1984;Ranjan et al., 1993;Bouvet and Wolffe, 1994). How FRGY2 interacts with mRNA and how mRNA packaged by the FRGY2 protein is rendered accessible to the translational machinery is not understood.
The eukaryotic Y-box proteins have been reported to bind with some selectivity to a wide variety of duplex and singlestranded nucleic acids including the Y-box (CTGATTGGCCAA) (Marello et al., 1992;Wolffe et al., 1992;Murray, 1994) (reviewed by Wolffe (1994a)); however, the exact nucleic acid binding specificity was not determined. The eukaryotic Y-box proteins recognize DNA through the conserved CSD (Kolluri et al., 1992;Tafuri and Wolffe, 1992). The recognition of RNA has been proposed to be either predominantly via the CSD (Ladomery and Sommerville, 1994) or through interactions with islands of basic/aromatic amino acids in the C-terminal domain of those Y-box proteins (Murray, 1994).
In this work we establish that both the oocyte-specific Y-box protein FRGY2 and a somatic homolog FRGY1 can have highly specific interactions with RNA. We determine that the cold shock domain is required for specific recognition of RNA. The C-terminal domain of FRGY2 has nonspecific interactions with RNA that may account for the more efficient association of wild type FRGY2 with RNA in comparison with the cold shock domain alone. We discuss the significance of these observations for the potential roles of FRGY2 in transcriptional and translational control.
Two steps of PCR amplification were performed to generate the constructs for point mutants of FRGY2 protein. Primers K12 (5Ј-GTT-GATAGCTCCGGCGCCGTTGCGAACGTTAAACC-3Ј; mutations corresponding to aa 55 and 57 of FRGY2 are underlined) and K13 (5Ј-AACGGCGCCGGAGCTATCAACAGAAATGACACCAAA-3Ј) were synthesized to obtain constructs for PM1 and CP1 proteins, and K14 (5Ј-GTTGATAA(G/A)TCCCAAGCCGTTGCG AACGTTAAACC-3Ј) and K15 (5Ј-AACGGCTTGGGA(T/C)TTATCAACAGAAATGACACCAAA-3Ј) were for PM2, CP2, and CP3 proteins. FRGY2 cDNA was amplified by PCR with primer sets of K9/K12 and K13/K8 for PM1 or K9/K14 and K15/K8 for PM2. PCR products were isolated and used for secondary PCR amplification with the primer set of K9/K8. The products from secondary PCR were used as the templates of PCR to generate constructs for CP1, CP2, and CP3 with the primer set of K20/K19. The final products were cloned into pGEX-4T-3 vector as described above. All constructs were completely sequenced to exclude the possible introduction of unintended fortuitous mutations during amplification and cloning.
To obtain GST-fusion FRGY2 protein and its derivatives, BL21(DE3) pLysS was transformed with each pGEX construct. A 300 -500-ml culture was induced by isopropyl-1-thio-␤-D-galactopyranoside to synthe-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ These authors contributed equally to this work.  B1A-13, NICHD, NIH, Bethesda, MD 20892-2710. Tel.: 301-402-2722Fax: 301-402-1323. 1 The abbreviations used are: CSD, cold shock domain; RNP, ribonucleoprotein; GST, glutathione S-transferase; PCR, polymerase chain reaction; aa, amino acids; nt, nucleotide(s); wt, wild type; YRS, FRGY recognition sequence. size GST-fusion proteins. C-terminal deletion mutants D4, CSD, CP1, CP2, and CP3 were purified from bacterial lysate using glutathione-Sepharose 4B according to the method suggested by the manufacturer (Pharmacia). Since wild type, D5, D6, PM1, and PM2 proteins did not bind to glutathione-Sepharose efficiently, they were purified from the pelleted inclusion bodies of Escherichia coli lysate through SP-Sepharose chromatography as described , except that the step of precipitation with ammonium sulfate was omitted and the proteins were eluted from SP-Sepharose with phosphate buffer containing 0.75 M NaCl after washing the column with the buffer containing 0.5 M NaCl.
The full-length FRGY1 and FRGY2 proteins with T7 gene 10 leader peptides (T7-FRGY2) were expressed in E. coli BL21(DE3) transformed with the pET constructs . The Y-box proteins were purified from the pelleted inclusion bodies of E. coli lysate through an SP-Sepharose column as described above.
Random RNA and Selection of Preferred Sequences for FRGY2 Binding (Selex)-RNA was prepared as described previously (Tsai et al., 1991). Briefly, double-stranded transcription template was prepared by subjecting 5 ng of linear N25 oligodeoxynucleotide (Tsai et al., 1991) to 35 cycles in an Ericomp temperature cycler (1 min at 94°C, 1 min at 50°C, 2 min at 72°C) in the following buffer: 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , 0.01% gelatin, 0.1 g of primer T7Univ, 0.1 g of primer RevUniv, 200 M dNTP, and 2.5 units of Taq DNA polymerase. Any tandem products were reduced to monomers by cutting with BamHI. The DNA template was then transcribed using T7 RNA polymerase.
RNA was selected from this random pool by coimmunoprecipitation with the FRGY1 and FRGY2 fusion proteins (Tsai et al., 1991). 1-5 g of full-length T7 gene 10 fusion proteins was bound to 4 mg of protein A beads using the antibody against the gene 10 epitope (Novagen). After three washes with NT2 buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 0.05% Nonidet P-40, 1 mM MgCl 2 ), the protein and 200 -500 ng of RNA were incubated for 7 min in 100 l of reaction mixture (20 mM KCl, 150 mM NaCl, 50 mM Tris, pH 7.5, 0.05% Nonidet P-40, 2.5% polyvinyl alcohol, 1 mM MgCl 2 , 1 mM EGTA, 50 g/ml poly(A), 2 g/ml vanadyl ribonucleoside complex, 0.5 mg/ml tRNA, 125 g/ml bovine serum albumin, 1 FIG. 1. FRGY1 and FRGY2 bind to RNA with sequence selectivity. A, DNA sequences obtained by Selex (see "Materials and Methods") for FRGY1 and FRGY2. The DNA sequence corresponding to the RNA initially selected is shown. B, RNA footprinting of FRGY1 and FRGY2 bound to RNA containing a binding consensus sequence. Methylation protection (see "Materials and Methods") is shown for FRGY1 (lanes 1-4) and FRGY2 (lanes 9 -12). Sequencing markers are shown (lanes 5-8). The sequence containing the Y-box consensus sequence (YRS) is indicated by the vertical bar. Dimethyl sulfate (DMS) alkylates accessible A and C nucleotides; sites of protection within the YRS are indicated by asterisks. mM dithiothreitol, 80 units/ml RNasin (Promega). Following this incubation, the RNA was washed 5 times with NT2 buffer. Ten g of carrier tRNA was then added to the immunoprecipitated RNA. It was then subjected to phenol extraction and ethanol precipitation. Rounds two to five had identical binding conditions except that 0.5 M urea was added to the final NT2 buffer wash to reduce nonspecific binding as described previously.
Reverse transcription and reconstitution of the transcription template were as described previously (Tsai et al., 1991). RNA was reverse transcribed using 0.1 g of primer RevUniv with avian myeloblastosis virus reverse transcriptase (1 h at 42°C using conditions recommended by the supplier). Following reverse transcription, the cDNA was resuspended in 10 l of double distilled water. Three l of the cDNA was subjected to 35 cycles of PCR under the conditions already noted. The new transcription template was then used to repeat the above transcription, coimmunoprecipitation, and reverse transcription steps for four additional rounds. After the last PCR the cDNA was digested with BamHI and then subcloned in pGem/BamHI. Then the clones were sequenced; 13 sequences for Y1 and 10 for Y2 contained a 6-nucleotide (nt) sequence that is very similar in all these clones (see Fig. 1).
To introduce mutations into the FRGY2-binding site, three doublestranded oligonucleotides containing the sequences shown (5Ј-AATTC- EcoRI and SacI sites are underlined) were cloned in EcoRI-and SacI-digested pGEM vector (Promega). Plasmid DNA was sequenced, and one of the mutants, M3/M4 -1, was used to generate the nonspecific RNA for competition used in this study.
Filter Binding Assay-Labeled in vitro transcribed RNA (0.05 pmol) was incubated in the presence of different amounts of GST-FRGY2 or T7 FRGY2 protein in a final volume of 100 l of RNA binding buffer (20 mM Tris-HCl, pH 7.4, 4 mM MgCl 2 , 100 mM KCl, 30% glycerol, 1 mM dithiothreitol, 0.1 mg/ml (GST-FRGY2) or 0.5 mg/ml (T7 FRGY2) yeast tRNA, 2 g/ml vanadyl ribonucleoside complex, 20 g/ml bovine serum albumin) for 20 min at room temperature. RNA-protein complexes were separated from unbound RNA by filtration through a nitrocellulose filter (BA85, Schleicher and Schuell). After three washes with 100 l of 20 mM Tris-HCl, pH 7.4, 4 mM MgCl 2 , 100 mM KCl, the filter was dried and then exposed with x-ray film. The bound RNA was calculated by counting the radioactivity retained on the filter.
Dimethyl Sulfate Footprinting-RNA-protein complexes were formed as described above. Then dimethyl sulfate (1 l) was added for 5 min at room temperature. Alkylation reactions are stopped by adding 50 l of stop solution (1 M Tris-HCl, pH 7.5, 1 M ␤-mercaptoethanol, 0.1 mM EDTA). RNA was immediately precipitated with 2 volumes of 100% EtOH and 0.1 volume of sodium acetate. RNA pellets were resuspended in 0.3 M sodium acetate, 0.5% SDS, 5 mM EDTA. After extraction twice with phenol/CHCl 3 , RNA was precipitated again. Alkylation sites were then analyzed by primer extension. RNA sequencing was also performed by introducing dideoxy-NTP in the primer extension reaction. Extended products are analyzed on an 8% polyacrylamide gel.
Gel Retardation Assay-32 P-Labeled 2.14 wt RNA (95 nt) was prepared by transcribing the EcoRI-and HindIII-digested pGEM construct that contains a FRGY recognition sequence (YRS) with T7 RNA polymerase. For the standard assay, 10 fmol of [ 32 P]RNA was incubated with FRGY2 protein or its derivatives for 15 min on ice in a 10-l RNA binding buffer described above. The mixture was electrophoresed in an 8% polyacrylamide gel (acrylamide:bis ϭ 60:1) containing 5% glycerol in 1 ⁄2 ϫ TBE buffer at room temperature. The gel was dried and subjected to autoradiography.

RESULTS AND DISCUSSION
Selection of a 6-Nucleotide Consensus Sequence Recognized by FRGY1 and FRGY2-We used the methodology of Keene and colleagues (Tsai et al., 1991) to select specific RNA sequences bound by FRGY1 and FRGY2 from a degenerate pool of transcripts using immunoprecipitation of protein-bound RNA and PCR. Sequence analysis of multiple clones, each representing a distinct RNA species coimmunoprecipitated with FRGY1 or FRGY2, revealed a favored recognition sequence (Fig. 1A). The consensus for FRGY1 and FRGY2 is very similar (Table I) and contains the hexanucleotide 5Ј-AACAUC-3Ј. We next wished to confirm the apparent specificity of RNA binding by FRGY1 and FRGY2 and performed RNA footprinting using one of the isolated RNA sequences (Table I, RNA 2.14 wt), which contains this consensus sequence. Footprinting with dimethyl sulfate of complexes of FRGY1 (Fig. 1B, lanes 2-4) and FRGY2 (Fig. 1B, lanes 9 -11) with the RNA 2.14 wt revealed specific interactions with both Y-box proteins. We conclude that both FRGY1 and FRGY2 have a sequence-specific interaction with the hexanucleotide 5Ј-AACAUC-3Ј, a YRS.
Domains of FRGY2 Required for RNA Binding-We wished to determine which domains of FRGY2 were required for stable interaction with RNA. The FRGY2 protein can be divided into three putative domains ( Fig. 2A); these include an N-terminal region, the CSD, and a C-terminal region containing peptide sequences that are alternately enriched in basic or acidic amino acids (Tafuri and Wolffe, 1990). The CSD contains a known RNA recognition motif, RNP-1 (Landsman, 1992). The CSD is the only highly conserved region of amino acid sequence between FRGY1 and FRGY2 (Tafuri and Wolffe, 1990) and therefore would a priori represent the domain most likely to confer sequence-specific recognition of RNA (Fig. 1). Using a variety of homopolymeric ribonucleotides, Ladomery and Sommerville (1994) determined that both the CSD and regions within the C-terminal tail of FRGY2 could interact with RNA. Murray (1994) suggested that the C-terminal tail was the primary site of interaction with RNA. These experiments made use either of chemically cleaved FRGY2 (Ladomery and Sommerville, 1994) or recombinant polypeptides (Murray, 1994) that contain partial fragments of the FRGY2 protein. We compared the properties of the intact C-terminal domain to the CSD and other deletion mutants of FRGY2 ( Fig. 2A). The RNP-1 motif (-GYGFI-) within the CSD (Fig. 2B) is a probable site of contact between FRGY2 and RNA (Burd and Dreyfuss, 1994a); therefore we introduced a series of point mutations into this domain (Fig. 2B). These mutations were based on previous work showing an essential role for conserved aromatic amino acids in the RNP-1 motif (Brennan and Platt, 1991;Merrill et al., 1988;Mayeda et al., 1994).
Full-length FRGY2 was purified either as a GST-fusion protein or as a T7 gene 10 fusion protein (Fig. 2C) (see "Materials and Methods"). The mutant proteins were purified as GST fusions (Fig. 2C). The C-terminal tail of FRGY2 is an extended hydrophilic structure extremely sensitive to proteolysis (Deschamps et al., 1992). Thus, immunoblotting using polyclonal antibodies against FRGY2 reveals that the other proteins of smaller size than the full-length expressed protein are degradation products (data not shown). Gel retardation experiments (Fig. 2D) using a 95-nucleotide RNA probe containing the specific YRS show the GST-FRGY2 protein and the deletion mutants retaining the CSD together with either the N-terminal (D4) or C-terminal domain (D5) bound efficiently to the probe (Fig. 2D, lanes 3 and 4, lanes 15 and 16, and lanes 9 and 10, respectively). The CSD or the C-terminal tail domain alone (D6) bound less well to the probe (Fig. 2D, lanes 17 and 18 and  lanes 13 and 14, respectively). Point mutations of both aromatic amino acids in the RNP-1 motif (PM1) reduced RNA binding (Fig. 2D, lanes 5 and 6), whereas a single point mutation (PM2) had relatively little effect on RNA binding within the context of the full-length protein (Fig. 2D, lanes 7 and 8). Control experiments (not shown) indicated that the GST moiety has no RNA binding activity in a gel retardation (or filter binding) assay. We conclude that both the CSD and the C-terminal tail can contribute to the association of FRGY2 with RNA. Within the context of the full-length protein, the RNP-1 motif within the CSD has an important role. Both the N-and C-terminal do- FIG. 3. RNA binding curves for FRGY2 and mutants. A filter binding assay (see "Materials and Methods") was performed with GST-FRYG2 (q), PM1 (Ⅺ), D4 (E), D6 ( ), and the CSD (ϫ). A, 32 P-labeled RNA containing the consensus sequence (2.14 wt, AACAUC) was used as the probe. B, one of the point mutants in the consensus sequence (M3/ 4 -1, AACAUG, see Table II) was used as the probe. The data show the fraction of RNA bound as a percentage.
mains of FRGY2 that flank the CSD can stabilize interaction with RNA. We wished to establish a more quantitative assessment of the relative binding affinity of FRGY2 compared with the various deletion and point mutations. In order to minimize the possibility that differential binding is influenced by differential recovery of active protein in the various recombinant protein preparations rather than true differences in binding affinity, we prepared all the proteins in parallel and examined their interaction with the YRS consensus (the 2.14 probe) and a mutant RNA sequence (the M3/4.1 probe, see Table II and Fig.  5) in parallel. The filter binding analysis of protein binding to the YRS consensus (Fig. 3A) reveals that GST-FRGY2 recognizes the 2.14 wt RNA probe with highest affinity (K d ϭ 0.2 ϫ 10 Ϫ7 M). The deletion mutants containing the CSD or tail domain alone (D6), the N-terminal domain including the CSD (D4), or point mutations in the RNP-1 motif all have reduced affinity (K d ϭ 0.8 -1 ϫ 10 Ϫ7 M). The use of a nonspecific RNA sequence (M3/4.1) reduced the binding of GST-FRGY2 (to a K d of 1.0 ϫ 10 Ϫ7 M) (Fig. 3B). We conclude that the filter binding experiments confirm the general indications of affinity derived from gel shift analysis (Fig. 2).
We extended this analysis to the properties of the CSD in isolation. Using the gel retardation assay, we found that point mutations of the RNP-1 motif (Fig. 4A) dramatically reduced stable interaction with RNA probes (Fig. 4B). Quantitative measurements revealed a reduction in K d from ϳ 10 Ϫ7 M for CSD to less than 10 Ϫ6 M in the mutant proteins (not shown). A single point mutation of the RNP-1 motif has a more severe consequence for the CSD in isolation (Fig. 4B, lanes 10 -12) than for the full-length FRGY2 protein (Fig. 2D, lanes 7-8). These differences might be due to the stabilizing influence of the C-terminal tail domain on FRGY2 interaction with RNA. Comparable effects are obtained when FRGY2 binds to DNA .
The association of FRGY2 with DNA is dependent on the length of the DNA probe used to assess binding . In our experiments we use RNA probes of 95 nt in length. Since earlier studies on selective RNA binding have generally made use of short 47-nt probes (Murray, 1994) or homopolymers of ill defined length (Ladomery and Sommerville, 1994), we explored whether RNA binding depends on the length of RNA probe. We find that little or no stable association of FRGY2 occurs with an RNA probe of 30 or 15 nt in length even though the YRS is present, whereas efficient binding occurs with a probe of 95 nt (Fig. 4C). Thus some of the reported discrepancies in nucleic acid binding selectivity may depend on the length of the probe and/or competitor in nucleic acid binding studies.
The CSD Confers Sequence-specific RNA Binding-Our experiments have defined an RNA sequence that is specifically recognized by the FRGY2 protein (Fig. 1, Table I). We have also determined that at least two RNA-binding domains exist within FRGY2 (Fig. 2). One of these is within the CSD (Figs.  2-4), and the other is within the C-terminal tail domain (Fig.  2). We next examined whether either of these domains conferred sequence-specific interactions with RNA. We initially determined what constituted a nonspecific RNA for FRGY2 by a combination of filter binding and gel retardation experiments in which the YRS consensus was systematically mutated (Fig.   FIG. 4. Requirements for RNA binding of FRGY2. A, diagram of GST-fusion FRGY2 deletion mutants containing the CSD. B, gel retardation assay using the 2.14 wt RNA probe (lane 1) was performed with 0. 1 g (lanes 2, 4, 7, 10, and 13), 0.25 g (lanes 3, 5, 8, 11, and 14), and 0.625 g (lanes 6, 9, 12, and 15) of GST-fusion FRGY2 proteins. C, binding of FRGY2 protein to RNA probes of various lengths. Gel retardation assays were performed with 10 fmol of 95 nt (lanes 2-5), 50 fmol of 30 nt (lanes 6 -9), or 50 fmol of 15 nt (lanes 10 -13) RNA probe containing a 2.14 wt binding site. The amount of T7-FRGY2 was 0 g (lanes 2, 6, and 10), 0.04 g (lanes 3, 7, and 11), 0.1 g (lanes 4, 8, and  12), and 0.25 g (lanes 5, 9, and 13). The positions of free probes and RNA-protein complexes are shown by asterisks and dots, respectively. MspI-digested pBR322 was electrophoresed in parallel (lane 1). 5) (see "Materials and Methods"). In these experiments we increased the excess of competitor tRNA in the binding reaction from that in the earlier filter binding experiments (Fig. 3) (see "Materials and Methods"). This was done in order to provide a clearer discrimination in the binding of FRGY2 to different RNA sequences. Our results (shown in Fig. 5 and summarized in Table II) reveal that every ribonucleotide in the YRS has an important role in FRGY2 binding to RNA. In these experiments we make use of T7-FRGY2 (see "Materials and Methods"). Our estimation of binding affinity for the YRS containing RNA probe (Fig. 5, K d ϭ ϳ0.4 ϫ 10 Ϫ7 M) is very similar to that for the GST-FRGY2 protein (Fig. 3A, K d ϭ 0.2 ϫ 10 Ϫ7 M). Thus, independent expression systems yield comparable binding values, indicative of equivalent recovery of active protein. We chose a 100-nt RNA in which the sequence AACAUG (Table II,  Our gel retardation experiments with T7-FRGY2 reveal that two complexes assemble on the 95-nt YRS RNA as the excess of protein over probe is increased (Fig. 6A, lanes 1-5). Since identical bands are seen with nonspecific RNA that does not contain a YRS (Tafuri and Wolffe, 1993a;Tafuri et al., 1993), this may reflect both specific and nonspecific protein-RNA interactions. In vivo the FRGY2 protein interacts with mRNA such that one protein binds every 40 nucleotides (Darnbrough and Ford, 1981). We chose an excess of FRGY2 over the RNA probe that led to the accumulation of a single complex (Fig. 6A, lane 6) and competed this complex with an increasing excess of RNA competitors (Fig. 6A, lanes 7-12). Competition with the 95-nt RNA containing the YRS was much more effective than when this sequence was altered. We next examined whether deletion mutants of FRGY2 containing either the CSD alone, the N terminus including the CSD, or the C-terminal portions of the protein would have specific interactions with RNA. The C-terminal tail domain of FRGY2 binds to the specific RNA probe, but this interaction is  Table II. FIG. 6. FRGY2 protein binds to a 2.14 wt site in a sequence-dependent manner. A, gel retardation assay was performed with 32 P-labeled 2.14 wt RNA as the probe and titration of T7-FRGY2 (lanes 1-5). The amount of FRGY2 was 0 g (lane 1), 0.04 g (lane 2), 0.1 g (lane 3), 0.25 g (lane 4), and 0.625 g (lane 5). Sequence specificity of RNA binding was examined in lanes 6 -12. Ten fmol of RNA probe were incubated with 0.1 g of T7-FRGY2 in the absence (lane 6) or presence of an increasing amount of non-radiolabeled 2.14 wt RNA (lanes 7-9) or M3/M4 -1 RNA (lanes 10 -12). The amount of competitor RNA was 0.6 pmol (lanes 7 and 10), 2 pmol (lanes 8 and 11), and 6 pmol (lanes 9 and 12). B, competition gel retardation assay was performed using the 2.14 wt RNA probe with 0.5 g of GST-Y2 CSD (lanes 1-3), 0.25 g of GST-Y2 D4 (lanes 4 -6), and GST-Y2 D6 (lanes 7-9). Fifteen pmol (lanes 2 and 3) or 10 pmol (lanes 5, 6, 8, and 9) of non-radiolabeled 2.14 wt RNA (lanes 2, 5, and 8) or M3/M4-1 RNA (lanes 3, 6, and 9) were added to the binding reaction as competitors. S is with specific competitor. NS is with nonspecific competitor. competed equivalently by both specific and nonspecific RNA competitors (Fig. 6B, lanes 7-9). In contrast, the CSD alone (Fig. 6B, lanes 1-3) and the N terminus of FRGY2, which includes the CSD (Fig. 6B, lanes 4 -6), bind to the specific RNA probe such that binding is selectively competed by the YRS sequence. This suggests that the CSD contributes to the specific recognition of RNA but requires the flanking N-and Cterminal domains of FRGY2 for a high affinity interaction (see Fig. 3). The CSD is the only conserved amino acid sequence in the N-terminal portion of FRGY1 and FRGY2 (Tafuri and Wolffe, 1990) leading to recognition of the same YRS consensus. Since RNA binding is weakened by mutation of the RNP-1 motif (Fig. 2D, lanes 5 and 6, Fig. 3), it appears that the RNP-1 motif may have a role in maintaining the structural integrity of the CSD necessary for discrimination between various RNA sequences.
Biological Implications-The Y-box proteins have the capacity to interact with RNA with sequence specificity (Figs. 1, 3, 5, and 6 and Tables I and II). The highly conserved cold shock domain is required for this sequence-specific recognition of RNA (Figs. 2-6). This domain differs in only four out of seventy-eight amino acids between FRGY1 and FRGY2, whereas these proteins are highly divergent at the N terminus and display an overall identity of 65% (Tafuri and Wolffe, 1990). Importantly, other metazoan Y-box proteins all retain the same cold shock domain sequences as FRGY1 or FRGY2 and will thus have the capacity to interact specifically with RNA. This might influence their capacity to regulate both transcriptional and translational events (Ranjan et al., 1993;Bouvet and Wolffe, 1994).
In the oocyte, FRGY2 and a highly related protein mRNP3 package mRNA (Deschamps et al., 1992;Murray et al., 1992;Wolffe, 1993a, 1993b). These proteins interact with mRNA with a stoichiometry such that one protein binds approximately every 40 nucleotides (Darnbrough and Ford, 1981;Marello et al., 1992). This repetitious binding requires that the FRGY2 protein binds to a wide variety of RNA sequences, and thus FRGY2 must also retain the capacity to package mRNA with relatively little sequence specificity. By analogy with the packaging of the vast majority of nuclear DNA with the histone proteins (Wolffe, 1992), the Y-box proteins might package the vast majority of cytoplasmic mRNA Wolffe, 1993a, 1993b;Wolffe, 1994a). Just as the histone proteins provide a default mechanism that directs the repression of basal transcription (Grunstein, 1990;Wolffe, 1992) so might the Y-box proteins provide a default mechanism for the repression of translation Wolffe, 1993a, 1993b;Ranjan et al., 1993;Bouvet and Wolffe, 1994). However, just as the histones can have highly selective interactions with DNA in spite of their general packaging function and thereby influence the events of transcription (Wolffe, 1994b), we have found that the Y-box proteins can also bind to RNA with sequence selectivity in spite of their general packaging function (Figs. 1, 3, 5, and 6). Thus the exact sequence of mRNA will influence the context and affinity with which the Y-box proteins bind and thus the association of other proteins necessary for the translational activation or silencing of maternal mRNA. Moreover, the selective association of Y-box proteins with particular RNA sequences might nucleate the assembly of masking proteins on mRNA. Future experiments will explore these possibilities.
FRGY1 is found in the nuclei of somatic cells (Tafuri and Wolffe, 1990;Ranjan et al., 1993). The function of the somatic Y-box proteins is unknown. They have been proposed to also influence the translation of mRNA (Evdokimova et al., 1995); however, other experiments suggest a role in transcriptional control Wolffe, 1990, 1992;Ting et al., 1994;Kashanchi et al., 1994;Kerr et al., 1994;MacDonald et al., 1995). Our definition of the Y-box proteins as capable of sequenceselective RNA recognition provides a potential explanation for their selective influence on the transcription process independent of specific association with DNA. It is possible that the Y-box proteins might interact with specific nascent RNAs and communicate with the basal transcriptional machinery in a manner analogous with that of TAT-TAR (Berkhout et al., 1989). Although RNA binding by the prokaryotic Y-box proteins has not yet been examined, it is possible that sequenceselective RNA binding might contribute to regulation of the cold shock response (Goldstein et al., 1990). A requirement to maintain the capacity to interact with a variety of specific nucleic acids might contribute to the striking evolutionary conservation of the cold shock domain itself.