Vigilin, a ubiquitous protein with 14 K homology domains, is the estrogen-inducible vitellogenin mRNA 3'-untranslated region-binding protein.

RNA-binding proteins containing KH domains are widely distributed. One KH domain protein of unknown function, vigilin (also known as the high density lipoprotein-binding protein), contains 14 KH domains and is ubiquitous in vertebrate cells. We previously used RNA gel mobility shift assays to describe an estrogen-inducible protein which binds specifically to a segment of the 3'-untranslated region (3'-UTR) of vitellogenin mRNA, an area which has been implicated in the estrogen-mediated stabilization of vitellogenin mRNA. Here we show that the vitellogenin mRNA-binding protein (VitRNABP) is vigilin. The VitRNABP was isolated as a 150-155-kDa protein on a vitellogenin mRNA 3'-UTR affinity column. Peptide microsequencing revealed that the purified protein was vigilin, a conclusion confirmed in Western blot analysis with antibodies to vigilin. Direct confirmation that vigilin is the VitRNABP was obtained from RNA gel mobility shift assays which demonstrated that antibodies to chicken vigilin supershifted the Xenopus VitRNABP band. Xenopus liver vigilin mRNA and the VitRNABP exhibited similar induction by estrogen, providing additional confirmation that vigilin is the estrogen-inducible protein which binds to the 3'-UTR of estrogen-stabilized vitellogenin mRNA. These data support a role for vigilin in the hormonal control of mRNA metabolism.

such biologically significant KH domain-containing proteins as the FMR protein, which is involved in Fragile X mental retardation (3), the Drosophila bicaudal C protein which is important in development (4), and the ␣-poly(C)-binding protein which is found in the ␣-globin mRNA ribonucleoprotein complex (5). One notable but little understood KH domain protein is vigilin (also identified as the human high density lipoproteinbinding protein, HDL-BP), a ubiquitous, highly conserved protein with 14 KH domains (6,7). Since vigilin has been found in all cell lines and tissues examined (8), appears to be regulated by diverse factors (7)(8)(9)(10)(11), and its 14 KH domains represent the largest number of KH domains in any known protein (3,12), it likely plays an important role in RNA metabolism. While vigilin has been used as a model protein in solving the structure of a KH domain uncomplexed to RNA (12), its RNA binding properties have been elusive, and its function(s) have remained obscure.
We have been studying a protein which binds to a segment of the 3Ј-untranslated region (3Ј-UTR) of the mRNA encoding the egg yolk precursor protein, vitellogenin. In male Xenopus liver, vitellogenin mRNA levels increase Ͼ10,000-fold following estrogen administration (13,14). The estrogen-mediated induction of vitellogenin mRNA is brought about both by an increase in the rate of vitellogenin gene transcription and by stabilization of cytoplasmic vitellogenin mRNA (13,14). Hepatic vitellogenin mRNA is degraded with a half-life of 16 h in the absence of estrogen, and 500 h following addition of estrogen to the culture medium (14). The estrogen-mediated stabilization of vitellogenin mRNA requires association of the mRNA with ribosomes (15) and involves the 3Ј-UTR of the mRNA (16). We identified a protein which binds specifically to a segment of the 3Ј-UTR of vitellogenin mRNA in an estrogen-inducible manner and named it the vitellogenin mRNA 3Ј-UTR-binding protein (VitRNABP) (17,18). Here we describe the isolation of the VitRNABP and demonstrate by several criteria that this protein is Xenopus vigilin.

EXPERIMENTAL PROCEDURES
Protein Isolation, Sequencing, and FPLC-A protein mixture from salt-extracted polysomes from livers of estrogen-treated male Xenopus laevis was prepared as we have described (17). The pB1-15pA used in affinity chromatography was made by cloning the HindIII-DraI fragment of pB1-15 (17) into the HindIII-SmaI site of pSP64pA (Promega). B1-15pA RNA was made by in vitro transcription from the SP6 promoter of the pB1-15pA vector linearized with EcoRI yielding a 146nucleotide RNA. Poly(U)-agarose beads (Pharmacia Biotech Inc.) were prepared by rinsing (i) once in 4 volumes of water, (ii) once in wash buffer (25 mM HEPES, pH 7.4, 10 mM EDTA, 1 mM EGTA, 0.05 M NaCl), and (iii) six times in binding buffer (25 mM HEPES, pH 7.4, 10 mM EDTA, 1 mM EGTA, 1 M LiCl). pB1-15pA RNA (approximately 4 g of RNA/l of beads) was added to the beads in 2 volumes of binding buffer and allowed to hybridize for 5-6 h at room temperature. The beads were kept in suspension by rotation. The beads were washed twice in hybridization buffer (HB) (6 mM Tris, pH 7.6, 6% glycerol, 1 mM EDTA, 0.01 mM EGTA, 0.25 mM magnesium acetate) with 50 mM NaCl and once in HB with 50 mM NaCl containing 1 g/l yeast tRNA, 1 g/l heparin, 0.3 unit/l RNasin, and protease inhibitors (17). The beads were then incubated with the polysome extract in HB with tRNA, heparin, RNasin, and protease inhibitors with 200 g of polysome extract per 20 l of beads in a 300-l total volume for 30 min at room temperature. The beads were then washed in buffer containing 200 mM KCl, 2.5 mM magnesium acetate, 10 mM Tris, pH 7.6, 10% glycerol, 0.1 mM EGTA, 1 mM EDTA, 1 mM dithiothreitol, and protease inhibitors. Proteins were eluted in the above buffer containing 500 mM KCl. Eluted proteins were resolved on a 7.5% SDS-PAGE gel (19) and visualized using a silver stain (20). For microsequencing, resolved proteins were transferred to a ProBlott membrane (Applied Biosystems) in 10 mM CAPS, pH 10, with 10% methanol at 50 V for 3 h and stained in Amido Black (21). The 150-kDa protein was cut from the membrane and sent to the Rockefeller University Technology Center (New York, NY) for digestion with endoproteinase Lys-C and internal sequencing (21).
For FPLC analysis, protein extracts were concentrated using Microcon 30 or 50 filters to about 0.25 original volume. Protein (1.2 mg) was loaded onto a Superdex 200 HR 10/30 column (Pharmacia) and resolved using a Pharmacia FPLC System at a flow rate of 0.5 ml/min at 4°C. A total of 25 ml of buffer (10 mM Tris, pH 7.6, 200 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 5 g/ml phenylmethylsulfonyl fluoride) was used to elute proteins in 0.5-ml fractions. Proteins eluting from the column were monitored at 280 nm. Fractions were assayed using the RNA gel mobility shift assay, and band intensities were determined using a PhosphorImager (Molecular Dynamics) or a densitometer.
RNA Isolation and Northern Blot Analysis-Total RNA was isolated from livers of control or estrogen-treated X. laevis as described (22). RNA was resolved on a 0.75% formaldehyde-agarose gel, transferred to a Nytran membrane (Schleicher and Schuell) and hybridized to cDNA probes as described (23). cDNA was radiolabeled with [ 32 P]dCTP by nick translation or random priming using standard methods (20). Ribosomal RNA band intensity was used to control for loading of samples. Intensities of labeled bands were determined using a PhosphorImager.
cDNA Library Screening-Oligonucleotides for PCR were made to two conserved regions in the 5Ј portion of vigilin corresponding to nucleotide numbers 213-243 and 315-340 in the chicken vigilin sequence (GenBank TM accession number X65292) (6). Xenopus liver RNA (10 g) was PCR-amplified using the above oligonucleotides (24), gelpurified, and radiolabeled by nick translation. The labeled 128-nucleotide cDNA was used to screen a X. laevis liver cDNA library (25). A positive clone was identified, plaque-purified, and excised, and its 980 nucleotide insert was subcloned into pGEM3 (Promega). This clone is referred to as pVIG10.
Gel Mobility Shift Assay, Western Analysis, and Antibodies-Gel mobility shift assays were performed as we have described (17) with the following modifications. Antiserum (42 g) was added to the hybridization mixture in the absence of RNA probe for 45 min at 4°C. RNA probe was then added, and the incubation was continued for an additional 45 min before gel analysis. Rabbit antibody to a segment of recombinant vigilin was a kind gift from Drs. S. Kü gler and P. K. Mü ller (26). Rabbit anti-BSA was a gift from S. Miklasz. Western blot analysis was done as described previously (27), except that sodium azide was not added to buffers, SDS (0.1%) was added to the transfer buffer, and blots were transferred at 60 mA (14 V) overnight at 4°C. Anti-vigilin antiserum was diluted 1:1500 and horseradish peroxidase-conjugated goat antirabbit second antibody (EY Laboratories) was used at a 1:2000 dilution. Bands were visualized with an ECL kit (Amersham).

RESULTS
Isolation of the Vitellogenin 3Ј-UTR RNA-binding Protein-We described the VitRNABP as an estrogen-inducible protein which bound to a segment of the vitellogenin mRNA 3Ј-UTR in RNA gel mobility shift assays and in UV crosslinking experiments (17). We used RNA affinity chromatography to isolate the VitRNABP. A 94-nucleotide segment of the vitellogenin mRNA 3Ј-UTR with a 30-nucleotide poly(A) tail, which is efficiently bound by the VitRNABP, was immobilized on poly(U)-agarose beads by hybridization. A salt extract of polysomes prepared from the livers of 14-day estrogen-treated male X. laevis was incubated with the RNA affinity column, and control extracts were incubated with the poly(U)-agarose. Bound proteins were eluted in 500 mM salt and resolved on an SDS-PAGE gel. As shown in Fig. 1A, a predominant protein band of approximately 150 -155 kDa, which appeared as a very closely spaced doublet, was eluted from the agarose beads containing the vitellogenin mRNA segment (Fig. 1A, lane 2), but not from control poly(U)-agarose beads (Fig. 1A, lane 3). RNA binding activity and the presence of the 150 -155-kDa band were well correlated since the eluate from the B1-15pA RNA beads, but not the eluate from the poly(U) beads, bound RNA in the RNA gel mobility shift assay (data not shown).
We next determined whether the vitellogenin mRNA 3Ј-UTR binding activity was in the same 150 -155-kDa molecular mass range as the protein purified by the RNA affinity column. Since we have been unable to reconstitute mRNA binding after denaturing the protein, we size-fractionated a concentrated polysome extract under native conditions. Proteins eluted from the FPLC sizing column were assayed using the RNA gel mobility shift assay. As shown in Fig. 2, peak binding activity corresponded to a molecular mass range of 145-190 kDa, which was consistent with the 150 -155-kDa protein being the VitRNABP.
Identification of the 150 -155-kDa Band as Vigilin by Microsequencing and Western Blot Analysis-The 150 -155-kDa protein was enriched by RNA affinity chromatography (Fig. 1A), size-fractionated on an SDS-polyacrylamide gel, transferred to a membrane, and submitted for microsequencing. Internal sequencing of one of the peptide fragments yielded the sequence; N(R/V)IRIEQDPQXVQQA. This sequence was used to search a protein sequence data base (Wisconsin Sequence Analysis Package of the Genetics Computer Group). Twelve out of fifteen amino acids were identical to those in chicken vigilin (amino acids 478 -492) (6) and the human HDL-BP (amino acids 479 -493) (7). Since vigilin and HDL-BP are the independently identified avian and human forms of the same protein, we subsequently refer to this protein as vigilin. To verify that the protein we purified as the VitRNABP was vigilin, a second peptide was sequenced. The sequence of this peptide, VI(S/ T)QIR had a 5-out of 6-amino acid match with both the chicken FIG. 1. A 150 -155-kDa protein is specifically eluted from a vitellogenin mRNA 3-UTR affinity column. The polysome extract (SEP), the eluate from the RNA affinity column (B1-15pA), and the control poly(U)-agarose (POLY-U) were fractionated by SDS-PAGE gels and either silver-stained (A) or subject to Western blot analysis using antiserum to chicken vigilin (B) as described under ''Experimental Procedures.'' The 150 -155-kDa protein band is indicated by an arrow. The relative mobility of proteins used as molecular mass standards is shown.
Vigilin Binds to the Vitellogenin mRNA 3Ј-UTR 12250 and human sequences, confirming that the protein sequenced was vigilin.
To ensure that the isolated protein was Xenopus vigilin, a Western blot of the protein from the RNA affinity beads was probed with antibodies to a segment of recombinant chicken vigilin, expressed in Escherichia coli and affinity-purified (26). In Western blots using crude polysome extracts, the antibody reacts with a single protein (Fig. 1B, lane 1) which is the same size as the protein shown to be vigilin by microsequencing. The protein eluted from the B1-15pA affinity beads is also specifically recognized by the antibody (Fig. 1B, lane 2) and on shorter exposures can be seen to be a doublet (data not shown). Although this is a polyclonal antibody, in agreement with previous reports (26), we see no cross-reactivity with other proteins. Since the elution pattern was identical in both the SDS-PAGE gels and Western blots, we were confident that the isolated protein was indeed vigilin.
Antiserum to Chicken Vigilin Binds to the VitRNABP-While these data demonstrated that the protein purified by affinity chromatography was vigilin, and that vigilin was approximately the same size as the VitRNABP, it was still important to directly establish that the protein bound to the vitellogenin mRNA 3Ј-UTR in our gel mobility shift assays was vigilin. If the VitRNABP was indeed Xenopus vigilin, antibodies to vigilin added to the RNA gel mobility shift assays should alter the mobility of the VitRNABP⅐RNA complex. In an RNA gel mobility shift assay, the VitRNABP in a polysome extract formed a specific complex with a radiolabeled probe containing a segment of the vitellogenin mRNA 3Ј-UTR (Fig. 3A, lane 2). Addition of antibody to vigilin supershifted this complex (Fig. 3A,  lane 4). Since this is a polyclonal antibody, the number of antibody molecules that can bind to each gel-shifted complex is variable, and some tailing of the supershifted band is observed (Fig. 3A, lane 4 and Fig. 3B, lane 2). This supershift was specific for anti-vigilin, since it did not occur when a control antibody to bovine serum albumin was used (Fig. 3A, lane 6). Control experiments also showed that the antibodies alone did not interact with the RNA probe (Fig. 3A, lanes 3 and 5). By increasing the ratio of anti-vigilin antibody to polysome extract, we were able to almost completely supershift the VitRNABP⅐RNA complex (Fig. 3B), indicating that vigilin is present in all of the complexes with the vitellogenin mRNA 3Ј-UTR segment.
Vigilin mRNA Is Induced by Estrogen in Xenopus Liver-Since we had previously reported that the VitRNABP was estrogen-inducible (17), we wanted to determine whether Xenopus vigilin mRNA was also regulated by estrogen. We therefore isolated and sequenced a 980-nucleotide Xenopus vigilin cDNA clone (see "Experimental Procedures") with a high homology to sequences at the 5Ј-end of chicken and human vigilin. The Xenopus vigilin cDNA showed a 74% and 76% nucleotide identity with the chicken and human vigilin cDNAs, respectively, and an 86% identity to amino acid sequences from both species.
To compare the regulation of Xenopus vigilin mRNA levels and vitellogenin mRNA 3Ј-UTR binding activity, three animals each were treated with vehicle or estrogen, total RNA was isolated from half the liver, and whole cell protein extracts were prepared from the other half. Consistent with our earlier work (17,18), binding activity in RNA gel shift assays from liver extracts from estrogen-induced Xenopus liver was about 3-fold higher than in control extracts (Fig. 4B). In agreement with the RNA gel shift data, Northern blot analysis showed that a predominant 5.7-kilobase vigilin mRNA was induced 2-3-fold after estrogen administration (Fig. 4, A and B). DISCUSSION Its broad distribution and the presence of 14 KH domains suggested an important, but unidentified, role for vigilin in RNA metabolism. In this work we show that vigilin is the estrogen-inducible protein we previously demonstrated binds to a segment of the vitellogenin mRNA 3Ј-UTR. This conclusion is supported by our findings that: vigilin is retained on a vitellogenin mRNA 3Ј-UTR affinity column; in gel shift assays, vigilin antiserum binds to the VitRNABP; the mRNA binding Vigilin has been proposed to be an RNA-binding protein because of its numerous KH domains (3) but to date had only been shown to bind tRNA (28). Our work demonstrates for the first time that vigilin can bind to an mRNA. We previously reported that the affinity for the vitellogenin mRNA 3Ј-UTR was orders of magnitude greater than that for tRNA (17). Since vigilin is present in a wide variety of tissues and cell types and is regulated by diverse factors, it almost certainly binds to RNAs other than vitellogenin mRNA. The relative binding affinity of vigilin for different RNA sequences may play a role in its function in different tissues. The exact determinants of RNA binding specificity remain to be determined and may involve specific KH domains within vigilin and proteins which associate with vigilin. The nucleic acid binding properties of individual KH domains vary (29). The 14 KH domains in vigilin may therefore interact on different RNA templates to create different binding surfaces, allowing vigilin to bind to a diverse array of RNAs. It is possible that other proteins may associate with the vigilin⅐RNA complex and play a role in creating binding specificity. Although other proteins which might play a role in RNA binding are not yet identified, there are several minor proteins in the vitellogenin mRNA affinity column eluent which may contribute to this function.
The ability of vigilin to bind the vitellogenin mRNA 3Ј-UTR suggests that it may be involved in mRNA stabilization. Since the 3Ј-UTR of vitellogenin mRNA is critical for estrogen-mediated stabilization (16), and vigilin is the major protein which binds to this region (17), we have proposed that vigilin is involved in the stabilization of the message. Other KH domaincontaining proteins have been shown to be involved in RNA stabilization (5), and vigilin has been reported to protect tRNA from RNase (28). Other functions have been proposed for vigilin based on its intracellular distribution (26). Additional studies will be required to determine the role of vigilin in mRNA stabilization and any additional functions it may serve.
While a clear role for vigilin has not been established, its tight regulation accentuates its importance in RNA metabolism. We have shown that vigilin levels are increased by estrogen in liver ( Fig. 4 and Refs. 17 and 18) and by testosterone in muscle, and that it is decreased by testosterone in testis (18).
Our findings are consistent with reports in several experimental models that vigilin can be regulated by several factors and that increased levels of vigilin correlate with increased protein production (8,9,11,30). Vigilin has been localized immunocytochemically to polysomes (8), and we have extracted vigilin from liver polysomes indicating its association with cytoplasmic RNA. In a situation such as the estrogen-induced liver in Xenopus, where vitellogenin is 50% of the total cellular mRNA (31), a substantial fraction of intracellular vigilin may be bound to the relatively high affinity binding site in the vitellogenin mRNA 3Ј-UTR. Our findings are also in agreement with the ubiquitous distribution of vigilin, since we have found vigilin in all Xenopus tissues we have examined and in several cell lines (18).
In conclusion, we have identified vigilin, a widely distributed protein containing 14 RNA binding domains, as the estrogenregulated protein which binds to the 3Ј-UTR of vitellogenin mRNA. Our data provide the first evidence that vigilin can bind a specific mRNA and suggest a role for vigilin in mRNA metabolism.

FIG. 4. Vigilin mRNA is inducible by estrogen in Xenopus liver.
A, Northern blot analysis of total RNA from livers of Xenopus treated with vehicle (Ϫ) or with estradiol (ϩ) and probed with vigilin cDNA. B, comparison of the induction of vigilin mRNA and vitellogenin mRNA 3Ј-UTR binding activity from livers of Xenopus treated with vehicle (Ϫ) or estrogen (ϩ). RNA levels (hatched bars) were determined by Northern analysis as described under ''Experimental Procedures.'' Relative binding activity of vigilin in liver extracts (solid bars) were determined from RNA gel mobility shift assays. Gel-shifted bands were quantified using a PhosphorImager. The data represent the average from three animals Ϯ S.E.