Assembly of the Yin Yang 1 Transcription Factor into Messenger Ribonucleoprotein Particles Requires Direct RNA Binding Activity*

The early stages of vertebrate development depend heavily on control of maternally transcribed mRNAs that are stored for long periods in complexes termed messenger ribonucleoprotein particles (mRNPs) and utilized selectively following maturation and fertilization. The transcription factor Yin Yang 1 (YY1) is associated with cytoplasmic mRNPs in vertebrate oocytes; however, the mechanism by which any of the mRNP proteins associate with mRNA in the oocyte is unknown. Here we demonstrate the mechanism by which YY1 associates with mRNPs depends on its direct RNA binding activity. High affinity binding for U-rich single-stranded RNA and A:U RNA duplexes was observed in the nanomolar range, similar to the affinity for the cognate double-stranded DNA-binding element. Similar RNA binding affinity was observed with endogenous YY1 isolated from native mRNP complexes. In vivo expression experiments reveal epitope-tagged YY1 assembled into high molecular mass mRNPs, and assembly was blocked by microinjection of high affinity RNA substrate competitor. These findings present the first clues to how mRNPs assemble during early development.

The early stages of vertebrate development depend heavily on control of maternally transcribed mRNAs that are stored for long periods in complexes termed messenger ribonucleoprotein particles (mRNPs) and utilized selectively following maturation and fertilization. The transcription factor Yin Yang 1 (YY1) is associated with cytoplasmic mRNPs in vertebrate oocytes; however, the mechanism by which any of the mRNP proteins associate with mRNA in the oocyte is unknown. Here we demonstrate the mechanism by which YY1 associates with mRNPs depends on its direct RNA binding activity. High affinity binding for U-rich single-stranded RNA and A:U RNA duplexes was observed in the nanomolar range, similar to the affinity for the cognate double-stranded DNA-binding element. Similar RNA binding affinity was observed with endogenous YY1 isolated from native mRNP complexes. In vivo expression experiments reveal epitope-tagged YY1 assembled into high molecular mass mRNPs, and assembly was blocked by microinjection of high affinity RNA substrate competitor. These findings present the first clues to how mRNPs assemble during early development.
Yin Yang 1 (YY1) 3 is a highly conserved transcription factor of the GLI-Kruppel family that can function as an activator, repressor, or initiator of transcription at a number of cellular and viral promoters (1)(2)(3). The activity of YY1 is modified by numerous interactions with other proteins, including c-Myc (4), Sp1 (5,6), the polycomb group protein EED (7), cAMPresponse element-binding protein (8), the viral protein E1A (9), retinoblastoma protein (10), the GATA1 transcription factor (11), and the specific activator YY1AP (12). YY1 is subject to post-translational modifications such as glycosylation (13), acetylation (14), and phosphorylation (15). It contains four C2H2-type zinc fingers near the C terminus responsible for YY1 DNA binding activity (16), a bipartite activation domain near the N terminus (16), and a transcriptional repression domain near the C terminus (17). The Xenopus laevis homologue is a 43-kDa, 373-amino acid protein sharing a high degree of sequence and structural conservation with mammalian YY1 (18). The second zinc finger knuckle of YY1 is highly homologous with the RNA-binding zinc finger knuckles of the doublestranded RNA-binding protein TFIIIA, particularly in the spacing of key cysteine and histidine residues (19).
YY1 clearly plays an important role in early embryonic development; however, there is compelling evidence that it functions through mechanisms other than transcriptional regulation. YY1 is localized entirely to the cytoplasm of mouse oocytes and has a mosaic pattern of nucleocytoplasmic distribution in cells of early embryos (20). Homozygous deletion of YY1 in mice causes peri-implantation lethality, and heterozygotes display severe neurulation defects (20). Studies in Xenopus show YY1 misexpression affects survival, neurulation, and patterning (7,21,22). Biochemical analysis has shown that YY1 is entirely restricted to the cytoplasm during early development, and an absence of YY1 transcriptional regulatory activity has been observed with promoter constructs containing YY1 consensus elements (19,23). In oocytes and early embryos, YY1 is a component of cytoplasmic messenger ribonucleoprotein particles (mRNPs) (19,23). Several groups have reported cytoplasmic localization of YY1 in mammalian cultured cells and differentiated cells in vivo (24 -26).
The control of early vertebrate development depends heavily on mRNAs stored in the oocyte cytoplasm in ribonucleoprotein complexes termed mRNPs (27). The Xenopus oocyte has been the primary model system for studies of mRNPs. During oogenesis, mRNPs become highly concentrated in the cytoplasm and function in the storage and stabilization of maternally transcribed mRNAs required for subsequent embryonic development (27)(28)(29)(30). The heterogeneous population of mRNAs associated with cytoplasmic mRNPs originate from a subset of maternally transcribed, intronless genes (31). Despite a relatively extensive body of experimental work from decades past, the structure and function of oocyte mRNPs are rather poorly understood. Common sequence features among these mRNAs that could possibly serve as sites of recruitment of mRNP proteins have not emerged (31). Thus, mRNPs constitute a highly heterogeneous population of complexes, which have thus far resisted traditional biochemical and genetic characterization (27,30,32).
The association of YY1 and other mRNP proteins with maternal transcripts could potentially be mediated by proteinprotein interactions, through direct interaction with mRNA molecules, or through a combination of protein-nucleic acid interactions. Given that YY1 possesses a potential RNA-binding zinc finger motif (19), we examined the ability of YY1 to bind directly to RNA. Here we report that both native mRNPassociated YY1 isolated from oocytes and recombinant bacterially expressed YY1 possess high affinity RNA binding activity. In vitro binding assays conducted with RNA probes exhibiting a range of sequence and structural features reveal that YY1 binds with highest affinity to U-rich single-stranded RNA and to A:U RNA duplexes. Nascent YY1 expressed in vivo assembled into high molecular mass mRNPs recoverable by oligo(dT) chromatography. Microinjection of high affinity RNA substrates effectively blocked the assembly of YY1 into mRNPs in vivo, demonstrating that association with mRNPs is dependent on RNA binding activity. We conclude the mechanism of YY1 assembly into mRNPs requires direct association with maternal mRNA.

EXPERIMENTAL PROCEDURES
Preparation of Recombinant Protein-The coding sequence of X. laevis YY1 (GenBank TM accession number NM_001087615) was inserted in-frame, downstream of the His 6 tag of pRsetB (Invitrogen), to produce pRsetBYY1. For bacterial expression, pRsetBYY1 was transformed into BL21(DE3) pLysS Escherichia coli (Invitrogen) and mid-log phase cultures induced with 2 mM isopropyl 1-thio-␤-D-galactopyranoside for 4 h at 37°C. Bacterial lysates were prepared, and recombinant protein was isolated by chromatography on nickel-charged NTA-agarose columns according to the manufacturer's directions (Bio-Rad). Isolated protein was renatured by buffer exchange via spin dialysis, using Macrosep 10K Omega tubes (Pall Corp., East Hills, NY) into dialysis buffer (250 mM NaCl, 50 mM NaH 2 PO 4 /Na 2 HPO 4 (pH 8.0), 25 mM EDTA pH 8.0, and 1 mM DTT) and then exchanged into renaturation buffer (50 mM KCl, 50 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 , 0.2 mM ZnCl 2 , 1 mM DTT, and 20% v/v glycerol). Some aliquots of protein were instead exchanged into renaturation buffer lacking magnesium or zinc and containing 0.1 mM EDTA. Protein concentrations were determined by the method of Bradford (44).
Oligo(dT)-cellulose and Size Exclusion Chromatography-Oligo(dT) isolation of mRNPs was performed essentially as described previously (19). Lysates (1 ml) were applied to 100 mg of pre-equilibrated type VII oligo(dT)-cellulose (New England Biolabs) and incubated at 4°C for 2 h with gentle agitation. Following incubation matrix was washed twice with 5 volumes ice-cold binding buffer and treated as follows. For preparation of native mRNPs, matrix was eluted twice with 2 ml of elution buffer (300 mM NaCl, 50 mM Tris-HCl (pH 7.6), 5 mM EDTA, 30% (v/v) deionized formamide). Eluted mRNPs were exchanged into renaturation buffer (50 mM KCl, 50 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 , 0.2 mM ZnCl 2 , and 1 mM DTT) and concentrated to a volume of 0.5 ml by spin dialysis as described above, yielding an equivalent of one oocyte per 10 l. For analysis of HA-YY1expressing oocytes, matrix was eluted directly in 200 l of 1ϫ SDS-PAGE loading buffer. For analysis of recombinant YY1, binding reactions were carried out in the presence of 100 mg of oligo(dT), with 0.5 g of poly(A ϩ ) oocyte mRNA and 10 pmol of rYY1 and either 1 g of RNase A or 20 pmol of rU or rC as indicated. Size exclusion chromatography was performed as described previously (19).
Preparation of Radiolabeled Nucleotide Probes-Oligonucleotide probes (supplemental sequences) were radiolabeled and purified as follows. Double-stranded DNA probes were end-fill labeled using [␣-32 P]dCTP (PerkinElmer Life Sciences) and MLuV reverse transcriptase. Single-stranded DNA and RNA probes were end-labeled using [␥-32 P]ATP (PerkinElmer Life Sciences) and T4 polynucleotide kinase. Double-stranded RNA probes were prepared by labeling reciprocal strands and then purified and annealed to the complementary unlabeled strand. Double-stranded RNA probe integrity was further verified by insensitivity to single-stranded ribonucleases (data not shown). Labeled probes were purified over Sephadex-G-50 (Sigma), extracted with phenol/chloroform (1:1) and chloroform/isoamyl alcohol (19:1), ethanol-precipitated, resuspended in TEN buffer (100 mM NaCl, 20 mM Tris-HCl (pH 7.6), 2 mM EDTA), and spin-dialyzed with TEN buffer five times using Microsep 3K Omega spin tubes (Pall Corp., Waltham, Ontario, Canada). The state of purity was monitored by electrophoresis on 20% polyacrylamide, 0.5ϫ TBE gels containing 8.3 M urea. Once no trace of unincorporated label could be detected by 12 h of radiography using Kodak XB-1 film, oligonucleotides were quantified by UV spectroscopy on a Bio-Rad SmartSpec using millimolar extinction coefficients specific to the oligonucleo-tide being measured. Only samples with an A 260 /A 280 ratio greater than 1.8 were used.
Nitrocellulose Filter Binding Assays-Nitrocellulose filter binding assays were performed according to Romaniuk (45) with slight modifications. Binding reactions were carried out with 0.5 pmol probe at a concentration of 10 nM. After 20 min of incubation at room temperature, reactions were diluted with 50 l of Filter Binding Buffer (50 mM NaCl, 50 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 ) and applied immediately to pre-equilibrated 0.45-m nitrocellulose membranes (Bio-Rad). Filters were washed twice with 100 l of Filter Binding Buffer and then dried for 90 min at 80°C in vacuo. Filters were dissolved in 500 l of 0.5 M benzethonium hydroxide in methanol and mixed with15 ml of scintillation fluid, and radioactivity was measured in a Tri-Carb 3100TR scintillation counter (Packard Biosciences, Meriden, CT). Presented data are the average of three replicate experiments.
Determination of Minimal Binding Site-RNA fragments were generated by digestion of 5 pmol of end-labeled U 20 probe with 1 ng of RNase A in 200 l of TMN (100 mM NaCl, 20 mM Tris-HCl (pH 7.6), 2 mM MgCl 2 ) for 20 min at room temperature and purified as described above for RNA probes with the exclusion of spin dialysis. Binding reactions were assembled by resuspending probe in 200 l of binding buffer (50 mM NaCl, 50 mM NaH 2 PO 4 /Na 2 HPO 4 (pH 7.5), 2 mM MgCl 2 , 0.5 mM 2-mercaptoethanol, 1 unit/ml RNasin), and addition of 50 l of nickel-charged NTA-agarose beads and 0.2 nmol of YY1. Reactions were incubated 30 min at room temperature, and beads were then recovered by centrifugation (1000 ϫ g, 1 min) and washed (2 ϫ 1 ml) with binding buffer. Bound probe was eluted with 50 l of 80% (v/v) formamide and electrophoresed on 20% DNA sequencing gels followed by autoradiography.

RESULTS
Analysis of Recombinant YY1-Following bacterial expression and purification of His 6 -tagged Xenopus YY1 (Fig. 1A), activity of the recombinant protein was verified by DNA-binding assays using the consensus promoter element. YY1 required the presence of both zinc and magnesium ions for DNA binding (Fig. 1B). Formation of YY1-DNA complexes was competed with both oligonucleotide and antibody competitors (Fig. 1B), in a manner similar to that of endogenous YY1 (23). Recombi-  DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 37915 nant YY1 bound with high affinity to consensus DNA (Fig. 1C) as observed with previous studies using mammalian YY1 (46,47). No binding was observed with single-stranded forward and reverse strands of either consensus or mutant DNA probes (data not shown). Therefore, YY1 did not possess affinity for single-stranded DNA.

YY1-mRNP Assembly Requires RNA Binding
Nitrocellulose filter binding reactions were used for comparative determination of dissociation constants of various probes and to quantitatively verify results of the EMSA analysis. The filterbinding assay measures the proportion of the total radioactivity bound, which is converted to a molar quantity of bound probe as a function of the specific activity. High affinity of YY1 for consensus DNA probe was verified by filter binding assay (K d ϭ 3.2 Ϯ 0.1 nM, see Table 1). Moderate to low affinity binding to mutant probe (K d ϭ 54.9 Ϯ 1.2 nM) and no detectable binding to heat shock element were observed ( Fig. 1C and Table 1). YY1 did not exhibit free nucleotide binding activity to adenosine, uridine, or deoxycytidine triphosphates (data not shown).
Examination of YY1 RNA Binding Activity-We examined the possibility that assembly of YY1 into mRNPs may involve direct RNA binding activity. Initially we tested RNA sequences complementary to the consensus promoter element, both single-stranded (forward and reverse) and duplex consensus as well as mutant probes. YY1 binding to single-stranded consensus as well as mutant RNA probes was barely detectable, and YY1 bound very weakly to both consensus and mutant duplex RNA probes (Fig. 2 and Table 1).
The next experiments measured YY1 affinity for short (20 nucleotides) synthetic oligoribonucleotides to examine sequence and structural specificity. EMSA and filter-binding assays revealed high affinity for U 20 (K d ϭ 1.2 Ϯ 0.2 nM) and weak affinity for A 20 (K d ϭ 24.8 Ϯ 1.3 nM) and G 20 (K d ϭ 114 Ϯ 4 nM), and no detectable binding was observed with C 20 (Fig. 2 and Table 1). Analysis of binding to G 20 by EMSA was not in accordance with filter binding data. We surmise that gel shift assays were complicated by the presence of the excessive secondary structure present in this probe (data not shown), and we

TABLE 1 Dissociation constants of YY1-probe complexes
Multiple series of binding reactions containing a range of concentrations of recombinant YY1 (1 nM to 10 M) with 10 nM of the indicated RNA or DNA probe (probe sequences given in Supplemental Material) were analyzed by filter binding assay. Experiments were carried out in triplicate. ds, double-stranded; ss, single-stranded; NA, not applicable.

YY1-mRNP Assembly Requires RNA Binding
note that this was the only case in any of the RNA-binding experiments in which apparent complex formation was not concentration-dependent. We next analyzed YY1 affinity for the single-stranded dinucleotide combinations (AC) 10 , (AG) 10 , (CU) 10 , and (GU) 10 . The dinucleotides (AU) 10 and (GC) 10 form self-complementary duplexes and are analyzed below. YY1 bound to (GU) 10 with moderate affinity (K d ϭ 38.0 Ϯ 1.5 nM) and (CU) 10 and (AG) 10 with low affinity, and no binding to (AC) 10 was detected ( Fig. 2 and Table 1). These data reveal a preference for U-rich single-stranded probes.
The effect of RNA secondary structure on YY1 binding was examined in greater detail using duplex-forming RNA oligonucleotides. Six possible duplexes can be made from the homopolymeric and dinucleotide RNAs described above (see Table 1). Each of these was prepared by labeling one of the two strands and annealing to the unlabeled complementary strand. YY1 bound with high affinity to the A:U duplex (K d ϭ 3.4 Ϯ 0.1 nM), whereas binding to the C:G duplex (Fig. 2, Table 1) was barely detectable. No significant binding of YY1 to (AU):(AU), (GC):(GC), (AG):(CU), or (AC):(GU) RNA duplexes was detected (data not shown). Overall, the results of Fig. 2 clearly demonstrate that YY1 exhibits direct, high affinity RNA binding activity with some degree of sequence specificity.
To determine unequivocally the identity of the complexes described above, YY1 was combined with radiolabeled RNA in the presence or absence of zinc and magnesium as well as with specific and nonspecific competitors (Fig. 3A). Complex formation depended on the presence of both magnesium and zinc ions. Furthermore, addition of cold specific DNA or RNA competitor, or anti-YY1 antibody, reduced or abolished complex formation. Nonspecific competitors and antibodies had no effect.
To determine the minimal binding site of YY1, singlestranded U 20 RNA was partially digested with RNase A, generating a population of probe fragments ranging in size from 1 to 20 nucleotides. Binding reactions were assembled with digested probe mixture and recombinant YY1 in the presence of nickelcharged NTA-agarose beads. YY1-RNA complexes formed in solution were recovered by binding of YY1 to the affinity matrix via the N-terminal His tag. The smallest fragment retained corresponds to the minimal binding site of YY1. Sequencing gel analysis of bound probe revealed YY1 retained fragments of 11 nucleotides and larger (Fig. 3B). Probe was not retained in control reactions without YY1 (data not shown). A minimum of 11 consecutive nucleotides are therefore required for binding of YY1 to RNA, and the YY1-RNA complexes observed with 20-mers (Fig. 2) likely contain a single YY1 molecule.
Because numerous cellular factors are known to associate with the 5Ј-cap structure of both polysomal and quiescent mRNAs (48,49), we also tested the potential effect of the 7-methylguanosine cap structure on YY1 RNA binding by determining the affinity of YY1 for capped versus uncapped synthetic RNA probes. No significant difference in affinity was detected between capped and uncapped probes (data not shown), suggesting YY1 does not bind to the 5Ј-terminal region of RNA.
RNA Binding of Endogenous YY1 from Native mRNPs-The results obtained with recombinant protein were confirmed using native preparations of the endogenous protein isolated from oocyte mRNPs. Complexes isolated by oligo(dT)-cellulose chromatography (19) were eluted and renatured, yielding an active preparation of isolated mRNP proteins, including YY1 (Fig. 4A). Control analysis of one oocyte equivalent of this preparation by EMSA using YY1 DNA consensus probe revealed the presence of DNA binding activity (Fig. 4B). YY1-DNA complexes were competed by specific oligonucleotide competitor   DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 or anti-YY1 antibody but were not affected by nonspecific competitors. Next, one oocyte equivalent of the mRNP preparation was analyzed for binding to A:U duplex RNA probe. Two prominent retarded bands were observed that were competed specifically by A:U RNA duplex, YY1 consensus DNA, or anti-YY1 antibody (Fig. 4B). Addition of C:G, for which recombinant YY1 displayed low affinity (Fig. 2), did not compete for complex formation. Similar results were obtained using U 20 probe RNA, demonstrating this phenomenon was not unique to the A:U duplex. The results of nucleic acid binding experiments were consistent between both recombinant and native mRNP-associated YY1, with the exception that native YY1 formed two retarded bands. It is possible that other mRNP proteins, isolated by oligo(dT) chromatography, are present in the retarded complexes. Together, these data clearly demonstrate that YY1 has the capacity to bind RNA directly and with high affinities similar to that for the DNA consensus.

YY1-mRNP Assembly Requires RNA Binding
Assembly of YY1 into Maternal mRNPs in Vivo Is Dependent on RNA Binding Activity-To demonstrate YY1 can associate directly with maternal mRNAs, in vitro binding reactions were assembled containing poly(A ϩ ) mRNA isolated directly from oocytes and recombinant YY1. Following incubation, reactions were subjected to oligo(dT)-cellulose chromatography to test for protein interaction with the maternal mRNA pool. Retention of YY1 on oligo(dT) demonstrated that YY1 was able to associate directly with maternal mRNA (Fig. 5A). Treatment of reactions with RNase abolished recovery of YY1 in bound fractions. The formation of YY1-mRNA complexes was effectively blocked by addition of U 20 to binding reactions but not by C 20 , results that are consistent with the direct RNA-binding experiments shown in Fig. 2 and Table 1. Addition of U 20 to reactions at the concentrations utilized did not significantly reduce total recovery of mRNA/mRNPs in bound fractions (data not shown). These in vitro data demonstrate that YY1 is able to bind directly to native maternal mRNA in the absence of other mRNP constituents.
The process of YY1-mRNP assembly was investigated in vivo by expression of HA-tagged YY1 from a plasmid construct (HA3-YY1) microinjected into oocytes, and subsequent examination of cytoplasmic extracts by size exclusion and oligo(dT) chromatography (Fig. 5, B and C). The results show that nascent HA-YY1 assembled into poly(A ϩ ) mRNPs with a mean molecular mass of 480 kDa, identical in size to native YY1containing mRNPs (Fig. 5B) (19). Additionally, HA-YY1 was retained on oligo(dT)-cellulose via interaction with poly(A ϩ ) mRNAs. RNase treatment of samples abolished retention on oligo(dT). Time course experiments revealed HA-YY1 could be detected in mRNPs as early as 2 h after microinjection with peak levels of mRNP-associated HA-YY1 appearing after 12 h (data not shown). This indicates that the molecular machinery responsible for the assembly of mRNPs is completely functional in the mature oocyte and that epitope-tagged YY1 is competent to assemble into mRNPs. In some experiments, oocytes were enucleated after 8 h of expression, and nuclear and cytoplasmic extracts were analyzed by Western blot. HA-YY1 was found to be completely localized to the cytoplasm (data not shown), identical to what has been observed for native YY1 (19,23). The assembly of HA-YY1 into mRNPs was significantly and specifically reduced by cytoplasmic microinjection of competitor RNA (U 20 ) (Fig. 5C). Microinjection of RNAs for which YY1 displayed low affinity in vitro did not affect assembly of HA-YY1-mRNPs. The observation that high affinity RNA substrates are able to inhibit YY1-mRNP assembly in vivo clearly demonstrates that the RNA binding activity of YY1 is not only biologically relevant but is an absolute requirement for its assembly into mRNP complexes. These observations also show that protein-protein interactions between YY1 and other mRNP proteins are not sufficient for association of YY1 with mRNPs.

DISCUSSION
Here we report that a novel activity for YY1, the capacity to bind RNA directly with high affinity, is required for assembly of YY1-containing mRNPs. The results of in vitro analysis of RNA binding activity (Fig. 2) were confirmed using native YY1 isolated from oocyte mRNPs (Fig. 4). YY1 bound to poly(A ϩ ) mRNA in vitro, and assembly of YY1 into mRNPs was specifically blocked in vivo by high affinity RNA substrates (Figs. 3 and 5). The major conclusions arising from these experiments are that YY1 has sequence-specific RNA binding activity for U-rich single-stranded and A:U duplex RNA and associates with maternal mRNA in vivo via direct RNA binding activity. The minimal length of single-stranded RNA substrate was found to be 11 nucleotides. Our findings reveal that the mechanism by which cytoplasmic YY1 associates with maternally derived mRNPs is through direct RNA binding activity. YY1 displayed highest affinity for single-stranded poly(U) RNA (Table 1), on the same order as affinity for its DNA consensus site (46,50,51). YY1 displayed sequence specificity in its RNA-binding characteristics, with high affinity for A:U duplexes and single-stranded U-rich regions. We hypothesize that these sequences are likely targets of YY1 in mRNPs in vivo. Binding to several single-stranded substrates other than U 20 was also observed; however, intramolecular A, G, or C residues significantly reduced or abolished binding for U-rich regions in both single-and double-stranded substrates ( Fig. 2 and Table  1). Low affinity binding was observed for several substrates, with affinities substantially lower than for the DNA consensus ( Fig. 2 and Table 1). No affinity was observed for RNA sequences such as C 20 , (AC) 10 , or any of the duplexes with the exception of A:U. Furthermore, very low affinity was observed for RNA complementary to YY1 promoter consensus sequences (Fig. 2). It is therefore unlikely that the high affinity RNA binding activities observed for poly(U) and A:U observed in this study are solely the result of nonspecific or backbone interactions between YY1 and RNA. The data taken together demonstrate sequence specificity in YY1 RNA binding. The general nonspecific RNA-binding characteristics of mRNP proteins Xp54 (33,39), FRGY2a/b (32), and Rap55 (37) contrast with the apparent sequence specificity of YY1. The functional consequences of these differences remain to be determined.
We conclude that YY1 recognizes internal regions of mRNAs, perhaps at U-rich or A:U-rich duplex regions within mRNAs (Fig. 2). This is supported by data showing YY1 did not bind appreciably to the 7-mGpppG 5Ј-cap structure (not shown) or single-stranded poly(A) probes (Table 1), which would mimic the poly(A) tail of maternal mRNAs. In any event, it is clear that binding to mRNA by YY1 is the primary mechanism by which YY1 is assembled into mRNPs.
Of all known zinc finger RNA-binding proteins, the precise mode of RNA-interaction has only been elucidated for three, including the Xenopus C2H2 zinc finger proteins TFIIIA and ZFa and the human CCCH zinc finger protein TS11d (52)(53)(54). Whereas the mode of DNA recognition by zinc finger proteins invariably involves recognition helix-major groove interactions, the same general principles do not apply to RNA recognition. The RNA-binding zinc fingers of TFIIIA make specific contacts with RNA bases in highly structured helical and nonhelical regions, and it binds a double-stranded region of the 5 S RNA with high sequence specificity (54). ZFa, in contrast, binds double-stranded RNA with little apparent sequence specificity, primarily through interaction with the RNA backbone (52). YY1 displays less apparent RNA sequence specificity than TFIIIA but more than ZFa (Fig. 2 and Table 1). YY1 displays high affinity for a single-stranded substrate, implying divergence in the mode of RNA recognition between C2H2 proteins. YY1 behavior more closely resembles the CCCH zinc finger protein TS11d. TS11d has high affinity for AU-rich duplex RNA, and the solution structure of this protein in complex with RNA has been determined (54). Sequence specificity of TS11d is apparently mediated entirely by peptide backbone contacts with Watson-Crick base pairs in helical RNA. Given the capacity of YY1 to bind single-stranded substrates (poly(U)), it would appear there is great variability in the mechanisms by which zinc finger proteins recognize RNA.
The present study constitutes the most comprehensive analysis of the RNA binding activity of any of the known Xenopus mRNP proteins and the first demonstration that RNA binding is an absolute requirement for assembly of mRNPs. Biochemical analysis of mRNPs has been difficult because they constitute a highly heterogeneous population of proteins and mRNAs. Moreover, they include a preformed, static population in the oocyte, and so genetic analyses and knockdown type experiments are not possible. Thus the major finding that YY1 assembles into mRNPs via direct RNA binding is an important first step in understanding the biochemical mechanisms of mRNP assembly and metabolism.