AUF1 binding affinity to A+U-rich elements correlates with rapid mRNA degradation.

Rapid degradation of many labile mRNAs is regulated in part by an A+U-rich element (ARE) in their 3′-untranslated regions. Extensive mutational analyses of various AREs have identified important components of the ARE, such as the nonamer motif UUAUUUAUU, two copies of which serve as a potent mRNA destabilizer. To investigate the roles of trans-acting factors in ARE-directed mRNA degradation, we previously purified and molecularly cloned the RNA-binding protein AUF1 and demonstrated that both cellular and recombinant AUF1 bind specifically to AREs as shown by UV cross-linking assays in vitro. In the present work, we have examined the in vitro RNA-binding properties of AUF1 using gel mobility shift assays with purified recombinant His6-AUF1 fusion protein. We find that ARE binding affinities of AUF1 correlate with the potency of an ARE to direct degradation of a heterologous mRNA. These results support a role for AUF1 in ARE-directed mRNA decay that is based upon its affinity for different AREs.

Rapid degradation of many labile mRNAs is regulated in part by an A؉U-rich element (ARE) in their 3-untranslated regions. Extensive mutational analyses of various AREs have identified important components of the ARE, such as the nonamer motif UUAUUUAUU, two copies of which serve as a potent mRNA destabilizer. To investigate the roles of trans-acting factors in ARE-directed mRNA degradation, we previously purified and molecularly cloned the RNA-binding protein AUF1 and demonstrated that both cellular and recombinant AUF1 bind specifically to AREs as shown by UV cross-linking assays in vitro. In the present work, we have examined the in vitro RNA-binding properties of AUF1 using gel mobility shift assays with purified recombinant His 6 -AUF1 fusion protein. We find that ARE binding affinities of AUF1 correlate with the potency of an ARE to direct degradation of a heterologous mRNA. These results support a role for AUF1 in ARE-directed mRNA decay that is based upon its affinity for different AREs.
Control of mRNA stability is an important component of eukaryotic gene expression and involves cis-acting elements that can be found in the coding region and/or UTRs 1 of mRNAs (reviewed in Refs. [1][2][3][4][5][6]. One type of cis-acting instability element is comprised of the AREs found in the 3Ј-UTRs of many unstable mRNAs (reviewed in Ref. 7). Many ARE-containing mRNAs are degraded by a sequential pathway involving removal of the poly(A) tract followed by degradation of the mRNA body (8 -10). In most cases the poly(A) tract is thought to protect the mRNA from ribonuclease attack so that its removal permits degradation of the mRNA body (reviewed in Ref. 11). While it has been known for almost a decade that AREs are important for mRNA instability (12)(13)(14)(15), the mechanism(s) by which they mediate mRNA turnover is still unknown.
Despite the presence of AREs in many different mRNAs, there is no single evolutionarily conserved AϩU-rich instability sequence. Typically AREs contain multiple copies of the pen-tanucleotide AUUUA, often in conjunction with one or more U-rich regions (14). In addition, transfection studies indicate that as the number of tandemly repeated AUUU motifs is increased in a reporter mRNA, its instability increases. Likewise, two copies of the nonameric motif UUAUUUAUU act as a more potent destabilizer than a single nonameric motif (16,17). Together, these analyses suggest that potent destabilizing AREs are high affinity binding sites for a mRNA decay factor(s).
In order to investigate how AREs function in mRNA turnover, we utilized a cell-free mRNA decay system to identify proteins that may be relevant to ARE-directed mRNA decay (8,18,19). To this end, we previously reported the purification, molecular cloning, and characterization of the ARE-binding protein AUF1 (20). Cellular AUF1 purified from cytoplasmic extracts of K562 human erythroid leukemia cells consists of a 37-and a 40-kDa isoform. Cloning of the 37-kDa isoform, p37 AUF1 , revealed two nonidentical RNA recognition motifs (21) and a short glutamine-rich region in the predicted amino acid sequence. Cloning of murine cDNAs suggests that the 40-kDa isoform may also contain 19 additional amino acids N-terminal to RNA recognition motif 1 (22). Both cellular and recombinant p37 AUF1 (hereafter referred to as AUF1) bind the AREs present in the c-fos and c-myc proto-oncogene mRNAs and the granulocyte-macrophage colony-stimulating factor cytokine mRNA as shown by UV cross-linking assays in vitro.
The potential influence of AUF1 on ARE-directed mRNA decay extends beyond the control of cytokine and proto-oncogene expression, however. Many mRNAs encoding components of G protein-coupled receptors, such as ␤-adrenergic receptors (␤-ARs), contain AREs. Moreover, receptor levels are frequently subject to regulatory control. For example, exposure of smooth muscle cells to agonist down-regulates ␤ 2 -AR mRNA levels by inducing degradation of the mRNA (23). Similarly, agonist-mediated destabilization of the human ␤ 1 -AR mRNA appears to be dependent upon an ARE (24), and for both the human ␤ 1 -AR and hamster ␤ 2 -AR mRNAs decay occurs concomitantly with an increase in the cytoplasmic levels of AUF1 (25). Since both cellular and bacterially expressed AUF1 bind the ␤ 2 -AR ARE (25), the reciprocal relationship between the half-life of ␤ 2 -AR mRNA and the abundance of AUF1 suggests that the half-lives of ARE-containing mRNAs may be dependent in part upon ARE-specific RNA binding affinity of AUF1.
Here, we test the hypothesis that the binding affinity of AUF1 for an ARE should reflect the potency of that ARE as a mRNA destabilizer. Using purified recombinant His 6 -AUF1 fusion protein, we find a direct relationship between the apparent K d for ARE binding by AUF1 and the potency of the ARE to direct mRNA decay. These results support a role for AUF1 in ARE-directed mRNA decay that is based upon its affinity for different AREs.

MATERIALS AND METHODS
All enzymes and plasmid vectors were obtained from Promega Corp. (Madison, WI) unless otherwise noted. All plasmid constructions were confirmed by both restriction enzyme analyses and dideoxy sequencing with Sequenase (version 2.0, U.S. Biochemical Corp.).
Expression and Purification of His 6 -AUF1 Fusion Protein-His 6 -AUF1 fusion protein was expressed in bacteria and purified as described by Pende et al. (25). The concentration of purified recombinant His 6 -AUF1 was estimated by comparison with dilutions of BSA using Coomassie-stained SDS-polyacrylamide gels.
Gel Filtration Analysis of His 6 -AUF1-A 1 ϫ 19-cm column of Sephacryl S-300 (Pharmacia Biotech Inc.) was equilibrated in factorbinding buffer (10 mM Tris, pH 7.5, 5.5 mM magnesium acetate, 100 mM potassium acetate). Four micromolar purified recombinant His 6 -AUF1 in a final volume of 300 l of factor-binding buffer was loaded onto the column; 60 300-l fractions were collected. Twenty microliters of each fraction was assayed for His 6 -AUF1 by Western blot analysis using anti-AUF1 antisera, and 5 l of each fraction was analyzed for RNAbinding activity by mobility shift assay with radiolabeled c-fos ARE. Similarly, 400 nM purified recombinant His 6 -AUF1 in 300 l of factorbinding buffer was loaded onto the column, and 20 900-l fractions were collected. Protein in these fractions was precipitated with 10% trichloroacetic acid, 20 g/ml lysozyme by incubation on ice for 30 min and centrifugation at 4°C for 30 min. Each precipitate was resuspended in 10 mM Tris (pH 7.5) and assayed for His 6 -AUF1 by Western blot. The void volume was determined using blue dextran.
The XbaI-SalI fragment of p␣19R␤ϩATϫ5 (described above) containing (ATTT) 5 was removed, and a BglII site was created using a synthetic linker, creating plasmid p␣19R␤⌬AU.
Purified recombinant His 6 -AUF1 fusion protein was incubated with 1 fmol of 32 P-labeled RNA probe in a final volume of 10 l containing 10 mM Tris-HCl (pH 7.5), 5 mM magnesium acetate, 100 mM potassium acetate, 2 mM dithiothreitol, 0.1 mM spermine, 0.1 g/l bovine serum albumin, 8 units of RNasin, 0.2 g/l tRNA, 5 g/l heparin, and 0.1 g/l poly(C). Reaction mixtures were incubated on ice for 10 min. Complexes were resolved by electrophoresis through nondenaturing 6% polyacrylamide gels (acrylamide/bisacrylamide ratio of 60:1) in 45 mM Tris borate (pH 8.3), 1 mM EDTA. Gels were prerun for 30 min at 13 V/cm prior to sample loading. Gels were then run at 13 V/cm for 2-3 h, dried, and visualized on a PhosphorImager (Molecular Dynamics). In some experiments there was a loss of shifted products during electrophoresis. This was observed as smears migrating between the bound and free RNA bands. Thus, free RNA bands were routinely used for quantitation using ImageQuant image analysis software (Molecular Dynamics). Free probe concentration was plotted versus His 6 -AUF1 concentration, and apparent K d values were determined as the protein concentration at which 50% of the RNA was bound (27). For each RNA substrate tested, binding assays were performed in triplicate, and the average apparent K d and standard deviation were determined. In addition, the Newman-Kuels analysis of variance test was applied to each set of apparent K d values to identify significant differences (p Ͻ 0.05) between K d values for binding to the various RNA substrates.
UV Cross-linking-Twenty-five nanograms of purified recombinant His 6 -AUF1 and 40 fmol of 32 P-labeled RNA containing the c-fos ARE were incubated as described above for mobility shift reactions. The binding mixture was then treated with 2500 J of UV light for 3 min and digested with 10 g of RNase A at 37°C for 30 min. The binding reaction was fractionated on an SDS, 10% polyacrylamide gel using prestained molecular weight standards as markers, fixed in the gel, and detected by autoradiography.

RESULTS
Binding of His 6 -AUF1 to Synthetic AREs Containing Tandem Repeats of AUUU-We previously reported the purification, molecular cloning, and characterization of the ARE-binding protein AUF1. Our hypothesis is that the binding affinity of AUF1 for an ARE should reflect the potency of that ARE as a mRNA destabilizer. To address this hypothesis biochemically, we have examined the RNA-binding properties of AUF1. For this purpose we expressed a His 6 -AUF1 fusion protein in Escherichia coli and purified the recombinant protein, which has an apparent molecular mass of about 51 kDa by SDS-polyacrylamide gel electrophoresis (Fig. 1, left panel). To confirm that this polypeptide was the ARE-binding protein, a binding reaction containing radiolabeled c-fos ARE was treated with UV light, digested with RNase A, fractionated by SDS-PAGE, and detected by autoradiography. The 51-kDa His 6 -AUF1 polypeptide was the major cross-linked species observed (Fig. 1, right  panel). As a control, a lysate prepared from bacteria expressing the empty vector had no detectable ARE-binding activity by UV cross-linking assay (data not shown). Therefore, the RNA-binding protein in these studies is His 6 -AUF1. In addition, the gel filtration profile of His 6 -AUF1 in 10 mM Tris-HCl (pH 7.5), 5 mM magnesium acetate, 100 mM potassium acetate was the same at both 400 nM and 4 M protein (data not shown), suggesting that the protein does not form high molecular weight aggregates that could affect its ARE-binding activity. RNA-binding activity of His 6 -AUF1 was assayed by nondenaturing gel mobility shift assays using radiolabeled substrates containing either one AUUUA motif or from two to five tandem repeats of AUUU. (See Fig. 2 for RNA sequences.) AUUUA motifs are present in the AREs of most unstable proto-oncogene and cytokine mRNAs, and the potency of an AUUU-containing sequence to act as a mRNA destabilizer is proportional to the number of tandemly repeated AUUU motifs (16,17,26). Radiolabeled RNA substrates were incubated with increasing concentrations of His 6 -AUF1, and the products were resolved by native polyacrylamide gel electrophoresis. Shown in Fig. 3 are representative gels along with plots of free RNA concentration versus fusion protein concentration. His 6 -AUF1 ARE binding affinities increase as the number of tandemly repeated AUUU motifs is increased (Table I), with lowest affinity binding to R␤ϩAUUUA (apparent K d ϭ 210 Ϯ 50 nM) and highest affinity binding to R␤ϩ(AUUU) 5 (apparent K d ϭ 19 Ϯ 7 nM). The control RNA substrate R␤⌬AU, which has all AUUU repeats deleted, is bound by His 6 -AUF1 with low affinity (apparent K d ϭ 660 Ϯ 90 nM (data not shown)).
Since multiple tandem repeats of AUUU constitute primarily U-rich sequence, the increasing binding affinity by His 6 -AUF1 observed with increasing AUUU copy number could result from an increase in the number of uridylate residues. To test His 6 -AUF1 binding affinity for U-rich RNA, binding to a substrate containing U 32 was performed, since this sequence has been tested for mRNA destabilizing activity (17). The apparent K d for His 6 -AUF1 binding to the R␤ϩU 32 substrate is Ͼ500 nM (Fig. 4). While low affinity binding occurs to the R␤ϩU 32 substrate, no binding is detected to the control ␤-globin substrate lacking the U 32 sequence (Fig. 4). Nonetheless the estimated binding affinity to R␤ϩU 32 (Ͼ500 nM) is at least 20-fold lower than that for binding to R␤ϩ(AUUU) 5 (K d ϭ 19 Ϯ 7 nM). Likewise, heterologous mRNAs containing poly(U) tracts in their 3Ј-UTRs are stable (16,17,28). We conclude that His 6 -AUF1 binding affinities for the AUUU-containing RNAs parallel their potencies as mRNA destabilizers. The low affinity binding to the U 32 substrate is also consistent with its inability to promote mRNA degradation.
Binding of His 6 -AUF1 to Authentic c-fos and c-myc ARE Sequences-Our results show that His 6 -AUF1 has a relatively low affinity for a substrate with two tandem AUUU motifs ( Table I). The c-fos ARE has two tandem AUUU motifs and a single AUUUA motif separated by 19 nucleotides, while the c-myc ARE has two AUUUA motifs separated by 25 nucleotides (see Fig. 2). However, the c-fos and c-myc AREs are very potent mRNA destabilizers (17,28). We therefore examined binding of His 6 -AUF1 to RNA substrates containing these AREs using the mobility shift assay. A plot of free RNA concentration versus fusion protein concentration revealed an apparent K d of 7.8 Ϯ 0.4 nM for the c-fos ARE (R␤ϩfosARE; Fig. 5A). High affinity binding does not require intact AUUUA motifs, since His 6 -AUF1 binds the ARE3 mutant c-fos substrate containing single U-to-A substitutions in each AUUUA motif with an apparent K d of 20 Ϯ 4 nM (R␤ϩARE3; Fig. 5B). (The difference in binding affinity between the wild-type and mutant c-fos ARE is not statistically significant (p Ͼ 0.05).) Likewise, His 6 -AUF1 binds the c-myc ARE with an affinity (K d ϭ 21 Ϯ 3 nM) similar to that for the c-fos ARE (Table I). (The difference in binding affinity between the c-myc and c-fos ARE is not statistically significant (p Ͼ 0.05).) We conclude that AUF1 binds authentic c-fos and c-myc AREs with high affinity even though these AREs lack multiple tandem repeats of AUUU. Thus, AUF1 is capable of binding a number of different AREs with high affinity.

DISCUSSION
The importance of AREs for mRNA turnover was first realized in 1986 (12,14), yet it is still unclear how AREs function in mRNA decay. We have utilized biochemical approaches to identify trans-acting factors that bind AREs in order to relate such RNA binding to mRNA degradation. We previously purified, characterized, and molecularly cloned the ARE-binding protein AUF1, and in the present study we have examined binding affinities of a His 6 -AUF1 fusion protein for AϩU-rich sequences with defined relative potencies as mRNA destabilizers. Here, by determining apparent K d values for His 6 -AUF1 binding, we demonstrate that AUF1-ARE binding affinity is directly related to the potency with which an ARE destabilizes a heterologous mRNA (Table I). Additionally, the affinity of His 6 -AUF1 for the most potent destabilizing AREs is within the average range (10 Ϫ9 M) of affinities exhibited by several other RNA-binding proteins that recognize specific sequences or FIG. 3. His 6 -AUF1 binding affinity for AUUU-containing sequences increases as the number of tandemly repeated AUUU motifs is increased. His 6 -AUF1 binding to RNA substrates containing AUUUA or tandem repeats of AUUU (see Fig. 2) was analyzed by electrophoretic mobility shift assays and apparent K d for His 6 -AUF1 binding to each RNA was determined as described under "Materials and Methods." Representative binding reactions using His 6  structures (29).
Certain AϩU-rich sequences are more potent mRNA destabilizers than others, suggesting that the potencies of destabilizers are proportional to the binding affinities of a cellular decay factor(s). For example, when placed in the context of a heterologous, normally stable mRNA, AUUUA and (AUUU) 2 are relatively ineffective as destabilizing elements; (AUUU) 3 has a modest destabilizing effect; (AUUU) 4 increases the decay rate further; and (AUUU) 5 is the most potent destabilizer of the five (16,17,26). In fact, (AUUU) 5 increases the degradation rate of a reporter mRNA to about the same extent as does the c-fos ARE (17). Likewise, His 6 -AUF1 binds the c-fos ARE (K d ϭ 7.8 Ϯ 0.4 nM) and the (AUUU) 5 substrate (K d ϭ 19 Ϯ 7 nM) with similar affinities. (The differences are not statistically significant (p Ͼ 0.05).) Statistical analyses were used to determine significant differences (i.e. p Ͻ 0.05) between K d values for His 6 -AUF1 binding to various RNA substrates, and as a result the RNA sequences used in this study can be grouped into three general classes: (i) RNAs that are either not bound or bound with low affinity by AUF1 and are not mRNA destabilizers (␤-globin, U 32 , and AUUUA); (ii) RNAs that are bound with gradually increasing, moderate affinities by AUF1 and have a gradually increasing, partial destabilizing effect ((AUUU) 2 Ͻ (AUUU) 3 Ͻ (AUUU) 4 ); and (iii) RNAs that are bound with the highest affinity by AUF1 and are potent mRNA destabilizers (c-fos and c-myc AREs and (AUUU) 5 ). Based upon these ranges of AUF1 binding affinities for various RNA substrates (low, moderate, high) and the relationship of high affinity binding to mRNA decay, the affinity of AUF1 for a mRNA may dictate the rate at which it is degraded. Therefore, cellular AUF1 concentration may be one determinant of mRNA half-life. In this regard we found that by comparing Western blots of K562 cytoplasmic extracts with known amounts of purified recombinant p37 AUF1 (the isoform used in these studies) that there are approximately 3.2 ϫ 10 4 cytoplasmic molecules of p37 AUF1 /cell (data not shown). Assuming a diameter of 20 m for K562 cells and 50% of the cell volume as cytoplasm (30), the concentration of p37 AUF1 is approximately 25 nM. This value is comparable with the apparent K d for binding to the c-myc ARE. Thus low cellular concentrations of active AUF1 may be sufficient for binding to a mRNA that contains a high affinity AUF1-binding site such as the c-myc and c-fos AREs. Based upon our results, such mRNAs should have very short half-lives. Likewise, mRNAs with AREs bound with lower affinities by AUF1 might require a higher concentration of active AUF1 for binding; these mRNAs should be degraded at a slower rate than those with high affinity binding sites. Thus, the availability of active AUF1 for ARE binding is a potential mechanism by which cells could control mRNA turnover rates and one in which the decay of multiple mRNAs could be differentially regulated by AUF1 concentration. Support for this hypothesis is the relationship between AUF1 levels and ARE-directed mRNA destabilization observed in DDT1-MF2 hamster smooth muscle cells treated with (Ϫ)isoproterenol. In this case, (Ϫ)isoproterenol induces an increase in cellular AUF1 protein and mRNA levels. This increase in turn correlates with a faster decay rate for ␤ 2 -adrenergic receptor mRNA, which contains an AUF1 binding site(s) in the 3Ј-UTR (25).
Recently, two groups reported that the functional sequence within an ARE appears to be the nonamer sequence UUAUUUAUU (16,17). One copy of the sequence UUAUUUAUU in the 3Ј-UTR of a normally stable mRNA can increase its degradation rate compared with the wild-type mRNA, while two copies of the nonamer motif act as a very potent mRNA destabilizer. As depicted in Fig. 2, (AUUU) 3 contains one copy of the nonamer (underlined in Fig. 2); (AUUU) 4 contains two overlapping copies; and (AUUU) 5 contains two copies that overlap by a single nucleotide. Consistent with the potencies of two copies or one copy of the nonamer as destabilizers, binding affinity of His 6 -AUF1 for (AUUU) 5 is 3-fold and 5-fold greater than the binding affinities for (AUUU) 4 or (AUUU) 3 , respectively (Table I). Thus, AUF1 may function in part via recognition of the nonamer motif.
Despite the potential importance of the nonamer motif UUAUUUAUU in ARE-directed mRNA decay, it is important to note that not all AREs found in unstable mRNAs contain this motif. For example, the portion of the c-myc ARE that functions as a very potent mRNA destabilizer does not contain this motif (28). While the c-myc ARE does contain noncontiguous AUUUA motifs, destabilizing AREs that contain no AUUUA motifs have also been identified. In addition, the presence of one or more AUUUA motifs in an ARE may not be sufficient for effective FIG 16 and 17), are depicted in the middle column as a series of plus symbols, where the number of symbols is proportional to the degradation rate of the mRNA (one is slowest and four is fastest). The apparent K d values (average Ϯ S.E.) for His 6 -AUF1 binding to each RNA are shown in the right column. mRNA destabilization (28). Moreover, analysis of a c-fos ARE mutant, ARE3, with single U-to-A substitutions in all three AUUUA motifs showed that intact AUUUA motifs are not required for rapid mRNA deadenylation but are important for rapid degradation of the mRNA body (9). The apparent K d for His 6 -AUF1 binding to the c-fos ARE and the ARE3 mutant were not statistically different (7.8 Ϯ 0.4 nM and 20 Ϯ 4 nM, respectively; p Ͼ 0.05). Likewise, His 6 -AUF1 bound a mutant c-myc ARE with single U-to-A mutations in both AUUUA motifs with affinity similar to wild-type c-myc ARE (data not shown). Thus, intact AUUUA motifs are not required for high affinity binding of His 6 -AUF1 to the c-fos and c-myc AREs. Therefore, although the study of AUUUA and UUAUUUAUU motifs has contributed greatly to understanding of ARE-directed mRNA decay, it is evident that these motifs may constitute only a subset of important motifs within various AREs.
In conclusion, our results suggest that the affinity of AUF1 for particular ARE sequences is related to their potency as mRNA destabilizers. Future experiments will utilize AUF1 as a tool to define multiple classes of AREs and to define specific nucleotide requirements for AUF1 binding by selection of high affinity binding substrates from combinatorial libraries of RNA sequences (e.g., SELEX; reviewed in Ref. 31).