Human PABP binds AU-rich RNA via RNA-binding domains 3 and 4 Rosemary T. Sladic 1 , Cathy A. Lagnado 1 , Christopher J. Bagley 1,2 and Gregory J. Goodall 1,2 1 Division of Human Immunology and Hanson Institute, Institute of Medical and Veterinary Science, Adelaide, Australia; 2 Department of Medicine, The University of Adelaide, Australia Poly(A) binding protein (PABP) binds mRNA poly(A) tails and affects mRNA stability and translation. We show here that there is little free PABP in NIH3T3 cells, with the vast majority complexed with RNA. We found that PABP in NIH3T3 cytoplasmic lysates and recombinant human PABP can bind to AU-rich RNA with high affinity. Human PABP bound an AU-rich RNA with K d in the n M range, which was only sixfold weaker than the affinity for oligo(A) RNA. Truncated PABP containing RNA recognition motif domains 3 and 4 retained binding to both AU-rich and oligo(A) RNA, whereas a truncated PABP containing RNA recognition motif domains 1 and 2 was highly selective for oligo(A) RNA. The inducible PABP, iPABP, was found to be even less discriminating than PABP in RNA binding, with affinities for AU-rich and oligo(A) RNAs differing by only twofold. These data suggest that iPABP and PABP may in some situations interact with other RNA regions in addition to the poly(A) tail. Keywords: PABP; iPABP; poly(A) binding protein; RNA- binding protein; AU-rich element. Poly(A) binding protein (PABP) occupies the poly(A) tails of mRNA transcripts in eukaryotic cells and has important influences on both the stability and the translation of mRNA [1]. Studies with in vitro systems are consistent with PABP acting as a physical impediment to limit the access of exoribonuclease to the 3¢-end of polyadenylated RNA [2–4], while more complex interactions also impact on mRNA stability. Through interactions with proteins that assemble on the 5¢-cap structure of mRNA, PABP participates in the effective circularization of mRNA. One consequence of this is that PABP helps limit the access of degradative enzymes to the 5¢-end of the mRNA, by helping to stabilize the binding of proteins such as the translation initiation factors eIF4G and eIF4E [5–8], and possibly other proteins as well [9]. In addition to its global effects on mRNA stability, PABP participates in the action of certain cis-acting elements that target particular mRNAs for rapid degradation. Some vertebrate mRNAs contain AU-rich elements (AREs) that act by a mechanism that includes acceleration of poly(A) shortening [10,11]. As PABP is bound to the poly(A) tail, it is likely to influence the poly(A) shortening rate, and consistent with this, an in vitro system of AU-mediated poly(A) shortening requires prior titration of PABP by addition of poly(A) [2]. AU-rich elements can also target mRNA in yeast for rapid degradation in a deadenylation dependent manner [9,12]. In vertebrates, poly(A) shortening is also the target of destabilizing elements that are structur- ally distinct from the ARE [13,14]. The initiation of translation, and possibly translation termination, are also influenced by PABP. The circulari- zation of mRNA that results from the physical interaction of PABP with the translation initiation factor eIF4G [5–7,15] enhances the recruitment of both the 40S and 60S ribosome subunits [16,17], resulting in an enhanced rate of translation. PABP also interacts with the translation termination factor eRF3, which may further enhance translation by promoting the recycling of ribosomes on the same mRNA [18]. Structurally, PABP consists of four RNA recognition motif (RRM) domains that constitute the N-terminal half of the molecule, and a C-terminal domain that has been implicated in intermolecular interactions between PABP molecules bound to a common poly(A) tail [19]. Studies of the binding of yeast and Xenopus PABPs to homopolymeric RNAs have shown that RRM domains 1 and 2 (RRM1,2) are sufficient for high affinity binding to poly(A), while RRM domains 3 and 4 can also bind poly(U) [19–21]. The crystal structure of the RRM1,2 pair-complexed with oligo(A) RNA has been determined [22], revealing a continuous RNA binding trough in which lays an 8 nucleotide length of oligo(A) RNA. We report here that PABP can bind with high affinity to AU-rich RNA in vitro, in agreement with some other recent reports [23,24]. We identified the region within PABP that is responsible for binding to an AU-rich RNA and also found that the closely related protein, iPABP [25], binds to AU-rich RNA with almost as high an affinity as it binds to oligo(A) RNA. This suggests that PABP in vivo could bind not only to poly(A), but also to other sites on some mRNAs,ifpresentatalevelinexcessofthatrequiredto occupy all the available sites on poly(A) tails. Correspondence to G. J. Goodall, Hanson Institute, IMVS, Frome Road, Adelaide, S.A., 5000, Australia. Fax: + 61 88232 4092, Tel.: + 61 88222 3430, E-mail: greg.goodall@imvs.sa.gov.au Abbreviations: ARE, AU-rich element; PABP, poly(A) binding protein; REMSA, RNA electrophoretic mobility shift assay; RRM, RNA recognition motif. (Received 31 October 2003, accepted 28 November 2003) Eur. J. Biochem. 271, 450–457 (2004) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03945.x Materials and methods Preparation of cell extracts NIH3T3 fibroblasts were grown to near-confluence in Dulbecco’s modified Eagles’ medium with 7.5% (v/v) fetal bovine serum (Gibco). For preparation of cytoplasmic extracts (essentially according to [26]), cells were harvested by scraping, washed in cold NaCl/P i and lysed on ice in buffer containing 40 m M KCl, 10 m M Hepes, pH 7.9, 3 m M MgCl 2 ,1m M dithiothreitol, 0.2% (v/v) Nonidet P-40 (NP-40), 5% (v/v) glycerol, 8 lgÆmL )1 aprotinin, 2 lgÆmL )1 leupeptin and 0.5 m M phenylmethanesulfonyl fluoride. Nuclei were removed by centrifugation at 14 000 g for 2min.CaCl 2 was added to the supernatant, followed by micrococcal nuclease (Worthington Biochemical Corpora- tion, Lakewood, NJ, USA) 1 to give final concentrations of 1m M and 12.5 lgÆmL )1 , respectively. Following incubation at 25 °C for 10 min, the nuclease was inactivated by addition of EGTA to a final concentration of 4 m M . UV crosslinking assay NIH3T3 cell extracts (0.3–0.5 lg of protein) were incubated for 10 min at room temperature in 40 m M KCl, 10 m M Hepes, pH 7.6, 3 m M MgCl 2 ,1m M dithiothreitol and 5% (v/v) glycerol with 25 ng yeast RNA (Boehringer) prior to the addition of probe RNA (26 fmol), various molar excesses of cold competitor RNA and 270 ng tRNA, then further incubated for 30 min at room temperature. The reaction mixes (final volume 12 lL) were irradiated in 96-well polypropylene microtitre plates on ice with 100 mJ of 254 nm UV light ( 35 s) using a Stratalinker 1800 apparatus (Stratagene), then electrophoresed on 11% (w/v) SDS/PAGE under reducing conditions. Gels were fixed, dried and analysed using a Molecular Dynamics Phos- phorImager and IMAGEQUANT version 3.3 (Molecular Dynamics, Sunnyvale, CA, USA). Immunoprecipitation For immunoprecipitation with the anti-human PABP monoclonal Ig 10E10 [27] or control monoclonal anti- bodies, 10 lg of cytoplasmic extract from HEL 299 human embryonic lung fibroblasts (ATCC, Manassus, VA, USA) 2 was digested with micrococcal nuclease and UV crosslinked as described above but with a UV dose of 750 mJ, in the presence of 78 fmol of probe. Samples were then made to 100 lL with buffer A containing 0.2% (v/v) NP-40 and 0.6 UÆmL )1 RNasin, after which 50 lLofa 25% slurry of Protein A-Sepharose (Pharmacia) equili- brated in the same buffer was added. The tubes were then placed on a rotating wheel at 4 °C for 1 h. The resulting precleared supernatant was incubated on ice for 1 h in the presence of 0.5–5 lL (as shown) of ascites fluid prior to the addition of a further 50 lL of Protein A-Sepharose slurry and incubated on a rotating wheel at 4 °Cfor1h. The beads were washed three times in buffer A containing 0.2% (v/v) NP-40 and twice in buffer A containing 0.5% (v/v) NP-40 and 100 m M NaCl. SDS/PAGE loading dye was added and the samples heated at 98 °Cfor5min before SDS/PAGE on an 11% (w/v) gel. Preparation of RNA oligonucleotides RNA was prepared by in vitro transcription performed using partial duplex oligonucleotide templates encoding the T7 RNA polymerase promoter, essentially according to the procedure of Milligan et al. [28]. For example, the oligo- nucleotides 5¢-TAATACGACTCACTATAGG-3¢ (univer- sal upper strand) and 5¢-CT (25) CCTATAGTGAGTCGT ATTA-3¢ (sequence specific lower strand for A25) direct transcription of the A25 sequence when annealed and used as a template in the in vitro transcription reaction. The other RNA oligonucleotides were similarly encoded using appro- priate lower strands. In vitro transcription was performed for 4 h at 37 °C in a reaction volume of 30 lL containing annealed template at 2.5 pmolÆmL )1 ,40mgÆmL )1 PEG 8000, 40 m M MgCl 2 ,1m M spermidine, 50 m M dithio- threitol, 0.01% (v/v) Triton X-100, 40 m M TrisCl, pH 8.1, 5mgÆmL )1 BSA, 40 m M each of CTP, ATP, GTP and UTP, [ 32 P]GTP[cP] to a final specific activity of 0.17 CiÆ mmol )1 ,4.5UÆlL )1 T7 RNA polymerase (Promega) and 8UÆmL )1 pyrophosphatase. Following template removal by digestion with RQ1 DNase (Promega), the transcripts were purified on 16% (w/v) polyacrylamide/8 M urea gels. All transcripts were treated with calf intestinal phosphatase (Promega) under manufacturers’ recommended conditions before the incorporation of c 32 P into 5 pmol of transcripts from [ 32 P]ATP[cP] by T4 polynucleotide kinase (Promega) under manufacturer’s recommended conditions. Transcripts were purified on 16% (w/v) polyacryamide/8 M urea gels andstoredin1m M EDTA. PABP constructs Sequence encoding amino acids 10–636 of the major cytoplasmic PABP, PABPC1 (NCBI RefSeq NM_002568) was subcloned from plasmid containing PABP cDNA [27] into the end-filled NdeI site of the histidine tagged vector pET28a+ (Novagen) to produce the construct pEThPABP. Two stop codons were introduced by site directed mutagenesis at amino acids 385 and 386 (end of RRM domain 4) into pBSPABP (BamHI cloned fragment containing PABP cDNA in pBlueScript vector) from which sequence encoding PABP amino acids 10–636 was sub- cloned into the end-filled NdeI site of the histidine tagged vector pET28a+ (Novagen) to produce the construct pEThPABP1234. Sequence encoding PABP RRM1 and RRM2 (amino acids 10–179) was PCR amplified from pBSPABP using oligonucleotides incorporating an NdeI restriction site at the 5¢-end and a stop codon followed by a SacI site at the 3¢-end. The resultant NdeI–SacI digested PCR product was cloned into similarly digested pET28a+ vector to produce the construct pEThPABP12. Sequence encoding PABP RRM3 and RRM4 (amino acids 190–374) was PCR amplified from pBSPABP using oligonucleotides incorporating an NdeI at the 5¢-end and a stop codon and a BamHI site at the 3¢-end. The resultant NdeI–SacI digested PCR product was cloned into similarly digested pET28a+ vector to produce the construct pEThPABP34. Two alternate changes were introduced at PABP amino acids 337 in pBSPABP by site directed mutagenesis. Sequence encoding PABP amino acids 10–636, including the change F337V, was subcloned from pBSPABP into the end-filled Ó FEBS 2003 Human PABP Binds AU-rich RNA via RRM 3 and 4 (Eur. J. Biochem. 271) 451 NdeI site of pET28a+ to produce the construct pEThPABP-F337V. Sequence encoding PABP amino acids 10–636, including the change F337D was subcloned from pBSPABP into the end-filled NdeI site of pET28a+ to produce the construct pEThPABP–F337D. Sequence encoding iPABP (PABPC4, NCBI RefSeq NM_003819) amino acids 1–644 was subcloned as an NcoI–XbaI fragment from plasmid piPABP [25] into the NcoI–NheI site of pET28a+ to produce the construct pEThiPABP. Electrophoretic mobility shift assays Electrophoretic mobility shift assays were performed in a total volume of 15 lL. Protein (1.5 lL) diluted to an appropriate concentration in protein buffer (50 m M NaPO 4 , pH 7.8, 300 m M NaCl, 10% glycerol, 2 m M dithiothreitol and 0.05% NP-40) was added to 10 fmol of radioactively labeled RNA in 13.5 lL binding buffer [10 m M Tris/HCl 3 , pH 8.0, 10% (v/v) glycerol, 0.05% (v/v) NP-40, 1 m M dithiotheitol, 70 m M KCl, 100 ngÆmL )1 BSA] and incuba- ted on ice for 15 min. Binding reactions were electrophore- sed in nondenaturing 7% (w/v) polyacrylamide gels [60 : 1, 1· Tris/borate/EDTA 4 and 0.1% (v/v) Triton X-100] at 4 °C at 220 V for 120 min. Gels were visualized with the use of a Molecular Dynamics PhosphorImager. Free RNA bands were quantitated using IMAGEQUANT image analysis soft- ware (Molecular Dynamics). The amount of free probe vs. log His-PABP concentration was plotted using PRISM version 1.03 (Graph Pad Software). Apparent K d was determined as the EC 50 , which is the protein concentration at which 50% of RNA was bound. Purification of recombinant protein Transformed strains of BL21(DE3) were grown at 37 °Cin superbroth (3.2% bactotryptone, 2.0% yeast extract, 0.5% NaCl, pH 7.2) and 30 lgÆmL )1 kanamycin to an A 600 value of 0.5. His-tagged protein expression from the transformed plasmid was induced by the addition of isopropyl thio-b- D -galactoside (0.5 m M final concentration). After 2 h of growth at 37 °C, cells were harvested by centrifugation, frozen and resuspended in 30 mL lysis buffer (50 m M NaPO 4 , 300 m M NaCl, 10 m M imidazole, pH 8.0). Thawed cells were sonicated and cell debris was removed by centrifugation. For Ni-nitriloacetic acid purification, the cleared supernatant was incubated with 1 mL of Ni-nitrilotriacetic acid agarose (Qiagen) for 2 h at 4 °C with constant mixing on a rotating wheel. The Ni-nitrilo- triacetic acid resin was transferred into a column and washed twice with 10 mL wash buffer (50 m M NaPO 4 , 300 m M NaCl, 20 m M imidazole, pH 8.0) before the His- tagged protein was eluted from the column with elution buffer (50 m M NaPO 4 ,300m M NaCl, 250 m M imidazole, pH 8.0). Ni-nitrilotriacetic acid purified proteins were buffer-exchanged into poly(U) binding buffer (10 m M Tris/HCl, pH 8.0, 1 m M dithiothreitol, 200 m M NaCl) with Centripreps (Amicon) to reduce the imidazole concentration to below 10 m M and bound to 1 mL of equilibrated poly(U) Sepharose (Pharmacia) at room temperature. The protein bound Sepharose was transferred to a column, washed twice with 10 mL poly(U) binding buffer and eluted with poly(U) elution buffer (10 m M Tris pH 8.0, 1 m M dithiothreitol, 800 m M NaCl). Protein concentrations were determined by comparison with dilutions of BSA on Sypro Orange (Molecular Probes) stained SDS/polyacrylamide gels using Molecular Dynamics IMAGEQUANT software. Results PABP binds to AU-rich RNA We employed a UV cross-linking assay to detect proteins in cytoplasmic extracts from NIH3T3 cells that bind to AU-rich RNAs. To ensure that proteins in the extract that are tightly bound to endogenous mRNA and consequently not readily available to bind the probe in the UV cross- linking assay would also be detected, we treated the cytoplasmic extract with micrococcal nuclease and then inactivated the nuclease with EGTA before performing the UV cross-linking assay. The nuclease treatment released a protein that, when UV cross-linked to the probe, migrated at approximately 90 kDa in SDS/PAGE (Fig. 1A). When the time of UV cross-linking was extended, a second, faster migrating band from micrococcal nuclease-treated extracts was also seen. Digestion of the crosslinked complexes with RNase A converted both bands to a single species migrating at  79 kDa, suggesting both bands are derived from the same protein (data not shown). Using competitor RNAs of various sequence composition and with a range of effect- iveness in destabilizing mRNA in vivo [11], we found that this protein preferentially bound to AU-rich RNA (Fig. 1C), although the selectivity of the protein for destabilizing AU-rich sequences in comparison to non- destabilizing AU-rich sequences was not sufficient to strongly suggest that it is involved directly in the destabil- izing function of AREs in vivo (data not shown). The protein also bound with high affinity to poly(A) RNA (Fig. 1C). Furthermore, the UV crosslinked complex could be immunoprecipitated with a monoclonal antibody to PABP (Fig. 1D), suggesting that that the crosslinked protein is most likely PABP. Identification of the domains responsible for binding to AU-rich RNA To further verify that PABP can bind AU-rich RNA, we prepared a recombinant His 6 -tagged form of the major cytoplasmic PABP (PABPC1), and measured its affinity for poly(A) and AU-rich RNA in an RNA electrophoretic mobility shift assay (REMSA). The AU-rich RNA used is similar in sequence to an ARE from the GM-CSF 3¢-UTR. To estimate the relative affinity for each RNA, a constant low concentration of radiolabelled RNA was incubated with a range of protein concentrations, after which the bound and unbound RNAs were separated by native PAGE. Under these conditions, the apparent K d can be determined from the concentration of protein at which 50% of the RNA probe is bound [29]. We compared the binding of PABP to a 26 nucleotides AU-rich probe, AU4, an oligo(A) RNA (A25) and a mixed sequence RNA, M8 (Fig. 2). The recombinant PABP bound the A25, AU4 and M8 RNAs with apparent affinity constants of 0.7, 3.9 and 7.7 n M , respectively 452 R. T. Sladic et al. (Eur. J. Biochem. 271) Ó FEBS 2003 (Fig. 2). Thus, purified recombinant PABP binds prefer- entially to poly(A) as expected but can also bind other RNAs with considerable affinity. To determine which domains of PABP contribute to binding the AU-rich RNA, we prepared truncated forms of PABP as His6-tagged proteins. PABP1234 contains the four RRM domains but not the large C-terminal domain, while PABP12 and PABP34 contain RRM domains 1 and 2, and RRM domains 3 and 4, respectively (Fig. 3). The Fig. 1. Detection of an AU-binding protein by UV crosslinking assay and identification as PABP. (A) UV crosslinking assays were per- formed with 0.5 lg of NIH3T3 cytoplasmic extract, with and without prior digestion of the extract with micrococcal nuclease, using UV doses of 100 mJ or 750 mJ as shown. Lane 1 extract was preincubated without addition of micrococcal nuclease. Lane 2 extract was pre- incubated with micrococcal nuclease but in the presence of an inhibi- tory level (4 m M )ofEGTA.Lane3extractwaspreincubatedwith micrococcal nuclease for 10 min before the nuclease was inactivated by addition of EGTA. (B) Sequences of the probe and competitor RNAs used in UV crosslinking assays. The M8 sequence has UUU motifs converted to CUC motifs. (C) UV crosslinking assays using micro- coccal nuclease treated extract in the presence of 0, 2.6, 6.4, 16, 40 and 100–fold molar excess of competitor RNA as indicated. (D) Immu- noprecipitation of the UV cross-linked complex with anti-PABP mAb 10E10, or irrelevant control mAb. Fig. 2. Affinity of recombinant PABP for oligo(A), AU-rich and mixed sequence RNAs measured by REMSA. (A) Sequences of RNA probes used in REMSAs. (B) Representative REMSAs in which each probe was incubated with a twofold dilution series from 1.25 to 80 n M of recombinant PABP. (C) Binding curves of free probe vs. protein concentration. The data shown are from an experiment in which the PABP concentrations were adjusted to give a high density of points in the vicinity of the EC 50 . K d values calculated from three replicate experiments are shown in Table 1. Fig. 3. Full length and truncated forms of recombinant PABP. (A) Schematic of the His 6 -tagged proteins. (B) Coomassie stained SDS/ polyacrylamide gel of recombinant proteins purified by nickel chro- matography and poly(U) affinity column. Lane 1, PABP; lane 2, PABP1234; lane 3, PABP12; lane 4, PABP34. Ó FEBS 2003 Human PABP Binds AU-rich RNA via RRM 3 and 4 (Eur. J. Biochem. 271) 453 His 6 -tagged proteins were purified by chromatography on Ni-nitrilotriacetic acid columns, and subsequently by affinity chromatography on poly(U) sepharose to ensure that the proteins were correctly folded and competent to bind RNA. Removing the C-terminal domain from PABP made little change to the affinity for A25, and resulted in a small increase in the affinity for AU4 and M8 (PABP1234, Fig. 4, Table 1). Thus, the primary determinants for RNA binding in human PABP are within the four RRM domains, as they are in the Xenopus protein [19]. Further truncation to remove domains 3 and 4 (producing PABP12) caused a large decrease in the binding of the AU-rich and mixed sequence RNAs. The apparent K d of PABP12 for AU4 was 113 n M (87-fold change compared to PABP1234). The affinity for M8 was also severely reduced, with apparent K d of 308 n M (128-fold change compared to PABP1234). However the apparent K d forA25was1.8n M , which is only slightly decreased compared to PABP1234. As PABP12 is less than half the size of PABP1234, two molecules of Fig. 4. Measurement of apparent K d values of truncated forms of PABP for various RNA sequences. REMSAs were performed using a fixed probe concentration (0.67 p M ) and with a range of protein concentrations, the limits of which are indicated in n M units. Binding experiments with PABP1234, PABP12 and PABP34, are shown in the left, middle and right columns, respectively. The top row shows REMSAs using A25 probe, the second row shows REMSAs with AU4 probe and the third row shows REMSAs with M8 probe. Above each gel a vertical arrow indicates the point at which 50% of probe has been bound by protein. Free probe in each lane was quantitated by phosphorimage analysis. The bottom panel in each column shows the plots of free probe vs. protein concentration, from which the apparent K d values were determined. At high protein concentration, the PABP12 and PABP34 proteins can bind probe with 2 : 1 stoichiometry, which perturbs the binding profile at the higher concentrations. The K d determination is slightly affected by this, but only to a very small degree, because at the protein concentration that leaves 50% free probe, very little trimeric complex is formed. Table 1. Apparent K d values for PABP proteins. Apparent K d values were determined by REMSA as described in Materials and methods. The data for each K d determination are pooled from three experiments. 95% Confidence limits are shown in brackets. Protein K d for A25 (n M ) K d for AU4 (n M ) K d for M8 (n M ) PABP 0.67 (0.61–0.74) 3.9 (3.7–4.1) 7.7 (6.5–9.1) PABP1234 0.74 (0.61–0.81) 1.3 (1.2–1.4) 2.4 (2.2–2.6) PABP12 1.8 (1.6–1.9) 113 (82–157) 308 (88–1074) PABP34 1.5 (1.3–1.6) 2.9 (2.8–3.1) 12 (11–13) 454 R. T. Sladic et al. (Eur. J. Biochem. 271) Ó FEBS 2003 PABP12 can be accommodated per probe molecule, resulting in a second, lower mobility band at higher concentrations of PABP12. This has only a slight effect on the apparent K d , which is calculated from measurement of free probe, because at the protein concentration that leaves 50%freeprobe,verylittleofthe2:1complexisformed. The results suggest that much of the binding affinity for poly(A) results from binding to domains 1 and 2, but that these domains do not contribute much to the binding of AU-rich RNA. In contrast, PABP34, containing just domains 3 and 4, bound the AU4 RNA almost as well as it bound the A25 RNA. The apparent K d of PABP34 for A25 was 1.5 n M and for AU4 was 2.9 n M , while the apparent K d for M8 was 11.7 n M . Thus, the major contribution to the binding of PABP to AU-rich RNA comes from domains 3 and 4. Effect of mutation of a residue predicted to contact AU-RNA In an investigation of the RNA binding properties of yeast PABP, Deardorff and Sachs [30] found that mutation of a single Phe residue in RRM4 to Val caused a large reduction in binding to a U-rich RNA (UUUUGUUGUUUU UUUUCUAG), without having a drastic effect on binding to oligo(A). This result is consistent with our findings with human PABP that the RRM3,4 pair is important for binding a U-rich RNA (AU4) and furthermore suggests that an equivalent mutation in the human PABP may affect binding to AU-rich RNA without a major effect on binding to poly(A). We aligned the sequences of the human and yeast PABPs and identified a Phe residue in human PABP (Phe337) that is equivalent to the yeast Phe366 (Fig. 5). We also modeled the structures of human and yeast RRM3,4 on the published crystal structure of human RRM1,2 [22], which indicated that the RNA binding pocket in RRM4 is similar in the yeast and human PABPs (data not shown). To test whether Phe337 in human PABP is crucial for binding to AU-rich RNA, we made mutant forms of the protein, PABP-F337V and PABP-F337D, in which Phe337 was mutated to Val and Asp, respectively. The recombinant His 6 -tagged mutant proteins were purified and their affinity for oligo(A), AU-rich and mixed sequence RNA was measured by REMSA and compared to that of the wild- type protein (Fig. 5). In contrast to the effect of this mutation in the yeast protein, mutation of this residue in the human protein did not reduce the affinity for any of the RNAs tested. Thus, yeast and human PABP may differ in their interaction with U-rich RNAs, despite their similarity in sequence and predicted structure, although the possibility also remains that differences in the sequences of the U-rich probes used in the two studies were responsible for the different outcomes. Binding AU-rich RNA is also a property of iPABP The inducible poly(A) binding protein, iPABP, is closely relatedtoPABPinsequence(79%identity).RRM domains 1 and 2 are especially conserved between the two proteins, sharing 95% sequence identity but RRM domains 3 and 4 are slightly less conserved (78% and 88%, respectively), raising the possibility that the iPABP may have similar affinity to PABP for poly(A) but have different affinity for AU-rich RNA. iPABP is expressed at low levels in resting human T cells, but its mRNA is induced rapidly following T cell activation [25], although the function of iPABP is unknown. To assess whether iPABP binds AU-rich RNA we prepared His 6 -tagged iPABP and measured its binding to oligo(A) and AU-rich RNA by REMSA (Fig. 6). The apparent K d of iPABP for the oligo(A) RNA was 1.1 n M , which is a slightly lower affinity than that of PABP for oligo(A). However, the apparent K d of iPABP for AU-rich RNA was 2.4 n M , which is a slightly higher affinity than PABP has for the AU-rich RNA. Thus the affinity of iPABP for AU-rich RNA is almost as high as its affinity for oligo(A) (2.4 n M and 1.1 n M , respectively). Fig. 5. Measurement of apparent K d values of PABP Phe337 mutants for different RNA sequences. (A)Alignmentofthesequencesofyeast and human PABPs in the vicinity of yeast Phe366, showing the high degree of sequence conservation in this region of the proteins. The yeast Phe336 and the equivalent residue in human PABP that was chosen for mutation are indicated by asterisks. (B) REMSAs were performed and binding curves constructed as in Fig. 4. The binding curves for PABP-F337V and PABP-F337D are shown in the upper and lower graphs, respectively. (C) Apparent K d values, with 95% confidence limits shown in brackets, for the mutant PABPs binding to each RNA. The data are pooled from three experiments. For com- parison, the apparent K d for PABP binding to each RNA, determined in Fig. 2, is also shown in the table. Ó FEBS 2003 Human PABP Binds AU-rich RNA via RRM 3 and 4 (Eur. J. Biochem. 271) 455 Discussion We found that both the poly(A) binding protein, PABP1, and the inducible poly(A) binding protein, iPABP, bind to an AU-rich RNA that is predominantly U-rich, and thus does not resemble the major in vivo RNA substrate for PABP, which is poly(A). The affinity of the interaction of these proteins with the AU-rich RNA is comparable to the affinities that some other RNA-binding proteins have for their in vivo RNA substrates. For example, the AU-binding protein AUF1/hnRNPD, which is involved in the rapid degradation of some mRNAs [31–33], binds to AU-containing RNAs with K d values of 10–20 n M [34]. Thus, although PABP binds primarily to mRNA poly(A) tails, PABP or iPABP could potentially bind additional RNA sites if the amount of cytoplasmic PABP exceeded that required to occupy the available poly(A) tails. When carrying out the UV cross-linking assay with cytoplasmic extract from NIH3T3 cells we did not detect cross-linking of the RNA probe to PABP if the extract was not first treated with micrococcal nuclease. This suggests that the majority of PABP in growing but unstimulated NIH3T3 cells is tightly complexed to RNA [most likely to poly(A)] and that there is very little PABP available for binding to RNA sequences of lower affinity than poly(A). This is consistent with the finding that a negative feedback regulation circuit, involving an A-rich region in the PABP mRNA 5¢-UTR, normally limits the expression of PABP [35,36]. However it is possible that higher levels of PABP are present in some situations. For example, iPABP mRNA is markedly induced in activated T cells [25], although it is not known whether this results in an increase in the availability of either PABP or iPABP to bind RNA sequences other than poly(A). We found that iPABP shares the property of binding AU-rich RNA with PABP1. iPABP barely distinguishes between an oligo(A) RNA, A25, and an AU-rich RNA, while the affinity of recombinant PABP1 for an oligo(A) RNA was only sixfold greater than for the AU-rich RNA. The binding of PABP to poly(A) in vivo may be further stabilized by interactions with translation initiation factors bound at the 5¢-cap [6] and also by cooperative interactions between multiple PABP molecules bound to long poly(A) tails [19]. Whether such protein–protein interactions might affect binding to other RNA sequences is not known. In addition, it is possible that post-translational modifications such as phosphorylation [24,37] or methylation [38], might differentially affect the affinity of PABP for different RNA sequences, thereby changing the degree of selectivity of PABP for the different RNA sequences. It is interesting that the selectivity of RNA binding by PABP is markedly partitioned between the RRM1,2 and RRM3,4 domain pairs. The RRM1,2 pair is highly selective for poly(A) over the AU-rich RNA, whereas the RRM3,4 pair barely distinguishes between the two RNA sequences. Our results are consistent with studies of PABP from Xenopus [19], which showed the domain 1,2 pair to have greater selectivity than the domain 3,4 pair for poly(A) over other homopolymeric sequences, although in this case affinity constants were not determined. That the selectivity has been maintained during evolution suggests it may play a role in the function of PABP, although we have no indication to date what that function may be. One possible role for PABP binding to an AU-rich RNA region of a viral mRNA has been proposed recently. Human papillomavirus type 1 late mRNA contains an AU-rich region in its 3¢-UTR that negates enhancement of trans- lation by the poly(A) tail. This AU-rich element was found to bind PABP, leading to the suggestion that the element may interfere with the interaction of PABP with eIF4G, thereby preventing circularization of the mRNA and reducing its rate of translation [23]. However, it remains to be determined whether this is the mechanism through which the papillomavirus AU-rich element functions, and whether such an effect might occur with any cellular mRNAs. Acknowledgements We thank Jeff Ross for initially suggesting PABP as the identity of the major band in the UV cross-links, Matthias Goerlach and Gideon Dreyfuss for providing PABP cDNA, Tullia Lindsten for iPABP cDNA and Gideon Dreyfuss for PABP mAb. This work was supported by a program grant from the National Health and Medical Research Council of Australia. References 1. Mangus, D.A., Evans, M.C. & Jacobson, A. (2003) Poly(A)- binding proteins: multifunctional scaffolds for the post- transcriptional control of gene expression. Genome Biol. 4, 223. 2. Ford, L.P., Bagga, P.S. & Wilusz, J. (1997) The poly(A) tail inhibits the assembly of a 3¢-to-5¢ exonuclease in an in vitro RNA stability system. Mol. Cell Biol. 17, 398–406. 3. Bernstein, P.L., Peltz, S.W. & Ross, J. (1989) The poly(A)- poly(A)-binding protein complex is a major determinant of mRNA stability in vitro. Mol. Cell Biol. 9, 659–670. 4. Wang, Z., Day, N., Trifillis, P. & Kiledjian, M. (1999) An mRNA stability complex functions with poly(A)-binding protein to stabilize mRNA in vitro. Mol. Cell Biol. 19, 4552–4560. Fig. 6. Measurement of apparent K d values of iPABP for the A25, AU4 andM8RNAs.REMSAs were performed and binding curves con- structed as in Fig. 4. (A) The graph shows representative binding curves for iPABP with each RNA. (B) Apparent K d values with 95% confidence limits shown in brackets. The data are pooled from three experiments. For comparison, the apparent K d for PABP binding to each RNA, determined in Fig. 2, is also shown in the table. 456 R. T. Sladic et al. (Eur. J. Biochem. 271) Ó FEBS 2003 5. Tarun, S.Z. Jr & Sachs, A.B. (1996) Association of the yeast poly (A) tail binding protein with translation initiation factor eIF-4G. EMBO J. 15, 7168–7177. 6. Le, H., Tanguay, R.L., Balasta, M.L., Wei, C.C., Browning, K.S., Metz, A.M., Goss, D.J. & Gallie, D.R. (1997) Translation initia- tion factors eIF-iso4G and eIF-4B interact with the poly (A)- binding protein and increase its RNA binding activity. J. Biol. Chem. 272, 16247–16255. 7. Imataka, H., Gradi, A. & Sonenberg, N. (1998) A newly identified N-terminal amino acid sequence of human eIF4G binds poly(A)- binding protein and functions in poly(A)-dependent translation. EMBO J. 17, 7480–7489. 8. Tharun, S. & Parker, R. (2001) Targeting an mRNA for decap- ping: displacement of translation factors and association of the Lsm1p-7p complex on deadenylated yeast mRNAs. Mol. Cell 8, 1075–1083. 9. Wilusz, C.J., Gao, M., Jones, C.L., Wilusz, J. & Peltz, S.W. (2001) Poly (A)-binding proteins regulate both mRNA deadenylation and decapping in yeast cytoplasmic extracts. RNA 7, 1416–1424. 10. Xu, N., Chen, C.Y. & Shyu, A.B. (1997) Modulation of the fate of cytoplasmic mRNA by AU-rich elements: key sequence features controlling mRNA deadenylation and decay. Mol. Cell Biol. 17, 4611–4621. 11. Lagnado, C.A., Brown, C.Y. & Goodall, G.J. (1994) AUUUA is not sufficient to promote poly (A) shortening and degradation of an mRNA: the functional sequence within AU-rich elements may be UUAUUUA (U/A) (U/A). Mol. Cell Biol. 14, 7984–7995. 12. Vasudevan, S. & Peltz, S.W. (2001) Regulated ARE-mediated mRNA decay in Saccharomyces cerevisiae. Mol. Cell 7, 1191– 1200. 13. Putland, R.A., Sassinis, T.A., Harvey, J.S., Diamond, P., Coles, L.S., Brown, C.Y. & Goodall, G.J. (2002) RNA destabilization by the granulocyte colony-stimulating factor stem-loop destabilizing element involves a single stem-loop that promotes deadenylation. Mol. Cell Biol. 22, 1664–1673. 14. Grosset, C., Chen, C.Y., Xu, N., Sonenberg, N., Jacquemin- Sablon, H. & Shyu, A.B. (2000) A mechanism for translationally coupled mRNA turnover: interaction between the poly (A) tail and a c-fos RNA coding determinant via a protein complex. Cell 103, 29–40. 15. Jacobson, A. (1996) Poly (A) metabolism and translation: the closed loop model. In Translational Control (Hershey, J.W., Mathews, M.B. & Sonenberg, N., eds), pp. 451–480. Cold Spring Harbor Press, Cold Spring Harbor. 16. Tarun, S.Z. Jr & Sachs, A.B. (1995) A common function for mRNA 5¢-and 3¢ ends in translation initiation in yeast. Genes Dev. 9, 2997–3007. 17. Munroe, D. & Jacobson, A. (1990) mRNA poly (A) tail, a-3¢ enhancer of translational initiation. Mol. Cell Biol. 10, 3441–3455. 18. Uchida, N., Hoshino, S., Imataka, H., Sonenberg, N. & Katada, T. (2002) A novel role of the mammalian GSPT/eRF3 associating with poly (A)-binding protein in Cap/Poly (A) -dependent trans- lation. J. Biol. Chem. 277, 50286–50292. 19. Kuhn, U. & Pieler, T. (1996) Xenopus poly (A) binding protein: functional domains in RNA binding and protein–protein inter- action. J. Mol. Biol. 256, 20–30. 20. Burd, C.G., Matunis, E.L. & Dreyfuss, G. (1991) The multiple RNA-binding domains of the mRNA poly (A)-binding protein have different RNA-binding activities. Mol. Cell Biol. 11, 3419– 3424. 21. Nietfeld, W., Mentzel, H. & Pieler, T. (1990) The Xenopus laevis poly (A) binding protein is composed of multiple functionally independent RNA binding domains. EMBO J. 9, 3699–3705. 22. Deo, R.C., Bonanno, J.B., Sonenberg, N. & Burley, S.K. (1999) Recognition of polyadenylate RNA by the poly (A)-binding protein. Cell 98, 835–845. 23. Wiklund, L., Sokolowski, M., Carlsson, A., Rush, M. & Schwartz, S. (2002) Inhibition of translation by UAUUUAU and UAUUUUUAU motifs of the AU-rich RNA instability element in the HPV-1 late 3¢ untranslated region. J. Biol. Chem. 277, 40462–40471. 24.Bollig,F.,Winzen,R.,Gaestel,M.,Kostka,S.,Resch,K.& Holtmann, H. (2003) Affinity purification of ARE-binding pro- teins identifies polyA-binding protein 1 as a potential substrate in MK2-induced mRNA stabilization. Biochem. Biophys. Res. Commun. 301, 665–670. 25. Yang, H., Duckett, C.S. & Lindsten, T. (1995) iPABP, an inducible poly (A) -binding protein detected in activated human T cells. Mol. Cell Biol. 15, 6770–6776. 26. Bohjanen, P.R., Petryniak, B., June, C.H., Thompson, C.B. & Lindsten, T. (1991) An inducible cytoplasmic factor (AU-B) binds selectively to AUUUA multimers in the 3¢ untranslated region of lymphokine mRNA. Mol. Cell Biol. 11, 3288–3295. 27. Gorlach, M., Burd, C.G. & Dreyfuss, G. (1994) The mRNA poly (A) -binding protein: localization, abundance, and RNA-binding specificity. Exp. Cell Res. 211, 400–407. 28. Milligan, J.F., Groebe, D.R., Witherell, G.W. & Uhlenbeck, O.C. (1987) Oligoribonucleotide synthesis using T7 RNA poly- merase and synthetic DNA templates. Nucleic Acids Res. 15, 8783–8798. 29. Carey, J. (1991) Gel retardation. Methods Enzymol. 208, 103–117. 30. Deardorff, J.A. & Sachs, A.B. (1997) Differential effects of aromatic and charged residue substitutions in the RNA binding domains of the yeast poly (A)-binding protein. J. Mol. Biol. 269, 67–81. 31. Loflin, P., Chen, C.Y. & Shyu, A.B. (1999) Unraveling a cyto- plasmic role for hnRNP D in the in vivo mRNA destabilization directed by the AU-rich element. Genes Dev. 13, 1884–1897. 32. Wilson,G.M.&Brewer,G.(1999)Thesearchfortrans-acting factors controlling messenger RNA decay. Prog. Nucleic Acid Res. Mol Biol. 62, 257–291. 33. Sarkar, B., Xi, Q., He, C. & Schneider, R.J. (2003) Selective degradation of AU-rich mRNAs promoted by the p37 AUF1 protein isoform. Mol. Cell Biol. 23, 6685–6693. 34. DeMaria, C.T. & Brewer, G. (1996) AUF1 binding affinity to A+U-rich elements correlates with rapid mRNA degradation. J. Biol. Chem. 271, 12179–12184. 35. de Melo Neto, O.P., Standart, N. & Martins de Sa, C. (1995) Autoregulation of poly (A)-binding protein synthesis in vitro. Nucleic Acids Res. 23, 2198–2205. 36. Bag, J. & Wu, J. (1996) Translational control of poly (A)-binding protein expression. Eur. J. Biochem. 237, 143–152. 37. Le, H., Browning, K.S. & Gallie, D.R. (2000) The phosphoryla- tion state of poly (A)-binding protein specifies its binding to poly (A) RNA and its interaction with eukaryotic initiation factor (eIF) 4F, eIFiso4F, and eIF4B. J. Biol. Chem. 275, 17452– 17462. 38. Lee, J. & Bedford, M.T. (2002) PABP1 identified as an arginine methyltransferase substrate using high-density protein arrays. EMBO Report 3, 268–273. Ó FEBS 2003 Human PABP Binds AU-rich RNA via RRM 3 and 4 (Eur. J. Biochem. 271) 457 . chro- matography and poly(U) affinity column. Lane 1, PABP; lane 2, PABP1 2 34 ; lane 3, PABP1 2; lane 4, PABP3 4. Ó FEBS 20 03 Human PABP Binds AU-rich RNA via RRM 3 and 4 (Eur A25 (n M ) K d for AU4 (n M ) K d for M8 (n M ) PABP 0.67 (0.61–0. 74) 3. 9 (3. 7 4. 1) 7.7 (6.5–9.1) PABP1 2 34 0. 74 (0.61–0.81) 1 .3 (1.2–1 .4) 2 .4 (2.2–2.6) PABP1 2 1.8