Feedback inhibition of poly(A)-binding protein mRNA translation. A possible mechanism of translation arrest by stalled 40 S ribosomal subunits.

An adenine-rich cis element at the 5'-untranslated region (UTR) of Pabp1 mRNA is able to inhibit translation of its own mRNA. Similar inhibition of translation of a reporter beta-galactosidase mRNA is observed when the adenine-rich auto regulatory sequence (ARS) is placed within the 5'-UTR of this mRNA. For this translational control the distance of the ARS from the 5' cap is not important. However, it determines the number of 40 S ribosomal subunits bound to the translationally arrested mRNA. Inhibition of mRNA translation by this regulatory sequence occurs at the step of joining of the 60 S ribosomal subunit to the pre-initiation complex. Translational arrest of the ARS containing mRNA in a rabbit reticulocyte lysate cell-free system in the presence of exogenous Pabp1 protects the 5'-flanking region of the ARS from nuclease digestion. This protection depends on the binding of the 40 S ribosomal subunit to the mRNA. The size and the sequence of the nucleotide-protected fragment depends on the location of the ARS within the 5'-UTR. When the ARS is located at a distance of about 78 nucleotides from the 5' cap, a 40-nucleotide long region adjacent to the ARS is protected. On the other hand, when the ARS is moved further away from the 5' cap to a distance of approximately 267 nucleotides, a 100-nucleotide-long region adjacent to the ARS is protected from nuclease digestion. Nuclease protection is attributed to the presence of one or more stalled 40 S ribosomal subunits near the Pabp1-bound ARS.

Poly(A) binding protein 1 (Pabp1) 1 is a conserved, abundant, and ubiquitous polypeptide (1)(2)(3)(4)(5)(6)(7). It performs multiple functions related to mRNA translation and stability in eukaryotic cells. Pabp1 contains two distinct structural domains. Four tandemly repeated RNA-binding domains (RBD) at its N-terminal end are joined by a variable length linker region to the protein binding domain at the C-terminal end (3, 8 -11). The four RBDs differ in their poly(A)-binding ability. The first two RBDs show high affinity toward poly(A), whereas the third and fourth RBDs have high affinity for polypyrimidines (8,12). In addition, Pabp1 may bind to a heptameric consensus sequence ACUAAYA (13). However, its known functions can be attributed to its ability to bind poly(A) and several polypeptides (14 -18). The polypeptide partners of Pabp1 include eIF4G, eIF4A, eIF4B, eukaryotic release factor eRF3, Pabp-interacting proteins Paip1 and Paip2, Rna15, and poly(C) binding protein ␣CP1 (19 -26). A conserved highly structured C-terminal domain consisting of 74 amino acids is responsible for its ability to bind Paip1, Paip2, and eRF3 (10). The RBDs of Pabp1 are probably involved in binding to the eIF4G subunit of eIF4F complex, and the 5Ј cap to circularize the translating mRNA (10,17,18). The circularization process is believed to facilitate re-initiation of terminating ribosomes (10,11,17,18). Pabp1 also functions as a determinant of mRNA stability/degradation by interacting with the poly(C) binding protein ␣CP1 and a poly(A) nuclease (27)(28)(29). Considering these important functions, it is not surprising that the Pabp1 gene is essential for viability of yeast cells (15). In the absence of Pabp1 expression, joining of the 60 S ribosomal subunit to the 48 S pre-initiation complex is inhibited (15). Besides circularizing translating mRNA, Pabp1 may participate in additional steps of mRNA translation (30,31).
Considering the important cellular functions of Pabp1, it is not unexpected that its expression is regulated in eukaryotes. In fact, the Pabp1 gene behaves like a primary response gene. Stimulation of proliferation of mammalian cells in culture or in vivo results in increased Pabp1 synthesis (32)(33)(34)(35)(36)(37). Control of the cellular level of Pabp1 usually occurs by regulating the translation of its mRNA. Stimulation of proliferation of quiescent 3T3 cells by serum (34,35,37) or the same in liver cells by partial hepatectomy (37) results in the activation of translation of stored non-translated mRNAs. It is thought that an increase in the cellular Pabp1 level could alter translation and stability of some mRNAs. In addition to translational control, the stability of Pabp1 mRNA is altered when proliferating myoblasts withdraw from the cell cycle and differentiate into myotubes (38). Changes in the Pabp1 level is not the only way Pabp1 controls mRNA translation and stability. The interaction between Pabp1 and eIF4G could also be a target for regulation. In serum-stimulated Xenopus kidney cells, the association of Pabp1 with eIF4G is enhanced. This occurs by a distinct mechanism, independent of eIF4E phosphorylation-mediated increased interaction between eIF4G and eIF4E (39). Furthermore, the stability of Pabp1 is also regulated. Shut-off of host protein synthesis following poliovirus or coxsackievirus infection is achieved at least in part through the cleavage of Pabp1 and eIF4G by viral 2A proteases (40 -42). Pabp1 expression also responds to stress. In the nematode Caenorhabditis elegans the Pabp1 mRNA level increased 3-to 6-fold in response to stress (43). Therefore, several pathways exist to control the function of this important cellular protein.
To examine how Pabp1 mRNA translation is regulated in eukaryotic cells, several laboratories have focused their attention to its unusually long (ϳ500 nucleotides) 5Ј-UTR. Within its long 5Ј-UTR, it has at least two translational control elements. An oligopyrimidine tract is located at the 5Ј terminus (37). This makes Pabp1 mRNA a member of the TOP (terminal oligopyrimidine) family of mRNAs. Several members of this family code for various components of the translation apparatus like the ribosomal proteins and elongation factor Tu. The oligopyrimidine tract within the first 32 nucleotides of Pabp1 mRNA is necessary for translational control in mouse 3T3 fibroblasts but not in HeLa cells (37). The C residue at the 5Ј cap is essential for translational control. The second regulator of translation, a highly conserved oligo(A)-rich region, is present near the TOP sequence. In vitro protein binding studies show that Pabp1 binds to this region of its own mRNA (44 -46). This oligo(A)rich region acts as an autoregulatory sequence (ARS) for Pabp1 mRNA translation. Deletion of the ARS region enhances Pabp1 mRNA translation in both in vitro (44,46) and in vivo systems (47). The ARS region is sufficient to confer translational arrest of the heterologous reporter ␤-galactosidase mRNA (47). TOP and ARS may function in a cell-specific manner (37). It is likely that the two cis elements of translational control can work independently from one another. There is no detectable synergistic effect on the translational repression of the reporter ␤-galactosidase mRNA in HeLa cells when both TOP and ARS are present at its 5Ј-UTR (47).
In the present study we have investigated how binding of Pabp1 to the ARS prevents translation. Results presented here suggest that the ARS⅐Pabp1 complex prevents the movement of the 40 S ribosomal subunit past the ARS. Consequently, the 40 S subunit stalls near the Pabp1-bound ARS and fails to reach the initiation codon to form the 80 S initiation complex.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Several reporter plasmids suitable for expression in mammalian cells were constructed by fusing different regions of the 5Ј-UTR of Pabp1 mRNA with the ␤-galactosidase coding region. The vector-derived 5Ј-UTR of the ␤-galactosidase mRNA was deleted from the CMV-SPORT-␤-Gal plasmid (Life Technologies, Inc., Burlington, Ontario, Canada) by digesting it with NcoI and EcoRI restriction enzymes. PCR was used to amplify the regions from nucleotides 132 to 350 and 58 to 350 of Pabp1 mRNA, and the amplified products were ligated to the NcoI/EcoRI-digested vector fragment to generate the p350-Gal and ARS p350-Gal plasmids, respectively (47). The forward (sense) and the reverse (antisense) primers used in PCR for constructing the p350-Gal plasmid were catgaccatggTCCGCGTCT-CCCCCG and acgatgaattcCGCAGAACGGGGTCG, respectively. Similarly, the forward primer for constructing the ARS p350-Gal plasmid was catgaccatggCACATTTATTATTAAA, and the reverse primer was the same as described before for making the p350-Gal plasmid. The lowercase letters indicate the NcoI (ccatgg) and EcoRI (gaattc) recognition sequences and 5 additional nucleotides at the 5Ј-end of each primer. To generate the third reporter plasmid, where the ARS was moved closer to the initiation codon, the region between nucleotides 132 and 360 of Pabp1 mRNA was amplified with the same forward primer as the one used for generating the p350-Gal clone, and the sequence of the reverse primer was ACTCAACGGCCGCAGAAC. The 228-bp PCR product was digested with XmaIII (3). In a separate reaction the ARS region between nucleotides 58 and 146 was amplified by PCR. The forward primer with the XmaIII recognition sequence and additional 5 nucleotides at the 5Ј-end (shown by lowercase letters) and the reverse primer with the EcoRI recognition sequence and additional 5 nucleotides at the 5Ј-end were used for PCR. The sequences of these oligomers are catgacggccgCACATTTATTATTAAA and acgatgaattcCGGGGGAG-ACGCGGA, respectively. The PCR-generated ARS region was ligated to the XmaIII-digested 228-bp PCR product. The 316-bp-ligated product was purified from the agarose gel and ligated to the NcoI/EcoRI-digested CMV-SPORT-␤-Gal vector for cloning.
The PCR reactions were carried out in a 50-l volume containing 10 mM Tris-HCl, pH 9.0, 1.5 mM MgCl 2 , 50 mM KCl, 0.2% Triton X-100, and 200 M dNTPs, 200 ng of each primer, 2 units of Taq polymerase, 20 ng of pHU73 plasmid as the template for 35 cycles. Each cycle was at 94°C for 15 s, 50°C for 30 s, and 72°C for 1 min, and the final cycle was at 72°C for 5 min. The amplified products were purified using a PCR purification kit (Qiagen, Valencia, CA) and digested with the appropriate restriction enzymes. The digested product of the correct size was gel-purified using a kit (Qiagen) and used for constructing the reporter plasmids. In addition to the ␤-galactosidase reporter constructs, several plasmids were made for in vitro RNA synthesis. The p350-Gal and p350 ARS-Gal plasmids were digested with the NcoI and EcoRI restriction enzymes, and the Pabp1 mRNA 5Ј-UTR containing insert was isolated. The gel-purified insert was ligated to the NcoI/EcoRI-digested PGEM-T Easy (Promega, Madison, WI) vector to generate PGEM-p350 and PGEM p350-ARS clones. For measuring the ␤-galactosidase and Pabp1 mRNA levels by using RNase protection assay, a selected region of these mRNAs was cloned into the PGEM-T Easy vector. The region between nucleotides 734 and 1049 of the CMV-SPORT-␤-Gal vector was amplified by PCR. The sequence of the forward sense primer and the reverse antisense primers, including the NcoI and EcoRI adaptors (lowercase) were catgaccatggAGAGTCCAGGCCGAG and acgatgaattc-CGCCGGTCGCTACCA, respectively. The PCR product was gel-purified and ligated to the NcoI/EcoRI-digested PGEM-T Easy vector to generate the PGEM-␤-Gal plasmid. Transcription of this plasmid with SP6 polymerase in vitro produced an RNA complementary to the 3Ј-UTR of ␤-galactosidase mRNA. Similarly the PGEM-Pabp1 plasmid was made by subcloning the 539-bp-long NcoI/EcoRI fragment of the pHu73 plasmid into the PGEM-T Easy vector. Transcription of the NcoI-digested PGEM-Pabp1 by SP6 polymerase produced an RNA complementary to the Pabp1-mRNA. All clones were maintained in Escherichia coli DH5␣ cells, and the plasmid DNA was isolated using a endotoxin-free plasmid isolation kit (Qiagen) as per the manufacturer's instructions. The nucleotide sequence of the PCR-derived region of all plasmids was confirmed by automated DNA sequencing. The schematic maps of the plasmids are shown in Fig. 1 (see below).
Transfection of Cells-Approximately 1 ϫ 10 6 HeLa WS cells were grown on a 12-well dish in Dulbecco's modified Eagle's medium (Life Technologies, Inc., Burlington, Ontario, Canada), supplemented with 10% fetal calf serum (Life Technologies, Inc., Burlington, Ontario, Canada), and used for transfection. Subconfluent cultures were transfected in a serum-free Opti-MEM medium (Life Technologies, Inc.) with plasmid DNA. Approximately 1-2 g of DNA was incubated with 10 g of LipofectAMINE (Life Technologies, Inc.) in 400 l of Opti-MEM medium at 20°C for 30 min before being added to the cells. Cells were incubated with the DNA/liposome mixture for 6 h at 37°C, and then the DNA-containing medium was removed, and transfected cells were allowed to grow for 24 h in Dulbecco's modified Eagle's medium with 10% fetal calf serum. Cells were also co-transfected with 0.75 g of pCAT3 plasmid (Promega, Madison, WI) to monitor the transfection efficiencies between experiments by measuring SV40 promoter driven expression of the chloramphenicol acetyl transferase (CAT) gene.
Measurement of the ␤-Galactosidase and CAT Protein Levels-The transfected cells were washed three times with phosphate-buffered saline (phosphate-buffered saline, 150 mM NaCl, 16 mM Na 2 HPO 4 , 4 mM NaH 2 PO 4 , pH 7.2) and lysed directly with 200 l of a buffer containing 25 mM Tris-HCl, pH 6.8, 2% SDS, 1% 2-mercaptoethanol, 10% glycerol, and 0.01% bromphenol blue. The samples were subjected to SDS-polyacrylamide gel electrophoresis as described previously (33). The separated polypeptides were transferred to a nitrocellulose membrane, and levels of ␤-galactosidase and CAT polypeptides were determined by using mouse anti ␤-galactosidase and rabbit anti-CAT antibodies (Sigma Chemical Co., Oakville, Ontario, Canada), respectively. The antigen⅐antibody complex was visualized by horseradish peroxidaseconjugated anti-mouse or anti-rabbit antibodies using Lumi-Light ϩ Western blotting kit (Roche Molecular Biochemicals, Laval, Quebec, Canada) according to the manufacturer's instructions.
Subcellular Fractionation-Following transfection, the cells were lysed in 200 l of lysis buffer (10 mM MOPS, pH 7.2, 250 mM NaCl, 2.5 mM MgOAc, 0.5% Nonidet P-40, 0.1 mM phenylmethylsulfonyl fluoride, 200 g/ml heparin, 50 g/ml cycloheximide). After removing the nuclei and cell debris by centrifugation at 12,000 ϫ g for 10 min, the polysomes were pelleted from the 12,000 ϫ g supernatant by ultracentrifugation at 100,000 ϫ g for 1 h in a 75 Ti rotor (Beckman) at 4°C. The polysomal pellet was suspended in 200 l of lysis buffer (44). The supernatant fraction was used as the source for post-polysomal material. The RNA from both fractions was isolated in the presence of 4 M guanidine thiocyanate (48). Polysomal and post-polysomal RNA samples corresponding to equal numbers of cells were used for analysis. In some experiments, the 12,000 ϫ g supernatant faction was centrifuged in a 12-ml 10 -50% sucrose gradient, containing 25 mM HEPES, pH 7.0, 50 mM KCl, 2 mM MgOAc, 50 g/ml cycloheximide, 15 mM 2-mercaptoethanol at 40,000 rpm in a Beckman SW 41Ti rotor for 3 h (35). Gradient fractions of 1 ml each were collected using an Auto Densi-Flow IIC apparatus (Buchler Instruments). Total RNA from each fraction was isolated as described before and precipitated with ethanol using 5 g of yeast tRNA as the carrier.
Measurement of mRNA Levels-Total cellular RNA was isolated by using the High Pure RNA isolation kit (Roche Molecular Biochemicals) according to the manufacturer's instruction. The integrity of the RNA samples was analyzed by 1% agarose gel electrophoresis as previously described (35). Samples containing equal amounts of 28 S rRNA were used for measuring the level of ␤-galactosidase, Pabp1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNAs by RNase protection assay (47) using the Hybspeed RPA kit (Ambion, Austin, TX). The RNA probes were produced by SP6 polymerase transcription of the PGEM-Gal and PGEM-Pabp1 plasmids. Due to the presence of sequence derived from the vector, hybridization of the 381-nucleotide-long ␤-galactosidase probe and 605-nucleotide-long Pabp1 probe, respectively, produced 315-and 539-nucleotide-long RNase protected fragments. The RNA probe for measuring the GAPDH mRNA level was generated by T7 polymerase transcription of a commercially available clone (Ambion). Again, due to the presence of a short sequence that does not hybridize to the GAPDH mRNA, the 374-nucleotide-long RNA probe produced a shorter 316-nucleotide-long RNase protected fragment. The shorter RNase-protected fragment of the probe permitted us to assess the presence of any undigested probe in the RNase protected sample.
Cell-free Translation and Translation-dependent Micrococcal Nuclease Protection of RNA-To investigate the presence of stalled 40 S ribosomal subunits at a site downstream from the ARS in the translationally arrested RNA, this region's resistance to nuclease digestion was examined. The methods used in these studies were similar to those previously described to examine the ribosome binding sites of translating mRNA (50 -52). For the nuclease protection analysis of translating RNA-truncated Pabp1 mRNA corresponding to nucleotides 1-350 (H350) and 132-350 (p350) and the ARS region (nucleotides 58 -146) joined to the p350 (p350 ARS) were generated by in vitro transcription using T7 RNA polymerase (44). The pHu73 plasmid (3) was linearized with XmaIII, and the PGEM-p350 and PGEM-p350 ARS plasmids were linearized with EcoRI to use as a template in the in vitro reactions to produce, respectively, the H350, p350, and p350 ARS RNAs. In vitro transcription was performed using a Transprobe T kit (Amersham Pharmacia Biotech, Quebec, Canada) containing m7GpppG capping solution according to the manufacturer's protocol. A small amount of [␣-32 P]UTP (1 Ci) was added in a 25-l reaction to examine the synthesis of RNA. The synthesized RNA was separated on a 1.5% agarose gel and purified by using a Qiagen column. Approximately 50 ng of RNA was used in 50 l of mRNA-dependent rabbit reticulocyte lysate (RRL) cell-free translation system (44) (Promega, Madison, WI). Due to the absence of downstream initiation codon in the capped RNA samples, only the 40 S ribosomal subunits could join to form the preinitiation complex (49). However, due to the binding of 40 S subunits to the RNA, certain regions of the RNA were expected to be protected from micrococcal nuclease treatment. Following a 10-min incubation of RRL programmed with different RNA at 30°C, 20 units/l micrococcal nuclease (in 5 mM CaCl 2 , 50 mM glycine, pH 9.2) was added to the samples as indicated and incubated at 25°C for 30 min as previously described (50). The nuclease-protected RNA was phenol chloroform-extracted and precipitated with ethanol using 5 g of E. coli tRNA as carrier. The RNA was dissolved in 10 l of nuclease free water and used for oligonucleotide protection and primer extension analyses. Similar studies were also carried out in the presence of either exogenous excess Pabp1 or the cap analogue m7GMP during translation.
Oligonucleotide Protection and Primer Extension Analyses of Micrococcal Nuclease-resistant RNA Fragments-To determine whether the nucleotides downstream from the ARS were protected from micrococcal nuclease, the presence of this region in the nuclease-resistant RNA samples was examined. This was performed by oligonucleotide protection assay. The first oligonucleotide used in our studies was complementary to the downstream site (nucleotides 45-69) of the authentic ARS region, and the second oligonucleotide was complementary to nucleotides 326 -50 of Pabp1 mRNA. The second oligomer was complementary to the region downstream from the ARS of p350 ARS mRNA where the ARS was near the initiation codon. Each oligonucleotide had 10 additional bases at its 3Ј-end that were not complementary to the Pabp1 mRNA. The presence of the additional bases allowed us to distinguish between the nuclease-protected fragment and the probe due to the loss of non-hybridized nucleotides. The sequences of the oligomers were ATAATAAATGTGTGTTCCGAGCCCGagcacgtcga (CAP oligomer) and CGCAGAACGGGGTCGATCCACTGCCagcacgtcga (AUG oligomer). The lowercase letters indicate the non-complementary nucleotides. These probes were 5Ј-end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase (44). The radiolabeled probe (5 ng Ϸ 5 ϫ 10 5 cpm) was hybridized to the micrococcal nuclease-protected RNA samples and subjected to S1 nuclease treatment using the Multi-NPA kit (Ambion) according to the manufacturer's protocol. The S1 nuclease-protected fragment was analyzed by 8 M urea-15% polyacrylamide gel electrophoresis and autoradiography.
Micrococcal nuclease-protected RNA samples were also subjected to primer extension analyses as previously described (50). The AUG and CAP oligomers lacking the mismatched nucleotides were used for primer extension. The oligomer was 5Ј-end-labeled with [␥-32 P]ATP as described before and mixed with the RNA sample in a 25-l volume containing 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, and 0.25 mM of all four deoxynucleoside triphosphates. The reaction mixture was kept on ice before addition of 200 units of Superscript II reverse transcriptase (Life Technologies, Inc., Gaithersburg, MD). The reverse transcription was carried out at 45°C for 60 min. After completion of the reaction, an aliquot of the sample was analyzed by 8 M urea-polyacrylamide gel electrophoresis and autoradiography.

RESULTS
The Effect of ARS Position on Translation Arrest-We have earlier shown that the ARS is sufficient for translational arrest of a heterologous reporter mRNA and the majority of the translationally repressed mRNA migrated slower than the 80 S initiation complex in a sucrose gradient (47). This distribution profile of the non-translated mRNA suggests that the 80 S initiation complex formation was inhibited by the presence of the ARS at the 5Ј-UTR. To investigate the mechanism of translational arrest by the Pabp1⅐ARS complex in more detail, we tested the effect of the distance between the 5Ј cap and the ARS on mRNA translation and scanning of this region by the 40 S ribosomal subunit. Two ␤-galactosidase reporter constructs containing the Pabp1⅐ARS were made. For the first construct (ARS p350-Gal), the ARS was placed near the 5Ј cap. But in the second construct (p350 ARS-Gal) the ARS was moved further away from the 5Ј cap and was placed near the initiation codon without significantly changing the length or the sequence of the 5Ј-UTR (Fig. 1). We hypothesized that the Pabp1-bound ARS impedes scanning of the 5Ј-UTR by the 40 S ribosomal subunit. Therefore, the number of 40 S ribosomes attached to the translationally arrested mRNA may depend on the distance between the 5Ј cap and the ARS. One 40 S subunit covers ϳ24 -30 (50) nucleotides of the mRNA, therefore, several 40 S subunits could be accommodated between the 5Ј cap and the ARS of the p350 ARS-Gal mRNA. On the other hand, in the ARS p350-Gal mRNA, only one or two 40 S subunits may be accommodated due to the proximity of the 5Ј cap to the ARS. In both cases, the mRNA translation should be suppressed. However, due to the binding of several 40 S subunits to the p350 ARS-Gal mRNA, it may migrate faster than the ARS p350-Gal mRNA in a sucrose gradient.
To test this we have analyzed the ␤-galactosidase protein and mRNA levels and the distribution of mRNA in a sucrose gradient in transfected cells. The p350-Gal construct produced nearly eight times more ␤-galactosidase than that observed in the ARS p350-Gal-and p350 ARS-Gal-transfected cells ( Fig.  2A). There was no detectable difference in the ␤-galactosidase level in cells transfected with either of the two ARS-containing plasmids. For each experiment the cells were co-transfected with the pCAT3 plasmid, and the level of CAT was determined by immunoblotting ( Fig. 2A). The results show that the difference in ␤-galactosidase levels observed here was not due to variations in transfection efficiencies between experiments. To investigate whether translational control of the two ARS-containing mRNAs was responsible for the low level of ␤-galactosidase polypeptide synthesis in transfected cells, we measured the mRNA levels by RNase protection analysis. Because the ␤-galactosidase mRNA produced from all three constructs had common 3Ј-ends, we used the same RNA probe to measure these mRNA levels as described under "Experimental Procedures." As an internal standard we also measured the GAPDH mRNA levels in the same RNA samples. The results show that the ␤-galactosidase mRNA level in all three sets of transfected cells were similar. In each case the ␤-galactosidase mRNA level was approximately three times higher than the endogenous GAPDH mRNA level (Fig. 2B). These results suggest that the ␤-galactosidase polypeptide synthesis was controlled at the level of mRNA translation. Translation of both ARS p350-Gal and p350 ARS-Gal mRNAs was similarly affected by the ARS. Therefore, the effect of the ARS on mRNA translation was not influenced by its distance from the 5Ј cap.
Because translation of both ARS-containing mRNAs was repressed, we investigated whether the nature of their translation-arrested states was similar. Therefore, we examined whether these mRNAs can form the 48 S pre-initiation complex by binding to the 40 S ribosomal subunit. We also examined whether the number of 40 S subunits joined to the ARS-containing mRNAs was influenced by the distance between the ARS and the 5Ј cap. To achieve this, we first examined the distribution of mRNAs in a sucrose gradient under normal conditions. The RNA from different fractions of the gradient was isolated, and the level of individual mRNAs in each fraction was measured by RNase protection analysis using the appropriate probe. In transfected HeLa cells, the majority of ARS Ϫ galactosidase mRNA (Fig. 3, p350-Gal) was found in the polysomal fractions (7-12) of the gradient. In contrast, both ARS-containing mRNAs (Fig. 3, ARS p350-Gal and p350 ARS-Gal) were predominantly present in the non-polysomal fractions. There was, however, a difference in the distribution profile between the ARS p350-Gal and the p350 ARS-Gal mRNA. The ARS p350-Gal mRNA was present mostly in the 40 -60 S region of the gradient. A large amount of the p350 ARS-Gal mRNA was found in the slightly faster sedimenting 60 -80 S region of the gradient. This observation suggests that the precise translation initiation status of these two mRNAs Twenty-four hours after transfection, cells were lysed with appropriate buffer and used for determining either the ␤-galactosidase and CAT polypeptide levels by Western immunoblotting (A) or the ␤-galactosidase and GAPDH mRNA levels by RNase protection assay (B). The intensity of each band was determined by scanning using an imaging device (Bio-Rad, Mississauga, Ontario, Canada) and used for calculating the levels of different polypeptides and mRNAs. The ratios between ␤-galactosidase and CAT polypeptides and ␤-galactosidase and GAPDH mRNAs are shown at the bottom of the top panels. The arrows in A show the ␤-galactosidase and CAT polypeptide levels. The arrows in B show the RNase-protected fragments, and the arrowheads represent the probe prior to RNase treatment. Complete digestion of the probe by RNase in absence of hybridization with the target RNA is shown as (ϪRNA). may be different. We also examined how the endogenous Pabp1 mRNA was translated in exponentially growing HeLa cells. The distribution profile of this mRNA in a sucrose gradient (Fig. 3, Pabp1) shows that the Pabp1 mRNA was translated, albeit less efficiently than the ARS Ϫ p350-Gal mRNA and the endogenous GAPDH mRNA. A significant (nearly 30%) amount of the native Pabp1 mRNA was also present in the non-translated fractions (4 -6) of the gradient (Fig. 3). This amount was higher than those observed for the translated ARS Ϫ , p350-Gal, and GAPDH mRNAs. These observations suggest that the effect of the ARS on the endogenous Pabp1 mRNA translation was weaker than that observed for the reporter ␤-galactosidase mRNAs. The reason for this difference is not clear. But it is possible that an ARS-neutralizing downstream sequence may be present in the Pabp1 mRNA. In a previous study (47) we showed that a reporter ␤-galactosidase construct containing nearly the full-length 5Ј-UTR (487 nucleotides out of 502) had the same inhibitory effect on the reporter mRNA translation as that of the short oligo(A)-rich region located between nucleotides 58 and 132 of the Pabp1 mRNA 5Ј-UTR. Therefore, the neutralizing sequence may be present at a site further downstream from the region of the 5Ј-UTR used in our studies. Evidence for the presence of an ARS-neutralizing sequence was also observed by Hornstein et al. (37).
The results discussed above show that the ARS-containing reporter mRNAs were poorly translated. Although the distance of the ARS from the 5Ј cap had no detectable effect on the ␤-galactosidase polypeptide synthesis, it produced a small difference in sedimentation profiles of the two ARS-containing mRNAs. To investigate whether the mRNA found in fraction number 6 of the sucrose gradient (Fig. 3, p50 ARS-Gal) was present as the 80 S initiation complex, we used an inhibitor to block this step and examine its effect on the mRNA distribution profile. To achieve this, the transfected cells were treated with 10 mM NaF for 4 h before harvesting. Under these conditions, translation of both ARS-containing mRNAs should be blocked prior to the joining of the 60 S ribosomal subunit to the 48 S pre-initiation complex. Therefore, most of the mRNAs were expected to migrate as the 40 S ribosomal subunit bound 48 S pre-initiation complex. The results show that, as expected, most of the ARS p350-Gal mRNA migrated throughout the 40 and 60 S region (fractions 4 and 5) of the gradient (Fig. 4A). In contrast, most of the p350 ARS-Gal mRNA still migrated in the slightly faster sedimenting region (fractions 6 and 7) of the gradient (Fig. 4A). This result may be due to binding of more than one 40 S ribosomal subunit to the p350 ARS-Gal mRNA. In the ARS p350-Gal mRNA the ARS is only ϳ78 nucleotides away from the 5Ј cap. Therefore only one 40 S ribosomal subunit may be present in the translationally arrested ARS p350-Gal mRNA. On the other hand, the distance of ϳ230 nucleotides between the ARS and the 5Ј cap in the p350 ARS-Gal mRNA provides a longer scanning region prior to the ARS, which may allow binding of more than one ribosomal subunit to the mRNA. The sedimentation profile of these two mRNAs in NaF-treated cells agrees with this model. However, to clearly ascertain whether the faster sedimenting p350 ARS-Gal mRNA was associated with the 40 S ribosomal subunit, we used an inhibitor to prevent the formation of the 48 S preinitiation complex. Aurin tricarboxylic acid (ATA) was used to prevent the joining of the 40 S ribosomal subunit to the mRNA (53). Analysis of the mRNA distribution profile in ATA-treated cells by sucrose gradients shows that the majority of both ARS p350-Gal and p350 ARS-Gal mRNAs was present in fractions 3 and 4 of the sucrose gradient. This is where the non-translated free mRNA⅐protein complexes for these large mRNAs should migrate (54). The results also show that the difference in the distribution profiles of these ARS-containing mRNAs could be abolished by preventing the joining of the 40 S ribosomal subunit to the mRNA. Therefore, the difference in the migration properties between the ARS p350-Gal and p350 ARS-Gal mRNAs in normal and NaF-treated cells (Figs. 3 and 4A)  mRNAs, the number of 40 S ribosomal subunits that will scan the 5Ј-UTR before stalling at the Pabp1 bound ARS may depend on the distance between the 5Ј cap and the ARS. Our strategy to evaluate this was to examine the protection of the 5Ј-flanking region of the ARS from nuclease digestion by the initiating 40 S ribosomal subunits. When the ARS is bound to Pabp1, the 40 S subunit may stall before scanning the ARS. If this takes place, the 5Ј-flanking region adjacent to the ARS of the translationally arrested mRNA will be protected from micrococcal nuclease digestion (50). The presence of this protected RNA fragment could then be determined by S1 nuclease protection assay using an appropriate probe (50 -52). In addition, to determine how many 40 S ribosomal subunits were present near the ARS, the size of the protected fragment can be examined by primer extension analysis. In these studies we used short RNA fragments that lacked the initiation codon for in vitro translation. These RNAs should not bind to the 60 S ribosomal subunit. Therefore, any observed nuclease protection of the ARS adjacent region could be due to the binding of the 40 S ribosomal subunit.
For these studies we have used three different RNAs where the ARS region was either lacking or present in different locations within the 5Ј-UTR (Fig. 5A). One of these RNAs contained the first 350-nucleotide-long region of the 5Ј-UTR of the native Pabp1 mRNA (H350). The ARS in native Pabp1 mRNA is located at ϳ70 nucleotides from the 5Ј cap. The second RNA contained the region between nucleotides 132 and 350 of Pabp1 mRNA (p350) and lacked the ARS. The third RNA was similar to the second RNA, but the ARS region was placed at the 3Ј-end of this RNA (p350 ARS). This was done to check if moving the ARS to a different location than that in the native Pabp1 mRNA would result in the protection of its new 5Ј-flanking sequence. The micrococcal nuclease digestion of translating RNA was performed under normal translating conditions and also in the presence of exogenous Pabp1. To test whether the nuclease protection was translation-dependent, the cap analogue, m7GMP, was used in some experiments to prevent the joining of the 40 S ribosomal subunit to the capped RNA. The micrococcal nuclease protection of the sequences adjacent to the ARS was determined by the presence of an RNA fragment in the nuclease-treated samples, capable of hybridizing to a complementary oligodeoxynucleotide probe. The hybridization between the 32 P-labeled oligomer, and the micrococcal nuclease-protected RNA fragment was then examined by S1 nuclease protection of the probe. The S1 nuclease-protected probes are shown in Fig. 5B. The results of these studies (Fig. 5B) show that in the presence of excess Pabp1 the ARS containing H-350 and p350-ARS RNAs showed protection of the region adjacent to the ARS (lanes a and j). Prevention of translation by the cap analogue abolished the nuclease protection (lanes b and l). However, when translation was carried out without the addition of exogenous Pabp1 to the RRL, some protection was also observed (lanes c and k). Protection of the same region was not observed if the oligomer probe was not complementary to the immediate downstream region of the ARS of the translating RNA. Therefore, no detectable RNA fragment containing the sequence complementary to the CAP oligomer was observed in the translating p350 ARS and p350 RNA (lanes f and g). Also no protection of the region complementary to the AUG oligomer was found in the translating H350 and p350 RNAs (lanes h and i). Control experiments were also performed to test the ability of both CAP and AUG oligomers to hybridize to the untreated target RNAs. Results show that the presence of the RNA can be detected in our samples prior to micrococcal nuclease treatment (lanes d and m). The larger size undigested probes are also shown (lanes e and n). Results of our studies suggest that the presence of the ARS confers micrococcal nuclease protection to the 5Ј-adjacent region in a translationdependent manner. Furthermore, protection of this region was significantly enhanced by the addition of exogenous Pabp1 to the RRL. This observation is consistent with the presence of stalled 40 S subunits at the 5Ј-side of the ARS.
To determine the size of the micrococcal nuclease-protected fragment derived from the region adjacent to the ARS, we used nuclease-treated RNA samples for primer extension. The RNA fragment derived from the micrococcal nuclease digestion of the translating p350 ARS RNA was used for primer extension using the AUG oligomer, which is complementary to nucleo- tides 326 -350 of Pabp1 mRNA. Small amounts of a primerextended product ϳ100 nucleotides long was observed when the nuclease-treated RNA samples were derived from normal RRL translation conditions (Fig. 6, lane a). Addition of exogenous Pabp1 to the RRL enhanced the level of the primer-extended product (lane b). Formation of the primer-extended product was dependent on translation, because addition of the cap analogue almost abolished the product formation (lane c). The micrococcal nuclease-protected RNA fragment from the H350 and p350 RNA translation reactions did not yield any primer-extended product (lanes d and e). In all reactions, a smaller sized primer-extended product was also observed. This may represent the presence of some smaller RNA fragments in the samples resulting from nuclease protection by the binding of RRL proteins to the RNA. However, only the 100-nucleotidelong product showed stimulation by Pabp1 and inhibition by the cap analogue. Similar analysis using the CAP oligomer, which is complementary to nucleotides 45-69 of the Pabp1 mRNA showed a primer-extended product of ϳ40 nucleotides for H350 RNA samples. This primer extension was also enhanced by the addition exogenous Pabp1 in the RRL before translation and was inhibited by the cap analogue (lanes g and  h). Furthermore, primer extension of the CAP oligomer was not observed when micrococcal nuclease-treated RNA samples from p350 and p350 ARS RNA translation were used (lanes i and j). Collectively, these results suggest that a longer RNA fragment was protected from micrococcal nuclease digestion by the 40 S ribosomal subunits when the distance between the ARS and the 5Ј cap was increased.
As one 40 S subunit could protect approximately a 30-nucleotide-long region of the RNA (50), three to four 40 S subunits FIG. 5. Analysis of RNA fragments protected by translating ribosomes from micrococcal nuclease digestion. RNA containing the ARS at different distances from the 5Ј cap and RNA lacking the ARS were synthesized in vitro using T7 RNA polymerase. The corresponding region of the 5Ј-UTR of Pabp1 mRNA is shown by nucleotide numbers, and the location of the ARS is shown by the dark box in A. The H350 RNA had ϳ8 and the p350 ARS and p350 ARS RNA had ϳ37 vectorderived nucleotides at their 5Ј ends. The in vitro transcribed RNA was used in an mRNA-dependent rabbit reticulocyte lysate translation system as described under "Experimental Procedures." Following a 10-min incubation at 30°C to allow binding of the 40 S ribosomal subunits, the samples were treated with micrococcal nuclease at 25°C for 30 min, and the nuclease-protected regions of the RNA were isolated by phenol/ chloroform extraction. This RNA sample was used for S1 nuclease protection assay using appropriate 32 P-labeled oligodeoxynucleotide probes to determine the presence of nuclease-protected RNA fragments derived from the region adjacent to the ARS of the Pabp1 mRNA 5Ј-UTR (B). The conditions of translation are indicated at the top of the figure. The S1 nuclease-protected labeled oligonucleotide was analyzed by 8 M urea 15% polyacrylamide gel electrophoresis followed by autoradiography. A, the probe used in these studies was either the CAP or the AUG oligomer. The position of their hybridization to the RNA is shown below the RNA. The number refers to the nucleotides number of the Pabp1 mRNA. B, S1 nuclease protection of the oligomer probes under different translation conditions and in the absence of translation is shown by ϩ and Ϫ, respectively. ϩ, presence of 2.5 g of exogenous Pabp1 (44) and 10 mM m7GMP (50) during translation; Ϫ, absence during translation. Micrococcal nuclease-resistant RNA was hybridized with either the CAP oligomer (lanes a-d, f, and g) or with the AUG oligomer (lanes h-m). Hybridization of the probe with the RNA prior to incubation with RRL and micrococcal nuclease treatment is shown as controls (lanes d and m). The cap and AUG probes before S1 nuclease treatment are shown in lanes e and n, respectively. will be necessary to protect the 100-nucleotide-long region of the p350 ARS RNA. On the other hand, only one 40 S subunit may be sufficient to protect the 40-nucleotide-long region of the H350 RNA. These observations are consistent with the model of stacking of about three to four 40 S subunits prior to reaching the ARS of the p350 ARS RNA. Therefore, it appears that greater distance between the 5Ј cap and the ARS allows more 40 S subunits to scan this region before stalling at a site close to the ARS. It should be noted that the number of stalled 40 S subunits near the ARS was fewer than the maximum number that could be stacked between the 5Ј cap and the ARS of both H350 and p350 ARS RNAs. If the 40 S subunits were stacked to cover the entire 5Ј-UTR between the 5Ј cap and the ARS of these RNAs, longer primer extension products than those observed in our experiments should be produced. Why longer primer extension products were not seen is unclear. However, the number of 40 S subunits that will bind to an mRNA depends on many factors, including its intrinsic ability to bind the 43 S complex (49). Also the presence of secondary structures in the region between the 5Ј cap and the ARS may limit the rate of scanning and affect the clearance of the 40 S subunit from the 5Ј cap to allow binding of another 43 S complex (49). In addition, the in vitro cell-free translation conditions and the lack of a 3Ј-poly(A) tail may limit the rate of the initiation process. Therefore, a stacking of fewer 40 S subunits than that of the theoretical maximum in our experiments can be expected.

DISCUSSION
Pabp1 mRNA has two well characterized cis elements for regulating its translation. One of these consists of the terminal oligopyrimidine tract (TOP), and the second cis element is the oligo(A)-rich autoregulatory sequence (37,47). A third control element, which is not yet well characterized, may be located near the initiation codon and may be involved in neutralizing the effect of the ARS on Pabp1 mRNA translation (37). Under what circumstances each of these elements function to control Pabp1 mRNA translation is not clear. There appears to be a cell-specific preference for the choice of TOP or the ARS as the primary regulator of Pabp1 mRNA translation. How the ARS controls Pabp1 mRNA translation is better understood than that of the TOP-mediated inhibition of mRNA translation. The ARS mediates translational control by a feedback mechanism. Pabp1 itself binds to the ARS and represses translation of its own mRNA. It is expected that TOP binding proteins may be involved in TOP-mediated inhibition of mRNA translation. This protein may also be involved in providing the observed cell specificity toward the effect of TOP on Pabp1 mRNA translation. Although TOP binding proteins have been shown for some TOP mRNAs (55)(56)(57)(58)(59)(60)(61), their role in mRNA translation is uncertain, because the level of TOP binding activity does not change under conditions that inhibit translation of TOP mRNAs (37). It is not clear whether all TOP elements behave similarly. One unique feature of the TOP sequence of Pabp1 mRNA is the presence of the second cis element, the ARS, almost within the distance covered by one 40 S ribosomal subunit. Given the proximity of these two cis elements, their influence on each other cannot be ruled out. Furthermore, Pabp1 also shows binding affinity through the third and fourth RBDs toward polypyrimidines. So it is possible that Pabp1 also binds to the TOP sequence. It is intriguing to speculate that Pabp1, through its ability to recruit other proteins, forms a larger mRNA⅐protein complex to fine tune the cell-specific nature of its translational efficiency. Given the complexity of multiple cis regulators of Pabp1 mRNA translation, it is necessary to examine how each of the control elements functions in detail. The work presented here focused on examining the mechanism of ARS-mediated inhibition of translation. Also we have evaluated the influence of the location of the ARS within the 5Ј-UTR of mRNA on its translation and the scanning process.
We have shown here, using two different constructs, that the ARS was able to inhibit mRNA translation when the distance of the ARS from the 5Ј cap and AUG was altered. The results of our studies indicate that the 40 S ribosomal subunit stalls near the ARS. Although more 40 S subunits were able to bind when the distance between the ARS and 5Ј cap was increased, it did not affect the ability of the ARS to repress translation. The amount of protein produced by translation of the ARS-containing reporter ␤-galactosidase mRNA was independent of the position of the ARS within their 5Ј-UTR. It should be noted that a low level of translation of the ARS-containing ␤-galactosidase mRNA was always observed. This leaky translation arrest was not surprising, because the endogenous Pabp1 mRNA is also translated under normal growth conditions in HeLa cells. Under these conditions, generally more Pabp1 is found in HeLa cells than what is required for binding to all the available 3Ј-poly(A) tail (13), yet Pabp1 mRNA translation is not completely inhibited.
It is believed that when the cellular growth rate is reduced, leading to decreased mRNA and protein synthesis, less Pabp1 may be required to sustain the Pabp1-dependent vital cellular functions. Therefore, under these conditions the cellular level of Pabp1 is adjusted by repressing the translation of its mRNA using either the TOP or the ARS (or both) cis control elements. This regulation may be necessary to prevent overproduction of Pabp1, which could influence the growth cycle of cells. However, it is not clear why translational control of Pabp1 mRNA does not limit the production of Pabp1 to a level sufficient to bind all the available 3Ј-poly(A) tail. It is possible that Pabp1 is FIG. 7. Model for Pabp1-mediated regulation of its own mRNA translation. In the absence of excess Pabp1 to bind to the 5Ј-UTR of its own mRNA, normal translation proceeds. The 40 S ribosomal subunit migrates to the initiation codon, and the initiation complex is formed (a). Presence of excess Pabp1 results in its binding to the ARS region of its own mRNA. This prevents migration of the 40 S subunit. The presence of a stalled 40 S subunit is shown (b). More stalled 40 S subunits are present at the 5Ј-UTR of a mutant Pabp1 mRNA in which the distance between the 5Ј cap and the ARS is increased (c). involved in other cellular functions that do not require its binding to the 3Ј-poly(A) tail. This contention is supported by a recent report (62) that Pabp1 binds to a 395-nucleotide-long localization signal of vasopressin mRNA in nerve cells and aids in sorting this mRNA to dendrites. Therefore, considering the involvement of Pabp1 in many cellular processes, it is likely that its level must reach above a threshold to trigger feedback inhibition of its mRNA translation. Because the oligo(A)-rich region of the ARS has less than the ideal 12 adenine residues in tandem for efficient binding to Pabp1 (63), it might effectively compete with the 3Ј-poly(A) tail for binding to Pabp1 only after its cellular level is sufficiently high. Furthermore, the binding of Pabp1 to the poly(A) tail was shown to be dynamic in nature (64). This may permit Pabp1 to shuttle between the ARS and 3Ј-poly(A) tail allowing the 40 S subunit to occasionally pass the ARS to form the 80 S initiation complex and permit low level of mRNA translation. The different scenarios for autoregulation of Pabp1 mRNA translation are shown in Fig. 7. In the presence of excess Pabp1 the ARS is occupied by Pabp1. This may impede scanning of the 5Ј-UTR by the 40 S ribosomal subunit, resulting in the presence of stalled subunits near this site. When the ARS is moved further downstream from the 5Ј cap, the 40 S subunit may scan further along the 5Ј-UTR could permit more 40 S subunits to bind and migrate further downstream. Therefore, the number of 40 S subunits bound to this UTR may depend on the distance between the 5Ј cap and the ARS (Fig. 7).
In support of this model (Fig. 7) we have shown that, when the distance between the 5Ј cap and the ARS was increased in the reporter ␤-galactosidase mRNA (p350 ARS-Gal), faster sedimenting repressed mRNAs were formed. The faster sedimenting complexes were resistant to inhibition of translation by NaF, which prevents joining of the 60 S ribosomal subunits. However, their sedimentation profile was sensitive to ATA, which prevents joining of the 40 S ribosomal subunit. Therefore, it is believed that these faster sedimenting complexes were composed of mRNAs bound to multiple 40 S subunits. We have also shown that the micrococcal nuclease-protected sequence of the translating mRNA can be changed by moving the location of the ARS within the 5Ј-UTR. Irrespective of the distance between the 5Ј cap and the ARS, the 5Ј sequence adjacent to the ARS was protected from the micrococcal nuclease digestion in a translation initiation-dependent manner. Furthermore, when the distance between the ARS and 5Ј cap was sufficient to accommodate multiple 40 S subunits in translating mRNA, an RNA fragment of ϳ100 nucleotides in length showed nuclease protection. A smaller sized nuclease-protected RNA fragment was observed only when the ARS was much closer to the 5Ј cap.