The Expression of Poly(A)-binding Protein Gene Is Translationally Regulated in a Growth-dependent Fashion through a 5′-Terminal Oligopyrimidine Tract Motif*

Poly(A)-binding protein (PABP) is an important regulator of gene expression that has been implicated in control of translation initiation. Here we report the isolation and the initial structural and functional characterization of the human PABP gene. Delineation of the promoter region revealed that it directs the initiation of transcription at consecutive C residues within a stretch of pyrimidines. A study of the translational behavior of the corresponding mRNA demonstrates that it is translationally repressed upon growth arrest of cultured mouse fibroblasts and translationally activated in regenerating rat liver. Furthermore, transfection experiments show that the first 32 nucleotides of PABP mRNA are sufficient to confer growth-dependent translational control on a heterologous mRNA. Substitution of the C residue at the cap site by purines abolishes the translational control of the chimeric mRNA. These features have established PABP mRNA as a new member of the terminal oligopyrimidine tract mRNA family. Members of this family are known to encode for components of the translational apparatus and to contain an oligopyrimidine tract at the 5′ terminus (5′TOP). This motif mediates their translational control in a growth-dependent manner.

lator of gene expression that has been implicated in control of translation initiation. Here we report the isolation and the initial structural and functional characterization of the human PABP gene. Delineation of the promoter region revealed that it directs the initiation of transcription at consecutive C residues within a stretch of pyrimidines. A study of the translational behavior of the corresponding mRNA demonstrates that it is translationally repressed upon growth arrest of cultured mouse fibroblasts and translationally activated in regenerating rat liver. Furthermore, transfection experiments show that the first 32 nucleotides of PABP mRNA are sufficient to confer growth-dependent translational control on a heterologous mRNA. Substitution of the C residue at the cap site by purines abolishes the translational control of the chimeric mRNA. These features have established PABP mRNA as a new member of the terminal oligopyrimidine tract mRNA family. Members of this family are known to encode for components of the translational apparatus and to contain an oligopyrimidine tract at the 5 terminus (5TOP). This motif mediates their translational control in a growth-dependent manner. PABP 1 is the major cytoplasmic RNA-binding protein in eukaryotes that exhibits a preferential affinity for poly(A). This highly conserved protein has been implicated in regulating the initiation of translation ((1, 2) and references therein), mRNA stability (3), regulation of poly(A) tail length during the polyadenylation reaction (4,5), or poly(A) shortening (6,7).
Study of PABP gene expression in various vertebrates has established the respective mRNA as translationally controlled. Thus, serum stimulation of quiescent Swiss 3T3 cells seems to up-regulate the translation of PABP mRNA, as indicated by the resistance of the induction to actinomycin D treatment (8) and the lack of change in the level of PABP mRNA (9). Likewise, PABP mRNA is essentially sequestered in messenger ribonucleoprotein in quiescent duck reticulocytes (10), in mouse testis (11), and during early Xenopus embryogenesis (12).
TOP mRNAs encode for various components of the translational apparatus, like ribosomal proteins (rp) and elongation factors 1␣ and 2 (EF1␣ and EF2). These mRNAs are candidates for growth-dependent translational control mediated through a translational cis-regulatory element. This approximately 30nucleotide-long element is composed of a cytidine residue at the cap site followed by an uninterrupted stretch of up to 13 pyrimidines (13)(14)(15) and sequences immediately downstream (16,17).
Growth-dependent translational control of an mRNA generally correlates with the presence of an oligopyrimidine stretch at its 5Ј terminus. Yet, the linkage between these functional and structural features is not an absolute one. Thus, human ␤-tubulin mRNA is refractory to growth arrest in all of the examined cell lines, although it contains a bona fide translational cis-regulatory element including a 5ЈTOP, which is able to confer translational control on heterologous mRNA (18). In contrast, the 6.0-kb transcript of insulin-like growth factor II is translationally regulated in a growth-dependent manner (19), yet it does not contain a 5ЈTOP (20). In light of these exceptions, the structural basis for the translational control of PABP mRNA could not be unequivocally predicted. Moreover, the cumulative experience in assessing the regulatory role of 5ЈTOP motifs in the expression of various TOP genes has underscored the prerequisite to delimit the corresponding promoter regions (16,(21)(22)(23). Hence, to establish the structural base for the translational behavior of PABP mRNA, we set out to clone and characterize the corresponding human gene. Our structural and functional analyses have established the transcription start site of PABP gene within a stretch of pyrimidine leading to the production of a TOP mRNA with an exceptionally long 5Ј-UTR. The translational efficiency of PABP mRNA tightly correlates with growth conditions of cultured fibroblasts as well as rat liver. The translational cis-regulatory element has been delimited to reside within the first 32 nucleotides of this mRNA and involves the 5ЈTOP motif. * This work was supported from United State-Israel Binational Science Foundation Grant BSF-93-00032 and in part by Grant 3599 from the Chief Scientist's office of Ministry of Health, Israel. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ A recipient of awards from the Foulkes Foundation (London) and from the Kornfeld Foundation.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U68093-68105 for the promoter and the 15

EXPERIMENTAL PROCEDURE
Isolation of Genomic Clones-A human genomic library from W138 cell line of lung fibroblasts (Stratagene) was screened with a 32 P-labeled probe (1588-bp ScaI-EcoRI fragment) derived from the human PABP cDNA (24). The initial screen of 2 ϫ 10 5 plaques (one genome equivalence) yielded 33 positives, of which three different clones were isolated following a more stringent wash regime (0.3 M NaCl, 0.03 M sodium acetate at 67°C) in the second and third cycles of plaque purification. Restriction enzyme mapping and sequencing of the corresponding inserts established that one is pseudogene 1 and the other two are processed pseudogenes 2 and 3. Processed pseudogene 2 contains the entire sequence of PABP cDNA, whereas the nucleotide sequence determined in processed pseudogene 3 corresponds to positions 1941 to 2848 in this cDNA. Pseudogene 1 contains a sequence that corresponds to nucleotides 1357 to 1953 of PABP cDNA and is flanked by a nonrepetitive sequence.
A second screen of 8 ϫ 10 5 plaques of the same library was carried out using replicate Nytran filters (Schleicher & Schuell) with either a 263-bp EcoRI-StuI 5Ј-terminal region of PABP cDNA or a 200-bp XhoI-PstI fragment derived from a region flanking the PABP insert in pseudogene 1. Sixteen positive clones were isolated, and their DNAs were subjected to restriction enzyme digest and Southern blot hybridization with seven nonoverlapping probes spanning the entire PABP cDNA. Two overlapping phage clones were selected for further analysis: E, which hybridized with 5Ј-end and mid-region probes, and I, which hybridized with mid-region and 3Ј-end probes.
After preliminary mapping with restriction enzymes, selected fragments were excised and subcloned into plasmid vectors. To determine the sequence of the promoter as well as exon-intron junctions, all exons were sequenced using primer walking. Upon identification of each exonintron junction, a complementary primer was synthesized and used to sequence the opposite strand.
DNA Sequencing-Double-stranded plasmid DNA was sequenced either manually by the dideoxy method (25) using a Sequenase kit (U. S. Biochemical Corp., Cleveland, Ohio) or automatically by a DNA sequencer (ABI 377, Applied Biosystem Inc., Foster City, Ca).
Primer Extension-Determination of the transcription start site in hGH chimeric transcripts was carried out by primer extension as described previously (16).
Cell Culture and DNA Transfection-NIH 3T3 mouse fibroblasts were grown, transfected, and arrested as described (16).
Animals-Adult (6-to 8-week-old) male Sabra rats (Wistar origin) were obtained from the Hebrew University breeding center. Partial hepatectomy, resulting in the removal of 70% of the liver mass, was performed on male rats as described by Higgins and Anderson (26). Sham-operated rats were laparotomized, and their livers were manipulated but not excised.
Polysomal Fractionation and RNA Analysis-Harvesting and lysis of cells as well as size fractionation of polysomes by sedimentation through sucrose gradients were performed as described (27). When polysomal gradients were divided into two fractions (polysomal and the subpolysomal), RNA was extracted from each fraction by RNAzol B (Biotecx Laboratories, Houston, Texas), according to the supplier's instructions, and the poly(A) ϩ mRNA was isolated as described (28). In all cases where sucrose gradients were divided into 12 fractions, RNA was extracted as described (28) and analyzed without further enrichment through a oligo(dT) column. RNA (Northern) blot analysis was performed as described (29). Quantification of the radioactive signals on the blots was carried out by BioImaging Analyzer (Fujix BAS 1000, Fuji, Japan). To assess the effectiveness of the growth arrest treatment and the selectivity of the effect on TOP mRNAs, we compared in each case (even if not shown) the polysomal association of a chimeric mRNA with that of endogenous rp mRNA and non-rp mRNA from the same polysomal gradient. Only experiments in which both these controls exhibited their typical translational behavior (repressed and unrepressed, respectively) were included.
Plasmid Constructions-Standard protocols were used for all recombinant DNA technology (30).
pPABP-GH1 was constructed through the following steps. (i) An ϳ1.0-kbp PstI fragment, spanning positions Ϫ378 to ϩ635 of human PABP gene, was inserted into PstI site of pUC18 to yield pPABP-5Јf. (ii) A 450-bp fragment was generated by PCR (31) using oligonucleotide primers pUC-1 (16) and PABP-8 (CGGGATCCCACTCTCAGGACTA-ACC; boldface letters correspond to nucleotides ϩ32 to ϩ15 of PABP gene and are preceded by an underlined BamHI recognition site) and pPABP-5Јf as a template. (iii) The PCR-generated fragment was cleaved by HindIII and BamHI. The resulting 410-bp fragment was cloned between the HindIII and BamHI sites of the promoter-less plasmid p0GH, which contains the hGH gene (32). The resulting construct contains the hGH gene preceded by a human PABP sequence spanning positions Ϫ378 to ϩ32.
pPABP-GH2 was constructed through the following steps. (i) An ϳ2.45-kbp filled-in NotI XbaI fragment, spanning positions Ϫ378 to ϩ2070 of human PABP gene, was inserted between filled-in SalI site and SmaI site of pUC18 to yield pPABP-5Ј. (ii) pPABP-5Ј was cut by StuI and EcoRI, leaving a PABP sequence spanning positions Ϫ378 to ϩ265 linked to the plasmid. (iii) A 2.1-kbp BamHI (filled-in) EcoRI fragment containing the hGH gene from p0GH was inserted between StuI and EcoRI sites of pPABP-5Ј.
pPABP-GH3 was constructed by inserting a 266-bp filled-in AvaI-BamHI fragment (derived from pPABP-GH1), which spans position Ϫ234 to ϩ32 of the human PABP gene, inbetween filled-in SalI and BamHI sites of p0GH.
pPABP-GH4 was constructed by inserting a 683-bp ApaI-SacI (derived from pPABP-GH1) between filled-in SalI and SacI sites of p0GH. The GH in the resulting plasmid is preceded by the promoter region (positions Ϫ72 to ϩ32) of human PABP gene.
pPABP-GH6 was constructed through the following steps. (i) A 217-bp fragment was generated by PCR using oligonucleotide primers hGH-3 (16) and PABP-41 (GCTCTAGAGTGCGGCGCGGGGTAT; boldface letters correspond to nucleotides Ϫ43 to Ϫ28 of PABP gene and are preceded by an underlined XbaI recognition site) and pPABP-GH1 as a template. (ii) The PCR-generated fragment was made blunt-ended and then cleaved by BamHI, and the resulting 80-bp fragment was cloned between the HincII and BamHI sites of p0GH. The resulting construct contains the hGH gene preceded by a human PABP sequence spanning positions Ϫ43 to ϩ32.
The strategy used for the construction of pPABP-GH7 was similar to that described for pPABP-GH6 except for the usage of oligonucleotide PABP-42 (GCTCTAGAGCTCTTTCCTCCTGTT; boldface letters correspond to nucleotides Ϫ156 to Ϫ141 of PABP gene and are preceded by an underlined XbaI recognition site) instead of PABP-41. The resulting construct contains the hGH gene preceded by a human PABP sequence, spanning positions Ϫ156 to ϩ32.
pPABP-GH8 was constructed through the following steps. (i) An ϳ440-bp fragment was generated by PCR using oligonucleotide primers pUC-1 (16) and PABP-43 (CGGGATCCAAGCACCGCCTCCTGCA; boldface letters correspond to nucleotides Ϫ1 to Ϫ17 of PABP gene and are preceded by an underlined BamHI recognition site) and pPABP-GH1 as a template. (ii) The PCR-generated fragment was then cleaved by BamHI and HindIII, and the resulting 383-bp fragment was cloned between the HindIII and BamHI sites of p0GH. The resulting construct contains the hGH gene preceded by a human PABP sequence spanning positions Ϫ378 to Ϫ1.
pPABP-GH11s and pPABP-GH11(as) were constructed by annealing two complementary oligonucleotides: PABP-47 (GATCCTTCTC-CCCGGCGGTTAGTGCTGAGAGTGC; boldface letters correspond to nucleotides ϩ3 to ϩ33 of PABP gene) and PABP-48 (GATCGCACTCT-CAGCACTAACCGCCGGGGAGAAGG; boldface letters correspond to nucleotides ϩ34 to ϩ4 of PABP gene. The underlined nucleotides correspond to the protruding ends of BamHI site). The resulting doublestranded oligonucleotide was inserted into the BamHI site of PABP-GH8 in sense and antisense orientations, respectively. The structure of all constructs described here was confirmed by DNA sequencing.
Quantitative Analysis of Human Growth Hormone-The amount of growth hormone secreted by cells to the medium during the last 3 h of incubation was determined in an aliquot of the medium by radioimmunoassay with a commercial kit (Nichols Institute, San Juan Capistrano, CA).
Molecular Probes-The isolated fragment probes used in the Northern blot analysis were a 1.15-kb PstI fragment containing mouse ␣-actin cDNA (33), a 1.56-kb SacI-EcoRI fragment containing the 3Ј half of human PABP cDNA (24), a 1.2-kb HindIII fragment containing rat albumin cDNA (34), a 1.8-kb BglI fragment containing mouse EF1␣ cDNA (kindly provided by L. I. Slobin), and a 0.8-kb HindIII fragment containing hGH cDNA (kindly provided by T. Fogel, Bio-Technology General).

RESULTS
Organization of the Human PABP Gene-A human genomic library in Fix vector was screened as described under "Experimental Procedures." We isolated two overlapping clones, E and I (Fig. 1a), which collectively span an ϳ25-kb region that encompasses the PABP gene. Restriction fragment mapping and Southern blot analysis of the phages and various subclones together with DNA sequencing of selected segments disclosed the complete organization of the PABP gene (Fig. 1b).
The gene consists of 15 exons totaling 2.86 kb and 14 introns covering about 22 kb. The sequences flanking the exon-intron junctions conform to the consensus for 5Ј and 3Ј splice sites ( Table I).
The entire exonic sequence is 12 nucleotides longer than that of the published cDNA sequence (24). This dissimilarity is part of several differences summarized in Table II, of which the major one is an insertion of 9 nucleotides leading to the addition of 3 amino acids (Lys-Phe-Gly) following amino acid residue 212 in the published sequence (24). The sequence and the relative location of these three residues is identical to that found in mouse PABP protein and is similar to the corresponding sequence in Xenopus laevis protein (Table II). The isolated human gene encodes for a protein that is identical in size to that deduced from the mouse cDNA and differ by only three conservative replacements: Lys to Arg, Glu to Asp, and Ser to Thr at positions 176, 259, and 576, respectively (35). In light of the apparent evolutionary conservation of PABP amino acid sequence, it is conceivable that the missing three residues in the original report (24) is because of a sequencing error, whereas other differences might simply reflect polymorphism in the human population.
Interestingly, the PABP genes from Xenopus and human not only show a high degree of conservation of the amino acid sequence (93% of identical residues) but also an identical distribution of the four RNA binding domains among the different exons. Each of these functional domains are divided between two or three exons, which are the same exons in both species (Fig. 1a and Ref. 36).
5Ј and 3Ј Boundaries of the PABP Gene-Primer extension analysis of human PABP mRNA was carried out twice using lymphoblastoid cells. The extended DNA was electrophoresed on a sequencing gel alongside with sequencing reaction of either the cDNA clone or the pPABP-GH2. In both cases, they comigrated with nucleotides corresponding to position Ϫ2 and Ϫ1 with respect to the published sequence of human PABP cDNA (data not shown). Circumstantial evidence concerning the transcription start site has been obtained from sequence analysis of PABP processed pseudogene 2. This gene is flanked on both ends by direct repeats. It has been previously shown that in many cases the 5Ј repeat is separated by one nucleotide from the established cap site (37,38). Indeed, the 5Ј repeat is located just three nucleotides upstream of the 5Ј-end of the cDNA (Fig. 2). Taken together, we have concluded that the transcription of PABP gene initiates at two consecutive C residues, and the resulting mRNA contains a 5Ј-terminal oligopyrimidine tract of 12 or 11 residues (Fig. 2). Like most other TOP mRNAs, the distal cap site is preceded by a few pyrimidine residues, and the entire pyrimidine tract is flanked by GC-rich sequences (13). It appears, therefore, that PABP mRNA is a bona fide TOP mRNA.
The 3Ј-end of the PABP gene has been determined to reside at the nucleotide immediately preceding the poly(A) tail in both the PABP cDNA and the processed pseudogene 3 (Fig. 2).
It should be noted that scanning of the promoter and the intronic as well as the 3Ј flanking sequences of the human PABP gene against the EMBL data base has revealed significant homology with two ESTs. Thus, a sequence of the antisense strand within the promoter region (positions ϩ98 to Ϫ205) show 94.3% homology with a rat ovary mRNA of unknown function (accession number AI176738). Similarly, a sequence of the antisense strand spanning the first 108 nucleotides of exon 15 and the last 132 nucleotides of intron 14 exhibits 96.6% homology with a human testis mRNA of unknown function (accession number AI140680). Conceivably, these sequences result from reverse transcription of PABP mRNA molecules derived by the rare usage of an alternative upstream transcription start site (AI176738) or of a splicing intermediate (AI140680).
Functional Characterization of the PABP Promoter-Conceivably, the 5ЈTOP identified in the PABP gene might explain the previous observation concerning growth-dependent translational control of PABP mRNA. However, examining this possibility is based on assessment of the translational efficiency of chimeric mRNAs containing the 5ЈTOP of PABP mRNA. Formation of an mRNA with an authentic PABP 5ЈTOP requires the construction of a chimeric gene containing PABP promoter and the first transcribed nucleotides of PABP gene followed by the sequence of a reporter gene. Previous such analyses of several other TOP genes have disclosed critical transcriptional regulatory elements within the first exon or even the first intron (21). Hence, a prerequisite for functional analysis of the 5ЈTOP element is a functional characterization of the promoter region. To this end, we first sequenced the promoter region from position Ϫ378 (Fig. 3). PABP promoter includes a canonical TATAA box spanning positions Ϫ30 to Ϫ26. In that respect it is similar to the promoters of EF1␣ and EF2 genes but differs from those of most rp genes (13). Next, we constructed various chimeras harboring the hGH gene preceded by sequences con- taining portions of various lengths of the 5Ј-flanking sequences and 5Ј-UTR of the PABP gene. These constructs were transiently transfected into HeLa cells, and their expression was assessed by monitoring both the relative abundance of the resulting transcripts and the amount of hGH secreted into the medium. Fig. 4 shows that the promoter activity was not affected when the 5Ј-flanking sequence was shortened from 378 to 234 nt (PABP-GH3). It was decreased about 2-fold when further shortened to 156 nt (PABP-GH7), dropped abruptly when it included just 72 nt (PABP-GH4), and was completely abolished when only 43 nt were left (PABP-GH6). The complete loss of the promoter activity apparent in the last construct might reflect the omission of sequences perfectly matching the binding sites of ATF and SP1, two ubiquitous transcription factors (Fig. 3). Sequential deletion of sequences within the 5Ј-UTR led to 1.5fold decline in the abundance of hGH mRNA and the synthesis rate of hGH, when this region decreased from 486 nt (PABP-GH5) to 32 nt (PABP-GH1). A further decrease by 3-to 5-fold was detected upon complete deletion of transcribed PABP sequence (compare PABP-GH1 and PABP-GH8 in Fig. 4). It appears that a regulatory element residing within the first 32 nt of the transcribed region modulates the abundance of the GH transcript in an orientation-dependent manner, as its reinsertion in opposite orientation into pPABP-GH8 does not resume its activity (compare PABP-GH11(as) and pPABP-GH8 in Fig. 4). These results suggest that PABP promoter extends into the transcribed region, as has been previously shown for several other TOP genes (21,39). Yet, we cannot formally exclude the possibility that the first 32 nucleotides of PABP mRNAs play a role in stabilization of the transcript. Whatever the mechanism, based on these observations, we used in subsequent experiments a promoter region spanning positions Ϫ378 to ϩ32 or further downstream.
The Translational cis-Regulatory Element of PABP mRNA Resides within the First 32 Nucleotides and Is 5ЈTOP-dependent-Analysis of the polysomal distribution of PABP mRNA in NIH 3T3 cells reveals that this mRNA is subject to translation control in a growth-dependent manner. Thus, PABP mRNA is mostly associated with polysomes in growing cells and sequestered in subpolysomal fraction (messenger ribonucleoprotein particles) upon growth arrest (Fig. 5, endogenous PABP). Pre-   Ile  Ile  2406  3Ј-UTR  A  2621ʈ2622 3Ј-UTR A a The coordinates refer to those published for the cDNA (accession number y00345). b The sign ʈ depicts a site of insertion.

FIG. 2. The 3 and 5 boundaries of PABP gene.
Sequences at the 5Ј and 3Ј termini of human PABP cDNA were aligned with the corresponding genomic sequences of the active PABP gene, processed pseudogene 2, and processed pseudogene 3. White letters represent conserved nucleotides. Dots depict sequences not included in this alignment, and boxed sequences are repetitive elements delimiting the processed pseudogene. The 5Ј-end was determined by primer extension analysis of PABP mRNA, and the 3Ј-end was determined by the nucleotide preceding the poly(A) sequence in both PABP cDNA and processed pseudogene 3. vious attempts to delimit the translational cis-regulatory element of various TOP mRNAs have shown that it is confined to within their first 30 nt (16,18,23). However, the 5Ј-UTR of PABP mRNA (505 nt long) is considerably larger than those of other TOP mRNAs (an average of 40 nt). Hence, to delineate the corresponding region in PABP mRNA, we set out to examine the translational behavior of two mRNAs containing the first 265 nt (PABP-GH2) or just 32 nt (PABP-GH1) of PABP mRNA. A prerequisite for such an experiment with chimeric TOP mRNAs is the establishment of the transcription start site by primer extension analysis. Fig. 6 shows indeed that the major cap sites of the mRNA encoded by PABP-GH1 are at two consecutive C residues, coinciding with positions ϩ2 and ϩ3 in the human PABP gene. Likewise, the same start sites were identified for the PABP-GH2 construct, although they are embedded within a much larger PABP sequence and their identification required the use of a different primer (Fig. 6).
Analysis of the polysomal distribution of the mRNAs encoding PABP-GH1 and PABP-GH2 demonstrates that they are both translationally repressed upon growth arrest of NIH 3T3, as do endogenous mRNAs encoding PABP and EF1␣ (Fig. 5).
These results indicate that the first 32 nucleotides of PABP mRNA include all the regulatory elements required for conferring growth-dependent translational control on a heterologous mRNA. Our primer extension analyses have demonstrated that endogenous PABP mRNA and mRNAs encoding PABP-GH1 and PABP-GH2 do not start at the same nucleotide. However, in all cases, the major transcription start sites are at one or two C residues within a stretch of four consecutive pyrimidines. To examine whether substitution of the C residue at the cap site by purines affects the translational control, we constructed the PABP-GH11s gene. This gene is similar to that of PABP-GH1, except for the replacement of the C residues at positions ϩ1 and ϩ2 by the tetranucleotide GGAT. This change leads to the selection of the new A residue as the major transcription start site (Fig. 6). The fact that the A at the cap site is followed by a stretch of 11 pyrimidines is not sufficient to render this mRNA translationally regulated (Fig. 5). It appears, therefore, that translational control of PABP mRNA is strictly dependent on the location of the oligopyrimidine tract at the very 5Ј-end.
PABP mRNA Is Translationally Regulated during Liver Regeneration-To study the translational behavior of PABP mRNA in whole animals, we exploited the fact that when the liver reaches the adult stage it becomes quiescent, yet it retains the capacity to resume proliferation after partial hepatectomy. Hence, polysomes from sham-operated and regenerating liver were size-fractionated by sucrose gradient centrifugation, and the polysomal association of various mRNAs was assessed by Northern blot analysis of each gradient fraction (Fig. 7). Our The amount of hGH secreted during the next 3 h was measured in an aliquot of the medium. The relative abundance of hGH mRNA was assessed by Northern blot hybridization of mRNAs extracted from the cells immediately after aliquoting the medium. The radioactive signals were quantified by phosphorimaging. In the schematic presentations of PABP-GH constructs, the 5Ј-flanking sequence was denoted as a thin line, and exon 1 was denoted as a dotted box. The relative promoter activity of the chimeric genes was expressed either as the amount of secreted hGH or as the abundance of the corresponding mRNA. The results were normalized to those obtained for PABP-GH5 and are presented as average ϮS.E. of the number of measurements indicated in parenthesis. results demonstrate that the proportion of PABP mRNA associated with polysomes increased from 37% in sham-operated liver to 72% within 15 h after partial hepatectomy (Fig. 7). A similar recruitment into polysomes (from 40% in control to 88% in regenerating liver) is apparent also for EF1␣ mRNA, another TOP mRNA. In contrast, albumin mRNA was efficiently translated under both these conditions, despite a slight unloading from polysomes following the operation (from 94% in polysomes to 83%). It should be noted that the translational behavior of albumin mRNA is not exceptional, as any of the other four non-TOP mRNAs previously examined in the regenerating liver exhibited no translational activation (40). DISCUSSION Structural analysis of the human PABP gene as well as of chimeric genes driven by PABP promoter revealed that the transcription starts mainly at one or two C residues within an oligopyrimidine stretch. The presence of 12-nucleotide long 5ЈTOP appears to provide the structural basis for its growthdependent translational control in fibroblasts ( Fig. 5 and Refs. 8 and 9) or adult liver (Fig. 7), and thus, has established PABP mRNA as a new member of the TOP mRNA family. Interestingly, PABP 5ЈTOP sequence is unique among the TOP mRNAs, as it contains an exceptionally high ratio of C to T (3 to 1). In comparison, the average ratio in 31 vertebrate rp mRNAs is 1.03 Ϯ 0.08, and in five non-rp TOP mRNAs, it is 0.76 Ϯ 0.19 (13). Whether this high C content has an additional regulatory role is yet to be determined. Nevertheless, shifting the first C of PABP 5ЈTOP into position ϩ3 completely abolished the translational response to growth arrest (PABP-GH11s, Fig. 5). This is consistent with our previous report that the pyrimidine tract fails to exert its effect even when preceded by a single A residue (16). Formally, we cannot rule out the possibility that the loss of translational control is because of the concomitant shortening of the pyrimidine stretch from 12 to 11 residues. However, the latter explanation is less likely, as most 5ЈTOP motifs contain eight pyrimidines. Moreover, several 5ЈTOPs with just five pyrimidines have been shown to suffice the translational control mechanism (16 -18).
RNA-protein binding experiments have recently provided some clues concerning putative specific trans-acting factors that might be involved in the translational control of TOP mRNAs (41)(42)(43)(44)(45)(46). Nevertheless, the relevance of proteins that specifically bind to the 5ЈTOP or the adjacent downstream FIG. 5. The first 32 nucleotides of PABP mRNA are sufficient, and the integrity of the 5TOP is critical for the translational control. NIH 3T3 cells were stably transfected with the indicated chimeric GH constructs. Cytoplasmic extracts were prepared from growing (G) or nongrowing cells (NG) because of 24 h aphidicolin (5 g/ml) treatment. These extracts were centrifuged through sucrose gradients and separated into polysomal (P) and subpolysomal (S) fractions. RNA from equivalent aliquots of these fractions was analyzed by Northern blot hybridization with hGH cDNA for detection of the chimeric transcripts and the cDNAs for PABP, actin, and EF1␣ for the corresponding endogenous mRNAs. The sequences around the transcription start sites (designated by arrows) of the human PABP gene or of the respective chimeric construct are indicated at the left of the autoradiograms.
FIG. 6. Determination of the transcription start site of PABP-GH transcripts by primer extension. 5 g of poly(A) ϩ mRNA from NIH 3T3 stably transfected with the indicated chimeric PABP-GH constructs were annealed to 5Ј-end-labeled synthetic oligonucleotides complementary to nucleotides ϩ31 to ϩ12 of hGH gene (PABP-GH1 and PABP-GH11s) or to nucleotides ϩ48 to ϩ29 of human PABP gene (PABP-GH2). The primers were extended with avian myeloblastosis virus reverse transcriptase, and the extended products (P) were analyzed on a 6% acrylamide-urea gel alongside a dideoxy sequencing reaction (A, C, G, T) in which the same primer (unlabeled) was used. Large and small asterisks indicate major and minor transcription start sites, respectively. FIG. 7. PABP mRNA is translationally activated during liver regeneration. Polysomal (1 to 8) and subpolysomal (9 to 12) fractions from sham-operated (C) and regenerating (PH) liver (15 h postoperative) are shown. RNA isolated from these fractions was applied to Northern blot analysis and hybridized with labeled cDNAs encoding PABP, EF1␣, and albumin. For each treatment, the same RNA preparations were hybridized with the different probes. The autoradiographic signals were quantified by phosphorimaging, and the relative amounts of the mRNAs in each fraction are graphically depicted in the right panels.
sequences is still unclear, as the binding activity remains unchanged under various growth conditions, at which the translational efficiency of rp mRNAs is repressed or derepressed. In contrast, a recent report has demonstrated that direct interaction of repressor molecules with pyrimidine-rich sequences within the 3Ј UTR of 15-lipoxygenase mRNA mediates its translational silencing (47).
Another putative component of the translational control mechanism, which has recently attracted much attention, is p70 S6 kinase (p70 s6k ). Numerous studies have shown that mitogenic or hormonal stimulations induce phosphorylation of rpS6 by this enzyme with a concomitant derepression of the translation of TOP mRNAs (48). Furthermore, inhibition of p70 s6k by the immunosuppressant rapamycin or by expression of a dominant-negative p70 s6k mutant selectively repressed the translation of this class of mRNAs (49,50). These correlations have led to the assumption that growth-dependent translational control of TOP mRNAs involves a p70 s6k -mediated signal transduction pathway.
It is tempting to speculate that the growth-dependent translational control of PABP mRNA in fibroblasts and liver, as well as of those encoding rps and EF1␣, reflects corresponding variations in the amounts or the activities of the pyrimidine-binding protein and/or p70 s6k .
The 5Ј-UTR of PABP mRNA (505 nt) is substantially larger than the average length (44 Ϯ 4) of this region in other 36 vertebrate TOP mRNAs rigorously analyzed (13). An intriguing possibility is that sequences downstream of the 5ЈTOP serve an additional regulatory function(s). Indeed, human PABP mRNA contains an A-rich region spanning position 73 to 123, which is evolutionarily conserved from yeast (Ref. 51 and references therein). The identification of this motif led to the hypothesis that PABP mRNA is autogenously regulated at the translational level through binding of the resulting protein to the 5Ј-UTR (52). In vitro experiments have shown that in the addition of PABP to a cell-free translation system selectively inhibits the translation of PABP mRNA and that this repression is mediated through the A-rich region (51,53). Nevertheless, inactivation of poly(A) polymerase in yeast, which is followed by the loss of poly(A) and consequently to an increase in the ratio of PABP to poly(A), had only a minor effect on the total level of PABP (54). The regulatory role, if any, of this conserved motif in mammalian cells has yet to be determined.