Inhibition of Translation by UAUUUAU and UAUUUUUAU Motifs of the AU-rich RNA Instability Element in the HPV-1 Late 3′ Untranslated Region*

The human papillomavirus type 1 (HPV-1) late mRNAs contain a 57-nucleotide adenosine- and uridine-rich RNA instability element termed h1ARE in their late 3′ untranslated regions. Here we show that five sequence motifs in the h1ARE (named I–V) affect the mRNA half-life in an additive manner. The minimal inhibitory sequence in motifs I and II was mapped to UAUUUAU, and the minimal inhibitory sequence in motifs III–V was mapped to UAUUUUUAU. We also provide evidence that the same motifs in the AU-RNA instability element inhibit mRNA translation, an effect that was entirely dependent on the presence of a poly(A) tail on the mRNA. Additional experiments demonstrated that the h1ARE interacted directly with the poly(A)-binding protein, suggesting that the h1ARE inhibits translation by interfering with the function of the poly(A)-binding protein.

Human papillomaviruses (HPVs) 1 are a group of non-enveloped, double-stranded DNA tumor viruses with tropism for epithelial cells (1,2). Expression of the late mRNAs is restricted to the terminally differentiated cells in the upper layers of the epithelium and at least four papillomaviruses (bovine papillomavirus type 1 (BPV-1), HPV-1, -16, and -31) have been shown to contain cis-acting inhibitory RNA elements located in the late 3Ј UTR (reviewed in Refs. [3][4][5][6]. In addition, negative RNA elements have been identified in the HPV-16 L1 and L2 open reading frames (4,7,8).
We have previously identified and characterized an inhibitory AU-rich element (ARE) located in the HPV-1 late 3Ј UTR region named h1ARE (4 -6, 9) (Fig. 1). Using actinomycin D we showed that the presence of the h1ARE reduced the mRNA half-life (10). The minimal inhibitory sequence termed XB spans 57 nucleotides (nt) and contains 93% A and U. The element contains two AUUUA-and the three UUUUU-containing sequences (9,10). Replacing two uridines (U) with cytidines (C) in each motif inactivated the h1ARE (10). The h1ARE interacts with cellular factors (11,12), which bind to the c-fos ARE (10). Two of the h1ARE binding factors interacted with the wild type h1ARE but not with a functionally inactive mutant of the h1ARE (10). These proteins were identified as HuR and hnRNP C (10,13), and we later showed that binding of the HuR protein correlates with inhibitory activity of a panel of h1ARE mutants (13). While HuR binds to both AUUUA-and UUUUU-motifs (13), hnRNP C binds exclusively to the UUUUU-motifs (14). The role of hnRNP C in HPV-1 late gene expression is unclear. The HuR protein shuttles between the nucleus and the cytoplasm (15), and we observed that there was an inverse correlation between the levels of HuR in the cell cytoplasm and the inhibitory activity of the h1ARE (16), suggesting that the presence of high levels of HuR in the cytoplasm antagonizes the inhibitory effect of the h1ARE, whereas a primarily nuclear association of HuR is associated with inhibition of HPV-1 late gene expression. Interestingly, the HIV Rev and RRE, and the SRV-1 CTE can overcome the inhibition (9), suggesting that the h1ARE traps the HPV-1 late mRNAs in the nucleus and that this may lead to rapid mRNA degradation. Interestingly, the inhibitory effect of the h1ARE was greater at the protein level than at the mRNA level, suggesting that the h1ARE also inhibited the utilization of the mRNA. Here we present results of a mutational analysis of the h1ARE, and we provide evidence that the h1ARE inhibits mRNA translation.

Plasmid Constructions
CMV Promoter-driven Plasmids-All the eukaryotic expression plasmids containing the h1ARE mutants are derived from p⌬KX (11) (Fig.  1B). Oligonucleotides were annealed and cloned into p⌬KX digested with KpnI and XbaI, which resulted in the insertion in p⌬⌲X of the wild type and mutant h1ARE sequences displayed in the figures. All mutants were sequenced. pCMVlacZ was constructed by replacing the chloramphenicol acetyltransferase (CAT) gene in p⌬KX by the lacZ gene.

In Vitro Transcription and Transfections
DNA transfections were performed with FuGENE (Roche Molecular Biochemicals) as described previously (10). Transfections were performed in triplicates, and mean values and standard deviations are displayed in the figures. For RNA analysis triplicate samples were pooled and analyzed by Northern blot. Each plasmid was analyzed in at least three independent transfection experiments. RNA synthesis and transfections were performed as described previously (18).
To generate RNAs with a poly(A) tail of fixed length, two PCR fragments were first amplified with the following primer pairs: T7CATS (5Ј-GTAATACGACTCACTATAGGGTACTGCGATGAGTGGCAGGG-3Ј) and HPV1ANTIPA (5Ј-TTTTTTTTTTTTTTTTTTTTTTTTTTTTTT-TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTCACACTTGTGTATAATG-CACCGG-3Ј) (which encodes a A 60 tail) or T7CATS and HPV1ALW (5Ј-CACACTTGTGTATAATGCACCGG-3Ј). The PCR fragments that were generated from plasmid p⌬KX or p⌬KX containing the 57-nt h1ARE fragment using the two primer pairs described above were gel-purified and used for synthesis of capped RNAs with and without the A 60 tail. Plasmids or RNA encoding CAT, human growth hormone (hGH), or ␤-galactosidase were included in all transfections to monitor the transfection efficiency.

CAT-ELISA and hGH-ELISA
To monitor CAT protein levels, RNA-transfected HeLa cells were harvested as described previously (8) at indicated time points. For DNA transfections, cells were harvested at 40 h posttransfection. The levels of CAT, hGH, and ␤-galactosidase proteins were quantified using CAT, hGH and ␤-galactosidase antigen capture ELISA (Roche Molecular Biochemicals), respectively. All CAT quantitations were normalized to the protein concentration of the cell extract as determined by the Bradford method.

Primer Extension and Northern Blotting
Cytoplasmic RNA extraction and primer extension were performed as described previously (8). To perform Northern blotting, total cytoplasmic RNA was extracted at various times posttransfection as previously described (18). Northern blot analysis was performed as described (10). Briefly, 10 g of total or cytoplasmic RNA was separated on 1% agarose gels containing 2.2 M formaldehyde, followed by transfer to a nitrocellulose filter and hybridization. Random priming of the DNA probe was performed using a DECAprime kit (Ambion) according to the manufacturer's instructions.

UV Cross-linking and Preparation of Recombinant Protein
UV cross-linking and synthesis of radiolabeled RNA were performed as previously described (11,19). GST-PABP, GST-HuR, and GST-PCBP were purified on GS beads according the manufacturers recommendations (Amersham Biosciences).

RESULTS
Mutational Analysis of the h1ARE-To study the HPV-1 late 3Ј UTR element named h1ARE ( Fig. 1A) further, the 57-nt minimal element (XB) (Fig. 1B) or a functionally inactive mutant thereof (AUM/UM) (Fig. 1B) was inserted between the CAT reporter gene, driven by the human cytomegalovirus promoter, and the late HPV-1 poly(A) signals resulting in pXB and pAUM/UM (Fig. 1B). As controls, we used p⌬KX (which lacks the major part of the late 3Ј UTR), pCCKH1 (which contains the entire late 3Ј UTR), and pCCKH1(A) (which contains the region of the late 3Ј UTR containing the XB sequence in antisense orientation). HeLa cells were transfected in triplicate, the CAT production was monitored in each plate, and mean values and standard deviations are displayed in the figures. pCMVlacZ was included as an internal control in all transfections, and the variation was less than 20% between the samples in the triplicates. For RNA analysis, cytoplasmic extracts from the triplicates were pooled and subjected to RNA extraction and Northern blotting.
Analysis of CAT production in transient transfections of HeLa cells and calculation of fold difference between pCCKH1(A)/pCCKH1 and pXB/pAUM/UM revealed that the fold difference between pCCKH1(A)/pCCKH1 and pXB/ pAUM/UM were similar and demonstrated that the 57-nt XB contains the major inhibitory sequence (data not shown). The mRNAs containing the 57-nt XB fragment have a short half-life ( Fig. 1C). In agreement with our previous findings, the presence of the h1ARE on the mRNA also results in a higher ratio of nuclear versus cytoplasmic mRNA (data not shown). We concluded that pXB and pAUM/UM could be used for further studies of the AU-rich element.
The XB sequence contains two AUUUA motifs and three AUUUUUA motifs that were numbered I-V ( Fig. 2A). These motifs were all mutated in AUM/UM ( Fig. 2A). To investigate if all motifs were required for inhibition, the motifs were mutated one by one ( Fig. 2A). However, there was only a modest increase in CAT RNA and protein levels (Fig. 2, B and C) for each mutant. To obtain a clearer answer on the importance of each motif, consecutive mutations were introduced in the motifs (Fig. 3A). The results revealed that higher expression levels were obtained as more motifs were mutated (Fig. 3, B and C), demonstrating that each motif contributed to inhibition. However, mutations in four of the motifs (pM4) yielded as high expression levels as mutations in all five motifs (pAUM/UM) (Fig. 3, B and C). To investigate if the fifth motif contributed to inhibition, it was mutated in pM2 ( Fig. 3A), resulting in pM2V (Fig. 3A). The results revealed that pM2V expressed higher RNA and protein levels than did pM2. These levels were similar to those produced from pM3 (Fig. 3, B and C), thereby demonstrating that also the fifth motif contributed to inhibition. Interestingly, the effect was greater at the protein level than at the RNA levels, for all analyzed mutants (Fig. 3, B and C). We concluded that all five motifs contributed to inhibition in an additive manner.
Determination of the Minimal Inhibitory Sequence of Each Sequence Motif within the h1ARE-To determine the minimal inhibitory sequence of the AUUUA-containing motifs, point mutations were introduced in motifs I and II (Fig. 4A). Mutations in the tri-U nucleotides, the flanking As or the Us immediately flanking the As were not well tolerated (Fig. 4, B and C). However, mutations in the second nucleotide position outside the As (Fig. 4A), did not significantly affect the inhibitory activity of these motifs (Fig. 4, B and C), indicating that the smallest inhibitory motif was UAUUUAU. The two UAUUUAU were separated by a four-nucleotide spacer sequence (Fig. 4A). Substituting this sequence with four Cs did not affect the inhibitory activity of the h1ARE (Fig. 4, B and C), indicating that this spacer sequence did not contribute to the inhibitory activity.
To determine the minimal inhibitory sequence of the penta-U motifs, the nucleotides flanking the penta-Us were mutated (Fig. 5A). The results revealed that mutations in both the As flanking the penta-Us and the Us flanking the As resulted in higher CAT protein and RNA expression levels (Fig.  5, B and C). Substituting two Us in the penta-U sequence with two Cs had the strongest effect, whereas mutations in the Us flanking the As had the smallest effect on the inhibitory activity (Fig. 5, B and C). Therefore, the minimal motif was UAUUUUUAU.
UAUUUAU and UAUUUUUAU Motifs Can Functionally Substitute for One Another-The h1ARE may be divided into the B2 region with the two UAUUUAU motifs and the C1 region with the three UAUUUUUAU motifs (Fig. 6A). The B2 region, although less inhibitory than the XB sequence, as expected, inhibited CAT production and reduced mRNA levels to the same extent as the did C1 region (Fig. 6, B and C). Two B2 or two C1 regions were as inhibitory as the entire XB (Fig. 6, B and C), demonstrating that one type of motif (UAUUUAU or UAUUUUUAU) could substitute for one another. Furthermore, if the two AUUUA motifs were extended to two AUUUUUA motifs by insertion of two Us in each motif, resulting in pAUUUUUA (Fig. 7A), or if the three penta-U motifs were all shortened to contain only three Us, as in pAUUUA (Fig. 7A), the resulting inhibitory elements were nearly as inhibitory as the wild type h1ARE (Fig. 7, B and C). Both B2 regions and C1 regions acted by reducing mRNA steady state levels and protein production. Inhibition was also greater at the protein level than at the mRNA level. Taken together, the results demonstrated that both motifs acted in a similar manner.
Multiple Copies of the HPV-1 AU-rich Element Inhibits Protein Production Ͼ99%-For all mutants that retained inhibitory activity, we observed that the inhibitory effect was greater at the protein level than at the mRNA level (see Fig. 2, B and C, Fig. 3, B-D, Fig. 4, B and C, Fig. 5, B and C, Fig. 6, B and C, and Fig. 7, B and C). In other words, the mRNAs that contained the h1ARE, or partially active mutants thereof were utilized less efficiently by the translation machinery than mRNAs lacking the h1ARE or mRNAs containing functionally inactive sequences. To better compare the effect on protein production and the effect on the mRNA levels, multiple XB sequences were inserted into the reporter plasmid pXB, resulting in p2XB, p3XB, and p4XB (Fig. 8A), and the fold inhibition at the protein and mRNA levels was separately plotted against the number of XBs on the mRNA. As can be seen, there is a gradual decrease in CAT protein and mRNA levels for every inserted copy of the XB fragment (Fig. 8, B and C), and the fold inhibition was greater at the protein level than at the RNA level (Fig. 8, B and  C). Extrapolation of the data in a semilog plot allowed a better estimate of the inhibitory activity of one single XB sequence at protein and RNA levels (Fig. 8D) within the context of these plasmid constructs. The results clearly demonstrated that XB had a greater effect at the protein level than at the RNA level (Fig. 8D). On average from multiple experiments, we found that CAT protein levels were reduced 3.7-fold per XB and RNA levels 1.4-fold per XB in the context of the mRNAs with multiple XBs (Fig. 8D). We concluded that in addition to the effect on mRNA half-life, mRNAs carrying the h1ARE are inefficiently utilized for translation.
Deadenylation (20,21) could potentially cause the inhibition of translation observed here. To determine the poly(A) tail length of the mRNAs shown in Fig. 8C, they were subjected to oligonucleotide-directed RNaseH cleavage with the "RNaseH oligo" shown in Fig. 8A followed by Northern blot using a probe located downstream of the RNaseH oligo. The results revealed that all mRNAs contained poly(A) tails of similar length (Fig.  8E), demonstrating that inhibition of translation was not a result of deadenylation.
The HPV-1 AU-rich Element Inhibits Translation of the Transfected mRNAs-To study the effect of the h1ARE on translation further, we replaced the CMV immediate-early promoter with the bacteriophage T7 promoter and the cleavage and polyadenylation signal with the XhoI restriction site in the reporter constructs pCCKH1(A) and pCCKH1 (Fig. 1B) used previously to study the h1ARE. These cloning steps resulted in pCC(A) and pCC (Fig. 9A) that were linearized with XhoI and used as templates for in vitro synthesis of capped and polyadenylated CC(A) and CC mRNAs as described under "Experimental Procedures." These mRNAs contained the same sequences as the mRNAs produced by the pCCKH1 and pCCKH1(A) plasmids in the nuclei of transfected cells (Fig.  1B). Capped and polyadenylated CC and CC(A) mRNAs were transiently transfected in triplicates into HeLa cells by electroporation, and the CAT levels produced at 20 h posttransfection were quantified by using a CAT ELISA. The results revealed that CC mRNAs that contain the h1ARE in sense orientation produced ϳ15-fold lower CAT protein levels than the CC(A) mRNAs, which contain the h1ARE in antisense orientation (Fig. 9A). Cotransfected hGH-encoding mRNAs included in all samples as an internal control produced similar hGH protein levels (Fig. 9A).
Next, aliquots of electroporated cells were harvested at different time points, and the levels of CAT protein were monitored and plotted against time (Fig. 9B). Two interpretations of the results shown in Fig. 9B were appropriate: either the mRNAs with the HPV-1 late 3Ј UTR were rapidly degraded, preventing further CAT protein synthesis, or the mRNAs were not available for further rounds of translation as a result of a direct inhibition of translation, presumably by factors binding to the h1ARE.
To investigate if the half-lives of the transfected mRNAs were reduced by the presence of the HPV-1 late 3Ј UTR, CC and CC(A) mRNA levels at 1, 3, 5, and 23 h posttransfection were determined by primer extension (Fig. 9C). CC and CC(A) mRNAs decayed at similar rates, as detected at 1, 3, and 5 h posttransfection (Fig. 9C). Longer exposures also detected similar amounts of CC and CC(A) mRNA at 23 h posttransfection (data not shown). The mRNA half-lives were calculated to be 2.6 h for CC mRNAs and 2.7 h for CC(A) mRNAs. A number of control experiments were performed using CC(A) and/or CC mRNA. These experiments verified that electroporated RNAs were not sticking to the outside of the cell and that the majority of the transfected mRNAs are normally utilized by the translation machinery (data not shown). We concluded that the inhibitory effect on the transfected mRNAs, mediated by the HPV-1 late 3Ј UTR, was not a result of reduced mRNA half-life in the cytoplasm and that the HPV-1 late 3Ј UTR acted by inhibiting mRNA translation. In addition, the previously observed effect on the mRNA half-life was dependent on a nuclear experience of the mRNA and was not seen when in vitro synthesized mRNAs were transfected into cells.
Inhibition of Translation by the h1ARE Requires Intact UAUUUAU or UAUUUUUAU Motifs in the h1ARE-To confirm that the 57-nt minimal XB sequence containing the two UAUUUAU and the three UAUUUUUAU motifs inhibited translation in the RNA transfection experiments, CCXB and CCAUM/UM mRNAs (Fig. 10A) were transfected into HeLa cells in parallel. The CCAUM/UM mRNAs produced 41-fold higher levels of CAT than the CCXB mRNAs (Fig. 10B). hGH protein levels produced from the internal control mRNAs were similar in both samples (Fig. 10B). Next, capped and polyadenylated CCXB, CCAUM/UM, CCB2, and CCC1 mRNAs ( quantified. hGH mRNAs were included in all samples as an internal control. Fig. 10C shows that XB-containing mRNAs produced lower CAT protein levels than the AUM/UM-containing mRNAs as expected, whereas B2-and C1-containing mRNAs showed similar intermediate inhibition of CAT protein production compared with AUM/UM-and XB-containing mRNAs (Fig. 10C). hGH protein accumulation in the cell culture medium was similar in all transfected samples at each time point (Fig. 10C).
The CAT protein production peaked at the 23 h time point for all four mRNAs, after which the CAT protein levels decreased. This is the expected result if the mRNAs have similar halflives. The mRNA decay rates in the cells transfected with CCXB and CCAUM/UM mRNA were determined by primer extension on RNA extracted from the transfected cells. Fig.  10D shows that the levels of CCXB and CCAUM/UM mRNAs were similar at 1 and 2.5 h posttransfection and decayed with a rate comparable to that observed for CC and CC(A) mRNAs (compare Figs. 10D and 9C). The results confirmed that the h1ARE acts primarily by reducing mRNA translation in the RNA transfection experiments performed here and demonstrated that both the two UAUUUAU motifs and the three UAUUUUUAU motifs inhibit translation.
The h1ARE Inhibits Translation of mRNAs Carrying a Cap and a poly(A) Tail-The poly(A) tail is required for efficient mRNA translation (23). We therefore wished to investigate if the HPV-1 AU-rich element interfered with the function of the poly(A) tail. We transfected polyadenylated and unpolyadenylated capped CC and CC(A) mRNAs and monitored the levels of CAT protein in the cells at 24 h posttransfection. Fig. 11A shows that the polyadenylated CC mRNA, which contains the HPV-1 AU-rich element in sense orientation, produced 56-fold lower CAT protein levels than polyadenylated CC(A) mRNA, which contains the HPV-1 AU-rich element in antisense orientation. In contrast, the presence of the h1ARE on the unpolyadenylated CC mRNA did not inhibit CAT production significantly (Fig. 11A). The stimulatory effect of the 3Ј-poly(A) tail, when added to the capped CC mRNA, which contains the h1ARE, was only 1.5-fold, compared with 58-fold when added to the CC(A) mRNA. Translation of co-transfected hGH mRNA was similar in all samples (Fig. 11A). Similar results were obtained in a time course experiment using the XB sequence and the inactive AUM/UM mutant (Fig. 11B), demonstrating a connection between the 57-nt XB sequence and the poly(A) tail. Taken together, the results demonstrated that inhibition of CAT production by the h1ARE was dependent on a 3Ј-poly(A) tail. Therefore, the results showed that the stimulatory effect on translation mediated by the 3Ј-poly(A) tail on cap-dependent translation was perturbed by the h1ARE.
Also here could the inhibitory effect of the AU-rich RNA element on translation be indirect through deadenylation. To investigate if the h1ARE promoted deadenylation of the in vitro synthesized mRNAs that were transfected into the HeLa cells, mRNAs with a poly(A) tail of fixed length were transfected into the cells and the length of the poly(A) tail was determined at 1 or 4 h posttransfection by Northern blotting. Capped mRNAs with and without the h1ARE or poly(A) tail were analyzed. The results revealed that the mRNAs were not deadenylated at 1 h posttransfection (Fig. 11C). In addition, analysis of the polyadenylated mRNAs containing the h1ARE at 4 h posttranfection showed that the poly(A) tails remained intact (Fig. 11C). Therefore, inhibition of translation was not caused by deadenylation.
The h1ARE Interacts with the Poly(A)-binding Protein-Having established that the h1ARE inhibited the function of the poly(A) tail, it was reasonable to speculate that the h1ARE, or h1ARE binding factors, interact directly with the poly(A)-binding protein (PABP). We therefore tested if PABP bind directly to the XB RNA. GST-PABP was UV cross-linked to XB RNA or AUM/UM RNA. UV cross-linking of GST-PABP revealed that GST-PABP bound strongly to the XB RNA but only weakly to the AUM/UM RNA (Fig. 12A); GST-PABP did not bind to an unrelated RNA derived from the L1 coding region in HPV-16 (Fig. 12A). The GST-HuR protein was used as positive control and interacted only with XB, as expected (13), and GST-PCBP did not bind to any of the RNA sequences (Fig. 12A). To confirm that the binding of GST-PABP to the XB sequence was sequence specific, a competition experiment was performed. The XB RNA competed efficiently with the RNA probe for binding to GST-PABP, whereas the HPV-16 L1-derived RNA did not (Fig. 12B), demonstrating that the interaction with XB was sequence-specific. Analysis of the deletion mutants B2 and C1 that were shown to inhibit translation to a similar extent also interacted with the PABP (Fig. 12A). Therefore, binding of PABP to the different mutants (B2, C1, and AUM/UM)) correlated with their inhibitory effect on translation. DISCUSSION The mutational analysis presented here revealed that the HPV-1 AU-rich element consists of the two UAUUUAU heptamers and the three UAUUUUUAU nonamers. Previous studies on the c-fos ARE led to the conclusion that the UAUUUUAU motif was sufficient for mRNA destabilization but not optimal, and that multiple copies were needed for significant destabilization (24). This is in agreement with the results presented here. Based on a mutational analysis and sequence alignments, these authors concluded that the minimal motif may be UUAUUUA(U/A)(U/A). In another article on the c-fos ARE, a deletion analysis of the c-fos ARE led to the conclusions that the UUAUUUAUU nonamer, and not the UAUUUAU heptamer, was the shortest destabilizing motif (25). These authors also found that mRNAs containing multiple copies of the nonamer are degraded more rapidly than mRNAs with only one copy. Using the HPV-1 ARE, we also found that multiple copies of the motif were more inhibitory than one copy. However in our system, mutations outside of the heptamer UAUUUAU did not affect its ability to reduce mRNA levels. In the context of the HPV-1 ARE, the UAUUUAU was the shortest motif with inhibitory activity, indicating that the sequence context may affect the potency of an AU-rich element. In contrast to the HPV-1 AU-rich element that contains two UAUUUAU and three UAUUUUUAU motifs, the AU rich element on the IL-3 mRNA contains six AUUUA motifs. Mutations in three of these motifs had the same effect as deleting the entire element (26). The HPV-1 AU-rich element contains five motifs, and we show that all five contribute to inhibition in an additive manner. Similarly, mutations in all three AUUUA motifs in the c-fos ARE resulted in mRNA stabilization (27). It appears that multiple copies of the "AUUUA"-related motifs are required for full function of the various AU-rich RNA elements.
We have previously shown that the mRNA half-life is reduced by the presence of the h1ARE when using DNA transfections in which the mRNAs are synthesized in the cell nuclei (9,10). In contrast, when the same mRNAs were introduced directly into the cytoplasm as described here, bypassing the nucleus, there was no effect on the mRNA half-life by the h1ARE. These results indicate that a nuclear experience of the h1ARE-containing mRNAs is necessary for rapid mRNA degradation, suggesting that the mRNAs are either modified in the nuclei or interact with nuclear factors that induce premature mRNA degradation. Recent results on the ARE-containing c-fos mRNA showed that HuR mediates nuclear export of c-fos mRNAs (28). HuR also increases the c-fos mRNA half-life, suggesting that nuclear export and mRNA stability are connected and that inefficient mRNA export leads to premature degradation. In a previous article, we reported that the HIV-1 mRNA export factor Rev in combination with RRE or the SRV-1 CTE could overcome the inhibitory effect of the h1ARE (9), demonstrating that export of the h1ARE-containing mRNAs through an alternative, productive pathway overcomes inhibition and results in high expression. These results also demonstrated that the h1ARE has an inhibitory function in the nucleus, in addition to its inhibitory effect of translation in the cytoplasm described here.
The h1ARE binds PABP and may inhibit the interaction between PABP and eIF4G, thereby preventing circularization of the mRNA and the subsequent loading of ribosomes on the mRNA. This is not without precedent since it was recently shown that the rotavirus mRNA 3Ј-end-binding protein NSP3 interacts with eIF-4G and that NSP3 competes with the PABP for the eIF4G (29). Alterations of either the poly(A)-PABP or cap-eIF4E complexes lead to access to the poly(A) tail and cap by the poly(A) ribonuclease (PARN/DAN), resulting in deadenylation (30). Deadenylation was not observed here. It has FIG. 11. A, the histogram shows mean values and standard deviation of the CAT levels produced at 20 h posttransfection in HeLa cells transfected with capped CC and CC(A) mRNAs in the absence or presence of a poly(A) tail as indicated in the figure. A representative experiment is shown. Mean values and standard deviation of the quantified hGH protein levels produced from the hGH-encoding mRNA included as a internal control are shown below the histogram. B, the graph shows the CAT levels produced from capped CCXB or CCAUM/UM mRNAs in the absence (ϪAn) or presence (ϩAn) of a poly(A) tail at 7.5, 24, and 32 h posttransfection into HeLa cells. C, the presence of the h1ARE does not result in rapid deadenylation of the transfected mRNAs. Capped RNAs with (ϩ) or without (Ϫ) the 57-nt XB h1ARE were synthesized in the absence (Ϫ) or presence of a poly(A) tail (ϩ) of fixed length (A 60 ). The in vitro synthesized RNAs were transfected into HeLa cells, total cytoplasmic RNA were harvested at 1 or 4 h posttransfection, and the RNAs were analyzed by Northern blotting to investigate if h1ARE-containing mRNAs were rapidly deadenylated. U, RNA from untransfected cells.
FIG. 12. A, UV cross-linking of GST-PABP, GST-HuR, and GST-PCBP to the XB probe, the AUM/UM probe, an unrelated HPV-16 L1-derived RNA probe named L1 (nt 5732-5768 in the HPV-16R genome) and UV cross-linking of GST-PABP to RNAs B2 and C1 (See Fig.  6A and 10A). B, GST-PABP was UV cross-linked to XB RNA in the presence of serially diluted competitor RNA. XB, serially diluted XB competitor; L1, serially diluted unrelated HPV-16 L1-derived competitor RNA. also been proposed that the shuttling elav-like proteins are involved in mRNA translation directly (31,32). Similarly, redistribution of HuR protein from the nucleus to the cytoplasm is associated with increased protein production from mRNAs containing AREs with HuR binding sites (33,34). Therefore, HuR may act similarly to HuB and promote polysomal loading of ARE-containing mRNAs. HIV-1 Rev protein that overcomes the inhibitory effect of the h1ARE in HeLa cells also induces polysomal loading of target mRNAs (35,36). Perhaps elav-like proteins such as HuR lead mRNAs onto a productive pathway that includes efficient nuclear export and polysomal loading. The role of the h1ARE, HuR, and the PABP in the HPV-1 life cycle remains to be determined.