Nuclear Import of Cytoplasmic Poly(A) Binding Protein Restricts Gene Expression via Hyperadenylation and Nuclear Retention of mRNA

Abstract

Poly(A) tail length is emerging as an important marker of mRNA fate, where deviations from the canonical length can signal degradation or nuclear retention of transcripts. Pathways regulating polyadenylation thus have the potential to broadly influence gene expression. Here we demonstrate that accumulation of cytoplasmic poly(A) binding protein (PABPC) in the nucleus, which can occur during viral infection or other forms of cellular stress, causes mRNA hyperadenylation and nuclear accumulation of poly(A) RNA. This inhibits gene expression but does not affect mRNA stability. Unexpectedly, PABPC-induced hyperadenylation can occur independently of mRNA 3′-end processing yet requires the canonical mRNA poly(A) polymerase II. We find that nuclear PABPC-induced hyperadenylation is triggered by multiple divergent viral factors, suggesting that altering the subcellular localization of PABPC may be a commonly used mechanism to regulate cellular gene expression in a polyadenylation-linked manner.

Mammalian mRNA poly(A) tails are decorated with two types of nonhomologous poly(A) binding proteins, with distinct functions. Within the cytoplasm, the cytoplasmic poly(A) binding protein (PABPC) helps modulate the rate of mRNA deadenylation, both through its protective interactions with the poly(A) tail and by interfacing directly with factors involved in deadenylation, including Pan3, TOB, and GW182 (18, 19, 60). These interactions play key roles in transcript silencing, for example, upon microRNA-mediated repression, because poly(A) tail removal is the rate-limiting step in mRNA degradation and restricts translational competence (20). PABPC enhances translation efficiency by bridging the mRNA termini via its simultaneous interactions with the poly(A) tail and the cap-binding complex through eIF4G (31). Formation of this “closed loop” is hypothesized to promote translation of full-length transcripts, protect mRNAs from exonucleolytic attack, and facilitate recycling of ribosomes (72). Through additional interactions with the translation release factor eRF3 (12, 29), PABPC is also proposed to enhance the efficiency of termination and inhibit nonsense-mediated mRNA decay (NMD) (5, 16, 33, 62), a quality control pathway essential for destruction of messages containing premature termination codons (32, 53).

PABPC is a nuclear-cytoplasmic shuttling protein (1), but its steady-state localization is cytoplasmic. Although it has been shown to interact with nuclear pre-mRNA (30), distinct nuclear roles for PABPC remain largely enigmatic. Stimulation of poly(A) polymerase II (PAPII) activity and regulation of mRNA polyadenylation in the nucleus are instead carried out by the nuclear poly(A) binding protein (PABPN), which shares little sequence homology with PABPC (38, 40, 42, 69, 70).

While poly(A) tail length clearly has implications for mRNA translation and stability in the cytoplasm, emerging evidence indicates that the extent of polyadenylation in the nucleus also influences RNA fate (2, 23, 44, 56). mRNA poly(A) tail length is generally ∼200 to 250 nucleotides (nt) in mammals and ∼70 to 90 nt in Saccharomyces cerevisiae (49). However, very short poly(A) tails can be found on RNAs that are targets for rapid RNA degradation via nuclear quality control pathways such as the exosome. In yeast, these tails are added by the TRAMP polyadenylation complex, generally upon recognition of RNA processing errors (43, 68, 74). Conversely, messages with poly(A) tails that extend beyond the canonical length, termed hyperadenylated, accumulate in yeast mutants defective in mRNA export (27, 28, 34, 52), although it remains to be established whether hyperadenylation is a cause or a consequence of inefficient nuclear-cytoplasmic trafficking. Errors in RNA processing or ribonucleoprotein (RNP) complex remodeling have also been proposed to trigger hyperadenylation (26, 52). Thus, mRNA poly(A) tail extension is linked to an increased duration of nuclear residence, perhaps as a result of failed quality control checkpoints.

Hyperadenylation is documented primarily in yeast, although recently it has been shown in mammalian cells as well, for example, upon expression of the gammaherpesviral SOX protein (45). During lytic Kaposi's sarcoma-associated herpesvirus (KSHV) infection, SOX promotes a global restriction of cellular gene expression, through both widespread cytoplasmic mRNA degradation and hyperadenylation and retention of cellular messages in the nucleus (22, 45). Hyperadenylation of nuclear mRNAs is orchestrated exclusively by the cytoplasmic pool of SOX (13), indicating that SOX must stimulate hyperadenylation indirectly, perhaps via another cellular cofactor. Interestingly, an additional SOX activity is the prominent relocalization of PABPC into the nuclei of infected cells (45), although a functional connection between this phenotype and hyperadenylation has not been described. Recently, infection with several other viruses, including herpes simplex virus (HSV), rotavirus, and bunyavirus, as well as additional nonviral stresses such as heat shock, have also been reported to drive PABPC relocalization (1, 8, 15, 25, 48).

In this report, we demonstrate that a functional consequence of accumulation of PABPC in the nucleus is mRNA hyperadenylation and inhibition of poly(A) RNA export, resulting in a restriction of protein expression. Messenger RNAs engineered to bypass cellular 3′-end processing can still be hyperadenylated by the canonical poly(A) polymerase when nuclear PABPC levels are elevated, suggesting 3′-end processing-independent polymerase activity. Interestingly, several divergent viral proteins that restrict host gene expression promote nuclear relocalization of PABPC and hyperadenylation of transcripts in the nucleus. We therefore propose that the manipulation of PABPC localization represents a novel mechanism to globally regulate gene expression both in the nucleus and in the cytoplasm and has been exploited by diverse viruses perhaps as a means of commandeering cellular resources.

MATERIALS AND METHODS

Plasmids.

The PABPC1 cDNA (GenBank accession number Y00345) was cloned into the BamHI and XbaI sites of pCDEF3 and subsequently 5′-tagged with a 1× hemagglutinin (HA) or 1× Flag tag to generate pCDEF3-HA-PABPC1 and pCDEF3-Flag-PABPC1, respectively. An 8-glycine (G8) linker, followed by either an hnRNPC1-derived wild-type (WT) nuclear retention signal (NRS) (amino acids [aa] 88 to 165) or a mutant version of NRS (NRSmut) (61) that contained aa 98 to 146 of the original NRS fused to 29 aa of green fluorescent protein (GFP) to ensure the same size as that of the WT (13), was fused to the C terminus of Flag-PABPC1 by standard PCR methods to generate pCDEF3-Flag-PABPC1-NRS and pCDEF3-Flag-PABPC1-NRSmut. Likewise, a G8 linker, followed by a cytoplasmic retention signal derived from APOBEC3G (aa 51 to 128) (6), was fused to the C terminus of Flag-PABPC1 by standard PCR methods to generate pCDEF3-Flag-PABPC1-CRS. HA-tagged PABPC1 deletion mutants fused to the NRS were produced by overlap extension PCR and cloned into the BamHI and XbaI sites of pCDEF3 to generate pCDEF3-HA-PABPC1-ΔRRM1+2-NRS (lacking nt 1 to 176), pCDEF3-HA-PABPC1-ΔRRM3+4-NRS (lacking nt 191 to 370), pCDEF3-HA-PABPC1-ΔRRM1-4-NRS (lacking nt 1 to 370), pCDEF3-HA-PABPC1-ΔLinker-NRS (lacking nt 371 to 538), and pCDEF3-HA-PABPC1-ΔC-terminus-NRS (lacking nt 539 to 636). Plasmids pd2-eGFP-HR and pd2-eGFP-A₆₀-HR, (45), pCDEF3-SOX and pCDEF3-muSOX (13), pCDNA3.1-vhs and pCDNA3.1-vhsmut (35) have been previously described. Plasmids pCAGGS-NSP1 and pCDNA3-PABPC4-HA were kindly provided by Shinji Makino (University of Texas Medical Branch) and Tullia Lindsten (University of Pennsylvania), respectively.

Cells, transfections, and infections.

HEK 293T cells, HeLa cells, and COS-7 cells (American Type Culture Collection) were maintained in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). DNA transfections were performed using Effectene (Qiagen), as per the manufacturer's instructions. Cells to be processed for in situ hybridization or immunofluorescence (and the corresponding Western blotting) were transfected in a 12-well plate with 1 μg total plasmid DNA (900 ng empty vector plus 100 ng of the indicated PABPC plasmid). Cells to be harvested for Northern blotting (and the corresponding Western blotting) were transfected in a 12-well plate with 1 μg total plasmid DNA (900 ng of the indicated PABPC plasmid or empty vector plus 100 ng GFP reporter), unless indicated otherwise in the figure legends.

siRNA knockdown.

PABPC1 and PAPII knockdowns were achieved using previously described small interfering RNA (siRNA) oligonucleotides and methods (45). Knockdowns of PABPC4 (inducible PABPC [iPABPC]) were carried out using the following PABPC4-specific siRNA duplex oligonucleotides: siRNA no. 1, 5′AGGAGAGAAUUAGUCGAUAUCAGGG, and siRNA no. 2, 5′GGAAUUCAACUCAAGGUUUGAAGAC. Nonspecific control siRNAs were purchased from Ambion. HEK 293T cells were transfected with 200 nM siRNA oligonucleotides using Lipofectamine 2000 (Invitrogen) for 24 h, followed by transfection with DNA for an additional 24 h prior to harvesting for protein analysis, in situ hybridization, or Northern blotting.

Immunofluorescence assays (IFA) and in situ hybridization.

Cells were grown on coverslips coated with 100 μg/ml poly-l-lysine and processed for IFA as previously described (4). Briefly, cells were fixed in 4% formaldehyde, incubated for 10 min in permeabilization buffer (1% [vol/vol] Triton X-100 and 0.1% [wt/vol] sodium citrate) and then for 30 min in block buffer (1% [vol/vol] Triton X-100, 0.5% [vol/vol] Tween 20, and 3% [wt/vol] bovine serum albumin [BSA]), and incubated with mouse monoclonal PABPC 10E10 (1:25 dilution; Santa Cruz Biotechnology), rabbit polyclonal SOX J5803 (1:500 dilution) (21), mouse monoclonal HA 12CA5 (1:500 dilution; Abcam), or mouse monoclonal Flag (1:500 dilution; Sigma) primary antibodies for 3 to 12 h in blocking buffer, followed by incubation with Alexa Fluor 488- or 546-conjugated goat anti-mouse or goat anti-rabbit secondary antibodies (1:1,500 dilution) (Molecular Probes). Coverslips were mounted in DAPI (4′,6-diamidino-2-phenylindole)-containing Vectashield mounting medium (Vector Labs) to visualize nuclei.

For in situ hybridization (http://www.singerlab.org/protocols), cells were fixed with 4% formaldehyde for 10 min, washed 2 times with phosphate-buffered saline (PBS), and permeabilized by treatment with 70% ethanol for 2 h to overnight. Cells were next treated with the following for 5 min each: 1× PBS, 1 M Tris (pH 8.0), and 1× PBS. Cells were hybridized overnight at 37°C in 200 μl of hybridization buffer (50% [vol/vol] formamide, 10% [vol/vol] dextran sulfate, 0.02% [wt/vol] BSA, 200 μg Escherichia coli tRNA, and 2× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]) using 2 ng/μl of Alexa Fluor 546-labeled oligo(dT)₁₅ (Molecular Probes). Cells were then processed for IFA as described above.

Cell extracts, fractionation, immunoblotting, and Northern blotting.

Cell lysates were prepared in RIPA buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1% [vol/vol] Nonidet P-40, 0.5% [wt/vol] sodium deoxycholate, 0.1% [wt/vol] sodium dodecyl sulfate [SDS]) containing protease inhibitors (Roche) and quantified by Bradford assay (Bio-Rad). Equivalent micrograms of each sample were resolved by SDS-polyacrylamide gel electrophoresis (PAGE), transferred to a polyvinylidene difluoride membrane, and subjected to immunoblotting with either mouse monoclonal GFP (1:2,000 dilution), rabbit polyclonal PABPC 4992 (1:2,000 dilution) (Cell Signaling Technology), or rabbit polyclonal SOX J5803 (1:5,000 dilution) (21) primary antibodies, followed by incubation with horseradish peroxidase-conjugated actin antibodies (for loading control) and goat anti-mouse or goat anti-rabbit secondary antibodies (1:5,000 dilution) (Southern Biotechnology Associates).

Total cellular RNA was isolated for Northern blotting using RNA-bee (Tel-Test), resolved on 1.2% agarose-formaldehyde gels, and probed with a ³²P-labeled GFP DNA probe made using the Rediprime II random prime labeling kit (GE Healthcare). Where indicated, cells were treated with 5 ng/ml leptomycin B (LMB; Sigma) for 6 to 15 h prior to RNA isolation to stabilize hyperadenylation products (45) or with 5 μg/ml actinomycin D (Act D) for the indicated times to monitor the mRNA half-life. Cellular fractionation was carried out using NE-PER nuclear and cytoplasmic extraction reagents (Thermo Scientific), as per the manufacturer's instructions, followed by RNA isolation from fractionated extracts using RNA-bee (Tel-Test).

RNaseH digestions were carried out with 5 to 7 μg of total RNA combined with 500 pmol of the oligo(dT)₁₅ primer in a 25.8-μl reaction volume. After incubation at 65°C for 8 min, 1 U of RNaseH (New England Biolabs) and 40 U of RNasin (Promega) were added to 1× RNaseH buffer, followed by incubation at 37°C for 30 min. Reactions were terminated by the addition of 1 μl of 0.5 M EDTA (pH 8.0), and the RNA was ethanol precipitated and subjected to gel electrophoresis. Northern blots were analyzed using a Typhoon 8600 phosphorimager, and for half-life experiments, GFP mRNA signal intensity at each time point was quantified using ImageJ software and normalized to an 18S signal.

RESULTS

KSHV SOX-induced nuclear retention of endogenous mRNA is dependent upon PABPC.

To better understand the mechanisms of hyperadenylation in human cells, we initially focused our studies on the KSHV SOX protein, which induces mRNA hyperadenylation and a coincident mRNA export block. We predicted that if there was a connection between this phenotype and PABPC relocalization, we should be able to block SOX-induced hyperadenylation by depleting PABPC. However, our prior attempts at depleting the major isoform of PABPC (PABPC1) using siRNA oligonucleotides failed to yield consistent inhibition of SOX-induced hyperadenylation or nuclear retention of endogenous mRNA (45). However, there are 3 additional cytoplasmic isoforms of PABPC, two of which (PABPC4 and PABPC5) are widely expressed and thus might compensate for PABPC1 function in its absence (9, 75). Indeed, we detected accumulation of a slightly higher-molecular-weight protein upon siRNA-mediated PABPC1 knockdown (Fig. 1 A). Expression of this protein was induced in multiple different cell lines with 2 independent PABPC1 siRNA oligonucleotides (Fig. 1A). Based on its size, we hypothesized that this protein might be the PABPC4 isoform (also known as inducible PABPC [iPABPC]), which shares 79% aa identity with PABPC1 (75). The identity of this band as PABPC4 was confirmed by showing its disappearance upon transfection of PABPC4-specific RNA oligonucleotides (Fig. 1B). Additionally, anti-HA immunofluorescence assays (IFA) on HEK 293T cells expressing an HA-tagged PABPC4 expression plasmid confirmed that, like PABPC1, HA-PABPC4 was relocalized from the cytoplasm to the nucleus upon coexpression with SOX (Fig. 1C). Our prior failure to detect mRNA export defects in the presence of SOX upon PABPC1 knockdown could therefore have been masked by the concomitant induction of PABPC4.

FIG. 1.

PABPC4 is induced upon PABPC1 depletion and is directed to the nucleus by SOX. (A) COS7, HEK 293T, and HeLa cells were transfected with either control siRNAs (si) or 2 independent siRNAs specific for PABPC1 (PABPC1#1 and PABPC1#2). siRNA transfections for COS7 and HEK 293T cells were performed in duplicate. At 72 h posttransfection, cell lysates were resolved by SDS-PAGE and immunoblotted with anti-PABPC antibodies. (B) HeLa and HEK 293T cells were transfected with control siRNAs or the indicated siRNAs specific for PABPC1 and/or PABPC4. Two independent PABPC4-specific siRNAs were used (PABPC4#1 and PABPC4#2). Lysates were then harvested and immunoblotted as described in the legend to panel A. (C) HEK 293T cells were transfected with a plasmid expressing HA-tagged PABPC4 alone or together with a plasmid expressing SOX and, 24 h later, subjected to immunofluorescence assays with anti-HA and anti-SOX antibodies. DAPI staining was used to visualize nuclei.

Hyperadenylation correlates with a failure to export mRNA from the nucleus, which can be visualized upon oligo(dT) in situ hybridization as an accumulation of endogenous nuclear poly(A) sequences. Thus, to explore whether PABPC played a role in mRNA export defects in SOX-expressing cells, siRNAs against both PABPC1 and PABPC4 or control siRNAs were transfected into HEK 293T cells (Fig. 2 A). In control siRNA-treated cells expressing SOX, oligo(dT) in situ hybridization shows a clear enhancement of endogenous nuclear poly(A) RNA (Fig. 2B). However, SOX expression failed to promote nuclear poly(A) RNA accumulation in cells for which the levels of the PABPC1 and PABPC4 protein were decreased (Fig. 2B). These data suggest that PABPC, perhaps upon nuclear import, plays an essential role in SOX-mediated retention of mRNA in the nucleus.

FIG. 2.

PABPC is required for SOX-induced nuclear poly(A) RNA accumulation. (A) HEK 293T cells were transfected with control siRNAs or siRNAs against PABPC1 and PABPC4. Lysates were harvested 24 h posttransfection and immunoblotted with anti-PABPC antibodies to detect the efficiency of PABPC protein depletion. In parallel, antiactin immunoblotting was performed to control for loading. (B) Cells described in the legend to panel A were transfected with either empty vector or a plasmid expressing SOX for 24 h and then processed for in situ hybridization with oligo(dT), followed by staining with anti-SOX antibodies and DAPI to visualize nuclei.

Nuclear accumulation of PABPC1 drives hyperadenylation and inhibits mRNA export.

Two possibilities followed the above-described results: (i) nuclear accumulation of PABPC is the principal driver of mRNA retention in SOX-expressing cells or (ii) PABPC is one necessary component of this phenotype, but other SOX activities are also required. To distinguish between these, we sought to induce nuclear accumulation of PABPC in the absence of SOX and monitor whether hyperadenylation and an mRNA export block ensued. To this end, PABPC1 was fused to an hnRNPC1-derived nuclear retention signal (NRS) (50, 61) and a Flag tag to differentiate it from endogenous PABPC. Importantly, immunoblotting of cell lysates with a PABPC antibody that recognized both endogenously and exogenously expressed proteins showed that our transfected PABPC is not expressed above physiologic levels in these cells (Fig. 3 A). Immunofluorescence assays with anti-Flag antibodies confirmed that the NRS tag restricted PABPC1 to the nucleus, and these cells exhibited a dramatic increase in the nuclear oligo(dT) signal, as revealed by in situ hybridization (Fig. 3B). It is striking that only a slight elevation in the levels of nuclear PABPC is sufficient to cause nuclear poly(A) RNA accumulation (Fig. 3A and B). Several observations indicate that this increase in nuclear poly(A) RNA is specifically dependent on the accumulation of PABPC in the nucleus. First, it rarely occurred upon expression of Flag-PABPC1 lacking an NRS fusion, except in the occasional cells in which Flag-PABPC1 expression was high enough for the protein to enter the nuclei even without the NRS. Second, it did not occur in cells expressing a control Flag-PABPC1 fused to a mutant NRS incapable of nuclear restriction (Flag-PABPC1-NRSmut). Finally, it did not occur upon expression of Flag-PABPC1 fused to a cytoplasmic retention signal (CRS) derived from APOBEC3G (Flag-PABPC1-CRS) (Fig. 3B) (6). Similar transfection efficiencies for each of the PABPC constructs were observed, and thus, this is not a contributing factor toward differences in their downstream effects.

FIG. 3.

Nuclear accumulation of PABPC causes mRNA retention and hyperadenylation. (A) HEK 293T cells were transfected with the indicated PABPC1 expression plasmids for 24 h. Lysates were then harvested, resolved by SDS-PAGE, and immunoblotted with anti-PABPC antibodies to detect both endogenous (bottom arrow) and exogenous (upper arrow) PABPC proteins. Note that Flag-PABPC1 comigrates with endogenous PABPC1 in this blot. In parallel, antiactin immunoblotting was performed to control for loading. (B) HEK 293T cells were transfected as described in the legend to panel A for 24 h and then processed for oligo(dT) in situ hybridization, followed by immunofluorescence assays with anti-Flag antibodies. (C) HEK 293T cells were cotransfected with the indicated plasmids for 24 h. After treatment with leptomycin B (LMB) for 12 h to stabilize hyperadenylated species, total RNA and proteins were isolated. Total RNA was incubated in the presence or absence of oligo(dT) and digested with RNaseH. Products were resolved on a 1.2% agarose-formaldehyde gel and Northern blotted with ³²P-labeled GFP and 18S probes (top). Hyperadenylated species are indicated by the labeled bracket [hyp(A) GFP]. Total protein was resolved by SDS-PAGE and immunoblotted using anti-Flag and anti-SOX antibodies (bottom). Actin served as a loading control. (D) Cells were transfected with the indicated plasmid as described above and treated with LMB for 7 h. Total RNA was then isolated from whole cells (W) or cytoplasmic fractions (C) and nuclear fractions (N) fractions and Northern blotted with ³²P-labeled GFP and 18S probes. Hyperadenylated species are indicated by the labeled bracket [hyp(A) GFP].

To confirm that nuclear PABPC1 indeed directs mRNA hyperadenylation, we used a second assay with which we monitored the length of a GFP reporter message by Northern blotting. SOX-transfected cells were included as a positive control, as we have shown previously that they exhibit hyperadenylated GFP mRNA (45). In the presence of either Flag-PABPC1-NRS or SOX, the length of the GFP mRNA was extended in a heterogeneous manner compared with that of GFP expressed alone (Fig. 3C). The mRNA size difference was a result of hyperadenylation, as removal of the mRNA poly(A) tails by hybridization to oligo(dT) followed by RNaseH digestion caused the GFP mRNAs to comigrate (Fig. 3C). Furthermore, fractionation of the nuclear and cytoplasmic RNA populations confirmed that the hyperadenylated GFP mRNAs were largely confined to the nucleus (Fig. 3D). Nuclear accumulation of PABPC therefore induces hyperadenylation and nuclear retention of mRNAs.

PABPC1 exerts differential effects on gene expression, depending on its localization.

We conducted the above-described Flag-PABPC1-NRS experiments with cells retaining endogenous PABPC so as not to confound the effect of removal of PABPC from the cytoplasm with the effect of increasing its concentration in the nucleus. Our data suggest that increasing the levels of PABPC in the nucleus causes hyperadenylation and nuclear retention of messages, which presumably would be detrimental to expression of the encoded proteins. Given that this contrasts with established cytoplasmic roles of PABPC in enhancing translation (49), we compared the effects of increasing the concentrations of PABPC in the cytoplasm and the nucleus using CRS- and NRS-fused PABPC1. Even when attached to nuclear proteins, the CRS has been shown to override an NLS and retain proteins in the cytoplasm (6). Immunofluorescence assays with anti-Flag antibodies confirmed that Flag-PABPC1-CRS remained exclusively cytoplasmic (Fig. 3B). Northern blotting of the GFP reporter was then used to monitor mRNA length and abundance in cells expressing CRS- or NRS-tagged PABPC1. In a dose-dependent manner, cytoplasmic Flag-PABPC1-CRS increased WT-length GFP mRNA levels and failed to promote hyperadenylation (Fig. 4A). Conversely, nuclear Flag-PABPC1-NRS expression instead decreased the abundance of WT-length GFP mRNA in a dose-dependent manner and induced the accumulation of higher-molecular-weight GFP mRNA species (Fig. 4B).

FIG. 4.

Cytoplasmic PABPC and nuclear PABPC have opposing effects on gene expression. (A, B) HEK 293T cells were transfected with GFP and increasing amounts of Flag-PABPC1-CRS or Flag-PABPC1-NRS (100 to 900 ng) for 24 h. Total RNA was then isolated and visualized by Northern blotting with ³²P-labeled GFP and 18S probes. Hyperadenylated species are indicated by the labeled bracket [hyp(A) GFP]. To monitor protein levels, protein lysate from the transfected cells was resolved by SDS-PAGE and immunoblotted with anti-Flag or antiactin (loading control) antibodies. (C) HEK 293T cells were transfected with GFP and empty vector, Flag-PABPC1-CRS, or Flag-PABPC1-NRS. At 24 h posttransfection, 5 μg/ml of actinomycin D was added to block transcription, and total RNA was harvested at the indicated time points thereafter. RNA was then visualized by Northern blotting with GFP and 18S probes, and the GFP mRNA half-life was calculated after 18S normalization. Error bars indicate standard errors between samples. Data were derived from five independent experiments. (D) HEK 293T cells were transfected with plasmids expressing GFP and empty vector, Flag-PABPC1-NRS, or Flag-PABPC1-CRS for 24 h. Equivalent amounts of protein lysate were resolved by SDS-PAGE and immunoblotted with antibodies against GFP, Flag, and actin (as a loading control).

To determine whether PABPC1-NRS and PABPC1-CRS alter mRNA stability, we measured their effects on the half-life of the GFP message. Cells were treated with actinomycin D (Act D) at 24 h posttransfection, and RNA was isolated at the indicated times thereafter. Expression of Flag-PABPC1-NRS did not significantly alter the overall stability of the GFP message (Fig. 4C). In contrast, expression of Flag-PABPC1-CRS lead to a marked increase in the GFP mRNA half-life from ∼17 h to >100 h (Fig. 4C). These cell data are in agreement with prior observations showing that cytoplasmic PABPC1 stabilizes mRNAs in cell extracts and when overexpressed in Xenopus oocytes (7, 71, 73). In addition, our results suggest that the levels of PABPC in the cytoplasm may be limiting. Alternatively, the CRS fusion could interfere with mRNA deadenylation by Pan2-Pan3, which are recruited via interactions with the C terminus of PABPC (60).

Although expression of nuclear Flag-PABPC1-NRS and subsequent mRNA hyperadenylation do not lead to changes in the half-life of GFP mRNA, we found that GFP protein levels were dramatically decreased in Flag-PABPC1-NRS-expressing cells (Fig. 4D). As endogenous PABPC is present in these cells, the reduced protein levels are likely not a consequence of decreased PABPC in the cytoplasm but rather are specific to the ability of nuclear PABPC to induce hyperadenylation and mRNA retention in the nucleus. Conversely, enhanced expression of PABPC in the cytoplasm might boost GFP production, at least in part as a consequence of increased mRNA stability. This was indeed the case, as immunoblotting shows enhanced GFP levels in cells coexpressing cytoplasmic Flag-PABPC1-CRS (Fig. 4D). Collectively, these results indicate that PABPC has profound but opposite effects on gene expression depending on its levels in the nucleus versus the cytoplasm.

Poly(A) binding motifs are required for hyperadenylation.

PABPC1 is a modular protein consisting of four amino-terminal RNA recognition motifs (RRMs), an unstructured linker region, and a conserved carboxyl-terminal helical (5H) region (Fig. 5A). RNA binding is primarily carried out by the first 2 RRM domains, although RRM3 and RRM4 can also associate with RNA, albeit with reduced affinity (11, 41, 42, 51, 63). Protein-protein interactions with factors involved in translation occur via both the RRM regions and the 5H domain (12, 14, 29, 31, 42). We constructed a variety of deletion mutants to identify the regions of PABPC1-NRS required for promoting hyperadenylation and mRNA retention in the nucleus. Immunoblotting with anti-HA and anti-Flag antibodies confirmed that the PABPC mutants are expressed to equivalent levels upon transfection into HEK 293T cells (Fig. 5B). Using oligo(dT) in situ hybridization and Northern blotting assays, we found that a mutant lacking both RRM1 and RRM2 failed to promote hyperadenylation and mRNA nuclear retention (Fig. 5C and E). In contrast, mutants lacking RRM3 and RRM4, the linker region, or the C-terminal domain retained the ability to hyperadenylate and restrict export of mRNA (Fig. 5C and E).

FIG. 5.

PABPC RRM1 and RRM2 are necessary and sufficient to induce hyperadenylation. (A) Diagram of PABPC, showing the 4 RRMs, followed by the linker region and conserved helical carboxyl terminus (square). (B) Expression levels from the indicated PABPC1 WT and mutant constructs were monitored following transfection into 293T cells for 24 h. Protein lysates harvested, resolved by SDS-PAGE, and immunoblotted with a mixture of anti-Flag and anti-HA antibodies or with antiactin antibody as a loading control. (C and D) HEK 293T cells were transfected with the indicated plasmids for 24 h and then subjected to oligo(dT) in situ hybridization and immunofluorescence assays with anti-HA antibodies. Nuclei were stained with DAPI. (E) HEK 293T cells were transfected with plasmids expressing GFP and either empty vector or the indicated PABPC construct for 24 h. Total RNA was then resolved by agarose-formaldehyde gel electrophoresis and Northern blotted with ³²P-labeled GFP and 18S probes. Hyperadenylated species are indicated by the labeled bracket [hyp(A) GFP].

After having shown PABPC1 RRM1 and RRM2 to be necessary for hyperadenylation, we next asked whether these domains were sufficient for hyperadenylation to occur. Strikingly, expression of RRM1 and RRM2 alone fused to an NRS was sufficient to retain poly(A) RNA in the nucleus and drive hyperadenylation (Fig. 5D and E). This region of PABPC1 must be present in the nucleus to block mRNA export, as it failed to cause this phenotype when restricted to the cytoplasm by fusion to a CRS (Fig. 5D). Furthermore, expression of RRM3 and RRM4 fused to an NRS also failed to promote nuclear retention of mRNA (Fig. 5D), even though these domains are capable of RNA binding (11, 42). Therefore, we conclude that both RRM1 and RRM2 are necessary and sufficient for PABPC1-induced hyperadenylation and nuclear retention of mRNAs.

Dissociation of cellular mRNA 3′-end processing from hyperadenylation.

Although hyperadenylation has been shown to occur in both yeast and mammals in association with defects in either mRNA 3′-end processing or export, these two processes are closely linked, and the underlying mechanisms governing aberrant polyadenylation remain unknown. We therefore assessed the ability of Flag-PABPC1-NRS to promote hyperadenylation of GFP mRNAs that do not undergo cellular 3′-end processing but instead are cleaved by a hammerhead ribozyme (45). We tested GFP constructs either terminating via ribozyme cleavage just 3′ proximal to the stop codon (GFP-HR) or containing a templated 60-nt poly(A) tail upstream of the ribozyme cleavage site (GFP-A₆₀-HR). Unlike GFP, which was processed by the cellular 3′-end machinery, neither GFP-HR nor GFP-A₆₀-HR appeared efficiently hyperadenylated in the presence of Flag-PABPC1-NRS (Fig. 6A). Notably, however, in the presence of Flag-PABPC1-CRS, both GFP and GFP-A₆₀-HR mRNA levels increased to a much greater extent than those of GFP-HR. This is presumably due to mRNA stabilization (Fig. 4C). Thus, while cellular 3′-end processing enhanced the susceptibility of mRNA to PABPC-induced hyperadenylation in the nucleus, the presence of a poly(A) tail is the critical determinant for PABPC-induced mRNA stabilization in the cytoplasm.

FIG. 6.

Cellular mRNA 3′-end processing enhances but is not required for PABPC-induced hyperadenylation. (A) HEK 293T cells were transfected with the indicated plasmids for 24 h, and then total RNA was isolated and Northern blotted with ³²P-labeled GFP and 18S probes. (B) Same protocol as described in the legend to panel A, but cells were treated with 5 ng/ml LMB for 12 h prior to RNA isolation to enhance detection of hyperadenylated species. (C) HEK 293T cells were transfected with the indicated plasmids for 24 h and then treated with 5 ng/ml LMB for 12 h. Total RNA was isolated and incubated in the presence or absence of oligo(dT) and then digested with RNaseH. RNA was visualized by Northern blotting with ³²P-labeled GFP and 18S probes. (D) HEK 293T cells were transfected twice over 48 h with either PAPII siRNAs or control siRNAs and then subsequently transfected in duplicate with DNA plasmids expressing GFP-A₆₀-HR alone or together with Flag-PABPC1-NRS for 24 h. Samples were treated with 5 ng/ml LMB for 6 h prior to harvesting either protein (top) or RNA (bottom). Protein lysates were resolved by SDS-PAGE and immunoblotted with antibodies against PAPII, Flag, or actin (as a loading control). RNA from each sample was Northern blotted with ³²P-labeled GFP and 18S probes.

We previously showed that treatment of cells with leptomycin B (LMB) stabilized hyperadenylated mRNAs in SOX-expressing cells, greatly facilitating their detection (45). We therefore tested whether LMB treatment might likewise reveal weak or unstable hyperadenylated GFP-HR or GFP-A₆₀-HR mRNAs in cells expressing Flag-PABPC1-NRS. Surprisingly, we were able to observe nuclear PABPC1-induced accumulation of higher-molecular-weight GFP-A₆₀-HR species in LMB-treated cells (Fig. 6B), although at significantly reduced efficiency relative to normal GFP mRNAs. The fact that the higher-molecular-weight mRNA species represented hyperadenylated products was confirmed by showing their disappearance upon incubation of the RNA with oligo(dT) and digestion with RNaseH (Fig. 6C). Hyperadenylation required the presence of the templated A₆₀ tail, as it was never detected on GFP-HR mRNA, which lacked any poly(A) tail (Fig. 6B). Additionally, we observed that, similar to the case with GFP, nuclear expression of PABPC1 RRM1 and RRM2 alone was sufficient to drive hyperadenylation of GFP-A₆₀-HR in LMB-treated cells, although to reduced levels relative to the full-length protein (Fig. 6B). These data show that hyperadenylation can occur in cells, albeit less efficiently, via a mechanism uncoupled to cellular mRNA 3′-end processing.

Polyadenylation normally occurs in conjunction with cleavage and polyadenylation specificity factor (CPSF) complex-induced mRNA 3′-end cleavage. The canonical poly(A) polymerase PAPII is recruited to nascent transcripts via interactions with PABPN and CPSF to processively polyadenylate mRNAs (40). PAPII also participates in hyperadenylation, as SOX-induced hyperadenylation of CPSF-processed GFP is reduced upon depletion of PAPII (45). Our current observations, however, suggested that polyadenylation can occur on mRNAs that are not processed by CPSF complex, as long as they already possess some poly(A) sequence. To determine whether this noncanonical polyadenylation is also carried out by PAPII, we monitored nuclear PABPC1-induced hyperadenylation of GFP-A₆₀-HR upon siRNA-mediated depletion of PAPII (Fig. 6D). Indeed, removal of PAPII significantly reduced GFP-A₆₀-HR hyperadenylation (Fig. 6D), suggesting that in cells with elevated levels of nuclear PABPC, PAPII can act on mRNAs after or independently of 3′-end processing to cause hyperadenylation.

Multiple divergent viral proteins cause nuclear import of PABPC and mRNA accumulation in the nucleus.

Finally, we sought to determine whether PABPC relocalization and subsequent hyperadenylation and nuclear retention of mRNA might be conserved mechanisms used by other viral proteins to restrict host gene expression. In addition to KSHV SOX and its gammaherpesvirus homologs such as murine gammaherpesvirus 68 (MHV-68) muSOX, the vhs protein of herpes simplex virus (HSV; an alphaherpesvirus), and the NSP1 protein of severe acute respiratory syndrome coronavirus (SARS-CoV) are well-characterized host shutoff factors that target mRNA (17, 36, 37, 55, 58, 67). We therefore monitored by immunofluorescence assays the endogenous PABPC localization in HEK 293T cells transfected with each of these viral factors. Similar to the gammaherpesviral SOX and muSOX proteins, HSV-1 vhs and SARS-CoV NSP1 also induced prominent nuclear relocalization of endogenous PABPC (Fig. 7A). For vhs, this relocalization is linked to its host shutoff activity, as a vhs mutant unable to restrict host gene expression (35) failed to induce PABPC import (Fig. 7A), even though it is expressed at levels higher than those of the WT protein (data not shown). The HSV-1 ICP27 and UL47 proteins have also recently been implicated in PABPC relocalization, suggesting that multiple viral factors may collectively target PABPC during HSV-1 infection (15).

FIG. 7.

Nuclear relocalization of PABPC, nuclear retention of mRNA, and hyperadenylation are phenotypes induced by multiple independent viral factors. (A) HEK 293T cells were transfected with either empty vector or with plasmids expressing SOX, HA-muSOX, vhs, vhs mutant (vhs-mut), or SARS-CoV NSP1 for 24 h and then subjected to immunofluorescence assays (IFA) with anti-PABPC antibodies and stained with DAPI to visualize nuclei. Right panels represent a merge between the IFA and DAPI signals. (B) HEK 293T cells were transfected as described in the legend to panel A, subjected to oligo(dT) in situ hybridization, and stained with DAPI. Right panels represent a merge between the in situ and DAPI signals. (C) HEK 293T cells were cotransfected with a plasmid expressing GFP and either empty vector or plasmids expressing the indicated viral protein for 24 h. Cells were treated with LMB for 15 h, and total RNA was then isolated and digested with RNaseH in the presence or absence of oligo(dT). Products were resolved by agarose-formaldehyde gel electrophoresis and Northern blotted with ³²P-labeled GFP and 18S probes.

Given their ability to cause PABPC import, we predicted that HSV-1 vhs and SARS-CoV NSP1 should also induce hyperadenylation and nuclear retention of mRNA. Indeed, expression of HSV-1 vhs or SARS-CoV NSP1 resulted in a robust accumulation of endogenous poly(A) RNA in the nucleus, similar to that in SOX and muSOX-expressing cells (Fig. 7B). In agreement with the PABPC import data, this mRNA export block was not observed in cells expressing mutant vhs (Fig. 7B). Additionally, Northern blotting coupled with oligo(dT)/RNaseH treatment showed that the GFP reporter message was hyperadenylated in the presence of SOX, muSOX, vhs, and NSP1 (Fig. 7C). Thus, four independent viral proteins known to globally inhibit host gene expression restrict mRNA export via a common mechanism of PABPC relocalization and hyperadenylation.

DISCUSSION

Although PABPC is a shuttling protein and can be found at low levels in the nuclei of uninfected cells (1), we showed that increasing its nuclear abundance drives hyperadenylation and nuclear retention of mRNAs, thereby inhibiting their expression. Hyperadenylation requires the PABPC poly(A) binding motifs, suggesting that it may be triggered upon binding of PABPC to mRNA poly(A) tails either during or after the normal polyadenylation process. In support of this idea is our observation that PABPC can direct PAPII-induced hyperadenylation of ribozyme-terminating transcripts only if the mRNAs contain templated poly(A) sequences. These data additionally indicate that hyperadenylation is not necessarily coincident with 3′-end processing and may be carried out by PAPII after the initial round of polyadenylation has completed.

We envision at least two nonmutually exclusive mechanisms by which nuclear PABPC could trigger hyperadenylation. First, if enhanced levels of PABPC binding to transcripts disrupted or significantly altered the RNP composition of mRNAs, nascent messages might fail nuclear quality control checkpoints required for export. Transcripts that are not efficiently exported may then become susceptible to poly(A) tail extension by PAPII, perhaps because of the increased duration of nuclear residence. Reports showing that hyperadenylated mRNAs accumulate in yeast export factor mutants or in mammalian cells depleted of TAP/NXF1 would be consistent with this model (28, 34, 46, 52). Additionally, the KSHV ORF57 protein, which binds and stabilizes the viral noncoding RNA PAN, can also induce PAN hyperadenylation, perhaps as a consequence of increasing its half-life in the nucleus (57). That said, hyperadenylation and retention of mRNAs in the nucleus are interconnected outcomes, making it difficult to distinguish the initiating event. The interplay between mRNA 3′-end processing and export is likely to be complex and involve extensive RNP remodeling and recycling of processing factors (52). Delineation of which (if any) of these events are disrupted by PABPC will be an important future challenge.

An alternate possibility is that PABPC more directly recruits or affects the regulation of poly(A) polymerase on mRNAs. Recruitment would have to occur through the first two PABPC RRMs, as these are sufficient to stimulate hyperadenylation. A recent report identifying PABPC as a copurifying component of the mRNA 3′-end processing complex may support this model, although whether it actually plays a role in mRNA processing remains to be established (59). We have yet to observe an interaction between PABPC and PAPII (G. R. Kumar and B. A. Glaunsinger, unpublished observations), indicating that if PABPC participates in PAPII recruitment, it is likely to be via other intermediate interactions. Indeed, in vitro biochemical experiments with purified proteins have shown that yeast and mammalian PABPC inhibit the polyadenylation reaction (47, 70). Competition between PABPC and PABPN for binding nascent poly(A) tails could also disrupt the interplay between PAPII and PABPN, which is proposed to govern poly(A) tail length control (40). However, this is unlikely to be the sole mechanism of hyperadenylation, given the susceptibility of GFP-A₆₀-HR to this phenotype.

Many viruses globally restrict cellular gene expression, in part as a mechanism of resource reallocation. By dampening cellular mRNA translation, competition is lessened for the gene expression machinery critical for efficient viral gene expression and replication. An additional important advantage of virus-induced restriction of cellular gene expression is immune evasion, as many of the genes whose expression is inhibited are important effectors of the immune response (54, 64). A diverse group of viruses have thus often evolved to target the same key host proteins and pathways that allow them to commandeer cellular gene expression machinery. PABPC is a clear cellular target of numerous viruses, many of which encode proteases that cleave it as a means of restricting host translation (65). However, recent findings that herpesviruses, rotaviruses, and bunyaviruses all drive nuclear import of PABPC underscores the likely importance of its nuclear functions, in addition to its roles in translation, in the global regulation of gene expression (8, 15, 25). In the case of gammaherpesvirus SOX, herpes simplex virus vhs, and SARS coronavirus NSP1, host gene expression is repressed on at least two fronts. In the cytoplasm, these viral proteins inactivate host mRNAs through cleavage and/or translational repression (17, 36, 37, 55, 58, 67). The gene expression block is then likely magnified via PABPC relocalization and ensuing hyperadenylation, which would prevent repopulation of the cytoplasm with newly transcribed mRNAs.

For most RNA viruses, whose viral mRNAs are transcribed and processed in the cytoplasm, triggering PABPC-induced hyperadenylation of nuclear mRNAs would be an effective means to selectively block nascent cellular gene expression. Additionally, RNA viruses such as rotaviruses and bunyaviruses produce transcripts lacking poly(A) tails and have evolved PABPC-independent mechanisms of translation (8, 24, 39). For these viruses, removal of PABPC from the cytoplasm should be detrimental to host, but not viral, gene expression. However, other RNA viruses such as coronaviruses encode mRNAs that are polyadenylated and thus may still depend on some residual cytoplasmic PABPC for efficient expression (66). In the case of herpesviruses, the advantage of PABPC relocalization is not immediately clear. These DNA viruses utilize the host RNA transcription and processing machinery to produce their transcripts, which must then be exported and translated in a manner presumably analogous to that of cellular mRNAs. One important future direction will therefore be to reveal how nuclear-replicating viruses such as herpesviruses are able to subvert this nuclear PABPC-induced gene expression block to allow efficient production of their own genes. Interestingly, a recent report suggests that during KSHV infection, translation of viral mRNAs in the absence of abundant cytoplasmic PABPC may be at least partially compensated by enhanced assembly of eIF4F through preferential eIF4G recruitment to the 5′ cap (3).

It is notable that nuclear relocalization of PABPC is also triggered during certain cellular stresses, including heat shock or transcriptional block (1, 48). This phenotype is thus perhaps a normal cellular response to some forms of stress, but which can also be usurped during viral infection. From a cellular perspective, PABPC relocalization could be a mechanism to pause gene expression via inhibition of both translation in the cytoplasm and nascent RNA transport from the nucleus. Under these conditions, cytoplasmic mRNAs may be shuttled to stress granules (10), whereas nuclear transcripts are retained and hyperadenylated. Our half-life analyses indicate that in the absence of viral infection, these hyperadenylated transcripts do not undergo enhanced degradation, perhaps suggesting that tail trimming might reinstate their export competence (28). An important distinction for nonviral stresses that cause PABPC import would therefore be that upon removal of the stress, PABPC could reenter the cytoplasm, perhaps allowing resumption of mRNA export and translation. Indeed, this is the case following the release from heat shock (48). However, during infection, PABPC remains nuclear until completion of the viral replication cycle (8, 25, 45), which generally coincides with cell death.

In summary, our data support a novel role for PABPC in the restriction of cellular gene expression, in addition to its established functions in regulating cytoplasmic mRNA stability and enhancing translation. The ability to globally influence gene expression by controlling the localization of this key regulatory protein would be an efficient means for cells to respond to physiologic stresses but also represents a potential Achilles heel for viral attack. Further exploration of PABPC functions in the nucleus should therefore shed new light on how viruses coopt host gene expression pathways and provide mechanistic insight into the expanding roles of polyadenylation in determining RNA fate.