Evidence that Poly(A) Binding Protein C1 Binds Nuclear Pre-mRNA Poly(A) Tails

Abstract

In mammalian cells, poly(A) binding protein C1 (PABP C1) has well-known roles in mRNA translation and decay in the cytoplasm. However, PABPC1 also shuttles in and out of the nucleus, and its nuclear function is unknown. Here, we show that PABPC1, like the major nuclear poly(A) binding protein PABPN1, associates with nuclear pre-mRNAs that are polyadenylated and intron containing. PABPC1 does not bind nonpolyadenylated histone mRNA, indicating that the interaction of PABPC1 with pre-mRNA requires a poly(A) tail. Consistent with this conclusion, UV cross-linking results obtained using intact cells reveal that PABPC1 binds directly to pre-mRNA poly(A) tails in vivo. We also show that PABPC1 immunopurifies with poly(A) polymerase, suggesting that PABPC1 is acquired by polyadenylated transcripts during poly(A) tail synthesis. Our findings demonstrate that PABPC1 associates with polyadenylated transcripts earlier in mammalian mRNA biogenesis than previously thought and offer insights into the mechanism by which PABPC1 is recruited to newly synthesized poly(A). Our results are discussed in the context of pre-mRNA processing and stability and mRNA trafficking and the pioneer round of translation.

The 3′ ends of almost all eukaryotic mRNAs and their precursors consist of homopolymeric tails of adenosine, or poly(A), that are added by poly(A) polymerase (PAP) during the process of 3′ end formation. There are two classes of poly(A) binding proteins (PABPs) in mammalian cells (32, 40). One class is exemplified by PABPN1, formerly called PABP2. PABPN1 is primarily nuclear and plays a role in the synthesis of poly(A) tails, but it also shuttles between the nucleus and cytoplasm (5, 8, 9, 31, 53). The other class consists of the primarily cytoplasmic PABPs, of which PABPC1, formerly called PABI or PABP1, is the major form in mammalian somatic cells (40). In humans, at least four separate PABPC genes and four pseudogenes have been identified (40). PABPC1 influences mRNA translation and decay (18, 20, 23, 27, 29, 30, 50, 54-56), and it shuttles between the nucleus and cytoplasm of at least some mammalian cells (1, 57, 58). Consistent with the preferential compartmentalization of PABPN1 to the nucleus and PABPC1 to the cytoplasm, a physical interaction has been detected between PAP and PABPN1 but not PABPC1 (28). Furthermore, PABPN1 associates with RNA polymerase II during transcription and accompanies the released transcript to the nuclear pore (2). Given that PABPC1 can exist within nuclei, a key issue is whether PABPC1 binds to transcripts inside the nucleus, and if it does, at which step in mRNA maturation.

The 5′ ends of eukaryotic mRNAs and their precursors are capped, and the cap is initially bound by the mostly nuclear cap binding protein (CBP) heterodimer of CBP80 and CBP20 (25, 35, 38), which will be called CBP80/CBP20. CBP80/CBP20 is detectably replaced by the mostly cytoplasmic eukaryotic translation initiation factor 4E (eIF4E) only after introns have been removed by splicing (35). Evidence indicates that PABPC1 is a component of CBP80/CBP20-bound mRNA. First, PABPC1, like PABPN1, coimmunopurifies with CBP80 (11, 24). Second, PABP-interacting protein 2, which inhibits the interaction of PABPC1 with poly(A), inhibits nonsense-mediated mRNA decay (NMD) (11). NMD occurs as a result of nonsense codon recognition during a pioneer round of translation (11, 22, 24). The pioneer round is defined as the translation of CBP80/CBP20-bound mRNA, and it is distinct from steady-state translation, which is defined as the translation of eIF4E-bound mRNA. Therefore, PABPC1 is a functional component of CBP80/CBP20-bound mRNA. Whether PABPC1 is acquired prior to the completion of splicing and whether PABPN1 remains associated with mRNA during the pioneer round of translation have never been reported.

In this communication, we provide the first evidence that PABPC1 can bind the poly(A) tails of unspliced pre-mRNA. First, cell fractionation verifies the presence of PABPC1 within the nuclei of both HeLa CCL2 and Cos 7 cells, which we use in our studies. Second, not only PABPN1 but also PABPC1 coimmunopurifies with unspliced pre-mRNA. Since this finding was unexpected, we provide evidence in control experiments that the association of PABPC1 with unspliced pre-mRNA occurs in vivo, rather than after cell lysis as an artifact of immunopurification (IP). Furthermore, results of UV cross-linking using intact cells reveal that PABPC1 interacts directly with the poly(A) tail of unspliced pre-mRNA in vivo. Third, PABPC1 coimmunopurifies with PAP. These findings indicate that PABPC1 associates with transcripts earlier in mRNA biogenesis than previously thought. Therefore, it became important to understand how long PABPN1 remains associated with mRNA before it is completely replaced by PABPC1. We find that NMD reduces the abundance of nonsense-containing mRNA that is bound not only by PABPC1 but also by PABPN1. This result provides the first direct evidence that PABPN1 remains associated with mRNA during the CBP80/CBP20-mediated pioneer round of translation.

In summary, our results extend current views by providing evidence that PABPC1 is acquired by polyadenylated transcripts during poly(A) tail synthesis. Roles for PABPC1 in pre-mRNA processing and stability and mRNA trafficking and the pioneer round of translation are discussed.

MATERIALS AND METHODS

Plasmid constructions.

To construct pCMV-Myc-PABPC1, cDNA encoding human PABPC1 was generated using PCR, pSHREKK-PABPC1 (47), and the primer pair 5′ GCCAGATCTCTATGAACCCCAGTGCCCCCAGG 3′ (sense) and 5′ TGAGGTACCTTAAACAGTTGGAACACCGG 3′ (antisense). The resulting PCR product was cleaved with BglII and KpnI and inserted into the corresponding sites of pCMV-Myc (Clontech).

To construct pCMV-Myc-PABPN1, pEGFP-C2-PABPN1 (16) was digested with BglII and BamHI, and the resulting cDNA fragment encoding human PABPN1 was inserted into the BglII site of pCMV-Myc.

To construct pGEX6P2-PABPN1 for PABN1 production in Escherichia coli, pCMV-Myc-PABPN1 was digested with SalI and NotI, and the resulting cDNA encoding human PABPN1 was inserted into the corresponding sites of pGEX6P2 (Amersham Biosciences).

Cell transfections and protein and RNA purifications.

Cos 7 cell transfections using calcium (see Fig. 7) and HeLa CCL2 cell transfections using Lipofectamine 2000 (Invitrogen) (see Fig. 2 to 6) were as previously described (24, 36). Protein and RNA were purified from total, nuclear, or cytoplasmic fractions also as previously presented (3, 24, 35).

FIG. 7.

Both PABPN1 and PABPC1 are present on nonsense-containing mRNA that was reduced in abundance by NMD. Cos 7 cells (4 × 10⁷/150-mm dish) were transfected with a pmCMV-Gl test plasmid (10 μg), either nonsense-free (Norm) or containing a TAG nonsense codon at position 39 (Ter), and the phCMV-MUP reference plasmid (3 μg). Protein and RNA from nuclear fractions were purified either before (−) or after IP using anti-PABPC1, anti-PABPN1, or, as a control for nonspecific IP, NRS. (A) Diagram of pmCMV-Gl. ATG(0) indicates the translation initiation codon, TAG(39Ter) indicates the nonsense codon that typifies pmCMV-Gl Ter, and TAA(147) indicates the normal translation termination codon. The thick gray bar represents the CMV promoter, boxes represent exons, and lines between boxes represent introns. (B) Western blotting of nuclear fractions using anti-PABPC1 or anti-PABPN1. IP efficiencies using anti-PABPC1 or anti-PABPN1 were 24% ± 9% or 38% ± 6%, respectively. The four leftmost lanes, which analyzed twofold protein dilutions, indicate that the analysis is semiquantitative. (C) RT-PCR of nucleus-associated Gl and MUP mRNAs. The three leftmost lanes, which analyzed twofold dilutions of RNA, demonstrate that the analysis is quantitative. Numbers immediately below the figure represent the level of Gl mRNA normalized to the level of MUP mRNA, where the normalized level of Gl Norm mRNA prior to or after IP using anti-PABPC1 or anti-PABPN1 was defined as 100%. Results are representative of three independently performed experiments.

FIG. 2.

Both PABPC1 and PABPN1 bind to intron-containing transcripts that derive from either transiently introduced plasmids or the HeLa cell genome. HeLa CCL2 cells (10⁷/150-mm dish) were transiently transfected with pmCMV-Gl (10 μg) and phCMV-MUP (3 μg). Protein and RNA were isolated before (−) and after IP of the nuclear extract as described in the legend to Fig. 1 and analyzed by Western blotting (data not shown; IP efficiencies were 9% ± 2% when anti-PABPC1 was used and 38% ± 4% when anti-PABPN1 was used) and RT-PCR. To the right of each panel, sense (1) and antisense (2) PCR primers are specified by arrows above the introns to which they correspond within each transcript diagram. Transcript exons are indicated as boxes, and transcript introns are specified as lines. (A) RT-PCR of Gl pre-mRNA produced from pmCMV-Gl and cellular histone H4 mRNA. PCR of pmCMV-Gl or pmCMV-H4 provided a control for transcript size. (B) RT-PCR of MUP pre-mRNA produced from phCMV-MUP and, as a control, PCR of phCMV-MUP. (C) RT-PCR of cellular RPL36 pre-mRNA. (D) RT-PCR of cellular TPI pre-mRNA and, as a control, PCR of pmCMV-TPI. Results are representative of two independently performed experiments.

FIG. 6.

PABPC1, like PABPN1, coimmunopurifies with PAP-γ. (A) Protein from nuclear extracts of untransfected HeLa CCL2 cells (10⁷/150-mm dish) was immunopurified using anti-PAP-γ or, as a control for nonspecific IP, NRS. Western blotting using anti-PAP-γ or mAb414, which recognizes nucleoporin p62, demonstrated that the IP efficiency for anti-PAP-γ was 53% ± 7% and that the nucleoporin p62 was not detectable in the IP. (B) Protein from nuclear extracts of untransfected HeLa CCL2 cells (10⁷/150-mm dish) was immunopurified using anti-PABPC1, anti-PABPN1, or, as a control for nonspecific IP, NRS. Western blotting revealed that the IP efficiency for anti-PABPC1 or anti-PABPN1 was 11% ± 6% or 35% ± 4% and that the nucleoporin p62 was not detectable in either IP. (C) HeLa CCL2 cells (10⁷/150-mm dish) were transiently transfected with 30 μg of pCMV-Myc-PABPC1 (lanes 2 and 5), 3 μg of pCMV-Myc-PABPN1 plus 27 μg of pCMV-Myc (lanes 3 and 6) or, as a control, 30 μg of pCMV-Myc (lanes 1 and 4). Nuclear extracts were generated and analyzed either before (−) IP (lanes 1 through 3) or after IP using anti-Myc (lanes 4 through 6). Western blotting using anti-Myc demonstrated that the IP efficiency for Myc-PABPC1 or Myc-PABPN1 was 45% ± 5% or 56% ± 8%, respectively. The specificity of each IP was evident from the absence of detectable p62. Samples before IP (lanes 1 through 3) consist of half the cell equivalents of samples after IP. The three leftmost lanes (A and B) or four leftmost lanes (C), which analyzed threefold dilutions of protein, indicate that the analysis is semiquantitative. Results are representative of two independently performed experiments.

IPs.

IPs were as previously reported (24) with antibodies described below. When specified, extracts were mixed using end-over-end rotation for 2 h at 4°C.

UV cross-linking.

HeLa CCL2 cells (10⁷/150-mm dish) were exposed to UV for 5 min in 4 ml of ice-cold Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum with a Stratalinker UV Cross-Linker (Stratagene). Nuclear extracts were subsequently prepared, and a fraction of each extract was either incubated at 80°C for 10 min to dissociate noncovalent interactions or, as a control, left on ice. IPs were as previously reported (24) except anti-PABPC1 (33) was used.

Western blotting.

Proteins were electrophoresed in 10% polyacrylamide, transferred to a nitrocellulose membrane (Amersham Biosciences), and probed using rabbit polyclonal anti-PABPC1 (33), rabbit polyclonal anti-PABPN1 (raised against the peptide NH₂-RGSGPGRRRHLVPGAGGEC-COOH; D. Bear, M. Becher, T. Howard, and B. Reinert, unpublished data), rabbit polyclonal anti-phospholipase C gamma (anti-PLC-γ) (Santa Cruz), mouse monoclonal antibody mAb414 (which recognizes the nucleoporin p62; BAbCO), mouse monoclonal anti-Myc (BD Biosciences), goat polyclonal anti-glutathione S-transferase (anti-GST; Amersham Biosciences), or rabbit polyclonal anti-PAP-γ (34).

Reverse transcription-PCR (RT-PCR).

β-Globin (Gl) and major urinary protein (MUP) mRNAs were amplified as previously described (36). Intron-containing Gl transcripts were amplified using the following primer pairs: (i) 5′ GCCTATTGGTCTATTTTCCC 3 ′ (intron 1, sense) and 5′ GAGGAGGGGAAGCTGATATC 3′ (intron 2, antisense), (ii) intron 1, sense (see above) and 5′ GGGTTTAGTGGTACTTGTGAGC 3′ (exon 3, antisense), or (iii) 5′ ACCACCGTAGACGCAGATCG 3′ (exon 1, sense) and intron 2, antisense (see above). Cellular histone H4 mRNA was amplified using 5′ CCTGCGGTCATGTCCGGCCTGTG 3′ (sense) and 5′ GCGCTTGAGCGCGTTAACCACATCCATGGCTGTGACGG 3′ (antisense). Cellular triosephosphate isomerase (TPI) pre-mRNA was amplified using 5′ ACCTTGGCTTCATCTCTTCC 3′ (intron 2, sense) and 5′ GTGTCTGTCCAAACCTATTG 3′ (intron 5, antisense), MUP pre-mRNA was amplified using 5′ AGATAGAAGATAATGGCAAC 3′ (intron 1,sense) and 5′ AGGCGTGAGACCATACCAGG 3′ (intron 2, antisense), cellular ribosomal protein L36 (RPL36) pre-mRNA was amplified using 5′ GAGAGAAGCTGCTTAACTAG 3′ (intron 1, sense) and 5′ GGCCCTGACTCCCATCCCAC 3′ (intron 2, antisense), and cellular RPL36 mRNA was amplified using 5′ AAAGTGACCAAGAACGTGAG 3′ (exon 1, sense) and 5′ TCTTGGCGCGGATGTGCGTC 3′ (exon 3, antisense).

RESULTS

PABPC1 is detected in the nuclear fraction of HeLa CCL2 cells.

Although PABPC1 has been characterized as exclusively cytoplasmic in selected HeLa and Cos cell lines (48), more recent studies demonstrate that PABPC1 shuttles between the nucleus and cytoplasm in at least some HeLa and NIH 3T3 cell lines (1, 57, 58). We utilized cell fractionation and rabbit polyclonal anti-PABPC1 (33) to determine if PABPC1 was present in the nuclear fraction of HeLa CCL2 cells, since we used these cells in subsequent studies (see below). As a control, we also used rabbit polyclonal anti-PABPN1 (33) to assess the cellular distribution of PABPN1. Importantly, anti-PABPC1 specifically reacts with PABPC1, and anti-PABPN1 specifically reacts with PABPN1 based on two criteria. First, Western blotting of E. coli-produced and purified PABPC1, which is ∼71 kDa (19), reacted only with anti-PABPC1, and E. coli-produced and purified PABPN1, which is ∼33 kDa but migrates in sodium dodecyl sulfate-polyacrylamide as ∼49 kDa (46), reacted only with anti-PABPN1 (data not shown). Second, anti-PABPC1 and anti-PABPN1 reacted with appropriately sized HeLa CCL2 cell or Cos 7 cell proteins by Western blotting (data not shown). Cos 7 cells were also used in subsequent studies (see below).

Anti-PABPC1 may react not only with PABPC1 but also with other members of the PABPC class. However, we concluded that PABPC1 is the primary if not sole PABPC that was detected in our experiments for the following three reasons. First, PABPC1 is the most abundant of the PABPCs in somatic cells (17, 40). Second, RT-PCR analysis of PABPC1, PABPC3, PABPC4 (also called inducible PABP), and PABPC5 mRNAs (40) in HeLa CCL2 and Cos 7 cells revealed that, in addition to PABPC1 transcripts, only PABPC4 transcripts were detected (data not shown). Furthermore, the ratio of PABPC4 and PABPC1 transcripts in the two cell types was only ∼1:8 and ∼1:20, respectively (data not shown). Third, and most important, the anti-PABPC1 used in this study was raised against a peptide that is specific to PABPC1 (33).

Nuclear and cytoplasmic fractions of HeLa CCL2 cells were prepared (35) and analyzed by Western blotting using anti-PABPC1, anti-PABPN1, anti-PLC-γ (which controls for the purity of the nuclear fraction), and mAb414 (which recognizes the p62 component of the nuclear basket and controls for the purity of the cytoplasmic fraction). The nuclear fraction proved to be devoid of detectable PLC-γ, which is essentially exclusively a cytoplasmic protein (Fig. 1) (12). The cytoplasmic fraction contained at most 7% ± 5% of total-cell p62, which is largely nuclear (Fig. 1) (14). While PABPC1 was primarily in the cytoplasmic fraction, 25% ± 3% of total-cell PABPC1 was in the nuclear fraction (Fig. 1). In contrast, PABPN1 was largely in the nuclear fraction, although 18% ± 2% of total-cell PABPN1 was in the cytoplasmic fraction (Fig. 1). We conclude that PABPC1 exists within the nuclear fraction of HeLa CCL2 cells. PABPC1 also exists within the nuclear fraction of Cos 7 cells (data not shown; D.G. Bear, personal communication).

FIG. 1.

Not only PABPN1 but also PABPC1 is present in the nuclear fraction of HeLa CCL2 cells. Nuclear (lane 1) and cytoplasmic (lane 2) fractions were analyzed by Western blotting using anti-PLC-γ, mAb414 (which recognizes the nucleoporin p62), rabbit polyclonal anti-PABPC1, or rabbit polyclonal anti-PABPN1. The four leftmost lanes, which analyzed threefold dilutions of total-cell protein, indicate that the analysis is semiquantitative. The amount of total-cell PLC-γ, p62, PABPC1, or PABPN1 that was nuclear was 5% ± 1%, 93% ± 5%, 25% ± 3%, or 82% ± 2%, respectively. Results are representative of two independently performed experiments.

PABPC1, like PABPN1, copurifies with unspliced pre-mRNA, partially spliced pre-mRNA, and fully spliced mRNA but not nonadenylated RNA.

Since PABPC1 is present in nuclear fractions and is a functional component of the pioneer translation initiation complex, it was of interest to determine if PABPC1 binds the intron-containing precursors of the pioneer translation initiation complex. HeLa CCL2 cells were transiently transfected with two plasmids: (i) a pmCMV-Gl test plasmid that encodes Gl RNA (61) and (ii) a phCMV-MUP reference plasmid that encodes MUP RNA and controls for variations in transfection efficiency and RNA recovery (4). Neither RNA was expressed from the HeLa cell genome. Protein and RNA were isolated before and after IP of nuclear extract using anti-PABPC1, anti-PABPN1, or, to control for nonspecific IP, normal rabbit serum (NRS).

RT-PCR was used to amplify the region that extends from the first intron into the last intron of Gl pre-mRNA or the region that extends from the first intron into the second intron of MUP pre-mRNA. Results demonstrated that each intron-containing pre-mRNA was immunopurified using anti-PABPC1 or anti-PABPN1 but not NRS (Fig. 2A and B, compare lanes 2 and 3 to lane 1 in each). The identity of each RT-PCR product was verified by PCR analysis of pmCMV-Gl or phCMV-MUP using the same primer pair that was used to amplify the corresponding transcript (Fig. 2A and B, lanes 4).

RT-PCR was also used to amplify sequences that extend from the first intron into the last intron of RPL36 pre-mRNA or sequences that extend from the second intron into the penultimate intron of triosephosphate isomerase (TPI) pre-mRNA, each of which was derived from the HeLa cell genome. Results demonstrated that each intron-containing pre-mRNA was likewise immunopurified with anti-PABPC1 or anti-PABPN1 but not NRS (Fig. 2C and D, compare lanes 2 and 3 to lane 1 in each). The identity of the TPI RT-PCR product was verified by PCR analysis of pmCMV-TPI (60) using the same primer pair that was used to amplify the corresponding HeLa cell TPI transcript (Fig. 2D, lane 4).

To assess if IP of each pre-mRNA using either anti-PABPC1 or anti-PABPN1 depends on the presence of a poly(A) tail, histone H4 mRNA that derived from the HeLa cell genome was also assessed. Histone H4 transcripts undergo neither splicing nor polyadenylation (15). RT-PCR was used to amplify almost the full length of histone H4 mRNA. Results demonstrated that this nonadenylated mRNA was not detectably immunopurified with either anti-PABPC1 or anti-PABPN1 (Fig. 2A, lanes 2 and 3). Here again, the identity of the RT-PCR product was corroborated by PCR analysis of pmCMV-H4 (44) using the same primer pair that was used to amplify HeLa cell histone H4 mRNA (Fig. 2A, lane 5).

We conclude that PABPC1, like PABPN1, can bind intron-containing pre-mRNA. We also conclude that binding depends on a poly(A) tail, since nonadenylated RNA fails to bind either PABP. Consistent with this view, while splicing can occur cotranscriptionally, unspliced but polyadenylated pre-mRNA does exist in cells (6, 26, 42, 43). Furthermore, the presence of unspliced but polyadenylated Gl pre-mRNA in samples analyzed in Fig. 2 was corroborated by RT-PCR using primers that amplified sequences extending from Gl intron 2 into the poly(A) tail (data not shown).

An association of PABPC1 with intron-containing pre-mRNA has never been reported. Therefore, it was particularly important to determine if the association occurred in vivo or only after cell lysis as an artifact of the experimental procedure (45). If the association occurred in vivo, then it should be detected after IP of nuclear extract from cells that coexpressed Myc-PABPC1 and Gl pre-mRNA. However, the association should not be detected after IP of a mixture of (i) nuclear extract from cells that expressed Myc-PABPC1 but not Gl pre-mRNA and (ii) nuclear extract from cells that expressed Gl pre-mRNA but not Myc-PABPC1.

To determine if PABPC1 and Gl pre-mRNA interact in vivo, HeLa CCL2 cells were transiently transfected so that the total amount of introduced DNA was constant (25 μg) per transfection (see the legend to Fig. 3). In these transfections, pCMV-Myc (12 μg) controlled for the absence of pmCMV-PABPC1 (12 μg), pCMV-Myc (3 μg) controlled for the absence of phCMV-MUP (3 μg), and pCMV-Myc (10 μg) controlled for the absence of pmCMV-Gl (10 μg). Nuclear extracts were prepared, and protein and RNA were isolated prior to and after IP using anti-Myc. IPs were performed before or after mixing specific nuclear extracts.

FIG. 3.

IP of PABPC1 with Gl pre-mRNA is attributable to binding in vivo rather than to binding during extract preparation. HeLa CCL2 cells (10⁷/150-mm dish) were transiently transfected with the following plasmids so that the total amount of introduced DNA was constant: lanes 1 and 6, pmCMV-Gl (10 μg), phCMV-MUP (3 μg), and pCMV-Myc (12 μg); lanes 2 and 7, pmCMV-Gl (10 μg), phCMV-MUP (3 μg), and pCMV-Myc-PABPC1 (12 μg); lane 3, pmCMV-Gl (10 μg), pCMV-Myc (3 μg), and pCMV-Myc (12 μg); lane 4, pCMV-Myc (10 μg), phCMV-MUP (3 μg), and pCMV-Myc (12 μg); lane 5, pCMV-Myc (10 μg), phCMV-MUP (3 μg), and pCMV-Myc-PABPC1 (12 μg). Protein and RNA were isolated from nuclear extracts before (−) or after IP using anti-Myc. IPs were performed prior to (lanes 6 and 7) or after the specified extracts were mixed (lane 8 analyzed a mixture of the extracts that were analyzed separately in lanes 3 and 4; lane 9 analyzed a mixture of the extracts that were analyzed separately in lanes 3 and 5). (A) Western blotting using anti-Myc. (B) RT-PCR of Gl pre-mRNA from intron 1 into intron 2 (introns 1 and 2) and MUP mRNA to determine the extent of PABPC1 binding to Gl pre-mRNA that occurs after cell lysis. This extent was calculated to be 9% ± 2% (lane 9) of PABPC1 binding to Gl pre-mRNA that occurred in cells that had been cotransfected with pmCMV-Gl, phCMV-MUP, and pCMV-Myc-PABPC1, which was defined as 100% (lane 7). The asterisk indicates an uncharacterized RT-PCR product that did not interfere with experimental interpretations. Results are representative of two independently performed experiments.

Western blotting using anti-Myc revealed that Myc-PABPC1 was expressed at a similar level in cells that had been cotransfected with pmCMV-Gl, phCMV-MUP, and pCMV-Myc-PABPC1 and in cells that had been cotransfected with pCMV-Myc, phCMV-MUP, and pCMV-Myc-PABPC1 (Fig. 3A, compare lanes 2 and 5). Furthermore, IP using anti-Myc demonstrated that Myc-PABPC1 was immunopurified with similar efficiency from the corresponding nuclear extracts (Fig. 3A, compare lanes 7 and 9).

Using RT-PCR, the amount of Gl pre-mRNA that was immunopurified using anti-Myc and extracts of cells cotransfected with pmCMV-Gl, phCMV-MUP, and pCMV-Myc-PABPC1 was defined as 100% (Fig. 3B, lane 7). Only 9% ± 2% of this pre-mRNA was immunopurified using anti-Myc and a mixture of (i) extract from cells cotransfected with pCMV-Myc, phCMV-MUP, and pCMV-Myc-PABPC1 and (ii) extract from cells cotransfected with pmCMV-Gl and pCMV-Myc (Fig. 3B, lane 9). As expected, anti-Myc immunopurified neither Gl pre-mRNA nor MUP mRNA with extract from cells that did not express Myc-PABPC1 (Fig. 3B, lanes 6 and 8). These data indicate that only 9% ± 2% of Gl pre-mRNA that was bound by Myc-PABPC1 was due to binding in vitro. In other words, the bulk (i.e., 89 to 93%) of PABPC1 that copurified with Gl pre-mRNA reflected binding in vivo rather than binding after cell lysis. Notably, the level of Myc-PABPC1 was threefold the level of endogenous PABPC1 when cells were harvested (data not shown). Considering that pre-mRNA is newly synthesized, there was therefore adequate Myc-PABPC1 to compete with endogenous PABPC1 for binding to pre-mRNA.

If the association of PABPC1 with pre-mRNA typifies pre-mRNA that is productively processed to translationally active mRNA, then the partially and fully spliced products of pre-mRNA should also be bound by PABPC1. To test this hypothesis, the association of PABPN1 and PABPC1 with unspliced, partially spliced, and fully spliced Gl transcripts was assessed using the same nuclear IPs that were analyzed in Fig. 2. RT-PCR was used to amplify from the first to the last exon of Gl RNA. Results demonstrated that fully spliced (Gl mRNA), partially spliced (intron 1 only) pre-mRNA, and an electrophoretically inseparable mixture of unspliced (i.e., containing introns 1 and 2) and partially spliced (containing intron 2 only) pre-mRNAs were immunopurified using anti-PABPC1 or anti-PABPN1 but not NRS (Fig. 4A, left; compare lanes 2 and 3 to lane 1). The same RT-PCR products were subsequently analyzed using a lower-percentage acrylamide gel that separated pre-mRNA containing introns 1 and 2 from pre-mRNA containing intron 2 only. Results revealed that unspliced pre-mRNA and all three possible spliced variants were immunopurified using anti-PABPC1 or anti-PABPN1 but not NRS (Fig. 4A, right, compare lanes 2 and 3 to lane 1).

FIG. 4.

PABPC1, like PABPN1, binds to unspliced, partially spliced, and fully spliced Gl transcripts. Gl transcripts were analyzed before (−) and after IP using the same nuclear extract that was analyzed in Fig. 2 and the specified antibody. To the right of each panel, sense (1) and antisense (2) PCR primer pairs are indicated by arrows above each uppermost transcript diagram, where exons are shown as boxes and introns are shown as lines. (A) RT-PCR of Gl exon 1 through exon 3 and, as a control, PCR of the same region of pmCMV-Gl or pmCMV-Gl Δ(intron 2). Samples were electrophoresed in either 5% (left) or 3.5% (right) polyacrylamide, the latter of which separated unspliced pre-mRNA (introns 1 and 2) from partially spliced pre-mRNA (intron 2 only). Dots indicate intron 1-only RNA, and triangles indicate an inseparable mixture of Gl RNA containing introns 1 and 2 and Gl RNA containing intron 2 only. (B) RT-PCR of Gl exon 1 through intron 2 and, as a control, PCR of pmCMV-Gl. Results are representative of two independently performed experiments.

RT-PCR was also used to amplify sequences from the first exon to the last intron of Gl transcripts. Results showed that partially spliced (intron 2 only) pre-mRNA and unspliced (introns 1 and 2) pre-mRNA were immunopurified using anti-PABPC1 or anti-PABPN1 but not NRS (Fig. 4B, compare lanes 2 and 3 to lane 1). Notably, the identity of RT-PCR products that derived from Gl transcripts that were unspliced or contained only intron 1 was verified by PCR analysis of, respectively, pmCMV-Gl or pmCMV-Gl Δ(intron 2) (61) using the same primer pair that was used to amplify the corresponding RNA (Fig. 4A, lanes 4 and 5, and B, lane 4). These findings indicate that PABPC1 binds unspliced Gl pre-mRNA and remains associated during splicing.

PABPC1 can be UV cross-linked in vivo to form a heat-resistant complex with Gl pre-mRNA and Gl mRNA but not with histone mRNA.

To determine if the association of PABPC1 with Gl pre-mRNA is direct, the ability of PABPC1 to form covalent UV cross-links with Gl pre-mRNA was examined. HeLa CCL2 cells that had been transiently transfected with pmCMV-Gl were either exposed to UV at 0°C, which induces covalent RNA-protein cross-links in vivo, or not exposed to UV. Nuclear extracts were subsequently generated, and a fraction of each extract was either incubated at 80°C for 10 min, which dissociates noncovalent interactions, or left at 0°C. Protein and RNA were then isolated from the variously treated extracts before or after IP using anti-PABPC1 or, to control for nonspecific IP, NRS.

Western blotting of samples prior to IP using anti-PABPC1 revealed that exposure to UV, heat, or both reduced the recovery of PABPC1 (Fig. 5A, compare lane 1 to lanes 2 through 4). However, IP efficiencies for all extracts were comparable (Fig. 5A, compare lane 6 to lanes 8, 10, and 12; Fig. 5, legend). RT-PCR of samples prior to IP revealed that exposure to UV, heat, or both also reduced the recovery of Gl mRNA, Gl pre-mRNA, and histone H4 mRNA (Fig. 5B, compare lane 1 to lanes 2 through 4). Additionally, anti-PABPC1 immunopurified more Gl mRNA than Gl pre-mRNA in the absence of UV or heat, consistent with the relative steady-state levels of the two RNAs (Fig. 5B, lane 6).

FIG. 5.

PABPC1 can be UV cross-linked in vivo to form heat-resistant complexes with Gl pre-mRNA and Gl mRNA but not histone H4 mRNA. HeLa CCL2 cells (10⁷/150-mm dish) that had been transiently transfected with pmCMV-Gl (10 μg) were either exposed to UV for 5 min to induce covalent RNA-protein cross-links in vivo (+UV) or not exposed to UV (−UV). Nuclear extracts were subsequently prepared, and a fraction of each extract was either incubated at 80°C for 10 min (+Heat) to dissociate noncovalent interactions or left on ice (−Heat). Protein and RNA were isolated from the variously treated extracts before IP (−IP) or after IP using anti-PABPC1 or, as a control for nonspecific IP, NRS. (A) Western blotting using anti-PABPC1 demonstrated that IP efficiencies were 25% ± 6%. The four leftmost lanes, which analyzed threefold dilutions of nuclear protein prior to IP, indicate that the analysis is semiquantitative. (B) RT-PCR of Gl mRNA, Gl pre-mRNA, or histone H4 mRNA. Samples before IP (lanes 1 to 4) consist of half the cell equivalents of samples after IP. The four leftmost lanes, which analyzed twofold dilutions of nuclear RNA prior to IP, demonstrate that the analysis is quantitative. (C) Quantitation of Gl mRNA and pre-mRNA levels that were immunopurified using anti-PABPC1. After UV and heat treatments, the level of immunopurified Gl mRNA was normalized to the level of Gl mRNA prior to IP. The normalized level was subsequently expressed as a percentage of the normalized level of the Gl mRNA that had not been exposed to UV or heat treatment, which was defined as 100. The same analysis was performed for Gl pre-mRNA. Values derive from two independently performed experiments.

As described above (Fig. 2A), anti-PABPC1 failed to immunopurify histone H4 mRNA because it lacked a poly(A) tail (Fig. 5B, lane 6). Anti-PABPC1 immunopurified Gl mRNA and Gl pre-mRNA after UV treatment without subsequent heat treatment (Fig. 5B, lane 8), as well as with subsequent heat treatment (Fig. 5B, lane 12), but not with heat treatment in the absence of UV treatment (Fig. 5B, lane 10). When we compensated for the reduced recovery of Gl mRNA and Gl pre-mRNA that occurred after UV and heat treatments, we found that the efficiency with which Gl mRNA coimmunopurified with PABPC1 after treatment was essentially the same as the efficiency with which Gl pre-mRNA coimmunopurified with PABPC1 after treatment (compare 73% ± 9% to 75% ± 13%) (Fig. 5C).

We conclude that nuclear PABPC1 can be UV cross-linked to Gl mRNA and Gl pre-mRNA in a way that is resistant to heat that dissociates noncovalent bonds. The simplest interpretation of these results is that nuclear PABPC1 interacts directly with the poly(A) tail of not only Gl mRNA but also Gl pre-mRNA in intact cells.

PABPC1 coimmunopurifies with PAP-γ.

To gain a better understanding of when in pre-mRNA metabolism PABPC1 associates with newly synthesized poly(A) tails, we tested if HeLa cell PABPC1 coimmunopurifies with HeLa cell PAP. Bovine PABPN1 that was synthesized in vitro using rabbit reticulocyte lysates has been shown to copurify with bovine PAP that was synthesized in E. coli as a GST-tagged protein (28). This and other findings indicate that PABPN1 stimulates PAP activity, and thus poly(A) tail elongation, by recruiting PAP to substrate RNAs (28). In theory, PABPC1 could also coimmunopurify with PAP.

The ability of PABPC1 to coimmunopurify with PAP-γ, one of several HeLa cell PAP isoforms, was tested for two reasons. First, this isoform is known to function in a mechanism that depends on both cleavage-polyadenylation specificity factor and the AAUAAA polyadenylation signal (34). Second, high-quality anti-PAP-γ is available (34). Protein was isolated before or after IP using nuclear extracts from untransfected HeLa CCL2 cells and rabbit polyclonal anti-PAP-γ or, as a negative control, NRS.

As expected, anti-PAP-γ immunopurified PAP-γ and PABPN1 but failed to immunopurify the nucleoporin p62 (Fig. 6A, lane 2). Anti-PAP-γ also immunopurified PABPC1 (Fig. 6A, lane 2), demonstrating that PAP-γ and PABPC1 do, indeed, interact. Even though the interaction may not be direct, these data suggest that PABPC1 may be acquired by newly synthesized, polyadenylated transcripts during poly(A) tail synthesis. Anti-PAP-γ reproducibly immunopurified a larger percentage of nuclear PABPN1 (13% ± 3%) than nuclear PABPC1 (5% ± 1%). These percentages indicate that there is likely to be more of each PABP associated with poly(A) tails after polyadenylation than with PAP-γ during polyadenylation, which makes sense, given that most poly(A) tails in a cell are not associated with PAP.

The co-IP of both PABPN1 and PABPC1 with PAP-γ implies that antibody to either PABP should immunopurify the other PABP. Using nuclear fractions from untransfected HeLa CCL2 cells, anti-PABPN1 immunopurified not only PABPN1 but also PABPC1 and, as expected, failed to immunopurify p62 (Fig. 6B, lane 3). Unexpectedly, however, while anti-PABPC1 immunopurified PABPC1, it failed to immunopurify PABPN1 (Fig. 6B, lane 2). Failure may be experimentally induced, e.g., because anti-PABPC1 binding to PABPC1 interferes with PABPC1 binding to PABPN1. Alternatively, the fraction of cellular PABPC1 that associates with PABPN1 may be sufficiently low to preclude easy detection.

As shown by a different experimental approach, the second interpretation appears to be correct. In this approach, HeLa CCL2 cells were transiently transfected with pCMV-Myc-PABPC1, pCMV-Myc-PABPN1, or pCMV-Myc. These plasmids produced Myc-tagged PABPC1, Myc-tagged-PABPN1, or Myc tag, respectively. Protein was purified from nuclear fractions before and after IP using anti-Myc. Results of Western blotting using anti-Myc indicated that although Myc-PABPN1 was expressed at a higher level than Myc-PABPC1, both Myc-PABPN1 and Myc-PABPC1 were readily detectable both before IP (Fig. 6C, lanes 2 and 3) and after IP (Fig. 6C, lanes 5 and 6). Western blotting using anti-PABPN1 confirmed that Myc-PABPC1 coimmunopurifies with cellular PABPN1 (Fig. 6C, lane 6). Western blotting using anti-PABPC1 revealed that Myc-PABPN1 indeed coimmunopurifies with cellular PABPC1, although a relatively dark exposure was required for detection (Fig. 6C, lane 5, darker exposure). Additionally, Western blotting using anti-PAP-γ showed that both Myc-PABPC1 and Myc-PABPN1 coimmunopurified with PAP-γ (Fig. 6C, lanes 5 and 6).

PABPN1, like PABPC1, is present on mRNA that is a substrate for NMD.

Given our finding that PABPC1 binds polyadenylated transcripts earlier than was previously appreciated, it became important to gain insight into how long PABPN1 remains associated with these transcripts. CBP80 coimmunopurifies with the C-terminal domain of RNA polymerase II (35) and, in the insect Chironomus tentans, binds cotranscriptionally to RNA caps (52). Additionally, Chironomus tentans PABPN1 can be visualized on elongating transcription complexes (2), and mammalian PABPN1 binds to RNA during poly(A) tail synthesis (53). However, though PABPN1 copurifies with CBP80, it may be removed by the time CBP80/CBP20-bound mRNA undergoes the pioneer round of translation (11, 24, 37).

Since NMD in mammalian cells targets CBP80/CBP20-bound mRNA but not detectably eIF4E-bound mRNA, one way to determine if PABPN1 remains associated with poly(A) during the pioneer round of translation is to assess if anti-PABPN1 immunopurifies nonsense-containing mRNA that has been reduced in abundance by NMD. If it does, then PABPN1 must be a component of the pioneer translation initiation complex. To this end, Cos 7 cells were transiently transfected with two plasmids: (i) a pmCMV-Gl test plasmid that encodes either nonsense-free (Norm) or nonsense-containing (Ter) β-Gl mRNA (Fig. 7A) (61) and (ii) the phCMV-MUP reference plasmid. Protein and RNA were purified from nuclear fractions before and after IP using anti-PABPC1, anti-PABPN1, or, as a control for nonspecific IP, NRS.

Western blotting of proteins from the nuclear fraction demonstrated that anti-PABPC1 immunopurified PABPC1 (Fig. 7B, top, lanes 3 and 4) and anti-PABPN1 immunopurified PABPN1 (Fig. 7B, bottom, lanes 5 and 6), whereas NRS immunopurified neither PABP (Fig. 7B, lanes 1 and 2). Furthermore, anti-PABPN1 immunopurified PABPC1 (Fig. 7B, top, lanes 5 and 6). Therefore, the two PABPs coimmunopurify has expected (Fig. 6). As in Fig. 6B, anti-PABPC1 failed to immunopurify PABPN1 (Fig. 7B, bottom, lanes 3 and 4), although an interaction between Myc-PABPC1 and cellular PABPN1 was detectable, which is consistent with data indicating that the two proteins do copurify (Fig. 6C).

RT-PCR of RNA from the nuclear fraction, where the NMD of nonsense-containing Gl mRNA occurs (36, 61), demonstrated that, prior to IP, the level of Gl Ter mRNA was 22% ± 5% the level of Gl Norm mRNA (Fig. 7C, compare lane 2 to lane 1). The level of Gl Ter mRNA that was immunopurified using either anti-PABPC1 or anti-PABPN1 was similarly reduced (Fig. 7C, compare lane 6 to lane 5 or lane 8 to lane 7, respectively). As expected, NRS immunopurified neither Gl mRNA nor MUP mRNA (Fig. 7C, lanes 3 and 4). These data indicate that NMD targets mRNA that is bound by PABPC1 and PABPN1 and that PABPN1 remains bound during the pioneer round of translation.

DISCUSSION

Our results provide the first evidence that PABPC1, like PABPN1, can associate in vivo with intron-containing, polyadenylated transcripts that go on to be productively spliced. Furthermore, PABPC1, like PABPN1, binds directly to the poly(A) tail of these transcripts. First, PABPC1 is present within the nuclei of both cell types that we examined (Fig. 1; data not shown). Second, anti-PABPC1, like anti-PABPN1, immunopurifies every unspliced pre-mRNA, partially spliced pre-mRNA, and fully spliced mRNA that we tested (Fig. 2 and 4). Third, cellular PABPC1 binding to these transcripts is dependent on a poly(A) tail, as evidenced by the failure of anti-PABPC1 to immunopurify nonadenylated histone H4 mRNA (Fig. 2 and 5). Fourth, anti-PABPC1 immunopurifies pre-mRNA, largely as a consequence of binding in vivo and not because of an experimental artifact (Fig. 3). Fifth, cellular PABPC1 remains UV cross-linked to pre-mRNA under conditions that dissociate noncovalent bonds, and cross-linking depends on the presence of a poly(A) tail (Fig. 5). Sixth, cellular PAP-γ coimmunopurifies with cellular PABPC1, and Myc-PABPC1 coimmunopurifies with cellular PAP-γ (Fig. 6). Our ability to detect an interaction between PAP-γ and PABPC1 contrasts with data demonstrating that purified bovine PAP-γ and Xenopus laevis PABPC1 do not detectably interact (28). However, Xenopus PABPC1 may be unable to interact with human PAP because of incompatibility between species, because mammalian PABPC1 functions in ways that Xenopus PABPC1 does not, or because in vitro binding conditions were insufficient to support the interaction of PABPC1 with PAP-γ.

We also find that E. coli-produced and purified PABPC1 copurifies with E. coli-produced and purified GST-PABPN1 but not GST alone (data not shown). However, considering there are several examples of proteins that interact in vitro but not apparently in cells (e.g., Staufen1 and Staufen2) (49; H. A. Kuzmiak and L. E. Maquat, unpublished data), it would be premature to conclude that PABPN1 and PABPC1 interact directly in mammalian cells. We favor the interpretation that PABPC1 associates with unspliced pre-mRNA directly via the poly(A) tail because PABPC1 can be UV cross-linked to pre-mRNA in intact cells. Nevertheless, we cannot exclude the possibility that PABPC1 also associates via PABPN1.

In view of the unexpectedly early step at which PABPC1 is acquired during mRNA biogenesis, it became important to determine how long PABPN1 remains associated with mRNA, given that PABPN1 is ultimately replaced by PABPC1. We have found that PABPN1 remains associated with mRNA during the pioneer round of translation, for which CBP80/CBP20-bound mRNA serves as a template. This was evidenced by the ability of anti-PABPN1 to immunopurify nonsense-containing mRNA that had already been reduced in abundance by NMD (Fig. 7). Even though anti-CBP80 immunopurifies PABPN1 and PABPC1 (11, 24) and anti-eIF4E immunopurifies only PABPC1 (11, 24), we cannot be certain that PABPN1 is no longer present after CBP80 and CBP20 are replaced by eIF4E, since the fraction of eIF4E-bound mRNA that is bound by PABPN1 may be too small to detect.

We conclude that both PABPN1 and PABPC1 bind to the poly(A) tail of newly synthesized transcripts in the nucleus, at least in some cases prior to splicing. Consistent with this, not all splicing occurs cotranscriptionally in mammalian cells and, consequently, unspliced or partially spliced polyadenylated transcripts do exist (6, 26, 42, 43; data not shown).

PABPN1 has been shown to function in nuclear polyadenylation by recruiting PAP to transcripts and, by so doing, increasing the processivity of PAP and controlling the length of the newly synthesized poly(A) tail (5, 28, 53). PABPN1 has also been functionally implicated in mRNA transport, since it shuttles between the nucleus and the cytoplasm (9, 10), and immunoelectron microscopy demonstrates that it is present on nuclear messenger ribonucleoprotein particles (mRNPs) that transit the nuclear envelope (2). A functional homolog to mammalian PABPN1 in Saccharomyces cerevisiae has not been found. The S. cerevisiae homolog to mammalian PABPC1, Pab1p, is known to play a role in both cytoplasmic mRNA metabolism, including mRNA translation and decay, as well as in nuclear RNA metabolism, including polyadenylation (32, 40) and mRNA export (7). Recent studies have shown that PABPC1, like yeast Pab1p (7), shuttles between the nucleus and the cytoplasm (1, 58).

Our data indicate PABPC1 associates with nuclear pre-mRNP prior to intranuclear transport by directly binding poly(A), most likely simultaneously with PABPN1. PABPC1 bound to newly synthesized poly(A) tails may function in pre-mRNA metabolism. For example, it could protect nuclear pre-mRNA from decapping, which is known to occur within nuclei, much as it protects cytoplasmic mRNA from decapping (30, 55, 56). As another example, PABPC1 could promote nuclear poly(A) nuclease 2 (Pan2)-mediated pre-mRNA poly(A) trimming, as it does in the cytoplasm (51), since both Pan2 and Pan3, the latter of which tethers Pan2 to PABPC1, are known to shuttle between the nucleus and the cytoplasm (59). In fact, Pan2p and Pan3p, the S. cerevisiae orthologs of mammalian Pan2 and Pan3, are known to regulate nuclear poly(A) tail length (39, 41).

PABPC1 may also influence the metabolism of nuclear mRNP. For example, as polyadenylated mRNP approaches the nuclear pore and is unfolded for transit through the nuclear pore complex (13), poly(A)-bound PABPC1 may play a role during transit by further nucleating PABPC1 assembly and concomitantly removing PABPN1, a process that would be completed after the pioneer round of translation. Another possibility is that PABPC1 primarily assembles with RNP and functions in the nucleus at a step that is much earlier than transit across the pore. In either scenario, PABPC1 may also be involved in the localization of particular mRNAs to specific regions of the cytoplasm. For example, PABPC1 was recently shown to bind directly to paxillin, which is an abundant protein of focal complexes at the leading edges of migrating cells (57, 58). The PABPC1-paxillin complex localizes to the perinuclear endoplasmic reticulum and the leading edge of the migrating cell plasma membrane in mouse fibroblasts. Thus, nuclear PABPC1 could play a role in assembling proteins at the 3′ untranslated region of an mRNP that are subsequently required for site-specific cytoplasmic localization of that mRNP.

PABPC1 function within nuclei would occur prior to its earliest certified role, which is during the pioneer round of translation. This round of translation involves CBP80/CBP20-bound mRNA, supports the decay of nonsense-containing mRNAs, precedes the translation of eIF4E-bound mRNA, and most often occurs in association with nuclei, probably during mRNA export to the cytoplasm (11, 24, 35, 37, 58). It is likely that PABPC1 augments the translation of CBP80/CBP20-bound mRNA in a manner similar to its augmentation of the translation of eIF4E-bound mRNA. PABPC1 binding to eIF4G increases the efficiency of translation initiation (27, 32), and eIF4G is a functional component of the pioneer translation initiation complex (37). Furthermore, PABPC1 binding to eukaryotic translation release factor 3 (21) may increase the efficiency of translation termination.

Future studies aim to determine at what step in RNA metabolism the bulk of PABPC1 binding to poly(A) tails occurs and specifically how PABPC1 functions prior to its role in the pioneer round of translation.