Identification of a Human Cytoplasmic Poly(A) Nuclease Complex Stimulated by Poly(A)-binding Protein*

The poly(A) tail shortening in mRNA, called deadenylation, is the first rate-limiting step in eukaryotic mRNA turnover, and the polyadenylate-binding protein (PABP) appears to be involved in the regulation of this step. However, the precise role of PABP remains largely unknown in higher eukaryotes. Here we identified and characterized a human PABP-dependent poly(A) nuclease (hPAN) complex consisting of catalytic hPan2 and regulatory hPan3 subunits. hPan2 has intrinsically a 3 (cid:1) to 5 (cid:1) exoribonuclease activity and requires Mg 2 (cid:1) for the enzyme activity. On the other hand, hPan3 interacts with PABP to simulate hPan2 nuclease activity. Inter-estingly, the hPAN nuclease complex has a higher substrate specificity to poly(A) RNA upon its association with PABP. Consistent with the roles of hPan2 and hPan3 in mRNA decay, the two subunits exhibit cytoplasmic co-localization. Thus, the human PAN complex is a poly(A)-specific exoribonuclease that is stimulated by PABP in the cytoplasm.

The poly(A) tail shortening in mRNA, called deadenylation, is the first rate-limiting step in eukaryotic mRNA turnover, and the polyadenylate-binding protein (PABP) appears to be involved in the regulation of this step. However, the precise role of PABP remains largely unknown in higher eukaryotes. Here we identified and characterized a human PABP-dependent poly(A) nuclease (hPAN) complex consisting of catalytic hPan2 and regulatory hPan3 subunits. hPan2 has intrinsically a 3 to 5 exoribonuclease activity and requires Mg 2؉ for the enzyme activity. On the other hand, hPan3 interacts with PABP to simulate hPan2 nuclease activity. Interestingly, the hPAN nuclease complex has a higher substrate specificity to poly(A) RNA upon its association with PABP. Consistent with the roles of hPan2 and hPan3 in mRNA decay, the two subunits exhibit cytoplasmic co-localization. Thus, the human PAN complex is a poly(A)-specific exoribonuclease that is stimulated by PABP in the cytoplasm.
Eukaryotic mRNAs have two major features, a 5Ј-terminal cap and a 3Ј-terminal poly(A) tail. Both of them play important roles in eukaryotic gene expression, especially in translation and mRNA decay processes. mRNAs are synergistically translated in the presence of both the 5Ј-cap and the 3Ј-poly(A) tail (1)(2)(3)(4). During translation, the 5Ј-cap and the 3Ј-poly(A) tail are recognized by eIF4E and the poly(A)-binding protein (PABP), 1 respectively, and eIF4G mediates their association (5,6). These result in the formation of a circularized mRNA (7), which provides a structural basis for the hypothetical machinery of efficient translation; ribosomes after translation termination are recruited to the next cycle of translation initiation (8 -11). The removal of the poly(A) tail from mRNA leads to translation inhibition and is used as a strategy to silence certain maternal mRNAs during oocyte maturation and early embryonic development. On the other hand, both the 5Ј-cap and 3Ј-poly(A) tail are also involved in the regulation of mRNA decay (12). The removal of the poly(A) tail is the first rate-limiting step in the degradation of most mRNAs (13)(14)(15)). This step is followed by the removal of the 5Ј-cap and exonucleolytic degradation of the mRNA body. Thus, deadenylation greatly affects gene expression in regard to abundance as well as the translation of mRNA.
Many factors involved in mRNA decay have been identified. The 5Ј-cap is removed by decapping enzymes termed Dcps. Dcp1, Dcp2, and DcpS were identified in Saccharomyces cerevisiae and metazoans (16 -22). On the other hand, the 3Јpoly(A) tail is degraded by deadenylases, and the Ccr4/Pop2 and PAN nuclease (consisting of Pan2 and Pan3) were identified as the two major mRNA deadenylases in S. cerevisiae (23)(24)(25)(26)(27). A strain lacking both Ccr4 and Pan2 exhibits no deadenylating activity (27). Yeast PAN nuclease is characterized by the requirement of Pab1, the yeast PABP, for its deadenylating activity. In addition, two other deadenylases in metazoans were reported, namely the poly(A)-specific ribonuclease PARN (28,29) and nocturnin (30). PARN is the most extensively investigated deadenylase in higher eukaryotes and is related to the enzyme activity involved in default deadenylation during Xenopus oocyte meiotic maturation (31). The most prominent feature of PARN is that its deadenylating activity is stimulated by the 5Ј-cap on mRNA (32)(33)(34). Nocturnin is a novel gene that is rhythmically expressed in the cytoplasm of retinal photoreceptor cells in a circadian clock-dependent manner (35) and is structurally related to the C-terminal deadenylase domain of Ccr4 (36). However, the orthologues of PARN and nocturnin are not present in S. cerevisiae.
The roles of PABP in mRNA decay are enigmatic. Biochemical experiments showed that PABP can protect the poly(A) tail from degradation (37,38) and that, in higher eukaryotes, PABP inhibits the deadenylating activity of PARN under physiological conditions (28,31). On the other hand, genetic approaches indicate that poly(A) tail shortening rates of mRNA are significantly reduced in S. cerevisiae strains lacking the Pab1 (39 -41).
In this study, we cloned the full-length human Pan2 (hPan2) and Pan3 (hPan3) cDNAs from HeLa cells. Biochemical analysis revealed that hPan2 is a Mg 2ϩ -dependent exoribonuclease and that hPan3 interacts with both hPan2 and PABP simultaneously. The intrinsic deadenylating activity of hPan2 is stimulated by PABP in the presence of hPan3, and the hPAN complex has a higher substrate specificity to poly(A) RNA when it associates with PABP. hPan2 and hPan3 exhibit cytoplasmic co-localization, consistent with their role in mRNA decay.

EXPERIMENTAL PROCEDURES
Plasmids-To express hPan2 in mammalian cells, the full-length hPan2 cDNA was amplified by reverse transcription PCR using total RNA from HeLa cells and inserted between the EcoRI and SalI sites of pFLAG-CMV-2 (Eastman Kodak Co.) to produce pFLAG-hPan2. pHA-hPan2, which contains an HA tag instead of FLAG, was also constructed. To express the deletion mutants of hPan2, cDNA fragments encoding the indicated amino acid sequences of hPan2 were inserted in pFLAG-CMV2 to construct pFLAG-hPan2 1-1028, 1-330, 311-1198, 311-1028, and 841-1198. pFLAG-hPan2 D1083A was generated by converting the corresponding GAC codon (Asp-1083) of pFLAG-hPan2 to GCC (alanine). To clone the full-length hPan3 cDNA, the 5Ј-end of XP170737 was amplified using the 5Ј-Full RACE core set (TaKaRa Bio Inc.) and the total RNA derived from HeLa cells. The obtained fulllength hPan3 cDNA was inserted between the HindIII and PstI sites of pFLAG-CMV-2 to construct pFLAG-hPan3; pHA-hPan3, which contains an HA tag instead of FLAG, was also constructed. pFLAG-hPan3⌬N contains the cDNA fragment encoding amino acids 291-688 of hPan3.
Cell Culture and DNA Transfection-COS-7 and HeLa cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum and maintained at 37°C in 5% CO 2 . Transfection was performed using Lipofectin or LipofectAMINE 2000 (Invitrogen).
Immunoprecipitation-The transfected cells were lysed in buffer A consisting of 20 mM Tris-HCl (pH 8), 50 mM NaCl, 1% Nonidet P-40, 1 mM dithiothreitol, 2.5 mM EDTA-sodium (pH 8), 100 M phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, and 2 g/ml leupeptin with 10 g/ml boiled RNase A. After centrifugation at 15,000 ϫ g for 20 min, the lysate was incubated at 4°C for 30 min with anti-FLAG IgG agarose (Sigma), and then the resin was washed with buffer A. When necessary, recombinant proteins were added, and the resin was further incubated at 4°C for 60 min. After washing with buffer A, proteins retained on the resin were subjected to SDS-PAGE and immunoblot analysis.
Immunofluorescence-HeLa cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) and 10% fetal calf serum on a polylysinecoated cover glass. The transfected cells were fixed with 4% paraformaldehyde for 15 min at room temperature. After a 15-min quenching with 100 mM glycine, the fixed cells were incubated for 15 min with 0.1% Triton X-100 and washed three times with phosphate-buffered saline. The cells were then incubated with an appropriate primary antibody for 1 h, washed three times with phosphate-buffered saline, and incubated with the appropriate secondary antibody (Alexa Fluor 488-conjugated anti-mouse IgG, Alexa Fluor 568-conjugated antimouse IgG, or Alexa Fluor 568-conjugated anti-rat IgG) with or without Pico Green (Molecular Probe). After three washes with phosphatebuffered saline, confocal images were obtained using a Zeiss LSM-510 confocal laser-scanning microscope.
Production of Recombinant Proteins-The purification of N-terminally GST-fused human PABP (PCBPC1) was described previously (42). Briefly, the recombinant PABP was induced by adding 0.1 mM isopropyl-1-thio-D-galactopyranoside at 37°C for 3 h to the Escherichia coli JM109 culture. The cells were resuspended in buffer B consisting of 50 mM Tris-HCl (pH 8), 1 mM EDTA, 150 mM NaCl, 1% Nonidet P-40, 2 g/ml aprotinin, 100 M phenylmethylsulfonyl fluoride, and 2 g/ml leupeptin. After incubation with 1 mg/ml lysozyme at 4°C for 30 min, the cells were lysed by sonication for 3 min on ice, and the resulting lysate was then centrifuged at 100,000 ϫ g for 60 min. The supernatant was subjected to glutathione-Sepharose 4B (Amersham Biosciences).
In Vitro Nuclease Assay-The production of internally 32 P-radiolabeled poly(A) RNA was described previously (26). Briefly, poly(A) RNA (23-mers) was extended using [␣-32 P]ATP and yeast poly(A) polymerase (USB Corp.). The probe was purified by gel filtration using S-400HR (Amersham Bioscience) and ethanol precipitation. For the production of 32 P-5Ј-radiolabeled poly(A) RNA and poly(dA) DNA, the nonlabeled nucleotides (Amersham Bioscience) were incubated with [␥-32 P]ATP and T4 polynucleotide kinase (TaKaRa Bio, Inc.) and then purified by elution from polyacrylamide gel and ethanol precipitation. For the immunopurification of recombinant proteins, extracts prepared from COS-7 cells expressing FLAG-tagged hPan2 with or without HA-tagged hPan3 were immunoprecipitated using anti-FLAG antibody in buffer A. After three washes with buffer A, the precipitated proteins were eluted with buffer C consisting of 10 mM HEPES-sodium (pH 7.5) and 2 mM MgCl 2 with 100 g/ml FLAG peptides (Sigma). The immunopurified proteins were incubated at 37°C for the indicated times in buffer C containing 2 mM dithiothreitol, 1 unit/l RNasin (Promega), and 1 ϫ 10 5 cpm of radiolabeled poly(A) RNA or poly(dA) DNA in the presence or absence of 1 mM spermidine. When necessary, recombinant GST-fused PABP at the indicated concentrations was added to the reactions with or without poly(A) or poly(C) RNA (Amersham Bioscience). The extent of nucleolysis was measured as follows. 1) Aliquots were precipitated with 25% trichloroacetic acid by centrifugation at 15,000 rpm, and the radioactivity of soluble nucleotide was measured using a liquid scintillation counter as described previously (26,43). 2) The liberated reaction product was analyzed by TLC on a polyethyleneimine-cellulose F plate (Merck, number 5579) using 0.75 M KH 2 PO 4 -H 3 PO 4 (pH 3.5) as the solvent (34). The positions of the unlabeled 5Ј-AMP and 3Ј-AMP (Sigma) were determined using UV light. 3) Aliquots were analyzed on a denaturing 12% polyacrylamide, 7 M urea gel. The radioactive molecules were visualized by autoradiography.
Nucleotide Sequence Accession Numbers-The sequence data of hPan2 and hPan3 have been submitted to the DDBJ/EMBL/Gen-Bank TM databases under accession numbers AB107584 and AB107585, respectively.

Molecular
Cloning of Human Pan2-To identify mammalian PAN nuclease, we first searched for the human homologue of the yeast Pan2 protein in nucleotide databases. The sequence of KIAA0710 in the human expressed sequence tag (EST) data base was used to design reverse transcription PCRs for isolating the full-length hPAN2 cDNA from HeLa cells. As schematized in Fig. 1A, the deduced hPan2 contains 1198 amino acids and shows sequence similarities to putative Pan2 homologues from Drosophila melanogaster, Caenorhabditis elegans, and S. cerevisiae (43.6, 25.2, and 23.3%, respectively). hPan2 contains RNase D motifs within its C-terminal 180-amino acid region, which is conserved among the RNase D family of 3Ј to 5Ј exoribonucleases ( Fig. 1, A and B), and this domain shows the highest degree of conservation among several species (compared with human, 71.7, 44.2, and 48% similarity in the order of D. melanogaster, C. elegans and S. cerevisiae, respectively). The invariant amino acid residues common to the family members are all conserved in hPan2 (Fig. 1B).
Identification of hPan2 as a Mg 2ϩ -dependent 3Ј to 5Ј Exoribonuclease-To examine whether hPan2 has nuclease activity, we measured the RNase activity of immunopurified hPan2 from transiently transfected COS-7 cells. The N-terminally FLAGtagged full-length hPan2 and its C-terminally deleted mutant, ⌬C170, whose exoribonuclease domain was expected to have been damaged ( Fig. 2A), were expressed in COS-7 cells, and the cell extracts were immunoprecipitated with an anti-FLAG antibody. The precipitated beads were washed with NaCl at high concentrations, and the precipitated proteins were eluted with FLAG peptide (Fig. 2B). Because a previous study showed that spermidine enabled yeast PAN nuclease to degrade poly(A) RNA even in the absence of Pab1 (23), RNase activity of hPan2 was first measured in the presence of the polyamine. As indexed by the release of 32 P-radioactivity from the internally 32 P-radiolabeled poly(A) RNA, the full-length hPan2 was co-purified with RNase activity (Fig. 2C). In sharp contrast, ⌬C170 had no significant RNase activity compared with the mock transfectant. To confirm that the observed RNase activity is derived from the immunopurified hPan2 rather than from a co-purifying unknown factor, we also prepared a hPan2 mutant in which a key conserved catalytic residue (Asp) in the RNase D domain had been replace by Ala (Fig. 2D). As shown in Fig. 2E, the single point mutation (D1083A) completely abolished the RNase activity of hPan2. We next analyzed ribonucleolytic products released by hPan2 by means of thin-layer chromatography analysis. Fig. 2F shows that 5Ј-AMP but not 3Ј-AMP was detected, consistent with the reported characteristics of members of the RNase D family. In addition, this family is also characterized by a requirement of Mg 2ϩ for its activity (29). As shown in Fig. 2G, RNase activity of hPan2 is completely inhibited in the presence of EDTA but not of EGTA, indicating that hPan2 is also a Mg 2ϩ -dependent RNase.
Next, we analyzed the reaction mode of hPan2 using 5Јterminally 32 P-radiolabeled poly(A) as a substrate. There was a progressive decrease in the length of the 5Ј-labeled poly(A) as the incubation time of the substrate with the full-length hPan2 was prolonged (Fig. 3A). However, such poly(A) shortening activity was not observed in ⌬C170. Thus, hPan2 was suggested to be a 3Ј to 5Ј exoribonuclease as assumed from the characteristics of the RNase D family. Generally, the actions of exoribonucleases are classified as either processive or distributive in terms of their catalytic mode (44). To determine which mode is responsible for the hPan2 action, we performed an experiment in which a constant amount of poly(A) was incubated with varying amount of hPan2. As shown in Fig. 3B, the poly(A) RNA was more rapidly shortened from the 3Ј-end with increasing amounts of hPan2, suggesting that hPan2 is a distributive type of exoribonuclease.
Under the present assay conditions, there was a non-degraded fraction (ϳ30%) in the 5Ј-labeled poly(A) substrate (see Fig. 3, A and B). This might be due to the 3Ј-end blocking of the RNA substrate. A previous study reported that some proportions of a commercially available poly(A) RNA are blocked at its 3Ј-end, which leads to resistance to a poly(A) nuclease (28). To test this possibility, the poly(A) substrate used in this study was incubated with yeast poly(A) polymerase and ATP under conditions for unspecific polyadenylation. As a result, ϳ35% of the poly(A) RNA was not elongated by the polyadenylation reaction (data not shown). Because the polymerase reaction requires 3Ј-OH group in the RNA substrate, this result indicates that a proportion of the poly(A) RNA used was blocked at its 3Ј-end. Thus, the resistance of the poly(A) substrate to hPAN is explainable by the 3Ј-end blocking of the RNA.
Several RNases also exhibit DNase activity (44). Thus, we examined whether hPan2 is capable of acting on poly(dA) DNA. The 5Ј-terminally 32 P-radiolabeled poly(dA) DNA was incubated with hPan2, and the reaction products were analyzed. However, poly(dA) DNA did not serve as the substrate of hPan2 (Fig. 3C).
Cloning of Human Pan3-Next, we searched and identified the partial cDNA fragment (XP170737) of hPan3 in the expressed sequence tag data base. To clone the full-length hPAN3 cDNA, both 5Ј-RACE and reverse transcription PCR were performed. The deduced hPan3 contains 688 amino acids and shows sequence similarities to putative Pan3 homologues from D. melanogaster, C. elegans, and S. cerevisiae (43.4, 39.3, and 22.4%, respectively) as shown in Fig. 4. hPan3 contains a C-terminal region in the segment of amino acids 552-676, which is highly conserved among several species (64.8, 52.1, and 41.6% similarity in order of D. melanogaster, C. elegans, and S. cerevisiae relative to the human homologue). Interestingly, motif analysis by RPS-BLAST revealed the presence of a kinase domain, which is conserved between human and fly (51% sequence similarity) but not in worm and yeast, although the functional significance of this domain is not yet clear at present.
Human PAN Complex Associates with PABP through Its hPan3 Subunit-To examine whether hPan2 and hPan3 form a complex similar to yeast PAN nuclease (25,26), FLAG-tagged hPan2 and HA-tagged hPan3 were co-expressed in COS-7 cells, and the cell extracts were subjected to immunoprecipitation using an anti-FLAG antibody. As shown in Fig. 5A, HA-hPan3 co-precipitated with FLAG-hPan2, and, at the same time, endogenous PABP was also detected in the precipitated fraction (lane 2). On the other hand, HA-hPan3 and PABP were not detected when COS-7 cells were mock transfected only with HA-hPan3 (lane 1). These results show that hPan2 and hPan3 form a complex, hPAN, and the hPAN complex also interacts with PABP. In addition, a comparative analysis of the deletion mutants of hPan2 revealed that FLAG-hPan2 mutants, which are capable of interacting with hPan3, also precipitate PABP (lanes 3-7). These results prompted us to hypothesize that hPan3 may function as a PABP-binding subunit for hPAN nuclease. To test this hypothesis, FLAG-tagged full-length hPan3 and its mutant lacking the N-terminal 290 amino acids (FLAG-hPan3⌬N) were co-expressed with HA-hPan2, and the cell extracts were subjected to immunoprecipitation using anti-FLAG antibodies. As shown in Fig. 5B, FLAG-hPan3 co-precipitated with HA-hPan2 as well as PABP; however, FLAG-hPan3⌬N could interact with hPan2 but not with PABP. These results suggest that hPan3 interacts with hPan2 and PABP through its C-and N-terminal regions, respectively. That is, hPan3 is the PABP-binding subunit of hPAN nuclease.
PABP Stimulates RNase Activity of hPAN2 in Its Associated Form with hPan3-To investigate whether the RNase activity of hPAN is stimulated by PABP, we prepared an hPAN complex consisting of hPan2 and hPan3. COS-7 cells were cotransfected with pFLAG-hPan2 and pHA-hPan3, and the cell extracts were subjected to immunoprecipitation with anti-FLAG IgG beads. To deplete the co-purified PABP, the precip- FIG. 2. hPan2 is a Mg 2؉ -dependent ribonuclease that produces nucleoside 5-monophosphate. A, the fulllength hPan2 and its ribonuclease-deficient mutant, ⌬C170, are illustrated. B, FLAG-tagged full-length hPan2 and its mutant, ⌬C170, purified from transfected COS-7 cells were silver-stained (left) or immunoblotted with anti-FLAG antibodies (right). As control, the same procedures were performed against COS-7 cells mock transfected with an empty vector. C, FLAG-hPan2 and FLAG-⌬C170 were incubated with internally 32 P-radiolabeled poly(A) RNA at 30°C for 30 min. Nuclease activity was measured by quantifying the release of 32 P-radioactivity, which is soluble in 25% trichloroacetic acid. D, FLAG-tagged full-length hPan2 and its mutant, D1083〈, purified from transfected COS-7 cells, were immunoblotted (IB) with anti-FLAG antibodies. E, nuclease activities of FLAG-hPan2 and FLAG-hPan2 D1083〈 were measured as described for panel C. F, the products of the RNase reaction were analyzed by TLC. The positions of the unlabeled 5Ј-AMP and 3Ј-AMP (Sigma) were determined using UV light. G, the effects of chelating agents, EGTA or EDTA, on nuclease activity of hPan2 were examined. In this experiment, 10 mM EGTA or EDTA was added to reactions in the presence of 2 mM MgCl 2 .
itated beads were washed with 200 mM NaCl. This washing markedly attenuated the interaction between hPan3 and PABP without affecting the interaction between hPan2 and hPan3 (Fig. 5C). Thus, Ͼ80% of PABP could be removed from the immunopurified preparation as compared with that prepared without high salt wash. As shown in Fig. 6A, hPan3 was copurified with hPan2 as a complex (hPAN complex). For comparison, we also used the nuclease-deficient ⌬C170 mutant of hPan2, which was also co-purified with HA-hPan3 (⌬C170 complex). The RNase activities of the purified proteins were examined in the presence or absence of spermidine. Fig. 6B shows that the purified hPAN complex (lane 3) and hPan2 (lane 2) could degrade poly(A) RNA in the presence of spermidine, whereas the ⌬C170 complex had no RNase activity (lane 4). However, in the absence of spermidine, none of the proteins exhibited any detectable RNase activity. Under the same assay conditions, the effect of PABP on RNase activity was investigated. As shown in Fig. 6C, the recombinant GST-fused PABP prepared from E. coli could markedly stimulate the RNase activity of the hPAN complex. In sharp contrast, GST-PABP could not stimulate the RNase activity of the hPan2 subunit or the ⌬C170 complex. These results indicate that PABP stimulates hPan2 nuclease activity through its association with hPan3, consistent with the above finding that hPan3 is the PABP-binding subunit.
PABP Enhances Substrate Specificity of hPAN Nuclease to Poly(A) RNA-The substrate specificity of the hPAN nuclease complex was investigated by measuring the extent of inhibition of the catalytic activity in the presence of increasing unlabeled polyribonucleotides, poly(A) and poly(C), in a reaction containing a constant level of radiolabeled poly(A) as described previously (43). As shown in Fig. 7A, both poly(A) and poly(C) inhibited the spermidine-stimulated RNase activity of hPAN. Poly(A) showed stronger inhibition than poly(C); the apparent K i values were ϳ0.07 and Ͼ2 g/ml for poly(A) and poly(C), respectively. On the other hand, the PABP-stimulated RNase activity of hPAN was inhibited strongly by poly(A), although only slightly by poly(C) (Fig. 7B). Thus, hPAN appears to be intrinsically specific for poly(A) RNA, and its substrate specificity for poly(A) is enhanced by its interaction with PABP. Because the RNA binding by PABP is highly specific for poly(A) RNA, of which the K d for poly(A) is two to three orders of magnitude lower than those for other polynucleotides (45), the binding specificity of PABP might contribute to the substrate specificity of hPAN.
Cytoplasmic Localization of hPAN Nuclease-The physical interaction of hPAN with cytoplasmic PABP suggests that the poly(A)-specific ribonuclease has a role in cytoplasmic mRNA degradation. Consistent with this view, the hPAN nuclease complex appeared to be present in the cytoplasm. HeLa cells expressing FLAG-hPan2 or FLAG-hPan3 were immunostained using anti-FLAG antibodies with concomitant staining of nuclei using Pico Green. As shown in Fig. 8A, FLAG-hPan2 and FLAG-hPan3 were detected exclusively in the cytoplasm but not in the nucleus. When HA-Pan2 and FLAG-Pan3 were coexpressed in HeLa cells, they exhibited cytoplasmic co-localization (Fig. 8B), which is consistent with the physical interaction between Pan2 and Pan3 (Fig. 5B).

DISCUSSION
Identification of Human PAN Nuclease-In this study, we have identified novel human cDNAs encoding an hPAN nuclease complex and analyzed its biochemical properties in terms of its regulation by PABP and substrate specificity. The catalytic subunit of human PAN nuclease, hPan2, is a Mg 2ϩ -dependent 3Ј to 5Ј exoribonuclease that functions in a distributive manner. On the other hand, hPan3 is a regulatory subunit that interacts with both hPan2 and PABP. Furthermore, the deadenylating activity of hPAN nuclease is stimulated by PABP; this stimulation requires hPan3, which is consistent with the finding that hPan2 interacts with PABP through hPan3. These characteristics are clearly distinct from those of another deadenylase, PARN, which has been reported to be a major mRNA deadenylase in mammalian cells. Eukaryotic mRNAs have two features, a 5Ј-cap and a 3Ј-poly(A) tail. eIF4E binds to the 5Ј-cap and PABP binds to the poly(A) tail during translation. eIF4E competes with PARN on 5Ј-cap and inhibits the deadenylating activity of PARN (46), because this activity is stimulated by 5Ј-cap. Moreover, PABP was also reported to inhibit the deadenylating activity of PARN under physiological conditions (28,31). Therefore, the dissociation of eIF4E and PABP from the 5Ј-cap and 3Ј-poly(A) tail, respectively, may be a necessary step for deadenylation by PARN. The biological regulation of PARN activation is an important issue. Because human PAN nuclease preferentially degrades poly(A) RNA in the presence of PABP and does not absolutely require the 5Ј-cap for its deadenylating activity, it could degrade the poly(A) tail of mRNAs even during translation.
In yeast, the Pop2/Ccr4 complex is identified as the major mRNA deadenylase (27). Pop2 encodes a member of the RNase D family of 3Ј to 5Ј exonucleases (43). Yeast Pop2 has recently been shown to exhibit deadenylating activity in vitro (43), but other groups reported that Pop2 is not required for the deadenylase activity (36,47). On the other hand, Ccr4, which is FIG. 3. hPan2 is a 3 to 5 exoribonuclease that functions in a distributive manner. A, the FLAG-tagged hPan2 and ⌬C170 were incubated with 32 P-5Ј-radiolabeled poly(A) RNA at 30°C for the indicated times. Products of the RNase reaction were separated by 7 M urea, 12% polyacrylamide gel, and the radioactive molecules were visualized by autoradiography. The positions of denatured 100-bp double-stranded DNA markers (Invitrogen) are indicated on the right. B, a constant amount of 32 P-5Ј-radiolabeled poly(A) RNA was incubated with varying hPan2 concentration at 30°C for 40 min. The amount of hPan2 is indicated as the fold increase for that used for panel A. C, 32 P-5Јradiolabeled poly(A) RNA or poly(dA) DNA was incubated with FLAG-hPan2, and the reaction products were analyzed as described for panel A. structurally homologous to apurinic endonucleases, was also shown to be the catalytic component of the 3Ј to 5Ј deadenylase in yeast and humans (36,47). Similarly to hPan2, Ccr4 functions in a distributive manner with strong preferences for poly(A) and shows Mg 2ϩ dependence (48). On the other hand, it contains a unique non-poly(A)-specific binding site and becomes processive with longer RNA substrates (48). In sharp contrast to that of hPAN nuclease, the deadenylating activity of Pop2/Ccr4 is inhibited by PABP (47).
Recently, the fifth mRNA deadenylase has been reported (30). Nocturnin is a homologue of Ccr4 and specifically degrades the 3Ј-poly(A) tail in a processive manner. In contrast to other deadenylases, nocturnin is located exclusively in the rods and cones of photoreceptor cells in Xenopus laevis and is specifically expressed in the early night (35). Therefore, it has been suggested that nocturnin functions in the regulation of circadian rhythm. However, in mammalian cells, it may also be involved in the general mechanism of mRNA decay, because a nocturnin homologue has been found to be expressed widely in mouse tissues, including the liver, kidney, brain, lung, and heart in addition to the retina (49,50). The requirement of the cap and PABP has not yet been elucidated. In combination, mammalian PAN is a unique deadenylating nuclease in the sense that it is specifically activated by PABP. FIG. 5. The interaction of heterodimeric human PAN nuclease with PABP is mediated through the hPan3 subunit. A, the FLAG-tagged hPan2 or its deletion mutants were expressed in COS-7 cells together with HA-tagged hPan3, and the cell extracts were subjected to immunoprecipitation (IP) using an anti-FLAG antibody. Immunoblot (IB) analyses with anti-FLAG (lower), anti-HA (middle), and anti-PABP (upper) antibodies were performed. As control, mock-transfected COS-7 cells with an empty vector were used. B, the FLAG-tagged hPan3 and its mutant lacking the N-terminal 290 amino acids, hPan3⌬N, were expressed in COS-7 cells together with the HA-tagged hPan2, and a coimmunoprecipitation assay was performed as described for panel A. C, FLAG-hPan2 and HA-hPan3 were co-expressed in COS-7 cells, and a co-immunoprecipitation assay was carried out as described for panel A. The immunoprecipitated beads were washed with NaCl at the indicated concentrations.

Possible Roles of Mammalian PAN in Poly(A)
Metabolism-In yeast, the primary function of PAN was suggested to be the nuclear trimming of poly(A) tails to message-specific lengths (26). However, it was also reported that PAN contributes to cytoplasmic mRNA turnover as an alternative mRNA deadenylase to the Ccr4/Pop2 complex (27). In both cases, Pab1 plays pivotal roles. However, in mammals, two types of poly(A)binding protein, PABP and PABPN1, function in the cytoplasm and nucleus, respectively, and PABPN1 is considered to be required for mammalian poly(A) tail-length control in the nucleus (51,52). We present here the data indicating that both hPan2 and hPan3 are localized exclusively in the cytoplasm and that the hPan2/hPan3 complex interacts with PABP. Therefore, these results support the latter function suggested in yeast, that is, mammalian PAN functions in cytoplasmic mRNA deadenylation. However, our data cannot totally ex-clude the possibility that hPAN also interacts with PABPN1 and functions in the nuclear trimming of poly(A) tails.
A previous study from Shyu and co-workers using c-fos mRNA indicated that mRNA undergoes synchronous poly(A) shortening (53). Thus, at least in the representative example, distributive ribonucleolytic digestion of poly(A) tails is implied in the mRNA decay in mammalian cells, and Shyu and coworkers predicted that the mammalian counterpart of the yeast PAN might be a distributive enzyme and may be involved in the mRNA deadenylation reactions (53). Our observation that the hPan2 functions in a distributive manner is consistent with this notion and further strengthens the possibility that the hPan2 is involved in the cytoplasmic mRNA deadenylation.
Implication of a Phosphorylation-dependent Regulation in Deadenylation-In this study, we have identified a putative kinase domain in the segment amino acids 302-549 of hPan3.
FIG. 6. PABP stimulates RNase activity of human PAN nuclease in a hPan3-dependent manner. A, the FLAG-tagged hPan2 and ⌬C170 were expressed in COS-7 cells together with the HA-tagged hPan3, and the cell extracts were subjected to immunoprecipitation (IP) using anti-FLAG IgG agarose. Proteins were eluted with the FLAGpeptide from the immunoprecipitated beads. The immunopurified proteins were immunoblotted (IB) with anti-FLAG (lower) and anti-HA (upper) antibodies. B, the purified proteins (as shown in panel A) were incubated with internally 32 P-radiolabeled poly(A) RNA at 30°C for 30 min in the presence or absence of 1 mM spermidine, and 32 P radioactivity release was measured. C, the purified proteins were incubated with the 32 P-radiolabeled poly(A) RNA at 30°C for 30 min with or without 10 ng of GST-fused PABP in the absence of spermidine. FIG. 8. Cytoplasmic co-localization of the two subunits of human PAN nuclease. A, HeLa cells expressing FLAG-hPan2 (red) or FLAG-hPan3 (red) were stained with anti-FLAG antibodies. The immunoreactions with the primary antibodies were visualized by staining the secondary antibody, Alexa Fluor 568-conjugated anti-mouse IgG. The cells were also stained with Pico Green (green) as a marker of nuclei. B, HeLa cells expressing HA-hPan2 (red) and FLAG-hPan3 (green) were stained with the anti-HA antibody and anti-FLAG (3F10) antibodies, respectively. As the secondary antibodies, Alexa Fluor 568conjugated anti-mouse IgG and Alexa Fluor 488-conjugated anti-rat IgG were used. The merged images of the two signals are displayed in yellow.
The functional significance of this domain in the regulation of hPAN nuclease is not yet clear at present. However, Hammet et al. have shown recently that yeast Pan3 interacts with Dun1 kinase, which has complex checkpoint functions, including the DNA damage-dependent cell cycle arrest in G 2 /M, transcriptional induction of repair genes and the regulation of postreplicative DNA repair pathways (54). The interaction between Pan3 and the Dun1 kinase is necessary in the posttranscriptional regulation of the RAD5 DNA repair gene, possibly in the regulation of the poly(A)-tail length of mRNA. As indicated in Fig. 4A, the sequence analysis of Pan3 homologues from yeasts to humans revealed that the putative kinase domain is conserved in D. melanogaster and humans but not in yeast and C. elegans. Therefore, it is intriguing to determine whether the built-in kinase domain of hPan3 is functional and behaves similar to Dun1 kinase.