A 54-kDa Fragment of the Poly(A)-specific Ribonuclease Is an Oligomeric, Processive, and Cap-interacting Poly(A)-specific 3′ Exonuclease*

We have previously identified a HeLa cell 3′ exonuclease specific for degrading poly(A) tails of mRNAs. Here we report on the purification and identification of a calf thymus 54-kDa polypeptide associated with a similar 3′ exonuclease activity. The 54-kDa polypeptide was shown to be a fragment of the poly(A)-specific ribonuclease 74-kDa polypeptide. The native molecular mass of the nuclease activity was estimated to be 180–220 kDa. Protein/protein cross-linking revealed an oligomeric structure, most likely consisting of three subunits. The purified nuclease activity released 5′-AMP as the reaction product and degraded poly(A) in a highly processive fashion. The activity required monovalent cations and was dependent on divalent metal ions. The RNA substrate requirement was investigated, and it was found that the nuclease was highly poly(A)-specific and that only 3′ end-located poly(A) was degraded by the activity. RNA substrates capped with m7G(5′)ppp(5′)G were more efficiently degraded than noncapped RNA substrates. Addition of free m7G(5′)ppp(5′)G cap analogue inhibited poly(A) degradationin vitro, suggesting a functional link between the RNA 5′ end cap structure and poly(A) degradation at the 3′ end of the RNA.

In recent years it has become clear that poly(A) removal is an important step during mRNA decay, and it has been found that mRNA degradation in many cases is initiated by degrading the mRNA poly(A) tail (reviewed in Ref. 1). In yeast two major deadenylation-dependent pathways of mRNA degradation have been defined, the deadenylation-dependent decapping pathway and the 3Ј-5Ј decay pathway (for reviews see Refs. 2 and 3). The former pathway is initiated by poly(A) tail removal followed by decapping and 5Ј-3Ј exonucleolytic degradation of the decapped mRNA by the Xrn1p 5Ј exonuclease. The latter is initiated by deadenylation followed by 3Ј-5Ј exoribonucleolytic degradation of the mRNA. The participating 3Ј-5Ј exonucleases have not yet been unambiguously identified. A multicomponent complex, termed the exosome (4), has been identified in yeast as well as in human cells, and it contains at least 10 different 3Ј-5Ј exoribonucleases (reviewed in Ref. 5). The exosome is involved in 3Ј end processing of several different stable RNAs (4,6) and 3Ј-5Ј degradation of mRNA (7). Thus, these exoribonucleases could be responsible for the degradation of mRNA including the poly(A). However, it remains to be established whether any of these exonucleases preferentially or exclusively degrade poly(A) of mRNA.
In mammalian cells mRNA decay has been studied extensively, and it has been found that gene expression to a large extent is regulated by changing the stability of an mRNA. From these studies it has been shown that cis-acting elements located in the mRNA body are important elements involved in regulating mRNA decay. One of the first elements to be identified was an AU-rich element located in the 3Ј-untranslated region of the human lymphokine granulocyte macrophage colony-stimulating factor mRNA (8). Shaw and Kamen (8) showed that the AU-rich element destabilized an mRNA and that the element could confer the destabilization effect if it was transferred to another mRNA. Furthermore, it has been found, equivalent to the case in yeast, that mRNA decay in mammalian cells in many cases is initiated by poly(A) removal (9). However, most of these studies (1) have been performed in vivo, and therefore, it has been difficult to identify participating components and to address mechanistic details. In the last years several in vitro cell-free mRNA stability systems have been developed, and these systems will make it possible to elucidate mechanistic details and to identify factors regulating mRNA decay in mammalian cells (10 -14). Using such a system Wilusz and colleagues (13) have been able to stimulate poly(A) removal and RNA degradation by the presence of an AU-rich sequence element (8) in the RNA body and to reproduce regulated aspects of mRNA decay. Similarly, Brewer (11) has shown that the 3Ј-untranslated region of c-Myc affects poly(A) removal and RNA degradation in a different in vitro system. The poly(A)-specific exonucleases responsible for degrading the mRNA poly(A) tails in these in vitro decay systems remains to be identified.
Poly(A) tail removal and poly(A) degrading nuclease activities have been studied in several eukaryotic systems (for reviews see Refs. 1, 3, and 15). In mammalian cells several different poly(A) degrading activities have been characterized over the years. Earlier studies (reviewed in Ref. 16) were hampered by the lack of RNA substrates resembling polyadenylated mRNA, making it difficult to assay, purify, and characterize nucleases that specifically degraded only the poly(A) tail of an mRNA. By the use of an in vitro transcribed and polyadenylated mRNA mimic we were able to define a poly(A)-specific 3Ј exonuclease in HeLa cell free extracts (17) and propose a reac-tion pathway for mRNA poly(A) tail removal (18). Important properties of the HeLa cell activity were its high selectivity for degrading only 3Ј-located poly(A) tails, its requirement for a 3Ј-located hydroxyl group, and release of 5Ј-AMP as the mononucleotide product. Recently, a poly(A)-specific ribonuclease (PARN) 1 associated with a 74-kDa polypeptide was purified and characterized in calf thymus extracts by Körner and Wahle (19). Human PARN has been molecularly cloned, and PARN activity has been recovered from recombinant 74-kDa polypeptide (20). PARN has also been found in Xenopus and shown to play a role in poly(A) tail removal during meiotic maturation of Xenopus oocytes (20,21). A polyribosome-associated 3Ј exoribonuclease, having a molecular mass of 33 kDa has been described and purified by Caruccio and Ross (22). However, this 3Ј exoribonuclease is not specific for poly(A), although poly(A) can be degraded. In addition to these metazoan poly(A) degrading nucleases Sachs and colleagues have described a poly(A)binding protein-dependent poly(A)-specific nuclease in yeast (23)(24)(25). Poly(A)-specific nuclease is a 3Ј-5Ј exoribonuclease and is composed of at least two polypeptides, Pan2p and Pan3p (26,27). Pan2p contains a conserved RNase D 3Ј exonuclease domain (28), suggesting that it harbors the catalytic activity.
In this paper we report on the purification of an activity present in calf thymus extracts having the same properties as the poly(A) tail-specific 3Ј exonuclease activity we previously described in HeLa cell free extracts (17,18). The calf thymus activity has been purified to apparent homogeneity and shown to copurify with a single 54-kDa polypeptide, which we identify as a fragment of the PARN 74-kDa polypeptide. We show that the purified nuclease most likely is oligomeric, acts in a highly processive fashion, and functionally interacts with the m 7 G(5Ј)ppp(5Ј)G cap structure at the 5Ј end of the RNA substrate.

EXPERIMENTAL PROCEDURES
Preparation of Cell-free Extracts-All steps were performed at 4°C. Crude whole cell extract was prepared from bovine calf thymus, essentially according to Wahle (29). In short, frozen calf thymus (2-15 kg) obtained from a local slaughter house was thawed on ice, cut into pieces, and homogenized in an approximately equal volume of buffer 1 (50 mM Tris-HCl, 10 mM K 3 PO 4, 1 mM EDTA, 10% glycerol, 50 mM KCl, 0.1 mM dithiothreitol at pH 7.9) using a Waring blender (50 s at low speed and 50 s at high speed). Solid material was precipitated by centrifugation in a Sorvall GSA rotor at 16,000 ϫ g in 4°C for 60 min. The resulting crude extract was filtered through a test sieve mesh 7-normal (Pascal Eng. Co. Ltd.), 0.134 g/ml NH 4 (SO 4 ) 2 (25% saturation) was added and stirred on ice for 2 h, and the precipitate was removed by centrifugation in a Sorvall GSA rotor at 16,000 ϫ g in 4°C for 60 min. The supernatant was further supplemented with an additional 0.115 g/ml NH 4 (SO 4 ) 2 (45% saturation) and then treated as for the previous precipitation step. The pellet collected after centrifugation was dissolved in 2-4 volumes buffer D (20 mM HEPES-KOH, 100 mM KCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 20% glycerol at pH 8.2) and dialyzed for 10 h at 4°C using dialysis tubing with a molecular mass cut-off at 6000 -8000 (Spectra/Por 1 number 132650). After dialysis the 45% ammonium sulfate fraction was frozen in liquid nitrogen and stored at Ϫ70°C. Starting with 2.8 kg of calf thymus the obtained extract (ϳ1038 ml) contained approximately 103 g of protein. The subsequent ammonium sulfate fraction contained 13.7 g of protein in 400 ml. Protein concentration was determined using the Bio-Rad protein assay kit (number 500-0001) and bovine ␥ globulin as reference.
Partial Purification of Poly(A)-specific 3Ј Exonuclease Activity-All steps were performed at 4°C. A typical protocol starting with 400 ml of crude ammonium sulfate fraction (13.7 g of protein) is given below and summarized in Table I. The ammonium sulfate fraction was added to DEAE-Sepharose CL-6B (Amersham Pharmacia Biotech, 17-0710-01) ion exchange medium (350 ml of packed matrix) equilibrated with buffer D (see above). Protein was allowed to bind to the matrix under slow stirring for 30 min at 4°C. Unbound material was removed by washing the matrix with buffer D three times. The suspended matrix was recovered by centrifugation (Sorvall H4000 rotor at 800 rpm for 3 min) after each washing step. The washed matrix was then packed in a column (diameter, 70 mm) and washed (2 column volumes) with buffer D at a flow rate of 12 cm/h. The column was eluted by two salt steps (buffer D supplemented with 0.17 M and 1.0 M KCl), and eluted protein from each step was collected. The nuclease activity, which eluted in the 0.17 M KCl step fraction (1.67 g of protein, 440 ml), was dialyzed against buffer D for 10 h and subsequently fractionated by heparin-Sepharose CL-6B (Amersham Pharmacia Biotech, 17-0467-01) chromatography. A column with bed volume of 130 ml and a diameter of 50 mm was equilibrated with buffer D at flow rate 9 cm/h. The dialyzed DEAE-Sepharose 0.17 M KCl fraction was applied to the heparin-Sepharose CL-6B column at a flow rate of 9 cm/h. The flow through protein fraction was collected, and then the matrix was washed with 350 ml of buffer D. The nuclease activity was subsequently eluted by buffer D supplemented with 0.5 M KCl. The eluate (441 mg of protein, 96 ml) was dialyzed against buffer D for 10 h at 4°C and subsequently fractionated by fast protein liquid chromatography Mono Q HR16/10 (Amersham Pharmacia Biotech, 17-0506-01) chromatography. Protein was loaded (ϳ48 ml of 0.5 M KCl heparin-Sepharose fraction/turn) onto the buffer D equilibrated Mono Q column using a 50-ml superloop (Amersham Pharmacia Biotech, 19-7850-01) at flow rate 2 ml/min. After washing with buffer D, the column was developed with a two-step linear gradient (0 -25% in 300 ml and 25-55% in 100 ml of buffer D, supplemented with 1.0 M KCl), 9-ml fractions were collected, and active fractions (eluting at ϳ10% 1.0 M KCl) were identified, pooled (81.8 mg of protein, 115 ml), and dialyzed against buffer D for 10 h at 4°C. A 36-ml packed Blue-Sepharose CL-6B (Amersham Pharmacia Biotech, 17-0830-01) column, 26 mm in diameter, was prepared according to the instructions of the manufacturer. The column was equilibrated with buffer D at a flow rate of 34 cm/h. The Mono Q fraction (115 ml) was loaded onto the column followed by a three-step wash procedure of the matrix (100 ml of buffer D, 100 ml of buffer D supplemented with 0.17 M KCl, and 100 ml of buffer D supplemented with 0.3 M KCl). The column was developed with a linear gradient between 0.3 and 1.0 M KCl (4.5 column volumes). Active fractions were identified (eluting at ϳ0.45 M KCl), pooled (7.88 mg of protein, 105 ml), and dialyzed against buffer D for 10 h at 4°C. The dialyzed fraction was frozen in liquid nitrogen and stored at Ϫ70°C.
Poly(A)-Sepharose Affinity Chromatography-All steps were performed at 4°C. A standard protocol is given below. The poly(A)-Sepharose 4B matrix (Amersham Pharmacia Biotech, 17-0860-01) was prepared according to the manufacturer. A HR10/10 column with bed volume of 8 ml was equilibrated with buffer D containing 25 mM KCl at pH 7. 12 ml of the Blue-Sepharose CL-6B fraction was dialyzed for 4 h against 2 ϫ 2 liters of buffer D containing 25 mM KCl at pH 7. The dialyzed fraction was applied to the column at flow rate of 1 ml/min. The column was first washed with 5 bed volumes of buffer D containing 25 mM KCl at pH 7 followed by a second wash with 5 bed volumes of buffer D containing 200 mM KCl at pH 6 and a third wash with 5 bed volumes of buffer D containing 280 mM KCl at pH 6. Subsequently, the column was developed with a gradient (5 bed volumes) from 280 to 600 mM KCl at pH 6. Poly(A)-specific exonuclease activity eluted between 300 and 550 mM KCl. This procedure purifies the specific activity of the poly(A)specific exonuclease approximately 14-fold.
5Ј-AMP-Sepharose and 7-Methyl-GTP-Sepharose Affinity Chromatography-All steps were performed at 4°C. A standard protocol is given below. The 5Ј-AMP-Sepharose 4B matrix (Amersham Pharmacia Biotech, 17-0620-01) was prepared according to the manufacturer. A HR 5/10 column (Amersham Pharmacia Biotech, 18-0388-01) with bed volume of 2 ml was prepared and equilibrated with buffer D containing 50 mM KCl at pH 7. The Blue-Sepharose CL-6B fraction (27.5 ml) was dialyzed against buffer D supplemented with 50 mM KCl at pH 7. The dialyzed fraction was loaded on the column at flow rate 0.2 ml/min. The protein containing flow through fraction was collected. The matrix was washed with 16 ml of buffer D supplemented with 60 mM KCl followed by a 20-ml wash with buffer D supplemented with 0.3 M KCl. Subsequently the column was developed with buffer D using a linear gradient between 0.3-2.0 M KCl (7.5 column volumes). Active fractions (eluting between 0.7-2.0 M KCl) were identified, pooled (15 ml), and dialyzed against buffer D. The matrix 7-methyl-GTP-Sepharose 4B (Amersham Pharmacia Biotech, 27-5025-01) was prepared according to the manufacturer. A HR 5/5 column (Amersham Pharmacia Biotech, 18-0383-01) with bed volume of 1 ml was prepared and equilibrated with buffer D containing 50 mM KCl at pH 7. The dialyzed 5Ј-AMP-Sepharose fraction was loaded at flow rate 0.2 ml/min. The protein-containing flow through fraction was collected, and subsequently the matrix was washed (10 column volumes) with buffer D at pH 7 supplemented with 50 mM KCl. The bound protein fraction was eluted with a salt step (10 column volume of buffer D supplemented with 2 M KCl). The nuclease activity was identified (eluted during the 2 M KCl step) and pooled (ϳ0.750 g of protein as determined by SDS-PAGE analysis and silver staining, 5 ml). To optimize recovery of nuclease activity 0.1 mg/ml of methylated bovine serum albumin (BSA) should be added during dialysis to the pooled fraction of the 5Ј-AMP-Sepharose chromatographic step and to the final 7-methyl-GTP fraction.
Analytical SMART Superdex 200 Gel Fitration-The poly(A)-Sepharose fraction (0.7 ml, 0.06 mg/ml protein) was first concentrated by SMART MonoQ PC 1.6/5 (Amersham Pharmacia Biotech, 17-0671-01) chromatography by the following procedure. The column was equilibrated by buffer D at pH 7 containing 50 mM KCl. The poly(A)-Sepharose fraction was dialyzed against buffer D containing 50 mM KCl at pH 7 and applied to the MonoQ column at flow rate 50 l/min. Bound material was eluted by increasing the KCl to 500 mM, and fraction size was 25 l. Active fractions (total volume, 100 l) were identified and pooled. Subsequently, 50 l of the concentrated poly(A)-sepharose fraction was fractionated by gel filtration using a SMART Superdex 200 PC 3.2/30 column equilibrated with buffer D at pH 7 containing 100 mM KCl. Flow rate was 40 l/min. Active fractions were identified by in vitro deadenylation. Molecular size markers were fractionated by the same procedure on the Superdex 200 column. Molecular size markers were ferritin, catalase, aldolase, and BSA having molecular masses 440, 232, 158, and 67 kDa, respectively. The 7-methyl-GTP-Sepharose fraction (50 l) supplemented with methylated BSA (0.1 mg/ml) was loaded directly onto a SMART Superdex 200 PC 3.2/30 column equilibrated in buffer D supplemented with methylated BSA (0.1 mg/ml).
Preparation of RNA Substrates-RNA substrates L3(A 30 ), L3(A 30 )X 15, L3(A 30 )X 49, L3(A 30 )X 164 capped at the 5Ј end were synthesized by in vitro transcription using T3 RNA polymerases (Promega number P208C) and plasmid pT3L3(A 30 ) (17), digested with NsiI, HincII, EcoRI, and PvuII, respectively, as DNA template. RNA substrates ML54(U 30 ) and ML40(C 32 ) were prepared as described previously (17). RNA substrates ML43(G 14 ) and ML40(C 16 ) were prepared as described previously (17) using plasmid pT3ML43(G 16 ), which is identical to the previously described plasmid pT3ML43(G 30 ) with the exception of the number of inserted G residues. RNA substrates were labeled either in their bodies or in their homopolymeric tails by inclusion of radioactively labeled mononucleotides during in vitro transcription as previously outlined (17,18). The specific radioactivities of the included radioactive mononucleotides were 40 Ci/mmol in the transcription mixture for body labeling or 5 Ci/mmol in the transcription mixture for tail labeling. Transcribed RNA was purified according to Moore and Sharp (30).
SDS-Polyacrylamide Gel Electrophoresis and Renaturation of Nuclease Activity-SDS-polyacrylamide (acrylamide:bisacrylamide 30:08) gels (5 and 7.5% acrylamide in spacer and separation gels, respectively) were prepared according to Laemmli (31) using a Mini-Protean II gel apparatus (Bio-Rad, 125BR). The indicated amount of protein was fractionated directly on the gel or precipitated before loading by the addition of 1 volume of 20% trichloroacetic acid followed by centrifugation at 13,000 ϫ g for 30 min at 4°C. The obtained pellet was dried, washed with acetone, collected again by centrifugation for 10 min at 13,000 ϫ g, and finally dissolved in 10 l of sample buffer (50 mM Tris-HCl, pH 6.8, 1% (w/v) SDS, 100 mM dithiothreitol, 8% (v/v) glycerol, 0.025% (w/v) bromphenol blue), and separated by gel electrophoresis. The resulting gel was fixed and stained as indicated. Silver staining was according to Oakley et al. (32). Coomassie Brilliant Blue staining was according to standard protocol (33). Elution of proteins from SDSpolyacrylamide gel and renaturation of nuclease activity was done as described by Hager and Burgess (34).
In-gel Digestion and Mass Spectrometry Analysis-The excised band was prepared for and subjected to in-gel digestion as described (35). In brief, after washing with ammonium bicarbonate and acetonitrile, the gel piece was completely dried, and a solution containing modified porcine trypsin, sequence grade (Promega Corp. Madison, WI) was allowed to soak into the gel piece. After overnight incubation at 30°C, generated peptides were recovered by extraction. The peptide mixture was analyzed by matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry, using a Bruker Biflex III instrument (Bremen, Germany), equipped with delayed extraction and reflector. The sample was prepared by the dried droplet technique, using a-cyano-4-hydroxycinnamic acid as matrix. The instrument was externally calibrated using angiotensin II (MHϩ 1046.54) and adrenocorticotropic hormone fragment 18-39 (MHϩ 2465.20). The peptide mass fingerprinting analysis was done using ProFound (version 4.7.5).
Molecular Cloning and Expression-The 74-kDa PARN polypeptide and a 54-kDa fragment of PARN were molecularly cloned by a standard reverse transcription-polymerase chain reaction procedure followed by subcloning into the pGem T vector (Promega Inc.). Inserts were subcloned into the pET-19 vector (Novagen Inc.) between the NdeI and BamHI sites. The following primer pairs were used 5Ј-TCGCATATGG-AGATAATCAGGAGCA-3Ј and 5Ј-TCAGATCTTTACCATGTGTCAGG-AACTTCA-3Ј for the 74-kDa PARN polypeptide and 5Ј-TCGCATATG-GAGATAATCAGGAGCA-3Ј and CGTGGATCCTCAGTTACCAAAGG-CACTGAA for the 54-kDa fragment. RNA template was obtained from HeLa cells. Recombinant and His-tagged polypeptides were expressed in the Escherichia coli strain BL21(DE3) and extracted according to the manufacturer of the pET vectors. Extracted recombinant polypeptides were purified by metal affinity chromatography using the TALON matrix (CLONTECH Inc. number 8901) as outlined by the manufacturer of the matrix.
Protein/Protein Cross-linking-Bis(sulfosuccinimidyl) suberate and dimethyl pimelimidate were purchased from Pierce, and cross-linking was performed according to the instructions of the manufacturer. Essentially, indicated amounts of purified recombinant polypeptides, in buffer D at pH 8.2, were mixed with 10 mM cross-linking reagents and incubated for 60 min at room temperature. Total volume was 100 l. The reactions were terminated by the addition of 11 l of 1 M Tris-HCl, pH 7.9.
One-dimensional TLC and Quantitation of Nuclease Activity-Deadenylation activity was quantified as follows: L3(A 30 ) RNA substrate labeled by inclusion of [␣-32 P]ATP during in vitro transcription was incubated in conditions for in vitro deadenylation as described above with the addition of 0.1 mg/ml BSA. Reactions were analyzed by onedimensional TLC using 0.75 M KH 2 PO 4 , pH 3.5 (H 3 PO 4 ), as the solvent. The resulting polyethyleneimine-cellulose F plate was dried and scanned by a 400 S PhosphorImager (Molecular Dynamics). The fraction of released [ 32 P]AMP was determined. Knowing the specific activity of [ 32 P]AMP in the RNA substrate, the amount of released AMP was calculated. One unit of deadenylation activity was defined as the release of 1 mol of AMP/min.

RESULTS
Partial Purification-To purify a poly(A)-specific 3Ј exonuclease activity, we used a dual assay strategy that monitored both the disappearance of the RNA substrate and the appearance of the two reaction products: deadenylated RNA and released AMP. Disappearance of the RNA substrate and accumulation of the deadenylated product were investigated by in vitro deadenylation using body labeled L3(A 30 ) RNA as the substrate, followed by analytical polyacrylamide gel electrophoresis of the reacted RNA. Release of AMP was investigated by in vitro deadenylation of L3(A 30 ) RNA substrate radioactively labeled in its poly(A) tail followed by detection of released mononucleotides by TLC. An important advantage of this dual assay strategy is that nucleases that degrade both the poly(A) tail and the RNA body of the substrate can be excluded.
The poly(A)-specific 3Ј exonuclease activity was first partially purified from calf thymus whole cell extract using ammonium sulfate precipitation followed by four chromatographic steps (see "Experimental Procedures," Table I, and Fig. 1). Using this purification protocol, a poly(A) removing activity fulfilling the requirements of the dual assay strategy was purified approximately 14,000-fold (Table I).
The partially purified Blue-Sepharose fraction was further purified by poly(A)-Sepharose affinity chromatography. This affinity step improved the purity of the poly(A)-specific exonuclease activity approximately 14-fold. The protein profile of the obtained poly(A)-Sepharose fraction, as detected by SDS-PAGE, revealed several polypeptides ( Fig. 2A). To determine whether any of those was associated with the activity, we eluted them from the SDS-PAGE gel matrix and subsequently denatured and renatured the eluted polypeptides. Exonuclease activity specific for degrading poly(A) and releasing AMP was recovered from polypeptides in the 50 -60-kDa range ( Fig. 2B and data not shown), suggesting that a polypeptide with this size was responsible for the poly(A)-specific 3Ј exonuclease activity.
A 54-kDa Polypeptide Is Responsible for the Activity-The Blue-Sepharose fraction was further purified by two consecutive steps of affinity chromatography, 5Ј-AMP-Sepharose followed by 7-methyl-GTP-Sepharose. SDS-PAGE of the final fraction revealed a prominent 54-kDa polypeptide (Fig. 3). Table I summarizes the quantitation of the purification. The two million-fold purification is probably an overestimation because of the presence of inhibitors of the activity in crude fractions. The increase in total activity during purification is in agreement with the presence of nuclease inhibitors. The nature of these nuclease inhibitors is at present unknown. Some of them could very well be abundant RNA binding proteins because we (data not shown) and others (19,21) have found that addition of an increasing amount of poly(A)-binding protein eventually will inhibit in vitro deadenylation.
The 54-kDa polypeptide was in-gel digested with trypsin and subjected to mass spectrometry analysis using MALDI-TOF. The mass spectrum revealed ten peptides that were used to scan the NCBI nonredundant data base, using the application ProFound. Six of the ten peptides were located in human PARN, and they represented the following tryptic peptides of PARN, amino acids 87-99, 236 -243, 260 -272, 350 -359, 414 -421, and 444 -454 (amino acids numbered according to Ref. 20). The existence of the remaining four peptides is most likely explained by differences between the human and bovine PARN amino acid sequences. We conclude that the poly(A)-specific 3Ј exonuclease activity that we purified was associated with a 54-kDa fragment of the 74-kDa PARN polypeptide.
Oligomeric Structure-The 7-methyl-GTP-Sepharose fraction was analyzed by gel filtration using a SMART Superdex 200 column as the matrix. Active fractions were identified (Fig. 4A). The native molecular mass of the 54-kDa active fragment of PARN was estimated to 180 -220 kDa. The poly(A)-Sepharose fraction was also analyzed by SMART Superdex 200 chromatography. The native molecular size was also in this case estimated to be 180 -220 kDa. Both preparations of the 54-kDa active fragment of PARN eluted consistently between the 158-kDa aldolase and the 232-kDa catalase molecular size markers. The big discrepancy between the molecular masses estimated by gel filtration and SDS-PAGE suggests that the nuclease associated with the 54-kDa active fragment of PARN is oligomeric. To further investigate the oligomeric structure of the nuclease, we molecularly cloned and expressed recombinant human 74-kDa PARN polypeptide and human 54-kDa PARN fragment. The 74-kDa polypeptide was found to be expressed in the soluble fraction of the bacterial host, whereas the 54-kDa fragment was present in inclusion bodies. The human 74-kDa PARN was affinity purified, and poly(A)-specific nuclease activity was recovered (data not shown). The purified recombinant human 74-kDa PARN nuclease was subjected to protein/protein cross-linking using the homobifunctional cross-linkers bis(sulfosuccinimidyl) suberate and dimethyl pimelimidate. The cross-linked polypeptides were fractionated by SDS-PAGE and subsequently revealed by either silver staining of the gel (data not shown) or by Western  Table I and "Experimental Procedures" for details.

TABLE I Purification of poly(A)-specific 3Ј exonuclease
Purification was monitored by the dual assay strategy and quantitated as described under "Experimental Procedures." One unit is defined as the release of 1 mol of AMP/min. The numbers, protein (mg) and activity (units), were obtained by multiplying the determined values by four.
Step blot analysis (Fig. 4B). The addition of the homobifunctional cross-linkers shifted the electrophoretic mobility of the 74-kDa polypeptide to two slower migrating forms: II and III (Fig. 4B). A careful inspection of the silver-stained gel (data not shown) revealed that none of the visible contaminating bacterial proteins was shifted in contrast to the 74-kDa PARN polypeptide, which was quantitatively shifted to slower migrating forms. Taken together these results strongly suggest that forms II and III were generated by the formation of protein/protein crosslinks between two and three 74-kDa PARN polypeptides, respectively. We conclude that the PARN nuclease is oligomeric and that it most likely consists of three subunits (i.e. homotrimer).
Conditions for in Vitro Deadenylation-The conditions for poly(A) tail removing activity were investigated by performing standard in vitro deadenylation reactions using the poly(A)-Sepharose fraction as the enzyme source and poly(A)-tailed labeled L3(A 30 ) RNA as the substrate. Activity was monitored by following the release of mononucleotides by the one-dimensional TLC assay. The requirements for monovalent (K ϩ and Na ϩ ) and divalent (Mg 2ϩ , Mn 2ϩ , Zn 2ϩ , and Ca 2ϩ ) cations were investigated. We found that monovalent cations were required and that the optimal concentration was around 100 mM. Poly(A) removing activity was higher in the presence of K ϩ than in the presence of Na ϩ . The nuclease activity was dependent on divalent cations, and Mg 2ϩ was found to be the preferred divalent ion with an optimal concentration around 1 mM. The nuclease was active in the presence of Mn 2ϩ . However, the activity decreased to 17% of the nuclease activity detected in the presence of Mg 2ϩ . No activity was detected in the presence  p(A)). The resulting gel was stained by silver. Molecular mass markers were separated in lane M. Numbers to the left indicate molecular masses of marker proteins in kDa. Capital letters (A-F) to the right indicate relative positions of gel slices that were cut out from a similar preparative gel in which 0.7 ml of the poly(A)-Sepharose 4B fraction was fractionated. Proteins were eluted from the gel slices and subsequently denatured and renatured (34). B, renatured proteins isolated from gel slices A-F (see A) were incubated together with uniformly labeled RNA substrate L3(A 30 ) under conditions for in vitro deadenylation for 90 min. In each reaction 12 l of renatured proteins from a total volume of 1 ml of renatured protein was used. Reacted RNA was recovered and fractionated by electrophoresis using a 10% polyacrylamide:bisacrylamide 19:1-7 M urea gel. The resulting flourogram is shown. Arrows to the right indicate the location of RNA substrate (S) and product (P).

FIG. 3. SDS-PAGE of proteins used for mass spectrometry.
Lane 7MeG, proteins (0.75 g in 1.65 ml) in fraction 5Ј-AMP/7-methyl-GTP was precipitated and fractionated by SDS-PAGE. The resulting gel was stained by silver. A gel slice containing the 54-kDa protein (marked with an arrow) was cut out from the gel and subsequently in-gel digested with trypsin. Tryptic peptides were analyzed by MALDI-TOF mass spectrometry. Molecular mass markers were separated in lane M. Numbers to the right indicate molecular masses of marker proteins in kDa.

FIG. 4. The nuclease is oligomeric.
A, the 5Ј-AMP/7-methyl-GTP fraction was fractionated by SMART Superdex 200 chromatography as described under "Experimental Procedures." Obtained fractions (14 -24) were incubated together with uniformly labeled RNA substrate L3(A 30 ) under conditions for in vitro deadenylation for 120 min. In lane L the RNA substrate was incubated with the 5Ј-AMP/7-methyl-GTP fraction. Reacted RNA was recovered and fractionated by electrophoresis using a 10% polyacrylamide:bisacrylamide 19:1-7 M urea gel. The resulting flourogram is shown. Arrows to the right denote the locations of RNA substrate (S) and product (P). Arrows at the top indicate the elution profile during SMART Superdex 200 chromatography of the molecular mass markers. Molecular mass is given in kDa. B, recombinant human 74-kDa PARN polypeptide (4 or 10 g, as indicated, in a 100-l cross-linking reaction, respectively) was treated with indicated homobifunctional cross-linkers bis(sulfosuccinimidyl) suberate and dimethyl pimelimidate as outlined under "Experimental Procedures" and fractionated by SDS-PAGE. The gel was subsequently analyzed by Western blot analysis using a His-tag-specific antiserum as the probe. The resulting flourogram is shown. Arrows to the left indicate the positions of forms I, II, and III. Numbers to the right indicate the position of molecular size markers in kDa. S/S denotes the boarder between the stacking and separation gels.
of Zn 2ϩ or Ca 2ϩ . The optimal pH was found to be around 7.
Substrate Specificity-The RNA substrate specificity of the poly(A) removing activity was investigated by three sets of experiments. First, RNA substrates L3(A 30 ), ML54(U 30 ), ML40(C 32 ), and ML43(G 14 ) radioactively labeled in their RNA bodies were incubated in conditions for in vitro deadenylation using either the poly(A)-Sepharose fraction or the final 7-methyl-GTP fraction as the source of enzyme. Reacted RNA was purified and analyzed by gel electrophoresis (Fig. 5). It was found that RNA substrate L3(A 30 ) was efficiently degraded by the 3Ј exonuclease activity present in both tested fractions and that an RNA product corresponding to the deadenylated RNA body accumulated (Fig. 5A and data not shown). The RNA substrate ML54(U 30 ) was degraded to some extent without the accumulation of the RNA body (Fig. 5A), whereas RNA substrates ML43(G 14 ) and ML40(C 32 ) were almost unaffected by the 3Ј exonuclease activity in both fractions (Fig. 5A and data  not shown). Secondly, the RNA substrate specificity was investigated by incubating RNA substrates having internal poly(A) stretches followed by plasmid encoded RNA sequences of increasing length with either the poly(A)-Sepharose fraction or the final 7-methyl-GTP fraction under conditions for in vitro deadenylation. It was found that RNA substrates L3(A 30 )X 15 , L3(A 30 )X 49 , and L3(A 30 )X 164 were almost unaffected by the 3Ј exonuclease activity in both preparations in contrast to the RNA substrate L3(A 30 ), which was efficiently deadenylated (Fig. 5B). Finally, RNA substrates L3(A 30 ), ML54(U 30 ), ML40(C 16 ), and ML43(G 14 ) radioactively labeled in their homopolymeric tails were incubated in conditions for in vitro deadenylation using the poly(A)-Sepharose fraction. The release of radioactive mononucleotides was monitored by onedimensional TLC assay. From this analysis we determined the apparent K m and V max values for each RNA substrate using Lineweaver-Burk formalism. Table II summarizes the obtained K m values and the calculated relative V max /K m values. Interestingly, the K m values for the different substrates were all in the 10 nM range, suggesting that substrate specificity is related to the V max parameter. Based on these three sets of experiments, we conclude that the nuclease activity associated with the 54-kDa active fragment of PARN is highly specific for degrading poly(A) and that it preferentially degrades 3Ј endlocated poly(A) tails.
5Ј-AMP Is the Released Mononucleotide-The nature of the released mononucleotide reaction product was investigated by two-dimensional TLC. The poly(A)-Sepharose fraction or the final 7-methyl-GTP fraction was incubated, under condition for in vitro deadenylation, together with RNA substrate L3(A 30 ) radioactively labeled in its homopolymeric adenosine tail. The reaction products were analyzed by two-dimensional TLC using nonlabeled 2Ј-AMP, 3Ј-AMP, and 5Ј-AMP as markers. The result ( Fig. 6 and data not shown) showed that the released mononucleotide comigrated with 5Ј-AMP. We conclude that 5Ј-AMP is a reaction product and that an exonuclease activity is responsible for the degradation.
The Nuclease Activity Is Highly Processive-To investigate whether the nuclease activity degraded RNA substrates in a processive or distributive fashion, we first titrated the exonuclease, using the final 7-methyl-GTP-Sepharose fraction as the source of enzyme and the RNA L3(A 30 ) labeled in its body as the substrate. Incubation time was 10 min. Fig. 7A (lanes 1-6) shows that nonreacted RNA substrate and fully deadenylated RNA product were present when a low amount of exonuclease was added. Next, six identical reactions using 2 l of the 7-methyl-GTP fraction as the exonuclease source were again incubated for 10 min, but in this case an increasing amount of poly(A) was added to each reaction. Fig. 7A (lanes 7-12) shows that addition of poly(A) inhibited the reaction and that both nondeadenylated and completely deadenylated RNA substrates were present when the concentration of poly(A) was 40 pg/l or 0.4 ng/l (lanes 10 and 11). The same results were obtained using the partially purified poly(A)-Sepharose frac-  tion as the source of enzyme (data not shown). The concentration of the RNA substrate in the experiments shown in Fig. 7A was approximately 0.7 nM, which is below the K m value of the RNA substrate. To investigate the disappearance of the RNA substrate and the appearance of fully deadenylated product at substrate concentration around the K m value and at saturation, we increased the RNA substrate concentration to 7 and 35 nM, respectively. We used 2 l of the 7-methyl-GTP fraction as the source of exonuclease and followed the time courses of these two reactions. The predictions for a highly processive exonuclease activity are: (i) the time point for the appearance of the first fully deadenylated product should not be affected by the RNA substrate concentration; (ii) unreacted RNA substrate and fully deadenylated product should be present at some time points; and (iii) a homogenous population of partially deadenylated RNA substrates should not be visible. The result of this experiment is shown in Fig. 7B. We conclude that the pattern of degradation during these sets of experiments was consistent with a highly processive exonuclease activity, i.e. the enzyme catalyzes multiple rounds of digestion without dissociating from the polymeric substrate. A Functional Link between the RNA 5Ј End Cap Structure and Poly(A) Tail Removal-The RNA substrate we have used during purification was capped at its 5Ј end by the inclusion of m 7 G(5Ј)ppp(5Ј)G cap analogue during in vitro transcription of the L3(A 30 ) RNA substrate. To investigate whether the presence of a cap at the 5Ј end affected the nuclease activity of the 54-kDa fragment of PARN, we compared the relative specific activity of the nuclease using either capped or noncapped L3(A 30 ) RNA, radioactively labeled in their homopolymeric tails, as the substrate. The released AMP was quantitated by the one-dimensional TLC assay. We found that the specific activity was approximately 6-fold higher using the capped L3(A 30 ) RNA substrate compared with the noncapped RNA substrate, suggesting that a cap structure at the 5Ј end stimulated the poly(A) tail removing activity of the 54-kDa fragment of PARN. To further investigate the role of the cap during poly(A) tail removal, we added in trans m 7 G(5Ј)ppp(5Ј)G, 5Ј GMP or 5Ј-AMP to in vitro deadenylation reactions using capped L3(A 30 ) RNA as the substrate. Fig. 8 shows that the free m 7 G(5Ј)ppp(5Ј)G cap analogue severely inhibited the deadenylation reaction already at 0.01 mM and completely inhibited the reaction at 0.1 mM, whereas 5Ј GMP and 5Ј-AMP only inhibited the reaction at much higher concentration. The efficient inhibition of the nuclease activity by the cap analogue acting in trans and the stimulatory role of the cap acting in cis strongly suggest that the nuclease interacts with the cap structure during poly(A) degradation establishing a functional link between poly(A) tail removal at the 3Ј end and the cap structure at the 5Ј end of the RNA substrate.

DISCUSSION
In this paper we report on the purification to apparent homogeneity of a poly(A)-specific 3Ј exonuclease activity (17,18) Tables I and II). A 54-kDa polypeptide (Fig. 3) corresponding to a fragment of the 74-kDa PARN nuclease (19,20) copurified with the 3Ј exonuclease activity. The molecular mass of the nuclease, as determined by gel filtration, was found to be 180 -220 kDa (Fig. 4A). The purification of the poly(A)-specific 3Ј exonuclease activity to apparent homogeneity is an important step toward mechanistic studies of the native nuclease. This work will therefore provide a solid platform for further studies using recombinant PARN nuclease as the source of activity.
The properties of the purified 54-kDa active fragment of minutes. Lanes 1-10 and 11-20 were treated as two separate objects (corresponding to two different exposure times using traditional autoradiography) during PhosphorImager analysis to visualize RNA substrate and deadenylated product. Thus, the relative darkness caused by radioactivity in lanes 1-10 versus lanes 11-20 should not be compared visually.
PARN were investigated and compared with a similar activity in HeLa cell free extracts that we previously have identified (17,18). Both activities release 5Ј-AMP as the reaction product (Fig. 6), preferentially degrade poly(A) (Table II and Fig. 5A), and can only efficiently degrade poly(A) located at the 3Ј end of the RNA substrate (Fig. 5B). Both activities generate a deadenylated RNA body during in vitro deadenylation (Fig. 5A), which upon prolonged incubation eventually will be degraded (data not shown). Both activities are strictly dependent on divalent cations, preferentially Mg 2ϩ , for their activities. The two activities have very similar chromatographic properties, and both can be purified by similar purification protocols (data not shown). We have not managed to obtain sufficient amount of the HeLa cell activity to unambigiously identify the 54-kDa PARN polypeptide by SDS-PAGE and silver staining. We conclude that the purified calf thymus poly(A) removing activity is a poly(A)-specific 3Ј exonuclease and that it corresponds to the poly(A)-specific 3Ј exonuclease activity that we previously identified in HeLa cell-free extracts.
The N-and C-terminal ends of the 54-kDa active fragment of PARN have not been unambigiously defined. However, we note that one of the peptides in the MALDI-TOF mass spectrum, 1461.69 in molecular weight, could correspond to the C-terminal end of the 54-kDa PARN fragment. Four lines of evidence support this suggestion: (i) a peptide of PARN consisting of amino acids 458 -470 has the corresponding molecular weight; (ii) a lysine residue is located at amino acid 457 of PARN, which should be expected after digestion with trypsin; (iii) no peptide corresponding to predicted tryptic peptides located C-terminally of amino acid 471 of PARN was identified in the mass spectrum; and (iv) the calculated molecular mass of a predicted polypeptide consisting of amino acids 1-470 of PARN is 54.2 kDa, which corresponds to the molecular mass of the purified polypeptide, as determined by SDS-PAGE analysis. It is note-worthy that a tentative exonuclease domain belonging to the RNase D family of 3Ј-exonucleases (28, 37) is located within the N-terminally located 389 amino acids of human PARN (20).
We have investigated by Western blot analysis the presence of the 74-kDa PARN polypeptide in our fractions by using a polyclonal antibody directed against the C-terminal part of the 74-kDa PARN polypeptide (20) as the probe. Our data showed that large amount of the 74-kDa PARN polypeptide was present in the initial crude extract. However, the 74-kDa polypeptide did not copurify with the nuclease activity we selected based on our dual assay strategy. 2 A very low abundant polypeptide in the size range of 80 kDa was visible in our purest fraction (Fig. 3). We have not been able to confirm that this polypeptide is related to the 74-kDa PARN polypeptide, neither by Western blot analysis or mass spectrometry. Unfortunately, it is not possible to investigate the presence of the 54-kDa polypeptide in the cruder fractions by Western blot analysis because of the inability of the antibody to recognize the 54-kDa PARN fragment. Thus, we cannot rule out the possibility that the 54-kDa active fragment of PARN was generated by proteolysis during purification. It is important to keep this possibility in mind when interpreting several of the properties of the nuclease activity that we have described in relation to the nuclease activity associated with the 74-kDa PARN polypeptide. However, high purity of an enzyme is a prerequisite for detailed mechanistic studies.
Körner et al. (20) detected two isoforms of PARN in Xenopus oocytes, one being 74 kDa in molecular size and the other 62 kDa. The 62-kDa form was the dominating form co-purifying with the nuclease activity after poly(A)-Sepharose chromatography. Interestingly, the two forms differed in subcellular distribution in Xenopus oocytes; the 62-kDa form being cytoplasmic, whereas the 74-kDa form was nuclear. It is therefore possible that the 54-kDa form that we purified represents an additional isoform of bovine PARN. Further experiments are required to resolve this issue (see also below).
The development of in vitro systems and the characterization of participating enzymes for polyadenylation and deadenylation will make it possible to reveal mechanisms regulating the poly(A) tail length. One interesting aspect is the competition between mRNA poly(A) tail addition and removal (reviewed in Refs. 38 and 39). The highly processive mode of degradation ( Fig. 7) will definitely influence this competition, because a highly processive nuclease will compete very efficiently with the opposing poly(A) synthesis reaction once degradation has been initiated. This is in sharp contrast to a situation where a distributive nuclease activity will be competed by the poly(A) synthesis reaction after removal of each single adenosine residue.
An intriguing possibility could be that the processive mode of degradation is a unique property of the nuclease activity associated with the 54-kDa polypeptide of PARN because Körner et al. (20) previously reported that the 74-kDa PARN nuclease activity degraded poly(A) in a distributive fashion. Thus, the nuclease activity associated with the 74-kDa PARN polypeptide that they purified differed in one important mechanistic aspect compared with the 54-kDa PARN fragment activity. One reason for purifying two mechanistically distinct nuclease activities could be that different assay strategies were used to follow the nuclease activity during purification. Körner et al. (20) selected fractions for further purification based on the release of mononucleotides, whereas we selected fractions based on a dual assay strategy. This difference in assay strategies could very well be responsible for preferentially selecting 2 J. Martinez, Y.-G. Ren, and A. Virtanen, unpublished observation. Arrows to the right denote the locations of RNA substrate (S) and product (P). Numbers above lanes denote concentration of 5Ј-AMP, GMP, or m 7 G(5Ј)ppp(5Ј)G, respectively (in mM). In lanes denoted S RNA substrate was incubated without enzyme. a distributive nuclease activity in one case and a processive nuclease activity in the other case.
Three lines of evidence suggests an interaction between the nuclease associated with the 54-kDa active fragment of PARN and the mRNA cap structure: (i) the specific activity of the nuclease is higher for capped RNA substrate than for noncapped; (ii) a free cap analogue added in trans inhibits the nuclease activity (Fig. 8); and (iii) the nuclease activity interacts very strongly with the 7-methyl GTP matrix, which partly resembles a methylated cap structure. This matrix is frequently used to affinity purify cap binding proteins (40). All of this evidence suggests that the 54-kDa fragment of PARN contains a cap-binding site. The presence of a cap-binding site implies that the 54-kDa polypeptide interacts both with the 5Ј and the 3Ј end of the mRNA during deadenylation. We can only speculate about the functional importance of this interaction: (i) a physical interaction between the cap structure and the poly(A) tail during deadenylation will most likely interfere with the translational initiation process (reviewed in Ref. 41); thus, the interaction may be part of a signal that informs the translational machinery that the mRNA is subjected to degradation and should not be used for protein translation; (ii) the interaction between the cap and the poly(A) tail may play a role in defining a deadenylation-dependent decapping pathway of RNA degradation; (iii) the cap structure and the poly(A) tail may represent two separate cis acting elements that together ensure that the PARN nuclease preferentially selects polyadenylated mRNA as the RNA substrate; and (iv) a physical interaction between the nuclease and the cap structure may stabilize the enzyme/substrate complex. The cap structure may therefore also play a role in making the 54-kDa PARN fragment a highly processive nuclease.