Structure-function relationships in the Saccharomyces cerevisiae poly(A) polymerase. Identification of a novel RNA binding site and a domain that interacts with specificity factor(s).

We have constructed deletions in the nonconserved regions at the amino and carboxyl ends of the poly(A) polymerase (PAP) of Saccharomyces cerevisiae and examined the effects of these truncations on function of the enzyme. PAP synthesizes a poly(A) tail onto the 3'-end of RNA without any primer specificity but, in the presence of cellular factors, is directed specifically to the cleaved ends of mRNA precursors. The last 31 amino acids of PAP are dispensable for both nonspecific and specific activities. Removal of the next 36 amino acids affects an RNA binding domain, which is essential for the activity of the enzyme and for cell viability. This novel RNA binding site was further localized using additional deletions, cyanogen bromide cleavage of PAP cross-linked with RNA or 8-azido-ATP, and a monoclonal antibody against a COOH-terminal PAP epitope. A deletion that partially disrupts this domain has reduced nonspecific activity but functions in specific polyadenylation. In contrast, deletion of the first 18 amino acids of PAP has no effect on nonspecific polyadenylation but completely eliminates specific activity. This region is essential for enzyme function in vivo and is probably involved in the interaction of PAP with other protein(s) of the polyadenylation machinery.

We have constructed deletions in the nonconserved regions at the amino and carboxyl ends of the poly(A) polymerase (PAP) of Saccharomyces cerevisiae and examined the effects of these truncations on function of the enzyme. PAP synthesizes a poly(A) tail onto the 3end of RNA without any primer specificity but, in the presence of cellular factors, is directed specifically to the cleaved ends of mRNA precursors. The last 31 amino acids of PAP are dispensable for both nonspecific and specific activities. Removal of the next 36 amino acids affects an RNA binding domain, which is essential for the activity of the enzyme and for cell viability. This novel RNA binding site was further localized using additional deletions, cyanogen bromide cleavage of PAP cross-linked with RNA or 8-azido-ATP, and a monoclonal antibody against a COOH-terminal PAP epitope. A deletion that partially disrupts this domain has reduced nonspecific activity but functions in specific polyadenylation. In contrast, deletion of the first 18 amino acids of PAP has no effect on nonspecific polyadenylation but completely eliminates specific activity. This region is essential for enzyme function in vivo and is probably involved in the interaction of PAP with other protein(s) of the polyadenylation machinery.
The yeast Saccharomyces cerevisiae shares mechanisms of mRNA maturation common to most eukaryotic species. One step in mRNA biogenesis is the formation of a modified 3Ј-end, a complex and highly regulated process that includes recognition of specific RNA sequence, cleavage and polyadenylation, and termination of poly(A) addition once a certain average tail length is reached (1,2). This is accomplished by a multiprotein complex, which in yeast can be separated into four factors (CF I, CF II, PF I, and PAP) 1 essential for reconstitution of accurate 3Ј-end formation in vitro (3). The yeast PAP is an important component of the 3Ј-processing machinery. When mixed with partially purified CF I and PF I, it is active only on RNA substrates containing specific polyadenylation signal sequences (3). However, replacement of magnesium for manganese or separation of PAP from CF I and/or PF I (3) completely eliminates this specificity. Components in mammalian cells, similar to CF I and PF I, provide analogous functions (1,2). In this sense, PAP can be considered as a catalytic subunit of a multicomponent enzyme, which requires at least two other polypeptides to confer specificity for RNA substrate. A similar situation has been described for vaccinia PAP, in which two proteins comprise the holoenzyme (4,5).
The nature of the interaction of PAP and other components, as well as the interaction of PAP with RNA and ATP during catalysis, is mostly unknown. Purified yeast PAP has been characterized (6), and the gene has been cloned and expressed in Escherichia coli (7). By using PAP as a bait in the two-hybrid system for screening yeast DNA libraries, it has been shown that PAP interacts with FIP1, a component of the PF I factor (8), and with PIP2, a protein homologous to ubiquitin-activating enzyme, and PIP3, a protein of unknown function (9). 2 The yeast PAP has strong homology in some regions with vertebrate PAPs (10) but also has unique nonconserved sequences located in the amino and carboxyl-terminal parts of the protein. There is a common RNA binding domain (RNP1) (7) whose role in PAP function has been confirmed by mutagenesis of the bovine gene (11). Surprisingly, similar conservation does not exist when the eukaryotic PAP sequences are compared to that of PAP from E. coli (12,13) or Vaccinia virus (5). A knowledge of the mechanism of poly(A) polymerase action and a correlation of structure with function would provide useful insights into the regulation of polyadenylation and its interaction with other components of the polyadenylation complex. In this paper, we describe a previously uncharacterized region at the amino terminus of the yeast PAP, which is responsible for specific protein-protein interaction, and a new and unsuspected RNA binding site at the carboxyl end. We also discuss the importance of these regions to in vitro polyadenylation and cell viability.
The pJPAP1 expression plasmid was a kind gift of J. Lingner and W. Keller (7). Serial deletions of the 5Ј-and 3Ј-ends of the PAP gene were made using polymerase chain reaction and natural restriction sites. To delete the last 67 amino acids of PAP, pJ⌬6PAP was created by digesting pJPAP1 with BstEII and PvuII, and to delete the last 20 amino acids (pJ⌬7PAP), pJPAP1 was digested with BglII and PvuII. The digestion products were blunted with DNA polymerase I (Klenow) fragment (New England Biolabs), and the large fragments were circularized using DNA ligase (New England Biolabs). pJ⌬8PAP (deleting the last 55 amino acids), pJ⌬9PAP (deleting the last 43 amino acids), pJ⌬10PAP (deleting the last 31 amino acids), pJ⌬14PAP (deleting amino-terminal amino acids [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18], and pJ⌬15PAP (deleting amino-terminal amino acids 3-44) were made by replacement of the PAP gene sequence between the SacI and PvuII sites in pJPAP1 with appropriate polymerase chain reaction fragments (Fig. 1). For testing function of the truncated PAPs in yeast, the same fragments were inserted into pHCp50 plasmid, a derivative of YCp50 (18), with the HindIII fragment of yeast chromosomal DNA containing PAP1 sequence. pHCp50 was a kind gift of Dr. W. Keller. All restriction and modifying enzymes were from New England Biolabs.
Protein Expression and Purification-All procedures were done at 4°C except where indicated. Truncated proteins were expressed with the T7 expression system (19), using BL21(DE3)pLysS cells grown for 2 h at 37°C after isopropyl-1-thio-␤-D-galactopyranoside induction. Cells from 1 liter of culture were harvested, washed in 100 ml of phosphatebuffered saline, 1 mM PMSF, pelleted, and quickly resuspended in lysis buffer (10 mM Tris-HCl, pH 8.0, 10% glycerol, 250 mM KCl, 0.5% Nonidet P-40, 1 mM EDTA, 0.5 mM DTT, 1 mM PMSF, 0.6 M leupeptin, 2 M pepstatin A), and frozen with liquid nitrogen. Cells were thawed at 30°C and incubated 15 min at this temperature. Lysates were centrifuged at 100,000 ϫ g for 1 h. The supernatant was briefly sonicated, diluted with DE 0 buffer (10 mM Tris-HCl, pH 8.0, 10% glycerol, 0.25 mM EDTA, 0.5 mM DTT, 1 mM PMSF) to a concentration of 100 mM KCl and loaded onto an equilibrated 20-ml DEAE-Sephacel (Pharmacia) column, and proteins eluted with either a loading buffer or a gradient of 100 -500 mM KCl, if necessary. Fractions containing PAP activity were combined. Potassium phosphate buffer, pH 6.8, was added to a final concentration of 30 mM, and the solution was loaded onto a 10-ml phosphocellulose P-11 (Whatman) column equilibrated with 30 mM potassium phosphate, pH 6.8, 10% glycerol, 100 mM KCl, 0.25 mM EDTA, 0.5 mM DTT, 1 mM PMSF. After washing the column with the same buffer, PAP was eluted with a salt gradient of 100 -700 mM KCl in the same buffer. Fractions containing PAP activity were dialyzed against DE100 buffer (DE 0 plus 100 mM KCl) and loaded onto a 1-ml HiTrap Heparin column (Pharmacia) equilibrated with DE100 buffer. After washing with DE100 buffer, proteins were eluted with a 100 -500-mM KCl gradient. Active fractions were combined, diluted with DE 0 buffer to 50 mM KCl, and loaded onto a 1-ml Mono Q column (Pharmacia). After washing with DE50 buffer, a 50 -500 mM KCl gradient was applied. Fractions containing PAP activity were combined and stored at Ϫ70°C.
Yeast whole extract was prepared as described (3) except that the spheroplasting step was omitted. Fractions containing CF II/PF I were concentrated to 4 mg/ml, and CF I was further purified about 100-fold and used at a concentration of 0.4 mg/ml. All extracts and factors were frozen in liquid nitrogen and stored at Ϫ70°C.
Protein concentrations were determined by the procedure of Bradford (20) using the Bio-Rad kit and bovine serum albumin as a standard. Protein markers were from New England Biolabs. After incubation at 30°C for 20 min, reactions were stopped by adding 100 l of 10% trichloroacetic acid. Trichloroacetic acid-precipitable counts were collected on GF/C filters (Whatman), and Cherenkov counts were measured. For kinetic studies, 2-l aliquots were taken at 2, 5, 10, and 20 min, stopped, and counted, and the initial rate of reaction was calculated.
To assay specific polyadenylation, the reactions (16 l) were assembled on ice and contained 1 mM magnesium acetate, 75 mM potassium acetate, 2% polyethylene glycol (8000), 2 mM ATP, 20 mM creatine phosphate, 10 nM (16 ng) labeled pre-cleaved mRNA, and 1.5 M (0.6 g) tRNA. The reactions contained either 1 l of yeast whole cell extract (20 g of protein) or 2 l of CF I fraction (0.8 g of protein), 2 l of CF II/PF I (8 g of protein), and 1 l of diluted PAP (13 ng of protein) and were incubated at 30°C for 20 min and stopped with proteinase K and SDS as described previously (3,21). The reaction products were fractionated on 5% polyacrylamide, 8.3 M urea gels and visualized by autoradiography.
Cross-linking of PAP with RNA and 8-Azido-ATP-Reactions were carried out in 10 l of 20 mM HEPES KOH, pH 7.5, 10% glycerol, 50 mM KCl, 0.25 mM EDTA, 0.5 mM DTT, 0.64 g (10 pmol) of PAP, and 10 -50 M [␣-32 P]8-azido-ATP or 10 6 cpm (10 -50 nM) of 32 P-labeled RNA. After incubation at room temperature for 5 min, samples were irradiated in a UV Stratalinker 1800 (Stratagene), 1 auto cycle for azido-ATP (120 millijoules) and 10 cycles for RNA (1.2 joules). For PAP/RNA crosslinking, samples were digested with RNase A (1 g/ml) for 1 h at 37°C. In the case of binding competition experiments, different nucleic acids were added to the reaction mixture prior to incubation as indicated in the figure legends. After treatment, samples were separated on a 10% polyacrylamide gel containing SDS, and proteins were visualized by autoradiography.
For cyanogen bromide (CNBr) cleavage (22), UV-cross-linked samples were precipitated with 10% trichloroacetic acid, rinsed with ethyl ether, and resuspended in 100 l of 70% formic acid containing 10 -100 g/ml freshly sublimated CNBr. After an overnight reaction at room temperature, samples were diluted with water and lyophilized twice. The resulting peptides were separated on a 16% polyacrylamide gel using Tris-tricine-SDS buffer (23,24), transferred to PVDF membrane, and detected by autoradiography. The membrane was treated for immunoblot analysis using PAP-specific monoclonal antibody as described previously (21).

RESULTS
Protein Expression and Purification-Truncated forms of PAP ( Fig. 1) were expressed in vitro using the T7 expression system (19), and crude E. coli lysates were tested for PAP activity and by immunoblotting, using PAP-specific monoclonal antibodies (21). Degradation was minimal if the expression time was limited to 2 h after induction. The nonspecific activity of the various truncated enzymes was measured and compared with the activity of wild type PAP from the same amount of cells. This assay estimates the enzyme's ability to incorporate adenosine monophosphate onto a primer in the presence of manganese. The average nonspecific activity from two independent expressions of truncated and wild type PAPs is shown in Fig. 1. The first 18 and the last 31 amino acids of PAP are dispensable for nonspecific polyadenylation. Deletion of 44 amino acids from the amino terminus of yeast PAP eliminates the epitope of an amino-terminal specific antibody and inactivates the enzyme. Serial deletions from the carboxyl terminus lead to a partial inactivation of the enzyme until a deletion of 67 amino acids, which completely inactivates the PAP's ability to catalyze nonspecific polyadenylation. The nature of this inactivation is discussed in a later section.
Truncated forms of PAP were purified using the protocol described under "Materials and Methods." It is interesting that some of the COOH-terminal truncated proteins exhibited chromatographic behavior different from that of the full-length enzyme. For example, on a DEAE-Sephacel column, wild type PAP comes out in the first half of the flow through fractions at a KCl concentration of 100 mM, but ⌬8PAP eluted from DEAE-Sephacel with 250 mM KCl. On a phosphocellulose column, this effect is even more dramatic. Wild type PAP elutes at 400 mM KCl, ⌬10PAP at 300 mM, and ⌬9PAP at 50 mM KCl; ⌬8PAP does not bind phosphocellulose at all and appears in the flowthrough fractions at 0 mM KCl. This altered behavior of ⌬9PAP and ⌬8PAP correlates with the large drop-off in nonspecific activity observed with these truncations and may indicate that we have deleted a phosphate binding site.
Enzymatic Properties of Purified Truncated PAPs in Nonspecific and Specific Polyadenylation-Using Michaelis-Menten enzyme kinetics, we determined some of the kinetic parameters of the active COOH-terminal truncated PAPs. We calculated the initial rates of reaction using time courses of oligo(A) polyadenylation at varying concentrations of primer or ATP, and the kinetic constants K m and V max were determined by a double reciprocal plot of 1/V 0 against 1/S. As shown in Table I, the K m for ATP is the same for wild type and the COOH-terminal truncated PAPs, indicating that the deletions do not affect ATP binding. In contrast, the K m for RNA increased 50-fold for ⌬9PAP. These results suggest that the decrease in the nonspecific activity of ⌬9PAP, despite the 2-fold increase of the V max compared to that of wild type, is due primarily to an inability to bind the RNA substrate.
Two different RNA substrates, both of which terminate at the GAL7 poly(A) site, were used to assay the ability of the enzyme to participate with other protein factors in specific polyadenylation. The first substrate contains 161 nucleotides of the wild type GAL7 sequence upstream of the poly(A) site, and the second is identical except for a deletion of a 12-nucleotide UA repeat, an element essential for specific activity (3) (Fig.  2A, lanes 1 and 2). Purified recombinant PAP was mixed with either of these substrates in the presence of CF I and PF I. As expected, in the absence of PAP, no poly(A) addition activity is observed (Fig. 2A, lane 3). The ⌬10PAP exhibits the same specific activity as the wild type enzyme ( Fig. 2A, lanes 4 and  5). As might be expected, the amount of polyadenylated product in reactions containing ⌬9PAP is less (Fig. 2A, lane 6), but all of this is the result of specific polyadenylation, since no product is observed using the mutated substrate ( Fig. 2A, lanes 7-9).
Elimination of amino acids 3 through 18 (⌬14PAP) from the amino-terminal end of PAP has no effect on nonspecific activity (Fig. 1). However, this truncation completely knocks out specific polyadenylation activity (Fig. 2B, lanes 3 and 4). The ⌬14PAP used in this experiment is active for nonspecific polyadenylation, since it exhibits wild type activity when magnesium is replaced with manganese (Fig. 2B, lanes 5 and 6).
Cross-linking of PAP to RNA-It was reported, based on sequence alignment and mutagenesis of bovine PAP, that an RNP1-type RNA binding domain was located in a highly conserved amino-terminal part of eukaryotic PAPs (7,11). However, our preliminary data discussed above pointed to the existence of another RNA binding site in the yeast PAP. We decided to use a more direct approach to determine whether the unique COOH-terminal part of the yeast PAP binds RNA and to explore the specificity of this binding site by cross-linking PAP to radioactive RNA using UV light. PAP could be UV cross-linked to 32 P randomly labeled precleaved GAL7 RNA, and this cross-linking can be competed with an excess of unlabeled tRNA (Fig. 3A). To localize the RNA cross-link, PAP was incubated with various substrates and exposed to UV light, and the complexes were treated with RNase A and partially cleaved with CNBr. The resulting peptides were separated on an 16% polyacrylamide SDS gel and blotted onto PVDF membrane, and labeled peptides were detected by autoradiography. These were matched with peptides visualized with a monoclonal antibody specific for the COOH terminus of yeast PAP (Fig. 3B). The epitope recognized by this antibody had been previously  [3][4][5][6] or mutant transcripts (lanes 2, 7-9) were assayed with yeast extract (lanes 1 and 2)  mapped by our laboratory using deletions of PAP as well as CNBr digests (21). The partial cleavage with CNBr generates a ladder of peptides of the size predicted from the location of methionines in the amino acid sequence (Fig. 3C). The positions of radioactive peptides correlated with immunostained peptides and indicated that the major RNA binding site detected using randomly labeled RNA was in the COOH-terminal part of yeast PAP (Fig. 3B, lane 3). An identical pattern was observed when the cross-linking was performed with precleaved GAL7 RNA labeled with 32 P only at the 5Ј-end (Fig. 3,  lane 2). When RNA labeled at the 3Ј-end was used, a different set of peptides was detected (Fig. 3, lane 1). The molecular mass of the major 3Ј-labeled peptide, calculated using the CNBr-generated peptides as standards, is about 17.5 kDa, which correlates well with the 17.8-kDa peptide containing the RNP1 motif (C3, in Fig. 3C). These results suggest that the region in the COOH-terminal peptide (C9, in Fig. 3C) binds the body but not the 3Ј-acceptor end of the RNA.
Cross-linking of PAP to 8-Azido-ATP-We wanted to use the same approach to identify the ATP binding site of yeast PAP. Because ATP UV-cross-links to PAP with very low efficiency (data not shown), we chose the ATP analog [␣-32 P]8-azido-ATP, which has been commonly used to identify nucleotide binding sites (26). Yeast PAP is readily labeled with this analog (Fig. 4,  lane 1).
To investigate the specificity of the azido-ATP cross-linking, we performed a competition assay by mixing different nucleic acids or nucleotides with PAP and azido-ATP before UV irradiation. RNA molecules, and even single-stranded DNA, competed with azido-ATP for PAP binding much more effectively than ATP and other nucleotide triphosphates (Fig. 4). These results suggest that azido-ATP is most likely interacting with an RNA binding site. To determine if this site corresponded to the one at the PAP carboxyl end, azido-ATP was cross-linked to wild type PAP, ⌬10PAP, and ⌬9PAP (Fig. 5). ⌬9PAP cross-links to 8-azido ATP much less efficiently than wild type PAP, and ⌬10PAP cross-links more efficiently, consistent with the results obtained for RNA binding by enzyme kinetic analysis.
To our surprise, 8-azido-ATP labeled the same peptide as did the randomly labeled RNA (Fig. 6, lanes 1 and 3). To further confirm that the carboxyl-terminal peptide was being labeled, we cross-linked azido-ATP to ⌬10PAP, which lacks the last 31 amino acids and would give a correspondingly shorter terminal peptide after CNBr cleavage. The labeled peptides after partial CNBr cleavage of ⌬10PAP shifted down the predicted distance (Fig. 6, lane 2), and this truncated PAP lost the epitope recognized by the COOH-terminal specific antibody (Fig. 6,  lane 4) (21).
The effect of PAP Truncations on Cell Viability-To test the ability of different PAP truncations to function in vivo, the wild type copy of the PAP1 gene on the plasmid pHCp50 was replaced with deleted forms of the gene. These plasmids were then transformed into a heterozygous diploid strain containing a LEU2 disruption in one copy of the endogenous PAP1 gene (7). The diploids were sporulated, and tetrad analysis was performed to determine if the truncated forms of PAP could rescue a lethal disruption of the chromosomal PAP1 gene. Eight or more tetrads were examined for each strain. Diploid strains containing the carboxyl deletions ⌬7PAP, ⌬10PAP, and ⌬9PAP all yielded tetrads with four viable spores. This is not surprising for ⌬7PAP and ⌬10PAP, which have wild type or nearly wild type nonspecific activity (Fig. 1). It is interesting that cells survived and grew normally with ⌬9PAP as the only PAP1 gene, since this deletion is only partially functional. On the other hand, tetrad analysis of ⌬8PAP yielded only two viable Leu Ϫ spores from each tetrad, indicating that the small amount of residual activity left in this enzyme was not sufficient for viability. ⌬14PAP, with a 16-amino acid deletion from the amino-terminal end and 100% of the wild type nonspecific activity, also could not support growth of cells, consistent with the hypothesis that an ability to associate with specificity factors would be essential for proper functioning of PAP in vivo.

DISCUSSION
The results described in this report present new features of the yeast poly(A) polymerase. A novel RNA binding site has been found and characterized in the carboxyl end of PAP, and a sequence involved in specific protein-protein interaction has been identified in the first 18 amino acids. From amino acids 80 -390, the yeast PAP has a strong, 70% homology to mammalian PAPs ( Fig. 1) (7). The RNP1 sequence, a conserved motif found in many RNA binding proteins (27), is located in this region. The first 79 amino acids and the last 178 amino acids of yeast PAP have very loose or very little homology to the mammalian counterparts. However, elimination of the first 44 amino acids (⌬15PAP) or the last 67 amino acids (⌬6PAP) completely inactivates the yeast enzyme. On the other hand, elimination of 16 amino acids from the amino-terminal end (⌬14PAP) or a COOH-terminal deletion of 31 amino acids (⌬10PAP) has no effect on the nonspecific activity of the yeast PAP.
Deletion of an additional 12 amino acids (⌬9PAP) from the COOH-terminal end of PAP causes a dramatic loss of nonspecific activity in crude extract, decreases the protein's ability to bind phosphocellulose, and increases the constant of RNA binding (K m ) by 50-fold. The sequence removed in ⌬9PAP (Fig. 7A) is highly basic; seven of the amino acids are lysine and arginine. Our data are consistent with a model in which this deleted sequence is part of an RNA binding site. After deletion of the next 12 amino acids (⌬8PAP), only a low level of activity remains, and deletion of the next 12 amino acids (⌬6PAP) completely inactivates the enzyme. A sequence related to that deleted in the yeast ⌬8PAP and ⌬9PAP is located in the aminoterminal region of mammalian PAPs (28,29) and Xenopus PAP (10) (Fig. 7A). Similarity can also be found in parts of the Vaccinia (5) and E. coli enzymes (12). However, this stretch does not resemble any of the previously described RNA binding motifs (27).
Because recombinant yeast PAP can use ATP instead of RNA as a primer 3 and 8-azido-ATP binds to the carboxyl RNA binding site, we suspected that this site might bind the 3Ј-end of the RNA primer in preparation to receive an adenosyl residue. However, by using RNAs labeled randomly or at either end, we could show that, even though this site is essential for activity, it is not specific for any portion of the RNA strand. This conclusion was supported by the binding competition assay where even single-stranded DNA competed effectively. These results also imply that the authentic ATP binding site is very specific and that the bulky azido group in this position of the analog interferes with binding.
A good candidate for a 3Ј-RNA binding site may be the RNP1 motif, which is very conserved among the eukaryotic, nonviral PAPs (10) and is adjacent to a potential catalytic domain (11). Mutations in the RNP1 of bovine PAP inhibit both specific and nonspecific activities, and this defect could be rescued by increasing the concentration of RNA (11), suggesting it is indeed involved in RNA binding, perhaps of the 3Ј-end. The fact that the major yeast PAP peptide cross-linked to 3Ј-end-labeled RNA after CNBr cleavage is similar in size to the yeast RNP1containing peptide lends support to this hypothesis.
Deletion of 16 amino acids (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18) from the amino terminus of the yeast PAP (⌬14PAP) completely knocked out the specific activity of the enzyme, leaving the nonspecific activity untouched. Regions with strong similarity (56%) to this yeast sequence are found in the amino-terminal ends of the higher eukaryotic enzymes (Fig. 7B), but this region has not been correlated with any particular function. The discovery of a specificity determinant in this part of the yeast PAP was surprising because previous mutagenesis of the bovine PAP indicated that a domain with similar function coincided with a nuclear localization signal (NLS1) in the COOH-terminal part of the protein (11). This region has no homology to the motif we have identified, and it is possible that two different domains contribute to the interactions of PAP with other polyadenylation factors.
Such a model invoking multiple contacts would not be implausible, since it has been shown that the processivity of the mammalian PAP is stimulated by either CPSF, the mammalian specificity factor, or PABII, a nuclear poly(A) binding protein, but is maximal when both factors are present simultaneously (30,31). Interestingly, the U1A protein can downregulate PAP activity, apparently through a domain which is further toward the carboxyl end of PAP than the NLS1 site, yet this interaction does not affect the ability of PAP to stimulate FIG. 5. The efficiency of 8-azido-ATP binding of different truncated PAPs was determined by UV cross-linking followed by separation on a 10% SDS-polyacrylamide gel, which was stained with silver ( lanes 1-3), and radioactive proteins were visualized by autoradiography ( lanes  5-7). Lanes 1 and 5, wild type PAP; lanes 2 and 6, ⌬10PAP; and lanes 3 and 7, ⌬9PAP. Lane 4, broad range marker proteins (New England Biolabs). 1 and 3) and ⌬10PAP (lanes 2 and 4) UV-cross-linked with radiolabeled 8-azido-ATP. Samples were separated on 16% polyacrylamide gel in Tris-tricine-SDS buffer and blotted onto PVDF membrane. Peptides were visualized by immunostaining with antibody specific for the PAP carboxyl end (lanes 3 and 4) and radioactive peptides by autoradiography (lanes 1 and 2). The size and position of CNBr-generated peptides is indicated on the right (see Fig. 3, panels B and C). The carboxylterminal peptides (C9) of wild type (a) and ⌬10PAP (b) are indicated with arrows. the cleavage reaction (32), again suggesting that two factors may contact PAP at the same time.

FIG. 6. CNBr digest of wild type PAP (lane
Recent studies are beginning to clarify the association of the yeast PAP with equivalent factors. For example, application of the two-hybrid system identified FIP1 as a protein that interacts with the yeast PAP (8). FIP1 is a component of PF I (8), a factor that works in conjunction with CF I, the yeast analog of CPSF, to catalyze specific poly(A) addition (3). It appears that FIP1 links PAP to RNA14, a subunit of CF I. Depletion of PAP from yeast-processing extracts with PAP-specific antibody leads to a loss of both cleavage and poly(A) addition activity (21,33). Both activities can be restored if only CF I is added to the depleted extracts, suggesting that a direct interaction of CF I with PAP may also exist. It will be interesting to determine which protein(s) interact with PAP through the amino-terminal domain identified in this study.