PNUTS, a protein phosphatase 1 (PP1) nuclear targeting subunit. Characterization of its PP1- and RNA-binding domains and regulation by phosphorylation.

PNUTS, Phosphatase 1 NUclear Targeting Subunit, is a recently described protein that targets protein phosphatase 1 (PP1) to the nucleus. In the present study, we characterized the biochemical properties of PNUTS. A variety of truncation and site-directed mutants of PNUTS was prepared and expressed either as glutathione S-transferase fusion proteins in Escherichia coli or as FLAG-tagged proteins in 293T cells. A 50-amino acid domain in the center of PNUTS mediated both high affinity PP1 binding and inhibition of PP1 activity. The PP1-binding domain is related to a motif found in several other PP1-binding proteins but is distinct in that Trp replaces Phe. Mutation of the Trp residue essentially abolished the ability of PNUTS to bind to and inhibit PP1. The central PP1-binding domain of PNUTS was an effective substrate for protein kinase A in vitro, and phosphorylation substantially reduced the ability of PNUTS to bind to PP1 in vitro and following stimulation of protein kinase A in intact cells. In vitro RNA binding experiments showed that a C-terminal region including several RGG motifs and a novel repeat domain rich in His and Gly interacted with mRNA and single-stranded DNA. PNUTS exhibited selective binding for poly(A) and poly(G) compared with poly(U) or poly(C) ribonucleotide homopolymers, with specificity being mediated by distinct regions within the domain rich in His and Gly and the domain containing the RGG motifs. Finally, a PNUTS-PP1 complex was isolated from mammalian cell lysates using RNA-conjugated beads. Together, these studies support a role for PNUTS in protein kinase A-regulated targeting of PP1 to specific RNA-associated complexes in the nucleus.

PNUTS, Phosphatase 1 NUclear Targeting Subunit, is a recently described protein that targets protein phosphatase 1 (PP1) to the nucleus. In the present study, we characterized the biochemical properties of PNUTS. A variety of truncation and site-directed mutants of PNUTS was prepared and expressed either as glutathione S-transferase fusion proteins in Escherichia coli or as FLAG-tagged proteins in 293T cells. A 50-amino acid domain in the center of PNUTS mediated both high affinity PP1 binding and inhibition of PP1 activity. The PP1-binding domain is related to a motif found in several other PP1-binding proteins but is distinct in that Trp replaces Phe. Mutation of the Trp residue essentially abolished the ability of PNUTS to bind to and inhibit PP1. The central PP1-binding domain of PNUTS was an effective substrate for protein kinase A in vitro, and phosphorylation substantially reduced the ability of PNUTS to bind to PP1 in vitro and following stimulation of protein kinase A in intact cells. In vitro RNA binding experiments showed that a C-terminal region including several RGG motifs and a novel repeat domain rich in His and Gly interacted with mRNA and singlestranded DNA. PNUTS exhibited selective binding for poly(A) and poly(G) compared with poly(U) or poly(C) ribonucleotide homopolymers, with specificity being mediated by distinct regions within the domain rich in His and Gly and the domain containing the RGG motifs. Finally, a PNUTS-PP1 complex was isolated from mammalian cell lysates using RNA-conjugated beads. Together, these studies support a role for PNUTS in protein kinase A-regulated targeting of PP1 to specific RNA-associated complexes in the nucleus.
Protein phosphatase 1 (PP1) 1 is a multifunctional serine/ threonine phosphatase that plays a key role in regulation of diverse cellular processes, including gene expression, muscle contraction, cell cycle progression, glycogen metabolism, and neurotransmission (1)(2)(3). The catalytic subunit of PP1, which exists as four different isoforms (PP1␣, -␥1, -␥2, and -␦), is widely distributed in various subcellular compartments (1,4) where it is regulated by association with a growing number of identified regulatory proteins. These regulatory proteins include several heat-stable inhibitors, such as inhibitor-1, its neuronal homologue DARPP-32, and inhibitor-2, which are controlled by phosphorylation (1,2). PP1 is also regulated by a family of proteins, termed targeting subunits, that direct the catalytic subunit of PP1 to specific subcellular locations and also influence the specificity of the enzyme at these sites (1,2,5). For example, the glycogen-binding proteins, G M and G L , target PP1 to glycogen and enhance the activity of PP1 toward glycogen synthase. Similarly, the myofibril-binding protein, M 110 , mediates the association of PP1 with the myofibrils of skeletal muscle and smooth muscle and stimulates the activity of PP1 toward phosphorylated myosin light chain (6). Despite little overall amino acid sequence homology, several studies have identified a common docking motif in many of the targeting proteins that binds to a defined region of PP1 removed from the active site of the enzyme (7,8).
A variety of studies have suggested an important role for PP1 in the nucleus (9). PP1 interacts with the retinoblastoma protein p110Rb (10) and is believed to act as a positive regulator of the interaction of p110Rb with the transcription factor, E2F (10,11). PP1 is likely to play an important role in dephosphorylation of the transcription factors CREB and Sp-1 (12)(13)(14). PP1 interacts with a protein termed NIPP1 that functions as a splicing factor at a late stage of spiceosome assembly (15,16). Other studies (17,18) have suggested a role for PP1 in modulation of mammalian splicesome assembly and in the subcellular distribution of pre-mRNA splicing factors. During the cell cycle, biochemical and genetic studies have shown that PP1 activity is regulated by phosphorylation (19,20) and that the enzyme plays a key role in the mitotic transition by dephosphorylating various nuclear phosphoproteins that are essential for driving structural reorganization of the nuclear envelope, spindle apparatus, and chromosomal DNA (21)(22)(23)(24)(25). PP1 also interacts with other nuclear proteins including the p53-binding protein, p53BP (26), Hox11 (27), and with sds22, a protein implicated in chromosome stability (28,29).
We and others have reported recently (30,31) the cloning and initial characterization of a novel nuclear PP1-binding protein named PNUTS (Phosphatase 1 NUclear Targeting Subunit) or p99. PNUTS exhibits a discrete nuclear compartmentalization and is found in a stable complex with PP1 in mammalian cell lysates. Recombinant PNUTS potently inhibits the catalytic activity of PP1 toward exogenous substrate in vitro. Primary sequence analysis indicates that the C terminus of PNUTS contains several closely spaced RGG sequences, motifs that are often found in RNA-binding proteins (32). PNUTS also contains a novel region of repetitive amino acid sequence that is rich in His and Gly, and a putative Zn 2ϩ finger domain with the signature CX 8 CX 5 CX 3 H. In the present study, we have characterized further the biochemical properties of PNUTS. PNUTS contains a short ϳ50-amino acid central region that contains closely associated PP1-binding domains and inhibitory domains. Moreover, the interaction of PNUTS with PP1 is regulated by phosphorylation within the binding domain. We have also found that PNUTS binds to homopolymeric RNA with high selectivity for poly(A) and poly(G) via the RGG motifs and the novel region rich in His and Gly. These studies support the conclusion that PNUTS may mediate the reversible association of PP1 with specific RNAs in the nucleus of mammalian cells.
pcDNA1/Neo-FLAG Plasmid Construction and Co-immunoprecipitation-DNA fragments of PNUTS were amplified by Pfu polymerase; PCR products were digested by SalI/NotI restriction enzymes and subcloned into pcDNA1Neo (Invitrogen) (encoding a FLAG epitope with a 5Ј SalI site in-frame with the FLAG sequence). Internal deletion mutants were prepared from the ligation of two individual PCR fragments. HEK293T cells grown in Dulbecco's modified Eagle's culture medium (10% fetal bovine serum) were transiently transfected with different plasmids (10 g) using the calcium phosphate method. Cells were washed with fresh medium 5 h after transfection and were cultured for 16 h. Cells were washed with PBS, harvested, and resuspended in lysis buffer (50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 0.6 mM PMSF, 20 g/ml leupeptin and antipain, 10 g/ml pepstatin A and chymostatin, 0.5% Nonidet P-40). Lysates were briefly sonicated and then centrifuged at 15,000 ϫ g for 20 min. Protein concentration of supernatants was determined using the BCA assay (Pierce). Lysates (1 mg) were incubated with anti-FLAG affinity beads (Sigma) for 2 h at 4°C. Immunocomplexes were washed with lysis buffer and eluted from beads by boiling in SDS sample buffer. Samples were analyzed by SDS-PAGE (12% polyacrylamide), and proteins were transferred to polyvinylidene difluoride membrane (Immobilon-P, Millipore) by electroblotting (200 mA, overnight). Blots were incubated with either an anti-FLAG antibody or an anti-PP1␣ antibody, followed by horseradish peroxidase-conjugated secondary antibody. Proteins were visualized using ECL.
Preparation of GST Fusion Proteins in Bacteria-The cDNAs encoding various deletion mutants of PNUTS were amplified by PCR with primers containing 5Ј EcoRI and 3Ј NotI restriction sites. PCR fragments were digested with EcoRI and NotI and subcloned into the pGEX-5x-1 expression vector. Plasmids were transformed into Escherichia coli (BL21 DE3), and bacteria were cultured in LB media in the presence of 50 mg/ml ampicillin to an A 600 value of 0.6 -0.8 at 37°C. Expression of GST fusion proteins was induced by addition of 0.3 mM isopropyl-1-thio-␤-D-galactopyranoside at 30°C for 3 h. Cells were collected by centrifugation, resuspended in lysis buffer (20% sucrose, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM EGTA, 0.2 mM PMSF, 150 mM NaCl, and 1% Triton X-100), and lysed by sonication on ice. Lysates were centrifuged at 12,000 ϫ g for 20 min, and supernatants were loaded onto a column containing glutathione-agarose beads (Sigma) and washed extensively with PBS. Fusion proteins were eluted with 5 mM glutathione, 50 mM Tris-HCl, pH 8.0. Protein purity was analyzed by SDS-PAGE and stained with Coomassie Brilliant Blue, and protein concentration was determined using the BCA assay.
Site Protein Phosphatase Assay-Purified rabbit muscle PP1 was kindly provided by Dr. Hsien-bin Huang. PP1 was assayed using [ 32 P]phosphorylase a as substrate essentially as described (34). Briefly, assays (final volume 30 l) contained 50 mM Tris-HCl, 0.15 mM EGTA, 15 mM 2-mercaptoethanol, 0.01% (w/w) Brij 35, 0.3 mg/ml bovine serum albumin, 5 mM caffeine, 10 M [ 32 P]phosphorylase a, various concentrations of GST fusion proteins, and Ͻ0.1 unit/ml of PP1␣ (ϳ50 pM). All components except [ 32 P]phosphorylase a were preincubated at 30°C for 2 min. Assays were initiated by addition [ 32 P]phosphorylase a; incubations were performed at 30°C for 10 min and then terminated by the addition of 100 l of 10% (w/v) trichloroacetic acid. Samples were centrifuged for 3 min, and 32 P in the supernatant was measured by Cerenkov counting.
PP1 Overlay Assay-PP1 overlay assays were carried out essentially as described (37). Briefly, proteins were separated by SDS-PAGE and transferred to nitrocellulose filters. Filters were incubated with a buffer containing 10 mM Tris-HCl (pH 7.4), 2% (w/v) dried milk, and 0.1% Tween 20. Filters were washed with PBS containing 0.2% Nonidet P-40 and then incubated with PBS/Nonidet P-40 containing 0.1 g/ml recombinant PP1 and 100 nM microcystin (to inhibit potential dephosphorylation of PNUTS) for 2 h at 4°C. Filters were washed with PBS/Nonidet P-40, and bound PP1 was detected by immunoblotting using PP1␣ antibody.
Metabolic Labeling-PC12 cells were incubated in 200 Ci/ml of [ 32 P]inorganic phosphate (PerkinElmer Life Sciences) and phosphatefree, serum-free Dulbecco's modified Eagle's medium for 2 h. After metabolic labeling, cells were washed three times with PBS, harvested by lysis in an immunoprecipitation buffer (50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 1 mM Na 3 VO 4 , 1 mM PMSF, 0.5% Nonidet P-40, 20 mM NaF), and clarified by centrifugation at 14,000 ϫ g for 10 min. Unlabeled HEK293 cells, incubated in the absence or presence of 8-Br-cAMP, were lysed in the same way. Lysates were incubated with control IgG or anti-PNUTS antibody in the absence or presence of antigen peptide, followed by the addition of protein Aagarose beads. Immunoprecipitated proteins were eluted from the protein A-agarose with SDS sample buffer and separated by SDS-PAGE (10% acrylamide). Gels were dried, and proteins were visualized by either autoradiography ( 32 P-labeled samples) or by immunoblotting with PNUTS or PP1␣ antibodies.
Phosphorylation of GST-PNUTS Fusion Proteins by PKA-Phosphorylation reactions were performed using the protein of interest and the catalytic subunit of PKA (40 g/ml) in an incubation mixture using 50 mM HEPES, pH 7.4, 10 mM MgCl 2 , 1 mM EGTA at 30°C. Reactions were initiated by the addition of ATP (50 M) in the absence or presence of [␥-32 P]ATP. Reactions were terminated at various time points by dilution of the reaction mixture into SDS-PAGE sample buffer, and the stoichiometry of phosphorylation was assessed after SDS-PAGE and autoradiography.
RNA Gel Retardation Assay-GST fusion proteins (10 -40 ng) were incubated with ϳ5 ng of 32 P-radiolabeled ␤-globin mRNA in binding buffer (total volume 40 l) containing 10 mM Tris-HCl, pH 7.4, 100 mM KCl, 1 mM MgCl 2 , and 40 ng of RNasin (Roche Molecular Biochemicals). Samples were incubated for 10 min on ice. After addition of 5 l of electrophoresis buffer containing 10% glycerol and 0.01% bromphenol blue, reaction mixtures were separated on a 4% native polyacrylamide gel for 2-3 h at 20 mA (about 6 V/cm) at room temperature. After electrophoresis, gels were dried and exposed to Hyperfilm MP.
In Vitro Transcription and Translation-cDNA encoding full-length PNUTS was subcloned into the pGEM T vector (Promega); the plasmid was linearized by digestion with ScaI and used as a template for RNA synthesis with T7 polymerase. The resulting RNAs were translated in rabbit reticulocyte lysate in the presence of [ 35 S]methionine according to the manufacturer's suggested conditions (Amersham Biosciences). Translated protein was analyzed SDS-PAGE (10% acrylamide) and autoradiography.
Ribonucleotide Homopolymer and ssDNA Binding Assays-Assays were initiated by addition of 25 l of ribonucleotide homopolymeragarose or ssDNA-agarose into binding buffer with various GST fusion proteins (total volume of 125 l of 10 mM HEPES, pH 7.4, 2 mM MgCl 2 , 0.1% Triton X-100, 3 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 0.05 mM EDTA, 0.1 M NaCl, or other salt at the indicated concentrations, 1 mM PMSF) at 4°C. After 30 min of incubation, beads were centrifuged and washed 6 times with 0.5 ml of binding buffer, and proteins were eluted with SDS sample buffer. Samples were analyzed by SDS-PAGE (10% polyacrylamide) and immunoblotting using an anti-GST antibody.
Binding of in vitro translated protein to ribonucleotide homopolymeragarose was performed essentially as described (39). An equivalent of 10 5 cpm of 35 S-labeled in vitro translation product and 25 l of homopolymer RNA beads were incubated at 4°C for 10 min in a total volume of 0.25 ml of binding buffer (10 mM Tris-HCl, pH 7.4, 2.5 mM MgCl 2 , 0.5% Triton X-100, at the salt concentrations indicated). The beads were pelleted by a brief centrifugation and washed 5 times with 0.5 ml of binding buffer, and protein was eluted with SDS sample buffer. Samples were analyzed by SDS-PAGE (10% acrylamide) and autoradiography.
Ribonucleotide Homopolymer Pull-down Assay and Poly(G)-Agarose Column Chromatography-HEK293T cells were lysed by brief sonication in buffer (50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, 150 mM NaCl, 0.6 mM PMSF, 20 g/ml leupeptin and antipain, 10 g/ml pepstatin A and chymostatin, 0.5% Nonidet P-40). Lysates were centrifuged at 15,000 ϫ g for 20 min. The supernatant (400 l, 500 g of protein) was mixed with 50 l of homopolymer RNA beads and incubated at 4°C for 1 h. The beads were centrifuged and washed 5 times with 0.5 ml of lysis buffer containing 0.25 M NaCl, and proteins were eluted in 50 l of SDS sample buffer. Samples were analyzed by SDS-PAGE (10% acrylamide) and transferred to polyvinylidene difluoride membrane (Immobilon-P, Millipore). Proteins were detected by immunoblotting using antibodies against PNUTS, PP1␣, and hnRNP C, and detection was by the ECL method.
For poly(G) column chromatography, 2 ml of the supernatants (2 mg of protein) were mixed with 200 l of poly(G) and incubated at 4°C for 1 h. Beads were washed once with 1 ml of lysis buffer and loaded onto a column, which was then washed extensively with lysis buffer. Bound proteins were eluted with a linear 0.1-2.0 M NaCl gradient in lysis buffer. Fractions (0.5 ml) were collected, and proteins in each fraction were analyzed by SDS-PAGE and immunoblotting with antibodies against PNUTS, PP1␣, PP1␥, and hnRNP C.

Characterization of Binding of PP1 to PNUTS and Effect on
Cell Viability-Previous studies (30) have shown that fulllength PNUTS and PNUTS-(309 -872) (the protein product of clone 14 originally isolated in the yeast two-hybrid screen) were able to interact with PP1␣. To characterize further the interaction of PP1 and PNUTS protein, various FLAG-tagged PNUTS fragments were expressed in HEK293T cells, proteins were immunoprecipitated with anti-FLAG antibody, and PP1␣ was detected in the immunoprecipitates (Fig. 1). All the The PP1-binding domain is localized approximately in the middle of the molecule (black shading), and a putative motif (KKKRK) for nuclear localization is at residues 157-161 (gray shading). The C terminus of PNUTS contains three distinct domains potentially involved in RNA binding (detailed in Fig. 6A). 293T cells were transiently transfected with various FLAG-tagged PNUTS mutants (N-and C-terminal amino acid number is indicated within each rectangle). Cell toxicity was estimated by counting viable cell numbers 24 h after transfection. PP1 binding was measured as shown in B. For cell toxicity, Ϫ indicates no effect; ϩ and ϩϩ indicate slight and potent toxicity, respectively. B, cells were lysed and anti-FLAG antibody was used to immunoprecipitate each PNUTS mutant. Immunoprecipitated (IP) samples were analyzed by immunoblotting (WB) using a PP1␣ antibody. Lane 1 shows untransfected cells; lanes 2-15 correspond to the mutants shown in A. Qualitative analysis of the amount of coprecipitated PP1 is shown in the 1st column in A. Results are representative of three experiments. PNUTS fragments were expressed at equivalent levels (data not shown). PP1 was bound to fragments containing residues 309 -872, 309 -589, 309 -433, 357-537, 357-486, 357-433, 143-433, and a fragment between residues 143 and 872 with an internal deletion of residues 434 -589. A low level of PP1 was found to bind to PNUTS-(309 -401), but no binding was detected for PNUTS-(404 -537), PNUTS-(590 -872), and PNUTS-(724 -872) or for two fragments with an internal deletion of residues 254 -589. Together, these results indicate that PP1 binds to PNUTS between residues 357 and 433, with the C-terminal boundary of the binding site close to residue 401.
The viability of the 293T cells was significantly affected by transfection with many of the PNUTS plasmids. Notably, all of the PNUTS fragments that bound strongly to PP1 caused cell death (Fig. 1). Neither a nuclear localization signal (a KKKRK motif at residues 157-161) nor the C-terminal region (residues 590 -872) was required for cell toxicity. Expression of PNUTS-(357-486) caused cell toxicity more effectively than PNUTS-(357-433) or PNUTS-(309 -433), suggesting that residues 434 -486 contributed to regulation of cell viability but were not required for PP1 binding.
PNUTS Contains Closely Associated PP1 Binding and Inhibitory Domains-Our previous studies showed that a GST fusion protein containing residues 309 -872 potently inhibited the phosphorylase phosphatase activity of PP1 in vitro (30). The results shown in Fig. 1 and other preliminary studies indicated that both PP1 binding and inhibition appeared to be contained within residues 382-537 of PNUTS. Various GST fusion proteins containing residues 382-537 of PNUTS were expressed in bacteria, purified, and incubated with a 293T cell extract.  Table I, and data not shown). Deletion of the PP1 docking motif (GST-PNUTS-(404 -537)) resulted in loss of PP1 inhibition. Surprisingly, deletion of only 6 residues from the C terminus of GST-PNUTS-(382-450) almost completely abolished PP1 inhibitory activity (GST-PNUTS-(382-444) at 1 M reduced PP1 activity by less than 5%, see Fig. 3A). Consistent with the binding studies, mutation of Trp 401 resulted in an almost complete loss of PP1 inhibition (an increase in IC 50 of more than 4 orders of magnitude), whereas mutation of Val 399 increased the IC 50 value by almost 2 orders of magnitude (Table I). Mutation of Lys 397 increased the IC 50 value by ϳ5fold, whereas mutation of Arg 396 had no effect. Together, these results suggest that inhibition of PP1 by PNUTS is mediated by two small regions between residues 382 and 450. Residues 399 -402 include Val 399 and Trp 401 that are necessary for binding to PP1, and residues 445-450 are necessary for inhibition of PP1.  Table I). However, mutation of Trp 401 (GST-PNUTS-(382-433;W401A)) resulted in a PNUTS fragment that was unable to antagonize the inhibitory action of thiophospho-DARPP-32-(1-39) (Fig. 3B). A shorter synthetic peptide encompassing residues 392-415 of PNUTS was also very effective at antagonizing the inhibitory action of phospho-DARPP-32 ( Fig. 3C and Table I) (note here that full-length thiophospho-DARPP-32 and higher concentrations of competing peptide were used). However, a shorter peptide PNUTS-(392-408) was much less effective in antagonizing the actions of thiophospho-DARPP-32, suggesting that residues 409 -415 (Glu-Tyr-Phe-Tyr-Phe-Glu-Leu) contribute to the binding of PNUTS to PP1. Similar results were obtained when PP1 was inhibited using various concentrations of spinophilin or inhibitor-2 (data not shown). As expected, mutation of Trp 401 in PNUTS-(392-408) rendered the peptide completely ineffective in competing with thiophospho-DARPP-32 (Fig. 3C). However, surprisingly, PNUTS-(392-415;W401Y) was very effective in antagonizing the actions of thiophospho-DARPP-32.
In the studies of the various PNUTS peptides as antagonists of the actions of PP1 inhibitors, we noted an unusual property of PNUTS-(392-415). PNUTS-(392-415) alone was able to activate consistently PP1 activity by ϳ30 -40% (Fig. 3D). PNUTS-(392-408) was much less effective as an activator, and PNUTS-(392-408;W401A) had no effect on PP1 activity. Consistent with the ability of the PNUTS peptide to activate PP1 via the C-terminal docking site where Trp 401 binds, PNUTS-(392-415) had no effect on a PP1/PP2A chimeric enzyme in which the C terminus of PP1 (residues 274 -330) was replaced by the equivalent residues of PP2A (data not shown) (see also Ref 40).
Phosphorylation of PNUTS Regulates Its Interaction with PP1-Examination of the amino acid sequence of residues 382-450 of PNUTS revealed the presence of several consensus sites for phosphorylation by PKA. In addition, in preliminary studies using PC12 cells metabolically labeled with [ 32 P]phosphate, full-length PNUTS was found to be phosphorylated (Fig. 4A).
To determine whether phosphorylation of PNUTS by PKA might affect the interaction with PP1, GST-PNUTS-(382-433) was phosphorylated by PKA and [ 32 P]ATP for various times (Fig. 5A, top panel). Maximal phosphorylation was reached between 40 and 70 min (a maximal stoichiometry of ϳ1 mol/ mol was determined). The phosphorylated samples were separated by SDS-PAGE, transferred to Immobilon-P membrane, and incubated with PP1␣. The amount of PP1 bound to GST-PNUTS-(382-433) decreased in parallel to the increase in phosphorylation (Fig. 5A, middle panel). These results indicate that phosphorylation of PNUTS by PKA within the site of PP1 binding blocks the association of PNUTS with PP1. We further investigated whether PKA regulates interaction of PNUTS with PP1 in intact cells. In intact HEK293 cells incubated in the presence of forskolin, PNUTS was phosphorylated largely on threonine (Fig. 4D). In addition, phosphopeptide mapping studies indicated that PNUTS was phosphorylated at the same site as that phosphorylated by PKA within GST-PNUTS-(382-433) (Fig. 4C). In parallel studies, HEK293 cells were incubated in the absence or presence of 8-Br-cAMP (500 M for 10 min), and the interaction between PP1␣ and PNUTS was examined following co-immunoprecipitation (Fig. 5B). By using antibodies specific for either PP1␣ or PNUTS, the interaction between PNUTS and PP1 was shown to be significantly reduced by activation of PKA with 8-Br-cAMP. Together these results suggest that phosphorylation of PNUTS by PKA negatively regulates the interaction of PNUTS with PP1.
The C Terminus of PNUTS Binds to mRNA and Singlestranded DNA-PNUTS contains multiple closely spaced repeats of the amino acid sequence, RGG, a motif often found in RNA-binding proteins (32) (Fig. 6A). The RGG motifs are followed by a region with several imperfect repeats of a sequence rich in histidine and glycine. The extreme C-terminal region then contains a putative zinc finger domain. These features together with its nuclear localization suggested that PNUTS might interact with nucleic acids. To initially examine the interaction with RNA, a gel retardation assay was performed with various GST fusion proteins and 32 P-labeled ␤-globin mRNA. GST-PNUTS-(617-872) retarded the mobility of ␤-globin mRNA, but deletion of the putative zinc finger domain and most of the region rich in histidine and glycine (GST-PNUTS-(617-762)) resulted in a marked decrease in RNA binding (Fig.  6B). Moreover, partial or complete deletion of the RGG motif (GST-PNUTS-(617-726) or GST-PNUTS-(404 -662)) led to an almost complete loss of RNA binding. We next examined the abilities of PNUTS fragments to bind to single-stranded DNA (ssDNA). Both GST-PNUTS-(617-872) and GST-PNUTS-(617-837) bound effectively to ssDNA (Fig. 6C). However, GST-PNUTS-(617-762), GST-PNUTS-(617-726), or GST-PNUTS-(404 -662) did not bind to ssDNA. These results suggest that the C-terminal region of PNUTS, including the histidine/glycine-rich motifs, is necessary for binding to both RNA and ssDNA and that the RGG motifs may contribute to a lesser extent to the binding to RNA.
Selective Binding of a PNUTS-PP1 Complex to Different Ribonucleotide Homopolymers-The entire rat PNUTS cDNA was transcribed in vitro using T7 polymerase and translated in a rabbit reticulocyte system in the presence of [ 35 S]methionine. Three radiolabeled bands were detected using SDS-PAGE, the largest corresponding to full-length PNUTS (apparent molecular mass ϳ110 kDa) (Fig. 7A). The other bands are likely to be proteolytic fragments or incomplete translation products. Fulllength PNUTS bound to poly(A) and poly(G) but not to poly(U) or poly(C) (Fig. 7A). As a control, in vitro translated hnRNP K, which is a known poly(C)-specific RNA-binding protein, bound selectively to poly(C) (Fig. 7A).
Because PNUTS is able to interact directly with PP1 as well as with RNA, we hypothesized that the PNUTS-PP1 complex may bind to RNA. To examine this possibility, 293T cell lysates were incubated with poly(A)-, poly(C)-, poly(G)-, and poly(U)agarose beads, and bound proteins were eluted and analyzed by immunoblotting. Consistent with the properties of the recombinant GST-PNUTS fusion proteins, endogenous PNUTS was efficiently precipitated by poly(A) and poly(G) but not by poly(U) and poly(C) (Fig. 7B). Moreover, PP1␣ was found only in the precipitates from poly(A)-or poly(G)-agarose beads. In addition, depletion of PNUTS from 293T cell extracts by immunoprecipitation using anti-PNUTS antibody significantly reduced the amount of PP1 bound to poly(G)-beads (data not shown). As a control, hnRNP C, one of the most abundant heterogeneous nuclear ribonucleoproteins (hnRNPs) and known to bind to pre-mRNA, was detected in all the precipitates.
To characterize further the PNUTS/PP1/RNA interaction, 293T cell lysates were loaded onto a poly(G)-agarose column (equilibrated in a buffer containing 100 mM NaCl), and the bound proteins were eluted with a linear salt gradient (Fig. 8). PNUTS and both PP1␣ and PP1␥ were detected in the same fractions at salt concentrations ranging from 0.5 to 0.7 M. hnRNP C was also detected in fractions that contained PNUTS and PP1 but was also found in fractions eluting at higher ionic strength that did not contain PNUTS/PP1.

Different Subdomains of PNUTS Bind to Poly(A) and
Poly(U)-In order to characterize further the ribonucleotidebinding properties of PNUTS, GST-PNUTS fragments were incubated with poly(A), poly(C), poly(G), and poly(U) at 100 mM NaCl. GST-PNUTS-(617-872) and GST-PNUTS-(617-837) bound with very similar efficiency to both poly(A) and poly(U) (Fig. 9). However, further deletion of the histidine/glycine-rich region (GST-PNUTS-(617-762)) resulted in a preferential decrease in binding to poly(A). Partial or complete deletion of the RGG motifs led to complete loss in binding to the poly(G) ribonucleic acid homopolymer. Increasing the bead volumes of poly(C) and poly(U) up to 4 times compared with poly(A) did not alter the inability of PNUTS to bind to poly(U) and poly(C) (data not shown). DISCUSSION In the present study, we have characterized the properties of the interactions between PNUTS and both PP1 and RNA. The Residues 751-834 (boxed) are rich in histidine and glycine and contain two perfect and one imperfect repeats of 14 residues. Following this, three additional pentamer repeats are found. Residues 844 -863 (underlined) contain a cysteine/histidine-rich putative Zn 2ϩ finger motif with the signature CX 8 CX 5 CX 3 H. B, the interaction of various GST-PNUTS fusion proteins (including those indicated) with ␤-globin mRNA was analyzed using a gel retardation assay. Each GST-PNUTS fusion protein (40 g) was incubated with 32 P-labeled ␤-globin mRNA for 90 min at 37°C. Samples were analyzed using non-denaturing gel electrophoresis and autoradiography. C, the interaction of various GST-PNUTS fusion proteins with ssDNA was measured. Each GST-PNUTS fusion protein (200 ng) was incubated with 25 l of ssDNA-agarose. Bound protein was analyzed by SDS-PAGE and immunoblotted by using anti-GST monoclonal antibody.
results obtained suggest that PNUTS may bind to specific types of RNA implicating PP1 in specific functions within the nucleus. Studies of the interaction with PP1 revealed that PNUTS contains two closely associated subdomains in the center of the protein within ϳ50 amino acids (residues 400 -450). These include a high affinity PP1-binding domain located within residues 397-401 of PNUTS and a distinct inhibitory region located within residues 445-450. The binding domain, Lys 397 -Thr-Val-Thr-Trp 401 , resembles the consensus motif (Arg/Lys-Arg/Lys-Val/Ile-X-Phe) found in many other PP1 regulatory subunits except for the replacement of Phe with Trp. Consistent with residues 397-401 of PNUTS binding to a common docking site in PP1, competition studies indicated that PNUTS peptides as short as residues 392-408 were potent antagonists of the inhibitory actions of thiophospho-DARPP-32. Site-directed mutagenesis studies of PNUTS indicated that the order of importance of amino acids contributing to the association with PP1 is Trp 401 Ͼ Val 399 Ͼ Ͼ Lys 397 , a pattern consistent with our previous studies of DARPP-32, where Phe 11 and Ile 9 play critical roles, with Lys 7 playing a lesser role in the interaction with PP1 (8,41). Therefore, it is likely that the conserved docking motif of PNUTS interacts with the exposed hydrophobic docking site in PP1 in a similar manner to DARPP-32 and other PP1-binding proteins.
A notable feature of the PP1 docking motif in PNUTS is the presence of Trp instead of Phe as the most important binding residue (30). In an analysis of a random peptide library that bound to PP1, Phe and Trp were identified with equal frequency in the interacting peptides (42). However, PNUTS ap-pears to be the first well characterized PP1-binding protein that contains Trp within the PP1-docking motif. The affinity of PNUTS for PP1 is suggested from the PP1 inhibition and competition assays to be very high, presumably being in the low nM range. It is possible that the presence of Trp 401 contributes to the high affinity, but it is clear that parts of PNUTS outside of the minimal PP1-docking motif also play a role. PNUTS-(392-408) was very effective in competing against the inhibitory action of thiophospho-DARPP-32, but addition of residues 409 -415 made PNUTS-(392-415) a much more effective antagonist. In addition, mutation of Trp 401 to Tyr had only a small effect on the antagonist properties of PNUTS-(392-415), and mutation of Phe 11 in DARPP-32 to Trp did not affect its inhibitory potency significantly (41), consistent with the idea that the presence of Phe or Trp as the key docking residue is not critical. The results with PNUTS also support the idea that like DARPP-32, inhibitor-1 and inhibitor-2, targeting proteins are likely to bind to PP1 via multiple subdomains (8,36,41,43,44).
PNUTS is a highly potent inhibitor of PP1, with an IC 50 value of ϳ10 Ϫ10 nM using phosphorylase a as substrate. Deletion mutagenesis studies identified a short sequence, ETARRL (residues 445-450), that was responsible for a greater than 10 4 -fold factor in inhibitory potency. Like DARPP-32 and inhibitor-1, the PP1 docking and inhibitory subdomains are contained in a short stretch of amino acids, but unlike phospho-DARPP-32 (phosphorylated at Thr 34 by PKA), PNUTS does not have to be phosphorylated in this subdomain to be an effective inhibitor. Possibly, residues 445-450 of PNUTS interact with the active site of PP1 in the manner of a pseudosubstrate, or these residues may act to block interactions of phosphorylase at a substrate-binding site that is situated close to but not within the active site of the phosphatase. Surprisingly, in contrast to the highly potent inhibitory actions of residues 382-450 of PNUTS, addition of peptides encompassing just the docking motif resulted in activation of PP1. This effect required the presence of Trp 401 but was more robust when residues 409 -415 were included. The molecular basis for this is not currently known, but may reflect an allosteric effect that stabilizes the substrate-binding site or alters the active site.
The PP1-binding domain of PNUTS contains a number of For analysis of RNA binding to hnRNP K, the binding buffer contained 500 mM NaCl. Samples were analyzed by SDS-PAGE and hnRNP K was detected by autoradiography (right panel). S indicates that in vitro translated 35 S-labeled protein that was loaded directly on the gel as a control. B, protein extract from 293T cells (ϳ500 g) was incubated with 50 l of the indicated ribonucleotide homopolymer (poly(U), -(A), -(G), or -(C)) immobilized to agarose beads at 4°C for 1 h. Bound proteins were analyzed by SDS-PAGE, and various proteins were detected by immunoblotting using the indicated PNUTS, PP1, and hnRNP C antibodies. Cell lysate (25 g, CL) was loaded directly on the gel as a reference. potential consensus sites for phosphorylation. These include a site at Thr 398 within the PP1-docking motif and at Ser 451 close to the inhibitory subdomain. Our results show that a polypeptide encompassing residues 382-433 of PNUTS was efficiently phosphorylated by PKA in vitro, and the resulting phosphorylation decreased its affinity for PP1 in vitro and also in intact cells following stimulation of PKA. Based on analysis of phosphorylation of different PNUTS fragments, and phospho-amino acid analysis, Thr 398 is a likely candidate for phosphorylation by PKA; notably Ser 451 was not phosphorylated in vitro. In the human homologue of PNUTS, Thr 398 is replaced by serine (31). Therefore, these results suggest that the interaction of PNUTS with PP1 may be negatively regulated by phosphorylation in an analogous manner to that observed in studies of a few other PP1-binding proteins. For example, phosphorylation of Ser 67 in the glycogen-binding G M subunit results in the dissociation of PP1-G M complex (5,45). Similarly, NIPP1 has phosphorylation sites for PKA (Ser 199 ) and CK2 (Ser 204 ) that flank the PP1binding motif, and the phosphorylation of either serine impairs PP1 binding and reduces the activity of NIPP1 as a PP1 inhibitor (46,47). We have also shown that phosphorylation of the brain-specific actin-binding protein, neurabin, at Ser 461 by PKA significantly reduces its binding to PP1 (48). These observations together with our present data suggest that reversible phosphorylation of a site at or near the Arg-Lys-Arg/Lys-Val/ Ile-X-Phe/Trp motif of PP1 regulatory proteins may be a common control mechanism that adds to the complexity of PP1 regulation in the nucleus and other cellular compartments.
The present study shows that the C-terminal region, including the seven closely repeated RGG boxes and particularly the histidine/glycine-rich domain, mediates the interaction of PNUTS with RNA and ssDNA. However, the zinc finger region did not appear to play any role in the interaction with RNA or ssDNA. The RGG box was first described as an RNA-binding domain in hnRNP U (32,49). Typically, multiple RGG boxes are found, with as few as 6 and as many as 18 being present in an RNA-binding protein. The RGG boxes are also frequently found adjacent to or interspersed with other RNA-binding motifs such as KH and RBD domains (32). In some cases such as the fragile X mental retardation protein (FMRP) or the TLS protein, the RGG boxes have intrinsic RNA-binding properties (50 -52), whereas in other cases such as nucleolin, the RGG boxes complement the specificity of additional RNA-binding domains (53).
Together, the histidine/glycine region, and to a lesser extent the RGG boxes, appear to be responsible for the high affinity interaction of PNUTS with RNA and may be responsible for specifying the interaction of PNUTS with specific mRNAs in the nucleus. Our studies with RNA homopolymers indicated that PNUTS has high selectivity for binding to poly(A) or poly(G) but not for poly(U) or poly(C). The RGG boxes of PNUTS appeared to be involved in the preferential binding to poly(G). In other cases, the RGG boxes of hnRNP U exhibited highest binding to poly(G) and intermediate binding to poly(U) (49); Nopp44/46 has been shown to bind preferentially to poly(U) (54), and FMRP showed selective binding to poly(G) and poly(U) (54). Interestingly, in the splicing factor, TLS, an N-terminal group of RGG boxes showed selective binding to poly(U), whereas a C-terminal group of RGG boxes showed selective binding to poly(G) (52). The amino acid sequences surrounding the RGG boxes are often rich in aromatic amino acids. However, in PNUTS, the amino acid composition of the RGG domain differs from that of other proteins by the abundance of proline. Therefore, it is likely that the specific amino acid residues surrounding the RGG boxes influence the specificity and avidity of RNA interaction (54). The histidine/glycine region of PNUTS was also necessary for the interaction with native mRNA and ssDNA and perhaps is involved in the preferential binding to poly(A). The histidine/glycine-rich domain is unique to PNUTS, and presumably the repetitive feature of the domain plays some role in the specific interaction with RNA and ssDNA.
The results from our studies indicate that PP1 associates indirectly with RNA through its interaction with PNUTS and suggests a role for PNUTS in anchoring PP1 to RNA-associated complexes. Several lines of evidence indicate that PP1 plays a number of roles in mRNA splicing by reversing the actions of multiple protein kinases (9,55). PP1 can regulate initial steps in splicing by dephosphorylating factors necessary for spliceosome assembly (17,18). PP1 interacts with NIPP1 in nuclear speckles, and NIPP1 appears to play a role in a late stage of spliceosome assembly. However, the spliceosome function of NIPP1 seems not to be related to PP1 binding (16,56), suggesting the possibility that NIPP1 plays an additional role to target PP1 to the splicing machinery. The PP1␦ isoform has also been found to interact with the polypyrimidine tract-binding protein-associated splicing factor, although the function of this interaction is not known (57). Finally, in a recent study PP1 has been implicated in the control of alternative splicing of caspase 9 and bcl-x genes in lung adenocarcinoma cells (58). Initial studies of the localization of PNUTS indicate that the protein exhibits a discrete punctate nucleoplasmic staining pattern with some accumulation in the nucleolus (30,31). This pattern of localization and the fact that PNUTS interacts with RNA is consistent with a specific function for the protein in the nucleus. Most likely this would involve some aspect of RNA processing or transport of RNA within the nucleus and probably would involve the action of PP1 that is bound to PNUTS at these sites.
It is also possible that PNUTS serves to target PP1 to the nucleus for functions of the phosphatase in addition to, or instead of, regulation of RNA processing. Recent studies (59) have highlighted the close physical association of the transcriptional and splicing machinery. Moreover, although a subpopulation of PP1 is localized in the nucleus, recent studies (60) indicate that this localization is dynamic, and PP1 can rapidly move between subnuclear compartments. Within the nucleus PP1 plays an important role in the dephosphorylation of transcription factors such as CREB and Sp1 (9,12,61). PP1 dephosphorylates Ser 133 of CREB which is phosphorylated by a number of kinases, including PKA and multiple Ca 2ϩ -dependent kinases (12). During de-differentiation associated with liver regeneration, PP1 was also identified as the enzyme that dephosphorylated Sp1, reversing the action of CK2 (61). PP1 is also an important phosphatase involved in regulation of the retinoblastoma protein, pRb (62). PP1 can bind to pRb and selectively dephosphorylate specific sites of pRb and appears to be the phosphatase responsible for dephosphorylation of pRb at the time of mitotic exit. Moreover, biochemical studies have identified a high molecular complex that dephosphorylates pRb (63) and that contains PP1 and a 110-kDa protein that appears to be identical to PNUTS (62). Thus, PNUTS may not only serve to target PP1 to the nucleus but to influence its specificity toward nuclear substrates.