PIR1, a Novel Phosphatase That Exhibits High Affinity to RNA·Ribonucleoprotein Complexes*

Protein tyrosine phosphatases are involved in the regulation of important cellular processes such as signal transduction, cell cycle progression, and tumor suppression. Here we report the cloning and characterization of PIR1, a novel member in the dual-specificity phosphatase subfamily of the protein tyrosine phosphatases. PIR1 also contains two stretches of arginine-rich sequences. We have shown that the recombinant PIR1 protein possessed an intrinsic phosphatase activity on phosphotyrosine-containing substrate. A unique feature of this phosphatase is that it binds directly to RNAin vitro with high affinity. In addition, we have found that PIR1 interacted with splicing factors 9G8 and SRp30C, possibly through an RNA intermediate during a yeast two-hybrid screen. PIR1 exhibited a nuclear-staining pattern that was sensitive to RNase A, but not to DNase I, suggesting that PIR1 in the cells are associated with RNA and/or ribonucleoprotein particles. Furthermore, a fraction of PIR1 showed a speckle-staining pattern that superimposed with that of the splicing factor, SC35. Taken together, our data suggest that PIR1 is a novel phosphatase that may participate in nuclear mRNA metabolism.

Protein tyrosine phosphatases, in conjunction with protein tyrosine kinases, regulate the levels of protein tyrosine phosphorylation important for cell growth, differentiation, or transformation (1,2). Protein tyrosine phosphatases (PTP) 1 can be grouped as classic PTPs (including receptor-like PTPs and cytoplasmic PTPs), dual-specificity phosphatases, and low molecular weight (acid) phosphatases (2). Both classic PTPs and dual-specificity phosphatases contain a conserved signature motif (HCXXGXXRXG), which constitutes the active site in the phosphatase catalytic domain (3). The conserved cysteinyl residue in this motif is required for the formation of a thiophosphate intermediate during the phosphate transfer reaction (3).
Dual-specificity phosphatases, a subfamily of protein tyrosine phosphatases, play important roles in signal transduction, cell cycle regulation, and tumor suppression. Although dual-specificity phosphatases contain little primary sequence homology to classic PTPs, they share a similar structural folding, especially at the catalytic site, with classic PTPs (3). Some members of this subfamily of enzymes have been shown to be able to dephosphorylate both phosphotyrosine and phosphoserine/phosphothreonine. One well known member of the dualspecificity phosphatases is MKP-1/CL100, a highly selective phosphatase that dephosphorylates and inactivates mitogenactivated protein kinases (4,5). Another example is Cdc25 (6), which dephosphorylates the inhibitory phosphotyrosine and phosphothreonine residues in Cdc2, a cyclin-dependent kinase required for G 2 to M phase transition during cell cycle progression. More recently, PTEN/MMAC1/TEP1, a novel phosphatase that is encoded by a tumor suppressor locus on chromosome 10q23 and its mRNA level is regulated by transforming growth factor ␤ (7-10), has been added to this subfamily of protein tyrosine phosphatases.
Here we report the cloning and characterization of a novel phosphatase that structurally belongs to the dual-specificity phosphatase subfamily. Interestingly, this novel enzyme can bind to RNA in vitro, and associate with RNA and/or ribonucleoprotein (RNP) complexes in vivo. We have therefore named this novel enzyme PIR1, as phosphatase that interacts with RNA/RNP complex 1.

MATERIALS AND METHODS
cDNA Cloning-The ML1 ZAPII cDNA library (9) was screened using the EST clone H60626 as a probe. Twenty positive clones were identified and excised with ExAssist helper phage (Stratagene). By restriction mapping, clone 14 was shown to contain the longest insert. This clone was subjected to DNA sequencing on both strands. The region containing the putative open reading frame was amplified by the polymerase chain reaction (PCR) with a BamHI site added at the ends of both primers. The 1-kilobase BamHI fragment was subcloned into pCGT (11), pGBT9 (CLONTECH), and pAcG1 (PharMingen) and sequenced again. Site-directed mutagenesis was carried out to change cysteine 152 to serine (C 3 S) using the QuikChange method (Stratagene). The presence of the mutation was confirmed by DNA sequencing. T epitope-tagged PIR1 or PIR1(CS) was amplified from pCGT-PIR1 or pCGT-PIR1(CS) template by PCR with BglII linker added at the ends of the primers and subcloned into pVL1393 (PharMingen).
Cell Lines, Northern Blot, and Western Blot Analyses-Cells culture and Northern blot analysis were performed as described previously (9). For Western blot analysis, cell lysates were resolved on 10% SDS-PAGE, and proteins were blotted onto nitrocellulose membrane. Anti-T epitope antibody (Novagen) and 9E10 (Oncogene) were applied in TTBS (20 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20) with 5% nonfat milk. Blots were then washed extensively with TTBS and then incubated with appropriate secondary antibody conjugated with horseradish peroxidase. The immunoreactive proteins were detected with enhanced chemiluminescence (ECL; NEN Life Science Products  1 The abbreviations used are: PTP, protein tyrosine phosphatase(s); RNP, ribonucleoprotein; PCR, polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis; WT, wild type; GST, glutathione Stransferase; Pipes, 1,4-piperazinediethanesulfonic acid.
10 g/ml leupeptin, 10 g/ml aprotinin, 1 mM benzamidine) and lysed by Dounce homogenization. Lysates were centrifuged at 13,000 rpm for 20 min to remove the insoluble cell debris. The supernatant was then supplemented with either 100 mM KCl (for single-stranded DNA binding assay) or 140 mM KCl (for phosphatase activity assay) and 0.1% Tween 20.
Protein Tyrosine Phosphatase Activity Assay-The clarified Sf9 cell lysates containing glutathione S-transferase(GST)-PIR1 or GST-PIR1(CS) fusion protein (1 ml) were incubated with 100 l of glutathione-Sepharose beads (Amersham Pharmacia Biotech). After 2 h incubation at 4°C, beads were collected and washed extensively with buffer A plus 140 mM KCl, and then assayed for phosphatase activity using phosphotyrosyl-containing poly(Glu 4 Tyr 1 ) as substrate. Poly(Glu 4 Tyr 1 ) (Sigma) was phosphorylated with the kinase domain of the insulin receptor (BIRK), and the protein tyrosine phosphatase activity assays were carried out as described previously (9).
Single-stranded DNA Binding Assay-Cell extracts (400 l) were incubated with 150 l single-stranded DNA agarose (Life Technologies, Inc.) in buffer B (20 mM Tris, pH 7.4, 100 mM KCl, 5 mM dithiothreitol, 0.1% Tween 20, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM benzamidine) for 2 h at 4°C. The beads were washed 4 times with 1 ml of buffer B, and the proteins were then eluted sequentially with 400 l each of 0.25 M, 0.5 M, 1 M NaCl in 20 mM Tris buffer (pH 7.4). The proteins retained on the beads after 1 M NaCl elution were released by boiling in 800 l of Laemmli sample buffer. Equal volumes of input, flow-through, eluates at different salts, and bound proteins at 1 M salt were analyzed by SDS-PAGE and Western blot analysis.
Southwestern and Northwestern Blot Analyses-Baculovirus-expressed GST-PIR1 (WT or CS) or GST-MKP-1 were affinity purified by binding to glutathione-Sepharose beads as described above. The bound proteins were analyzed by 10% SDS-PAGE and then transferred onto nitrocellulose filter. The filters were washed with TTBS then phosphate-buffered saline. Proteins on the filter were denatured with 6 M guanidine hydrochloride in TBB solution (20 mM Tris pH 7.4, 5 mM MgCl 2 , and 75 mM KCl) and renatured by sequential incubation (10 min each) in 3, 1.5, 0.75, 0.375, 0.187, and 0 M guanidine hydrochloride in TBB solution. Filters were blocked with nucleic acid binding buffer (Tris pH 7.5, 1 mM EDTA, 50 mM NaCl, and 1ϫ Denhardt's solution) for 30 min. The filters were incubated with the radiolabeled DNA or RNA probes for 1 h at room temperature, then washed with the nucleic acid binding buffer (three times, 5 min each) before autoradiography. A single-stranded DNA probe was prepared with [␣-32 P]dATP using the random primer-labeling kit (Stratagene) with sheared single-stranded salmon sperm DNA as the template. To compare the binding preference of PIR1 for single-stranded DNA or RNA by Southwestern and Northwestern blot analyses, cDNAs encoding PTEN/MMAC1/TEP1, cyclin A, p45 SKP2 , p27 KIP1 , and p21 WAF1 were each PCR amplified with T3 and T7 primers from the corresponding pBluescript-based plasmid. The PCR products were gel-purified, pooled, and used as templates. The 32 Plabeled DNA probe was prepared by asymmetric PCR using Taq DNA polymerase, T3 primer, and [␣-32 P]dATP for 40 cycles. The 32 P-labeled RNA probe was generated with T3 RNA polymerase using [␣-32 P]ATP and the Riboprobe kit (Promega) for 1 h at 30°C. Both [␣-32 P]dATP and [␣-32 P]ATP were used at 4 ϫ 10 5 cpm/pmol in the labeling reactions. The DNA and RNA probes were each purified with G25 sizing column (Boehringer Mannheim) and adjusted to 10 6 cpm/ml for filter binding assays.
Yeast Two-hybrid Screen-YRG2 (Stratagene) cells were transformed with 400 g of HeLa pGADGH library DNA (12) and 400 g of pGBT9-PIR1 DNA. Transformants were selected for histidine, leucine, and tryptophan prototrophs according to Stratagene's protocol. Histidine prototroph colonies (His ϩ ) were then tested by the ␤-galactosidase assay. About 100 His ϩ ␤-galactosidase ϩ colonies were obtained, and 20 of them were further analyzed. Plasmids were recovered from the yeast cells and the pGADGH constructs were selected in the Escherichia coli MH4 strain (12) based on their ability to confer leucine prototroph. pGADGH plasmids were then re-transformed into YRG2 strain together with either pGBT9-PIR1 or pGBT9-CDK6 plasmid. Plasmids encoding proteins that showed specific interaction with PIR1 but not with CDK6 were sequenced. The positive cDNAs that were fused inframe with the upstream Gal4 activation domain include genes encoding splicing factor 9G8 or SRp30C.
Immunostaining-HeLa cells were transfected with 1 g of pCGT-PIR1 and 19 g of pUC18 carrier DNA by the calcium phosphate method. Cells were washed with phosphate-buffered saline 36 h posttransfection and permeabilized with CSK buffer (10 mM Pipes, pH 6.8, 300 mM sucrose, 3 mM MgCl 2 , 1 mM EGTA, 0.5% Triton X-100, 100 mM NaCl, 1 mM phenylmethylsulfonyl fluoride) (13). For the nuclease sen-sitivity experiments, cells were incubated at 27°C for 1 h with CSK buffer supplemented with either (i) 400 units/ml DNase I, 20 mM vanadyl-ribonucleoside complex and 50 mM NaCl; (ii) 0.1 mg/ml RNase A; or (iii) 20 mM vanadyl-ribonucleoside complex and 50 mM NaCl (control). The cells were then extracted with 250 mM (NH4) 2 SO 4 in CSK buffer at room temperature for 10 min and fixed with 3.7% formaldehyde in CSK buffer at 4°C for 30 min. Cells were stained with anti-T epitope antibody (Novagen), followed by rhodamine-conjugated donkey antibody to mouse IgG (Jackson ImmunoResearch Laboratories). For immunocolocalization of PIR1 and SC35, cells were permeabilized in CSK buffer in the absence of RNase A inhibitor (vanadyl-ribonucleoside complex) at 27°C for 1 h, extracted with 250 mM (NH4) 2 SO 4 in CSK buffer at room temperature for 10 min and then fixed with 3.7% formaldehyde in CSK buffer. Cells were stained with monoclonal antibody SC35 (ATCC), followed by rhodamine-conjugated donkey antibody to mouse IgG. The coverslips were then incubated with biotinylated anti-T epitope antibody (Novagen) followed by fluorescein-conjugated streptavidin (Jackson ImmunoResearch Laboratories). Cells were examined with a Bio-Rad confocal microscope MRC-600 and the imaging data was processed with computer program Adobe Photoshop 4.0.

Cloning of a Novel
Member of Dual-specificity Phosphatases-To identify novel dual-specificity phosphatases, we have used primers corresponding to the conserved sequences HCTH-GIN and FEQARGH in the catalytic domain of several dualspecificity phosphatases to amplify gene sequences from human cDNA libraries by polymerase chain reaction. In addition, we have used these sequences to directly search the Gen-Bank TM EST data base. These combined approaches have led us to identify a partial cDNA (EST clone H60626) that potentially encoded a novel phosphatase. To obtain its full-length cDNA, we used the EST clone H60626 as a probe to screen a cDNA library constructed from ML1 cells (a human myeloid cell line). The full-length cDNA sequence consists of 1593 nucleotides that potentially encode a protein of 329 amino acids (Fig. 1A). We have named this novel phosphatase PIR1 (phosphatase that interacts with RNA/RNP complex 1).
Sequence analysis of PIR1 shows that it is closely related to several dual-specificity phosphatases at the amino acid level (Fig. 1B), such as BVP, encoded by the baculovirus Autograph Californica (35% identity) (14); two open reading frames identified by the Caenorhabditis elegans genome project T23G7.5 (31% identity) and F54C8.4 (23% identity); and CEL-1 (25% identity). CEL-1 is recently shown to be a phosphatase in C. elegans that can remove 5Ј-phosphate from newly transcribed RNA molecules and is potentially involved in the mRNA capping reaction (15).
In addition to harboring the protein tyrosine phosphatase signature motif, PIR1 has several interesting features. It contains two stretches of arginine-rich sequences that are found in some of the RNA-binding proteins (Fig. 1A). PIR1 is also rich in lysine and histidine residues, and the protein is predicted to contain 17 basic charges at the neutral pH. PIR1 also carries a proline-rich region, a motif known to interact with the Srchomology 3 domain.
We have examined the expression pattern of PIR1 in various human cell lines by Northern blot analysis (Fig. 2). PIR1 was widely expressed with the most abundant messages in HaCaT (human keratinocyte cell line), A431 (human epidermoid cells), and ML1 (human myeloblastic). Interestingly, in several transformed cell lines that lack functional p53 and/or pRb, such as 293 (human embryonic kidney cells), HeLa (human cervical epitheloid carcinoma), and Saos2 cells (human osteosarcoma), the PIR1 level was quite low. Whether the PIR1 mRNA level may be regulated by p53 or pRb tumor suppressor protein needs to be further studied. No expression of the PIR1 message was observed in PC12 (rat pheochromocytoma) cells, although we cannot rule out the possibility that human PIR1 cDNA probe had failed to recognize its rat homolog. PIR1 Possesses Intrinsic Protein Tyrosine Phosphatase Activity-Although PIR1 carries the signature motif of protein tyrosine phosphatase, whether it bears protein phosphatase activity needs to be directly demonstrated. To do so, we have expressed PIR1 in insect Sf9 cells using the recombinant baculoviruses because the PIR1 protein is not stable in E. coli. We have expressed a fusion protein in which PIR1 was fused at the C terminus of GST (GST-PIR1). GST-PIR1 were affinity-purified by binding to glutathione-Sepharose beads. GST-PIR1 displayed protein tyrosine phosphatase activity toward tyrosylphosphorylated poly(GluTyr) (Fig. 3). Poly(GluTyr) is a random polymer of glutamate and tyrosine, and the tyrosyl-phosphorylated poly(GluTyr) has been shown to be an excellent in vitro substrate for dual-specificity phosphatases such as PTEN/ MMAC1/TEP1 (10). As control, we have assayed in parallel the PIR1 derivative carrying the cysteine 152 to serine mutation (C152S) in the tyrosine phosphatase signature motif. The tyrosine phosphatase activity of GST-PIR1 was abolished by the C152S mutation (Fig. 3). These studies suggest that PIR1 possesses an intrinsic protein tyrosine phosphatase activity. Under similar conditions, we could not detect phosphatase activity using phosphoseryl/threonyl casein as substrate (data not shown). This could be because of the substrate selectivity of PIR1. It remains possible that in vivo PIR1 may dephosphorylate phosphoseryl/threonyl residues in addition to phosphotyrosyl residues with its physiological substrates.
PIR1 Interacts Directly with Single-stranded DNA or RNA in Vitro-PIR1 contains two stretches of arginine-rich regions (Fig. 1A). Because the arginine-rich regions in the viral protein Rev and Tat are known to be involved in binding to RNA (16), we tested whether PIR1 can interact with RNA. Because many RNA-binding proteins can bind to both RNA and singlestranded DNA in vitro, we first tested whether PIR1 can bind to single-stranded DNA. Wild type PIR1 or PIR1(CS) mutant were expressed as T epitope-tagged forms by recombinant baculoviruses in Sf9 cells. When the cell lysates from such baculovirus-infected Sf9 cells were incubated with singlestranded DNA immobilized on agarose beads, most of the PIR1 or PIR1(CS) proteins were retained on the beads (Fig. 4A). The binding of PIR1 or PIR1(CS) to single-stranded DNA beads was resistant to a 0.25 M salt wash, and even after a 1 M salt wash, a substantial fraction of the proteins were still retained on the beads (Fig. 4A). The relative resistance of PIR1-DNA interactions to medium salt washes suggests that PIR1 or its CS mutant protein can bind to single-stranded DNA with high affinity.
Because the binding of PIR1 to single-stranded DNA in crude cell lysates can be mediated either by PIR1 directly or through its association with nucleic acid-binding proteins, we therefore performed Southwestern blot analysis to examine whether PIR1 can interact with single-stranded DNA directly. In this assay, GST-PIR1 or its CS derivative was affinity-purified, resolved on SDS-PAGE, and blotted to nitrocellulose. The proteins on the filters were subjected to denaturation and renaturation and then incubated with 32 P-labeled single-stranded DNA probe derived from the salmon sperm DNA template. After washing, the bound radiolabeled probe was detected by autoradiography. As shown in Fig. 4B, both GST-PIR1 or GST-PIR1(CS) bound to single-stranded DNA with comparable affinity in this assay. To examine the specificity of the binding, we used the GST-MKP-1 protein as a control. No singlestranded DNA binding was observed for GST-MKP-1, although Coomassie Blue staining confirmed that all GST-fusion proteins were present in about equal amounts (Fig. 4B). Our subsequent studies showed that similar single-stranded DNA binding could be achieved even when the denaturation and renaturation step was omitted (data not shown), suggesting that the native conformation of these proteins may not be required for the nucleic acid binding. In addition, this assay showed that both the wild type PIR1 and the catalytically inactive CS mutant protein bound to RNA with comparable affinity, suggesting that the PTP activity of the enzyme is not required for the binding.
To examine whether PIR1 shows binding preference for RNA or single-stranded DNA, we performed the filter binding assay using either 32 P-labeled RNA or 32 P-labeled single-stranded DNA as probes, referred to as Northwestern blot and Southwestern blot analysis, respectively. The RNA and DNA probes used were of defined sequences and were both generated with the same set of DNA templates. To increase the sequence complexity of the defined probes, we have used a mixture of several cDNAs clones as templates. These cDNAs include TEP1/PTEN/MMAC1 (9), cyclin A (17), p45 SKP2 (18), p27 KIP1 (19), and p21 WAF1 (20). Both 32 P-labeled RNA and DNA probes were of same specific activity and were used at 2 ϫ 10 6 cpm/ml for the filter binding assay. As shown in Fig. 4C, PIR1 exhibited higher affinity for RNA since more probe binding and less PIR1 protein required for the binding was observed for the labeled RNA than the DNA probe.
PIR1 Interacts with RNPs in the Yeast Two-hybrid System-To gain an insight into the cellular processes that are regulated by PIR1, we have employed the yeast two-hybrid screen method to identify proteins that may interact with PIR1. By screening the HeLa cDNA yeast two-hybrid library with PIR1 as a bait, we have identified ϳ100 positive clones in both histidine prototroph and ␤-galactosidase assay, and 20 of them were chosen for further analysis. These plasmids were recovered from yeast colonies. To test the specificity of the interaction with the bait protein, each plasmid was then retransformed into the yeast strain together with a bait plasmid encoding either PIR1 or CDK6. The plasmids that conferred interactions only with PIR1, but not with CDK6 control, were then sequenced. Two of the cDNA clones were found to encode splicing factors 9G8 or SRp30C, both as an in-frame fusion with the upstream Gal4 transcription activation domain (Fig. 5). Both 9G8 and SRp30C belong to the SR family splicing factors, which share a domain containing serine and arginine repeats and are components of the mRNA-splicing complexes (21). So far, we have not been able to detect physical association between PIR1 and splicing factors in mammalian cells. It is possible that the interaction of PIR1 with splicing factors 9G8 or SRp30C is mediated through an RNA intermediate, as PIR1 itself can bind to RNA directly, and our coimmunoprecipitation method may not be sensitive enough to detect such interaction.
Immunolocalization of the Ectopically Expressed PIR1-To examine the cellular localization of PIR1, we expressed the T epitope-tagged PIR1 in HeLa cells by the transient transfection method, followed by immunofluorescence staining with anti-T epitope antibody. PIR1 was localized to the nuclei with the exclusion of the nucleolus, and no staining was observed when cells were transfected with the vector alone (Fig. 6, A and B). To determine whether PIR1 in cells is associated with certain nucleic acids or protein complexes involved in nucleic acids metabolism, we subjected the permeabilized cells to DNase I or RNase A treatment before fixation. Although PIR1 staining was not affected by pretreatment with DNase I, it was greatly diminished by pretreatment with RNase A (Fig. 6, C and D). The sensitivity of PIR1 staining to RNase A suggests that PIR1 in mammalian cells is associated with RNA and/or RNP complexes involved in nuclear mRNA metabolism.
Because PIR1 interacts with splicing factors in the yeast two-hybrid system, we examined whether PIR1 is colocalized with proteins involved in mRNA splicing in mammalian cells. Several splicing factors, including SC35, are known to exhibit a speckled immunofluorescence pattern (22). Speckles are be- GST-PIR1 wild type (WT) or GST-PIR (CS) mutant, and GST-MKP-1 were expressed in Sf9 cells and purified by glutathione-Sepharose beads. The purified proteins were examined by SDS-PAGE and Coomassie Blue staining (left panel). All GST-fusion proteins were shown to be present in comparable amounts. Lysate from uninfected Sf9 cells were analyzed in parallel to serve as a control. Molecular standards are indicated on the left. The purified proteins were also analyzed by Southwestern blot analysis (right panel) in which a parallel set of proteins from SDS-PAGE gel were transferred to nitrocellulose filter. The filter was incubated with a 32 P-labeled single-stranded DNA probe derived from the salmon sperm DNA template. The bound probe was detected by autoradiography. C, PIR1 binds preferentially to RNA rather than to single-stranded DNA in vitro. Affinity-purified GST-PIR1(WT) proteins were analyzed by filter binding assay with either 32 P-labeled RNA probe (Northwestern blot) or 32 P-labeled singlestranded DNA probe (Southwestern blot) analysis. Both the RNA and the DNA probes were derived from the same pool of cDNA templates and were of the same specific activity and concentration (see "Material and Methods"). The protein amount loaded in each lane is indicated.
FIG. 5. Specific interaction of PIR1 with splicing factors in yeast two-hybrid system. Using PIR1 as a bait, positive clones were isolated that encode 9G8 or SRp30C. The plasmid expressing 9G8 or SRp30C (in pGADGH vector) was retransformed into yeast YRG2 strain either together with PIR1 or with CDK6 bait plasmid (in pGBT9 vector). Two independent colonies from each transformation were streaked out on histidine deficient plates. 9G8 (upper panel) or SRp30C (lower panel) interacted specifically with PIR1 but not with CDK6. lieved to be storage sites for certain splicing factors, and mRNA transcription, splicing, and maturation have been reported to take place in close proximity to speckles (21). Because the staining of PIR1 exhibits a nonuniform staining pattern (Fig.  6B), we wondered if removal of the diffused PIR1 staining would allow us to detect certain PIR1-associated subnuclear structures. We modified the immunostaining procedure so that the cells were permeabilized in the absence of RNase inhibitor to allow limited RNase digestion to take place. Under this condition, the ectopically expressed PIR1 was observed in speckles, which was reminiscent of the pattern known for the distribution of SC35 (22). We therefore performed a double immunostaining experiment in the PIR1-transfected HeLa cells and examined whether a fraction of PIR1 colocalizes with SC35 by confocal fluorescence microscopy. As shown in Fig. 7, both the exogenous PIR1 and endogenous SC35 exhibited the speckled pattern, which was superimposable, suggesting that PIR1 partially colocalizes with SC35 in these cells. These studies provide further support that PIR1 is associated with RNA and/or RNP in the mRNA processing or maturation process. DISCUSSION Our studies demonstrate that PIR1 is a novel member of the dual-specificity subfamily of protein tyrosine phosphatases that exhibits high affinity to RNA. PIR1 bound directly to the in vitro transcribed RNAs derived from cDNA templates. Using PIR1 as a bait, we have isolated cDNA clones encoding accessory splicing factors 9G8 and SRp30C that showed specific interactions with PIR1 in the yeast two-hybrid system. When ectopically expressed in HeLa cells, PIR1 manifested a nuclear staining pattern, and the staining was removed by pretreatment of cells with RNase A but not DNase I. Furthermore, a fraction of PIR1 was colocalized with the splicing factor SC35 in speckles. PIR1 is the first member in the protein tyrosine phosphatase family that shows high affinity for RNA both in vitro and in vivo. Our studies suggest that PIR1 may participate in nuclear mRNA metabolism.
Increasing evidence suggests the participation of protein kinases and phosphatases in mRNA processing (21,23). For example, a dual-specificity kinase Clk/Sty is found to be a partner with SR splicing factors in a yeast two-hybrid screen, and a catalytic inactive form of this kinase is colocalized with SR splicing factors in speckles in transfected cells (24). Serine/ threonine phosphatase 1, or a serine/threonine phosphatase-1like activity, has been shown to affect the subnuclear localization of the splicing factors (25). More recently, CEL-1, a phosphatase in C. elegans was shown to be a 5Ј triphosphatase for RNA molecules and was suggested to be involved in the mRNA capping reaction (15). Whether PIR1 is involved in regulating pre-mRNA splicing or other aspects of mRNA metabolism such as capping, polyadenylation, stability, or transport awaits further studies.
One interesting question arising from our studies is which region or domain in PIR1 mediates its interaction with RNA. Our observations that both the wild type PIR1 and its catalytically inactive C152S mutant bind to RNA with comparable affinity, and our data that binding can take place with denatured phosphatases raises the interesting possibility that interaction of PIR1 with RNA may not require the catalytic center nor the native conformation of the enzyme. One possibility is that PIR1 binds to RNA through its arginine-rich FIG. 6. Association of PIR1 with RNA-but not DNA-containing complexes in HeLa cells. HeLa cells were transfected with either pCGT vector alone (panel A) or with pCGT-PIR1 plasmid (panel B-D) and then processed for immunostaining with anti-T epitope antibody followed by rhodamine-conjugated secondary antibody. PIR1 immunostaining was observed in the nucleus of cells transfected with the PIR1 plasmid (panel B) but not in cells with the control vector (panel A). The PIR1 staining was not affected by treatment of permeabilized cells with DNase I (panel C) but was greatly diminished by treatment with RNase A before fixation (panel D).

FIG. 7.
Colocalization of a fraction of PIR1 with SC35 by confocal fluorescence microscopy analysis. HeLa cells transfected with pCGT-PIR1 plasmid were permeabilized in the absence of the RNase inhibitor to allow limited RNase digestion before fixation. Exogenous T epitope-tagged PIR1 was revealed with biotinylated anti-T epitope antibody followed by fluorescein-conjugated streptavidin (panel A, green). Endogenous splicing factor SC35 was detected with antibody SC35 followed by rhodamine-conjugated secondary antibody (panel B, red). The superimposed images of panels A and B are shown in panel C (yellow), indicating that PIR1 is colocalized with SC35. sequences. Several specific RNA binding proteins, such as Rev and Tat, are known to contain an arginine-rich motif (16). Rev, a protein encoded by HIV, can bind and facilitate nuclear export of intron-containing viral RNA. Tat, also encoded by HIV, is involved in the regulation of transcription by binding to viral mRNA. Very little identity is found between the argininerich motif sequences except for the richness of arginine residues. Further studies are required to determine whether the arginine-rich regions in PIR1 mediate its binding to RNA.