A heteromorphic protein-tyrosine phosphatase, PTP phi, is regulated by CSF-1 in macrophages.

A novel protein-tyrosine phosphatase, PTPφ, was cloned from a murine macrophage cDNA library. As a result of alternative splicing, macrophage PTPφ mRNAs are predicted to encode two membrane-spanning molecules and a cytosolic enzyme with identical catalytic domains. The membrane-spanning forms differ in the juxtamembrane region, while a start codon downstream of this region is utilized in the translation of the putative cytosolic form. Expression of PTPφ mRNA is low and restricted to macrophage cell lines, macrophage-rich tissues, and brain, kidney, and heart. The mRNA in macrophages and heart is ∼2.8 kilobases (kb). However, a ∼5.5-kb transcript in brain and kidney indicates a fourth isoform encoding a large extracellular domain. The ∼5.5-kb PTPφ brain mRNA encodes the mouse homolog of GLEPP1, a recently reported glomerular epithelial protein. The level of expression of the mRNA encoding the cytosolic form was very low, and only the membrane-spanning proteins (43 and 47 kDa) could be detected in macrophages. Following addition of colony stimulating factor-1 to quiescent BAC1.2F5 macrophages, PTPφ mRNA and protein were down-regulated. The restricted expression of the shorter isoforms of PTPφ and their regulation by colony stimulating factor-1 in macrophages suggest that PTPφ may play a role in mononuclear phagocyte survival, proliferation, and/or differentiation.

The importance of phosphorylation on tyrosine residues of proteins involved in the signaling of cell proliferation, differentiation, and transformation has been well established (reviewed in Ref. 1). Despite the fact that the protein-tyrosine phosphatases (PTPs) 1 were first described only relatively re-cently (2, 3), they have been found in species ranging from viruses to mammals (reviewed in Ref. 4). However, unlike the protein-tyrosine kinases, which exhibit sequence similarity with serine/threonine kinases, the PTPs do not show any sequence similarity with the serine/threonine phosphatases.
PTPs can be divided into two main groups: the low molecular weight, cytoplasmic, single catalytic domain-containing molecules and the high molecular weight, membrane-spanning, receptor-like forms, almost all of which contain two tandem repeats of the intracellular catalytic domain (4). The well conserved phosphatase domain consists of about 240 amino acids, within which is a particularly well conserved consensus sequence of 11 residues ((I/V)HCXAGXGR(S/T)G). The cysteine residue in this signature sequence is essential for the specific catalytic activity of the enzyme (5,6) and has been shown to participate in the formation of a covalent phosphoenzyme intermediate as part of the catalytic process (7).
The membrane-spanning PTPs frequently have very large extracellular domains containing immunoglobulin-like repeats and/or fibronectin III-like repeats, suggesting a receptor-like function. However, no ligands have yet been identified. In view of the similarity of the extracellular domains to neural cell adhesion molecules, it has been suggested that homophilic or heterophilic cell-cell interactions rather than soluble ligands may modulate the activity of some PTPs leading to cell contact inhibition (5), and this has been confirmed in the case of PTP (8) and PTP (9). The functional significance of the tandemly repeated catalytic domains present in all but a few of the receptor-like PTPs is uncertain. The amino-terminal domain alone is active with evidence for a complete lack of phosphatase activity in the carboxyl-terminal domain (5,6) or the presence of low activity (7,10). However, the carboxyl-terminal domain appears to regulate the substrate specificity of the first domain (5,6,7,10).
The cytosolic PTPs comprise a more disparate group. Subfamilies within this group are characterized by different amino-and carboxyl-terminal domains, many of which are important in the subcellular localization of these molecules (11)(12)(13). Recent evidence indicates that both subcellular localization (12) and phosphorylation status (14) are important in the cellular regulation of the activity of the cytosolic tyrosine phosphatases. Other methods of regulating activity include transcriptional control, which is responsible for the pattern of expression of an immediate, early gene whose product, 3CH134, is a dual specificity phosphatase that dephosphorylates mitogen-activated protein kinase (15) and alternative splicing in the catalytic domain, which down-regulates the in vitro affinity of PTP1D for substrates (16). * This work was supported by National Institutes of Health Grant CA 26504, Albert Einstein Core Cancer Grant P30-CA13330, a grant from the Lucille P. Markey Charitable Trust, a Leukemia Society of America Fellowship (to F. J. P.), and a Betty and Howard Isermann Scholarship and National Institutes of Health Medical Scientist Training Grant T32-GM07288 (to D. B. E.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) U37465, U37466, U37467.
‡ To whom all correspondence should be addressed: Dept. of Developmental and Molecular Biology, Albert Einstein College of Medicine, 1300 Morris Park Ave., Bronx, NY 10461. Tel.: 718-430-2344; Fax: 718-823-5877. 1 The abbreviations used are: PTP, protein-tyrosine phosphatase; PTP, protein-tyrosine phosphatase ; GLEPP1, glomerular epithelial protein 1; PCR, polymerase chain reaction; CSF-1, colony stimulating factor-1; RT-PCR, reverse transcriptase PCR; PAGE, polyacrylamide gel electrophoresis; GST, glutathione S-transferase; UTR, untranslated Alternative splicing is a relatively common phenomenon in PTPs resulting in extracellular variants (17,18) and intracellular variants (19,20) of the receptor-like PTPs. A membranespanning PTP isolated from rat pheochromocytoma cells has alternative splicing that results in two isoforms, one of which contains one catalytic domain while the other has the more usual tandemly repeated domains (21). Extracellular splicing in CD45 produces isoforms that are differentially expressed in lymphocytes (22), whereas the biological significance of the splice variants in the other molecules has not been clearly determined.
We have used the polymerase chain reaction (PCR) with degenerate oligonucleotide primers to identify the PTPs that are expressed in the colony stimulating factor-1 (CSF-1)-dependent murine macrophage cell line, BAC1.2F5. In this report, we describe a novel PTP, designated as PTP, whose primary transcript is alternatively spliced to yield messages encoding two membrane-spanning forms and a putative cytosolic form and whose expression is regulated by CSF-1. A recent study reported the cloning of a novel rabbit PTP derived from the renal glomerulus and designated glomerular epithelial protein 1, GLEPP1 (23), for which one isoform of PTP appears to be the mouse homolog.
Organs were collected from 17-and 20-day-old fetal, 2-day-old neonate, and adult C57B/6 mice and directly processed for RNA isolation or immediately frozen in liquid nitrogen prior to storage at Ϫ80°C.
RNA Isolation-Total cellular RNA was isolated from tissues and cells either by the shortened acid guanidinium thiocyanate-phenolchloroform protocol (29) or by using Trizol (Life Technologies, Inc.), a single-phase reagent for the extraction of RNA (30). When necessary, poly(A) ϩ RNA was purified using oligo(dT) cellulose columns (Stratagene).
Isolation of PTP Domains by PCR-A pair of oligonucleotide primers, 5Ј-TATATCTAGAAATGTGCACAGTACTGGCC-3Ј and 5Ј-AGGTA-AGCTTCC(G/C)AC(T/G/A)CC(G/A)GCXCT(G/A)CA(G/A)TG-3Ј, were designed with the bold letters corresponding to the highly conserved PTP domain sequences KCAQYWP and HCSAGVG, respectively. PTP sequences were amplified from BAC1.2F5 cell RNA using the SUPERSCRIPT Preamplification System (Life Technologies, Inc.) and a thermocycling protocol as follows: 5-min denaturation at 94°C, 2.5 units of Taq polymerase added, 3 low stringency cycles of 94°C for 1 min, 37°C for 30 s, then a 2-min ramp time heating to 72°C for 2 min, 35 higher stringency cycles of 94°C for 1 min, 50°C for 1 min, and 72°C for 2 min with a 2-s per cycle increase for extension, followed by a final extension of 5 min at 72°C. Amplification products were chloroform extracted, digested with XbaI and HindIII, purified by agarose gel electrophoresis and electroelution, subcloned into pGEMS7Z(ϩ)f (Promega), and then sequenced by the dideoxy chain termination method (31).
Isolation and Characterization of cDNA Clones-The novel PTP domain was labeled (Prime-It II kit, Stratagene) and used to screen a ZapII BAC1.2F5 cDNA library constructed by Stratagene from mRNA produced using the acid guanidinium thiocyanate-phenol-chloroform method. A 2.1-kb BamHI-EcoRV fragment of a positive clone (10 -1) obtained from the first screening was used to rescreen the library for additional clones. For screening, the DNA on the nitrocellulose filter lifts was denatured, neutralized, and UV-cross-linked, and the filters were prehybridized for 2 h followed by overnight hybridization with 2 ϫ 10 6 cpm of probe per ml of hybridization fluid in 50% formamide, 5 ϫ SSC, 25 mM KPO 4 , 5 ϫ Denhardt's, 50 g/ml herring sperm DNA, 10% dextran sulfate at 37°C. Filters were washed twice with 2 ϫ SSC and 0.1% SDS at 37°C and three times with 0.1 ϫ SSC and 0.1% SDS at 55°C. All positive clones in the pBluescript phagemid were excised from the ZapII vector using the protocol described by Stratagene, and seven clones were fully sequenced in both directions using the dideoxy chain termination method and by automated sequencing in the 373A automated sequencer (Applied Biosystems, Foster City, CA). To determine the full-length of the molecule, both 5Ј-and 3Ј-rapid amplification of cDNA ends (Life Technologies, Inc.) were performed using genespecific primers corresponding to regions at the 5Ј-and 3Ј-ends of the derived sequence. The following libraries were also screened: an adult mouse brain cDNA library (ML1024a, Clontech) and a neonatal mouse brain cDNA library (937319, Stratagene).
Reverse Transcriptase PCR to Confirm Alternative Transcripts-RT-PCR was carried out on BAC1.2F5 and mouse brain total RNA using the SUPERSCRIPT preamplification system (Life Technologies, Inc.) with the thermocycling program set for 30 high stringency cycles as described above for the initial experiment. Primers used in the amplification procedure were designed to frame the alternative splicing sites and were 5Ј-GCCACTGCCTACAACTGCAGTGTC-3Ј and 5Ј-AT-AGTCCGAGTCCTTGGCCATATC-3Ј. PCR products were subcloned immediately into the pCRII vector (Invitrogen) and sequenced in the 373A automated sequencer.
In Vitro Translation of PTP cDNA Clones-The translation products of the three different PTP clones were examined using the TNT coupled reticulocyte lysate in vitro translation system (Promega) with [ 35 S]methionine (Amersham). Translation products were analyzed using a 7.5-17.5% gradient SDS-polyacrylamide gel electrophoresis (SDS-PAGE) minigel.
Expression and Purification of Recombinant PTP Protein-A fulllength recombinant PTP protein was produced as a glutathione Stransferase (GST) fusion protein using the pGEX-KG Escherichia coli expression vector (a gift from K. Guan) (33) and a PCR-produced PTP sequence identical to the open reading frame of the long insert variant. The primers used were 5Ј-TGGTTCTAGAGGTGAACCCTAACGTG-GTG-3Ј and 5Ј-CGACCATGGATTTGCTGACGTTCTCGTA-3Ј. The vector and the PCR amplicon were digested with XbaI and NcoI and ligated. The DH5␣ strain of E. coli was transformed with the construct, and the GST-PTP fusion protein was expressed and purified as described by Guan and Dixon (33).
nonspecific binding. Due to the random length of the poly(Glu-Tyr) (4:1), substrate concentrations are expressed in terms of phosphotyrosine.
Casein was phosphorylated by incubating 100 units of the protein kinase A-catalytic subunit (P-8289, Sigma) with 660 g of casein (C-4032, Sigma) in 40 mM Hepes, pH 7.0, 20 mM MgCl 2 , 1 mM benzamidine HCl, 1 g/ml pepstatin A, 1 g/ml leupeptin, 5 g/ml aprotonin, 2 mM ATP with 600 Ci of [␥-32 P]ATP (3000 Ci/mmol) (total volume of 600 l) for 1 h at 37°C before termination of the reaction by the addition of 400 l of trichloroacetic acid at 4°C for 1 h. The trichloroacetic acid-precipitated protein was pelleted by centrifugation at 12,000 ϫ g for 1 h at 4°C, washed, and dialyzed against water in a Centricon-30 concentrator as described above.
Serine phosphorylation and conversion of phosphorylase b to phosphorylase a by phosphorylase kinase was carried out according to the protocol outlined in the Life Technologies, Inc. phosphatase assay system (Life Technologies, Inc.). The substrate was then precipitated in ice-cold 90% saturated ammonium sulfate at 4°C for 30 min, pelleted, washed five times with ice-cold 45% ammonium sulfate, and resuspended in 1 ml of buffer B (50 mM Tris HCl, pH 7.0, 0.1 mM EDTA, 15 mM caffeine, 0.1% ␤ mercaptoethanol (w/v)). The sample was then successively diluted and concentrated against buffer B in a Centricon-30 concentrator as described above.
Phosphatase Assays of Recombinant PTP-The purified recombinant GST-PTP protein was added to buffer A containing 2 M 32 Plabeled poly(Glu-Tyr) (4:1), 1 mg/ml bovine serum albumin, and 10 mM dithiothreitol (final volume of 100 l) and incubated at 22°C with constant agitation for 20 min. Acid-activated charcoal (700 l of 0.9 M HCl, 90 mM Na 4 P 2 O 7 ⅐ 10H 2 O, 2 mM NaH 2 PO 4 , 4% activated charcoal (C-5260, Sigma)) was added for 5 min at 22°C, and the charcoal-bound substrate was pelleted by centrifugation at 15,000 ϫ g for 8 min, as described previously (5). The released soluble 32 P i was measured by Cerenkov counting.
Serine phosphatase assays were performed as for the tyrosine phosphatase assays with the following changes: 1 mM MnCl 2 and 30 g/ml protamine sulfate were included in buffer A (34, 35) and 100 pmol of 32 P-labeled casein or 600 pmol of [ 32 P]phosphorylase a, prepared as stated above, replaced the 32 P-labeled poly(Glu-Tyr) (4:1).
The pH activity profile for PTP was determined using a previously described mixed buffer system (50 mM acetic acid, 50 mM Mes and 100 mM triethanolamine) that maintained constant composition and ionic strength within the pH range tested (5-8.5) (36). Tyrosine phosphatase assays were performed as described above with the mixed buffer replacing Hepes in buffer A and a decrease in the concentrations of NaCl and KCl to 35 mM. Assays to determine the optimum ionic strength for PTP activity were carried out in buffer A without KCl but containing from 0 to 500 mM NaCl.
Expression of PTP Protein in BAC1.2F5 Cells-Cells were washed in phosphate buffered saline and lysed in 2 ϫ SDS-PAGE sample buffer. Separation into membrane and cytosolic fractions and Western blotting were performed as described previously (37), with the following modifications: Immobilon-P SQ transfer membrane (ISEQ26260, Millipore) was used, and the transfer buffer contained 0.01% SDS. The blot was blocked with 5% bovine serum albumin and 1% ovalbumin (Sigma) in Tris-buffered saline containing 0.05% Nonidet P-40 and 0.1% azide and then washed with Tris-buffered saline containing 0.05% Nonidet P-40 prior to incubation with 1:400 ␣PTP antiserum overnight at 4°C.
Peroxidase-conjugated goat anti-rabbit antibody (Amersham) diluted at 1:8000 in the blocking solution (without azide) was used to detect the bound ␣PTP antibody by enhanced chemiluminescence. The ␣PTP rabbit antisera were raised to three different PTP-specific peptides and to the full-length PTP protein after its thrombin cleavage from GST-PTP. The peptide sequences corresponded to the juxtamembrane (SLEREGKLPYSWRRSVF) and carboxyl-terminal (CISDVIYENVSKS) sequences and a sequence amino-terminal to the catalytic domain (AKDSDYKFSLQFEE) of PTP.

Isolation of PTP-After
RT-PCR with BAC1.2F5 RNA using degenerate oligonucleotides to the catalytic domain, a discrete band of the expected length (ϳ300 bp) was observed. Subcloning and sequencing of the amplification products revealed 9 different PTP sequences in the 50 colonies screened, corresponding to 6 previously identified PTPs and 2 PTPs that were novel at that time. The first novel clone was simultaneously identified in several other laboratories and called PTP1C (38), SHP (39), SH-PTP1 (40), and HCP (41). The cloning and analysis of the cDNA encoding the second novel sequence, designated PTP, is reported here.
Screening of a BAC1.2F5 cDNA library with the subcloned PCR product yielded two overlapping partial clones. The library was rescreened with a 2.1-kb fragment of one of these, 10 -1, and five additional clones (10 -1/1, 10 -1/3, 10 -1/5, 10 -1/7, 10 -1/8) were obtained, all of which contained the 10 -1 PTP sequence. However, three of the five clones contained an insert of either 190 or 274 bp that was spliced in upstream of the catalytic domain (Fig. 1). The 274-bp insert in clones 10 -1/1 and 10 -1/7 had a unique 84-bp stretch at the 3Ј-end of the 190-bp insert in clone 10 -1/3 but was otherwise identical (Fig.  2). The sequences of the two groups of clones that have inserts differed from each other by a finite number of codons so that the reading frame was maintained, but the sequence of the original insert-free clone, 10 -1 (Fig. 1), differed by 63 codons ϩ 1 bp and 91 codons ϩ 1 bp from the short (10 -1/3) and long (10 -1/7) insert clones, respectively, resulting in the loss of reading frame. All clones had the same 5Ј-and 3Ј-untranslated region (UTR) sequences, the 5Ј-UTR containing an in-frame ochre stop codon at nucleotide 111. The insert-containing clones had a start codon with strong consensus for protein initiation (42) that was just upstream of the 5Ј-end of each alternatively spliced insert. Both inserts encoded the same 25-amino acid putative transmembrane domain, followed by a variable juxtamembrane domain upstream, and thus the insert-containing variants putatively coded for two very similar membrane-spanning PTPs with short extracellular domains. In contrast, the start codon for 10 -1 lay 78 bp downstream of the insert site (Fig. 2), and the sequence predicted a low M r cytosolic PTP. The alternative splice sites have been confirmed by mapping and sequencing of the exon-intron boundaries (Fig. 2). 3 Data base searching (GenEmbl, 1994) for the catalytic domain alone revealed close homology with the murine homolog of PTP␤ (49.8% similarity by Lipman-Pearson protein alignment), a single catalytic domain-containing PTP (43), with DPTP10d (50.2%), a Drosophila PTP expressed in the central nervous system (44,45), and with leukocyte antigen-related PTP (42.1%) (46).
Expression of PTP mRNA in Cell Lines and Tissues-Northern analysis of BAC1.2F5 RNA for PTP expression revealed a low-copy message of about 2.8 kb. It is not possible to resolve three separate PTP mRNAs corresponding to the variant cDNA clones, but the broad mRNA band shown covered the range (2.5-2.9 kb) of all three. PTP was also present in another well differentiated macrophage cell line, JPL3, derived from placenta (25)  mast cell line (MC) or in the fibroblastic L cell line (LC) (Fig. 3). It was also not expressed in a thymic lymphoma cell line (WEHI 222), a T-cell lymphoma cell line (WEHI 34), or a B-cell lymphoma cell line (WEHI 231) (data not shown). Northern analysis of adult C57B/6 mouse tissues showed moderately low expression of PTP in brain, weak expression in kidney, bone marrow, and heart, and barely detectable expression in lung, spleen, and thymus (Fig. 4). There was no expression in liver or intestine (Fig. 4) or in stomach, skeletal muscle, uterus, or testis (results not shown). The message in brain and kidney is much longer (ϳ5.5 kb) than the BAC1.2F5 message or the message from tissues containing large numbers of macrophages. Developmental Northern analysis of brain, kidney, lung, spleen, liver, and intestine from 17-and 20-day-old fetal and 2-day-old neonatal mice showed expression only in brain and kidney (data not shown).
Both an adult and a neonatal mouse brain cDNA library were screened to sequence the extra 2.7 kb of brain transcript revealed by Northern analysis. Of the six independent clones identified, only one contained sequence extending 5Ј of the short extracellular domain seen in BAC1.2F5 cells. Sequence analysis of this clone, which contained 116 bp of new 5Ј-sequence revealed a fibronectin-like repeat homology region (results not shown). All six brain clones spanned at least part of the alternatively spliced region, and all contained the insert characteristic of long insert (10 -1/7) variant.
In view of the failure to discern separate transcripts corresponding to the different cDNA clones obtained from the library screens by Northern analysis, RT-PCR was carried out on RNA extracted from BAC1.2F5 cells, mouse brain, kidney, and bone marrow using oligonucleotide primers that framed the insert region. BAC1.2F5 RNA showed three bands corresponding to the long (10 -1/7), short (10 -1/3), and no-insert (10 -1) variants (Fig. 5). The short insert variant appeared to be more highly expressed than the long insert variant, while the level of expression of the no-insert variant was substantially lower. Since all three amplicons used the same primer pair in the same PCR reaction tube, it is likely that the results reflected the relative levels of expression of the three variants. The major product seen on amplification of whole brain RNA was the long insert variant with a faint band corresponding to the short insert variant and no evidence of expression of the no-insert variant. The major product in kidney was the short insert variant with some expression of the long insert variant but no evidence of the no-insert variant. The pattern of expression of PTP mRNA in bone marrow was the same as the pattern of BAC1.2F5 macrophage expression.
To confirm that the sequences of the BAC1.2F5 mRNAs in the alternatively spliced regions corresponded to the sequences of the cDNA clones obtained by library screening, all three RT-PCR products were subcloned and sequenced. The mRNA sequences for the two insert-containing variants and the putative cytosolic variant were identical to those of their corresponding cDNA clones.
Expression and Activity of PTP Protein-To confirm that, as postulated, the original clone, 10 -1, initiated translation from a downstream start codon (ATG 2, Fig. 1), in vitro transcription-translation of the three full-length macrophage variant clones was carried out. The SDS-PAGE analysis shows that the most intense band for each cDNA clone corresponds to a protein product of expected molecular mass, 33 kDa for 10 -1, 43 kDa for 10 -1/3, and 47 kDa for 10 -1/7 (Fig. 6). Immunoprecipitation of these three proteins with an ␣PTP rabbit antiserum directed to a COOH-terminal peptide confirmed that they were PTP (data not shown). The next most intense band for the putative membrane-spanning variants corresponds to the 33-kDa band observed for the cytosolic variant and is explained by the reinitiation of protein synthesis at the downstream start codon (42). The protein produced by translation from the upstream start codon in the cytosolic variant would only be 6 amino acids long and therefore not visible in this analysis. In the case of the membrane-spanning variants, there is another start codon with a strong consensus at nucleotide 498, which is probably responsible for the less intense band running ϳ4 kDa faster than their major products.
The long insert variant (10 -1/7) was expressed as a GST-PTP fusion protein in E. coli to characterize its catalytic activity with respect to substrate specificity, pH, ionic strength, and kinetic parameters. GST-PTP possessed PTP activity for the tyrosine-phosphorylated poly(Glu-Tyr) substrate that was substantially decreased in the presence of 1 mM vanadate, while GST alone had no measurable activity (Fig. 7A). No GST-PTP activity could be detected toward the serine/threonine phosphatase-specific substrates, phosphorylated phosphorylase a, and casein (Fig. 7A). The pH optimum for GST-PTP of 6.5 was similar to that reported for leukocyte antigen-related PTP toward protein substrates (7) (Fig. 7B). The optimum ionic strength for GST-PTP activity was achieved with 100 mM NaCl, with activity declining at higher NaCl concentrations (Fig. 7C). Michaelis-Menton analysis revealed a K m of 2.6 M toward phosphotyrosine on the phosphorylated poly(Glu-Tyr) (4:1) with a k cat of 0.2 s Ϫ1 (Fig. 7D).
To confirm the presence of the various protein isoforms of PTP in BAC1.2F5 cells, whole cell lysates, membrane, and cytosolic fractions were Western blotted and probed with ␣PTP antisera. Consistent with our in vitro translation results and their predicted transmembrane domain, the ␣PTP fusion protein antibody detected a doublet at 43-47 kDa corresponding to the short and long insert forms in both whole cell lysate and the membrane fraction that was not detected by preimmune serum (Fig. 8). The ␣PTP antibody binding was inhibited by purified GST-PTP protein but not by GST alone. The three anti-peptide PTP antisera also detected the 43-and 47-kDa isoforms. Neither the ␣PTP antibody nor the two anti-peptide antibodies, which would be expected to recognize the putative cytosolic form, were able to detect the predicted 33-kDa band (see Fig. 6) in the whole cell lysate or cytosolic fraction either by Western blotting or by immunoprecipitation followed by Western blotting (data not shown). As PTP is not an abundant cellular protein and the relative expression of the 43-and 47-kDa isoforms reflects the relative expression of their predicted mRNAs assessed by RT-PCR (Fig. 5), the much lower expression of mRNA for the putative 33-kDa cytosolic isoform is consistent with the failure to detect the protein. The ϳ38 -42-kDa doublet detected by the fusion protein antibody in the whole cell lysate and cytosolic fraction (Figs. 8 and 9D) was detected by all the anti-peptide antibodies. This doublet may represent proteolytic cleavage products of the putative membrane-spanning isoforms since the extent of doublet band separation is unchanged and there are potential cleavage sites (dibasic residues) immediately after the transmembrane domain (Fig. 2).
Regulation of PTP mRNA Expression by CSF-1-CSF-1 is the major growth factor required for the survival, proliferation, and differentiation of macrophages. Since PTP is relatively selectively expressed in macrophages, we examined the regulation of PTP mRNA expression by CSF-1. BAC1.2F5 cells, rendered quiescent by overnight incubation in the absence of CSF-1, were then incubated with the growth factor for different lengths of time, and RNA was extracted and subjected to Northern analysis (Fig. 9A). Compared with PTP expression in log-phase growth cells, PTP expression was consistently much stronger in quiescent cells. Expression remained high for 2 h after the addition of CSF-1 before abruptly decreasing to a level below that observed in log-phase cells. Thereafter, levels remained low for the next 12 h before returning to about the level seen in log-phase cells (Fig. 9C). Quantitative phosphorimage analysis of the data from three separate experiments indicates that there was a 2.5 Ϯ 1.3-fold increase in PTP expression in quiescent over log-phase cells. RT-PCR of RNA samples extracted from proliferating and quiescent BAC1.2F5 cells revealed that there was no differential regulation of the expression of the alternatively spliced transcripts by CSF-1 (data not shown). Consistent with these observations, the expression of the 43-and 47-kDa PTP proteins, measured densitometrically, was also increased in quiescent cells compared with log-phase cells (2.4-fold for whole cell lysate, 1.8-fold for membrane fraction) or with quiescent cells that had been cultured with CSF-1 for 5 h (1.7-fold for whole cell lysate, 1.8-fold for membrane fraction) (Fig. 9D). DISCUSSION We have cloned and characterized a novel PTP from a murine macrophage cell line, BAC1.2F5, and have designated it PTP. Uniquely among PTPs, PTP encodes, as a result of alternative splicing, both membrane-spanning, single catalytic domain-containing PTPs and a putative low M r cytoplasmic PTP. The alternative splicing seen in PTP is unusual in that the transcript encoding the cytoplasmic variant has a change of reading frame compared with the membrane-spanning transcripts.
In the putative membrane-spanning isoforms an identical 5Ј-sequence is found within both the short and long inserts, while the long insert has an extra 28 codons spliced in at the 3Ј-end compared with the short insert providing evidence for two separate alternatively spliced regions. These putative membrane-spanning isoforms encode 43-and 47-kDa proteins that each have a very short extracellular domain of 8 residues.
In vitro transcription-translation demonstrates that a second, downstream start codon is responsible for initiation of protein synthesis in the putative cytosolic PTP, yielding a catalytic domain that possesses only short amino-terminal (32 residues) and carboxyl-terminal (22 residues) stretches (Fig. 1). RNA processing to produce both a single and a double catalytic domain-containing, receptor-like PTP has been reported previously (21), but there have been no reports of any PTPs that, through alternative splicing of a transmembrane domain, may exist as membrane-spanning and cytosolic forms.
Northern analysis reveals the presence in brain and kidney of a further splice variant that, at ϳ5.5 kb, is almost twice the length of the transcripts seen in BAC1.2F5 cells and bone marrow. Preliminary studies in our laboratory indicate that the 5.5-kb PTP message expressed in brain encodes a form of PTP with a large extracellular domain containing fibronectin type III repeats. 3 As this manuscript was being completed, a report of a membrane-spanning rabbit PTP, GLEPP1, containing a single catalytic domain and a large extracellular domain consisting of 8 fibronectin type III repeats was published (23). PTP appears to be the mouse homolog of GLEPP1 (Fig. 10), which was detected in the renal cortex and brain only. The failure to detect GLEPP1 mRNA in other tissues probably resulted from the use of a GLEPP1 extracellular domain probe in the ribonuclease protection assay, preventing detection of both the cytosolic and short extracellular domain-containing forms. Although 2 of the 13 GLEPP1 cDNA clones showed evidence of a short stretch of alternative splicing in the juxtamembrane region leading to the exclusion of 12 amino acids immediately 3Ј to the inserts described for PTP, none of these clones indicated alternative splicing of the extracellular domain or of the transmembrane domain (Fig. 10A). Thus, the ϳ2.8-kb messages encoding the membrane-spanning forms and the putative cytosolic form were not described. Moreover, as the sequence of GLEPP1 in the insert region corresponds to one of the short insert variants of PTP, the unique 3Ј 84-bp sequence of the long insert is not represented in the published GLEPP1 sequence (Fig. 10B). The cDNA sequence for the 5Ј-UTR of PTP is homologous to that of the extracellular domain of GLEPP1 in the juxtamembrane region for 270 bp before it diverges completely at nucleotide 111 for PTP (nucleotide 2340 for GLEPP1) (Fig. 10C). The point of divergence coincides with the in-frame stop codon in the 5Ј-UTR of PTP and is therefore likely to represent a splice site. Consistent with its homology to GLEPP1, the brain PTP clone that extended 5Ј of the BAC1.2F5 sequence diverged from the BAC1.2F5 sequence at this point also. The 3Ј-UTR of GLEPP1 and PTP exhibit a ϳ67.5% sequence similarity, and the ATTAAA polyadenylation site at nucleotide 2718 in PTP is not seen in GLEPP1, for which there is no reported polyadenylation site.
The pattern of expression of PTP is quite restricted and the level of expression very low with Northern blots requiring a 2-week autoradiograph exposure time. Like several other membrane-spanning PTPs (18,44,45), it is most highly expressed in brain, both developing and adult, but even there at low levels and at even lower levels in kidney. However, expression of the ϳ2.8-kb form is mostly limited to macrophages and tissues that contain significant numbers of macrophages such as bone marrow, lung, spleen, and thymus (28,47). The lack of detectable ϳ2.8-kb message in liver, which contains significant numbers of macrophages, may be a result of the large amounts of mRNA FIG. 10. Comparison of PTP and GLEPP1. (23). A, schematic comparison of the two membrane-spanning isoforms of PTP expressed in macrophages and the larger, receptor-like GLEPP1 expressed in kidney and brain. The signal sequence and transmembrane domains are represented by the filled boxes, the lighter cross-hatching indicates the fibronectin type III repeats, and the darker cross-hatching indicates the catalytic domains. B, alignment of the open reading frame of the long insert variant of PTP (10 -1/7) with the corresponding amino acid sequence of GLEPP1. Identical residues are boxed, and the dashes represent the lack of corresponding GLEPP1 sequence. C, alignment of the DNA sequence of the 5Ј-UTR of PTP with the juxtamembranous extracellular domain of GLEPP1. The start codon for PTP is boxed, and the in-frame stop codon is underlined.
produced by hepatocytes and a consequent dilution of PTP mRNA below the levels of detection by Northern blotting. A similar problem may exist in the brain where the proportion of macrophages (microglia) is lower. A ϳ2.8-kb message was also detected in heart but not skeletal muscle. It is notable that regulation of expression of PTP mRNA at a tissue-specific level is very complex, with the ϳ5.5-kb long insert variant almost exclusively expressed in the brain, the ϳ5.5-kb short insert variant expressed in renal glomeruli, and the shorter ϳ2.8-kb insert-containing variants expressed in macrophages. Expression of PTP mRNA in hematopoietic cells appears to be limited to mature cells of the monocyte-macrophage lineage. Primary macrophages express PTP at a level at least equivalent to the brain, and this, together with the regulation of PTP by CSF-1 in BAC1.2F5 macrophages, suggests that this enzyme plays an important role in these cells.
An antibody raised to the recombinant PTP fusion protein expressed in E. coli recognized 43-and 47-kDa proteins exclusively expressed in the membrane fraction of BAC1.2F5 macrophages. The predicted 33-kDa cytosolic PTP protein could not be detected by Western blotting, consistent with the very low level of mRNA expression.
Apart from the evidence for both restricted tissue expression and tissue-specific alternative splicing, we have shown that expression of macrophage PTP is regulated by CSF-1 with a pronounced increase in the level of expression of PTP mRNA in quiescent BAC1.2F5 macrophages compared with cells in logphase growth. The level of expression decreases rapidly 2 h after CSF-1 is reintroduced, to a level that is lower than that seen in log-phase cells. Commensurate with the increased expression of PTP mRNA in quiescent BAC1.2F5 macrophages, the expression of PTP protein is increased in cells deprived of CSF-1.
Thus, PTP is a heteromorphic PTP that is not only expressed in a restricted range of tissues but has differential tissue expression of several alternatively spliced mRNA variants encoding two membrane-spanning forms with very short extracellular domains and a putative cytosolic form detected in macrophages and bone marrow, a large membrane-spanning form that contains fibronectin type III repeats in the extracellular domain expressed in brain, and a slightly shorter fibronectin repeat-containing membrane-spanning form found in kidney (23). The restricted expression of the shorter isoforms of PTP and the demonstration that CSF-1 down-regulates their expression suggests a role for PTP in the regulation of macrophage survival, proliferation, or differentiation. It will be of interest to determine whether expression of PTP is induced in myeloid cells undergoing differentiation toward the monocyte/ macrophage lineage under the influence of CSF-1. It would also be of interest to determine whether the higher molecular weight forms of PTP expressed in kidney and brain are similarly regulated during proliferation and differentiation of the cells expressing them.