Isolation and Characterization of a Human Dual Specificity Protein-Tyrosine Phosphatase Gene*

Vaccinia phosphatase VH-1 and its mammalian counterparts, including protein-tyrosine phosphatases (PTPase) CLlOO and VHR, constitute a novel subfamily of protein-tyrosine phosphatases that exhibits dual substrate specificity for phosphotyrosineand phosphoserindthreonine-containing substrates. The expression of human VH-l-like PTPase CLlOO is rapidly inducible by mitogen stimulation and oxidative stress, suggesting that this gene is transcriptionally regulated. In order to study the mechanism underlying this transcriptional regulation, we isolated the first human gene of this subfamily, the CLlOO gene, and characterized its promoter. "he gene consists of four exons intervened by three short introns 400600 base pairs in length. Analysis of the protein sequence encoded by each exon revealed that there is a second region of similarity between CLlOO protein and cdc2S in addition to the PTPase catalytic domain. Promoter analysis of the CLlOO gene indicates that an 800-base pair region flanking the transcriptional initiation site is sufficient to confer  a transcriptional response to serum and 12-O-tetradecanoylphorbol-13-acetate stimulation. The CLlOO gene is expressed in numerous tissues, including nonmitotic cells in the brain. Within the brain, CLlOO mRNA is localized in discrete neuronal populations, suggesting that this PTPase is likely to play a key role in neurotransmission as well as in mitotic signaling. Finally, although extracellular signal-regulated kinase has recently been shown to act as substrate for CLlOO in vitro, we find no clear correspondence between the distribution of extracellular signal-regulated kinase and CLlOO mRNA in the brain. The potential significance of a second cdc2S homology domain of CLlOO is discussed.


Isolation and Characterization of a Human Dual Specificity
Protein-Tyrosine Phosphatase Gene* (Received for publication, August 17, 1993, and in revised form, September 17, 1993) Seung P. Kwak, David J. Hakes Vaccinia phosphatase VH-1 and its mammalian counterparts, including protein-tyrosine phosphatases (PTPase) CLlOO and VHR, constitute a novel subfamily of protein-tyrosine phosphatases that exhibits dual substrate specificity for phosphotyrosine-and phosphoserindthreonine-containing substrates. The expression of human VH-l-like PTPase CLlOO is rapidly inducible by mitogen stimulation and oxidative stress, suggesting that this gene is transcriptionally regulated. In order to study the mechanism underlying this transcriptional regulation, we isolated the first human gene of this subfamily, the CLlOO gene, and characterized its promoter. "he gene consists of four exons intervened by three short introns 400600 base pairs in length. Analysis of the protein sequence encoded by each exon revealed that there is a second region of similarity between CLlOO protein and cdc2S in addition to the PTPase catalytic domain. Promoter analysis of the CLlOO gene indicates that an 800-base pair region flanking the transcriptional initiation site is sufficient to confer a transcriptional response to serum and 12-O-tetradecanoylphorbol-13-acetate stimulation. The CLlOO gene is expressed in numerous tissues, including nonmitotic cells in the brain. Within the brain, CLlOO mRNA is localized in discrete neuronal populations, suggesting that this PTPase is likely to play a key role in neurotransmission as well as in mitotic signaling. Finally, although extracellular signal-regulated kinase has recently been shown to act as substrate for CLlOO in vitro, we find no clear correspondence between the distribution of extracellular signal-regulated kinase and CLlOO mRNA in the brain. The potential significance of a second cdc2S homology domain of CLlOO is discussed.
The vaccinia virus H1 (VH-1) gene encodes a protein phosphatase that exhibits dual specificity for phosphoserine-and phosphotyrosine-containing substrates (1). Other viruses of the pox family also express this class of phosphatase (21, all of which contain the common peptide motif, HCXAGXXR, within the catalytic domain. This motif is considered the signature sequence for protein-tyrosine phosphatases (PTPases),l a fam-* This work was supported in part by funding from the Walther Cancer Institute, Indianapolis, IN (to s. P. K. and J. E. D.), Grant M 0 1 RR00042 awarded to GCRC, University of Michigan, and Grant 18024 from the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases. 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.

to the GenBank-IEMBL Data Bank with accession number(s) 1701669.
The nucleotide sequence(s) reported in this paper has been submitted (CL100 gene) $ To whom correspondence should be addressed.
ERK, extracellular signal-regulated kinase; nt, nucleotide(s); "PA, 12-The abbreviations used are: PTPase, protein-tyrosine phosphatase; ily of enzymes that exhibit a n absolute substrate requirement for phosphotyrosines. Thus, the VH-l-like phosphatases constitute an emerging subfamily of PTPases that possess dual catalytic functions.
Three human PTPases of the VH-1 subfamily have been isolated to date (3-5). The protein structure of two of these VH-l-like PTPases are quite similar. Proteins encoded by PAC-1 and CLlOO mRNAs (pPAC-1 and pCL100, respectively) are similar in size, possess the catalytic domain at the C terminus, and have an extended N terminus. The third human VH-l-like PTPase, pVHR, is a smaller protein that lacks the long N terminus and resembles the viral VH-1 PTPase in structure (3). Additional characteristics shared between PAC-1 and CLlOO are the rapid transcriptional response to mitogen stimulation and rapid degradation of their respective mRNAs (5)(6)(7)(8). pPAC-1 becomes translocated into the nucleus after stimulation ( 9 , whereas the intracellular localization of pCLl00 is presently unknown. These observations suggest that the activity of some members of this PTPase subfamily is regulated at the transcriptional level and additionally by intracellular translocation. An interesting feature of this novel subfamily of PTPases is the ability to dephosphorylate phosphotyrosine and phosphoserindthreonine residues. Both VH-1 and pVHR PTPases exhibit this dual specificity toward artificial substrates such as casein and myelin basic protein (1-3), whereas pCLl00 catalyzes phosphotyrosine and phosphothreonine residues of extracellular signal-regulated kinase (ERK1) in vitro (9). The only other eukaryotic PTPase shown to possess dual substrate specificity is cdc25. Protein cdc25a is involved in the cell cycle and activates the cdc2-cyclin B complex by dephosphorylating residues Thr14 and of cdc2 (10)(11)(12)(13)(14). However, the sequence homology between cdc25s and the VH-l-like PTPases is low within the catalytic domain, suggesting that they constitute two separate subclasses of dual specificity phosphatases (15,16).
The existence of a novel class of PTPase whose expression appears to be induced by a variety of mitogenic signals allows one to speculate on the roles VH-l-like PTPases play during cell division. It is suggested from the transient and immediate early time course of mRNA expression that trans-acting and cis-acting elements provide exquisite control over the transcriptional activity of these PTPase genes. However, the gene structure of this relatively novel class of PTPases is not known. We therefore characterized the first gene of this subfamily of PTPases, the CLlOO gene, and partially defined the promoter region necessary for expression of this gene in HeLa cells. We subsequently determined the tissue distribution of CLlOO mRNA and investigated the potential relationship between O-tetradecanoylphorbol-13-acetate; PCR, polymerase chain reaction; CRE, cAMP-responsive element; CREB, cAMP-responsive elementbinding protein; PDGF, platelet-derived growth factor; EGF, epidermal growth factor; FGF, fibroblast growth factor.
CLlOO and its potential substrate, ERKl, by ascertaining their anatomical localization within the rat brain.
MATERIAL AND METHODS Genomic Library Screening-Specific primers 24 nt in length coding for the N terminus peptide sequence PVEILPFL (S'GGTGGAAATCCT-GCCCTITCT3') and complementary to the C-terminal peptide sequence PVSIPVHS (5'AGTGGACAGGcATGGAGACGGB') of pCLl00 were used to amplify a 560-bp CLlOO cDNA fragment from a human placental cDNA library (Stratagene, La Jolla, CA). The amplified fragment containing the catalytic region of pCLl00 was subsequently labeled with random primers in the presence of [s2PldCTP to be used as a probe to screen the genomic library. The genomic library was constructed from human leukocyte DNA cloned into EMBL3 (Clontech, Palo Alto, CA). Filters were prehybridized at 42 "C in 5 x SSPE (750 m~ NaCl, 5.5 n m EDTA, 48.7 m~ NaPOJ, 30% formamide, 5 x Denharts, 100 pg/ml sheared sperm DNA, and 0.1% SDS for 2 h, then further hybridized overnight after addition of the labeled fragment (2 x lo6 cpm) into the prehybridization solution. Filters were washed in 1 x SSC at 45 "C, 55 "C an finally at 60 "C for 1 h, then the filters were exposed on film (Kodak X -k a t A R ) for 2 days with intensifying screens. Positive plaques were identified, plaque-purified, and the inserts were excised, subcloned, and sequenced using Sequenase V2.0 (Stratagene).
Data Base Searches and Alignments-The amino acid sequences encoded by exon 1 (amino acids 1-1221, exon 2 (amino acids 123-1711, exon 3 (amino acids 172-2441, and exon 4 (amino acids 245-3673 were used separately in a search of the SwissPro protein data base using the FASTA program 117). Alignments between the CL100 protein and various cdc25 phosphatase proteins were done using the PILEUP command in GCG (version 7.0, Genetics Computing Group, Madison, WI).
Primer Extension Studies-Two oligonucleotides 17 nt in length complementary to nucleotide positions 210-227 (primer 210) and 234-251 (primer 234) on the cDNA were used as primers in a reverse transcription reaction (see Fig. 2). Primers (100 ng) were labelled with 50 pCi of [-y-32PlATP (>5000 Ci/mmol) and T4 DNA kinase (Life Technologies, Inc.). Poly(A)-purified RNA(1 pg) was extracted from HeLa cells as described (18) and annealed with a labelled primer (10 ng) at 75 "C for 3 min. The reaction was cooled slowly to room temperature, then incubated at 37 "C in the presence of 200 units of avian myeoloblastosis virus reverse transcriptase and 20 units of RNasin (Promega, Madison, WI) for 30 min. The reaction was organically extracted, precipitated with 0.3 M NaAc (pH 5.0), and 2.5 volumes of EtOH, resuspended 50% formamide (5 pl), then e~ectrophore~d on 8% acrylamide-urea gel in parallel with a sequencing reaction of the CLl00 gene primed by primer 210. The gel was exposed to film for 2 days with intensifying screens.
HeLu Cell Zkansfection-Human cervical carcinoma (HeLa) cells were seeded on a 35-mm culture plate and grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 10% fetal calf serum. Once the cells were grown to 70% confluence, plasmid DNA (5 pg) was transfected using 6 pl of Lipofectin according to the manufacturer's instructions (Life Technologies, Inc.). ARer 6 h of transfection, cells were washed in Dulbecco's phosphate-buffered saline, then allowed to recover in serum-free Dulbecco's modified Eagle's medium. Two days following transfection, cells were either stimulated with 20% fetal calf serum or with specific growth factors. PDGF ( A B heterodimeric subunit), EGF, and bovine FGF (Life Technologies, Inc.) were used at final concentrations of 50 ng/ml, 1 pg/ml, and 100 ng/ml, respectively. TPA (Calbi~hem) was used at a final concentration of 30 ma. All transfections were conducted on either duplicate or triplicate plates in order to minimize experimental variation.
Luciferase A s s a p A t various times following growth factor stimulation, cells were scraped in Dulbecco's phosphate-buffered saline, pelleted, then washed once in Dulbecco's phosphate-buffered saline before lysis. The luciferase assay was performed in the presence of 1 m~ ATP and luciferin (Sigma) as described previously (19). For each cell extract, light emission was measured for 20 s from duplicate reactions. A plasmid containing the CMV promoter fused to the @galactosidase gene was co-transfected in all assays such that the @-galactosidase activity could be used as an internal standard to correct for differences in transfection efficiency. Thus, luciferase activities were expressed as relative light units/@-galactosidase activity (Adm). The colorimetric assay for @-galactosidase activity was performed as described (20).
RNA BZot Analysis-Slot blot analysis was performed by denaturing 10 pg of total RNA extracted from HeLa cells, then transferring onto a Nytran membrane (Schleicher & Schiill) via vacuum suction. A human multiple tissue Northern blot was obtained commercially (Clontech). Both blots were hybridized overnight at 45 "C with the PCR-ampWed Mapping of VH-I-like PTPuses on the human genome fragment of CLl00 in 50% formamide, 4 x SSC, 5 x Denhardts, 10 pdml sheared salmon sperm DNA, and subsequently washed in a final condition of 0.1 x SSC, 0.1% SDS at 60 "C. Filters were exposed on film for 1 h at -70 "C with intensifying screens. In Situ Hybridization-A partial cDNA fragment of rat CLlOO was cloned by PCR using a pair of oligonucleotides 22 nt in length that corresponded to the human cDNA sequence at nucleotide position 92-114 (S'GCATCCCTGTGGAGAGAACCAC) and complementary to position 1165-1187 (5'AGCAGCTGGCCCATGAAGCTGA). A fragment of expected size (250 bp) was observed from a rat cortex cDNA library (a generous gift from Dr. M. J . Brownstein, National Institutes of Health), aRer 35 rounds of amplification (94-55-72 "C per cycle). Sequencing the amplified DNA fragment revealed that its proposed prutein sequence matched that of mouse pCLl00 (3CH134) perfectly over 84 amino acid residues and differed from that of the human pCLlOO by 2 residues. At the nucleic acid level, the rat clone exhibited 93 and 91% identity to mouse and human CLlOO cDNA, respectively. Thus, this DNAfragment was considered to be the rat homologue of human CLlOO and was subsequently used to synthesize riboprobes for in situ hybridization.
A partial fragment of the ERKl cDNA was cloned by PCR using a pair of primers 17 nt in length, corresponding to positions 660 bp (5'GC-CCCAGAGATCATGCT) and 1205 bp (5'CAGACCAGGTCCAAAAG) of the previously published cDNA sequence (21). Amplification proceeded as described above, and the resulting 545-bp fragment was subcloned and sequenced for verification.
In situ hybridization was performed as described previously (22). Briefly, frozen rat brains were sectioned on a cryostat at -20 "C (Bright-Hacker, F'rinceton, NJ). Coronal 10-)lm sections were thaw-mounted on polylysine-coated slides and stored at -70 "C. On the day of experimentation, sections were allowed to thaw from -70 ' C and post-fixed in 4% buffered p a r~o~a l d e h y d e for 1 h at room temperature. The sections were de-proteinated and acetylated to reduce background as described previously (22). The antisense riboprobe was labeled with 36S-UTP and applied to each slide a t an approximate concentration of 2 million dpm.
The sections were hybridized at 55 "C overnight, treated with RNase A (100 pg/ml) for 1 h, then washed in 0.1 x SSC, 0.1% SDS at 65 "C for 1 h. The sections were initially exposed to film (Kodak XAR) and later dipped in emulsion for analysis at higher resolution. The sections were either pre-RNase treated or hybridized using a sense strand cRNA probe as a control.

RESULTS
Screening 800,000 plaques of the human genomic library with the amplified cDNA of CLlOO at moderate stringency yielded 21 positive signals. Upon further characterization of these clones, the positive plaques were identified as overlap ping or sister clones of four PTPase genes. The four genes included our cognate gene CL100, PAC-1, and two novel genes that were most similar to the VH-1 subclass of PTPases (Table  I), suggesting that a large family of VH-1-like PTPases exist. Searches with both exon 1 and exon 2 yielded a significant alignment with the cdc25 phosphatase. This was unexpected as neither exon 1 nor exon 2 contained the catalytic residues of CL100. Upon alignment of the protein sequence encoded by exon 1 and exon 2 with various members of the cdc25 family, two regions of sequence similarity were identified (blue bores in  (red box, Fig. lB). However, there were no residues within the N terminus of pCLl00 that resembled the catalytic domain of phosphatases. In fact, the catalytic domain of pCLl00 was situated on the C terminus in a manner such that when the entire protein sequence of CLlOO was aligned with members of the cdc25 family, the catalytic regions of all proteins were preferentially aligned (red box, Fig. 1C). Thus, it appeared that pCLl00 possessed two domains with sequence similarity to cdc25. The first, at the C terminus, contained the active site motif common to all PTPases, and the second domain, which we termed CH2 (for the gdc-25 pomology domain similarity to the region surrounding cdc25 active site. These

Danscriptionul Start Site of CLlOO
Ge-The nucleotide sequence of the CLlOO gene is shown in Fig. 2. The intron splice junctions were sequenced for each exon, and an additional 1 kb of 5"flanking sequence was characterized. All splice junctions conformed to the donor and acceptor consensus. The ATl"A consensus that confers mRNA instability is repeated near the translation termination site (Fig. 2). The sequence immediately upstream from the TATA box consensus to position -150 was extremely high in GC content.
The transcriptional start site was identified by primer extension analysis using primers complementary to nucleotide position 210 and 234 of the cDNA (Fig. 3A). A control reaction Stimulation with calf serum increased the expression of CLlOO mRNA rapidly and transiently in HeLa cells. Consistent with the previous data obtained from the mouse homologue 3CH134 in Swiss 3T3 cells (7,8), the induction of CLlOO mRNA was maximal at 30 min and decreased afterwards such that mRNA levels were back to base line by 3 h (Fig. 4A). Cyclohexamide produced a prolonged rise in CLlOO mRNA content as reported previously for 3CH134 mRNA (7,8). Thus, human CLlOO gene and murine 3CH134 are similarly activated in an immediate-early manner and are independent of novel protein synthesis.
Since these observations indicate that a region on the CLlOO gene confers responsiveness to mitogen stimulation, we trans- fected HeLa cells transiently with gene fusion constructs to partially characterize this segment of the CLlOO promoter. Three fusion constructs were made with varying amounta of the CLlOO 5"flanking sequence and the luciferase reporter gene (Fig. 4B). The 1.6-kbXbaI fragment spanning -1500 nt to +lo0 nt of exon 1 responded to serum stimulation i n HeLa cells (Fig. 4 0 PDGF ( A B , 50 ng/ml), EGF (1 pglml), fetal calf serum (20%), and TPA (30 m) were added to serum-deprived cells for 2 h prior to harvest. Luciferase activity was normalized to a co-transfected internal standard as in Fig. 4C. B, endogenous CLlOO transcription activity was monitored during the treatment described above. HeLa cells were stimulated for 30 min by the indicated agents and total RNA (10 pg) were used for slot blot analysis. basal and stimulated levels were higher than the 1.6-kb fragment. Luciferase activity increased by approximately 2-fold in both 1.6-and 0.8-kb constructs 2 h after serum stimulation. Basal level of transcription of CL100, as measured by luciferase activity, was reduced 6-fold when a control construct consisting of the 1.6-kb fragment oriented in antisense was transfected into HeLa cells.
The 0.8-kb sense construct was subsequently used to determine the effects of a variety of growth factors. Stimulation with FGF or PDGF for 2 h appeared ineffective in eliciting a consistent response from the gene fusion construct (Fig. 5 A ) . Transcription was marginally increased by EGF (60% over control levels), whereas serum and TPAincreased the activity of CLlOO promoter by approximately 1.5-and 3-fold, respectively. This response was compared with the transcriptional activity of the endogenous CLlOO gene by measuring CLlOO mRNAcontent in HeLa cells 30 min after stimulation (Fig. 5B). The endogenous CLlOO transcription activity, as determined by CLlOO mRNA content, was generally consistent with that of the transfected fusion construct (Fig. 5B). Thus, the elements necessary to mediate "PA and serum responses on this gene reside within 800 bp of the transcription initiation site.
CLlOO mRNA Distribution-The distribution pattern of CLlOO mRNA in human tissues was determined by screening a Northern blot with the human CLlOO cDNA fragment. The expression of CLlOO mRNA was widespread among peripheral tissues (Fig. 6). High levels of mRNA were detected in lung, liver, placenta, and pancreas, whereas moderate levels were found in heart and skeletal muscle. Lower levels of gene expression was observed in brain and kidney upon longer exposure (data not shown).
Since the transcriptional activity of CLlOO gene appears to be modulated by mitotic factors, we asked whether the expression of CLlOO was restricted to dividing cells or whether it was expressed in terminally differentiated neurons as well. We initially confirmed the expression in the CNS by cloning a rat homologue of CLlOO from an adult rat brain library using primers based on human cDNA sequence (see "Materials and Methods" for further detail). Subsequent in situ hybridization analysis using the amplified rat CLlOO cDNA on rat brain sections revealed that mature neurons express CLlOO mRNAin discrete regions (Fig. 7). High levels of expression was found in the cingulate gyrus within the retrosplenial cortex, ventral, and medial divisions of the anterior thalamus and the medial geniculate nucleus (Fig. 7, A X ) . Parietal and temporal cortex also expressed moderate levels of this transcript. Nuclei associated with the limbic circuit, including the anteroventral and the anteromedial thalamic nuclei, the cingulate gyrus, and the hippocampus were intensely labeled. Analysis at higher resolution was performed from emulsion-dipped slides to provide evidence that the neurons in the suprachiasmatic nucleus, the primary circadian rhythm generator in most mammals, and the medial aspect of the cingulate gyrus expressed CLlOO mRNA (Fig. 7, D and E ) . A section was counterstained with cresyl violet to identify silver grains on the lightly stained neurons (black arrows) but absent on smaller and darker glial cells (Fig. 7F).
pCLl00 has recently been shown to dephosphorylate ERKl in vitro, raising the possibility that ERKl may be the endogenous substrate for pCL100. We asked whether these two proteins are associated in vivo by determining the distribution pattern of CLlOO and ERKl mRNAs within the brain. Hybridization of adjacent sections with CLlOO and ERKl antisense riboprobes revealed that ERKl mRNA was expressed in all areas of the brain (Fig. a). We found no exclusive correspondence between CLlOO and ERKl distribution in the rat brain, although CLlOO mRNA was expressed in a subset of areas that were positive for ERK1.

DISCUSSION
Transcription of CLlOO and PAC-1 genes appear to be induced rapidly during cell stimulation. Since this feature may be commonly exhibited by the mammalian VH-1-like subfamily of PTPases, characterization of a VH-1-like PTPase gene and its promoter activity may help us understand how this class of

3602
Characterization of PTPase CLlOO Gene RG. 7. In situ hybridization analysis of CLlOO mRNA in the rat brain. Rat CLlOO cDNA cloned by PCR was used to synthesize a SsS-labeled riboprobe. A X , coronal sections of rostral, middle, and caudal rat brain in low magnification. Analysis of emulsion-dipped sections a t higher magnification reveals mRNAsignal in the suprachiasmatic nucleus (D) and from single cells in the cingulate cortex ( E ) . It is apparent after counter-staining with cresyl violet that grains surround the lightly stained neurons (arrows in F ) but   Interestingly, we observe a second region of sequence similarity to cdc25, the CH2 domain, that is encoded by exons 1 and 2 of the CLlOO gene. This domain aligns well with the regions surrounding the catalytic domain of cdc25. We suspect that this N-terminal alignment was not detected previously, since the computer program preferentially aligned the PTF'ase motif of CLlOO with the PTPase domain of cdc25. The alignment of the CH2 domain in pCLl00 to cdc25 is interrupted into two segments by the existence of the cdc25 active site. The first segment is encoded on exon 1 of the CLlOO gene, whereas the second segment spans the intron splice junction of exons 1 and 2. Thus, the amino acid residues Lys and Glu of the conserved LKGGY motif are separated on different exons. It is conceivable that these 2 residues are conserved by the virtue of the intron splicing mechanism that requires a specific donor and acceptor sequences on exons 1 and 2, respectively.
The CH2-like sequence is also present at the N terminus of PAC-1 cDNA, suggesting that the CH2 domain is shared by several members of this PTPase subfamily. It is noteworthy that pVHR, a short VH-1-like protein isolated from human fibroblast, lacks the CH2 domain (3). Perhaps this clone d e h e s a subclass of VH-1-like PTPases that lacks the N-terminal cdc25 homology domain and may therefore possess functions that are distinct from CH2-containing PTPases.
The functional significance of the CH2 domain of pCLl00 is presently unclear. The region of homology corresponds to an amino acid stretch flanking the active site of cdc25, a cell cycle protein that dephosphorylates cdc2 at juxtaposed residues Thr14 and T y r Z 5 (13, 29, 30). It is conceivable that the CH2 domain of pCLl00 functions to specify substrates for catalysis. Although cdc25 is a born fide dual specificity phosphatase, the CH2 domain is not likely to confer dual substrate specificity on pCL100, since the vaccinia phosphatase VH-1 lacks this domain but still retains its ability to catalyze phosphotyrosines and phosphoserines. Perhaps the cdc25 homology domain acts to refine the substrate recognition process. The PTPases that possess this region may therefore be more "selective" for their substrates, whereas those that lack it may exhibit activity across a broader spectrum of substrates. This may perhaps explain why pCLl00 and its murine homologue exhibit low catalytic activity against artificial substrates (4, 31). In contrast, pCLl00 readily exhibits tyrosine phosphatase activity on extracellular signal-regulated kinase (ERK) (31) and dual phosphatase activity on ERK when phosphorylated by ERK kinase (MEK) (9).
Alternatively, this region of pCLl00 may be analogous to the noncatalytic SIT homology domains fSH2 and SH3) in that they are involved in the recruitment of other proteins (32,331. Association with other proteins may permit nuclear translocation of this class of FTPases or even regulate its half-life. Several PTPases are known to possess a second domain that specifies nuclear or cytoskeletal localization (34,35). Whether the cdc25 homology region of VH-1-like PTPases serves as a separate functional domain remains to be determined in future studies.
We determined that 800 bp of 5"flanking sequence contain the elements necessary for response to serum stimulation. The transcriptional response from the 1.6-kb construct was consistently lower during serum and TPA stimulation. It is therefore conceivable that a negative regulatory element resides between -1600 and -801. During the initial characterization of mouse CLlOO (3CH1341, Charles et al. (6) observed that cyclohexamide exaggerates and prolongs the increase of this transcript during serum stimulation. We observed a similar effect on the CLlOO gene, suggesting that like many other immediate early transcripts, activation of CLlOO gene is relatively independent of protein synthesis. Although our transfection studies suggest that 800 bp flanking the promoter encodes most of the elements necessary for response to mitogens, we do not find an serum response element-like consensus on this gene within -1.5 kb of the promoter. Thus, it is likely that the response to serum stimulation is mediated by a mechanism that involves a transcriptional activator other than the serum response factor or ternary complex factor.
The two CREs situated upstream of the promoter are of interest in this context since the CREBIATF family of DNAbinding proteins are known to be regulated primarily by a phosphorylation event (see Ref. 36 for review). Consistent with this notion, PAC 1 expression has been shown to be increased upon stimulation with Tax (37), a protein encoded by the X region of HTLV-1 gene that can transactivate through a CRE site (38). Further work is required to determine the involvement of one or both CREs in eliciting transcriptional response to mitogens.
The levels of CLl00 mRNA were high in several human tissues, although expression was widespread. Since CLlOO expression in Swiss 3"3 cells is inducible by a variety of mitogens (6), it appears likely that this gene product is involved in mitotic signaling. However, we suspect that CLlOO is important for signal transduction during a variety of other conditions. We provide evidence here that terminally differentiated neurons in the brain express CLlOO mRNA. The discrete localization of this transcript within the rat brain suggests that only a subpopulation of neurons utilizes pCL100-mediated signaling.
Since no obvious neurotransmitter is common to all the regions expressing CLlOO mRNA, it is likely that several transmitters' factors can elicit CLlOO expression in a region-specific manner.
Finally, recent reports have indicated that ERKl may be the endogenous substrate for CLlOO (9, 31). ERKl is abundantly expressed in the brain, multiphosphorylated on serine, tyrosine, and on threonine residues (39, 40), and reported to be dephosphorylated by pCLl00 in vitro (9,31). We examined the relationship between ERKl and CLlOO by in situ hybridization and found that the distribution of ERKl mRNA in the brain correlates poorly with CL100. ERKl mRNA expression is widespread in the brain whereas CLlOO mRNA is localized in a subset of these areas. The distribution of ERKl and ERK2 assessed by a Northern blot analysis is consistent with this notion (391, suggesting that ERKs cannot be paired exclusively with CL100. Perhaps other VH-1-like PTPases exist and catalyze ERK in specific areas of the brain or conversely an unobserved substrate for CLlOO exists in the brain.