Human SPA-1 Gene Product Selectively Expressed in Lymphoid Tissues Is a Specific GTPase-activating Protein for Rap1 and Rap2

Mouse Spa-1 gene with a region homologous to the human rap1GAP gene is transcriptionally induced in the lymphocytes by mitogenic stimulation. Herein we have cloned a cDNA for its human counterpart. SPA-1 cDNA encodes a 130-kDa protein (p130 SPA-1 ) consisting of proline-rich regions and rap1GAP-related domain followed by a coiled-coil stretch. Baculovirally expressed p130 SPA-1 exhibited GTPase-activating protein (GAP) activity for Rap1 and Rap2, but not for Ras, Rho, Cdc42, Rac, and Ran, with comparable specific activity to the rap1GAP gene product (p85/95 rap1GAP ). In the cells, p130 SPA-1 was mostly localized at the perinuclear membranous region co-localizing with Rap1 and Rap2. Expression of SPA-1 and rap1GAP genes tended to be segregate in various tissues, lymphoid tissues expressing abundantSPA-1 transcript without rap1GAP, while those such as brain, kidney, and pancreas exhibiting rap1GAPmRNA with little SPA-1. Promyelocytic HL-60 cells, which expressed p130 SPA-1 with little p85/95 rap1GAP in uninduced state, showed progressive decline in p130 SPA-1 and conversely drastic increase in p85/95 rap1GAP as they ceased from proliferation and differentiated into macrophages by 12-O-tetradecanoylphorbol-13-acetate. These results suggested that products of SPA-1 and rap1GAPgenes, albeit comparable GAP activity for Rap1 and Rap2, functioned in the distinct contexts depending on cell types and/or states.

Stimulation of immunocompetent lymphocytes via antigen receptors induces a series of signal transduction including activation of phosphatidylinositol-phospholipase C␥ leading to the phosphatidylinositol turnover as well as Ras-signaling pathway (see Ref. 1, for review), and the cells undergo extensive proliferation called clonal expansion. The activation of p21 ras is initiated by tyrosine kinases associated with receptors (2), which can be achieved in principle by two ways (3). The receptors phosphorylated by the kinases may interact with an adaptor molecule Grb2, either directly or indirectly via a second adaptor such as Shc, which binds and recruits mSos to the membrane (4 -6). mSos functions as a guanosine nucleotide exchange factor catalyzing the conversion of Ras-GDP to Ras-GTP (7). Alternatively, receptor-associated kinases may downregulate the function of Ras GTPase-activating protein (GAP), 1 thereby inhibiting the hydrolytic conversion of Ras-GTP to Ras-GDP (8,9). Recently, a docking protein p62 dok , which served as an excellent substrate for the cytosolic tyrosine kinases and was able to bind to p120 Ras-GAP , has been cloned (10,11). Vav, exclusively expressed in hematopoietic tissues (12), has been also shown to participate in the antigen receptorinduced signaling (13) and to play a crucial role in the development and activation of lymphocytes (14). Vav has been recently shown to function as a guanosine nucleotide exchange factor for Rac (15), a small G protein of Ras family required for the Ras-mediated signaling (16).
Rap1 is another close member of Ras family small GTPases (17). Unlike Ras, Rap1 does not have oncogenic potential, but rather is shown to inhibit Ras-mediated oncogenic or growthpromoting activity (18,19). Since Rap1 shares the same effector domain with Ras and can interact with several effector molecules of Ras pathway, it has been suggested that the Ras-antagonizing effects of Rap1 was due to the competitive inhibition for the Ras effectors such as Raf-1 and p120 Ras-GAP (20 -22). Rap1 is also implicated to be involved in the cAMPinduced inhibition of Ras-MAP kinase pathway (23). Recently, C3G, originally identified as a Crk-binding protein (24), has been shown to function as a specific guanosine nucleotide exchange factor for Rap1 (25) and suggested to be involved in the antigen receptor-induced signaling (26). On the other hand, GAP specific for Rap1 (rap1GAP) was previously identified (27,28) and molecularly cloned from the human brain (29). The rap1GAP, however, was reported to be expressed mostly in the brain with little expression in the lymphoid tissues where the highest Rap1 GAP activity was detected (29).
We have previously cloned a new gene, Spa-1, transcriptionally induced in the mouse lymphocytes by mitogenic stimulation (30). The original Spa-1 cDNA contained an ORF encoding 693 residues, the N-terminal portion (190 residues, ⌬N GRD) of which was highly homologous to human rap1GAP. The GST fusion protein of ⌬N GRD was indicated to show GAP activity for Rap1 and Ran, although activity of the complete protein remained to be seen (30). It was recently found that the original Spa-1 cDNA clone had 1-base deletion in its N terminus, and * The work was supported in part by grants from Ministry of Education, Japanese Government. 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 the corrected cDNA sequence was revealed to contain a N terminally extended ORF longer than originally reported (1,038 residues). In the present study, we have cloned a human SPA-1 cDNA that contained an ORF encoding 1,042 residues 90% identical to that of the revised mouse cDNA. We now show that the SPA-1 gene product in its complete form (130 kDa) exhibits a specific GAP activity for Rap1 and Rap2, but not for Ran or other small GTPases. It was indicated that the expression patterns of SPA-1 and rap1GAP genes were quite distinct and tended to be segregate in various human tissues, in that lymphoid tissues abundantly expressed SPA-1 transcript in the absence of rap1GAP, while tissues such as brain, kidney, and pancreas strongly exhibited rap1GAP transcript with little SPA-1 expression. Furthermore, the expression of p130 SPA-1 and p85/95 rap1GAP was regulated in an opposite manner during the growth arrest and differentiation of promyelocytic HL60 cells into macrophages, raising a possibility that these proteins albeit the comparable Rap1 GAP activity exhibited distinct functions in the cells.

EXPERIMENTAL PROCEDURES
cDNA Cloning-Poly(A)(ϩ) RNA was prepared from the human peripheral blood lymphocytes (PBL) stimulated with phytohemagglutinin (2 g/ml) and TPA (10 ng/ml) for 48 h and the cDNA library was made in a pcDL-SR␣296 vector. Using a 32 P-labeled mouse full-length Spa-1 cDNA as a probe, a cDNA was isolated by colony hybridization with high stringency condition. The insert of 3.6 kilobase pairs was subcloned into pBluescript KS(ϩ) vector (Stratagene, CA) and sequenced.
Plasmid Construction-To introduce a BamHI site just before the initiation codon, the fragment encoding the N-terminal region of human SPA-1 (residues 285 to 805) was amplified by polymerase chain reaction using the sense (5Ј-ATGGATCCACAGAGCATGCCCATGT) and the antisense (5Ј-CCGAGCTCACACACAAAGCCAGGTGC) primers. After digestion with BamHI and SacI, the fragment was ligated into the BamHI-XhoI site of pSP73 vector (Promega Corp., WI) together with the SacI-XhoI fragment of the pBluescript KS(ϩ) SPA-1 cDNA. The C-terminally 6 ϫ histidine (His)-tagged pBac SPA-1 was constructed as follows. After removing the NcoI-XhoI fragment from pSP SPA-1, the NcoI-BstXI fragment of pSP SPA-1 and the BstXI-XhoI synthetic oligonucleotides including 3Ј His and the stop codon (5Ј-CTGG-GCTCACCCACCGCCGACCTGGCGAATTCGATGCATCACCATCACA-TCACTGAC and 5Ј-TCGAGTCAGTGATGGTGATGGTGATGCATC-GAATTCGCCAGGTCGGCGGTGGGTGAGCCCAGCTGC) were ligated into the NcoI-XhoI site of the pSP SPA-1. Then the BamHI-XhoI fragment of pSP SPA-1-His was subcloned into the pBacPAK8 (CLON-TECH). To generate the C terminally His-tagged pBac rap1GAP, pAcRG9T (provided by Dr. B. Rubinfeld, Onyx Inc., Emeryville, CA) was digested with BglII and PvuII, and the fragment was ligated into the BamHI-XhoI site of pBacPAK8 together with the PvuII-XhoI synthetic oligonucleotides including 3Ј His and the stop codon 5Ј-CTGGA-AGCATCTGAGCAGCACATGCCCCAGCTGGGCTGTAGGAATTCGA-TGCATCACCATCACCATCACTGAC and 5Ј-TCGAGTCAGTGATGGT-GATGGTGATGCATCGAATTCCTACAGCCCAGCTGGGGCATGTGC-TGCTCAGATGCTTCCAG). pNeo SR␣ SPA-1 was constructed by subcloning the BamHI-XhoI fragment of pSP SPA-1 into the BamHI-XhoI site of pNeoSR␣. N terminally His-tagged mouse E1, provided by Dr. F. Yamao, National Institute of Genetics, Mishima, Japan, was subcloned into SmaI-EcoRI sites of a pVL1393 baculoviral transfer vector (Invitrogen Corp.). All constructs were confirmed by sequencing.
In Vitro Transcription/Translation--Using pSP SPA-1 as a template, in vitro transcription and translation was done by the TNTcoupled wheat germ extract system (Promega) with [ 35 S]methionine according to the manufacture's protocol.
Expression and Purification of Proteins-cDNAs for Rap1A and Ras were provided by Dr. H. Maruta, Ludwig Institute for Cancer Research, Melbourne, Australia; cDNA for Ran by Dr. M. Dasso, National Institutes of Health, Bethesda, MD; cDNAs for RhoA, Cdc42, and Rac by Dr. S. Narumiya, Kyoto University, Kyoto, Japan; and cDNA for Rap2A was supplied from Dr. J. Gunzburg, INSERUM, France. All small G proteins were expressed as GST fusion proteins and purified from Escherichia coli by single step purification using glutathione-Sepharose 4B (Pharmacia Biotech, Sweden) as described previously (30). The His-tagged SPA-1 and rap1GAP proteins were expressed by the Sf 9/ baculovirus expression system. To generate recombinant virus, 0.5 g of pBacPAK8 transfer vector containing each cDNA was cotransfected with 5 l of BacPAK6 viral DNA (Bsu36I digest, CLONTECH) into insect Sf 9 cells. Recombinant virus was then isolated by plaque purification. 4 ϫ 10 5 /ml Sf 9 cells were infected with the recombinant virus at 1-5 plaque forming units/cell and incubated at 27°C. After 48 h, the cells were harvested, washed with phosphate-buffered saline, and lysed with lysis buffer (20 mM Tris-HCl, pH 8.0, 0.5% Nonidet P-40, 0.5% sodium deoxycholate). The recombinant proteins were purified by single step purification using Ni-NTA-agarose (QIAGEN Inc., Chatsworth, CA), dialyzed against 50 mM Na-PO 4 , pH 7.0, containing 1 mM dithiothreitol, aliquoted, rapidly frozen in liquid N 2 , and then stored at Ϫ80°C.
Assay for GTPase Activity-[␥-32 P]GTP was loaded to the GST fusion proteins of various small GTPases as follows. Small G proteins (5 g) were added to 20 l of 4.4 nM [␥-32 P]GTP (ICN Biochemicals Inc.) diluted with exchange buffer (50 mM Tris-HCl, pH 7.5, 2.5 mM EDTA, 1 mM dithiothreitol, 0.5 mg/ml bovine serum albumin) and the mixtures were incubated at 30°C for 30 min for Rap1A, Rap2A, Ran, and Ras, or at room temperature for 10 min for RhoA, Rac1, and Cdc42. After adding MgCl 2 to final concentration of 15 mM, 20 l of the [␥-32 P]GTPloaded proteins were added to the GAP samples (100 l) in exchange buffer containing 15 mM MgCl 2 or bufffer control. Samples were incubated at 30°C (Rap1A, Rap2A, Ran, and Ras) or at 24°C (RhoA, Rac1, and Cdc42) for indicated periods. Reaction was stopped by adding 1 ml of ice-cold 50 mM Tris-HCl, pH 7.5, containing 15 mM MgCl 2 , followed by rapid vacuum filtration through NC45 nitrocellulose membrane filter. The filters were washed three times with 4 ml of the same buffer, dried, and the radioactivity remaining on the filter was counted by liquid scintillation counter.
Antibodies-Rabbit antiserum was raised against the N-terminal region of mouse Spa-1 (SpaN, residues 245 to 801) by immunizing with a recombinant thioredoxin (Trx)-SpaN fusion protein. The antiserum was affinity purified using the immobilized Trx-SpaN fusion protein on Affi-Gel 10 (Bio-Rad) followed by extensive adsorption with the immobilized Trx. Anti-rap1GAP antiserum was kindly provided by Dr. P. Polakis, Onyx Inc., Emeryville, CA, and anti-␤COP antiserum was a gift from Dr. J. Donaldson, National Institutes of Health, Bethesda, MD. Anti-Rap1, anti-Rap2, and anti-calnexin antibodies were purchased from the Transduction Laboratories Inc., Lexington, KY.
Immunoprecipitation, Immunoblotting, and Northern Blotting-Lysates of cell lines or Sf9 cells infected with recombinant baculovirus were incubated with 2 g/ml purified anti-SpaN antibody followed by precipitation with protein A-Sepharose 4B (Pharmacia). The beads were washed, boiled in SDS sample buffer, and subjected to SDS-PAGE. The gels were transferred to the polyvinylidene difluoride membranes, probed with the indicated antibodies, and visualized using horseradish peroxidase-coupled ECL detection system. Northern blot analysis was performed as described previously (30).
Transfection and Immunostaining-HeLa cells were transfected with a SPA-1 cDNA in pSR␣ expression vector with a CaPO 4 precipitation method. Two days later, the cells were rinsed with phosphatebuffered saline, fixed with 3% folmaldehide, permealized with 0.5% Nonidet P-40 in phosphate-buffered saline, blocked with 3% bovine serum albumin in phosphate-buffered saline, and then stained with various antibodies followed by fluorescein isothiocyanate-conjugated second antibodies. They were examined with a confocal laser microscopy.
Cell Cultures-HL60 cells were obtained from the American Type Culture Collection (CCL-240) and maintained in RPMI 1640 medium (Life Technologies, Inc., Rockville, MD) supplemented with 20% fetal calf serum and antibiotics. For the induction of differentiation, HL60 cells were plated at 1 ϫ 10 6 cells/ml and exposed with 1 ϫ 10 Ϫ7 M TPA (Sigma) for the indicated periods of time. Morphological examination was done by Wright-Giemza staining, and cell cycle analysis was performed using a flow cytometry as described before (30).
Subcellular Fractionation-5 ϫ 10 6 cells were resuspended in 500 l of hypotonic buffer (10 mM HEPES pH 7.5, 5 mM KCl, 2 mM MgCl 2 ), incubated at 4°C for 5 min, and then lysed with the Dounce homogenizer. After centrifugation at 2,500 rpm for 5 min to remove nuclear fraction, the supernatant was separated into the S100 and P100 fractions by ultracentrifugation at 100,000 ϫ g for 1 h.

RESULTS
Isolation of Human SPA-1 cDNA-A cDNA of 3.6 kilobase pairs long has been cloned from the cDNA library of phytohemagglutinin-stimulated human PBL using a murine Spa-1 cDNA probe. Despite the extremely high overall homology to the murine Spa-1 cDNA (83% identical at the nucleotides lev-el), the human cDNA contained an ORF (nucleotides 297 to 3425, 1,042 residues) much longer than that of previously reported mouse Spa-1 cDNA (nucleotides 1194 to 3277). This prompted us to re-examine the original mouse Spa-1 cDNA sequence carefully, and by the sequencing of independent cDNA as well as genomic clones the original mouse Spa-1 cDNA clone was found out to have one base deletion at nucleotide 378. Inclusion of the missing nucleotide negated the previous initiation codon and resulted in the N-terminal extension of the ORF retaining the reading frame, now encoding 1,038 residues. Alignment of the human and revised murine cDNAs revealed that the deduced amino acid sequences of the ORFs were 90% identical (Fig. 1A). A N-terminal region (residues 192 to 539) was highly homologous to human rap1GAP (GRD), 38% of residues being identical, and almost perfectly conserved between human and mouse (98% identical). The region covered entire reported catalytic region of rap1GAP (33). N terminally to the GRD was a proline-rich region including possibly SH3binding motifs (PPXXP), and a coiled-coil stretch with hydrophobic heptad repeats was noted at the C terminus following the probable PEST sequences (Fig. 1B). A leucine zipper-like motif (Lx 6 Lx 6 Lx 6 Ex 6 Lx 6 Lx 6 L) at the C terminus was also completely conserved between the two species. There are a number of possible phosphorylation sites, including those by Cdk2 in the N-terminal region.
SPA-1 Encodes a 130-kDa Protein-In vitro transcription/ translation of the full-length SPA-1 cDNA resulted in the single protein with relative molecular mass 130 kDa, comparable to the calculated molecular mass 112 kDa ( Fig. 2A). An identical protein was produced by the baculoviral expression system, and the recombinant human p130 SPA-1 was found to be cross-reactive to anti-mouse Spa-1 antibody (SpaN) by immunoprecipitation/immunoblotting (Fig. 2B). When expressed by baculovirus, a minor C-terminally truncated 85-kDa product devoid of His-tag was usually detected by SpaN (Fig. 2B).
Using the antibody, both human and mouse endogenous SPA-1 proteins could be indeed identified to be 130-kDa proteins in HL60 promyelocytic and 70z/3 pre-B cell lines (Fig. 2B).
p130 SPA-1 Is a Selective GAP for Rap1 and Rap2-Using the baculovirally expressed recombinant protein (Fig. 3A), we now examined GAP activity of the SPA-1 in its complete form (p130 SPA-1 ) for a series of small G proteins. As summarized in Fig. 3B, p130 SPA-1 potently stimulated the GTPase activity of Rap1A and Rap2A, both of which showed negligible intrinsic GTPase activity, without affecting that of other small G pro- FIG. 2. Human SPA-1 gene encodes a 130-kDa protein  (p130 SPA-1 ). A, human and mouse Spa-1 cDNAs were transcribed and translated in vitro in the presence of [ 35 S]methionine as described under "Experimental Procedures" and analyzed by the 6.5% SDS-PAGE followed by autoradiography. B, cell lysates of human HL60 promyelocytic cells and mouse 70z/3 pro-B cells as well as Sf9 cells infected with baculovirus containing human SPA-1 cDNA were immunoprecipitated with preimmune rabbit IgG or purified anti-mouse SpaN antibody using protein A-Sepharose 4B, electrophoresed in SDS-PAGE, and then immunoblotted with anti-SpaN antibody. teins such as Ras, Rho, Cdc42, and Rac. Unlike a GST-⌬N GRD (residues 346 to 539) fusion protein of mouse Spa-1, p130 SPA-1 failed to show any significant Ran GAP activity either. A control protein E1 similarly expressed in the baculoviral expression system showed no detectable GAP activity at all for any of the GTPases. We then compared the specific GAP activity of p130 SPA-1 with that of the baculovirally expressed p85/ 95 rap1GAP (Fig. 3A). As shown in Fig. 3C, the specific GAP activity of p130 SPA-1 for Rap1A was nearly equivalent to that of p85/95 rap1GAP , which was some 40 times more active than the previously reported GST-⌬N GRD fusion protein (30). Interestingly, p85/95 rap1GAP appeared to be significantly more effective for Rap2A than Rap1A in our hands, while p130 SPA-1 showed comparable activity against both (Fig. 3, C and D).
p130 spa-1 Is Localized Mostly in the Perinuclear Insoluble Fraction: Probable Co-localization with Rap1 and Rap2-Subcellular fractionation analysis was done to see the intracellular localization of p130 SPA-1 using HL60 cells. As shown in Fig. 4, the vast majority of p130 SPA-1 was detected in the insoluble membrane fraction (P100), although a minor portion was also found in the soluble fraction (S100). In the parallel experiments, both Rap1 and Rap2 were exclusively detected in the P100 fraction, as was endoplasmic reticulum-resident protein calnexin. A significant portion of p130 SPA-1 was also associated with the nuclear fraction, but it was most likely due to the contamination of undisrupted cells and/or insoluble cytosolic microsomes because a endoplasmic reticulum-resident protein calnexin was similarly detected in the preparation (see also below). Since immunostaining analysis was rather difficult in the hematopoietic cells, further study was performed by the transient expression of SPA-1 cDNA into HeLa cells, which exhibited undetectable endogenous SPA-1 transcript (data not shown). As shown in Fig. 5, the expression of p130 SPA-1 was detected with anti-SpaN in the cytosol in a speckled pattern with predominant localization in the perinuclear region (B) possibly including juxtanuclear Golgi area stained with anti-␤COP (E), and no staining was detected at all in the nucleus. This particular pattern did not seem to simply reflect overexpression of the protein, since predominant perinuclear staining was similarly observed in the cells only weakly expressing the protein (B). The staining pattern of p130 SPA-1 was quite similar to those of Rap1 and Rap2 as well as calnexin (Fig. 5, C, D, and  F). Taken together, it was suggested that p130 SPA-1 was in the cytosol associated mostly, if not exclusively, with the perinuclear endoplasmic reticulum and Golgi region, probably colocalized with Rap1 and Rap2.
Expression of SPA-1 Transcript in the Normal Human Tissues: Segregate Distribution from rap1GAP-Since it was indicated that p130 SPA-1 functioned as GAP for Rap1 with comparable specific activity with p85/95 rap1GAP , expression of both Lysates of HL60 cells were separated into nuclear, soluble (S100), and insoluble (P100) fractions as described under "Experimental Procedures," and each fraction along with unseparated lysate equivalent to the 10 6 cells was subjected to the immunoblotting analysis using anti-SpaN, anti-Rap1A, anti-Rap2, and anti-calnexin antibodies. Each specific band is indicated by an arrow.
transcripts was then compared in various normal human tissues. SPA-1 mRNA was expressed quite abundantly in the lymphoid tissues such as thymus, spleen, and PBL (Fig. 6A), conforming to the results in mice (30). It was also expressed in some other tissues albeit by far lower levels. It should be noted, however, that the preparations of these tissues such as intestine and colon were inevitably contaminated by the associated lymphoid tissues as well as PBL expressing abundant SPA-1 transcript, possibly leading to the overestimation. In the brain and kidney, on the other hand, SPA-1 transcript was hardly detected. In contrast, rap1GAP mRNA was significantly expressed in the brain, kidney, and strongly in pancreas, while it was undetectable in the lymphoid tissues and PBL (Fig. 6A).
Although not shown, among the nervous system, rap1GAP transcript was most abundant in the cerebral cortex with much less expression in the spinal cord, while significant SPA-1 mRNA was detected in the spinal cord with undetectable expression in the cerebral cortex. These results indicated not only that SPA-1 and rap1GAP showed distinct tissue distribution patterns, but also that their expression tended to segregate each other, the lymphoid system exclusively expressing SPA-1 while the nervous system and kidney mainly rap1GAP. To confirm the above notion, various human leukemia/lymphoma cells were also examined. As shown in Fig. 6B, SPA-1 transcript was expressed in all lines at varying degrees, while that of rap1GAP was hardly detected in any of them. Among them, SP49 lymphoma cells which had t(11q13;14q32) chromosomal translocation at the BCL-1 site expressed unusually high level of SPA-1 mRNA (Fig. 6B) as well as p130 SPA-1 (data not shown), which is interesting because human SPA-1 locus is mapped at the 11q13 in the neighborhood close to BCL-1 (31).
Opposite Regulation of p130 SPA-1 and p85/95 rap1GAP Expression during the Growth Arrest followed by Differentiation into Macrophages of Promyelocytic HL60 Cells-We have previously suggested that the expression of murine Spa-1 was regulated in association with the proliferative state of cells (30), and therefore intended to examine the possible regulation of p130 SPA-1 during the cellular growth and/or differentiation in comparison with p85/95 rap1GAP in a defined system. Human promyelocytic HL60 line is induced to differentiate into macrophage by TPA, which is manifested by gradual morphological changes from floating round cells to firmly adherent large spread cells (Fig. 7A). The process is preceded by the rapid arrest of cell cycle at G 1/0 stage (Fig. 7A). As shown in Fig. 7B, uninduced proliferating HL60 expressed a significant level of p130 SPA-1 . When the cells were induced by TPA, however, the level of p130 SPA-1 was progressively reduced after day 2 and remained very low thereafter. In quite a contrast, the expression of p85/95 rap1GAP , which was almost undetectable in uninduced cells, was progressively increased as cells differentiate into macrophages. On the other hand, the expression of both Rap1 and Rap2 proteins were roughly steady during the differentiation process of HL60 cells, although the latter tended to be transiently increased. Northern blot analysis revealed that the SPA-1 transcript was rapidly reduced as early as day 1 after the TPA stimulation (Fig. 7C). Although it tended to show slight rebound on day 2, the level remained to be lower level later on, indicating that the down-regulation of p130 SPA-1 was at least partly due to the transcriptional repression. In contrast, the transcript of rap1GAP, which was hardly detected in the uninduced HL60 cells, was potently induced by day 2 and sustained thereafter (Fig. 7C). Thus TPA stimulation induced clearly opposite effects on the transcription of SPA-1 and rap1GAP genes and subsequent expression of their products in HL60 cells.

DISCUSSION
Resting virgin lymphocytes initiate extensive clonal proliferation in response to specific antigens, and a portion of the progenies undergoes terminal differentiation with limited lifespan while the rest return to the quiescent state, which can reinitiate even grater proliferation upon re-exposure to the antigens (32). In an attempt to identify the molecules involved in such a characteristic growth regulation of lymphocytes, we have previously isolated a new gene, Spa-1, transcriptionally induced in the mitogen-stimulated mouse lymphocytes (30). In the present study, we have cloned a human counterpart of Spa-1 gene. Human SPA-1 cDNA contained a single ORF encoding 1,042 residues, which was 90% identical to the revised mouse Spa-1 cDNA. Both human SPA-1 and revised mouse Spa-1 cDNAs produced 130-kDa proteins in vitro, and the 130-kDa endogenous proteins could be detected in lymphohematopoietic cells of both species by specific antibody. A Nterminal region (residues 198 to 539, GAP-related domain, GRD) was highly homologous to human rap1GAP, which covered entire catalytic region of rap1GAP (33), and a portion of it (residues 295 to 480) was also homologous to human TSC-2 gene product, tuberin (34). 2 We have previously reported that a GST fusion protein of the ⌬N GRD of mouse Spa-1 (residues 346 to 539 of the revised sequence) exhibited weak but significant GAP activity for Rap1 and Ran (30). By using the baculovirally expressed protein, we now examined the GAP activity of SPA-1 gene product in its complete form (p130 SPA-1 ) for a series of small GTPases. The results clearly indicated that the p130 SPA-1 indeed exhibited GAP activity for Rap1A at the comparable specific activity to the previously reported human p85/95 rap1GAP (28), which was by far greater than that of a GST-⌬N GRD fusion protein. Both proteins also stimulated Rap2A GTPase conforming to the report that Rap2 shared the GAP with Rap1 (35). p130 SPA-1 showed no effect at all on Ras, Rho, Cdc42, and Rac. In contrast to the GST-⌬N GRD, a complete form of SPA-1 gene product failed to show any significant GAP activity for Ran either. We thus conclude that the complete form of SPA-1 gene product is a specific GAP for Rap1 and Rap2 but not for Ran. Most recently, another GRD-sharing p180 TSC-2 (tuberin), a GST-GRD fusion protein of which was also shown to have Rap1 GAP activity (36), was indicated to show GAP activity for Rab5 in its complete form (37). It thus still remains to be an open possibility that p130 SPA-1 can function as a GAP for some additional small GTPases.  Immunostaining study indicated that p130 SPA-1 was localized mostly at the perinuclear region in a speckled pattern similar to Rap1 and Rap2. Subcellular fractionation analysis also indicated that the majority of endogenous p130 SPA-1 in the cells was detected in the insoluble membrane fraction (P100) with a minor portion in the soluble fraction (S100). Both Rap1 and Rap2 were exclusively associated with the former as reported (38). It may be thus suggested that p130 SPA-1 is rather loosely associated with the intracellular membranes such as endoplasmic reticulum. Alternatively, p130 SPA-1 may shuttle between the cytosol and intracellular membrane compartments depending on the state of cells. Indeed, our preliminary experiments indicate that the ratio of p130 SPA-1 in soluble and membrane fractions in lymphocytes changes significantly following antigen-receptor stimulation, implying that its localization is affected by the external signals. Transfection of a cDNA deleted of N-terminal proline-rich region resulted in the protein expression diffusely in the whole cytosol, 3 suggesting that the membrane localization of p130 SPA-1 depended on its N-terminal region. On the other hand, our unpublished results 4 also indicated that truncation of the C-terminal region including a coiled-coil stretch and leucine zipper resulted in the exclusive nuclear localization even in the presence of N terminus. We have previously reported that a smaller form of the Spa-1 (p68) deleted of both N-and C-terminal regions was detected in the nuclei of activated lymphocytes (30). Although its relationship to the complete form of the Spa-1 gene product remained to be further investigated, it appears that the C-terminal region acts as a cytoplasmic retention signal to prevent nuclear translocation of p130 SPA-1 .
Albeit comparable Rap1 GAP activity in vitro, p130 SPA-1 and p85/95 rap1GAP exhibit several distinct features. Structurally, the regions other than homologous GRD are totally distinct each other. p130 SPA-1 possibly has a SH3-binding proline-rich region at the N terminus and a coiled-coil stretch at the C terminus, while no apparent motif is so far evident in p85/ 95 rap1GAP . It seems to be reminiscence of the Ras-GAP family, in which several molecules sharing the catalytic GRD with distinct regulatory elements are known including p120 Ras-GAP (39), neurofibromatoses-1 (40), GAP1 m (41), and IP4BP (42). In human, the SPA-1 gene locus is mapped at chromosome 11q13 (31) and rap1GAP at 1q. 36 -35 (43), and interestingly in Caenorhabditis elegans two distinct genes, CET27F2 and CELF47F6, highly homologous to SPA-1 and rap1GAP, respectively, are also identified. In terms of the expression patterns in human tissues, SPA-1 transcript was predominantly expressed in the lymphoid tissues including leukemia cell lines while by far less in most of the other tissues and practically undetect-  able in the brain and kidney. In contrast, rap1GAP transcript was significantly detected in many tissues, in particular brain, kidney, and pancreas, whereas lymphoid tissues hardly expressed it. Another GRD-sharing gene, TSC-2, was shown to be ubiquitously expressed in human tissues (34), 4 but it was recently reported that p180 TSC-2 (tuberin) was dominantly expressed in the particular cell types such as small blood vessels (44), partly explaining its ubiquitous tissue expression pattern. Thus there seems to be segregation in the expression of these genes among human tissues or cell types. And finally, it was rather surprisingly indicated that p130 SPA-1 and p85/95 rap1GAP exhibited quite distinct, actually opposite, behavior in promyelocytic HL60 cells as they are induced to differentiate.
Uninduced HL60 cells expressed rather abundant p130 SPA-1 while little p85/95 rap1GAP . Upon induction with TPA, the expression of p130 SPA-1 was progressively reduced as the cell cycle progression was arrested followed by the differentiation into typical macrophages. The down-regulation of p130 SPA-1 could be ascribed to the transcriptional repression, which seemed to be compatible with our previous finding that Spa-1 transcription was activated as resting lymphocytes were induced to proliferate by mitogenic stimulation (30). It, however, remained to be seen whether the reduction of p130 SPA-1 was due solely to the transcriptional repression, since the transcript could be still significantly detected on day 3 and later when little protein was detected by the anti-SpaN antibody. Our preliminary results implied the additional effect of modification of the p130 SPA-1 N-terminal region affecting the immunoreactivity of the protein to the antibody. 5 In contrast, the expression of p85/95 rap1GAP was greatly induced during the differentiation process, which was clearly ascribed to the transcriptional activation. Thus on day 4 when essentially all cells fully differentiated into typical macrophages, they now expressed little p130 SPA-1 with quite abundant p85/95 rap1GAP as opposed to the uninduced cells. It was previously reported that the p85/95 rap1GAP level was rather reduced as HL60 cells were induced to differentiate into granulocytes (29). Reasons for the discrepancy are unknown at present, except that HL60 cells in our hand were fully induced to differentiate into macrophages by TPA but unable to differentiate into granulocytes by dimethyl sulfoxide and dexamethasone at all. 5 The transcriptional repression of SPA-1 took place much earlier than the activation of rap1GAP after TPA stimulation. Since the growth arrest preceded the appearance of differentiated phenotypes, it may be implied that p130 SPA-1 was related to the proliferative state while p85/95 rap1GAP to the state of differentiation. In this aspect, we have reported that overexpression of mouse Spa-1 cDNA deleted of both N-and C-terminal region hampered the cell cycle progression (30), which might have reflected the dominant negative effect. Studies to clarify this point is underway in the HL60 system. The implication may be also compatible with the finding that the rap1GAP gene was dominantly expressed in the non-regenerative tissues such as brain and kidney.
Rap1 has been originally shown to induce reversion of the transformed phenotype in K i -ras-transformed NIH3T3 cells (18). Using Rat-1 cells, it was also shown that GTP-bound Rap1 could antagonize the Ras-mediated signaling for the proliferation (19). This Ras-antagonizing effect of Rap1 has been suggested to be due to the competitive inhibition for the Ras effectors (45). On the other hand, Rap1 was shown to directly induce the initiation of DNA synthesis in Swiss 3T3 cells (46), and to activate B-Raf (47). One possible explanation for such seemingly contradictory effects of Rap1 would be that it is involved as a molecular switch in the distinct contexts of the signaling pathways depending on the cell types or states. Taken together, it is suggested that p130 SPA-1 , p85/95 rap1GAP , and possibly p180 TSC-2 (tuberin) sharing Rap1 GAP activity might function in distinct manners and aspects via Rap1/Rap2 in the different types as well as states of cells.