SET-related Cell Division Autoantigen-1 (CDA1) Arrests Cell Growth*

We used an autoimmune serum from a patient with discoid lupus erythematosus to clone a cDNA of 2808 base pairs. Its open reading frame of 2079 base pairs encodes a predicted polypeptide of 693 amino acids named CDA1 ( c ell d ivision a utoantigen-1). CDA1 has a predicted molecular mass of 79,430 Daltons and a pI of 4.26. The size of the cDNA is consistent with its esti-mated mRNA size. CDA1 comprises an N-terminal pro-line-rich domain, a central basic domain, and a C-ter-minal bipartite acidic domain. It has four putative nuclear localization signals and potential sites for phosphorylation by cAMP and cGMP-dependent kinases, protein kinase C, thymidine kinase, casein kinase II, and cyclin-dependent kinases (CDKs). CDA1 is phosphorylated in HeLa cells and by cyclin D1/CDK4, cyclin A/CDK2, and cyclin B/CDK1 in vitro . Its basic and acidic domains contain regions homologous to almost the entire human leukemia-associated SET protein. The same basic region is also homologous to nucleosome assembly proteins, testis TSPY protein, and an uncharacterized brain protein. CDA1 is present in the nuclear fraction of HeLa cells and localizes to the nucleus lines, 40 (cid:3) 20:1, of incorporation rates (data not shown for

Circulating autoantibodies to antigens in the nucleus are associated with systemic autoimmune diseases and are useful diagnostic markers for these diseases (reviewed in Ref. 1). These autoantibodies typically react with highly conserved epitopes of biologically important molecules and have proven highly useful as probes for the molecular cloning and functional characterization of their cognate autoantigens. For instance, human autoantibodies played a pivotal role in the discovery of molecules implicated in the splicing of pre-mRNA and in the identification of proliferating cell nuclear antigen as an auxiliary protein of DNA polymerase ␦ (1). We have ourselves successfully used this approach to identify novel proteins implicated in ion and protein transport (2)(3)(4)(5)(6)(7).
Discoid lupus erythematosus is primarily a cutaneous subset of systemic lupus erythematosus associated with low titer autoantibodies to the nucleus in up to 40% of patients (8). We have shown previously that serum from a patient with discoid lupus erythematosus contains autoantibodies to the glycolytic protein enolase (9) and mitotic chromosomal antigens (10). Here, using serum from the same patient with discoid lupus erythematosus for immunoscreening a human cDNA library, we have isolated a novel nuclear antigen related to SET, a leukemia-associated protein. The transgenic expression of the novel nuclear antigen arrests cell growth. Mutation of its two consensus CDK 1 phosphorylation sites abolishes its ability to arrest cell growth. CDA1 is a phosphoprotein that can be phosphorylated by CDKs in vitro. We named the molecule CDA1 (for cell division autoantigen-1).

EXPERIMENTAL PROCEDURES
Autoimmune Serum-The serum was from a patient with discoid lupus erythematosus identified at the Immunology Laboratories of Gribble's Pathology (Melbourne, Australia).
Molecular Cloning and Sequence Analysis-cDNA expression libraries of human testis (CLONTECH, catalog number HL 101b) and HeLa cells (Stratagene, catalog number 937216) were used. The human testis library was screened with the autoimmune serum as described (4). Positive plaques were detected with horseradish peroxidase-labeled rabbit anti-human immunoglobulin (DAKO A/S, Denmark) and enhanced chemiluminescence (DuPont). 5Ј RACE products were generated on HeLa mRNA using 5Ј RACE (Life Technologies, Inc., catalog number 18374-058) and primers based on the sequence of clone hT6 or hTsl-9 (Fig. 1A). Sublibraries for PCR screening were prepared by plating 50,000 plaque-forming units/plate of the cDNA library and harvesting the phages in 10 ml of SM buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 8 mM MgSO 4 , and 0.01% (w/v) gelatin). Each sublibrary was tested for clones containing the 5Ј end of the target cDNA by PCR using primers located at the very 5Ј end of known cDNA sequences. PCR-positive sublibraries were replated and screened using 32 P-labeled DNA probes. DNA sequencing was carried out as described (4). The predicted amino acid sequence was analyzed for hydrophilicity according to Kyte and Doolittle (11) and for secondary structure by the methods of Chou and Fasman (12)(13)(14)(15) and Robson and co-workers (16,17).
Affinity-purified Rabbit Antibodies to CDA1-Escherichia coli GST fusion proteins were produced in E. coli DH5␣ strain using pGEX vector (20) and purified using a glutathione-Sepharose 4B column (Amersham Pharmacia Biotech). Antibodies were raised in rabbits by immunization with purified GST-hT4 fusion protein (aa 552-693) as described (4) and affinity-purified by sequential passage through GST and GST-hT4 fusion protein columns. Bound antibodies were eluted from a GST-hT4 column in HCl-glycine, pH 2.5, supplemented with 1 mg/ml bovine serum albumin, neutralized to pH 7.0 with 2 M Tris and dialyzed against PBS.
Serum Starvation and Stimulation-HeLa cells were serum-starved for 48 h or serum-starved for 48 h followed by stimulation with 10% fetal calf serum for 4 h and treated with thymidine or colcemid for 24 h to arrest cells at G 1 /S and M phases, respectively. M phase cells were removed by shake-off, and adherent cells were removed by scraping. CDA1 was detected by immunoblotting with affinity-purified rabbit anti-CDA1 antibodies. Blots were reprobed with an antibody to ␣-tubulin to assess protein loading (30 g/lane). CDA1 levels were quantified using the image analysis software Optimas.
Transient and Stable HeLa Cell Lines Transfected with CDA1 Constructs-pCDNA3 was modified by inserting a pair of complementary oligonucleotides containing a SacII restriction site, a Kozak consensus sequence for translational initiation (5ЈCCGCGGCCATGGAA3Ј) (21,22), and two consecutive 10-residue Myc epitopes (EQKLISEEDL), generating pCDNA3-2M. CDA1 (nucleotides 132-2238) or its N terminus (nucleotides 132-1487) was amplified from clone hTsl-9 using Pfu DNA polymerase (Stratagene) and primers designed to introduce flanking EcoRI sites. PCR products were digested with EcoRI and cloned into pCDNA3-2M. pCDNA3-2M containing Myc-tagged CDA1 or its N terminus was digested with SacII and XbaI and cloned into pTRE-tet (CLONTECH). The C terminus of CDA1 was amplified from hTsl-9, cloned into the pTRE-tet, and Myc-tagged at its N terminus. Purified pCDNA3-2M or pTRE-tet constructs were electroporated into HeLa or tTa-HeLa cells. To obtain stable cell lines, 40 g of DNA of pTRE-tet constructs were co-transfected with pTK-Hyg plasmid in a ratio of 20:1, and cells were selected with hygromycin (200 g/ml) in the presence of doxycycline (5 ng/ml). Hygromycin-resistant colonies were recovered individually, cultured in 24-well plates, and screened for CDA1 expression by immunofluorescence staining with an anti-Myc antibody. If necessary, cell lines were cloned further by limiting dilution in 96-well plates with untransfected HeLa cells as feeders that were then killed by culture in hygromycin. Clonal cell lines were used directly or frozen in liquid nitrogen.
Indirect Immunofluorescence-Untransfected HeLa cells or HeLa cells transfected with Myc-tagged CDA1 were reacted with anti-Myc antibody as described (4). Cells were also stained with Hoechst DNA dye 33342 and examined by phase-contrast and immunofluorescence microscopy.
Immunoblotting-Untransfected or transfected HeLa cells, HeLa subcellular fractions, and bacterial fusion proteins were reacted appropriately with anti-Myc antibody or affinity-purified antibody to CDA1 as described (4). Blots were reprobed with anti-␣-tubulin antibody to assess loading. To produce the ␤-galactosidase fusion protein from gt11, isopropyl-1-thio-␤-D-galactopyranoside-induced phage plaques were harvested and subjected to freeze/thawing, and the supernatants were collected.
Immunoprecipitation-Stable tTa-HeLa cell lines containing transgenes of CDA1 or its N terminus were cultured in the absence of doxycycline to induce transgene expression. Cells were metabolically labeled with either 35 S using EXPRE 35 S 35 S protein labeling mix (ϳ200 Ci, PerkinElmer Life Sciences, catalog number NEG072, 7 mCi) or inorganic 32 P (ϳ200 Ci, PerkinElmer Life Sciences, catalog number NEX053, 5 mCi) as described (23). The labeled cells were lysed and immunoprecipitated with either IgG1 anti-Myc antibody or IgG1 isotype control antibody (2B6) to the ␤-subunit of gastric H ϩ /K ϩ ATPase. Immunoprecipitates were subjected to SDS-PAGE followed by autoradiography.
Cell Growth, Cell Density, and Colony Outgrowth Assays-For cell growth assays, stably transfected HeLa cell lines (4 ϫ 10 4 or 5 ϫ 10 5 ) were plated in 6-well plates or 8-cm dishes and cultured for 4 -5 days with or without 5 ng/ml doxycycline. The cells were either counted or stained with 0.1% crystal violet, and the optical density (595 nm) of cell-associated dye extracted with 10% acetic acid was determined, because this reflects cell numbers (24). For cell density assays, the cells were examined by phase-contrast microscopy. For colony outgrowth assays, the 5 ϫ 10 3 HeLa cells were plated on each 10-cm dish, mixed with 5 ϫ 10 5 feeder HeLa cells, and cultured for 24 -48 h. The cells were then selected with hygromycin (200 g/ml) for 2 weeks to kill off feeder cells. Outgrowing colonies were fixed, stained with crystal violet, and photographed. Cell viability was assessed by Trypan blue dye exclusion.
Mutagenesis in Vitro-Mutagenesis was carried out to generate S20A and T340A mutants. Complementary mutant primers were designed to replace codons for serine at position 20 and threonine at position 340 with codons for alanine (S20A and S340A) and to create new NarI restriction sites. The forward primer for S20A was 5Ј-GCAGCTCCGAGgCgCCACAGCGCG-3Ј and for T340A 5Ј-GTGTCT-CACTCggCgCCAATCCGCTG-3Ј (mutant nucleotides are represented by lowercase and NarI sites are underlined). Mutant DNA strands were synthesized using Pfu DNA polymerase (Stratagene), two complementary mutant primers, and circular plasmid DNA of pTRE-2Myc-CDA1 as template. Parental wild-type DNA, produced and methylated in E. coli, was digested with DpnI, and newly synthesized mutant DNA was transformed into the DH5␣ strain of E. coli. Minipreps were digested with NarI to screen for mutants. Mutation sites were confirmed by DNA sequencing. The expression of mutant constructs was confirmed by transient transfection followed by immunoblotting with an anti-Myc monoclonal antibody. A single ClaI site is present between the two mutation sites and an XbaI site downstream of the multi-cloning sites of pTRE vector. A double mutant was constructed by digestion of both constructs with ClaI and XbaI followed by insertion of the T340A fragment into the S20A construct to replace its wild-type fragment. The presence of double mutations was confirmed by NarI digestion.
Phosphorylation of CDA1 by Cyclin/CDKs in Vitro-CDA1 or the CDK phosphorylation site double mutant of CDA1 (CDA1-DM) was immunoprecipitated from lysates generated from transfected HeLa cells. Immunoprecipitated CDA1 or CDA1-DM was incubated in a final volume of 30 l consisting of 20 mM Hepes, pH 7.0, 1.0 mM dithiothreitol, 10 mM MgCl 2 , 100 M ATP, 10 Ci [␥-32 P]ATP, 1.0 mM vanadate, 10 mM NaF, and 10 mM ␤-glycerophosphate for 60 min at 37°C in the presence or absence of 70 units of cyclin/CDK (one unit of cyclin/CDK activity is defined as the amount of kinase activity capable of transferring 1 pmol phosphate/g GST-pRb 773-928 /min at 37°C). Purified cyclin/ CDK complexes were prepared from the baculovirus expression system as described previously (26). Reactions were terminated by the addition of EDTA to a final concentration of 16 mM. Samples were washed five times with RIPA buffer (20 mM Tris, pH 7.6, 300 mM NaCl, 2 mM EDTA, 1% (w/v) Triton X-100, 1% (w/v) sodium-deoxycholate, and 0.1% (w/v) SDS) supplemented with 1 mM dithiothreitol and protease inhibitors (10 g/ml leupeptin, 10 g/ml aprotonin, and 300 M phenylmethylsulfonyl fluoride). After washing, the samples were resuspended in SDS buffer (60 mM Tris-HCl, pH 6.8, 10% (w/v) glycerol, 2% SDS, and 5% ␤-mercaptoethanol), heated at 100°C for 2 min, centrifuged, and electrophoresed on 10% SDS polyacrylamide gel. After electrophoresis, proteins were stained with 0.5% Coomassie Brilliant Blue R, dried under vacuum, and exposed to Kodak Biomax MR x-ray film for autoradiography. Phospho-amino acid analysis and tryptic phosphopeptide mapping of phosphorylated CDA1 or CDA1-DM was performed as described (27).

Molecular
Cloning-Clones hT4 and hT6 were isolated from a human testis cDNA library using the autoimmune serum (Fig. 1A). ␤-galactosidase fusion proteins produced by the clones hT4 (Fig. 1B) and hT6 (data not shown) were specifically immunoreactive with the autoimmune serum, confirming that they encoded the serum-reactive antigen. 5Ј RACE1 and 5Ј RACE2 products were obtained using HeLa mRNA. The clones HL59 and hTsl-9 were isolated by the PCR screening of sublibraries. Full-length cDNA comprising 2808 bp is consistent with the 2.8-kilobase size of HeLa and human testis mRNA (Fig. 1C). Expressed sequence tags corresponding to various regions of CDA1 were identified in the data base (data not shown).
Immunochemical Characterization-Two bacterial GST fusion proteins encoded by cDNA clones hT4 and hT6 both reacted with autoimmune serum (data not shown). Rabbit antibodies to GST-hT4 fusion protein was affinity purified by sequential passage through GST and GST-hT4 fusion protein columns (Fig. 4A). Affinity-purified antibodies to CDA1 reacted with human and monkey cells but not with dog, hamster, mouse, or chicken cells (Fig. 4B). The migration of CDA1 in SDS-PAGE (120 kDa) is slower than that predicted by its molecular mass (79.43 kDa). The apparent molecular mass of 120 kDa was also observed in HeLa cells transfected with Myc-tagged CDA1 (Figs. 6, 7, and 9). The migration of GST-hT4 and GST-hT6 fusion proteins in SDS-PAGE (52 kDa and 84 kDa) was also slower than that predicted by the molecular mass of the fusion proteins (44 kDa and 56 kDa, data not shown). Aberrant migration of proteins containing acidic amino acids has been reported previously (35).
CDA1 is present in the nuclear but not cytosol and membrane fractions of untransfected HeLa cells (Fig. 4C). Myctagged CDA1 or its N terminus containing all four putative NLSs localizes by immunofluorescence to the nucleus and nucleolus of HeLa cells transfected with these constructs (Fig.  4D). The same localization pattern was also observed in HeLa cells transfected with pEGFP-CDA1 (data not shown). In con-

FIG. 3. Homology of CDA1 to human SET and related proteins.
A, a diagram of CDA1 structure and its relationship to SET. Hom. 1 and Hom. 2 are homologous regions of CDA1 and SET, respectively (40% identity, 68% similarity). B, profiles of acidic and basic residues of CDA1 compared with SET. C, amino acid sequences of the homologous region of CDA1 (aa 216 -395) aligned for maximum homology by introducing gaps (dashed lines) with human SET, human TSPY, KIAA, an uncharacterized human brain protein, and human nucleosome assembly proteins NAP1L, NAP1L3, and NAP2. Residues identical to CDA1 are shadowed and conservatively replaced amino acids are boxed. Inserted NAP residues are represented by "xxxxxx." trast, the Myc-tagged C terminus of CDA1 predominantly localizes to the cytoplasm.
Expression of CDA1 after Serum Starvation or Stimulation-To determine whether CDA1 levels vary with cell growth, HeLa cells were serum-starved for 48 h or starved for 48 h and then stimulated with 10% fetal calf serum for 4 h followed by treatment for 24 h with either 2 mM thymidine to block cells at G 1 /S and S phase or with 0.15 g/ml colcemid to block cells at M phase. CDA1 was just detectable in serumstarved cells, increasing dramatically in stimulated cells peaking in S phase (Fig. 5).
Transgenic Expression of CDA1 Arrests Cell Growth-To further investigate the role of CDA1 on cell growth, we established stable HeLa cell lines transfected with CDA1 under the control of a "tetracyclin-off" responsive promoter. The withdrawal of doxycycline from the culture medium of stable cell lines transfected with CDA1 resulted in maximal expression of the CDA1 transgene after 3-4 days (Fig. 6A). Transgenic expression of CDA1 correlated with the complete arrest of cell growth over the entire 4 days of culture.
To determine whether the total arrest of cell growth observed is a consequence of expression of any transgene, we compared the growth of cells transfected with CDA1 to that of stable cell lines transfected with its N terminus (Fig. 6B). We confirmed that expression of the CDA1 transgene for 5 days completely arrested cell growth over this entire period. This was collaborated by morphological assessment of cell density. DNA synthesis in S phase assessed by the BrdUrd incorporation rate showed a gradual decline over the first 3 days with complete arrest by days 4 and 5. However, the DNA content histograms showed normal cell cycle profiles. In contrast, expression of the N-terminal transgene did not arrest cell growth (Fig. 6B) nor did it affect cell density or BrdUrd uptake (data not shown). Over 90% of growth-arrested cells remained viable over this time. The expression of CDA1 and its N-terminal transgenes was confirmed by immunoblotting (data not shown).
To address whether the ability of CDA1 to arrest cell growth depends on expression levels of the CDA1 transgene, stable cell lines transfected with CDA1 were cultured for 2 weeks for colony outgrowth assay or for 5 days for cell growth assay with doxycycline in concentrations ranging from 0 to 5 ng/ml (Fig. 7). The incremental increase in CDA1 expression induced by decreasing doxycycline concentrations was accompanied by a corresponding incremental decrease in colony numbers, cell growth, cell density, and BrdUrd incorporation rate but with retention of normal cell cycle profiles. Colony outgrowth, cell growth, cell density, and BrdUrd incorporation rates were dramatically suppressed with maximal expression of the CDA1 transgene. The inhibition of cell growth was not accompanied by any change in cell viability. In contrast, expression of the Nterminal transgene did not inhibit colony numbers, cell growth, cell density, or BrdUrd incorporation rates (data not shown for colony growth, cell density, and BrdUrd incorporation).

Transgenic Expression of CDA1 Containing Mutant CDK Phosphorylation Sites (CDA1-DM) Fails to Arrest Cell
Growth-To determine the role of the two CDK consensus phosphorylation sites for CDA1 activity, we generated a Myctagged CDA1 construct in which both consensus phosphorylation sites were mutated. The serine at position 20 and threonine at position 340 of wild-type CDA1 was replaced with alanine to generate the double mutant (CDA1-DM). A HeLa cell line, 2C6D3 stably expressing Myc-tagged CDA1-DM, was established using the same Tet-off system. In contrast to wildtype CDA1, overexpression of the double mutant failed to arrest HeLa cell growth determined by BrdUrd incorporation and DNA content (Fig. 8A). On days 4 and 5, after doxycycline was withdrawn from the medium to turn the transgene on, the cells were still actively up-taking BrdUrd and showed similar cell cycle profiles as those with the transgene turned off. Immunoblotting of cells on the same days showed overexpression of Myc-tagged CDA1-DM, whereas the cells with the transgene turned off had undetectable levels of transgene product (Fig.  8B). Immunofluorescence staining of this cell line using anti-Myc antibody showed identical staining patterns and intensities as a wild-type cell line (data not shown). NarI digestion of RT-PCR products confirmed the presence of the double mutation in transgene transcripts regulated by doxycyclin (Fig. 8C).
The S20A and T340A mutations generated additional NarI sites not present in wild-type CDA1. The S20A RT-PCR product (381 bp) was amplified from transcripts of cells stably transfected with mutant CDA1 using a vector primer and a CDA1-specific primer that detected transgene transcripts only. The T340A RT-PCR product (667 bp) was similarly amplified using instead a pair of CDA1-specific primers that detected endogenous as well as transgene transcripts. RT-PCR products derived from cells with the transgene turned on showed an identical-sized band as PCR products derived from wild-type and mutant CDA1 plasmids using the same primers. NarI digestion of RT-PCR products derived from mutant CDA1 transcripts gave the same digestion patterns as PCR products generated from mutant CDA1 plasmids with expected sizes (213 and 169 bp for S20A; 512 and 155 bp for T340A). The PCR products generated from wild-type CDA1 were not digested. With the S20A RT-PCR reaction, no product was seen using transcripts from cells with the transgene turned off, indicating an absence of detectable transgene transcripts. With the T340A RT-PCR reaction, the product from transcripts derived from cells with the transgene off was of a lower yield and was not digested by NarI, suggesting that it was the RT-PCR product of endogenous wild-type CDA1.
CDA1 Is Phosphorylated in Vivo-To determine whether CDA1 is phosphorylated in vivo, HeLa cells transfected with Myc-tagged CDA1 were metabolically labeled with either [ 35 S]methionine or 32 P. A 35 S-or 32 P-labeled 120-kDa protein was specifically immunoprecipitated by an anti-Myc antibody (Fig. 9A). The 32 P-labeled Myc-tagged N terminus of CDA1 containing putative phosphorylation sites was also specifically immunoprecipitated (data not shown). A, cell colony outgrowth assay. HeLa cells (5 ϫ 10 3 ) stably transfected with CDA1 were cultured for 14 days with doxycycline at the concentrations (ng/ml) indicated. The colonies were fixed and stained with crystal violet. B: a, cell growth assay. HeLa cells (4 ϫ 10 4 ) stably transfected with CDA1 or its N terminus were cultured with doxycycline at the concentrations indicated for 4 days and stained with crystal violet, and the optical density of cell-associated dye was determined. b, immunoblotting. Stably transfected HeLa cells (2 ϫ 10 5 cells) were cultured under the same conditions as described above, and CDA1 (arrow) was detected with anti-Myc antibody (50 g protein/sample). C, cell density and DNA synthesis. Stably transfected HeLa cells were cultured for 4 days with doxycycline concentrations as indicated. The cell density, BrdUrd uptake, cell cycle profiles, and cell viability are as described in the Fig. 7B legend.
CDA1 Is Phosphorylated in Vitro by Cyclin/CDKs-Because mutation of the two CDK phosphorylation sites of CDA1 abolished its growth-inhibitory activity, we sought to determine whether CDA1 is phosphorylated on these sites by cyclin/CDKs. We incubated CDA1 immunoprecipitated from HeLa cells with purified catalytically active cyclin/CDK complexes in in vitro phosphorylation reactions. These studies revealed that CDA1 was phosphorylated in the absence of added cyclin/CDK, which was probably caused by the activity of a co-immunoprecipitating kinase. The level of CDA1 phosphorylation was increased by cyclin D1/CDK4, cyclin A/CDK2, and cyclin B/CDK1 but not by cyclin E/CDK2 (Fig. 9B). CDA1-DM was also phosphorylated in the absence of added cyclin/CDKs. However, the level of CDA1-DM phosphorylation was not increased by any of the added cyclin/CDKs, indicating that phosphorylation depended on the presence of the CDK phosphorylation site(s). This was confirmed by tryptic phosphopeptide mapping, which demonstrated that CDA1 phosphorylated in the absence of cyclin/CDKs displayed four major phosphopeptides, 1-4 (Fig. 9C). CDA1 phosphorylated with cyclin A/CDK2 significantly increased the phosphorylation of these phosphopeptides and generated three major new phosphopeptides, 5-7, resulting from the phosphorylation of either one or both of the CDK phosphorylation sites. The generation of several tryptic phosphopeptides from the phosphorylation of one or two sites is caused by the generation of partial digests by tryptic cleavage. To define which site(s) on CDA1 was phosphorylated by cyclin A/CDK2, we performed phospho-amino acid analysis of CDA1 phosphorylated either in the absence or presence of cyclin A/CDK2. These studies revealed that CDA1 was phosphorylated basally on threonine and to a greater extent on serine (Fig. 9D). The level of serine and threonine phosphorylation was increased by cyclin A/CDK2, confirming that both serine 20 and threonine 340 were phosphorylated by this kinase in vitro. DISCUSSION We report the molecular characterization of a novel cell division nuclear autoantigen that we have named CDA1. CDA1 localization to the nucleus is supported by the following observations. First, the predicted amino acid sequence of CDA1 contains four putative NLSs. Second, CDA1 localizes to the nuclear fraction of HeLa cells. Third, Myc-tagged or enhanced green fluorescent protein-tagged CDA1 and its Nterminal segment containing all four nuclear localization signals localizes to the nucleus and nucleolus of transfected HeLa cells. The nuclear localization of CDA1 is supported further by observations of Ueki et al. (36), who used a nuclear transportation trap for isolating nuclear proteins. A partial cDNA clone isolated in this way (GenBank TM accession number AB015345) and encoding aa 208 -693 of CDA1 localizes to the nucleus in COS-7 cells. These latter observations indicate that the third and fourth NLSs contained in aa 208 -693 of CDA1 are sufficient to target CDA1 to the nucleus. The role of the first and second NLSs in targeting CDA1 to the nucleus or nucleolus remains unknown.
Endogenous CDA1 levels are low in resting serum-starved HeLa cells and dramatically elevated in serum-stimulated cells. These observations suggest a role for CDA1 in cell growth. The suggestion is supported by our findings in stable HeLa cell lines transfected with Myc-tagged CDA1 in which CDA1 levels can be regulated by a tetracyclin-off promoter. Maximal expression of the CDA1 transgene in cells cultured in the absence of doxycycline was observed after 3-5 days. CDA1 transgene expression was associated with dramatic arrest of cell growth and cell density over the 4 -5 days of culture. DNA synthesis assessed by BrdUrd uptake showed a gradual decline in the first 3 days followed by a virtual complete arrest by days 4 -5. The capacity of CDA1 to arrest cell growth was confirmed further in the stable HeLa cell lines in which we regulated CDA1 expression levels by culture in the presence of different doxycycline concentrations. An incremental decrease of HeLa colony numbers, cell growth, cell density, and BrdUrd uptake paralleled corresponding incremental increases in CDA1 expression, indicating that cell growth and DNA synthesis depends on the level of expression of the CDA1 transgene. Inhibitory effects of CDA1 were not accompanied by a change in cell viability or in cell cycle profiles. The latter results were surprising, because we had expected the cells in which DNA synthesis had been arrested to accumulate at the G 1 /S transition. One possible explanation for these results is that CDA1 may exert inhibitory effects on multiple stages of the cell cycle, acting as a negative regulator of cell cycle progression. The suggestion is consistent with the elevated levels of endogenous CDA1 observed in G 1 , S, and M phases of the cell cycle.
Stable transfectants of HeLa cells harboring the CDA1 Nterminal transgene lacking its acidic C-terminal tail did not arrest cell growth or DNA synthesis. These observations suggest that the inhibition of cell growth and DNA synthesis observed with the CDA1 transgene is not the consequence of expression of just any transgene. The observations also suggest that the inhibitory effects of CDA1 on cell growth and DNA synthesis requires its acidic C-terminal tail. The acidic C-terminal tail of CDA1 has ϳ40% identity and 68% similarity with the acidic C-terminal tail of the leukemia-associated protein SET. The central region of CDA1 shares the same level of identity and similarity with most of the remainder of the SET protein and is also homologous to the other SET-related proteins, namely human nucleosome assembly proteins (NAPs), TSPY testis protein, and an uncharacterized brain protein KIAA0721. NAPs also have single or multiple acidic regions, although they are not located consistently in the C-terminal tail.
Human SET is a 32-kDa ubiquitous nuclear protein first identified fused to CAN in acute undifferentiated leukemia (34,37). The fused set-can gene includes a large part of the acidic tail of SET. CAN has also been found fused with DEK in a subtype of acute myeloid leukemia (38), suggesting that CAN is an oncogene activated by fusion with SET or DEK. The only common motif of SET and DEK is their acidic domain. SET is a potent and specific inhibitor of protein phosphatase 2A (39). The acidic C-terminal tail of SET binds to a region of the HRX protein that is consistently retained in some 25 HRX fusion proteins (40). The reciprocal translocation of the HRX gene, resulting in HRX fusion proteins, is one of the most common chromosomal abnormalities in acute leukemia (41). SET and HRX fusion proteins form a heterocomplex with protein phosphatase 2A, suggesting that leukemic effects of these proteins may be related to their interactions (40). SET has been reported recently to associate directly with the CDK inhibitor p21 Cip1 and to reverse p21-mediated inhibition of cyclinE/ CDK2 activity. Based on these observations, it has been suggested that SET modulates p21 Cip1 -inhibitory activity and regulates G 1 /S transition by modulating cyclin E/CDK2 activity (42). SET has also been identified as the template-activating factor 1 required for adenovirus genome replication. The acidic tail of SET is required for this activity (43).
The predicted amino acid sequence of CDA1 contains putative phosphorylation sites for a variety of protein kinases including CDK1 and CDK2. Our demonstration that CDA1 is phosphorylated in HeLa cells suggests that one or more of these sites are phosphorylated, raising the possibility that CDA1 activity is regulated not only by expression level but also by phosphorylation. In this context, Xenopus SET has been shown to interact with B-type cyclins (44). Saccharomyces cerevisiae and Xenopus NAP1, a 60-kDa protein homologous to SET (45), also binds to B-type cyclins and is phosphorylated by the cyclin B/CDK1 complex. We observed that a CDA1 mutant with the two consensus CDK phosphorylation sites abolished (S20A and T340A) disabled its capacity to inhibit cell growth, indicating that these sites are important for the function of this protein. Furthermore, we showed that these sites are phosphorylated by cyclin/CDKs in vitro, suggesting that these kinases may regulate CDA1 function in vivo. These observations suggest that in addition to the expression levels of CDA1, phosphorylation of these two sites by cyclin/CDKs may play important roles in regulating the function of this protein in cellular proliferation. Interestingly, cyclin D1/ CDK4, cyclin A/CDK2, and cyclin B/CDK1 phosphorylated CDA1, whereas cyclin E/CDK2 did not. Because cyclin E/CDK2 is active during the G 1 to S phase transition, these studies suggest that CDA1 may be differentially phosphorylated throughout the cell cycle on serine 20 and threonine 340 to regulate its function and in turn cellular proliferation.
In the absence of NAP1 in S. cerevisiae, Clb2, a B-type cyclin, is unable to induce mitotic events or switch from polar to isotropic bud growth (45). Yeast cells lacking NAP1 also undergo a long delay at the short spindle stage with normal levels of Clb2/CDK1 kinase activity, suggesting that NAP1 is required for the regulation of microtubule dynamics and for Clb2/CDK1 kinase to amplify its own product. S. cerevisiae NAP1 binds tightly to Gin4 kinase and is required for kinase activation by phosphorylation by Clb2 as cells enter mitosis. In turn, Gin4 kinase is required for NAP1 and Clb2 to promote progression through mitosis and for switching from polar to isotropic bud growth (46). NAP1 and NAP2 are phosphorylated in vitro by casein kinase 2. NAP2 is phosphorylated in vivo at the G o /G 1 boundary but not in S phase (47).
Because CDA1 overexpression showed an activity opposite to that of SET and NAP, it is possible that phosphorylated CDA1 may inhibit activity of one or more CDKs that regulate cell cycle progression.
CDA1 contains an N-terminal Pr domain that is not present in SET or other related proteins. Pr domains are found in ligand domains of a number of regulatory proteins. These Pr domains bind to Src Homology 3 domains contained in a variety of intracellular and membrane-associated proteins and kinases (48 -53) involved in signal transduction. The common amino acid sequence motif of Pr domains for Src Homology 3 domain binding is PXXP with neighboring residues forming patterns specific for individual Src Homology 3 domain of different proteins (54 -57). CDA1 has three regions containing PXXP motifs, although no specific patterns are found for binding to known Src Homology 3 domains. The region also contains stretches of nine and five P residues. The observations suggest that CDA1 may interact, through its Pr domain, with unique cognate protein(s), the identification of which may provide more information on its function.