The Mechanism of Growth-inhibitory Effect of DOC-2/DAB2 in Prostate Cancer

DOC-2/DAB2 is a member of the disable gene family with tumor-inhibitory activity. Its down-regulation is associated with several neoplasms, and serine phosphorylation of its N terminus modulates DOC-2/DAB2's inhibitory effect on AP-1 transcriptional activity. We describe the cloning ofDIP1/2, a novel gene that interacts with the N-terminal domain of DOC-2/DAB2. DIP1/2 is a novel GTPase-activating protein containing a Ras GTPase-activating protein homology domain (N terminus) and two other unique domains (i.e. 10 proline repeats and leucine zipper). Interaction between DOC-2/DAB2 and DIP1/2 is detected in normal tissues such as the brain and prostate. Altered expression of these two proteins is often detected in prostate cancer cells. Indeed, the presence of DIP1/2 effectively blocks mitogen-induced gene expression and inhibits the growth of prostate cancer. Thus, DOC-2/DAB2 and DIP1/2 appear to represent a unique negative regulatory complex that maintains cell homeostasis.

DOC-2/DAB2 (differentially expressed in ovarian carcinoma 2/disabled-2) is identified in normal human ovarian epithelial cells but absent in ovarian cancer cell lines (1,2). An absence of DOC-2/DAB2 expression is associated with malignant cells including mammary, prostate, and ovary (2)(3)(4). Increased expression of DOC-2/DAB2 inhibits the growth of several cancer cells (4,5), which suggests that it functions as a tumor suppressor. DOC-2/DAB2 also appears to be a phosphoprotein, and its phosphorylation can be regulated by several stimuli (4,6). We recently demonstrated that DOC-2/DAB2 expression is significantly increased in the enriched basal cell population with stem cell potential of the degenerated rat prostate (4), suggesting that DOC-2/DAB2 may play an important role in regulating the homeostasis of epithelial differentiation. Amino acid sequence analysis predicts that DOC-2/DAB2 is a potential sig-naling molecule. Its N terminus shares 54% homology with mouse DAB1 protein that can be phosphorylated by Src, and the disruption of the DAB1 gene can cause developmental defects in central neurons (7,8). In addition, we and others also show that the C terminus of DOC-2/DAB2 containing prolinerich domains can bind to Grb2 proteins (9,10). Thus, it appears that DOC-2/DAB2 is involved in modulating the Ras signaling pathway.
To understand biochemical function of DOC-2/DAB2, we identified a key amino acid residue (Ser 24 ) in its N terminus. The phosphorylation of this residue by protein kinase C (PKC) 1 activator 12-O-tetradecanoylphorbol-13-acetate (TPA) can modulate its inhibitory activity on TPA-induced gene transcription (11). These data indicate that the DOC-2/DAB2 protein, particularly its N terminus, is a potent negative regulator for the PKC-elicited signal pathway. However, little is known about the downstream effector(s) mediated by DOC-2/DAB2.
In this study, we employed a yeast two-hybrid system to identify DIP1/2 as an immediate DOC-2/DAB2-interactive protein. DIP1/2, a novel Ras GTPase-activating protein (GAP), interacts with the N-terminal domain of DOC-2/DAB2. We cloned DIP1/2, characterized it as an immediate downstream effector of DOC-2/DAB2, and delineated its functional role in mitogen-induced gene expression and growth inhibition of prostate cancer.

EXPERIMENTAL PROCEDURES
Cell Cultures-Three human prostate cancer cell lines (TSU-Pr1, LNCaP, and C4-2) and COS cells were maintained in T medium supplemented with 5% fetal bovine serum (4). PrEC, a primary prostatic epithelial cell derived from a 17-year-old juvenile prostate, was maintained in a chemically defined medium purchased from Clonetics. PZ-HPV-7, a cell line derived from the peripheral zone of a normal prostate (12), was maintained in T medium containing 5% fetal bovine serum. MDAPCa 2a and MDAPCa 2b cell lines were derived from patients with bony metastasis (13). Three additional primary prostatic epithelial cells, derived from either cancer lesions (SWPC1, SWPC2) or adjacent normal tissue (SWNPC2), were obtained from patients with prostate cancer who had had radical prostatectomy. All these cells were maintained in the same medium as PrEC. Corresponding antibody staining indicated that all primary cells were cytokeratin-positive and vimentin-negative.
Yeast Two-hybrid Screening-Using primers 5Ј-GAATTCCCCGTC-ATGTCTAACGAA-3Ј and 5Ј-GGATCCTAACTGAGGCTTTGGTCGAG-G-3Ј and using DOC-2/DAB2 cDNA as a template we generated an 823-bp fragment corresponding to the 5Ј-end of the cDNA. The PCRamplified fragment was sequenced before it was subcloned in-frame into pVJL11 vector as a bait construct. Equal amounts of constructed bait vector and pVP16 rat brain cDNA library vector were co-transformed into the yeast strain of L40, and the transformed yeasts were plated on SD-L-T-H (synthetic medium lacking amino acids of leucine, tryptophan, and histidine) plates with 5 mM 3-aminotrizol. Only those colonies that had ␤-galactosidase activities were further analyzed. Plasmids from those positive clones were rescued and transformed into Escherichia coli HB101 strain for further amplification.
Four rounds of phage library screening were performed with a rat brain ZAP phage library (Stratagene) to clone the full-length cDNA of DIP1/2. After DNA sequencing for each positive clone, the full-length cDNA of DIP1/2 was assembled with the appropriate restriction enzyme digestion.
Northern Blot-Total RNA from various organs of the male rat was isolated with RNAzol (TEL-TEST). Twenty micrograms of total RNA/ lane were separated on 1% formaldehyde agarose gel, transformed onto Zeta-Probe membrane (Bio-Rad), and then hybridized with 32 P-labeled DIP1/2 or glyceraldehyde-3-phosphate dehydrogenase cDNA probe.
Generation of the Anti-DIP1/2 Polyclonal Antibody-A peptide sequence (CTNPTKLQITENGEFRNSSNC) corresponding to DIP1/2 amino acid residues 976 -996, with an extra cysteine at the N terminus as a linker, was synthesized and used as the antigen to immunize rabbits for generating polyclonal antibody by Zymed Laboratories Inc. Laboratory. After 7 weeks, rabbits were sacrificed to collect antiserum after the fourth boosting injection of antigen. SulfoLink gel (Pierce) was coupled with the synthetic peptide and then blocked with 50 mM cysteine in 50 mM Tris, 5 mM EDTA, pH 8.5; it was then washed with 1 M NaCl. Antibodies against DIP1/2 were first purified by slowly passing the antiserum to the coupled SulfoLink gel. After washing with five column volumes of 1 M NaCl, they were eluted with elution buffer from the Sulfolink kit (Pierce). Purified antibodies were dialyzed overnight against 4 liters of deionized water at 4°C.
Purification of the GST Fusion Protein and Ras GAP Activity Assays-The minimal GAP domain of DIP1/2 cDNA was amplified by PCR using primers 5Ј-GGGATCCCAGAACGCAACAGC-3Ј and 5Ј-AGAAT-TCTTAGCTTGAGCTGCGGGCAGG-3Ј. The amplified fragment was subcloned in-frame into the pGEX-5X vector and transformed into the E. coli strain of BL21. The bacteria culture was induced with 2.5 mM isopropyl-1-thio-␤-D-galactopyranoside at 37°C for 4 h. The bacterial pellet was washed once with cold PBS. After spin, the pellet was resuspended in PBS and subjected, five times, to 30-s sonication. The GST-GAP fusion protein was purified according to the manufacturer's manual (Roche Molecular Biochemicals), and the purified protein was analyzed using 10% SDS-PAGE analysis and followed by GELCODE Blue staining (Pierce). Assay of Ras GAP activity in vitro was performed according to Kim et al. (14). To prepare GTP-bound Ras protein, 0.25 M human recombinant Ha-Ras protein (Calbiochem) was incubated with 20 nM [␥-32 P]GTP (6000 Ci/mmol; PerkinElmer Life Sciences) in a buffer containing 20 mM HEPES (pH 7.3), 1 mM EDTA, 2 mM dithiothreitol, and 1 mg/ml bovine serum albumin for 5 min at room temperature. Up to 1 g of either GST-GAP or GST protein alone was added into a buffer containing 20 mM HEPES, pH 7.3, 5 mM MgCl 2 , and 1 mM dithiothreitol. The loaded Ras was then incubated with either GST-DIP1/2 or GST for the indicated time, and the reaction was stopped by adding 5 volumes of ice-cold 20 mM HEPES, pH 7.3, and 1 mM MgCl 2 . The reaction mixture was then filtered through 0.45-m HA membrane (Millipore Corp.). Filters were air-dried and then subjected to scintillation counting.
C4-2 cells were transfected with either HA-tagged Ras alone or HA-tagged Ras and DIP1/2. Two days after transfection, cells were treated overnight with T medium without serum, and 100 ng/ml EGF was added for 20 min. Then cells were lysed with lysis buffer (PBS with 5 mM MgCl 2 and 1% Triton X-100). The whole lysate was spun down, and the supernatants were added with 10 l of Raf-conjugated agarose beads (Upstate Biotechnology, Inc., Lake Placid, NY). The mixtures were incubated at 4°C for 30 min. Pellets were spun down and washed four times with the same lysis buffer, and 20 ml of 1ϫ SDS-PAGE sample buffer was added to the pellets and incubated at 100°C for 3 min. Treated samples were loaded on SDS-PAGE gel, and Ras-Raf binding was detected using Western blot.
Generation of pCI-DIP1/2 Mutants and pGEX-5X-DIP1/2 Constructs-To make mutants of DIP1/2 and the GST fusion DIP, the QuikChange TM site-directed mutagenesis kit was employed. Site-directed mutagenesis was performed by PCR according to the manufacturer (Stratagene). Oligonucleotide used for each mutant was 5Ј-GGACAATGAGCAC-CTCATCTTTCTGGAGAACACATTGGCCACCAAGG-3Ј. Briefly, after denaturing the wild type plasmids, the oligonucleotide primer was annealed with template DNA and then extended with Pfu Turbo DNA polymerase. After PCR, the methylated and nonmutated parental DNA template was digested with DpnI. The XL-1 Blue cells were then transformed with DpnItreated DNA for selecting the mutated DNA. Mutants were verified by sequencing.
In Vitro Characterization of the Effect of DIP1/2 on Prostate Cancer-To determine the effect of DIP1/2 on prostate cancer cells, we studied 1) gene transcription using either serum response element (SRE) or TPA response element (TRE) reporter gene assays and 2) cell growth using both crystal violet and colony formation assays.
For reporter gene assay, C4-2 cells were transiently transfected with either SRE or TRE reporter plasmids in the combination of DIP1/2, ⌬B, and ⌬B-S24A expression vectors. Two days after transfection, T medium with 0.5% FBS was changed for another 24 h, and then either 50 ng/ml EGF or 100 ng/ml TPA was added to the cells for an additional 14 h. Luciferase activity assays were performed as described previously (11).
Using LipofectAMINE (Invitrogen), C4-2 cells (2 ϫ 10 5 ) were transfected with 2 g of pCI-DIP1/2 or 2 g of pCI-neo (control). Two days after transfection, cells were selected with G418 (800 g/ml), and an individual colony was cloned by ring isolation (4). The in vitro growth rate of each clone was determined by plating cells in a 24-well plate at a density of 5000 cells/well with T medium containing 2% TCM™ (Celox) and 0.5% FBS. At the indicated days, cell numbers were determined by crystal violet assay (15).
For the colony formation assay, C4-2 cells (3 ϫ 10 4 ) were pleated on a 35-mm dish with T-medium containing 5% FBS and co-transfected with 0.2 g of ␤-galactosidase expression vector with 0.8 g of cDNAs as indicated. Twenty-four hours after transfection, cells were changed to T-medium containing 0.2% FBS. At the indicated time, cells were washed with cold PBS twice and fixed. The number of blue cells was counted by ␤-galactosidase staining according to Yeung et al. (16).

RESULTS
Identifying and Cloning DIP cDNA-DOC-2/DAB2's first 260 amino acids were used as a "bait" sequence in the yeast two-hybrid system to search for protein(s) that interacts with the N-terminal domain of DOC-2/DAB2 (17). Of 10,000 transformants screened, 36 positive clones were selected, and two positive clones (DIP1, DIP2) were further analyzed. These two clones shared overlapping sequence and were identical. However, since neither alone contained a full-length sequence, we designated the full-length sequence "DIP1/2." To obtain the full-length cDNA of DIP1/2, a ZAP cDNA library from a rat brain was screened. Eleven clones spanned about 6.3 kb and represented two different sizes of DIP1/2 mRNA transcripts with different 5Ј upstream sequences (Fig.  1A). The DIP1/2 was predicted to have an open reading frame of 996 amino acids and a calculated molecular mass of 110 kDa.
According to the deduced protein sequence, DIP1/2 appears to be a novel protein with several potential functional domains (Fig. 1B). Its key feature is the Ras GAP homology domain, which spans from residues 177 to 409 and is present in all members of the Ras GAP family (18). Also, DIP1/2 had a stretch of 10 proline repeats (residues 727-736) with the capacity to bind to proteins containing an Src homology 3 domain (19) and a leucine zipper dimerization domain (residues 842-861) for protein dimerization (20). The amino acid sequence alignment of DIP1/2's GAP domain with other Ras GAP pro-teins ( Fig. 1C), including p120 GAP , Homo sapiens neurofibromine (hNF1), Rattus norvegicus Ras GAP (rnGAP), a novel human Ras GAP (nGAP), and synaptic Ras GAP (SynGAP), shows that DIP1/2 contains all of the critical consensus amino acids for Ras GAP activity (21). This suggests that DIP1/2 can function as a Ras GAP.
Characterizing the DIP1/2 Expression Profile-Northern blot analysis indicated that steady-state levels of DIP1/2 mRNA (about 6.9 kb) are detected in brain, lung, thymus, bladder, and skeletal muscle tissue ( Fig. 2A). In both brain and kidney, a different size of RNA transcript with 9.6 kb was found that may represent DIP1/2a with an additional 5Ј-upstream sequence. Nevertheless, steady-state levels of DIP1/2 mRNA were not detected in several urogenital organs including the ventral prostate, dorsal lateral prostate, seminal vesicle, and coagulat-ing gland. However, detectable DIP1/2 mRNA levels ( Fig. 2A) were detected in Noble rat prostate epithelia (NbE) and Sprague-Dawley rat prostate epithelia (VPE) from basal cells of the ventral prostate. Increased levels were not detected in stromal cells (i.e. NbF and VPF) from the same animals. This indicates that DIP1/2 preferentially expresses in the prostatic basal epithelial cells. Further Northern analysis (Fig. 2B) indicated that expression of DIP1/2 mRNA increased in degenerated prostates in a time-dependent manner, and the expression patterns of DIP1/2 and DOC-2/DAB2 mRNA concur. These findings suggest that both genes co-express in the enriched basal cell population during prostate degeneration (Fig. 2B).
A polyclonal antibody was raised against a synthetic peptide derived from the C terminus of DIP1/2. With this antibody, a major band of 110 kDa was detected from the in vitro tran- scription and translation of DIP1/2 cDNA (Fig. 3A). To further test the specificity of this antibody, serum was incubated with increasing concentrations of synthetic peptide of DIP1/2, ranging from 20 to 200 g/ml, prior to probing with the blotted membrane. Results (Fig. 3A) indicated that the synthetic peptide effectively competes with the antibody in ability to bind to the DIP1/2 protein.
To further understand the profile of DIP1/2 and DOC-2/ DAB2 expression in human prostate cancer, we screened a variety of prostate cancer cell lines. In Western blot analysis, only a 110-kDa protein band was detected in cell lysate of human prostate cells, indicating that the antibody also recognizes the homologue of human DIP1/2 (hDIP1/2). The sequence of hDIP1/2 is very similar to that of rDIP1/2. 2 As shown in Fig.  3C, we found that DIP1/2 and DOC-2/DAB2 proteins were present in both normal primary epithelial cells (PZ-HPV-7, PrEC, and SWNPC2) and primary tumors (SWPC1 and SWPC2). However, a significant decreased expression of DIP1/2 was detected in several metastatic cell lines such as TSU-Pr1, LNCaP, C4-2, MDAPCa 2a , and MDAPCa 2b , cancer cell lines. We believe this indicates that DIP1/2 is involved in the progression of prostate cancer.
Specific Interaction between DIP1/2 and DOC-2/DAB2-Because data from the yeast two-hybrid screening indicated that DIP1/2 and DOC-2/DAB2 directly interact, we further examined whether these two native proteins interact with each other using brain and prostate as tissue sources. Using antibodies against either DOC-2/DAB2 (transfection) or DIP1/2 in a coimmunoprecipitation experiment, we demonstrated that endogenous DIP1/2 and DOC-2/DAB2 proteins were present in the same immune complex (Fig. 3D). Noticeably, there were two sizes of DIP1/2 protein in the rat brain (molecular masses of 110 and 135 kDa, respectively) (Fig. 3D), and the predominant protein appears to be 135 kDa, which may correspond to the 9.6 kb of DIP1/2 mRNA detected in brain tissue ( Fig. 2A).
To confirm whether DIP1/2 only interacts with the N-termi- 2 Hong, C., Pong, R-C., Wang, Z., and Hsieh, J. T., Genomics, in press. nal domain of DOC-2/DAB2, we used a series of T7-tagged DOC-2/DAB2 wild type (p82), slicing form (p59), N-terminal deletion mutant (⌬N), and C-terminal deletion mutant (⌬B)) as described previously (11), and an HA-tagged DIP1/2 construct. Cells were co-transfected with both vectors for 24 h, and then cell lysates were immunoprecipitated with T7-antibody (for DOC-2/DAB2) and then probed with HA-antibody (for DIP1/2). The presence of DIP1/2 was only detected in the cells co-transfected with DIP1/2 and DOC-2/DAB2 containing the N terminus, such as p59, p82, and ⌬B, but not in the cells co-transfected with DIP1/2 and the C terminus of DOC-2/DAB2 (⌬N) (Fig. 4A). Conversely, using HA-antibody for immunoprecipitation and then probing with T7-antibody, we demonstrated that DIP1/2 protein can be co-precipitated with p59, p82, and ⌬B protein but not with ⌬N protein (Fig. 4B). Furthermore, to rule out the possible experimental artifact that is due to differential protein expression in each transfection, levels of each recombinant protein were determined by Western analysis and appeared to be identical (Fig. 4, C and D). Taken together, these data indicate that DIP1/2 protein only interacts with the Nterminal, not the C-terminal, domain of DOC-2/DAB2 protein.
Function of DIP1/2 in Vitro and in Vivo-Due to the high sequence homology between DIP1/2 and other Ras GAPs, we thought it likely that DIP1/2 facilitates Ras GTPase activity. To test this, we prepared a GST-DIP1/2 fusion protein containing the minimal Ras GAP domain (23), and either this fusion protein or GST protein was incubated with human recombinant [␥-32 P]GTP-bound Ha-Ras protein. The increasing amounts of GST-DIP1/2 (ranging from 0.2 to 1 g) stimulated Ras GTPase activity in a dose-dependent manner (Fig. 5A). Conversely, control GST protein (1 g) had no effect on Ras GTPase activity.
To further compare the Ras GAP activity of DIP1/2 with a known Ras GAP protein (GAP120), we created a DIP1/2 cDNA construct (DIP1/2 m ) as a control with a point mutation (R220L) that may disrupt GAP activity. As shown in Fig. 5B, the overall GAP activity between p120 GAP and DIP1/2 is very similar, and the DIP1/2 m did not have any GAP activity toward Ha-Ras. These data clearly demonstrate that DIP1/2 can stimulate GTPase activity of Ha-Ras in vitro. To examine the specificity of DIP1/2 to other small G-proteins such as K-Ras, R-Ras, TC21, and Rap1A, we found that DIP1/2 has a similar GAP activity as p120 GAP by stimulating GTPase activity of K-Ras, R-Ras, and TC21 but not Rap1A (Fig. 5C). These data indicate that DIP1/2 is a typical Ras GAP.
Early study of Ras signal transduction indicates that Raf is an immediate downstream effector for Ras signaling (24). Because Raf binds tightly to the GTP-bound form of Ras but not to the GDP-bound form, such differential affinity can be used to determine the GTP-bound status of Ras. To analyze the GAP activity of DIP1/2 in vivo, C4-2 cells (25) were transfected with vectors expressing HA-tagged Ras, DIP1/2, or DIP1/2 m . After activating Ras using EGF, the GTP-bound form of Ras was precipitated with GST-RBD (GST-Raf containing Ras binding domain). Precipitated Ras was detected using the HA-antibody. As shown in Fig. 5D, in the presence of EGF, the amount of GTP-bound Ras increased over that of the control. Levels of the GTP-bound Ras significantly decreased in cells expressing DIP1/2; however, DIP1/2 m failed to stimulate Ras GTPase in cells treated with EGF. The whole cell lysates were examined for expression of DIP1/2 and Ras proteins, and results demonstrated that expressions of DIP1/2 and Ras were identical between each transfection. These results indicate that DIP1/2 can function as a Ras GTPase-activating protein in vivo. Therefore, we conclude that DIP1/2 functions as a Ras GAP in vivo and in vitro.
Regulation of the Ras-Raf Signaling Pathway by DIP1/2-The Ras protein functions as an essential component in many intracellular signaling pathways responsible for differentiation, proliferation, and apoptosis (26). The Raf-MEK-ERK pathway is a key signal transduction pathway modulated by Ras protein (27). The downstream components of this pathway, including AP-1, which binds to TRE, and EIK-1, which binds to SRE, subsequently activate gene expression (28 -30). Since PKC is able to activate the Raf/MEK/ERK axis (31,32), we investigated the impact of DIP1/2 on this cascade. As shown in Fig. 6A, in the absence of EGF, increased expression of DIP1/2 could inhibit the basal levels of SRE reporter gene activity in prostate cancer cells.
The presence of EGF increased the reporter gene activity at least 5-fold. However, DIP1/2 could inhibit the reporter gene activity in a dose-dependent manner. Using the same reporter gene assay, we found that DIP1/2 or ⌬B, a DOC-2/DAB2 cDNA FIG. 4. Direct interaction between DIP1/2 and DOC-2/DAB2 or DAB1. COS cells were cotransfected with different T7-tagged DOC-2/ DAB2 constructs and HA-tagged DIP1/2 constructs for 48 h. The supernatants were immunoprecipitated with either T7-antibody-conjugated agarose beads (A) or HA-antibody plus protein A/G-agarose beads (B). After centrifugation, pellets were subjected to immunoblotting analysis. The levels of protein expression from each transfection were determined by Western analysis (C and D). Cell lysate was prepared from COS cells transfected with each expression vector and subjected to pull-down by glutathione beads and then probed with DIP1/2 antibody (E). IP, immunoprecipitation; IB, immunoblotting.
containing the N-terminal domain (11), alone can suppress SRE activity (Fig. 6B), but co-expression of ⌬B and DIP1/2 had an additive effect on the inhibition of SRE activity in the presence of TPA. These data indicate that physical interaction between DIP1/2 and DOC-2/DAB2 has functional impact on Ras-mediated signal transduction.
Previously, we demonstrated that PKC-elicited DOC-2/ DAB2 phosphorylation can block TPA-induced gene activity (11). Therefore, we investigated whether DIP1/2 is a mediator involved in this action. We employed the C4-2 cell line because both DOC-2/DAB2 and DIP1/2 are not detectable. Either a high concentration of ⌬B or DIP1/2 alone was able to inhibit TPAinduced TRE reporter gene activity (Table I). At this concentration, we observed only an additive effect on inhibiting TRE reporter gene activity in the presence of ⌬B and DIP1/2. When the concentration of ⌬B or DIP1/2 decreased to 0.1 or 0.2 g, respectively, we noticed that the inhibitory effect of each individual cDNA reduced significantly. Transfecting both cDNAs at this concentration, we observed a synergistic inhibitory effect on C4-2 cells. On the other hand, combing the ⌬B-S24A mutant with DIP1/2 failed to have any synergistic effect; it appears that Ser 24, a PKC substrate, in DOC-2/DAB2 is the key amino acid to modulate this activity. Thus, the binding of DIP1/2 to ⌬B can be enhanced by TPA, whereas ⌬B-S24A cannot (Fig. 6C). Taken together, these data indicate that the interaction between DIP1/2 and the N-terminal domain of DOC-2/DAB2 has a significant functional impact on gene transcription.
Biological Effect of DIP1/2 on Prostate Cancer Cells-Because DIP1/2 appears to be a negative regulator for the Rasmediated pathway, it may function as a growth inhibitor. To test this, C4-2 cells (a tumorigenic human prostate cancer cell line) were transfected with a DIP1/2 expression vector. Initially, we observed that there were fewer surviving clones in the DIP1/2-transfected plate than in plasmid control-transfected cells despite the same number of cells being used in transfection. After isolating two independent colonies (D1, D2), the protein levels of DIP1/2 in the D2 subline were higher than those in the D1 subline (Fig. 7A). Data from Fig. 7B indicated that expression of DIP1/2 significantly inhibited the in vitro cell growth compared with the plasmid control. This inhibitory effect of both D1 and D2 correlated with their DIP1/2 protein levels (Fig. 7A).
To rule out possible artifacts from stable transfection, we examined the growth suppression of DIP1/2 in C4-2 cells using transient transfection (16). As shown in Fig. 7C, the elevated DIP1/2 expression, determined by Western blot, inhibited the colony formation of C4-2 cells in a time-dependent manner.
We also examined whether the growth suppression effect of C4-2 cells can be enhanced in the presence of both DIP1/2 and DOC-2/DAB2. In this experiment, we used half of the amount of DIP1/2 cDNA from the previous experiment (Fig. 7C) and only the N-terminal domain of DOC-2/DAB2 (i.e. ⌬B), since we found some additional inhibitory activity associated with the C terminus (10). As shown in Fig. 7D, 0.1 g of ⌬B alone did not exhibit any growth-inhibitory effect, and DIP1/2 alone had a growth-inhibitory effect. However, we observed a synergistic effect of growth suppression of C4-2 cells transfected with both DIP1/2 and ⌬B cDNAs compared with C4-2 cells transfected with either DIP1/2 or ⌬B alone (Fig. 7D). In contrast, C4-2 cells transfected with both DIP1/2 m and ⌬B did not exhibit enhanced growth suppression. These data indicated that the interaction between DIP1/2 and DOC-2/DAB2 is critical for the growth-inhibitory effect of DIP1/2 in prostate cancer cells.

DISCUSSION
The DIP1/2 expression profile in different organs appears to be diverse. Northern analysis ( Fig. 2A) indicates a high level of DIP1/2 mRNA expression in brain, thymus, and bladder tissue and a low level in skeleton muscles, kidney, and liver tissue. But no expression can be detected in several urogenital organs, including the ventral prostate, dorsolateral prostate, seminal vesicle, and coagulating gland. This unique pattern of tissue distribution implies that DIP1/2 may have a specific physiological function in each organ. For example, DIP1/2 expression is detected in the enriched basal cell population of degenerated prostate and in prostatic epithelial cell lines (such as NbE and VPE) derived from the basal cell population ( Fig. 2A), suggesting that DIP1/2 may be involved in prostate regeneration. This hypothesis can be supported by our results: 1) decrease or absence of either DIP1/2 or DOC-2/DAB2 is often detected in several metastatic human prostate cancer cell lines (Fig. 3C), and 2) DIP1/2 appears to be a potent growth inhibitor for human prostate cancer cells (Fig. 7). It is known that increased Ras activity is associated with high grade metastatic prostate cancer; however, RAS mutation is rarely detected (33,34). Our results suggest an underlying mechanism with which to account for this phenomenon. In addition to DIP1/2, we found that altered expression of p120 GAP is associated with prostate cancer cells (data not shown). Thus, our results indicate that altered Ras GAP expression plays a critical role in the progression of prostate cancer.
The phosphorylation of Ser 24 in DOC-2/DAB2, which is involved in inhibiting TPA-induced AP-1 activity (11), provides evidence for the underlying functional mechanism of DOC-2/ DAB2. In this study, our data indicate that DIP1/2 is an immediate downstream effector for DOC-2/DAB2 in both prostate and brain tissues (Fig. 3D), and the binding of DIP1/2 to DOC-   7. Growth-inhibitory effect of DIP1/2 on prostate cancer cells. Cells were transfected with either pCI-neo or DIP1/2 expression vector. After G418 selecting, two independent clones were isolated and characterized. A, increased protein expression was detected by DIP1/2 antibody. B, the in vitro growth rate for each clone was determined by crystal violet assay. C, the colony formation of cells transfected with DIP1/2 cDNA. D, the colony formation of cells transfected with DIP1/2 cDNA and/or DOC-2/DAB2 cDNA. Data represent the mean Ϯ S.D. from six determinants. *, significantly different from plasmid control (p Ͻ 0.01).
2/DAB2 can be enhanced when the Ser 24 residue in DOC-2/ DAB2 is phosphorylated (Fig. 6C). The most conserved region in DIP1/2 protein is the GAP domain, which has a high amino acid sequence homology (40 -90%) compared with the GAP domains of other Ras GAPs, and DIP1/2's GAP domain contains all 31 consensus amino acids of other Ras GAPs. These consensus amino acid residues in the Ras GAP domain modulate Ras GTPase activity (13). For example, Arg 789 of p120 GAP participates in catalysis and simultaneously stabilizes Gln 61 of Ras for optimal GTP hydrolysis (35). Our data (Fig. 5) indicate that DIP1/2 has Ras GAP activity in vitro and in vivo. Since Arg 220 of DIP1/2 is equivalent to Arg 789 of p120 GAP , once Arg 220 was altered (R220L), the single amino acid mutant of DIP1/2 lost its Ras GAP activity in vitro and in vivo (Fig. 5, B and D). Similar to the GAP activity of p120 GAP (36), DIP1/2 can stimulate the GTPase activity of several small G-protein of Ras family (Fig. 5C). In majority of PCa cell lines used in this study, we were able to detect the presence of R-Ras, K-Ras and Ha-Ras (data not shown). Therefore, we believe that loss of GAP protein in prostate cells may be an underlying mechanism leading to constitutive activation of Ras in these cells.
We demonstrate that co-expression of DIP1/2 and the Nterminal domain of DOC-2/DAB2 (i.e. ⌬B) has an additive effect on suppressing either TPA-induced SRE or TRE reporter gene activity ( Fig. 6B and Table I). Therefore, we believe that the in vivo interaction of DIP1/2 with the N-terminal domain of DOC-2/DAB2 acts as a feedback mechanism to modulate PKC-induced gene activation. It is known that the modulation of the Raf/MEK/ERK axis is also controlled by growth factors through their protein receptor tyrosine kinase, although our data indicate that DIP1/2 alone is also able to inhibit EGF-induced SRE reporter gene activity (Fig. 6A) and cell growth. However, the interaction between DIP1/2 and DOC-2/DAB2 certainly amplifies their individual inhibitory effect (Figs. 6C and 7D and Table I). Therefore, this protein complex containing both DOC-2/DAB2 and DIP1/2 represents a unique negative regulatory machinery to balance signals elicited by growth factors (such as EGF) or PKC activators (such as TPA).
In addition to the N-terminal domain of Ras GAP homology domain of DIP1/2, the proline-rich repeats and leucine zipper domains from its C terminus may contribute to DIP1/2 activity. Our preliminary data indicate that the proline-rich repeats (residues 727-736) in DIP1/2 interact with Grb2 (data not shown). Because Grb2 binds to SOS, a guanine nucleotide exchange factor critical for downstream signaling, the binding of DIP1/2 to Grb2 may interrupt Ras activation. It is also possible that DIP1/2 can interact with other proteins containing the Src homology 3 domain. On the other hand, the leucine zipper domain (residues 842-861) of DIP1/2 may form a homodimer or a heterodimer with other proteins. Although no direct evidence has been shown for DIP1/2 dimerization, we observed a self-dimerization of DIP1/2 using the yeast twohybrid experiment, which suggests that the dimerization of DIP1/2 may affect its activity. More detailed studies are under way to examine this hypothesis.
In summary, both DIP1/2 and DOC-2/DAB2 form a unique protein complex (Fig. 8) with a negative regulatory activity that modulates the Ras-mediated signaling pathway. This complex is operative in basal cells of the prostate and may orchestrate the differentiation and proliferation potential of these cells during prostate regeneration. In contrast, altered expression of any component of this complex may result in abnormal growth and/or the acquired malignant phenotypes of prostate cancer and perhaps other types of cancer such as ovarian and breast cancer. Further dissection and functional examination of each component in this complex is warranted.