Epigenetic inactivation of the candidate tumor suppressor gene RASSF1A is a frequent and critical event in the pathogenesis of many human cancers. The RASSF1A protein contains a Ras association domain, suggesting a role in Ras-like signaling pathways, and has also been implicated in cell cycle progression. However, the preliminary data suggests that the RASSF1A gene product is likely to have multiple functions. To identify novel RASSF1A functions, we have sought to identify interacting proteins by yeast two-hybrid analysis in a human brain cDNA library. We identified the E1A-regulated transcription factor p120E4F as a RASSF1A interacting partner in yeast and mammalian cells, and demonstrated that RASSF1A protein and p120E4F form a complex in vivo. The interaction between RASSF1A and p120E4F was confirmed by both in vitro and in vivo pull downs and coimmunoprecipitation assays. In addition, specific inactivation of RASSF1A by short interfering RNA disrupts binding of RASSF1A to p120E4F in coimmunoprecipitation assays. In addition, we demonstrated enhanced G1 cell cycle arrest and S phase inhibition by propidium iodide staining of p120E4F in the presence of RASSF1A. As p120E4F has been reported previously to interact with p14ARF, retinoblastoma, and p53, these findings provide an important link between the function of RASSF1A and other major human tumor suppressor genes.
Frequent loss of heterozygosity at 3p21.3 in many sporadic human cancers and the discovery of overlapping homozygous deletions in lung and breast tumor cell lines helped to define a critical region of 120 kb at 3p21.3. Eight candidate tumor suppressor genes were cloned from within this critical interval (1) . Mutation analysis, however, did not reveal frequent mutations in the candidates CACNA2D2, PL6, 101F6, NPRL2/G21, BLU, RASSF1, FUS1, and LUCA2 (1) . However, one isoform of the RASSF1 gene, RASSF1A, has been shown to be epigenetically silenced by methylation of its promoter in various cancers including lung (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13) , breast, ovarian, nasopharyngeal, kidney, gastric, bladder, prostate, and testicular cancer, as well as neuroblastomas and pheochromocytomas. RASSF1 is expressed as several splice variants. The expression of the longer isoform RASSF1A is lost in many tumor cell lines and primary tumors, whereas the shorter isoform RASSF1C is well expressed in tumor cell lines and primary tumors. Hence, RASSF1A belongs to an increasing list of tumor suppressor genes that undergo frequent promoter hypermethylation but rare somatic mutations (14) . Furthermore, RASSF1A has also been demonstrated to suppress tumorigenicity in nude mice, reduce colony formation, and suppress anchorage-independent growth (2 , 4) . Thus, the evidence to date suggests that inactivation of RASSF1A is important in the pathogenesis of many human cancers.
RASSF1A is a Mr 39,000 (340 amino acids) protein containing two putative functional domains including a diacylglycerol binding domain (50–101 amino acids) at the NH2 terminus (absent in isoform C) and a RAS association domain (194–288 amino acids) in the COOH terminus. The presence of a RAS association domain suggests that RASSF1 proteins may function as RAS-effectors. Many known Ras effectors are oncoproteins in their own right; however, less is known about Ras effectors possessing tumor suppressor properties Consequently, the characterization of proteins encoded by tumor suppressor genes such as RASSF1A, their binding partners, and their location is crucial to understanding their function and role in tumor development.
Recently, several additional RASSF homologues have been reported, namely NORE1, RASSF2, and RASSF3. The biological functions of these polypeptides are not known. Murine Nore1 binds to RAS and several RAS-like GTPases in a GTP-dependent fashion; however, RASSF1A and C are not able to bind RAS directly. RASSF1A, however, can heterodimerize with NoreI and thereby associate with Ras-like GTPases (15) . The dimerization of RASSF1A requires the NH2-terminal 119 amino acids, which are not found in the C isoform. Human Nore1 interacts with the proapoptotic protein kinase MST1 to mediate a novel RAS-regulated apoptotic pathway. Interestingly, RASSF1A also interacts with MST1 suggesting a network of interacting proteins in the RAS-mediated apoptosis pathway (16) . More recently, we and others have demonstrated that RASSF1A blocks cell cycle progression and inhibits cyclin D1 accumulation (17 , 18) . Furthermore, our microarray analysis identified RASSF1A gene targets involved in many other cellular processes important in tumorigenesis including apoptosis, cell adhesion, transcription, protein synthesis, signaling, and cell migration (18) . Clearly, RASSF1A has a profound effect on gene expression and, hence, tumor cell biology. However, how these effects are exerted is not yet known. To establish a clearer picture of what the functions might be, and to identify additional cellular proteins that might bind to RASSF1A and modulate its biological activity we used a yeast two-hybrid approach to screen a human brain cDNA library for RASSF1A binding proteins. In this study, we isolated the ubiquitously expressed E1A-regulated transcription factor p120E4F a GLI-kruppel-related zinc finger phosphoprotein involved in cell cycle progression as a RASSF1A interacting partner and show that when coexpressed cell cycle arrest is enhanced.
MATERIALS AND METHODS
The bait plasmid pGBKT7-RASSF1A and pGEX-4T-1-RASSF1A were constructed by restriction digest of full-length RASSF1A from pcDNA3.1 using EcoR1 and cloned in frame into yeast vector pGBKT7 (BD-Clontech) and the glutathione S-transferase (GST) expression vector pGEX-4T-1, respectively (Amersham Pharmacia Biotech). RASSF1A derivatives were originally cloned into pcDNA3.1 using the restriction enzymes BamHI and NotI (New England Biolabs). Each construct was generated using primers specific for each region required before cloning. The resulting proteins generated for GST assays were as follows: 1–119RASSF1A-pGex-4T-1, RASSF1C-pGex-4T-1, and 143–340RASSF1A-pGex-4T-1. Deletion constructs of p120E4F were made by PCR amplification of p120E4F using primers specific for each region required. For the generation of mammalian expression plasmids full-length RASSF1A and p120E4F clones were also cloned in frame into the mammalian expression vector pCMV-HA.
All of the deletion constructs were generated using PFU turbo (Promega) and checked by sequence analysis. In vitro translated proteins were generated using a TNT-coupled reticulocyte lysate system (Promega) according to the manufacturer’s instructions. pEGFP-F was generated by inserting the farnesylation signal for H-Ras in frame into pEGFP-C2 (Clontech).
Yeast Two-Hybrid Screen.
Yeast two-hybrid screening was performed using MATCHMAKER system 3 (BD-Clontech). The bait vector pGBKT7-RASSF1A was transformed in yeast strain Saccharomyces cerevisiae AH109 (RASSF1A-BD) using the LiAc/PEG method and selected on Trp-plates. Strain AH109 includes four reporter genes, ADE2, HIS3, lacZ, and MEL1, of which the expression is regulated by GAL4-responsive upstream activating sequences and promoter elements. The bait RASSF1A-BD was then used to screen a pretransformed MATCHMAKER brain library cloned in pACT2 and introduced into yeast strain S. cerevisiae Y187. A total of 6 × 106clones were screened, of which 103 were positive for α-galactosidase expression and turned blue. Clones were rescued and retransformed into yeast to reconfirm positive interactions. Positive clones were sequenced using automated DNA sequence analysis (ABI) and homologies identified using National Center for Biotechnology Information BLASTN/BLASTX.
Coimmunoprecipitation and Western Blot Analysis.
In vitro translated E4F and RASSF1A was generated using the TNT-coupled reticulocyte lysate system (Promega). The T7 promoter was incorporated into the sequence of library cDNAs using the following commercially available primers: AD HA forward Co-IP primer and AD Reverse Co-IP primer (BD-Clontech). PCR products were cleaned using a PCR clean up kit (Qiagen) according to the manufacturer’s instructions. For coimmunoprecipitation studies, 10 μl of in vitro translated (Transcend tRNA labeled) bait and library protein were mixed and incubated for 1 h at room temperature before the addition of either 1 μg of c-Myc monoclonal antibody or HA-polyclonal antibody, in the presence of Protein G/A agarose beads (Santa Cruz Biotechnology) and coimmunoprecipitation buffer [freshly prepared 20 mm Tris-HCL (pH7.5), 150 mm NaCl, 1 mm DTT, 5 μg/ml aprotinin, 0.5 mm phenylmethylsulfonyl fluoride, and 0.1% Tween 20 (v/v)]. Samples were then incubated for an additional 2 h at 4°C on a rotating wheel. The beads were then washed three times in 500 μl of TBST (Tris buffered saline and 0.1% Tween 20), and the bound protein was then solubilized by addition of 15 μl of 2× SDS sample buffer. Once eluted the samples were then denatured before loading onto a 7.5–14% SDS-PAGE minigel. Western blotting was performed according to standard procedures. Membranes were then probed with c-Myc monoclonal antibody, HA polyclonal antibody (Clontech), RASSF1A polyclonal antibody N15 (Santa Cruz Biotechnology), RASSF1A monoclonal antibody (eB114 eBiosciences), β- actin monoclonal (Sigma), or E4F polyclonal antibody-88.2 (generated using a NH2-terminal peptide of p120E4F corresponding to amino acids 50–64 EEDEDDVHRCGRCQA; Ref. 19 ) and the appropriate secondary antibodies. The bands were then visualized by enhanced chemiluminescence (Amersham). When no antibody was appropriate Coomassie blue staining was applied (according to standard procedures).
Fusion Protein Expression and GST Pull-Down Assay.
To prepare GST fusion proteins, Escheria coli strain BL29 was transformed with a pGEX expression vector (pGEX-4T-1) containing the appropriate insert. Briefly, overnight cultures were diluted 1:10 and grown to A600 = 0.7, then the generation of fusion proteins was induced by the addition of 0.4 mm isopropyl-d-thiogalactoside at 30°C overnight. Pelleted cells were lysed by mild sonication (3 × 30 s blasts) in ice-cold PBS (1 × PBS) containing 1% Triton X-100 and Complete protease inhibitor (Roche) before resuspension in Bugbuster protein extraction reagent. After centrifugation, supernatants were applied to glutathione Sepharose 4B beads (Amersham) at 4°C overnight, collected, and resuspended in 1× PBS and 10% glycerol before storage at −80°C. Immobilized GST protein was detected by SDS-PAGE and Western blotting using Coomassie blue staining and probing with a GST specific antibody (Amersham).
For the pull-down assay 10 μl in vitro translated proteins (generated using the TNT-coupled reticulocyte lysate system) was incubated with 25 μl GST-RASSF1A bound beads in a final volume of 440 μl of PBS and incubated for 1 h at 30°C, followed by an additional incubation for 2 h at 4°C with rotation. The beads were then pelleted, washed three times in 1 × PBS, before being resuspended in 2 × SDS sample buffer and analyzed by SDS-PAGE (10–13% gels). Western blotting was performed according to standard procedures. Membranes were then probed with the appropriate primary and secondary antibodies. The bands were then visualized by enhanced chemiluminescence.
After transfection (48 h; except for endogenous immunoprecipitations, which were harvested when confluent) cells were harvested by trypzination and washed with ice-cold 1 × PBS. Pelleted cells were lysed with modified radioimmunoprecipitation assay buffer (Mammalian Cell Lysis kit; Sigma) containing Complete protease inhibitor (Roche) and mild sonication (3 × 10 s). Protein content was standardized as determined by the Bradford assay using Coomassie Plus Protein Assay Reagent (Bio-Rad). Fifty-μg lysates were precleared with protein A/G PLUS agarose beads, and then incubated with appropriate antibody and beads overnight at 4°C on a rotating wheel. The beads were then pelleted, washed three times in modified radioimmunoprecipitation assay buffer, before resuspension in 2× SDS sample buffer and analyzed by SDS-PAGE (10–13% gels). Western blotting was preformed according to standard procedures. Membranes were then probed with the appropriate primary and secondary antibodies. The bands were then visualized by enhanced chemiluminescence.
Cell Lines Used in This Study.
All of the cell lines were maintained in DMEM containing 10% fetal bovine serum in a 37°C incubator and 5% CO2. For cell cycle distribution and protein analysis, cultured cells (2 × 105) were seeded in 25-cm2 flasks and cotransfected with the expression plasmids (2 μg). Cells were harvested using trypsin 48 h after transfection.
Small Interfering RNA (siRNA) Treatment of Cells.
The double-stranded siRNA oligonucleotide targeting RASSF1A was synthesized by MWG Biotech United Kingdom Ltd. (Milton Keynes, United Kingdom), and the sequences used were those published previously (17) : sense 5′-GACCUCUGUGGCGACUUCATT-3′ and antisense 5′-UGAAGUCGCCACAGAGGUCTT-3′.
A negative control duplex (Ambion Europe Ltd., Huntingdon, United Kingdom) was used to demonstrate that transfection did not induce nonspecific effects on gene expression. The day before transfection, cells were plated onto six-well cell culture plates in 2 ml of growth medium without antibiotics and grown to 30–50% confluence. On the day of transfection, for each transfection sample, the duplexes were diluted to give a final concentration of 20 nm in Opti-Mem I (Invitrogen Life Technologies, Inc., San Diego, CA). Sixty μl of Oligofectamine reagent (Invitrogen Life Technologies, Inc.; diluted 1:4 with Opti-Mem I) were added to the diluted duplex, and the mixture was incubated at room temperature for 20 min to allow the siRNA:Oligofectamine complexes to form. This mixture was then added to the transfection well and incubated for 72 h at 37°C before whole cell lysates were taken.
Cell Cycle Analysis.
All of the transfected cells were cotransfected with a 1:5 ratio of pEGFP-F, a GFP vector, and appropriate vector. This was used as a measure of transfection efficiency and allowed accurate gating (20) . Forty-eight h after transfection, cells were trypsinized and washed with PBS before fixing in 50% ethanol in 1% BSA at 4°C for 30 min. Cells were stained with propidium iodide (50 μg/ml) containing RNase A (1 mg/ml). DNA content from at least 25,000 events was analyzed by flow cytometry using a Coulter Epics Xl-MCL flow cytometer running system software. Data were collated using WINMDI software and then analyzed using Cylchred software. S phase inhibition was calculated from four independent experiments using the following formula: [(percentage of cells transfected with vector alone cells in S phase) − (percentage of transfected cells in S- phase in RASSF1A, p120E4F, or RASSF1A/p120E4F)]/(percentage of cells transfected with vector alone cells in S phase) × 100.
Isolation of E4F as a Candidate RASSF1A Interacting Protein in a Yeast Two-Hybrid Screen.
A yeast two-hybrid screen was used to identify proteins able to associate with RASSF1A. The bait vector pGBKT7-RASSF1A was constructed and transformed into yeast strain S. cerevisiae AH109. The bait RASSF1A-BD was then used to screen a pretransformed MATCHMAKER human adult brain cDNA library (BD-Clontech). Approximately 6 × 106 independent transformations were screened, of which 103 clones were positive for α-galactosidase expression. Nucleotide sequence analysis revealed that two separate clones (designated E4FΔ) contained a 1-Kb insert with an open reading frame of 1113 bp encoding a protein fragment of 324 amino acids, which corresponds to the COOH-terminus half of the E1A regulated transcription factor p120E4F (NP 004415). We also identified a number of other interesting targets including the proapoptotic protein kinases MST1 and MST2 (10 clones) already known to interact with RASSF1A (16) . The remaining novel binding partners are under additional investigation.
In the absence of either RASSF1A or E4FΔ, or the replacement of RASSF1A by the unrelated protein lamin, diploid cells generated no α-galactosidase signal. To determine the region of RASSF1A essential for interaction with E4FΔ, RASSF1C (the shorter variant missing amino acids 1–119) was expressed as a fusion protein with the yeast two-hybrid vector pGBKT7 and tested for interaction with E4FΔ. No α-galactosidase production was observed as an indicator of a direct interaction (data not shown). p120E4F contains 10 putative zinc finger domains, which are clustered in two separate regions (Fig. 1A) ⇓ . Four motifs are contained within an NH2-terminal domain that is also present in p50E4F (cleaved product), whereas the remaining 6 motifs are grouped within a central region (amino acids 277–437) found only in the full-length protein (1) . The isolated clone E4FΔ (3) encodes the last 5 zinc finger domains and all of the additional COOH-terminal residues that are specific to p120E4F (Fig. 1A) ⇓ .
To verify the interaction in vitro translated RASSF1A-myc and E4FΔ -HA were tested in a coimmunoprecipitation before resolving on a 10% SDS-PAGE gel and Western blotting. In agreement with the data obtained in the yeast two-hybrid system an interaction was observed generating specific bands close to the predicated sizes of Mr 41,000 and Mr 60,000. These bands were absent in the presence of empty vector. Therefore, these data demonstrate that RASSF1A can directly associate with p120E4F in yeast (Fig. 1B) ⇓ .
GST-RASSF1A Interacts with E4FΔ and p120E4Fin Vitro.
To confirm that RASSF1A interacted with E4FΔ and full-length p120E4F in vitro, a GST pull down assay was performed. A GST fusion protein corresponding to full-length RASSF1A (1–340 amino acids) was created and incubated with in vitro translated E4FΔ and p120E4F protein. E4FΔ and wild-type p120E4F specifically bound to the GST-RASSF1A (1–340) fusion protein but not to the GST alone (Fig. 1C) ⇓ , indicating a physical association between these two proteins in vitro.
RASSF1A and p120E4F Interact in Vivo.
Full-length RASSF1A and p120E4F-HA were overexpressed transiently in mouse NIH 3T3 fibroblast cells and human A549 non-small cell lung cancer line. Approximately 48 h after transfection, cell lysates were prepared and protein expression confirmed (Fig. 2A) ⇓ . Cell lysates were then subject to immunoprecipitation with RASSF1A polyclonal antibody and control IgG antibody before running on a 10% SDS-PAGE gel and Western blot analysis with E4F (88.2) antibody. As shown in Fig. 2A ⇓ , p120E4F-HA was efficiently coprecipitated in both NIH 3T3 and A549 cells with RASSF1A but not control IgG. In addition the specificity of the Co-IP was confirmed by Western blotting with anti-HA antibody also (data not shown). These data show that RASSF1A physically interacts with p120E4F in vivo when both proteins are exogenously overexpressed.
To determine whether endogenous RASSF1A interacts with p120E4F in vivo an immunoprecipitation was performed using human breast carcinoma cell line MCF7 (RASSF1A negative) and non-small cell lung tumor cell line NCI-H1792 (RASSF1A positive). Cell lysates were prepared and subjected to immunoprecipitation with RASSF1A-specific polyclonal antibody. As shown in Fig. 2B ⇓ endogenous levels of RASSF1A interact with p120E4F in vivo. In addition NCI-H1792 cells were treated with RASSF1A-siRNA before cell lysates were subject to immunoprecipitation (Fig. 2C) ⇓ . In the absence of endogenous RASSF1A no specific interaction was seen with p120E4F in NCI-H1792 cells.
NH2 Terminal of RASSF1A Interacts with p120E4F.
To additionally define the region of RASSF1A involved in the interaction with p120E4F, various GST-fusion constructs of RASSF1A deletions were created and tested for their ability to bind to in vitro translated p120E4F (see Fig. 3 ⇓ ). As summarized in Fig. 3 ⇓ A, full-length GST-RASSF1A protein (1–340 amino acids) bound strongly to p120E4F (construct 1). Likewise, p120E4F bound strongly to the smaller COOH-terminal truncated GST construct 1–119 but failed to bind with the NH2-terminal truncated construct GST 143–340. We observed a very weak band in GST-RASSF1C (Fig. 3A ⇓ , Lane 4). Together this data suggests that the residues 1–119 specific to RASSF1A are important for the interaction with p120E4F.
RASSF1A Interacts with Amino Acids 411–445 of p120E4F.
The previous data demonstrated that residues between 1 and 120 are required for RASSF1A to bind to p120E4F. To define the region of p120E4F required for binding to RASSF1A several p120E4F deletions were constructed and in vitro translated protein produced. These were then used for pull-down assays with GST-RASSF1A before resolving on 10 or 13% SDS-PAGE gel and Western blotting was performed (Fig. 4) ⇓ . As demonstrated previously, wild-type p120E4F bound strongly to GST-RASSF1A, as did Δ E4F, which is missing the first zinc finger domain. Deletion of the COOH-terminal 243 amino acids had no effect on the ability of p120E4F to bind GST-RASSF1A (construct 3). Removal of residues 299–683 that includes the central zinc finger domain (constructs 4–8) completely abolished the interaction indicating that the site of interaction lies within this region. Sequential deletion of each zinc finger within the second region indicated that the site of interaction lies within the COOH-terminal zinc finger. Only construct 9, which contained the last zinc finger motif, might be involved in mediating the interaction. Indeed removal of the last intact zinc finger motif (construct 10: 445–683 amino acids) resulted in no interaction. In addition construct 11 containing only the last zinc finger motif (417–560 amino acids) demonstrated interaction with RASSF1A. These data indicate that p120E4F contacts RASSF1A through a COOH-terminal region located between amino acids 411 and 445.
Coexpression of p120E4F and RASSF1A Induces Enhanced Cell Cycle Arrest.
The expression of p120E4F can suppress cell proliferation at the G1-S phase and G2-M phase transition. To assess the effect of RASSF1A and the effect of the interaction of p120E4F and RASSF1A on growth arrest, NIH 3T3 cells were transiently transfected with RASSF1A and/or p120E4F. At 48 h after transfection, p120E4F had a small effect on the cell cycle distribution of NIH 3T3 cells (Fig. 5, A and B) ⇓ . The percentage of cells in G1 was increased by 10%, G2 was reduced by ∼6%, whereas significant inhibition of S phase was evident 15.5 ± 3% (P < 0.05). RASSF1A overexpression had a small effect (>5%) on the percentage of cells in G1 compared with vector alone. In addition a small reduction in G2 was evident (∼5%), whereas no significant difference was seen on the percentage of cells in S phase (0.5 ± 1%; Fig. 5B ⇓ ). Coexpression of RASSF1A and p120E4F, however, caused S phase inhibition to increase to 40 ± 4%, whereas increasing the number of cells in G1 by ∼18%. Little additive effect was seen in the G2 peak. Thus, coexpression of p120E4F and RASSF1A induced enhanced cell cycle arrest.
The RASSF1A gene isoform is thought to be a major tumor suppressor gene on chromosome 3p21.3. RASSF1A is epigenetically inactivated in a significant number of lung, breast, kidney, and other cancers. Re-expression of RASSF1A is sufficient to revert the tumorigenicity phenotype of tumor cell lines (2 , 4) . Microarray analysis has identified a number of RASSF1A gene targets, which may be involved in this process (18) . However, the mechanisms of RASSF1A tumor suppression are not yet understood. Identification of RASSF1A interacting proteins is a pivotal step for understanding its function in tumorigenesis. We used the yeast two-hybrid method to identify proteins that interact with RASSF1A. Here we have isolated a COOH-terminal fragment specific to full-length p120E4F, a transcription factor regulated by E1A. We demonstrate that RASSF1A protein and p120E4F form a complex in vivo. This interaction between RASSF1A and p120E4F was confirmed by both in vitro and in vivo assays. Blockage of RASSF1A by siRNA eliminated binding of p120E4F, and no specific signal for p120E4F was detected. We were also able to demonstrate enhanced G1 cell cycle arrest and S phase inhibition by propidium iodide staining of p120E4F in the presence of RASSF1A. These data suggest that p120E4F may be an important RASSF1A effector in cell cycle control.
In mice and humans p120E4F is a GLI-kruppel-related zinc finger phosphoprotein that is ubiquitously expressed. p120E4F is synthesized as a Mr 120,000 protein that upon proteolytical cleavage gives rise to p50E4F, a Mr 50,000 NH2-terminal fragment. Full-length p120E4F (murine homologue φ AP3) is one of two related zinc finger proteins that differentially regulate the adenovirus E4F gene in an E1A-dependent manner (21) . The full-length protein, p120E4F, represses this promoter in the absence, but not in the presence of E1A. In response to adenovirus E1A or treatment with phorphol ester, p120E4F becomes hyperphosphorylated and undergoes a reduction in both DNA binding and transcriptional repressor activities (22 , 23) . The other E4F protein, p50E4F is a proteolytically derived Mr 50,000 NH2-terminal fragment of p120E4F that has been shown to stimulate the E4 promoter when coexpressed with E1A. Interestingly, although p50E4F and p120E4F can recognize the same DNA motif (RTGACGTC/AAY) in vitro, the additional residues specific to p120E4F confer functional and regulatory properties distinct from p50E4F. Expression of p120E4F significantly inhibits cellular growth and cell cycle progression. p50E4F has no such effect. p120E4F, but not p50E4F, blocked cell proliferation when expressed in mouse fibroblasts by inducing cell cycle arrest near the G1-S transition (19 , 21, 22, 23, 24, 25, 26, 27) .
p120E4F has been shown to be associated with a number of cell cycle regulating proteins including cyclin E, cyclin B, the post-transcriptional up-regulation of cyclin-dependent kinase inhibitors p21waf1 and p27kip1, reduced cdk2, cdk4/6, and cdc2 kinase activities, and the down-regulation of cyclin A gene expression (19 , 21, 22, 23, 24, 25, 26, 27) . Cyclin A is the only cellular target demonstrated to be regulated by p120E4F to date (25) . There are probably other genes regulated by p120E4F that are involved in cell cycle progression because p120E4F repression was not affected by a reduction in cyclin A protein and RNA levels. Cell cycle arrest by p120E4F is also enhanced by interacting with retinoblastoma, the p53 transcription factor (19 , 24) and the p14 ARF tumor suppressor (28) . Thus, although the precise cellular signals that regulate endogenous p120E4F have yet to be clearly defined the evidence to date suggests that p120E4F may regulate cell cycle progression in response to a number of different signals.
Here we demonstrate that p120E4F associates both in vitro and in vivo with RASSF1A protein in human and mouse cells to form a complex, which modulates cell cycle progression. The strong association with p120E4F occurs mainly through residues that are specific to RASSF1A and are required for its growth suppression activity, and amino acids 411–445 of p120E4F, a region that contains both p53 and retinoblastoma protein binding domains (19 , 24) .
p120E4F contacts RASSF1A in vivo and enhances RASSF1A- mediated growth arrest. At 48 h after transfection, RASSF1A alone induced a small increase in G1 phase cells, with no significant effect on S phase. Overexpression of p120E4F affected cell cycle consistent with previous published data (19 , 21, 22, 23, 24, 25, 26, 27) . Another study has demonstrated recently in the non-small cell lung cancer cell line NCI-H1299 that RASSF1A induced G1-S cell cycle arrest and blocked accumulation of cyclin D1 (17) . Although we did not observe a dramatic effect by RASSF1A alone in our experiments we cannot rule out contrasting results due to the cells lines used or culture conditions. More importantly, however, NIH-3T3 cells overexpressing both RASSF1A and p120E4F showed a much greater reduction in the percentage of S phase cells, with a consistent increase in the number of cells in G1 phase. Interestingly it has also been shown that p120E4F also affects G2 phase progression (27) , although the mechanisms underlying this effect are not clear. We also saw a small difference in G2 phase in the cell lines we studied.
Both proteins have been demonstrated to evoke changes in cell cycle regulatory proteins, including the post-transcriptional elevation of cyclin B1, p21Waf1, and p27Kip1 protein levels, reduced expression of cyclin A, and reduced levels of cdk2 and cdc kinase activities in the case of p120E4F, and down-regulation of cyclin D1 in the case of RASSF1A (17, 18) . It is conceivable that p120E4F may be involved in modulating the transcriptional function of RASSF1A. A better understanding of the role of the p120E4F/RASSF1A complex requires characterization and functional analysis of other factors that associate with the COOH half of p120E4F and the NH2 terminus of RASSF1A. Additional studies need to be focused on understanding the regulation of cellular RASSF1A, as well as the molecular mechanisms of cell cycle control and identification of cellular target genes. Ultimately control of gene expression generally requires the interaction of multiple factors, which make up the transcriptional complex, including several activators, coactivators, and repressors, which stabilize the transcriptional machinery. The evidence presented here that p120E4F can physically associate with RASSF1A suggests that p120E4F is a novel interacting partner in the signaling pathway(s) of RASSF1A.
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.
Grant Support: Breast Cancer Campaign, AICR (Association for International Cancer Research), The Wellcome Trust, SPARKS (Sport Aiding Medical Research for Kids), and Cancer Research United Kingdom.
Requests for reprints: Farida Latif, Department of Reproductive and Child Health, The Medical School, Edgbaston, Birmingham, B15 2TT, United Kingdom. Fax: 44-0-121- 627-2618; E-mail:
- Received August 21, 2003.
- Revision received October 30, 2003.
- Accepted October 31, 2003.
- ©2004 American Association for Cancer Research.