Anchorage-Independent Growth of Pocket Protein-Deficient Murine Fibroblasts Requires Bypass of G₂ Arrest and Can Be Accomplished by Expression of TBX2

Abstract

Mouse embryonic fibroblasts (MEFs) deficient for pocket proteins (i.e., pRB/p107-, pRB/p130-, or pRB/p107/p130-deficient MEFs) have lost proper G₁ control and are refractory to Ras^V12-induced senescence. However, pocket protein-deficient MEFs expressing Ras^V12 were unable to exhibit anchorage-independent growth or to form tumors in nude mice. We show that depending on the level of pocket proteins, loss of adhesion induces G₁ and G₂ arrest, which could be alleviated by overexpression of the TBX2 oncogene. TBX2-induced transformation occurred only in the absence of pocket proteins and could be attributed to downregulation of the p53/p21^CIP1 pathway. Our results show that a balance between the pocket protein and p53 pathways determines the level of transformation of MEFs by regulating cyclin-dependent kinase activities. Since transformation of human fibroblasts also requires ablation of both pathways, our results imply that the mechanisms underlying transformation of human and mouse cells are not as different as previously claimed.

Deregulation of the pRB tumor suppressor pathway is a frequent event in the development of cancer (18, 31). pRB and its close homologs p107 and p130 comprise the family of so-called pocket proteins and are widely known for their role in cell cycle regulation, especially during G₁ phase. In their active, hypophosphorylated form, pocket proteins restrict cell cycle progression by binding to E2F transcription factors. E2F-pocket protein complex formation inhibits the expression of E2F target genes both by blocking E2F's ability to induce transcription and by active repression. As a result, initiation of S phase is inhibited. Upon cell cycle stimulation, cyclin D-CDK4/6 (cyclin-dependent kinase 4 or 6) and cyclin E-CDK2 complexes are activated and hyperphosphorylate the pocket proteins, resulting in liberation and activation of E2F transcription factors and initiation of S phase. Further cell cycle progression requires cyclin A-CDK1/2 activity in S phase and cyclin A-CDK1/2 and cyclin B1-CDK1 activities in G₂/M phase. The activity of cyclin-CDK complexes can be inhibited by the INK4a (inhibitor of cyclin-dependent kinase 4a) and the CIP/KIP families of CDK inhibitors (reviewed in reference 2).

Consistent with a role for pocket proteins in G₁ control, we and others have shown that complete ablation of pocket proteins in mouse embryonic fibroblasts (MEFs) abrogated G₁ arrest in response to growth-inhibitory signals, such as cell-cell contact, growth factor depletion, and DNA damage. Additionally, upon prolonged culturing or expression of constitutively active Ras (Ras^V12), wild-type MEFs arrested in G₁ and displayed hallmarks of senescence, while MEFs deficient for both pRb and p107 or both pRb and p130 (double-knockout [DKO] MEFs) or all three pocket proteins (triple-knockout [TKO] MEFs) were refractory to replicative and Ras^V12-induced senescence (6, 7, 32, 38). Similar to ablation of pocket proteins, ablation of the tumor suppressor p53 or its upstream regulator p19^ARF also bypassed replicative and Ras^V12-induced senescence in MEFs (22). However, while p53- or p19^ARF-deficient MEFs could easily be transformed by Ras^V12, pocket protein-deficient MEFs expressing Ras^V12 were unable to exhibit anchorage-independent growth and did not form tumors in nude mice (32; this communication). This demonstrates that the loss of pocket proteins in primary MEFs is not sufficient for Ras^V12-induced transformation.

In the present study, we performed a cDNA screen aimed at identifying oncogenes that collaborate with the loss of pocket proteins and Ras^V12 in transformation of MEFs. We identified the oncogene TBX2 as a cooperating factor in transformation of pocket protein-deficient MEFs. TBX2 downregulates the p53/p21^CIP1 pathway both in wild-type and pocket protein-deficient MEFs, and indeed, downregulating the p53/p21^CIP1 pathway by RNA interference transformed pocket protein-deficient MEFs. In contrast to previous reports, our results demonstrate that Ras^V12-induced transformation of murine fibroblasts requires downregulation of both the pocket protein and p53 pathways. Since downregulation of both pathways is also required for transformation of human fibroblasts (3, 35, 46), our results indicate that the mechanisms underlying Ras^V12-induced transformation of human and mouse fibroblasts are not fundamentally different. Furthermore, we show that anchorage-independent growth not only requires bypass of G₁ control but also requires bypass of G₂ control.

MATERIALS AND METHODS

MEF isolation, cell culture, and retroviral infections.

Rb^−/−, p107^−/−, p130^−/−, p107^−/− p130^−/−, Rb^−/− p107^−/−, Rb^−/− p130^−/−, and Rb^−/− p107^−/− p130^−/− MEFs were isolated from chimeric embryos, which were generated by injection of mutant embryonic stem cells into blastocysts as previously described (7). MEFs were cultured in Glasgow minimal essential medium (GMEM) (Invitrogen/Gibco) containing 10% fetal calf serum, 1 mM nonessential amino acids (Invitrogen/Gibco), 10 mM sodium pyruvate (Invitrogen/Gibco), 100 units of penicillin/ml (Invitrogen/Gibco), 100 μg streptomycin/ml (Invitrogen/Gibco), and 0.1 mM β-mercaptoethanol and incubated at 37°C in the presence of 5% CO₂. Retroviral supernatants were produced by calcium phosphate transfection (Invitrogen) of phoenix cells with 16 μg of the desired construct and 4 μg pCL-eco. Forty-eight hours posttransfection, retroviral supernatant was filtered using 0.45-μm filters (mixed cellulose ester membrane; Millipore) and immediately frozen using a dry ice-ethanol bath and stored at −80°C. After viral supernatant was harvested, phoenix cells were supplemented with GMEM containing medium supplements as described above, and supernatant was again harvested using the same procedure, with an interval of at least 7 h. MEFs were twice infected with viral supernatant, supplemented with Polybrene to a concentration of 4 μg/ml, during a time span of at least 7 h per infection. For serial infections, MEFs were cultured in non-virus-containing medium for 48 h between infections and reseeded before infection to obtain optimal cell density. After the last infection, MEFs were cultured in non-virus-containing medium for at least 48 h.

Constructs.

The pEYK-MCF7 library and pEYK-GFP were a gift of G. Q. Daley. pCl-Eco was kindly provided by D. Peeper. pBABE-Ras^V12 was kindly provided by T. Brummelkamp, pRetroSuper-p53 and pRetroSuper-p16^INK4A by A. Dirac, and pBABE-p21^CIP1 by J. Dannenberg. The pRetroSuper-p21strong vector was previously generated and contains the following 19-mer p21^CIP targeting sequence: GCCCTCACTCTGTGTGTCT (11). The 19-mer p21^CIP targeting sequence in pRetroSuper-p21weak is ACAGGAGCAAAGTGTGCCG. The pRISC-p16^INK4A/p19^ARF vector was a gift of S. Huang; this vector is a variant of pRetroSuper (pRS) containing an additional chloramphenicol resistance marker under the control of a TET promoter and contains the following targeting sequence: ATCAAGACATCGTGCGATA.

Soft agar and methylcellulose assays.

For soft agar assays, 6 × 10⁴ MEFs were suspended in 2 ml of a 37°C, 0.35% soft agar solution (low gelling agarose type VII from Sigma) in GMEM containing 10% fetal calf serum, the same medium supplements as mentioned above plus gentamicin to a concentration of 0.02 mg/ml (Invitrogen/Gibco) and plated in one well of a six-well plate. To prevent cells from attaching to the bottom of the well, the 0.35% soft agar solution was poured into an ultra-low-attachment surface plate (catalog no. 3471; Corning Incorporated) coated with a 1% soft agar layer. To allow solidification of the agar, plates were incubated at 4°C for 30 min. Subsequently, cells were incubated at 37°C in the presence of 5% CO₂ for 2 to 4 weeks. A few drops of fresh medium were added twice a week. Pictures were taken using a non-phase-contrast lens (×2.5 magnification) and assembled using Axiovision 4.5. Detail images were taken using a phase-contrast lens (×5 magnification). For methylcellulose assays, 3 × 10⁵ MEFs were suspended in 4 ml of a 37°C, 1.3% methylcellulose solution in GMEM supplemented with fetal calf serum to a concentration of 5%, penicillin to a concentration of 100 units/ml (Invitrogen/Gibco), streptomycin to a concentration of 100 μg/ml (Invitrogen/Gibco), and gentamicin to a concentration of 0.02 mg/ml (Invitrogen/Gibco) and plated in one well of a six-well ultra-low-attachment surface plate (catalog no. 3471; Corning Incorporated). A 2.6% methylcellulose stock solution was obtained from Stem Cell Technologies (catalog no. H4100). Cells were harvested by suspending 4 ml of methylcellulose culture with 40 ml ice-cold phosphate-buffered saline (PBS) (Invitrogen/Gibco), followed by centrifugation and aspiration of methylcellulose-PBS.

Recovery of retroviral integrations from soft agar colonies.

Soft agar colonies were isolated using sterilized glass pipettes and propagated under attached conditions. Cells were lysed overnight (O/N) at 55°C in lysis mix containing 0.1 M Tris HCl (pH 8.5), 5 mM EDTA, 0.2 M NaCl, 0.2% sodium dodecyl sulfate (SDS), and 100 μg/ml proteinase K. Genomic DNA was isolated by phenol-chloroform-isoamyl alcohol extraction, precipitated using isopropanol, and dissolved O/N at 37°C in 10 mM Tris HCl (pH 8.0)-0.1 mM EDTA. Recovery of retroviral inserts was performed using a shuttle strategy modified from the method of Koh et al. (24). Five micrograms of genomic DNA was digested using 50 units of either NotI (Roche) or AscI (New England Biolabs) O/N in 100 μl, followed by an additional 3-h digestion upon the addition of 25 units. For each analyzed colony, both digestions were performed. Subsequently, fragments were purified using phenol-chloroform-isoamyl alcohol extraction and ethanol precipitated in the presence of 200 μg of type VII mussel glycogen (catalog no. G1508; Sigma) at −80°C. Ligation was performed O/N at 16°C in a total volume of 200 μl using 2 units of T4 DNA ligase (Roche) and in the presence of extra ATP to a concentration of 0.05 mM. An extra 3-h ligation was performed the next day upon the addition of 2 units of T4 DNA ligase. Ligated plasmids were purified as described above, electroporated in 5 μl of DH10B electromax competent cells per reaction (Invitrogen/Life Technologies), and plated on LB plates containing 5 μg of NaCl/liter and 100 μg of zeocin/ml (Invitrogen/Life Technologies). Colonies were propagated, and plasmids were isolated according to standard protocols. Sequencing was performed using the following primers: FW (5′ CAC CCC CAC CGC CCT CAA AGT AG 3′) and RV (5′ GGA ACG GCA CTG GTC AAC TTG G 3′).

MEF injections in immunocompromised mice.

A total of 1 × 10⁶ MEFs were suspended in 200 μl PBS and injected into immunocompromised BALB/c nude mice. Mice were inspected twice a week.

Protein isolation, immunoblot, immunoprecipitation, and in vitro kinase assays.

For protein isolations, cells were lysed for 30 min on ice in lysis buffer containing 150 mM NaCl, 50 mM HEPES (pH 7.5), 5 mM EDTA, 0.1% NP-40, 5 mM NaF, 0.5 mM vanadate, 20 mM β-glycerolphosphate, and 1 tablet complete protease inhibitor cocktail (Roche) per 50 ml. After centrifugation, protein concentration was determined using Bio-Rad protein assay. For immunoblot analysis, 20 to 30 μg protein was separated on 12% SDS-polyacrylamide gels. Blotting was performed by using standard protocols. For immunoprecipitation reactions, lysates were incubated with antibody and 50% beads in lysis buffer (either protein A [Pharmacia] or protein A/G [Santa Cruz Biotechnology], depending on the antibody used) and incubated O/N at 4°C while rotating. Prior to immunoblot analysis, lysates were washed five times in lysis buffer and loaded on 12% SDS-polyacrylamide gels. Prior to kinase assays, lysates were washed three times in lysis buffer and then washed two times in kinase buffer containing 50 mM HEPES (pH 7.5), 5 mM MgCl₂, 2.5 mM MnCl₂, and 1 mM dithiothreitol. Kinase assays were performed by incubating 50 μg protein (CDK2 assay) or 20 μg protein (cyclin B1 kinase assay) with 5 μg histone H1 (Roche) and 2.5 μCi [γ-³²P]ATP in kinase buffer for 30 min at 37°C while shaking. After the reaction was stopped by adding protein loading buffer, one-third of this mixture was separated on 10% SDS-polyacrylamide gels, blotted by using standard protocols, and exposed to Kodak BioMax XAR films. The antibodies used for immunoblot analysis were mouse monoclonal anti-p21^CIP (F-5, SC-6246; Santa Cruz), mouse monoclonal anti-Ras (catalog no. 610002; BD Transduction Laboratories), sheep polyclonal anti-p53 (Ab-7, PC35; Calbiochem), rabbit anti-TBX2 (kindly provided by M. van Lohuizen), rabbit polyclonal anti-p19^ARF (ab80; Abcam), and goat polyclonal anti-CDK4 (C-22, SC-260-G; Santa Cruz Biotechnology). Secondary antibodies were horseradish peroxidase-conjugated goat anti-mouse, goat anti-rabbit, and rabbit anti-goat antibodies (DakoCytomation). The antibodies used for immunoprecipitation reactions were rabbit anti-CDK2 (M2; Santa Cruz), mouse anti-cyclin B1 (GNS1; Santa Cruz) for kinase assays, and rabbit anti-cyclin B1 (H433; Santa Cruz) for immunoblots. Immunoprecipitation reactions for cyclin B1 kinase assay were performed using protein A/G beads (Santa Cruz). All other immunoprecipitation reactions were performed using protein A beads (Pharmacia).

FACS analysis.

For fluorescence-activated cell sorting (FACS) analysis, cells were trypsinized and fixed in 70% ethanol in PBS at 4°C. Subsequently, cells were washed in PBS, suspended in PBS containing 200 μg/ml RNase A and 20 μg/ml propidium iodide, incubated at 37°C for 15 min, and analyzed using Cell Quest and Summit software.

RESULTS

Gain-of-function screen for anchorage-independent growth of Ras^V12-expressing pocket protein-deficient MEFs.

We have previously shown that ablation of pRb and p107 rendered MEFs insensitive to replicative and Ras^V12-induced senescence but was not sufficient for Ras^V12-induced transformation: Ras^V12-expressing DKO MEFs were unable to exhibit anchorage-independent growth in culture and did not form tumors in nude mice (7, 32). To identify events enabling Ras^V12-induced transformation of pocket protein-deficient MEFs, we performed a gain-of-function screen using Rb^−/− p130^−/− (DKO) MEFs expressing Ras^V12. In short, DKO MEFs were retrovirally transduced with pBABE-Ras^V12 and subsequently with the pEYK-MCF7 cDNA library or with pEYK-GFP. Transduced cells were then seeded in soft agar. Although some colonies appeared in the control vector-infected cells, infection with the pEYK-MCF7 library increased the number of colonies three times. Colonies were picked, and the presence of a pEYK-cDNA integration was determined by both PCR analysis and by a shuttling method that enabled excision of the integrated pEYK-cDNA from the genomic DNA and its propagation as a plasmid in Escherichia coli (24). Sequencing of recovered pEYK vectors revealed many fragmented cDNA sequences in the sense or antisense orientation. Three independent colonies contained a full cDNA encoding the TBX2 oncogene. Only reintroduction of the TBX2-expressing vector into pocket protein-deficient MEFs expressing Ras^V12 caused robust colony formation in soft agar (Fig. 1A, compare left top and left bottom dishes).

FIG. 1.

TBX2 specifically transforms Ras^V12/pocket protein-deficient MEFs. (A) Primary MEFs of the indicated genotypes were first transduced with pEYK-TBX2 and subsequently with pBABE-Ras^V12 (TBX2 + Ras^V12) and plated in soft agar. The left bottom dish shows Rb^−/− p107^−/− p130^−/− MEFs infected with pEYK-GFP and pBABE-Ras^V12 (GFP + Ras^V12). Pictures of soft agar plates were taken using a non-phase-contrast lens (×2.5 magnification). (B) Detailed pictures of Rb^−/− p107^−/− p130^−/− (TKO) MEFs transduced with either pEYK-TBX2 or pEYK-GFP and subsequently with pBABE-Ras^V12. Pictures show the same cells as in panel A, taken using a phase-contrast lens (×5 magnification). (C) Ras^V12 and TBX2 protein levels in the cell lines depicted in panels A and B. (Left) MEFs infected with pEYK-TBX2 and pBABE-Ras^V12. (Right) MEFs infected with pEYK-GFP and pBABE-Ras^V12.

TBX2 transforms pocket protein-deficient MEFs expressing Ras^V12.

To determine whether transformation by TBX2 required the loss of specific pocket proteins, we retrovirally transduced MEFs deficient for all different combinations of pocket proteins with either pEYK-TBX2 or pEYK-GFP, followed by pBABE-Ras^V12, and tested their ability to form colonies in soft agar. In all the cell lines tested, expression of Ras^V12 alone hardly induced anchorage-independent growth (see Fig. 1A and B for TKO MEFs; also data not shown). Concomitant expression of TBX2 induced colony formation in TKO, DKO, and Rb^−/− MEFs (Fig. 1A and B). The strongest colony formation was observed with TKO MEFs, followed by Rb^−/− p107^−/− MEFs and then by Rb^−/− p130^−/−, p107^−/− p130^−/−, and Rb^−/− MEFs. In contrast, expression of TBX2 and Ras^V12 did not support anchorage-independent growth of p107^−/−, p130^−/−, and wild-type MEFs, the latter being consistent with previous results (21). Note that Ras^V12 and TBX2 were expressed to comparable levels in all cell lines (Fig. 1C). Expression of TBX2 alone in pocket protein-deficient MEFs did not induce colony formation (data not shown). Together, these results indicate that the combination of pocket protein ablation, Ras^V12 expression, and TBX2 expression is required for transformation of primary MEFs.

Transformation of TBX2/Ras^V12-expressing MEFs requires ablation of pocket proteins.

To provide independent proof for collaboration between TBX2 and the loss of pocket proteins, we made use of Rb^F/F p130^−/− MEFs. In these cells, exon 19 of both Rb alleles is flanked by LoxP sites and can therefore be deleted by Cre-mediated recombination, which inactivates the gene. Rb^F/F p130^−/− MEFs were transduced with retroviral vectors expressing TBX2 and Ras^V12 and subsequently with either pMSCV-Cre-ERT2 or empty pMSCV. Even without the addition of tamoxifen, transduction of Cre-ERT2 in Rb^F/F p130^−/− MEFs already resulted in nuclear Cre activity, yielding a mixed population of cells containing active (Rb^F) and/or inactive (Rb^Δ19) Rb alleles (Fig. 2A). Expression of Cre-ERT2 in TBX2/Ras^V12/Rb^F/F p130^−/− MEFs increased colony formation in soft agar, indicating that inactivation of the Rb allele promoted anchorage-independent growth (Fig. 2B). Conversely, expression of TBX2, Ras^V12, and Cre-ERT2 in Rb^+/+ p130^−/− MEFs did not enhance colony formation (data not shown). Note that some colonies appeared in the control vector-infected cells (Fig. 2B, right part), which was probably due to the prolonged culturing required for transduction with three retroviral vectors, creating a window for the selection for additional growth-stimulating mutations. Soft agar colonies derived from Rb^F/F p130^−/− MEFs transduced with TBX2, Ras^V12, and Cre-ERT2 were picked and analyzed for Rb status by PCR. Strikingly, while the starting population of cells consisted of a mixed population carrying active and inactive Rb alleles (Fig. 2A), all 14 soft agar colonies exclusively carried inactive Rb alleles (Fig. 2C; note that 5 colonies failed to produce PCR fragments, probably due to the loss of cells during picking). Together with the results presented in Fig. 1, these experimental results show that transformation of MEFs by Ras^V12 requires both ablation of pocket proteins and expression of TBX2.

FIG. 2.

Transformation by TBX2 and Ras^V12 requires ablation of pocket proteins. (A) Expression of Cre-ERT2 yields a mixed population of cells containing active (Rb^F) and/or inactive (Rb^Δ19) Rb alleles. Rb^F/F p130^−/− MEFs were transduced with TBX2, Ras^V12, and pMSCV-Cre-ERT2 viruses. Genomic DNA (gDNA) was isolated and analyzed for the presence of active and inactive Rb by PCR. Lane 1, 100 ng of input gDNA; lanes 2 to 10, genomic DNA diluted 2-, 4-, 8-, 16-, 32-, 64-, 128-, 256-, and 512-fold, respectively, showing that both alleles could be detected at a low input concentration of gDNA; lane 11, 1-kilobase ladder. (B) Expression of Cre-ERT2 enhances colony formation in Rb^F/F p130^−/− MEFs. Rb^F/F p130^−/− MEFs infected with TBX2 and Ras^V12 and either pMSCV-Cre-ERT2 (left) or pMSCV-empty (right) viruses were plated in soft agar. Pictures were taken using a non-phase-contrast lens (×2.5 magnification). (C) Nineteen soft agar colonies derived from Rb^F/F p130^−/− MEFs expressing TBX2, Ras^V12, and Cre-ERT2 were picked and immediately analyzed for the presence of active and inactive Rb by PCR. Each lane represents one soft agar colony; the leftmost lane contains a 1-kilobase ladder. Five colonies (−) failed to produce PCR fragments. The positions of active and inactive Rb alleles are indicated by Rb^F and Rb^Δ19.

Requirements for growth of MEFs in nude mice.

To study transformation in an in vivo system, Rb^−/− p107^−/−, Rb^−/− p130^−/−, and Rb^−/− p107^−/− p130^−/− MEFs were retrovirally transduced with either Ras^V12 and TBX2 or Ras^V12 and green fluorescent protein (GFP) and injected into immunocompromised mice (nude mice) (Fig. 3). TBX2 and Ras^V12 expression levels were comparable in the different cell lines (data not shown). Rb^−/− p107^−/− p130^−/− MEFs expressing Ras^V12 and TBX2 formed tumors within a shorter time window than Rb^−/− p107^−/− or Rb^−/− p130^−/− MEFs expressing Ras^V12 and TBX2; mice injected with TKO MEFs had to be sacrificed between 16 and 21 days, while mice injected with DKO MEFs were sacrificed after 26 days. This suggests, consistent with the soft agar experiments (Fig. 1), that MEFs deficient for all three pocket proteins were more easily transformed than MEFs partially deficient for pocket proteins. DKO and TKO MEFs expressing Ras^V12 and TBX2 formed robust tumors, weighing between 0.1 and 0.7 g upon dissection. In contrast, MEFs expressing Ras^V12 and GFP formed very small clumps of cells, in 9 out of 11 cases weighing less than 0.01 g. In the other two cases, small tumors were formed; these tumors were 10 and 4 times smaller than the corresponding tumors expressing Ras^V12/TBX2. These results demonstrate that also in vivo, expression of TBX2 strongly supports transformation of Ras^V12-expressing DKO and TKO MEFs.

FIG. 3.

Expression of TBX2 promotes tumor formation of pocket protein-deficient MEFs expressing Ras^V12. Immunocompromised mice (nude mice) were injected with MEFs of the indicated genotypes, which had been transduced with either pBABE-Ras^V12 and pEYK-GFP (Ras^V12 + GFP) (injected into the left flanks) or pBABE-Ras^V12 and pEYK-TBX2 (Ras^V12 + TBX2) (injected into the right flanks). Unless indicated, 10⁶ MEFs were injected per flank. Three mice were injected per cell line; two additional mice were injected with 10⁵ Rb^−/− p107^−/− p130^−/− (TKO) MEFs. (A) Graphical representation of tumor mass. Error bars denote standard deviations. (B) Pictures of immunocompromised mice injected with MEFs of the indicated genotypes. MEFS were transduced with either pEYK-TBX2 (+TBX2) or with pEYK-GFP (−TBX2).

Downregulating the p53 pathway transforms pocket protein-deficient MEFs.

Overexpression of TBX2 induces bypass of both replicative and Ras^V12-induced senescence. This has been attributed to downregulation of p19^ARF, and as a consequence, to downregulation of p53 (9, 21). Furthermore, TBX2 has been reported to directly repress the promoter of the p53 target gene p21^CIP1 (33, 44). Given the importance of p53 in transformation, we investigated whether TBX2-induced transformation could be attributed to downregulation of the p53/p21^CIP1 pathway. To this aim, we expressed a pBABE-p21^CIP1 vector in TKO MEFs expressing TBX2 and Ras^V12. Ectopic expression of p21^CIP1 dramatically reduced the number of soft agar colonies induced by TBX2 (see Fig. S1 in the supplemental material), indicating that TBX2-induced transformation involves downregulation of p21^CIP1.

To further address the involvement of downregulating the p53/p21^CIP1 pathway in transformation, we used pRetroSuper (pRS) vectors to suppress expression of p53 or p21^CIP1 in DKO MEFs by RNA interference. Similar to overexpression of TBX2, downregulation of either p53 or p21^CIP1 supported anchorage-independent growth of Ras^V12-expressing Rb^−/− p107^−/− MEFs (Fig. 4A). Knockdown of p53 was more potent in inducing anchorage-independent growth than knockdown of p21^CIP1 was. Residual levels of p21^CIP1 were comparable or even somewhat higher in p53 knockdown cells (see Fig. 6D), suggesting that p53 protects against anchorage-independent growth only partially via p21^CIP1. As observed for transformation by TBX2, transformation by p21^CIP1 knockdown required pocket protein loss; p21^CIP1 downregulation and Ras^V12 expression in wild-type MEFs did not induce colony formation in soft agar (Fig. 4B). Although p53 downregulation and Ras^V12 expression in wild-type MEFs induced a small number of colonies (Fig. 4B), this sharply contrasted to the robust colony formation observed for Ras^V12/Rb^−/− p107^−/− MEFs (Fig. 4C). These results show that in a Ras^V12 transformation assay, downregulation of the p53 pathway mimics overexpression of TBX2. Similar as observed for TBX2, transformation by downregulation of the p53 pathway requires the loss of pocket proteins.

FIG. 4.

Similar to TBX2 overexpression, downregulation of either p21^CIP1 or p53 specifically transforms pocket protein-deficient MEFs. (A) Rb^−/− p107^−/− MEFs were transduced with the indicated constructs and plated in soft agar. Pictures were taken using a non-phase-contrast lens (×2.5 magnification). (B) Wild-type MEFs were transduced with the indicated constructs and plated in soft agar. Pictures were taken using a non-phase-contrast lens (×2.5 magnification). (C) Detailed pictures of the indicated cell lines, plated in soft agar. Pictures show the same cells as in panels A and B, taken using a phase-contrast lens (×5 magnification).

FIG. 6.

Anchorage-independent growth of Ras^V12/pocket protein-deficient MEFs requires rescue of G₂ arrest. Rb^−/− p107^−/− MEFs (A) or Rb^−/− p107^−/− p130^−/− (TKO) MEFs (B and C) were infected with the indicated constructs and cultured either under adherent conditions or for 6 days in methylcellulose. Cells were harvested, stained with propidium iodide, and analyzed by FACS. The percentages of cells in G₁ and G₂ are shown in the top right corners of graphs. (D) Downregulation of p21^CIP1 and p53 in Ras^V12/Rb^−/− p107^−/− and Ras^V12/TKO MEFs upon infection with the indicated constructs. Cells were plated in suspension culture for 24 h, lysed, and immunoblotted for the indicated proteins.

A balance between the p53 pathway and the pocket protein pathway determines the level of transformation of MEFs.

Since colony formation in TBX2/Ras^V12-expressing MEFs was more robust upon the loss of two or three pocket proteins compared to the loss of pRB alone (Fig. 1) and transformation by TBX2 could be attributed to downregulation of the p53/p21^CIP1 pathway (Fig. 4), we hypothesized that a balance between the activities of the pocket protein and the p53/p21^CIP1 pathways determines the rate of transformation of MEFs. We therefore used pRS vectors downregulating p21^CIP1 to different levels, referred to as pRS-p21weak and pRS-p21strong. Figure 5A shows that infection with pRS-p21weak in Rb^−/− p107^−/− MEFs caused a mild decrease of p21^CIP1 levels compared to infection with control vectors (pRS or pRS-GFP), while infection with pRS-p21strong caused a strong decrease. We downregulated p21^CIP1 to different degrees by combined expression of the pRS-p21 vectors and the pEYK-TBX2 expression vector in MEFs deficient for different combinations of pocket proteins. Briefly, TKO MEFs, Rb^−/− p107^−/− MEFs, Rb^−/− MEFs, and wild-type MEFs were first infected with pEYK-GFP or pEYK-TBX2, subsequently with pRS-GFP or a pRS-p21 vector, and finally with pBABE-Ras^V12. The colony-forming capacity in soft agar was monitored, and the results are summarized in Table 1. Expression of Ras^V12 and TBX2, without additional p21^CIP1 downregulation, induced colony formation most robustly in TKO MEFs, followed by Rb^−/− p107^−/− and Rb^−/− MEFs, but expression of Ras^V12 and TBX2 did not induce colony formation in wild-type MEFs (Table 1 and Fig. 1). Strikingly, in Rb^−/− MEFs, coexpression of the pRS-p21 vectors enhanced TBX2-induced colony formation (Fig. 5B). These results show that relatively mild p21^CIP1 downregulation was sufficient to induce transformation of MEFs with complete pocket protein loss (pEYK-TBX2 expression in Ras^V12/TKO MEFs), while strong p21^CIP1 downregulation was required to induce transformation of MEFs partially deficient for pocket proteins (pEYK-TBX2 plus pRS-p21strong in Ras^V12/Rb^−/− MEFs). Thus, a balance between pocket protein and p21^CIP1 levels determines the rate of transformation of Ras^V12-expressing MEFs.

FIG. 5.

A balance between the activities of the pocket protein and p21^CIP1 pathways determines the level of transformation of MEFs. (A) Downregulation of p21^CIP1 using various constructs. Rb^−/− p107^−/− MEFs were infected with the indicated constructs and plated in suspension culture 6 days postinfection. After 24 h, cells were harvested, lysed, and immunoblotted for the indicated proteins. (B) Rb^−/− MEFs were first infected with pEYK-GFP or pEYK-TBX2, next with pRS-GFP or a pRS-p21 vector, and last with pBABE-Ras^V12. Forty-eight hours postinfection, cells were plated in soft agar. Pictures were taken using a non-phase-contrast lens (×2.5 magnification). + and − symbols refer to the quantification of colony formation as depicted in Table 1.

TABLE 1.

Summary of growth of the cell lines in soft agar^a

Genotype of cell line	Growthb in soft agar of cell line infected with the following construct:
	Control vector	pRS-p21weak	pEYK-TBX2	pEYK-TBX2 + pRS-p21weak	pEYK-TBX2 + pRS-p21strong
Wild-type	−	−	−	−	+/−
Rb−/−	−	−	+	++	+++
Rb−/−p107−/−	−	−	++++	++++	++++
TKO	+/−	+/−	+++++	+++++	+++++

^aA balance between the activities of the pocket protein and p21^CIP1 pathways determines the level of transformation of MEFs. Cells were infected as described in the legend to Fig. 5, plated in soft agar, and monitored for anchorage-independent growth.

^bSymbols: −, no growth; +/−, less than 20 colonies; +, ++, +++, ++++, and +++++, indicate the relative capacity for growth in soft agar, where each additional + indicates an approximately twofold increase. +, ++, and +++ correspond to the levels of soft agar growth shown in Fig. 5B.

Ras^V12-induced transformation of pocket protein-deficient MEFs requires rescue of G₂ arrest.

Whereas under adherent conditions, the loss of pocket proteins turned Ras^V12 expression from a growth-inhibiting signal to a growth-promoting signal, Ras^V12/TKO MEFs were largely unable to form colonies in soft agar. To determine why proliferation ceased upon removal of anchorage, we analyzed the cell cycle profile of anchorage-deprived DKO and TKO MEFs by FACS. For this experiment, we cultured cells in methylcellulose, which like soft agar, prohibits attachment but enables easy recovery of nonproliferating cells. Upon removal of anchorage, wild-type fibroblasts arrest in G₁ (10, 14, 15). Here, we show that the majority of Ras^V12/Rb^−/− p107^−/− MEFs arrested in both G₁ and G₂ upon removal of anchorage (Fig. 6A). Knockdown of p21^CIP1 reduced the fraction of G₂ cells (Fig. 6A). As the number of S-phase cells remained low, this indicates that a large fraction of the cells had arrested in G₁. In contrast, knockdown of p53 increased the number of S-phase cells, indicating rescue of both G₁ and G₂ arrest in nonadherent cells (Fig. 6A). Ras^V12/TKO MEFs predominantly arrested in G₂ upon removal of anchorage, which could be alleviated by knockdown of p21^CIP1 or p53 (Fig. 6B) or by overexpression of TBX2 (Fig. 6C). Note that G₂ arrest under nonadherent conditions in Ras^V12/TKO MEFs was not induced by expression of Ras^V12, as TKO MEFs also arrested in G₂ upon removal of anchorage (see Fig. S2 in the supplemental material). Similar to knockdown of p53 or p21^CIP1, knockdown of p19^ARF also rescued G₂ arrest in nonadherent Ras^V12/TKO MEFs (Fig. 7C; note that while the vector used targets the INK4A/ARF locus, the effect was due to knockdown of p19^ARF, as downregulation of p16^INK4A alone did not rescue G₂ arrest of Ras^V12/TKO MEFs). As TBX2 expression downregulated p19^ARF (Fig. 7A), these results suggest that transformation by TBX2 involved downregulation of p19^ARF, and subsequently p53 and p21^CIP1. In conclusion, our results demonstrate that the loss of anchorage induces checkpoint activation in G₁ and G₂, the latter becoming more prominent upon the loss of pocket proteins. Transformation requires rescue of cell cycle arrest, which can be achieved by expression of TBX2 or downregulation of the p19^ARF/p53/p21^CIP1 pathway.

FIG. 7.

Transformation by TBX2 involves downregulation of p19^ARF. (A) TBX2 downregulates p19^ARF in RAS^V12/TKO MEFs. (B) Downregulation of p19^ARF by a knockdown vector targeting the INK4A/ARF locus, but not by a vector targeting p16^INK4A. (C) p19^ARF downregulation rescues G₂ arrest in nonadherent Ras^V12/TKO MEFs. Cells were infected with the indicated constructs and cultured under adherent conditions or for 5 days in methylcellulose. Cells were harvested, stained with propidium iodide, and analyzed by FACS. The percentages of cells in G₁ and G₂ are shown in the top right corners of graphs.

Downregulating the p53/p21^CIP1 pathway by TBX2 expression or RNA interference rescues cyclin-dependent kinase activities under nonadherent conditions.

As the G₁/S and G₂/M transitions require cyclin E-CDK2 and cyclin B1-CDK1, respectively, we analyzed the activity of these complexes under nonadherent conditions. Figure 8A shows that expression of TBX2 induced cyclin B1- and CDK2-associated kinase activities in Ras^V12-expressing TKO cells under nonadherent conditions (Fig. 8, panels 1 and 2). Increased kinase activity in TBX2-expressing cells was related to reduced association of p21^CIP1 with cyclin B1 and CDK2 (Fig. 8, panels 3 and 4). This is most likely due to reduction of p53 levels by TBX2 (Fig. 8, panels 5, 6, and 7). These results indicate that expression of TBX2 relieves inhibition of cyclin-dependent kinase activity by p21^CIP1, leading to rescue of cell cycle arrest and anchorage-independent growth.

FIG. 8.

A combination of pocket protein loss and downregulation of the p53/p21^CIP1 pathway rescues cyclin-dependent kinase activities under nonadherent conditions. (A) Rb^−/− p107^−/− p130^−/− (TKO) MEFs infected with pBABE-Ras^V12 (Ras^V12) and either pEYK-TBX2 or pEYK-GFP were cultured in methylcellulose for 6 days. (Panels 1 and 2) Cyclin B1- and CDK2-associated kinase activities as determined by in vitro phosphorylation of histone H1. IP, immunoprecipitate; P-Histone H1, phosphorylated histone H1. (Panels 3 and 4) Cyclin B1 and CDK2 immunoprecipitates were immunoblotted for p21^CIP1. (Panels 5, 6, and 7) Total protein levels of TBX2, p53, and CDK4 as loading control. (B) Rb^−/− and Rb^−/− p107^−/− MEFs infected with either pRS-p21 or pRS-GFP and subsequently with pBABE-Ras^V12 were cultured under adherent conditions or in methylcellulose for 5 days. Cyclin B1- and CDK2-associated kinase activities as well as association of p21^CIP1 with CDK2 are shown in the top, middle, and bottom panels, respectively.

Downregulating the p53/p21^CIP1 pathway transformed DKO and TKO MEFs more effectively than Rb^−/− MEFs (Fig. 1 and Table 1). We therefore compared the regulation of cyclin-dependent kinase activities in Rb^−/− and Rb^−/− p107^−/− MEFs under both adherent and nonadherent conditions. Both cell lines were infected with either pRS-p21 or pRS-GFP and then with pBABE-Ras^V12 and plated either in methylcellulose or under adherent conditions. Cyclin B1- and CDK2-associated kinase activities in Rb^−/− p107^−/− MEFs were higher than in Rb^−/− MEFs, which was particularly visible under adherent conditions (Fig. 8B, compare lanes 1 and 3 and lanes 5 and 7). The loss of anchorage caused inhibition of kinase activities in Rb^−/− and Rb^−/− p107^−/− MEFs (Fig. 8B, compare lanes 1 and 5 and lanes 3 and 7). Knockdown of p21^CIP1 induced kinase activities under both adherent and nonadherent conditions, and as expected, no association of p21^CIP1 with CDK2 could be detected (Fig. 8B, lanes 2, 4, 6, and 8). However, in nonadherent Ras^V12/Rb^−/− MEFs (Fig. 8B, lane 6), the increase in CDK2 and cyclin B1-associated kinase activity was minor and apparently not sufficient to support anchorage-independent growth. In contrast, knockdown of p21^CIP1 in Ras^V12/Rb^−/− p107^−/− MEFs (Fig. 8B, lane 8) raised kinase activities to much higher levels than observed in Ras^V12/Rb^−/− MEFs (lane 6). Note that knockdown of p21^CIP1 induced cyclin B1-associated kinase activity more robustly than CDK2-associated kinase activity. This fits with the observation that knockdown of p21^CIP1 was more potent in rescuing G₂ arrest than in rescuing G₁ arrest (Fig. 6). We conclude that downregulation of p21^CIP1 sufficiently rescued the decrease in kinase activities induced by the loss of adhesion only in Ras^V12/Rb^−/− p107^−/− MEFs. This allowed bypass of G₂ arrest and, to a lower extent, G₁ arrest, enabling anchorage-independent growth.

DISCUSSION

Pocket protein depletion is not sufficient for Ras^V12-induced transformation of MEFs.

Functional inactivation of the pocket protein and p53 tumor suppressor pathways is one of the most frequent events in the development of cancer. Consistent with their importance in G₁ control, the loss of either p53 or pocket proteins bypasses replicative senescence in primary murine embryonic fibroblasts (7, 8, 19, 38). Furthermore, inactivation of the p53 pathway or the loss of pRb and at least one other pocket protein reversed the growth-inhibitory effect of Ras^V12 into a proliferative stimulus (32, 39). However, unlike abrogation of p53, the loss of pocket proteins was not sufficient to support oncogenic transformation by Ras^V12 (32). This observation contradicts a report by others (38) but is strongly confirmed in the present study; Ras^V12-expressing MEFs deficient for both pRb and p130, both pRb and p107, or all three pocket proteins were largely unable to exhibit anchorage-independent growth or to form tumors in nude mice. By screening a cDNA library, we have identified a missing link: overexpression of the TBX2 oncogene strongly induced transformation of pocket protein-deficient MEFs by Ras^V12, as evidenced by anchorage-independent growth in vitro and in vivo (Fig. 1 and 3). We have consistently shown that transformation of MEFs overexpressing TBX2 and Ras^V12 required the loss of pocket proteins (Fig. 2). These results show that the loss of pocket proteins and overexpression of TBX2 strongly act synergistically in Ras^V12-induced transformation of primary murine embryonic fibroblasts.

The pRB and p53 pathways cooperate in suppression of transformation.

TBX2 overexpression is known to attenuate the p53/p21^CIP1 tumor suppressor pathway by downregulating transcription of the Mdm2 inhibitor p19^ARF. Consistently, we show that partial RNA interference-mediated knockdown of p19^ARF, p53, or p21^CIP1 strongly supported Ras^V12-induced transformation of pocket protein-deficient MEFs (Fig. 4A, 6, and 7). Moreover, expression of p21^CIP1 inhibited TBX2-induced transformation (see Fig. S1 in the supplemental material). The complete loss of p53 was sufficient for Ras^V12-induced transformation of wild-type MEFs (32). Here we show that partial knockdown of p53 induced anchorage-independent growth only to a very limited degree in wild-type MEFs but was very effective in pocket protein-deficient MEFs (Fig. 4B and C). Furthermore, we found that anchorage-independent growth upon Ras^V12 expression could be achieved by either a combination of moderate p21^CIP1 downregulation and severe pocket protein loss or strong p21^CIP1 downregulation and mild pocket protein loss (Fig. 5 and Table 1). Apparently, the p53 and pRB pathways act synergistically in preventing anchorage-independent growth.

The complete loss of pocket proteins did not support anchorage-independent growth upon expression of Ras^V12, although it should be noted that a small number of colonies appeared in the soft agar culture. This may be explained by downregulation of the p53/p21^CIP1 pathway in a subset of the cells. Indeed, Sage and coworkers (38) reported that two out of five TKO MEF cultures displayed downregulation of p53 and p21^CIP1 during culturing under adherent conditions.

Regulation of cell cycle arrest and cyclin-dependent kinase activity by pRB, TBX2, and p53.

Anchorage-independent growth is a hallmark of transformed cells. Untransformed cells are dependent on anchorage via integrin signaling. Upon attachment to the extracellular matrix, integrin signaling induces cyclin D1 and inhibits p21^CIP1 and p27^KIP1 expression, thereby authorizing cell cycle progression (1, 37, 47). Conversely, detachment from the extracellular matrix is associated with downregulation of cyclin D1, induction of p21^CIP1 and p27^KIP1, and increased association of p21^CIP1 and p27^KIP1 with cyclin-CDK complexes (10, 49, 50). As a result, cells deprived of attachment downregulate CDK2 activity and arrest in G₁ (10, 15). We show here that in anchorage-deprived Ras^V12/DKO MEFs, there was only partial G₁ arrest, as a substantial fraction of cells arrested in G₂ (Fig. 6A). In Ras^V12/TKO MEFs, which have completely lost the G₁ restriction point, G₂ arrest predominated (Fig. 6B). Thus, anchorage dependence is regulated via both G₁ and G₂ checkpoints. Our results therefore indicate that the transforming activity of TBX2 not only relies on override of the G₁ checkpoint but also on override of the G₂ checkpoint and involves downregulation of the p53 pathway (Fig. 6).

Are G₂-arrested cells senescent? We were unable to perform a senescence-associated β-galactosidase staining on arrested cells, as both DKO and TKO MEFs were targeted with a lacZ reporter gene. However, we have strong indications that the arrest is reversible, which according to the current definition excludes senescence. To address this issue, we centrifuged Ras^V12/TKO MEFs harvested from methylcellulose at low speed to separate the low number of growing colonies from G₂-arrested cells. By FACS, we found that the population of cells in the supernatant was enriched for G₂-phase cells. These cells entered the cell cycle upon reattachment, as observed by growth of the cells and by FACS profiles (data not shown).

The level of p21^CIP1 knockdown required to induce transformation was dependent on the extent of pocket protein ablation: relatively moderate p21^CIP1 knockdown was sufficient to induce anchorage-independent growth in TKO MEFs, whereas strong p21^CIP1 knockdown was required to transform Rb-deficient MEFs. Our results indicate that these different requirements for anchorage-independent growth are due to different levels of kinase activities in pocket protein-ablated cells. Despite high levels of cyclins A, E, and B1 in Rb^−/− MEFs (6, 12, 20, 26), their associated kinase activities were lower than in Rb^−/− p107^−/− MEFs (Fig. 8), and this may be caused by E2F-dependent induction of p107 (48). As Rodier and coworkers (36) suggested that overexpression of p107 reduces the half-life of Skp2 protein, this could lead to downregulation of Skp2 in Rb^−/− MEFs and subsequently, inhibition of p27^KIP1 degradation. However, we did not observe different p27^KIP1 levels in Rb^−/− and Rb^−/− p107^−/− MEFs (data not shown). An alternative may be that high p107 levels in Rb^−/− cells, together with p130, inhibit kinase activities in Rb^−/− MEFs via direct binding to cyclin E-A/CDK2 (5, 13). This mechanism may explain why the extent of p21^CIP1 knockdown needed to achieve sufficient cyclin-dependent kinase activity to support anchorage-independent growth was lower in DKO and TKO MEFs than in Rb^−/− cells. In conclusion, we propose that the pocket protein and p53 pathways synergistically protect against anchorage-independent growth by their shared ability to regulate cyclin-dependent kinase activities.

Roles of pRB, TBX2, and p53 in tumorigenesis.

We identified TBX2 as a transforming oncogene in pocket protein-deficient cells. Several lines of evidence point to a role for TBX2 during in vivo tumorigenesis. TBX2 was found amplified in a range of human tumor samples and cancer cell lines of different developmental origin, such as melanoma cell lines (44), a subset of human breast tumors (21, 43), and pancreatic cancer cell lines (27). The combined loss of components of the pRB and p53 pathways cooperates in tumorigenesis in vivo. The loss of p16^INK4A accelerated tumorigenesis in p53^−/− mice (40), and similarly, INK4A/ARF knockout mice developed tumors with a shorter latency than p16^INK4A−/− or p19^ARF−/− mice (41). In a p16^INK4A null background, the loss of one p19^ARF allele accelerated tumorigenesis, while strikingly, the wild-type allele was retained in a subset of the tumors (25). This indicates that also in vivo, tumor development by abrogation of the pRB pathway can be accelerated by partial ablation of the p53 pathway. Moreover, various mouse tumor models have been generated by combined deletion of p53 and Rb, including tumors of the lung (29), central nervous system (28), breast (42), and pineal and pituitary glands (45). Finally, simultaneous mutations in both pathways have been found in a variety of mouse and human tumor samples (4, 30). Thus, the synergism between the loss of the pocket protein and p53 pathways, as observed in this study during transformation of MEFs, also applies to tumorigenesis in both mice and humans. While this synergism has been explained by alleviation of apoptosis by p53, our previous (11) and present findings indicate that the loss of p53 may also be required for alleviation of cell cycle arrest of cells with abrogated pocket protein function.

Comparison of the requirements for transformation of human and murine fibroblasts.

By using variants of simian virus 40 large T antigen (LT) that specifically target p53 or the pRB family, Rangarajan and coworkers (35) suggested that in both human and mouse cells, the loss of p53 or the loss of pocket proteins is sufficient for bypassing Ras^V12-induced senescence. However, Ras^V12-induced transformation met different requirements in human and murine fibroblasts. First, transformation of human fibroblasts required expression of hTERT and simian virus 40 small t antigen (ST) (16), which was dispensable for transformation of murine fibroblasts. Since murine cells contain long telomeres due to active telomerase (23, 34), mTERT overexpression is not required for transformation. The requirement for ST in transformation of human fibroblasts is caused by its interaction with protein phosphatase 2A (17), which possibly downregulates PTEN (phosphatase and tensin homolog) (3). Why ST expression is not required for murine transformation is currently unknown. Second, LT-mediated ablation of either the pRB or the p53 pathway appeared sufficient for Ras^V12-induced transformation of mouse embryonic fibroblasts, while ablation of both pathways was required for transformation of human fibroblasts (35). Similarly, Boehm et al. (3) claimed that expression of a dominant-negative version of p53, c-Myc, and Ras^V12 was sufficient for transformation of MEFs, whereas human fibroblasts needed expression of ST and suppression of pRB. Also, Ras^V12- and hTERT-expressing BJ primary human fibroblasts could exhibit anchorage-independent growth only upon concomitant knockdown of p53 and pRB (46). Taken together, a widely held view that suppression of both the p53 and pRB pathways is mandatory to transformation of human cells while either one of these events is sufficient for transformation of murine cells has emerged. This view was supported by Sage et al. (38), who reported anchorage-independent growth of TKO MEFs upon expression of Ras^V12. However, our previous (7, 32) and present results sharply contrast with this view: partial or complete pocket protein-deficient MEFs were largely unable to exhibit anchorage-independent growth upon Ras^V12 expression. Furthermore, we found that partial ablation of the p53 pathway strongly stimulated anchorage-independent growth of pocket protein-deficient MEFs. We therefore conclude that in murine cells, ablation of the p53 and pRB pathways synergistically supports oncogenic transformation and that in this respect, mouse fibroblasts are not fundamentally different from human fibroblasts.