Programmed cell death protein 4 suppresses CDK1/cdc2 via induction of p21Waf1/Cip1

R. Göke, P. Barth, A. Schmidt, B. Samans, B. Lankat-Buttgereit


We show that the recently discovered tumor suppressor pdcd4 represses the transcription of the mitosis-promoting factor cyclin-dependent kinase (CDK)1/cdc2 via upregulation of p21Waf1/Cip1. p21Waf1/Cip1 inhibits CDK4/6 and CDK2. Decrease of CDK4/6 and CDK2 enhances the binding of pRb to E2F/DP, which in turn together bind to and repress the cdc2 promoter. Upregulation of CDK1/cdc2 accompanied by a malignant change was previously reported in colon cancer. We show that expression of pdcd4 as an indirect suppressor of CDK1/cdc2 is lost in progressed carcinomas of lung, breast, colon, and prostate. Furthermore, it seems that localization and expression of pdcd4 directly correlate with tumor progression. Finally, the CDK1/cdc2 inhibitor roscovitine reduces the proliferation of several tumor cell lines, suggesting that inhibition of CDK1/cdc2 may be a useful strategy against malignant transformation. Therefore, pdcd4 might serve as a novel target for antineoplastic therapies.

  • tumor growth
  • cell cycle
  • tumor suppressor gene

investigating the mechanisms of programmed cell death, we cloned rat pdcd4 (=DUG, death upregulated gene) (8). Rat pdcd4 is highly conserved during evolution and almost identical to the murine [MA3 (25), TIS (22)] human [H731 (19), 197/15a (2)], and chicken (24) homologs. pdcd4 is upregulated after induction of apoptosis by different stimuli such as glucose/serum starvation and death receptor ligation (for review of pdcd4 see Ref. 15). Although the role of pdcd4 during apoptosis is still unclear, there is increasing evidence that pdcd4 might function as a tumor suppressor (32). Promotion-sensitive (P+) JB6 cells undergo neoplastic transformation in response to tumor promoters. Interestingly, (P−) JB6 cells, which are promotion resistant, show higher pdcd4 expression levels of mRNA and protein compared with P+ cells (4). Decrease of pdcd4 protein expression by an antisense approach resulted in a transformation-sensitive P+ phenotype. In accordance with these findings, higher pdcd4 mRNA levels have been found in preneoplastic and initiated keratinocyte cell lines than in further-progressed cell lines. Most likely, pdcd4 inhibits the transactivation of the transcription factor complex AP-1, which is essential for neoplastic transformation, by controlling an unknown modulator of c-fos and c-jun activation domains (33). This might, at least partially, explain its tumor-suppressive effect. However, there is evidence that pdcd4 also acts via other signaling pathways. Analysis of the amino acid sequence revealed that pdcd4 contains two conserved MA3 domains. The translation initiation factors eukaryotic initiation factor (eIF)4G I and eIF4G II directly interact with the RNA helicase eIF4A via the MA3 domain (23). eIF4A is part of the cap-dependent translation initiation complex and unwinds the mRNA. Recently, we showed (8) that, similarly to eIF4G, pdcd4 interacts directly with eIF4A. In addition, this was confirmed by Yang et al. (31), who showed that this interaction inhibits protein synthesis. Futhermore, in in vitro binding assays, pdcd4 prevented binding of eIF4A to the carboxy-terminal domain of eIF4G. The mechanisms of action of pdcd4 might even be more complex, because it was reported that it also interacts with eIF4G (12). The inhibitory effect of pdcd4 on translation might contribute to its tumor-suppressive effect, because translation initiation factors may function as oncogenes (7). However, although some progress has been made toward understanding pdcd4 function, the mechanisms of the tumor-suppressive effect of pdcd4 are still poorly understood. In the present study, we show that pdcd4 regulates cyclin-dependent kinase (CDK)1/cdc2 activity via the CDK inhibitor p21Waf1/Cip1, which might at least partially explain its antiproliferative effect on transformed cells.


Cell culture.

Bon-1 human carcinoid cells were cultured in DMEM-Ham's F-12 medium containing 10% fetal calf serum and 40 μg/ml gentamicin. INS-1 human insulinoma cells were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum, 1 mM sodium pyruvate, 10 mM HEPES, 0.05 mM 2-β-mercaptoethanol, and 40 μg/ml gentamicin. INR1-G9 cells were grown in RPMI 1640 medium containing 10% fetal calf serum and 40 μg/ml gentamicin. To establish cell lines stably expressing pdcd4, Bon-1 cells were transfected with pcDNA3.1-V5-His plasmid (Invitrogen) containing the human pdcd4 cDNA by the calcium coprecipitation method. For control, cells were transfected with empty vector. Selection of positive transfected cells was performed with G418. Ectopic pdcd4 expression was confirmed by performing Western blot analysis with a horseradish peroxidase (HRP)-coupled anti-V5 antibody.

Western blot.

Total proteins were extracted from cells, and protein concentration was determined with a protein assay from Bio-Rad according to the manufacturer's protocol. Proteins (30 μg) were separated by SDS-PAGE (10% gel) and electrotransferred onto nitrocellulose membrane (pore size 0.45 μm). Immunodetection was carried out with different antibodies in 50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20, followed by HRP-conjugated secondary antibodies. Detection was performed with an ECL kit (Amersham Bioscience). Except for anti-V5-antibodies (HRP coupled; Invitrogen) all antibodies were from Santa Cruz (Actin-HRP, cdc2p34, RbIF8, p21, Wee 1) and Cell Signaling (secondary antibodies, cdc2 Tyr15, cdc2 Thr161, pRb Ser807/811, pRb Ser795, p53, E2F1, cdk4, p-cdc25C Ser216, p27/Kip1, cyclin D1).


Total RNA from mock- and pdcd4-transfected Bon-1 cells was isolated with the RNeasy Mini Kit (Qiagen), and semiquantitative RT-PCR was performed with the One Step RT-PCR kit (Qiagen) according to the supplier's protocol. As an internal control β-actin/3 and /6 primers (Promega) were used, yielding a 285-bp product. Primers for amplification of CDK1/cdc2 were synthesized by Metabion, yielding a 203-bp product (5’-cdc2: CAG TCT TCA GGA TGT GCT TAT GC; 3’-cdc2: GAG GTT TTA AGT CTC TGT GAA GAA CTC). After 15, 20, 25, and 30 cycles, aliquots from each sample were removed and analyzed on a 2% agarose gel.

Gene expression analysis.

The 4.6K cDNA chips contained the GF200 set of Research Genetics cDNAs. Each cDNA was spotted in duplicate [ (Research, Microarray Unit)]. Chips were generated by the group of M. Krause (Institute of Molecular Biology and Tumor Research, Marburg, Germany). One microgram of total RNA was amplified with the MessageAmp aRNA kit (Ambion), yielding ∼20 μg of aRNAs. Cy3- and Cy5-labeled cDNAs were synthesized with the CyScribe cDNA Post-Labelling kit (Amersham Biosciences). A flip-color experiment was included. The appropriate Cy3- and Cy5-labeled samples were pooled and hybridized to microarrays under a glass coverslip for 16 h at 55°C and washed with 0.1× SSC. Chips were scanned with a GMS 418 fluorescent scanner (MWG-Biotech), and the images were analyzed with IMAGENE 3.0 software. The raw data were evaluated as described previously (18).

Methylthiazoletetrazolium assay.

To test the effect of the CDK inhibitor roscovitine on tumor cells, cells were incubated for 24 h in 500 μl of culture medium in the absence or presence of different concentrations of roscovitine. Subsequently, 100 μl of methylthiazoletetrazolium (MTT; 5 mg/ml phosphate-buffered saline) was added and incubated for an additional 2 h. After removal of the medium, cells were treated with 200 μl of DMSO for 1 h. Samples were centrifuged, and the absorbance at 550 nm was determined in the supernatants with a multiwell plate reader. Percent inhibition of cell proliferation was calculated with the following formula: 100 − (extinction of treated sample × 100/extinction of control sample). To test whether cell viability is affected by overexpression of pdcd4, mock-transfected or pdcd4-overexpressing cells were detached from culture plates, stained with propidium iodide, and analyzed by fluorescence-activated cell sorting (FACS). Analysis revealed that there was no difference between cell lines in cell viability (data not shown).

Anti-pdcd4 antibodies.

Polyclonal anti-pdcd4 antibodies were generated by Peptide Specialty Laboratories (Heidelberg, Germany). For this, rabbits were immunized with synthetic peptide corresponding to the amino-terminal 20 amino acids of the human pdcd4 protein coupled to keyhole limpet hemocyanin. Specific antibodies were affinity purified from the resulting serum with nitrocellulose-bound antigen as described by Hammerl et al. (10).


The present study included carcinomas of colon (n = 11), lung (n = 3), breast (n = 10), pancreas (n = 3), and prostate (n = 7) and adjacent tumor-free tissue, which were surgically resected when indicated. Specimens were fixed for diagnostic purposes in formalin (10%) and embedded in paraffin, and subsequently sections of ∼5 μm were prepared. Sections were mounted on glass slides and stained with hematoxylin and eosin for routine purposes by standard protocols. Consecutive sections of the specimens were used for immunohistochemical examination with antibodies.


Expression of pcdc4 was detected with a polyclonal rabbit antibody (dilution 1:10) by means of the standard avidin-biotin complex (ABC)-peroxidase method (ABC Elite Kit; Vector, Burlingame, CA) with 3,3’-diaminobenzidine (DAB) as chromogen after microwave pretreatment performed by heating the deparaffinized and rehydrated sections, immersed in 10 mM sodium citrate buffer (pH 6.0), in a microwave oven at 600 W three times for 5 min. Sections were counterstained with Mayer's hemalum. After dehydration in graded alcohols, sections were cleared in xylene and coverslipped with Entellan.


pdcd4 represses expression of maturation-promoting factor CDK1/cdc2.

In previous experiments we showed (16) that overexpression of pdcd4 in Bon-1 carcinoid cells suppresses cell proliferation, whereas cell viability was not affected. To elucidate the mechanisms of this effect, we performed gene array analysis with RNA isolated from mock-transfected and pdcd4-overexpressing Bon-1 cells. We found that CDK1/cdc2 expression was strikingly reduced in pdcd4-overexpres-sing cells. This finding was confirmed by semiquantitative RT-PCR (Fig. 1A) and was also detectable on the protein level by Western blot analysis (Fig. 1B). Cell division consists of two steps: interphase, including G1, S, and G2 phases, and mitosis (M). In late G2 and early M phase CDK1/cdc2 associates with cyclin A to promote entry into mitosis. Mitosis is further regulated by CDK1/cdc2 in complex with cyclin B (1, 14). Activation of CDK1/cdc2 requires dephosphorylation of Tyr15 and Thr14 by protein phosphatase cdc25C and phosphorylation of Thr161 by CDK-activating kinase (CAK) (17, 26, 28). Analyzing the phosphorylation status, we found decreased levels of Tyr15-CDK1/cdc2 in pdcd4-overexpressing Bon-1 cells, as expected (Fig. 1B). However, this does not indicate increased activation of CDK1/cdc2 but is most likely a consequence of reduced CDK1/cdc2 expression by pdcd4, because the effect is not compensated by elevated levels of Thr161-CDK1/cdc2 (Fig. 1B). Thus pdcd4 may repress CDK1/cdc2 activity on the transcriptional level and not by protein modification through phosphorylation.

Fig. 1.

A: semiquantitative RT-PCR with primers specific for β-actin (285 bp) and cyclin-dependent kinase (CDK)1/cdc2 (203 bp). After 15, 20, 25, and 30 cycles, aliquots were removed and analyzed on a 2% agarose gel. a, RNA isolated from mock-transfected Bon-1 cells; b, RNA from pdcd4-transfected Bon-1 cells. m, DNA marker. B: Western blot analysis of protein extracts from mock- and pdcd4-transfected Bon-1 cells. Equal amounts of protein were loaded and immunoblotted with antibodies recognizing CDK1/cdc2 regardless of phosphorylation (cdc2p34), phospho-Tyr15-cdc2 antibodies (cdc2 Tyr15), or phospho-Thr161-cdc2 antibodies (cdc2 Thr161). Each experiment was done with at least 3 independently isolated protein extracts. Each immunoblot was treated with antibodies to β-actin in a last step to verify equal protein amounts, and a representative experiment is shown. Exposure times were from 30 s to 3 h (for some of the phosphorylation-specific antibodies).

pdcd4 reduces phosphorylation of retinoblastoma gene product pRb.

pRb inhibits the cell cycle by trapping E2F transcription factors in quiescent cells. As the cell cycle progresses, pRb is phosphorylated by CDKs, releasing E2F that activates genes required for cell proliferation (20). In pdcd4-overexpressing cells, expression of pRb and E2F was unaffected (Fig. 2A). However, phosphorylation of pRb at Ser795 and Ser807/811 was drastically reduced (Fig. 2A). Because CDK1/cdc2 is one of E2F's target genes, hypophosphorylation of pRb and thereby inhibition of E2F activity could be responsible for reduced CDK1/cdc2 level in pdcd4-overexpressing cells. Moreover, pRb is substrate for CDK 4/6 and CDK 2. In accordance with the hypophosphorylation of pRb, CDK 4/6 was found to be decreased in pdcd4-overexpressing cells on the protein level by Western blot analysis (Fig. 2B). Furthermore, the expression of the cell cycle inhibitor p21Waf1/Cip1 was dramatically increased (Fig. 2B), whereas expression levels of p16 and p27 were unchanged (data not shown). This induction of p21 was not mediated by the p53 signaling pathway, because we could not detect any signifi-cant change of p53 protein amounts (Fig. 2B). This suggests that pdcd4 regulates CDK activity via the CDK inhibitor p21Waf1/Cip1. This assumption is supported by the finding that the protein expression level of CDK1 regulators Wee1 and p-cdc25C Ser216 were unaltered in pdcd4-overexpressing cells (data not shown). Furthermore, with gene array analysis we could not see any difference of expression of genes for CHK1, CHK2, cyclin D1, cyclin D2, and cyclin D3 in mock-transfected and pdcd4-overexpressing cells. Consistent with these data, cyclin D1 protein expression was also not different between cell lines (Fig. 2B).

Fig. 2.

Western blot analysis of protein extracts from mock- and pdcd4-transfected Bon-1 cells. A: equal amounts of protein were loaded onto SDS-PAGE gels and immunoblotted with antibodies recognizing E2F1, pRb regardless of phosphorylation (RbIF8), phospho-Ser795-pRb antibodies (Rb Ser795), or phospho-Ser807/811 antibodies (Rb Ser807/811). Experiments were performed as in Fig. 1B. B: equal amounts of protein were loaded and immunoblotted with antibodies recognizing cdk4, p21Waf1/Cip1 (p21), p53, and cyclin D1. Experiments were performed as in Fig. 1B.

CDK1/cdc2 inhibitor roscovitine inhibits endocrine tumor cell growth.

Because reduced CDK1/cdc2 levels might contribute to a pdcd4-induced inhibition of cell growth, we tested whether the CDK1/cdc2 inhibitor roscovitine would exert a similar effect on endocrine tumor cells. For this, Bon-1 carcinoid cells, INS-1 insulinoma cells, and INR1-G9 glucagonoma cells were incubated for 24 h in the absence and presence of different concentrations of roscovitine. As shown in Fig. 3, roscovitine concentration-dependently inhibited cell growth of all three cell lines. Therefore, CDK inhibitors might represent promising tools for the therapy of endocrine tumors.

Fig. 3.

Inhibition of endocrine tumor cell growth by the CDK1/cdc2 inhibitor roscovitine. Bon-1 carcinoid cells (A), INS-1 insulinoma cells (B), and INR1-G9 glucagonoma cells (C) were incubated for 24 h without and with different concentrations of roscovitine. Methylthiazoletetrazolium (MTT) assay was then performed as described in materials and methods. Data shown are means ± SE of 2 experiments. Statistical analysis by t-test. *P < 0.05.

pdcd4 exerts an important role in tumorigenesis of human cancer.

Overexpression of pdcd4 results in suppression of tumor cell growth. This effect is at least partially mediated by induction of p21Waf1/Cip1 and repression of CDK1/cdc2. Because pdcd4 obviously exerts a tumor suppressor function, it might play a role in tumorigenesis of human malignomas. Therefore, we investigated the expression of pdcd4 in carcinomas obtained from different anatomic sites compared with adjacent normal tissue.

Cells of normal tissues showed intense nuclear staining for pdcd4 in the majority of epithelia, whereas cytoplasmic staining was not observed. Normal colonic mucosa showed a characteristic local distribution of epithelia with nuclear pdcd4 staining, in that epithelia of the middle and apical portions of the crypts disclosed intensely stained nuclei. In contrast, epithelia in the lower portion of the crypt showed no pdcd4 staining, neither nuclear nor cytoplasmic. In normal prostate, breast, and lung, most epithelia disclosed intense nuclear staining and no pdcd4 in cytoplasm. In addition to epithelial cells, intense pdcd4 staining was also found in the nuclei of endothelia, stromal fibrocytes, and lymphocytes. No cytoplasmic staining was observed in these cells.

We investigated a total of 30 carcinomas obtained from different anatomic sites and immunohistochemically compared pdcd4 staining with that observed in adjacent tumor-free tissue. Depending on the primary tumor site and grade of differentiation, a decrease of nuclear pdcd4 staining accompanied by a gain of cytoplasmic pdcd4 staining was found (Table 1). Six of seven colon carcinomas showed a complete loss of nuclear pdcd4 staining (Fig. 4), and in four cases cytoplasmic pdcd4 staining was found. All colonic adenomas investigated showed a clear shift from nuclear pdcd4 localization to cytoplasmic staining.

Fig. 4.

Immunohistochemically normal colonic epithelium shows intense nuclear staining paralleled by negative cytoplasmic reactivity of the upper cryptal epithelium (right), whereas the base of crypts discloses no immunoreactivity (inset). Adjacent adenocarcinoma (left) shows neither cytoplasmic nor nuclear staining (pdcd4 immunohistochemistry, avidin-biotin complex method). Magnification, ×100.

View this table:
Table 1.

Expression of pdcd4 in tumor sites

Finally, small cell lung cancer exhibited a complete loss of nuclear as well as cytoplasmic pdcd4 staining. Invasive breast cancer showed preserved nuclear pdcd4 reactivity in 3 of 10 cases in addition to cytoplasmic staining. In all except one case, nuclear localization for pdcd4 was associated with cytoplasmic staining, distinguishing these cases from normal breast tissue. In one case of invasive breast carcinoma, nuclear staining combined with no cytoplasmic pdcd4 staining, a pattern similar to that observed in normal breast epithelium, was found. In prostate cancer, weak residual pdcd4 staining was found in three of seven cases, whereas the remaining four cases disclosed a complete loss of nuclear staining. However, nuclear pdcd4 staining was closely related to the Gleason grade, because all cases with preserved nuclear pdcd4 staining disclosed it in Gleason grade 2 areas; higher Gleason grades were associated with a loss of nuclear pdcd4 reactivity.


In the present study investigating the suppression of Bon-1 cell growth, we show that CDK1/cdc2 gene and protein expression is repressed by pdcd4. CDK1/cdc2 has also been termed mitosis-promoting factor because, associated with cyclin B, it represents the CDK required for G2-M transition during the cell cycle (6, 21). The activity of CDK1/cdc2 is regulated by phosphorylation: dephosphorylation at Tyr15 and Thr14 and phosphorylation at Thr161 activates CDK1/cdc2 (17, 26, 28). Tyr15-CDK1/cdc2 is suppressed in pdcd4-overexpressing cells, which could point to an enhanced activation of CDK1/cdc2. However, this may not be due to dephosphorylation but instead the result of a reduction of CDK1/cdc2 expression at the transcriptional level. This is supported by the fact that we also found decreased levels of Thr161-CDK1/cdc2.

The transcription of the CDK1/cdc2 gene is regulated by pRb/E2F/DP. Hypophosphorylated pRb recruits E2F-DP complexes, converting them from sequence-specific transcription activators to sequence-specific repressors (29). Despite the lack of change in total protein amounts, in pdcd4-overexpressing cells pRb was found to be hypophosphorylated, which might explain the repressed transcrip-tion of the CDK1/cdc2 gene. This assumption is sup-ported by the previous finding that Rb negatively regulates the cdc2 promoter (5). pRb is substrate for CDK4/6 and CDK2, and, accordingly, we found a reduction of CDK4/6 protein in pdcd4-overexpressing cells. Furthermore, the expression of p21Waf1/Cip1 is drastically increased. p21Waf1/Cip1 is a CDK inhibitor that inhibits CDK4/6 and CDK2 (11). Obviously, pdcd4 reduces the activity of CDKs via induced p21Waf1/Cip1. It is unlikely that other CDK1/cdc2 regulators are involved, because, analyzing gene arrays and protein expression pattern using Western blotting, we did not find any effect of pdcd4 on expression of p16, p27, Wee1, Cdc25C, CHK1, CHK2, or cyclin D1, D2, and D3. The induction of p21Waf1/Cip1 seems to be independent of p53, because we found no change or even a minor decrease in p53 protein. Hence, pdcd4 might suppress tumor progression even in p53-defective cells. Interestingly, we found previously (9) that pioglitazone, an activator of the nuclear transcription factor peroxisome proliferator-activated receptor (PPAR)-γ, upregulates the expression of p21Waf1/Cip1, inhibits cell growth, and sensitizes NCI-H727 carcinoid cells to tumor necrosis factor-related apoptosis-inducing ligand TRAIL-induced apoptosis. Influencing the p21Waf1/Cip1 signal pathway may be a promising therapeutic strategy. Moreover, in p53-arrested cells p21Waf1/Cip1 associates with CDK1/cdc2/cyclin B, representing an additional inhibitory mechanism (27) that might also play a role in pdcd4-overexpressing cells. We summarize the proposed pdcd4 signal pathway in Fig. 5.

Fig. 5.

Function of pdcd4. For details, see discussion.

Interestingly, pdcd4- and p53-dependent CDK1/cdc2 regulation are somewhat analogous. Similarly to pdcd4, p53 represses CDK1/cdc2 through mechanisms involving induction of p21Waf1/Cip1 that inhibits CDK activity, enhancing the binding of Rb family member p130 to E2F, which together bind to and repress the CDK1/cdc2 promoter (27). Therefore, p53-induced expression of p21Waf1/Cip1 results in both a G1 arrest and a G2 arrest of the cell cycle because of the association of p21Waf1/Cip1 with CDK1/cdc2/cyclin B and the repression of the CDK1/cdc2 promoter (3). The impact of pdcd4 on the cell cycle should be similar, because pdcd4 also induces p21Waf1/Cip1. Future studies must address whether pdcd4 exhibits more similarities to the regulation and biological effects of p53.

Because overexpression of pdcd4 in Bon-1 carcinoid cells repressed CDK1/cdc2, resulting in reduced cell proliferation, we speculated that CDK1 inhibitors should exert a similar effect. To test this hypothesis, different endocrine tumor cell lines were treated with the CDK1 inhibitor roscovitine. As expected, roscovitine concentration-dependently inhibited the growth of all endocrine tumor cell lines tested. This is of special interest because most endocrine tumors are resistant to radiation and common chemotherapeutic agents. In our experiments the IC50 of roscovitine was quite high (∼80 μM). However, CDK inhibitors more potent than roscovitine might represent a new approach for the treatment of endocrine tumors.

Previously, immunohistochemical studies revealed that CDK1/cdc2 is overexpressed in a subset of colonic adenoma (14 of 59; 28%), which was more obvious in focal carcinoma (13 of 15; 86.7%) (13, 30). These data suggested an upregulation of CDK1/cdc2 accompanied by a malignant change. In accordance with these findings, we show here that expression of the CDK1/cdc2 suppressor pdcd4 is lost in progressed carcinomas of lung, breast, colon, and prostate. Furthermore, it seems that both pdcd4 localization and expression level directly correlate with tumor progression. Our data indicate that pdcd4 might represent a novel target for antineoplastic therapies.


This study was supported by Wilhelm Sander-Stiftung Grant 2000.120.1 (to R. Göke).


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract