Matrix Metalloproteinases 9 and 10 Inhibit Protein Kinase C–Potentiated, p53-Mediated Apoptosis

Resistance to apoptosis is a hallmark of cancer. Although numerous molecular events can enable cells to survive in the face of proapoptotic stimuli, the full repertoire of mechanisms used by cancer cells to resist programmed cell death has yet to be established. In the physiologic setting, the transcription factor p53 plays a major role in regulating cell survival, in addition to several other cellular processes that include cell cycle checkpoint control (1), DNA repair and senescence (2, 3). It is therefore not surprising that p53 is among the most commonly affected molecules in cancer, >50% of malignant human tumors bearing mutations in the TP53 gene. p53 is activated in response to a multitude of signals, including oncogene activation, hypoxia (4), and DNA damage (5), and when required, induces a cell cycle checkpoint blockade and/or cell death through modification of the expression level of several target genes. p53 target genes include, among many others, Bax (6), Mdm2 (7), Apaf1 (8), Fas (9), p21^WAF1, and DR5 (10). The consequences of the induction of these target genes are a function of the cellular context.

Recently, a new mode of action has been proposed for the proapoptotic activity of p53 (11). According to this view, p53 may function in a transcription-independent manner and may induce cell death by modifying mitochondrial membrane permeability, thereby activating the intrinsic apoptotic pathway. Regulation of p53 activity occurs at several levels, including post-translational modification (phosphorylation and acetylation), induction of the Mdm2 protein, which in turn induces ubiquitin-mediated p53 degradation (12), and intracellular localization. To act as a transcription factor, p53 must translocate from the cytosol to the nucleus. In contrast, its transcription-independent activity requires its presence at the mitochondrial surface where it can associate with the mitochondrial regulators of apoptosis Bax and Bcl-X_L (13). Localization is regulated in part by Mdm2, which sequesters p53 and forces its translocation back toward the cytosol, augmenting its availability for mitochondrial association (14, 15).

Recent evidence suggests that protein kinase C (PKC) isozymes are involved in the regulation of the p53 DNA-binding activity by directly phosphorylating p53 on the Ser³⁷⁸ residue (16). It has also been proposed that PKCδ regulates p53 expression at the transcriptional level and contributes to the accumulation of the p53 protein (17). PKC family members are serine/threonine protein kinases that are divided into three subfamilies depending on their structure and mode of activation. The classic PKC subfamily includes the α, β_I, β_II, and γ isotypes, which are activated by diacylglycerol and phosphatidylserine in a calcium-dependent manner. The atypical PKCs, which constitute the second subfamily, include the ζ and λ isotypes that are activated by phosphatidylserine alone. Finally, the novel PKC subfamily is composed of the δ, ε, η, and 𝛉, isotypes, which are activated by diacylglycerol and phosphatidylserine in a calcium-independent manner. These kinases play a role in multiple cellular functions, one of them being apoptosis. However, the precise role played by each of the different isotypes in programmed cell death (PCD) remains to be fully elucidated. For example, the novel PKC, PKCδ, is considered proapoptotic and is cleaved by caspase-3 (18, 19), one of the effector caspases whose activation is common to several apoptotic pathways. The resulting cleaved fragment is proapoptotic and initiates an apoptotic amplification loop by targeting its own activator, caspase-3. On the other hand, the full-size PKCδ can mediate apoptosis at the mitochondrial level (20, 21). In contrast, the classic PKCs, PKCα, PKCβ_I, and PKCβ_II are believed to exert antiapoptotic activity. PKCα, for example, is able to phosphorylate and stabilize the antiapoptotic molecule Bcl-2 (22). Several other reports show data suggesting that classic PKCs may display proapoptotic activity. The overall picture emerging from these observations is that even if the role played by PKC family members in apoptosis differs from isotype to isotype, the consequences of their activation are most likely dependent on the molecular and cellular context and the type of apoptotic stimulus.

In the present study, we sought to determine the effect of PKC activity on p53-induced PCD and the physiologic mechanisms that underlie resistance to p53-mediated apoptosis in tumor cells. To this end, we engineered the SW480 human colon adenocarcinoma cell line, which contains an endogenous mutant p53 (23), to overexpress wild-type p53 (wt p53) under the regulation of a tetracycline (Tet-ON) conditional system. Using this model system as well as additional colon carcinoma cell lines that express inducible exogenous or constitutive endogenous p53, we show that PKC activation and p53 expression function in synergy to induce apoptosis in colon carcinoma cells and that resistance to the combined proapoptotic effect of these molecules is conferred, at least in part, by matrix metalloproteinase 9 (MMP-9) and MMP-10 expression.

The SW480 parental human colon adenocarcinoma cell line was obtained from European Collection of Animal Cell Cultures (Porton Down, United Kingdom) and the HCT p53^+/+, HCT p53^−/−, HCT Bax^+/− control, and HCT Bax^−/− cells were a gift from Bert Vogelstein (The Johns Hopkins University Medical Institutions, Baltimore, MD). The p53-inducible EB-1 colon adenocarcinoma cell line was a gift from Phil Shaw (Department of Pathology, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland; ref. 24). Cell stocks were maintained in DMEM (Life Technologies, Basel, Switzerland) supplemented with penicillin, streptomycin, and 10% fetal bovine serum (FBS, Life Technologies). Tetracyclin (Calbiochem, Schwalbach, Germany) and ZnCl₂ (Sigma Co., Buchs, Switzerland) were dissolved in water, whereas 5-fluorouracil (5-FU) and phorbol 12-myristate 13-acetate (PMA, both from Sigma) were dissolved in DMSO. z-VAD-fmk (BIOMOL Res. Lab., Butler Pike, PA) and 1,10-phenanthroline monohydrate (Sigma) were dissolved in DMSO. The metalloproteinase inhibitor GM6001 and its negative control (Calbiochem) were dissolved in DMSO, as were the PKC inhibitors rottlerin/mallotoxin, bisindolylmaleimide I, and Gö6976 (Calbiochem).

The human wt p53 cDNA sequence was subcloned as a 1,200-bp BamHI-EcoRI fragment into the pcDNA4/TO/myc-His A vector (Invitrogen, Basel, Switzerland). The resulting vector (p4TO-hwtp53) allowed expression of wt p53 under the control of a tetracycline operator 2 sequence–containing cytomegalovirus promoter. The SW480 cell line was first stably transfected with the pcDNA6/TR vector (Invitrogen), containing sequences encoding the tetracycline repressor protein, using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's recommendations. The bulk population was then stably transfected with the p4TO-hwtp53 plasmid. After selection with Zeocin and Blasticidin (Invitrogen), a monoclonal population was isolated and named SMF1.

SMF1 cells were treated with tetracycline and PMA (2 and 0.16 μmol/L, respectively) for 24 hours. Resistant cells were rinsed with PBS and cultured in DMEM/10% FBS for 2 weeks. The resulting population was subjected to additional rounds of tetracycline + PMA stimulation and selection. After the fifth round of selection, surviving cells were tested for p53/PKC resistance and preservation of tetracycline-inducible wt p53 expression. The resulting SMF1-derived cell line was named S1R6.

Total RNA was isolated from SMF1 or S1R6 cells after 8 hours of DMSO or PMA/tetracycline stimulation using the RNeasy RNA extraction kit (QIAGEN, Basel, Switzerland) according to the manufacturer's recommendations. The isolated RNA was linear-amplified using the Arcturus (Mountain View, CA) riboamp RNA amplification kit. For each experiment, 6 μg of sample 1 mRNA were labeled with dCTP-Cy3 and 6 μg of sample 2 mRNA were labeled with dCTP-Cy5. Labeled samples were pooled and hybridized in duplicate to human cDNA microarray (Hu10K) chips from NCCR/ISREC containing 11,552 spotted elements corresponding to 9,500 unique human cDNAs (for an accurate description see http://intranet.isrec.ch/microarrays/nccr_isrec.html). Hybridization was done in hybridization chambers (Corning, Baar, Switzerland) in a 64°C water bath for 16 hours. After several careful washing steps, the chips were scanned in a Perkin-Elmer/GSI Lumonics ScanArray 4000 scanner and scanned slide images were converted to a tagged image file format. Spot intensity was analyzed with the ScanAlyze software (http://rana.lbl.gov/EisenSoftware.htm) and further primary data analysis was done using com.braju.sma routines in the R statistical package (http://www.maths.lth.se/help/R/ and http://www.r-project.org/). We used the nonlinear Scaled print-tip normalization method (25). Quality control of slide hybridization was done using variables described on the corresponding web site: http://intranet.isrec.ch/microarrays/nccr_isrec.html.

When comparing SMF1-DMSO– and SMF1-PMA/tetracycline–treated cells, we considered spots with an average ratio value superior to 1.85 to correspond to positively regulated genes; a ratio of 1.00 was used for comparisons between PMA/tetracycline-treated SMF1 and S1R6 cells. The average ratio was determined for each set of experiments as the experiment background intensity plus two SDs.

Total RNA was isolated with the RNeasy extraction kit (QIAGEN) and cDNAs were produced by reverse transcription with the Moloney murine leukemia virus Reverse Transcriptase RNase H minus, point mutant (Promega, Wallisellen, Switzerland) and Random primers (Promega). Real-time PCR was done in an Applied Biosystems 7900HT Sequence Detection System using the ABI Taqman Master mix chemistry. The sets of primers and Taqman probes were as follows: cyclophilin A, ABI human PPIA endogenous control [VIC/minor groove binder (MGB) Taqman probe]; large ribosomal protein, ABI human RLPO endogenous control (VIC/MGB Taqman probe); MMP-9 (26): forward primer 5′-CCTGGAGACCTGAGAACCAATCT-3′, reverse primer 5′-TGCCACCCGAGTGTAACCA-3′, FAM/TAMRA-Taqman probe 5′-CAGCTGGCAGAGGAATACCTGTACCGCT-3′; MMP-10, ABI Assay on Demand Ref. Hs00233987_m1 (FAM/MGB Taqman probe); MMP-12, ABI Assay on Demand Ref. Hs00159178_m1 (FAM/MGB Taqman probe).

The resulting data were analyzed with the comparative C_t method for relative gene expression quantification.

Propidium iodide sub-G₁ peak quantification (Nicoletti stain). Following stimulation, dead cell–containing supernatants were collected and pooled with trypsinized living cells. Samples were then centrifuged and washed with PBS. After a second centrifugation step, cells were permeabilized with 70% ethanol, washed in PBS, and treated for 15 minutes at 37°C with RNase A (Sigma, 1 mg/mL in PBS) and for 15 additional minutes at room temperature with propidium iodide (PI, Sigma, 100 μg/mL in PBS). Samples where then analyzed on a FACS analyzer (Becton Dickinson, San Jose, CA). The percentage of specific cell death was calculated using the following formula:

Annexin V staining. The Annexin V-Biotin Apoptosis Detection Kit from Oncogene Research Products (Lucerne, Switzerland) was used according to the manufacturer's protocol. Briefly, dead and living cells were pooled and washed once with PBS. The Annexin V-Biotin conjugate was added to the sample in binding buffer [10 mmol/L HEPES (pH 7.4), 150 mmol/L NaCl, 2.5 mmol/L CaCl₂, 1 mmol/L MgCl₂, 4% bovine serum albumin] and incubated at room temperature in the dark for 15 minutes, centrifuged, and resuspended in 0.5 mL binding buffer supplemented with 15 μL streptavidin-FITC (eBioscience, San Diego, CA; 15 μg/mL) and 10 μL PI (Sigma, 1 μg/mL). Cells were kept at 4°C in the dark and analyzed immediately by flow cytometry.

Terminal deoxynucleotidyl transferase–mediated nick-end labeling staining. Terminal deoxynucleotidyl transferase–mediated nick-end labeling (TUNEL) assays were done using the In situ Cell Death Detection Kit, AP from Roche Diagnostics (Rotkreuz, Switzerland), according to the manufacturer's instructions. Briefly, cells were grown on glass chamber slides. Following stimulation, the slides were washed in PBS, air-dried, and fixed in 4% paraformaldehyde in PBS (pH 7,4). After a second washing step, cells were permeabilized in 0.1% Triton X-100 with 0.1% sodium citrate. Samples were labeled by adding fluorescein-dUTP conjugate and terminal deoxynucleotidyl transferase and incubating for 1 hour at 37°C in the dark. After several washes in PBS, the cells were examined under a fluorescence microscope. The signal was converted using an AP-conjugated anti-fluorescein antibody. Phosphatase activity was detected by immersion of the slides in AP substrate–containing staining solution [nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate mix in 0.1 mol/L Tris (pH 9,5), 0.05 mol/L MgCl₂, and 0.1 mol/L NaCl] for 45 minutes in the dark. The reaction was stopped by addition of double distilled water.

Mitochondrial transmembrane potential (ΔΨ_m) analysis. Briefly, DMSO- or PMA/tetracycline-stimulated cells (5 × 10⁵ cells per sample) were washed twice with cold PBS containing 1% serum. For each sample, two tubes were prepared, one containing cells stained with the carbocyanine dye 3,3′-dihexyloxacarbocyanine iodide (Molecular Probes, Eugene, OR) and the other containing unstained cells that served as negative controls for fluorescence quantification. After 15 minutes of incubation at 37°C, the samples were washed in PBS containing 1% serum and resuspended in a final volume of 200 μL PBS/1% FBS. Samples were kept at 4°C in the dark and analyzed immediately by flow cytometry.

Crystal violet staining. DMSO-, PMA-, 5-FU-, or 5-FU/PMA-treated cells were washed with PBS. Cells were stained with 0.5% Crystal violet solution (25% methanol) for 15 minutes. After two washes with distilled water, cells were lysed in 1% SDS and absorbance at 620 nm was measured. All 620-nm absorbance values were expressed as a percentage of values obtained from lysates of DMSO-treated cells, giving the percentage of cell viability. Cell death rate was calculated by subtracting this percentage from 1.

SDS-PAGE protein electrophoresis and Western blotting were done according to standard procedures. For immunodetection of the human p53 protein, we used the DO-1 mouse monoclonal IgG2a antibody (Santa Cruz Biotechnology, Santa Cruz, CA). For caspase-3/Yama/Apopain/CPP32, we used the Ab-1 mouse monoclonal IgG1 antibody (Oncogene Research Products), which recognizes both the zymogen and the cleaved form of the caspase-3. For human β-actin, we used the AC-74 mouse monoclonal IgG2_a antibody (Sigma). The secondary antibody was a horseradish peroxidase–conjugated antimouse IgG and the signal was revealed using the Enhanced Chemiluminescence PLUS substrate solutions (Amersham Biosciences., Otelfingen, Switzerland).

For specific gene silencing, we used the small interfering RNA (siRNA) technology. Cells grown to 30% to 40% confluence in 24-well culture plates were transfected with siRNA solution consisting of oligofectamine reagent and the Opti-MEM (both from Invitrogen). Following transfection, cells were maintained in DMEM/10% FCS without antibiotics for 48 hours. We used the following siRNA: control, sense sequence 5′-AAAGGAAACUGGAAAAAUGtt-3′ and antisense sequence 5′-CAUUUUUCCAGUUUCCUUUtt-3′; anti–MMP-9, sense sequence 5′-CAUCACCUAUUGGAUCCAAtt-3′ and antisense sequence 5′-UUGGAUCCAAUAGGUGAUGtt-3′; anti–MMP-10, sense sequence 5′-GGAGGACUCCAACAAGGAUtt-3′ and antisense sequence 5′-AUCCUUGUUGGAGUCCUCCtc-3′; anti–MMP-12, sense sequence 5′-GGAAACAUUAUAUCACCUAtt-3′ and antisense sequence 5′-UAGGUGAUAUAAUGUUUCCtc-3′.

All siRNAs were synthesized by Ambion, Inc. (Huntingdon, United Kingdom).

After cell stimulation, the serum-free supernatant was concentrated with centricon column with a 30-kDa cutoff value (Millipore, Volketswil, Switzerland). The concentrated solution was mixed with Laemmli sample buffer, without β-mercaptoethanol, and loaded onto a gelatin (1 mg/mL) containing 10% SDS-PAGE gel. Following the completion of electrophoresis, the gel was washed in Triton X-100 solution and the proteases were activated overnight at 37°C in 5 mmol/L CaCl₂ and 50 mmol/L Tris (pH 8). The gel was then stained with Coomassie blue.

To address the mechanisms that underlie tumor cell resistance to apoptosis and to elucidate the effect of PKC activation on p53-induced proapoptotic signaling in human cancer cells, we selected the SW480 cell line established from a primary human colon adenocarcinoma. These cells express a p53 protein containing two distinct mutations: G273A (ref. 23; converting an arginine to a histidine residue) and C309T (converting a proline to a serine residue; ref. 27), which result in p53 inactivation. To study the effect of wt p53 overexpression in SW480 cells, we applied a tetracycline-regulated wt p53 expression system (28). PKC activation was achieved by exogenous administration of the phorbol ester PMA, which is a major activator of PKCs as well as of other signaling pathways, including mitogen-activated protein kinases and nuclear factor κβ.

SMF1 cells overexpress wild-type human p53 under the control of tetracycline (Tet-ON system) and initiate a cell death programme accelerated by phorbol 12-myristate 13-acetate stimulation. SMF1 cells were generated by genetically modifying SW480 cells to stably express the tetracycline repressor protein allowing conditional expression of wt p53. Expression of exogenous p53 was achieved by addition of tetracycline to the culture medium (Fig. 1A). DMSO and PMA stimulation (Fig. 1A,, lanes 1 and 3, respectively) had no effect on endogenous p53 expression in SMF1 cell lysates, whereas lysates from SMF1 cells stimulated with either tetracycline alone or tetracycline and PMA show expression of both the endogenous and the wild-type exogenous protein (Fig. 1A,, lanes 2 and 4, respectively). Twenty-four hours following induction of expression, p53-dependent cell death was almost negligible (mean 1%, Fig. 1B). Stimulation with PMA alone was not accompanied by significant induction of cell death, but costimulation with tetracycline and PMA resulted in 73% cell death at 24 hours (Fig. 1B). Observations by light microscopy indicated the presence of detectable cell death 12 hours following combined p53 induction and PMA stimulation in SMF1 cells (data not shown). Thus, PMA treatment sensitized SMF1 cells to p53-dependent cell death. The notion that this effect was p53 dependent and not related to some property of the tetracycline molecule itself is supported by the observation that PMA-induced sensitization did not occur in tetracycline-treated parental SW480 cells that do not express wt p53 (Fig. 1B). In addition, tetracycline stimulation of SW480 cells stably transfected with tetracycline repressor and the p53-independent gene PLSCR1 did not result in a higher ratio of cell death than in the parental SW480 cells (data not shown).

To ensure that the observed sensitization of p53-mediated cell death by PMA was not a cell type–specific event, we tested the effect of p53 induction in the presence and absence of PMA in the EB-1 colon carcinoma cell line (24) engineered to express wt p53 gene under the control of the metallothionein MT-1 promoter. Unlike SMF1 cells, which contain mutated p53, EB-1 cells lack endogenous p53 expression. Treatment of EB-1 cells with ZnCl₂ led to a robust induction of p53 (Fig. 1C) accompanied by 20% cell death at 24 hours, whereas costimulation with ZnCl₂ and PMA resulted in 40% cell death at the same time point (Fig. 1C).

To determine whether the observed effect of p53/PMA costimulation occurs in a more physiologic context, we stimulated HCT116 colon carcinoma cells, which constitutively express wt p53 (HCT^+/+), and their p53^−/− derivatives (HCT^−/−) with the anticancer drug 5-FU, a known p53 inducer. Treatment of HCT^+/+ cells with 5-FU augmented endogenous p53 expression and induced 34% cell death at 24 hours (Fig. 1D). Consistent with the observations using SMF1 and EB-1 cells, 5-FU–induced cell death was potentiated by costimulation with PMA (Fig. 1D). 5-FU cytotoxicity was due, in part, to p53-independent events, as 5-FU treatment of HCT p53^−/− cells resulted in 15% to 19% cell death (Fig. 1D). Importantly, however, PMA costimulation did not enhance 5-FU–mediated death of HCT^−/− cells (Fig. 1D).

Cell death observed in phorbol 12-myristate 13-acetate/tetracycline stimulated cells shows features of apoptosis, including DNA cleavage, phosphatidylserine externalization, and mitochondrial transmembrane potential ΔΨ_m alteration. Differentiation between apoptotic and necrotic cell death is based on several morphologic and functional criteria. We therefore applied a series of approaches to determine the type of cell death that was induced by the combination of p53 and PMA. Analysis of genomic DNA from dying cells revealed a DNA ladder (data not shown) which is a hallmark of apoptosis. p53/PMA costimulation was observed to induce several other apoptosis-specific molecular and cellular features. Thus, TUNEL staining (Fig. 2A) showed the presence of nick-containing DNA in the PMA/tetracycline-stimulated cells, which occurs in most forms of apoptosis as a result of endonuclease activation. Annexin V-biotin conjugate reactivity revealed externalization of phosphatidylserine (29, 30), another characteristic feature of programmed cell death (Fig. 2B). Similar phosphatidyl serine externalization was observed in EB-1 and HCT^+/+ cells in response to ZnCl₂ and 5-FU stimulation, respectively (data not shown). Finally, using a mitochondrial membrane carbocyanine dye (Fig. 2C), we could show that p53/PMA altered the mitochondrial transmembrane potential ΔΨ_m (31, 32), in SMF1 cells, suggesting that activation of the intrinsic mitochondrial apoptotic pathway may contribute to cell death induced by the combined effect of p53 and PMA.

Selection of S1R6 cells that are partially resistant to p53/phorbol 12-myristate 13-acetate costimulation induced apoptosis. Because all three cell types displayed a similar response to p53/PMA costimulation, we selected SMF1 cells for subsequent experiments. Although most of the SMF1 cells treated with tetracycline and PMA underwent apoptosis, a small fraction (<10%) survived. To address the mechanisms that underlie tumor cell resistance to the synergistic cell death inducing effect of PMA and p53, we subjected the surviving cells to several rounds of tetracycline/PMA treatment and selection. The resulting S1R6 cell line displayed markedly different sensitivity to PMA/p53 compared with the parental SMF1 cells (Fig. 2). Thus, in the TUNEL assay (Fig. 2A), PMA/tetracycline stimulation induced only weak staining in S1R6 cells, whereas the majority of the SMF1 cells were strongly positive (DMSO treated SMF1 and S1R6 cells, used as controls, were TUNEL negative). Similarly, phosphatidylserine externalization in S1R6 was weaker than in SMF1 (18% compared with 45%, Fig. 2B). In addition, whereas there was little relevant mitochondrial transmembrane potential difference between DMSO- and PMA/tetracycline-treated S1R6 cells (DMSO: 14% and PMA/tetracycline: 16%), a major difference was observed in SMF1 cells (DMSO: 9% and PMA/tetracycline: 49%, Fig. 2C). A cell death time course (Fig. 2D) indicates that at 24 and 48 hours, the cell death observed in S1R6 represents 21% and 46%, respectively, of that induced in SMF1 cells.

Phorbol 12-myristate 13-acetate/tetracycline–induced apoptosis is caspase dependent. Most of the apoptotic signaling pathways described thus far are mediated by the caspase family of cysteine-proteases (33). The activated forms of these enzymes cleave diverse targets that are important for several key cellular functions, including DNA repair and cell structure maintenance, and are also able to activate other caspases. We therefore addressed their implication in PMA/p53-induced apoptosis. Caspase-3 (Apopain/YAMA/CPP32), a member of the effector caspase subfamily, was observed to be activated by PMA/p53, as shown by the cleavage of the 32-kDa zymogen (Fig. 3). The active 17-kDa fragment was detectable only in the PMA/tetracycline-stimulated SMF1 cell lysates after 24 hours (Fig. 3A). Cleavage was not detected in SMF1 cell lysates at 8 hours or in S1R6 cell lysates at any time. The 116-kDa form of poly(ADP-ribose) polymerase (PARP), a nuclear protein implicated in DNA repair and known to be a target of at least caspase-3 and caspase-7 was also found to be cleaved in SMF1 cells after PMA + tetracycline stimulation (data not shown). Finally, using the generic caspase inhibitor, z-VAD-fmk (34), which inactivates most members of the family, we observed complete inhibition of PMA/tetracycline-induced apoptosis in S1R6 cells but incomplete inhibition in SMF1 cells (Fig. 3B).

Because activation of Bax, which leads to the release of cytochrome c from mitochondria and caspase activation, is a key mechanism underlying p53-mediated cell death, we tested the effect of 5-FU in the presence and absence of PMA on HCT 116 Bax^−/− cells. Cell death induced by 5-FU and 5-FU/PMA was lower in HCT Bax^−/− cells than in HCT Bax^+/− counterparts (Fig. 3C). The observed cell death in HCT Bax^−/− cells most likely corresponds to p53-independent cytotoxicity of 5-FU consistent with the notion that Bax is implicated in apoptosis induced by p53 and p53/PMA.

Phorbol 12-myristate 13-acetate sensitization to p53-induced apoptosis is protein kinase C mediated. Because the phorbol ester PMA is known to activate several members of the PKC family in addition to other signaling pathways, we attempted to identify the PKC family members implicated in the observed PMA/p53-induced apoptosis using specific PKCs inhibitors, including rottlerin (35), a novel PKC subfamily inhibitor, Gö6976 (36, 37), a specific inhibitor of the classic PKCs α and β, and bis-indolylmaleimide I (38), a broad classic PKC inhibitor. When pretreated with either classic PKC inhibitor, both SMF1 and S1R6 cells displayed a marked reduction in cell death following PMA/tetracycline stimulation (Fig. 4). In contrast, the novel PKC inhibitor rottlerin enhanced PMA/tetracycline sensitivity of both cell lines. These observations suggest that the potentiation of p53-induced cell death by PMA in SW480 cells is mediated by PKCs of the classic subfamily and antagonized by PKCs of the novel subfamily.

S1R6 expresses higher levels of matrix metalloproteinases after phorbol 12-myristate 13-acetate/tetracycline stimulation. To address the mechanisms responsible for the observed resistance of S1R6 cells to PKC/p53-induced apoptosis, we did a microarray gene expression analysis of SMF1 and S1R6 cells. First, we compared gene expression in DMSO- and PMA/tetracycline-stimulated SMF1 cells. Among 474 genes that displayed a >3-fold induction in cells treated with PMA/tetracycline, were several p53 target genes, PKC-related genes, and genes implicated in apoptosis (data not shown). P53 target genes included PIG11, CDKN1A (encoding p21), and MDM2. The gene encoding p53 itself, TP53, was up-regulated, and as expected, genes encoding PKC family members, including PRKCA and PRKCZ, and genes encoding proapoptotic proteins, including BIK, CASP10, and APAF1 were also induced by PMA/p53 (data not shown).

A second round of microarray analysis was done to compare gene expression in SMF1- versus S1R6-PKC/p53–stimulated cells. Ninety-seven genes were found to display a >2 fold induction in S1R6 cells (Table 1 and data not shown). At least two groups of genes of potential interest were found up-regulated. The first includes genes encoding proteins involved in cell adhesion and cell-cell communication. The second includes genes encoding matrix metalloproteinases (MMP), and more specifically, MMP-9, MMP-10, and MMP-12. Because of their implication in the regulation of cell survival, we chose to address the functional role played by the MMPs in response to PMA/p53 stimulation.

Matrix metalloproteinase activity plays a role in protein kinase C/p53-induced apoptosis. To test the potential role of MMP-mediated proteolysis in colon carcinoma cell response to PMA/p53 signals, we first blocked MMP proteolytic activity using two broad spectrum MMP inhibitors: the metal chelating agent o-phenanthroline (PHE), a molecule known to interfere with the MMP-zinc ion fixation (39), and GM6001 (40), a hydroxamic acid isomer known to be a potent collagenase inhibitor. Figure 5A illustrates the effect of pretreatment of SMF1 or S1R6 cells with these drugs. PMA/tetracycline-induced apoptosis was augmented in both cell lines following PHE compared with DMSO treatment (64% and 36% for SMF1 and 53% and 24% for S1R6, respectively). Similarly, treatment with GM6001 increased the fraction of cells undergoing apoptosis in response to PMA/tetracycline to 75% and 35% among SMF1 and S1R6 cells, respectively. Thus, MMP activity seems to provide the tumor cells with partial protection against apoptosis induced by p53/PKC costimulation. To make sure that these effects were not specific to the DNA degradation marker of apoptosis, we did Annexin V binding analysis and found that both inhibitors affect phosphatidylserine externalization in a similar way (data not shown).

Matrix metalloproteinase 9– and matrix metalloproteinase 10–specific small interfering RNAs sensitize S1R6 cells towards phorbol 12-myristate 13-acetate/tetracycline–induced cell death. To identify the MMPs that confer protection to S1R6 cells from PKC/p53-induced cell death, we did siRNA-transient transfection experiments (41) to specifically silence MMP-9, MMP-10, and MMP-12 transcripts. Each siRNA induced a reduction of the corresponding mRNA expression as detected by real-time PCR analysis (Fig. 5B-D). MMP-9 and MMP-10 gene silencing was observed to be followed by corresponding cell sensitization to PMA/tetracycline-induced cell death (Fig. 5B and C). MMP-9– and MMP-10–specific siRNAs induced 257% and 161% sensitization, respectively. In contrast, MMP-12 siRNA failed to modify S1R6 cell sensitivity to PKC/p53-induced apoptosis (Fig. 5D). Thus, consistent with the results obtained using chemical MMP inhibitors, MMP-9 and MMp-10 expression may play a protective role in colon cancer against PKC/p53-induced apoptosis.

This notion is supported by the observation that sensitization of EB-1 cells, which express MMP-9, to ZnCl₂/PMA-mediated apoptosis could also be achieved by MMP-9–specific siRNA, with a 30% increase in cell death (data not shown). By contrast, HCT^+/+ cells, which are negative for MMP-9 expression, as assessed by gelatin zymography, displayed no sensitization to p53-mediated cell death following transfection with MMP-9 siRNA (data not shown).

p53 down-regulates expression of matrix metalloproteinase 9 at the transcriptional level. p53 is known to induce the expression of several MMPs, but its effect on MMP-9 has not been fully explored. To determine the effect of p53 on MMP-9 expression, we quantified the mRNA expression of the gene encoding MMP-9 by real-time PCR in response to p53 induction. Surprisingly, p53 was observed to be a negative regulator of the MMP-9 gene. Figure 6A shows relative quantification using the cyclophilin A mRNA as endogenous control. As expected, PMA induced MMP-9 (15.6-fold), but the addition of tetracycline together with PMA reduced this induction to 6.07-fold. It is important to note that the mRNA was extracted from PMA + tetracycline stimulated SMF1 cells before the onset of cell death. To eliminate the possibility of an effect due to intrinsic properties of the tetracycline molecule (independent of p53 induction), we did the same experiment in the SW480 parental cells. We observed a similar induction by PMA, but the addition of tetracycline failed to down-regulate MMP-9 (data no shown). These observations suggest that p53 is a negative regulator of MMP-9 mRNA expression. Accordingly, MMP-9 protein levels, as assessed by gelatin zymography, were decreased by p53 (Fig. 6B).

Inactivation of p53-dependent cell death by mutation or deletion of the TP53 gene is one of the key events in carcinogenesis. Nevertheless, many malignant tumors retain p53 expression and function, and an understanding of the mechanisms that potentiate p53-mediated tumor cell death as well as those that tumor cells deploy to survive in the face of p53 signals may have important implications in cancer treatment. One of the principal observations of the present work is that PMA treatment of three different colon carcinoma cell lines expressing inducible exogenous or constitutive endogenous p53 accelerates and enhances cell death induced by p53 expression alone. Importantly, cell death triggered as a result of 5-FU–induced endogenous p53 expression, which reflects a more physiologic situation than tetracycline- or ZnCl₂-mediated induction of exogenously introduced p53, was augmented in the presence of PMA.

Several criteria support the notion that the observed cell death was apoptotic. First, phosphatidylserine, normally present in the inner layer of the cytoplasmic membrane, was externalized in all three cell types. Second, SMF1 cells were positive for TUNEL staining. Third, chromatolysis and DNA fragmentation occurred in a nonstochastic manner. Fourth, caspases were involved in this process as shown by the cleavage of PARP and caspase-3. However, z-VAD-fmk was not able to completely abrogate SMF1 cell death even at a high concentration (100 μmol/L), suggesting that the cell death we observed is predominantly but not exclusively caspase dependent. Interestingly, mitochondria were also affected as shown by the modification of the mitochondrial transmembrane potential, suggesting that the intrinsic mitochondrial apoptotic pathway is also involved here. As might be expected, the use of HCT116 cells with homozygous deletion of Bax showed that Bax is involved in p53/PMA-mediated apoptosis.

The mechanism that underlies PMA-mediated potentiation of p53-induced cell death is consistent with PKC involvement. Phorbol esters activate the PKC family whose members phosphorylate and thereby regulate the activity of numerous cellular substrates, possibly including p53 itself. p53 activity is regulated, at least in part, by phosphorylation at several levels. First, p53 stability is increased by the activity of several kinases, including ATM, ATR, Chk1, and Chk2 (42), which directly phosphorylate p53 and/or its negative regulator MDM2 (43). Second, by phosphorylating sites at its NH₂ terminus, kinases can inactivate a nuclear export signal and thereby regulate the cellular localization of p53 (44). Third, the activity itself of p53 is prone to regulation by kinases. Thus, phosphorylation of p53 has been shown to control both its DNA binding ability and its transcriptional activity. In addition, p53 can be phosphorylated by the c-Jun NH₂-terminal protein kinases, known to be activated by PMA-induced PKC activity (45), and by the CAK (46) complex (CDK7, cyclin H, and p36MAT1). Changes in p53 phosphorylation may therefore provide at least one mechanism to explain the PKC-mediated sensitization of p53-dependent cell death observed in this study.

Members of the PKC family have been shown to interact with p53 and modulate p53-induced apoptosis and cell growth arrest. The classic PKCs are able to induce p53 translocation to the nucleus and the G₁ cell cycle control checkpoint (47) and PKCδ has been shown to induce an accumulation of p53 (17). Moreover, p53 contains a PKC binding site in the central oligomerization domain and can be directly phosphorylated by PKCs at Ser³⁷⁸ and by PKCα and PKCζ at Ser³⁷¹ (48). In addition, a recent study has shown activation by cyclin-dependent kinase 2/cyclin A and PKC of the p53-specific DNA sequence binding ability (16).

Because all novel PKCs and classic PKCs contain a phorbol ester binding motif, it would seem reasonable to expect that either or both of these subfamilies are implicated in the potentiation of p53-mediated cell death. Consistent with this notion, both classic PKC and novel PKC inhibitors affected p53-dependent apoptosis but in opposite directions. Thus, bis-indolylmaleimide I, which inhibits PKCα, PKCβ_I, PKCβ_II, PJCγ, PKCδ, and PKCε isozymes, was the strongest inhibitor of PMA-dependent cell death potentiation, with Gö6976, which inhibits PKCα and PKCβ_I isozymes, displaying a lesser degree of inhibition. Based on these findings, it would seem that the observed PMA-associated cell death enhancement was mediated by the PKCα, PKCβ_I, PKCβ_II, PKCγ, and PKCε isozymes. In contrast, PKCδ and PKC𝛉 seemed to have an antiapoptotic effect in p53-stimulated SW480 cells.

PKC targets independent of the p53 may contribute to the observed synergistic proapoptotic effect of PKC activation and p53 induction (Fig. 7). The function of PKCδ, which is both a substrate and an activator of caspase-3, has been suggested to be linked to the mitochondrial release of cytochrome c during apoptosis. Consistent with these observations, the phorbol ester 12-O-tetradecanoylphorbol-13-acetate induces the release of cytochrome c and activation of caspase-3. Our observation that inhibition of PKCδ by rottlerin sensitized colon carcinoma cells to p53 in our in vitro system may therefore be somewhat surprising and would seem to contradict the evidence suggesting that PKCδ is proapoptotic. However, rottlerin has been suggested to inhibit the intrinsic cell death pathway but to potentiate extrinsic cell death pathways (49). Thus, depending on the type of proapoptotic stimulus, rottlerin may be expected to exert a stimulatory or inhibitory effect. In the case of our model system, the observed effect of rottlerin is consistent with the possibility that even if there was activation of the intrinsic pathway, as suggested by changes in the mitochondrial transmembrane potential, the extrinsic proapoptotic pathway was likely to be dominant.

Selection of SMF1 cells that were resistant to PMA/p53-induced cell death (termed S1R6), allowed the identification of potential mechanisms that underlie the resistance. S1R6 cells were observed to retain tetracycline-regulated p53 expression, such that their resistance could not be attributed to loss of p53 induction. However, comparison of the gene expression profiles of S1R6 and SMF1 cells showed up-regulation of a relatively small number of genes among which two groups seemed of immediate interest: MMPs and cell adhesion molecules.

Members of the MMP family, also called matrixins, are cysteine proteases with zinc ion–dependent proteolytic activity (reviewed in refs. 50–52). Thus far, >20 MMPs have been identified in mammals and are currently grouped according to structural features (51). The majority of MMPs are secreted into the extracellular matrix (ECM), a subset being membrane bound by virtue of a transmembrane or PI-linked domain (51). All are produced as inactive zymogens and are activated by proteolysis. MMPs were originally described as being primarily involved in the proteolysis of ECM components and therefore believed to primarily mediate tissue remodeling. However, it is becoming increasingly clear that these enzymes play a critical role in a broad range of physiologic and pathologic processes. Thus, MMPs can cleave and regulate the activity of a host of growth factors, chemokines, and cytokines as well as cell surface adhesion receptors and proteoglycans involved in intercellular communication and cell migration that are directly implicated in tumor progression and metastasis (50–52).

Functional analysis in the present study showed that expression of MMP-9, and to a lesser extent that of MMP-10 but not MMP-12, confer partial resistance to PMA/p53-mediated cell death. A potential issue in our approach is the use of tetracycline to induce p53, because tetracycline and its derivative doxycyline have been shown to inhibit MMP activity and MMP-2 gene expression. However, inhibition occurs with an IC₅₀ of 30 to 300 μmol/L, whereas we used a concentration of 2 μmol/L, well below the reported inhibitory levels. Accordingly, gelatin zymography assays showed no difference between lysate and supernatant MMP activity from cells treated or not with tetracycline. The presence of tetracycline therefore seems to have no effect on the activity of MMP-9 in our system at the concentration used.

Similar to PKC, MMP activity may have both proapoptotic and antiapoptotic effects. Proapoptotic effects can be due to the proteolytic degradation of ECM proteins, such as laminin, that serve as ligands of the integrin class of cell surface adhesion receptors (50–52). In the absence of ligand, specific subclasses of integrins can no longer trigger signals necessary for normal epithelial and endothelial cell survival, resulting in a form of apoptosis known as anoikis. Conversely, ECM degradation can result in the release and augmented bioactivity of potent survival factors, such as insulin-like growth factor I (IGF-I) that may directly promote tumor cell resistance to proapoptotic signals and favor tumor growth. MMP-mediated cleavage of the integral membrane ligand of the death-inducing receptor Fas (FasL) can result in increased or reduced apoptosis depending on the physiologic context (50–52). The proapoptotic or antiapoptotic effect of any given MMP may therefore depend on the local cellular production and relative availability of proteolytically released and activated death-inducing and/or growth and survival factors.

There are several potential mechanisms whereby MMP-9 may promote tumor cell resistance to proapoptotic stimuli. MMP-9 can release and activate transforming growth factor β (TGF-β; ref. 50), cleave several chemokines and activate other MMPs including MMP-1 (49–51). TGF-β can promote survival in transformed cells, as can the chemokine IL-8, whose function is potentiated by MMP-9–mediated proteolysis (50–52). Activation of MMP-1 leads to the release of IGF-I from ECM-sequestered IGF binding proteins (50–52). However, the full range of MMP substrates is far from elucidated, and it is quite possible that proteolytic activation or inactivation of other regulators of cell survival may be implicated in the observed antiapoptotic effect of MMP-9 expression. High expression of MMP-9 is often associated with poor prognosis in cancer patients and, in at least some tumor types, correlates with increased angiogenesis and metastatic proclivity (50–52).

Interestingly, regulation of MMP expression is reported to be subjected, at least in part, to p53 control. Thus, p53 has been shown to induce transcriptional activation of the MMP-2 (53) gene promoter and to attenuate the production of MMP-1 (54) and MMP-13 (55). Consistent with earlier work by others (53), we have found that MMP-2 expression is enhanced by p53 (data not shown). In the present study, we observed that p53 represses the human MMP-9 gene, which in and of itself may contribute to p53-induced cell death. It is also interesting to note that the two MMPs (MMP-2 and MMP-9) that form the gelatinase subfamily of MMPs, based on structural similarities and shared substrate specificity (50–52), are differentially regulated by p53.

The second group of genes of potential interest found to be overexpressed in S1R6 cells encode proteins involved in cell-cell communication and/or adhesion. Several are components of tight junctions (zonula occludens) and contribute to the epithelial barrier function (claudin and occludin family of proteins), whereas E-cadherin is part of the zonula adherens junctions that generate signals sensed by p53 (56). Although further work will be required to address their role, it is conceivable that elevated expression of these proteins contributes to resistance to PKC/p53-dependent apoptosis.

Taken together, our observations provide evidence that PKCs are able to potentiate p53-induced apoptosis and that expression of MMP-9 and MMP-10 confers partial protection against the proapoptotic PKC/p53 signals. Understanding of the precise mechanism of PKC/p53 synergy and identification of the relevant PKC targets may help develop therapeutic strategies that, in combination with selective MMP inhibitors, could augment the effectiveness of tumor cell elimination.

Grant support: National Center for Competence in Research Molecular Oncology, Swiss National Scientific Foundation grant 31-65090.01, and Oncosuisse grant 1267-08-2002.

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.

We thank Josiane Wyniger and Otto Hagenbüchle for providing the cDNA microarrays, Bert Vogelstein for the HCT cell lines, Phil Shaw for EB-1 cells, Richard Iggo for the wt human p53 cDNA, and Nathalie Mayran for her advice regarding siRNA experiments.