Chromosome 20q13.2 is amplified in 20–30% of early-stage breast tumors and is associated with poor prognosis. Detailed mapping of the amplified region using molecular cytogenetics, positional cloning and genomic sequencing culminated in a detailed molecular description of the candidate oncogene ZNF217 . ZNF217 proteins resemble Kruppel-like transcription factors, localize predominately to the nucleus and associate with proteins involved in transcriptional repression. The findings that ZNF217 can immortalize human mammary epithelial cells and that its amplification is associated with poor prognosis suggest that it may play roles in both early- and late-stage breast cancer. We present evidence that ZNF217 can attenuate apoptotic signals resulting from telomere dysfunction as well as from doxorubicin-induced DNA damage and that silencing ZNF217 with siRNA restores sensitivity to doxorubicin. Moreover, elevated ZNF217 leads to increased phosphorylation of Akt, whereas inhibition of the phosphatidylinositol 3 kinase pathway and Akt phosphorylation decreases ZNF217 protein levels and increases sensitivity to doxorubicin. These results suggest that ZNF217 may promote neoplastic transformation by increasing cell survival during telomeric crisis and may promote later stages of malignancy by increasing cell survival during chemotherapy.
DNA amplification at chromosome position 20q13 is common in human breast tumors ( 1 ), appears to be an early event ( 2 ), correlates with poor prognosis ( 3 ) and may reflect the presence of at least two important, previously uncharacterized oncogenes ( 4 ). Amplification of 20q13 was initially discovered in 40% of breast cancer cell lines and 18% of primary breast tumors, using comparative genomic hybridization (CGH) ( 1 ). In a study of 132 primary breast carcinomas and 11 metastases, 20q13.2 amplification was found in 29% of the cases ( 3 ). Amplification was associated with high histologic grade, aneuploidy and high S-phase fraction. High-level amplification was associated with short, disease-free survival of patients with node negative breast cancer. Independent studies employing CGH have identified 20q13 amplification in colon, bladder, ovarian and brain cancers ( 5 – 9 ), suggesting this amplification may contribute to tumor progression in a wide spectrum of the most common cancers.
To identify the oncogene(s) driving amplification of 20q13.2, we pursued a positional cloning and genomic sequencing strategy that resulted in the identification of seven genes: ZNF217 , ZNF218 , NABC3 , NABC2 , NABC1 , CYP24 and PIC-like ( 10 ). Of these genes, only ZNF217 and NABC1 were expressed in breast cancer cell lines amplified at the 20q13.2 locus and only ZNF217 mapped within the 260 kb minimum common amplicon defined in over 300 tumors and cell lines ( 10 ). We have shown that ZNF217 is expressed at levels that correlate with the degree of amplification in specific cell lines and tumors ( 10 ). Thus, ZNF217 emerged as a strong candidate for the oncogene driving 20q13.2 amplification.
Full-length ZNF217 cDNAs encode two open-reading frames of 1062 and 1108 amino acids, because of alternative splicing of exon 4. Each predicted protein has eight C2H2 zinc fingers and a proline-rich domain. Sequence analysis of ZNF217 indicates a strong resemblance to members of the Kruppel-like family of zinc finger proteins ( 11 ). Other members of the Kruppel-like family have been implicated in several human malignancies, including WT1 in Wilms tumor and GLI1 in glioblastoma. The eight zinc finger domains in ZNF217 are interspersed throughout the ZNF217 sequence, and their pattern does not appear to fall into one of the three classes: triple-C2H2, multiple-adjacent-C2H2 and separated-paired-C2H2 finger proteins (reviewed in 12). The sixth and seventh zinc fingers in ZNF217 are separated by the conserved linker sequence, TGEKP, reported to bind DNA with high affinity ( 13 ). Database analysis indicates that this paired zinc finger region aligns with those in several members of the Delta-EF1/ZFH-1 family of two-handed zinc-finger homeodomain proteins, including Smad-interacting protein 1 ( 14 ).
As members of the Kruppel family are involved in many aspects of growth, differentiation and development, the precise function of ZNF217 and its possible role in tumorigenesis could not be predicted on the basis of homology alone. ZNF217 protein localizes to the nucleus ( 4 ) and co-immunoprecipitates with complexes containing the transcriptional corepressors CoREST and CtBP, histone deacetylases HDAC1 and 2 and histone methyltransferases G9a and Eu-HMTase1 ( 15 , 16 ). This strongly suggests that ZNF217 may function as part of a transcriptional repressor complex. Changes in CtBP expression, in turn, can modulate the cellular threshold for apoptotic responses by contributing to the repression of certain proapoptotic genes ( 17 ). These include well-established p53 target genes as well as genes that are not known to be regulated by p53. Although CtBP does not appear to be a direct antagonist of transactivation by p53, there is a significant overlap in target genes inversely regulated by the two factors.
Additional clues to a possible ZNF217 function come from a cell culture study that found that human uroepithelial cells immortalized after exposure to the HPV16 E7 oncogene routinely contained extra chromosome 20q material ( 18 ). This work suggested that ZNF217 or another gene within the amplified region might complement E7's role in immortalization. In addition, the average number of additional chromosome 20 signals increased with passage in one such E7 expressing line, suggesting that the amplification might confer a selective advantage in culture. An association between high-level 20q13.2 amplification and genome instability was also reported, despite the presence of wild-type p53, suggesting that 20q gain might generate or enable cells to tolerate increased genome instability.
To investigate the cellular consequences of ZNF217 over-expression in tumor cells, we transfected HeLa cells with a plasmid encoding a ZNF217 –enhanced green fluorescent protein (EGFP) fusion. Control HeLa cells express low levels of endogenous ZNF217 (data not shown). Cells accumulated faster in ZNF 217–EGFP transfected cultures than in either parallel control cultures of non-transfected cells or cells transfected with an EGFP construct alone (Fig. 1 A). In three independent experiments, ∼40% more EGFP positive cells were present in ZNF217 –EGFP transfected cultures when compared with control EGFP cultures 72 h after plating equal numbers of cells. Analysis of DNA content showed no apparent difference in cell-cycle distribution between the two populations of cells (data not shown). However, the proportion of floating cells accumulating propidium iodide (PI), an indicator of loss of viability, was significantly lower in ZNF217 –EGFP transfected cultures than that in either non-transfected cells or cells transfected with an EGFP construct alone (Fig. 1 B). Assays of Annexin-V staining, an independent indicator of cell death, confirmed that spontaneous cell death was suppressed in HeLa cells expressing ZNF217 –EGFP (data not shown).
To determine whether ZNF217 prevented cell death induced by different mechanisms, we assayed the responses of cells expressing different levels of either endogenous or exogenously introduced ZNF217 to doxorubicin. Doxorubicin is a chemotherapeutic agent known to inhibit topoisomerase II and to induce double-strand DNA breaks, resulting in ATM/p53-mediated apoptosis ( 19 ). The HBL100 cell line, which expresses relatively low levels of endogenous ZNF217 transcripts, was found to be much more sensitive to doxorubicin-induced cell death than either the MCF7 or the 600MPE (Fig. 2 A), both of which express high levels of endogenous ZNF217 transcripts ( 10 ). To directly assess the protective effect of ZNF217 against doxorubicin-induced cell death, we compared HeLa and HBL100 cells transfected with ZNF217 –EGFP with control cells. In multiple independent experiments in both cell types, ZNF217 –EGFP-expressing cells displayed statistically significant lower levels of doxorubicin-induced cell death than non-transfected or EGFP-transfected controls (Figs 2 B and C).
To establish whether ZNF217 plays a direct role in conferring resistance to doxorubicin, four siRNA molecules were selected to specifically down-modulate its expression in the MCF7 cells. SiRNAs 1, 2 and 3 all elicited increased cell killing by doxorubicin, and siRNAs 1 and 2 were shown to knock down ZNF 217 mRNA and/or protein (mRNA and protein levels were not evaluated for siRNA 3). Transfection of siRNA 4 into MCF7 had no effect on ZNF 217 mRNA or protein level or cell death (data not shown). Maximum knockdown of both ZNF 217 mRNA and protein was achieved with siRNA 1, which was used for further studies. Transient transfection with this ZNF217 -targeted siRNA resulted in reduction of both ZNF217 mRNA and protein levels when compared with control cells transfected with scrambled control RNA in HBL100, HBL100 transfected with ZNF217 –EGFP (Z-HBL100), BT474, MCF7 and three brain tumor cell lines with low (SF188), medium (U251) and high (U251+7.9) relative expression of ZNF217 (data not shown) (Fig. 3 A). In three independent experiments, transfection with ZNF217 siRNA 1 resulted in an increase in spontaneous cell death and cell death induced by doxorubicin (Fig. 3 B). The data are consistent with the hypothesis that in the cells in which it is highly expressed, ZNF217 supports basic cell survival as well as survival in response to DNA damage.
The ability of ZNF217 to confer resistance to spontaneous and doxorubicin-induced cell death in multiple transformed cell lines suggests that its expression may be selected for whenever cell survival is threatened during multiple stages of tumor progression. Our past work ( 20 ) showed that expression of exogenously introduced ZNF217 does not directly activate telomerase expression in human mammary epithelial cells, but that it allows a subset of such cells to continue growing when control cells have succumbed to growth arrest and cell death associated with critically short telomeres. We hypothesized that high ZNF217 expression protects against telomere dysfunction-induced growth arrest and/or cell death. To test this hypothesis directly, we manipulated the levels and function of telomere binding proteins, TRF1 and TRF2, to disrupt telomere integrity in cell lines expressing different levels of endogenous or exogenous ZNF217 . Expression of a dominant negative TRF2 mutant (TRF2 ΔBΔM ) has previously been shown to trigger ATM/p53 dependent apoptosis ( 21 ), whereas over-expression of wild-type TRF1 has been shown to cause ATM/p53 independent cell death in susceptible cells ( 22 ). Transiently transfected TRF2 ΔBΔM or TRF1 induced significantly less cell death in the MCF7 and 600MPE, which express endogenous ZNF217 at high levels, than that induced in HBL100 or HeLa cells, which express low levels of endogenous ZNF217 (Fig. 4 ). In HBL100 and HeLa cells transfected with both TRF2 ΔBΔM and ZNF217 –EGFP, the level of cell death was 3–5-fold lower than that observed in cells transfected with TRF2 ΔBΔM alone. Similarly, transient transfection with TRF1 alone induced significant cell death in HeLa and HBL100 cells, whereas cotransfection with ZNF217 –EGFP attenuated cell death in the same cell lines.
Akt activation is known to promote cellular survival in a variety of experimental systems. Activated Akt directly phosphorylates key mediators of the apoptotic cascade and indirectly regulates the expression of pro- and anti-apoptotic genes ( 23 , 24 ). To begin to elucidate the mechanism by which ZNF217 promotes cell survival, we asked whether increased cellular ZNF217 protein leads to activation of the phosphatidylinositol 3′ kinase (PI3K) pathway, as indicated by elevated levels of phosphorylated Akt. Transfection of HBL100 with ZNF217 –EGFP resulted in increased phosphorylation of Akt (Ser-473) relative to the non-transfected HBL100 cells that weakly express ZNF217 (Fig. 5 ). In the MCF7 cell line in which ZNF217 is amplified, we detected high levels of both ZNF217 protein and phosphorylated Akt (Ser-473), without a concomitant increase in total AKT levels, relative to HBL100 cells. Silencing ZNF217 with siRNA resulted in decreased Akt (Ser-473) phosphorylation. Finally, inhibition of PI3K and thus Akt phosphorylation with the PI3K inhibitor Ly294002 reduced both Akt phosphorylation and ZNF217 protein levels in the MCF7, and this was accompanied by increased apoptosis in response to doxorubicin (data not shown). In all cases, the levels of total Akt were relatively unchanged. The data are therefore consistent with the concepts that ZNF217 effects on cell survival are mediated at least partly through activation of the PI3K/AKT pathway, and that there is a feedback regulation between the Akt pathway and ZNF217 expression.
We have found that ZNF217 can attenuate apoptotic signals emanating from DNA damage by a common therapeutic agent and from functionally compromised telomeres. This finding is highly significant because it may explain how elevated ZNF217 expression contributes to both early and late stages of tumor progression. ZNF217 may act early to promote survival of preneoplastic cells with dysfunctional telomeres and later to confer resistance to chemotherapy. Absence of adequate TTAGGG repeat DNA or artificially disrupted binding of telomeric proteins may cause telomeric DNA to resemble double-stranded breaks, which are the predominant initiating signals for ATM-mediated p53 activation ( 22 ). Doxorubicin also causes double-stranded DNA breaks by inhibition of topoisomerase II. Therefore, the growth arrest and cell death caused by these two disparate stimuli may share common elements mediated by p53 activation. However, the ability of ZNF217 to protect against TRF1 induced cell killing suggests that ZNF217 can also attenuate an ATM/p53 independent pathway.
The mechanism by which ZNF217 expression promotes cell survival may involve phosphorylation of Akt, as high levels of endogenous ZNF217 were correlated with high levels of phosphorylated Akt. Moreover, artificially increasing ZNF217 protein levels increased Akt phosphorylation, whereas siRNA-mediated ZNF217 silencing resulted in decreased phosphorylation of Akt. Furthermore, there may also be a feedback between PI3K activity and ZNF217 expression. Upstream inhibition of Akt phosphorylation by blocking PI3K with Ly294002 reduced both phosphorylated Akt and ZNF217 protein in the MCF7 cell line, resulting in increased apoptosis in response to doxorubicin. The positive feedback loop between ZNF217 and the PI3K pathway could provide a self-reinforcing amplification of the effect of increased ZNF217 DNA copy number.
ZNF217 protein localizes to the nucleus ( 4 ) and co-immunoprecipitates with complexes containing the transcriptional corepressors CoREST and CtBP, histone deacetylases HDAC1 and 2 and histone methyltransferases G9a and Eu-HMTase1 ( 15 , 16 ). This suggests that ZNF217 may function as part of a transcriptional repressor complex and that aberrant transcriptional repression and recruitment of corepressors are mechanisms involved in ZNF217 -induced cancer associated phenotypes. The hypothesized target genes through which ZNF217 expression suppresses apoptosis have yet to be elucidated. The observation that ZNF217 may regulate phosphorylated Akt suggests that ZNF217 -regulated genes lie upstream of this important apoptosis regulator.
The current findings may have important clinical ramifications. We present data which demonstrate that aberrant ZNF217 expression promotes both survival and proliferation of cells in vitro by a mechanism that may occur both early in the neoplastic process when telomeres become limiting and then at a later stage by conferring resistance to double-strand breaks induced by chemotherapy. This hypothesis could explain why patients with tumors having amplification of 20q13.2 have such a poor prognosis ( 3 , 25 ). In addition, we provide evidence that the anti-apoptotic activity of ZNF217 may be mediated through activation of the PI3K pathway and Akt. Aberrant expression of ZNF217 may occur even in cells with normal ZNF217 gene copy number ( 4 , 10 ), possibly making cells less responsive to apoptotic signals. It will be important to determine the degree to which ZNF217 expression levels vary in human populations and whether there is a correlation between the level of expression and either a susceptibility to cancer or its clinical outcome. Our findings support a model that unifies clinical, genomic and cell biological data. Most importantly, our data suggest that abrogation of ZNF217 function in tumors with elevated ZNF217 protein levels may render these tumors more sensitive to existing anti-tumor therapies.
MATERIALS AND METHODS
Cells and cell culture
The HeLa, HBL100, MCF7, BT474 and U251 cell lines were obtained from American Type Culture Collection (ATCC) and cultured as suggested by the supplier. The 600MPE cell line was obtained and cultured as previously described ( 10 ). The SF188 cell line was obtained from the tissue bank at the Brain Tumor Research Center at the University of California, San Francisco. The U251+7.9 clone was produced by transferring normal human chromosome 7 into the U251 cell line (unpublished data).
Full-length ZNF217 cDNAs in the plasmid pEGFP-N1 (Clontech, Palo Alto, CA, USA) or pEGFP-N1 alone were transfected into cells using Fugene 6 (Roche, Indianapolis, IN, USA), according to the manufacturers' protocol. EGFP positive populations were sorted by flow cytometry and replated into medium containing 500 µg/ml Geneticin (G418; Gibco BRL) for continued culture. Experiments were performed with stably transfected cells in which the EGFP + percentage was greater than 80. For transient transfections of dominant negative TRF2 mutant (TRF2 ΔBΔM ) ( 21 ) and TRF1 ( 22 ), cells were plated at a density of 1.5×10 5 cells per well in six-well plates. On day 2 following plating, the cells were transfected with plasmids encoding TRF1 or TRF2 ΔBΔM , and after 24 h, the transfection reagent was removed. Apoptosis assays were performed 48 h later.
Measurement of cell death
At indicated times, adherent cells were harvested by trypsinization and combined with non-adherent cells in culture medium. After centrifugation, the cells were resuspended in phosphate buffered saline and filtered (70 µm nylon mesh) to remove aggregates. PI was added to a final concentration of 50 µg/ml before analysis by flow cytometry (FACSCalibur, Becton Dickinson, San Jose, CA, USA). Triplicate samples were assayed to determine percentage of live (PI − ) versus dead (PI + ) cells. For p ZNF217 –EGFP-N1 and pEGFP-N1 transfected cells, 30 000 events were first gated to exclude small debris, then analyzed for percentage of live (GFP + PI − ) versus dead GFP + (GFP + PI + ) cells.
Adherent and non-adherent cells were collected, washed twice with phosphate buffered saline, then resuspended in 100 µl of binding buffer (10 m m HEPES, 140 m m NaCl and 2.5 m m CaCl 2 , pH 7.4). Annexin-V antibody of 7.5 µl (Molecular Probes, Eugene, OR, USA) was added and the cells were incubated at room temperature for 15 min. After incubation, 400 µl of binding buffer was added and cell samples were analyzed by flow cytometry. For p ZNF217 –EGFP-N1 and pEGFP-N1 transfected cells, the events were gated to measure percentage of GFP + Annexin-V − versus GFP + Annexin-V + cells.
Commercial software (Qiagen; Valencia, CA, USA) was used to design four candidate siRNA molecules mapping outside ZNF217 zinc finger motifs and lacking detectable sequence similarity to other genes. The 19-nucleotide RNAs targeting ZNF217 corresponded to regions beginning 591- (1: 5′-CGAUCAACGAGGUCGUCCA-3′), 722- (2: 5′-ACUGCUUUCGGUACCAGCA-3′), 970- (3: 5′-CGACGAUUCGAGUUCCGAG-3′) and 2830-bp (4: 5′-GUACCAUAUGGUCAGAGGC-3′) downstream of the first nucleotide of the start codon. Synthesis, purification and pre-annealing of siRNAs were performed by the vendor. SiRNA was transfected into cell lines with Oligofectamine (Invitrogen, Carlsbad, CA, USA), according to the manufacturer's protocol. Maximal down-modulation of ZNF217 mRNA and protein was achieved with siRNA 1, and this siRNA was used for further studies. A scrambled duplex RNA (5′-AGUAUGGAUGGGGAGGCAC-3′) designed using Oligoengine software (Oligoengine, Seattle, WA, USA) and a commercial negative control siRNA (Qiagene Catalog no. 1022076, ACGUGACACGUUCGGAGAA) were used as negative controls. Control siRNA showed neither any silencing of ZNF217 gene expression nor effect on doxorubicin-mediated apoptosis (data not shown).
Growing cultures of MCF7 were treated with 10 µ m of the PI3K inhibitor Ly294002 (Calbiochem, La Jolla, CA, USA) for 24 h, after which cells were harvested for immunoblot analysis. For assays of cell death, cells were treated with the Ly294002, followed 4 h later by the addition of 100 ng/ml doxorubicin. After an additional 16 h, the apoptotic cells were stained with Annexin-V antibody and analyzed by flow cytometry, as described earlier.
Total cell lysates were made with lysis buffer (20 m m Tris–HCL, pH 7.5, 150 m m NaCl, 25 m m NaF, 1 m m EDTA, 1 m m EGTA, 1 m m DTT, 1 m m NaVO 4 , 0.5% NP-40, 0.2%SDS and 1× Protease Inhibitor Cocktail Set III (Calbiochem, La Jolla, CA, USA). Protein concentrations were determined. Twenty microgram aliquots were mixed with SDS–PAGE sample buffer and separated by gel electrophoresis. Proteins were transferred onto blotting grade PVDF membranes (Osmonics Inc., Minnetonka, MN, USA) using an XCell II Blot Module (Invitrogen, Carlsbad, CA, USA). ZNF217 , pAkt, total Akt and β-actin protein levels were detected by immunoblot analysis using a polyclonal antibody developed against a ZNF217 –GST fusion peptide, a monoclonal antibody against phosphorylated Akt (Ser-473) ( 26 ), a polyclonal antibody against total Akt (Cat. no. SC-8312; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and a monoclonal antibody against β-actin (Cat. no. A1978; Sigma, Saint Louis, MO, USA).
The significance of differences observed was calculated using Student's t -test. Values were judged to be significant when P ≤0.005.
Supplementary Material is available at HMG Online.
We thank J.W. Gray for critically reading the manuscript and Drs K.P. Lu and T. de Lange for providing the TRF1 and TRF2 ΔBΔM constructs. This work was supported by American Cancer Society Institutional Research Grant no. 97-150-01-IRG, an Avon Foundation Gift, NIH Grant R21 CA87522-01, UC-BCRP Grant 6JB0133, NS42927, CA85799 and Brain Tumor SPORE P50 CA 97527. G.B.M. was supported by P50 CA83639, PO1 CA64602 and PO1 CA099031.
Conflict of Interest statement . The authors have declared that no conflict of interest exists.
- positional cloning
- chemotherapy regimen
- 1-phosphatidylinositol 3-kinase
- breast neoplasms
- cell death
- cell nucleus
- cell survival
- cell transformation, neoplastic
- dna damage
- transcription factor
- breast cancer
- transcriptional repression
- epithelial cells
- rna, small interfering
- proto-oncogene proteins c-akt
- genome sequencing