Persistent Activation of Stat3 Signaling Induces Survivin Gene Expression and Confers Resistance to Apoptosis in Human Breast Cancer Cells

Materials and Methods

Tumor specimens. Primary tumors were obtained at H. Lee Moffitt Cancer Center and Research Institute from patients with stage III breast carcinoma before neoadjuvant chemotherapy with sequential docetaxel and doxorubicin in a phase II clinical trial.⁶ To preserve the in vivo activated phosphorylation state of Stat3 proteins in tumor cells, specimens were fixed in formalin or snap frozen in liquid nitrogen within 15 to 20 minutes of surgical excision (see Diaz et al., in this issue). Tissue specimens were obtained with written patient consent in this Institutional Review Board–approved clinical trial.

Cell lines and transfections. MDA-MB-435s, MDA-MB-468, MDA-MB-231, MDA-MB-361, HEK-293, and NIH 3T3 cells were grown in DMEM supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin. MCF10A cells were cultured in 1:1 mixture of Ham's F12 medium/DMEM with 2.5 mmol/L l-glutamine and supplemented with 20 ng/mL epidermal growth factor, 100 ng/mL cholera toxin, 0.01 mg/mL insulin and 500 ng/mL hydrocortisone, 5% horse serum, 100 units/mL penicillin, and 100 μg/mL streptomycin. Transfections were carried out by the Lipofectamine-Plus method as described by the supplier (Life Technologies, Grand Island, NY).

Expression constructs, oligonucleotides, and IL-6 treatment. The reporter pGL2-Survivin encodes the Survivin gene promoter driving expression of firefly luciferase in pGL2 (Promega, Madison, WI). Expression vectors for v-Src (pMvSrc) and Stat3 (pVRStat3) have been previously described (20). The Stat3 antisense (5′-GCTCCAGCATCTGCTGCTTC-3′), control mismatch (5′-GCTCCAATACCCGTTGCTTC-3′) and Survivin antisense (5′-CCCAGCCTTCCAGCTCCTTG-3′) oligonucleotides were synthesized using phosphorothioate chemistry. To increase stability, oligonucleotides were synthesized with 2′-O-methoxyethyl modification of the five or three underlined terminal nucleotides (21). The final concentration for Stat3 antisense and control mismatch oligonucleotides was 250 nmol/L; for Survivin antisense oligonucleotides, the final concentration was 300 nmol/L. For IL-6 stimulation, cells were serum-starved (DMEM supplemented with 0.1% fetal bovine serum) for 18 hours before IL-6 treatment (20 or 30 ng/mL in DMEM) for 30 minutes [electrophoretic mobility shift assay (EMSA) analysis] or 48 hours (Western blot analysis). Nuclear extracts and cell lysates were prepared for EMSA and Western blot analysis, respectively, as described below.

Western blot analyses. Cells were lysed in a buffer containing 10 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1% NP40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mmol/L EDTA (pH 8.0), 2 mmol/L phenylmethylsulfonyl fluoride, 2 μg/mL aprotinin, 2 μg/mL leupeptin, and 1 mmol/L Na₃VO₄. For Western blot analyses, 30 μg of total extracted proteins were applied per lane before SDS-PAGE. Following transfer to nitrocellulose membranes, protein expression levels were detected using polyclonal anti-Stat3 (Santa Cruz Biotechnology, Santa Cruz, CA), anti-Survivin (Alpha Diagnostics International, San Antonio, TX), or anti-poly-(ADP-ribose) polymerase (Cell Signaling Technology, Beverly, MA) antibodies. The expression of β-actin (Sigma-Aldrich, St. Louis, MO) was used as a normalization control for protein loading.

Immunohistochemistry and image analysis. Details of the immunohistochemical analysis are described by Diaz et al., in this issue. Survivin monoclonal antibody and polyclonal pY-Stat3 antibody (Cell Signaling Technology) were used for immunohistochemistry of formalin-fixed, paraffin-embedded tissue sections. The Optimas 6.5 software was used to quantify protein expression as described by Diaz et al., in this issue. Results are reported as an index reflecting the intensity of the marker in relationship to the total pixel intensity of the region of interest. The index is an average of the values of three images.

Statistical analysis. Pathologic response was classified as complete pathologic response or partial pathologic response based on the size of residual tumor after treatment (complete pathologic response if 0 cm; partial pathologic response if > 0 cm). Correlations between biomarkers were assessed by Spearman's rank correlation coefficient for each group and tests of the correlation coefficient equal to zero were carried out. All tests were two sided and declared significant at the 5% level. SAS 8.2 (SAS Institute, Cary, NC) was used for statistical calculations.

Nuclear extract preparation and EMSA. Nuclear extracts were prepared as previously described (9) by high-salt extraction into 30 to 70 μL buffer [20 mmol/L HEPES (pH 7.9), 420 mmol/L NaCl, 1 mmol/L EDTA, 20% glycerol, 20 mmol/L NaF, 1 mmol/L Na₃VO₄, 1 mmol/L Na₄P₂O₇, 1 mmol/L DTT, 0.5 mmol/L phenylmethylsulfonyl fluoride, 0.1 μmol/L aprotinin, 1 μmol/L leupeptin, and 1 μmol/L antipain]. For EMSA, 5 μg of total nuclear protein were used for each lane. EMSA was done using a ³²P-labeled oligonucleotide probe containing a high-affinity sis-inducible element (hSIE, m67 variant) derived from the c-fos gene promoter (sense strand 5′-AGCTTCATTTCCCGTAAATCCCTA-3′) that binds activated Stat3 proteins (11). Stat3 protein was supershifted in the EMSA by preincubation with Stat3 antibody (C-20X, Santa Cruz Biotechnology). The oligonucleotides containing the putative Stat3-binding sites in the Survivin promoter used in EMSA are as follows (sense strand): (−1,184) site #1, 5′-TGGAGACTCAGTTTCAAATAAATAAATAAAC-3′; (−1,143) site #2, 5′-TGAGTTACTGTATTAAAGAATGGGGGCGGG-3′; and (−1,105) site #3, 5′-TGTGGGGAGAGGTTGCAAAAATAAATAAAT-3′ (the bolded and underlined oligonucleotides, TG, are sequences added to the 5′ end to create overhangs for radiolabeling by Klenow reaction and are not part of the Survivin promoter). Competition analysis to determine specificity of Stat3 binding to the Survivin promoter was done by preincubating unlabeled hSIE probe with radiolabeled Survivin probe in the EMSA. Following incubation of radiolabeled probes with nuclear extracts, protein-DNA complexes were resolved by nondenaturing PAGE and detected by autoradiography.

Cytosolic extract preparation and luciferase assays. Cytosolic extract preparation and luciferase assays were done as previously described (20). Briefly, cells were lysed in 0.1 mL of low-salt HEPES buffer [10 mmol/L HEPES (pH 7.8), 10 mmol/L KCl, 0.1 mmol/L EGTA, 0.1 mmol/L EDTA, 1 mmol/L phenylmethylsulfonyl fluoride, 1 mmol/L DTT, and 20 μL of 10% NP40]. After centrifugation (10,000 × g, 1 minute, 4°C), cytosolic supernatant was used for luciferase assays as described by the vendor (Promega). Experiments were done in triplicate and the average values were determined. To control for transfection efficiency, firefly luciferase values were normalized to the values for β-galactosidase.

In situ terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling and cellular proliferation assays. MDA-MB-435s and MDA-MB-231 cells were transfected with antisense or control mismatch oligonucleotides. After 48 hours, cells were labeled for apoptotic DNA strand breaks by terminal deoxyribonucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) reaction using an in situ cell death detection assay (Roche Applied Science, Indianapolis, IN) according to the instructions of the supplier. TUNEL-positive nuclei were counted and the apoptotic index was expressed as the number of apoptotic cells in one microscopic field. To determine cellular viability, cells were harvested by trypsinization and counted by trypan blue exclusion assay at 24 and 48 hours after transfection. All experiments were done in triplicate.

RNase protection assay. Total RNA was isolated from MDA-MB-435s cells using the RNAeasy mini kit (Qiagen, Valencia, CA). RNase protection assays were carried out with the Riboquant hStress-1 template set containing Bcl-xL and Mcl-1 probes or custom-made multiprobe templates containing Stat3 and Survivin probes (BD PharMingen, San Diego, CA). Briefly, the multiprobe templates were synthesized by in vitro transcription with incorporation of [³²P]dUTP and purified on Quick Spin RNA columns (Roche Applied Science). Labeled probe (1 × 10⁶ cpm) was hybridized with 10 μg of total RNA through a temperature gradient of 90°C to 56°C over a 16-hour period. Unprotected probe was removed by RNase digestion at 30°C for 1 hour followed by separation of protected RNA fragments on a 5% polyacrylamide-urea gel and detection using autoradiography.

Chromatin immunoprecipitation assay. Chromatin immunoprecipitation assays were done as previously described (22). Briefly, asynchronously growing HEK-293 cells were incubated with formaldehyde to cross-link protein-DNA complexes. The cross-linked chromatin was then extracted, diluted with lysis buffer, and sheared by sonication. After preclearing with 1:2 mix of protein A/protein G-agarose beads (Life Technologies), the chromatin was divided into equal samples for immunoprecipitation with either anti-Stat3 or anti–immunoglobulin G (negative control) polyclonal antibody (Santa Cruz Biotechnology). The immunoprecipitates were pelleted by centrifugation and incubated at 65°C to reverse the protein-DNA cross-linking. The DNA was extracted from the eluate by the phenol/chloroform method and then precipitated by ethanol. Purified DNA was subjected to PCR with primers specific for a region (−1,231 to −1,009) in the human Survivin promoter spanning three putative Stat3-binding sites. The sequences of the PCR primers used are as follows: Survivin forward primer, 5′-CAGTGAGCTGAGATCATGCC-3′; Survivin reverse primer, 5′-TATTAGCCCTCCAGCCCCAC-3′.

Microarray sample processing and data analysis. Five micrograms of total RNA collected from MDA-MB-435s cells treated with antisense or control mismatch oligonucleotides served as the mRNA sources for microarray analysis. The poly(A) mRNA was specifically converted to cDNA and then amplified and labeled with biotin following the procedure initially described by Van Gelder et al. (23). Hybridization with the biotin-labeled DNA, staining, and scanning of the microarray chips followed the prescribed procedure outlined in the Affymetrix technical manual and has been previously described (24).

The oligonucleotide probe arrays were Human Genome U133A chips (Affymetrix, Santa Clara, CA). Scanned output files were visually inspected for hybridization artifacts and then analyzed using Affymetrix Microarray MAS 5.0 software. The MAS 5.0 software identifies the increased and decreased genes between any two samples with a statistical algorithm that assesses the behavior of oligonucleotide probe sets designed to detect the same gene. Probe sets that yielded a change at P < 0.0045 were identified as changed (increased or decreased). In addition, the data were processed using robust multiarray analysis (25). Genes that were significantly changed in their expression were identified. Empirical estimates of the null distribution were determined using permutation analysis, thereby controlling the number of false positives. The significance analysis of microarrays (26) implements this approach to address the multiple testing problem and was also applied to the data analysis. Genes were considered changed if consistent behavior (increase or decrease) was observed in each of three replicate experiments based on analyses of data by the multiple methods described above.

Results

Direct blocking of Stat3 induces apoptosis in breast cancer cells. The human breast cancer cell line MDA-MB-435s, harboring activated Stat3, was transfected with Stat3 antisense or mismatch oligonucleotides using Lipofectamine-Plus. Twenty-four and 36 hours after transfection, the mRNA levels and DNA-binding activities of Stat3 were measured by RNase protection and electrophoretic mobility shift (EMSA) assays, respectively. Figure 1 shows that Stat3 antisense diminished Stat3 mRNA expression compared with the mismatch oligonucleotides (Fig. 1A), accompanied by a significant decrease in Stat3 DNA-binding activity (Fig. 1B). Incubation of cells with Stat3 antisense oligonucleotides for up to 48 hours resulted in a marked increase in vacuolated cells and cellular debris (Fig. 2A), indicative of apoptotic cell death. The occurrence of apoptosis was confirmed by in situ TUNEL assay (Fig. 2B and C) and by cleavage of poly-(ADP-ribose) polymerase most notably at 48 hours after treatment with antisense oligonucleotides (Fig. 2D). Figure 2E shows that treatment with Stat3 antisense oligonucleotide also induced significant growth inhibition. Both apoptosis and inhibition of cellular proliferation correlated with blockade of Stat3 expression and activation (Fig. 1 and data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Blocking of Stat3 expression and activity by antisense oligonucleotides in breast cancer cells. A, MDA-MB-435s cells were transfected with Stat3 antisense or mismatch oligonucleotides. Twenty-four and 36 hours after transfection, total RNA was isolated and analyzed by RNase protection assay with a Stat3-specific probe. B, EMSA analysis of Stat3-specific DNA-binding activity using the hSIE probe and nuclear extract preparations from MDA-MB-435s cells transfected with Stat3 antisense (lane 1) or mismatch (lane 2) oligonucleotides for 24 hours. Cells treated with transfection reagent (LP+, Lipofectamine-Plus) alone (lane 3) or nontreated cells (lane 4) were used as controls.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Blocking of Stat3 signaling leads to apoptosis in breast cancer cells. MDA-MB-435s cells were transfected with Stat3 antisense or control random oligonucleotides for 48 hours. A, cell morphology visualized by light microscopy. B, in situ TUNEL assay detected by light microscopy. C, number of apoptotic cells. TUNEL-positive nuclei were counted and the apoptotic rate was expressed as the number of apoptotic cells per microscopic field. Representative of three independent experiments. D, apoptosis indicated by poly-(ADP-ribose) polymerase (PARP) cleavage. Western blot analysis for poly-(ADP-ribose) polymerase (top) and Stat3 (bottom). Blots were normalized for total protein loading. E, number of viable cells. Cells were harvested by trypsinization and counted by trypan blue exclusion assay at 24 and 48 hours after transfection. Representative of three independent experiments.

Inhibition of Stat3 decreases expression of Survivin in breast cancer cells. We did microarray gene expression profiling analyses to assess the gene expression changes associated with blockade of Stat3 activity. RNA samples derived from MDA-MB-435s cells treated with Stat3 antisense or mismatch oligonucleotides for 24 hours were processed for microarray analysis using Affymetrix Human Genome U133A GeneChips. Experiments were done in triplicates and overlap among the three sets of deregulated genes was determined by using a variety of different microarray data analysis methods (see Materials and Methods). Whereas the biological functions of the genes identified were diverse (see below), we focused on apoptosis-related genes. Previous studies revealed that in some cell types, Stat3 activation prevents apoptosis by regulating expression of the Bcl-2 family of antiapoptotic proteins (26–31). However, microarray analysis did not reveal consistent decreases in the expression of any Bcl-2 family proteins in antisense oligonucleotide–treated breast cancer cells (data not shown). In contrast, expression of Survivin, which is a member of the IAP family of antiapoptotic genes, was found to be diminished by microarray analysis as confirmed below by independent molecular approaches. The microarray analysis revealed no change in other IAP family members including X-linked IAP, cellular IAP-1, and cellular IAP-2 (data not shown).

The microarray data were validated using an RNase protection assay. Results show a decrease in mRNA expression of Survivin, but not of Bcl-xL or Mcl-1, in antisense oligonucleotide–treated breast cancer cells (Fig. 3A). The correlation between Stat3 and Survivin protein expression was further confirmed by Western blot analysis, showing decreased Survivin expression in breast cancer cells that are treated with Stat3 antisense oligonucleotide (Fig. 3B). Because activated Stat3 signaling promotes the survival of breast tumor cells, we determined whether Survivin expression protects breast cancer cells from apoptosis. In MDA-MB-435s and MDA-MB-231 cells transfected with antisense oligonucleotides directed against Survivin, poly-(ADP-ribose) polymerase cleavage (Fig. 3C) and in situ TUNEL staining (Fig. 3D) were evident following inhibition of Survivin expression, indicative of apoptosis. Thus, expression of the antiapoptotic protein Survivin is associated with constitutive Stat3 activity and survival in breast cancer cells.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Expression analysis of Stat3 and antiapoptotic genes in breast cancer cells. A, RNase protection assay for Bcl-xL, Mcl-1, and Survivin in MDA-MB-435s cells transfected with Stat3 antisense or mismatch oligonucleotides for 24 hours. B, Western blot analysis for Survivin and Stat3 in MDA-MB-435s cells transfected with Stat3 antisense or control random oligonucleotides for 48 hours. C, apoptosis indicated by poly-(ADP-ribose) polymerase cleavage. Western blot analysis for poly-(ADP-ribose) polymerase and Survivin in MDA-MB-435s cells transfected with Survivin antisense or control random oligonucleotides for 24 hours. D, apoptosis detected by in situ TUNEL assay in MDA-MB-435s and MDA-MB-231 cells transfected with Survivin antisense or control random oligonucleotides for 48 hours.

Stat3 directly binds to and regulates the Survivin promoter. We examined whether Stat3 regulates Survivin promoter activity. In transient transfection studies with a luciferase reporter gene driven by the human Survivin promoter, we show that cotransfection with a v-Src vector that activates endogenous cellular Stat3 induces expression of the Survivin reporter by 2-fold (Fig. 4A). Moreover, cotransfection with both v-Src and full-length Stat3 vectors further induced the Survivin reporter expression up to nearly 5-fold. By contrast, ectopic expression of the dominant-negative Stat3 variant, Stat3β, decreased basal levels of Survivin reporter expression by 50% (Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Stat3 binds and regulates the Survivin promoter. A, NIH 3T3 cells were transfected with pGL2-Survivin promoter reporter alone or together with v-Src, Stat3, or Stat3β expression constructs. Columns, mean luciferase activity values from three independent experiments; bars, SD. B, chromatin immunoprecipitation assay was done using anti-Stat3 antibody (lane 2) or irrelevant anti–immunoglobulin G (IgG) antibody as a negative control (lane 3). PCR primers for the Survivin gene promoter were used to detect promoter fragment in immunoprecipitates. Input lane, total genomic DNA used as control for the PCR reaction. C, EMSA analysis of Stat3 DNA-binding activity in MDA-MB-435s cells using radiolabeled oligonucleotide probes containing the putative Stat3-binding sites 1, 2, and 3 in the Survivin promoter (lanes 1-3). Competition of Survivin probe #1 with unlabeled hSIE probe (lane 6). Lanes 4 and 5, positive controls for Stat3 binding using the radiolabeled hSIE probe and the Survivin probe #1, respectively. D, schematic of the Survivin promoter region with positions of the three putative Stat3-binding sites.

A search for potential Stat3-binding sites within the Survivin promoter region (18) revealed five candidates with the consensus sequences TT(N₄)AA and TT(N₅)AA (28). To determine whether Stat3 could bind the Survivin promoter under physiologic conditions in intact cells, we did chromatin immunoprecipitation assays using three sets of primers that cover the five candidate Stat3-binding sites. Primers to the region of −1,231 to −1,009 upstream from the ATG translation initiation site yielded Survivin promoter DNA in chromatin immunoprecipitated with an anti-Stat3 antibody (Fig. 4B). This region contains three potential Stat3 binding sites (Fig. 4D). By contrast, primers to the regions of −358/−148 and −938/−759 in the Survivin promoter did not detect promoter DNA in the anti-Stat3 immunoprecipitates (data not shown).

EMSA was done to determine binding of Stat3 to the same Survivin promoter region in vitro. Results show that endogenous activated Stat3 protein, present in nuclear extracts of MDA-MB-435s breast cancer cells, binds to the Survivin promoter fragments −1,174/−1,166 (site #1) and −1,095/−1,087 (site #3) but not to fragments −1,133/−1,126 (site #2), −851/−844 (site #4), and −264/−256 (site #5; Fig. 4C and data not shown). Specificity of Stat3 binding to the Survivin promoter fragments #1 and #3 was shown by competition analysis with unlabeled hSIE probe (Fig. 4C and data not shown). Both site #1 and site #3 are located within the −1,231 to −1,009 region that was detected in the chromatin immunoprecipitation assays above with anti-Stat3 antibody, suggesting it is this region of the Survivin promoter that accounts for Stat3 binding. Taken together, these data provide evidence that Stat3 directly binds the Survivin promoter and induces its expression.

Stat3 activation correlates with expression of Survivin in breast cancer cell lines. A unique feature of Survivin is its differential expression in tumor versus normal tissues (17–19) and overexpression of Survivin has been found in many cancers, including breast cancer (17). We investigated whether Stat3 activation correlates with Survivin up-regulation in a panel of human breast cancer cells harboring constitutively active Stat3. Western blot analysis shows Survivin protein expression in all tested breast cancer cell lines with activated Stat3 (MDA-MB-231, MDA-MB-435s, and MDA-MB-468; Fig. 5A). By contrast, minimal Survivin expression was observed in breast cancer cells (MDA-MB-361) and in normal breast epithelial cells (MCF-10A) lacking detectable Stat3 activation.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Stat3 activation and Survivin protein expression in various human breast cell lines. A, EMSA analysis of Stat3 DNA-binding activity with the hSIE probe (top) and Western blot analysis of Survivin expression (middle) in breast cancer (MDA-MB-231, MDA-MB-435s, MDA-MB-468, and MDA-MB-361) or normal breast epithelial (MCF10A) cells. Bottom, β-actin represents loading control for Western blot. B, EMSA analysis of IL-6 induced Stat3 DNA-binding activity with the hSIE probe in MDA-MB-435s cells. Lane 1, serum-starved cells; lane 2, IL-6-treated cells (20 ng/mL for 30 minutes); lane 3, supershift control with Stat3 antibody. C, Western blot analysis of phospho-Stat3 (p-Stat3) and total Stat3 levels (top) as well as Survivin and β-actin expression levels (bottom) after IL-6 treatment in MDA-MB-435s cells.

We investigated whether cytokine-induced Stat3 activity correlates with increased Survivin expression in breast cancer cells. IL-6 treatment of serum-starved MDA-MB-435s cells increases Stat3 DNA-binding activity within 30 minutes as detected by EMSA (Fig. 5B). In addition, both phospho-Stat3 levels and Survivin protein expression were induced by IL-6 treatment of MDA-MB-435s cells after 48 hours (Fig. 5C). These findings indicate that constitutive and cytokine-induced Stat3 activation correlates with Survivin expression in breast cancer cell lines.

Stat3 activation and survivin expression in clinical breast tumor specimens. Based on the above results, we explored the association between constitutive Stat3 activation and Survivin expression in primary breast tumors. Our analysis involved 45 tumor specimens and matched nonneoplastic tissues, which are all from patients with invasive breast carcinoma, obtained before initiation of chemotherapy treatment in a phase II clinical trial with sequential doxorubicin and docetaxel (see Diaz et al., in this issue). Tissue specimens were analyzed by immunohistochemical staining of formalin-fixed, paraffin-embedded sections using phospho-Stat3 or Survivin antibodies. We observed moderate to strong predominantly nuclear staining for phospho-Stat3 and Survivin in a majority of the tumor specimens but not in normal breast epithelial cells (Fig. 6). Importantly, a statistically significant positive correlation (P = 0.001) was observed between elevated phospho-Stat3 levels and Survivin expression in 33 of the 45 breast cancer patients who displayed a partial pathologic response to this neoadjuvant chemotherapy regimen (see Diaz et al., in this issue). Thus, high levels of phospho-Stat3 and Survivin expression correlate with invasive breast cancer and resistance to chemotherapy.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Evaluation of phospho-Stat3 and Survivin expression in primary breast tumors. Immunohistochemical detection shows elevated phospho-Stat3 and Survivin antibody immunoreactivity (specific nuclear localization) in breast carcinoma compared with matched nontumor breast tissues from the same patient. Representative samples (from a total of 45 matched pairs) photographed at ×400 optical magnification.

Additional candidate Stat3-regulated genes in breast cancer cells. To increase the likelihood of identifying additional Stat3-regulated genes, we focused on reproducibility of microarray data and intersections between different statistical methods of analysis rather than on fold levels of changes. Microarray samples were obtained from three independent experiments and data were analyzed using standard Affymetrix Microarray MAS 5.0 software, as well as robust multiarray analysis combined with significance analysis of microarrays. The significance analysis of microarrays identifies the most statistically significant, differentially expressed genes between two groups of samples. Potentially interesting genes that were most consistently changed on inhibition of Stat3 signaling in MDA-MB-435s breast cancer cells (data not shown) include TACC3, SMARCA4, CDC37, and ZNF148 (candidate Stat3 up-regulated genes) and Caspase 4, Histone H2B, RNF6, and GADD45A (candidate Stat3 down-regulated genes).

Discussion

We have previously shown that inhibition of activated Stat3 signaling in breast cancer cells by Src and Janus kinase pharmacologic inhibitors induces apoptosis (11). Here, we confirm the critical role of activated Stat3 in breast tumor cell survival by directly blocking its function using antisense Stat3 oligonucleotides. These findings are consistent with earlier studies showing induction of apoptosis following blockade of Stat3 signaling in human head and neck squamous cell carcinoma, multiple myeloma, leukemic large granular lymphocytes, prostate cancer, melanoma, lung cancer, and lymphoma cells (14–16, 29, 32–36). In the present study, we investigated candidate Stat3-regulated genes involved in preventing apoptosis of breast cancer cells. Among the genes of which expression changed on blocking the Stat3 pathway is Survivin, a member of the IAP family of antiapoptotic proteins. Survivin is strongly expressed in embryonic and fetal organs (17–19) but undetectable in most terminally differentiated normal tissues (17, 19, 37). By contrast, dramatic overexpression of Survivin has been shown in many tumors, including breast cancer (19, 38). Our results are consistent with previous findings linking Stat3 activation and Survivin expression in primary effusion lymphoma and gastric cancer cells (36, 39), and provide new evidence that Stat3 directly regulates the Survivin gene promoter.

The present studies establish a correlation between Stat3 activation and Survivin expression in a set of 45 primary tumor specimens from patients with invasive breast carcinoma (see Diaz et al., in this issue). Notably, elevated Stat3 activity is significantly correlated (P = 0.001) with increased Survivin expression in 33 of the 45 breast cancer patients who exhibited a partial response to sequential neoadjuvant therapy with doxorubicin and docetaxel.⁶ Previous studies have indicated that Survivin expression correlates with taxol resistance in ovarian cancer (,40, 41), providing further support for the potential of Stat3-mediated Survivin expression to modulate response to chemotherapy in breast cancer. Moreover, studies suggest that Survivin directly interacts with the Smac/DIABLO protein in modulating apoptotic responses to taxol (40, 41). Our findings presented here indicate that activated Stat3 signaling in breast cancer cells induces Survivin expression and thereby prevents apoptosis, which potentially contributes to resistance to chemotherapy.

Several lines of evidence suggest that deregulation of Survivin expression occurs in cancer as a result of genetic (amplification of the Survivin locus on 17q25 in neuroblastoma), epigenetic (selective demethylation of Survivin exon 1 in ovarian cancer but not in normal ovaries), transcriptional (transcriptional repression by wild-type p53), and posttranslational (increased protein stability by phosphorylation on Thr34 of Survivin protein) molecular mechanisms (30, 31, 42–44). Whereas association of Survivin expression with Stat3 signaling has been previously reported (36, 39, 45, 46), our data suggest a new mechanism for overexpression of Survivin that involves direct regulation of the Survivin gene promoter by persistently activated Stat3 protein in tumor cells. This mechanism may be prevalent in other human cancers that exhibit both persistent Stat3 activation and overexpression of Survivin.

Additional candidate Stat3-regulated genes were identified by our microarray analysis.⁷ Genes of which expression was repeatedly down-regulated by Stat3 antisense treatment (potential “Stat3 up-regulated” genes) encode proteins with a variety of biological functions, including regulation of cell growth and differentiation (TACC3) and signal transduction (SMARCA4, CDC37, and ZNF148). One of the down-regulated genes was Stat3, as expected because this was the target of the antisense oligonucleotides. Genes of which expression increased on blocking Stat3 (potential “Stat3 down-regulated” genes) include Caspase 4, a member of the histone H2B family, and the potential tumor suppressors ring-finger protein (RNF6) and growth arrest/DNA-damage-inducible protein (GADD45A). However, the regulation of these genes by Stat3 remains to be confirmed by independent means. To our knowledge, the genes identified here have not been previously considered as potential targets for regulation by Stat3, and thus provide new directions for future study.

The differential expression of Survivin in tumors compared with normal cells and its requirement for cancer cell survival identify it as a potential marker in cancer diagnosis as well as an attractive therapeutic target (19). The finding that Stat3 induces Survivin expression provides a potential new therapeutic approach based on inhibition of Stat3 signaling and, consequently, Survivin expression. In this regard, we and others are currently developing novel Stat3 inhibitors for cancer therapy (47–50). Our results presented here encourage further exploration of the potential therapeutic value of targeting Survivin through inhibition of upstream Stat3 signaling. This approach may have dual benefits for cancer patients by directly inducing tumor cell apoptosis and conferring sensitivity to conventional chemotherapy agents.