Overexpression of Caspase-3s Splice Variant in Locally Advanced Breast Carcinoma Is Associated with Poor Response to Neoadjuvant Chemotherapy

Apoptosis is a genetically regulated form of programmed cell death that plays an important role in eliminating infected and/or damaged cells (1–4). Dysfunction of apoptotic pathways can contribute to both tumorigenesis and chemoresistance (5). One of the important components of the apoptotic machinery is a proteolytic system consisting of a family of cysteine proteases called caspases. Activated caspase-3 is considered to be a central protease in the execution of apoptosis (6, 7). Resistance to apoptotic stimuli was frequently reported in MCF-7 human breast cancer cells that lack expression of caspase-3 as a result of a 47 bp deletion in exon 3 of the CASP-3 gene (8–10). Caspase-3 is characterized by the presence of the conserved QACRG pentapeptide active site motif, specific to this family of structurally related proteases (11–13).

CASP-3 gene is located on chromosome 4q34 and possess 2,635 bp that encodes for seven exons resulting in a principal transcript with 834 bp size (14). Its alternative transcription gives rise to a shorter splice variant (caspase-3s) with 549 bp size resulting from a deletion of the exon 6 (15). This deletion results in an altered reading frame in the COOH terminus, leading to an altered amino acid sequence and a truncated polypeptide. Thus, the novel variant is 95-amino acid shorter but shares the same 161 residues with caspase-3 in the NH₂ terminus containing the prodomaine and the majority of the large subunit. However, the NH₂ terminus of the caspase-3s variant lacks the conserved QACRG sequence of caspase-3, suggesting that it might not be a functional protease. Indeed, it has shown that the caspase-3s variant may act as a dominant-negative regulator of the caspase-3 transcript (15).

Different experimental studies have shown in MCF-7 cells that caspase-3 is essential for breast tumor cells to undergo apoptosis in response to doxorubicin, etoposide, Taxol, cisplatin, and ionizing radiation, and that overexpression of caspase-3 restores sensitivity to these stimuli (10, 16–19). Cyclophosphamide was also shown to effect tumor cell death by stimulating apoptosis. Indeed, this anticancer alkylating agent prodrug, inactive until is metabolized by cytochrome P450 in the liver, acts via its activated 4-hydroxy metabolites and induces a mitochondrial caspase-9–dependent apoptotic pathway. This effect was determined in gliosarcoma cells as evidenced by the induction of DNA fragmentation and cleavage of the caspase-7 and caspase-3 substrates in drug-treated cells (20).

Several investigators revealed that expression of caspase-3 is up-regulated in breast carcinomas in parallel to the increase in the apoptotic index and tumor progression (21–24). These finding seem to conflict with the idea that apoptosis is generally reduced in tumor cells. Nevertheless, the expression of numerous genes involved in apoptosis regulation, such as CASP-3, CASP-2, Bcl-X, survivin, and death receptor families, is modulated by alternative splicing. These genes are transcribed as diverse mRNA species, which encode variants with sometimes opposite functions. This mechanism seems to play an important role in the regulation of programmed cell death, but recent evidences indicate that in several cancers the ratio of splice variants is dramatically altered and that deregulation of splicing favoring the expression of antiapoptotic variants contributes to tumorigenesis and chemoresistance (25, 26).

The present study was undertaken to determine the role of caspase-3 and its splice variant in breast cancer cells. For that, breast tumor cell lines deficient and proficient for CASP-3 were transfected with each transcript. In addition, their prognostic and predictive values were investigated in breast carcinomas.

Cell lines. Human breast cancer cell lines MCF-7, ZR-75-1, SK-BR-3, HBL-100, and MDA-MB-231 were purchased from the American Type Culture Collection (Manassas, VA) and UACC-3199 from the Arizona Cancer Centre Tissue Culture Shared Resource (Tucson, AZ). The cell lines were cultured according to the instructions from the manufacturer. Briefly, cells were routinely grown at 37°C with 5% CO₂ in RPMI medium supplemented with antibiotics and 10% fetal bovine serum (Invitrogen, Carlsbad, CA).

Patients and samples. We studied retrospectively a population of 129 patients with invasive ductal breast carcinomas, including 90 diagnosed with locally advanced tumors. One patient has bilateral lesion, and the analysis was done in both carcinomas. The surgically resected carcinomas were obtained during the period going from 1991 to 1997 at the Centre Georges François Leclerc (Dijon, France). Corresponded adjacent normal tissues were also obtained for 50 patients belonging to the locally advanced group. The median clinical follow-up was 11.5 years (range 7.5-15.3 years). The patients with locally advanced breast carcinomas received neoadjuvant chemotherapy (four or six courses each 21 days) with a regimen containing cyclophosphamide (500 g/m²), epirubicine (100 g/m²), and 5-fluorouracil (500 g/m²). In all cases, neither radiotherapy nor hormone therapy were applied before chemotherapy. However, adjuvant hormone therapy was administrated in 40 of the 90 locally advanced cases. This study was done with the approval of the local boards governing research on human subjects and all the examined patients are with well-known clinical history.

All tissue samples were frozen and stored in liquid nitrogen. Total RNA from a pool of four normal mammary tissues was purchased from Clontech (Palo Alto, CA) and was used as control.

Full-length cDNA synthesis and cloning. The full-length of caspase-3 and caspase-3s coding sequences were obtained using SuperScript One-Step long templates reverse transcription-PCR (Invitrogen) with 1.25 μg total RNA from the UACC3199 cell line (containing high levels of the two transcripts). Specific primers were used at 300 nmol/L (ACGCTAGATATCATGGAGAACACTGAAAACTCAGTG and GGCGGC GGTCTAGATTAGTGATAAAAATAGAGTTC). PCR program was done by one cycle at 45°C for 30 minutes, 94°C for 2 minutes followed by 35 cycles of 15 seconds at 94°C, 30 seconds at 50°C, 1 minute at 68°C, and one final cycle for 5 minutes at 72°C (Abi Prism 9700 thermocycler; Applied Biosystems, Foster City, CA).

Addition of 3′ adenines was realized for 15 minutes at 72°C by using Platinum Taq DNA polymerase. The inserts were cloned into pcDNA3.1/CT-GFP-TOPO and amplified in One Shot TOP10 Chemically Competent Escherichia coli with green fluorescent protein fusion TOPO TA expression kits (Invitrogen). The plasmids from a few randomly picked colonies were isolated. The orientations of the caspase-3 or caspase-3s fragments were tested by EcoRV and XbaI digestion. The products of digestion were separated on 2% agarose gel and revealed by SYBR green (Molecular Probes, Eugene, OR) staining on ChemiDoc XRS Analyser (Bio-Rad, Hercules, CA).

Stable transfection. For stable transfection, 5 × 10⁵ MCF-7 and HBL100 cells were grown in a medium without antibiotics in 12-well plates. Two days later, cells were transfected with 0.5 μg plasmid containing either caspase-3 or caspase-3s insert and with empty control vector. The stable transfections were done by LipofectAMINE 2000 (Invitrogen) according to the instructions of the manufacturer. The stable colonies were selected by 1,000 μg/mL geneticin. Each stable transfected clone was transiently transfected with the other variant as described. To ensure that the variants were sufficiently expressed, the transfection efficiency was controlled by quantitative real-time reverse transcription-PCR amplification. Relative change was calculated between control and transfected cells.

Measurement of apoptosis cell death by flow cytometry. Apoptosis was induced by cyclophosphamide in the two cell lines (MCF-7 and HBL100). Twenty-four hours after transfection, cells were cultured in MEM medium without antibiotics during 24 hours in the presence of 600 ng/mL cyclophosphamide for MCF-7 and 1.2 μg/mL for HBL100 cells. Assessment of apoptosis was accomplished by measuring the translocation of the membrane phosphatidylserine using the Annexin V-PE Apoptosis Detection kit I (BD PharMingen) according to instructions of the manufacturer. Briefly, after trypsinization and washing with PBS, cells were pelted 5 minutes at 1,500 × g and incubated 15 minutes with Annexin V-PE and 7-amino-actinomycin D (BD Biosciences, San Diego, CA). Cells (30,000) were counted by flow cytometry using Becton Dickinson LSRII and the experiments were done in triplicate in two different clones.

RNA extraction and cDNA synthesis. Total RNA was extracted with Trizol reagent (Invitrogen) and its quality was checked by 28S/18S ratio on agarose gel. One microgram of total RNA was reverse transcribed as described (27).

Quantitative real-time PCR amplification. Amplification was done in a total volume of 25 μL in the presence of 600 nmol/L of each primers, 200 nmol/L of each probes, 12.5 μL Universal Master Mix (Applied Biosystems), and 12 ng cDNA (or water as negative control). PCR was done with an initial denaturation step of 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. Sequences of primers and probes used are the following: 5′-CTGGACTGTGGCATTGAGACA-3′; 5′-AGTCGGCCTCCACTGGTATTT-3′ and 5′-TGGTGTTGATGATGACATGGCGTGTC-3′ for caspase-3 primers and probes, respectively, located in exon 6 and resulting in a 76 bp fragment; 5′-AGAAGTCTAACTGGAAAACCCAAACT-3′; 5′-CAAAGCGACTGGATGAACCA-3′; and 5′-ATTATTCAGGTTATTATTCTTGGCG-3′for caspase-3s primers and probes, respectively, located in exons 5 to 7 and resulting in a 95 bp fragment. Probes were labeled with FAM in 5′ and with TAMRA in 3′. All samples were amplified in duplicate and results were analyzed at the C_T level. Control 18S reactions (Applied Biosystems) were used to normalize ΔC_T values.

Sequencing of PCR products. The specificity of all the PCR amplifications was verified by sequencing of PCR products. Briefly, products were excited from 3% agarose gels and isolated by Geneclean Turbo gel extraction kit (Qiagen). The purified PCR products were sequenced using the Abi-Prism Big Dye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems) with the respective primers used in the initial PCR according to the protocol of the manufacturer. Sequence analysis was carried out using an ABI-Prism 310 as described (28).

Statistical analysis. All analyses were done with Stata software with a bilateral 5% type I error. Comparisons between cell lines were done with ANOVA test. Comparisons between transfected and vector control cells were made by Student’s t test. In breast carcinomas, analyses were realized among the 129 patients and among the 90 locally advanced patients. Qualitative variables were described with frequency and compared with Fisher's exact test. The quantitative variables were described with mean (SD) and with box plots. They were compared between subgroups with Kruskal-Wallis, Mann-Whitney, and/or two-sample Wilcoxon rank-sum tests. The longitudinal changes of caspases expression were tested by Wilcoxon matched pairs signed-rank tests. The overall survival was defined as the interval between the diagnosis and the last follow-up or death. Disease-free survival was defined as the time between the date of diagnosis and the date of distant metastases or local recurrence or death, whichever came first, or last follow-up. Univariate relative hazard ratio and 95% confidence interval were calculated by Cox proportional hazards model. Multivariate analyses were also done with Cox proportional hazards model. Univariate and multivariate logistic analyses (odds ratio and 95% confidence interval) were done to evaluate influence between caspases expression levels and pathologic response.

Coexpression of caspase-3 and caspase-3s decreased apoptotic induction in breast tumor cell lines. Breast cancer cell lines (ZR-75-1, SK-BR-3, HBL-100, UACC-3199, and MCF-7) were tested for the expression of the two caspase-3 transcripts and compared with normal mammary tissues. The results confirm that caspase-3 transcript was very weakly expressed in MCF-7 cells comparatively with normal mammary tissues (P < 0.0001, ANOVA test) and reveal that caspase-3s variant was absent. Mean expression levels of caspase-3 were variable among the remaining cell lines, whereas caspase-3s seem higher compared with normal mammary tissues, except for ZR-75-1 cells (Fig. 1A). Therefore, the ratio between caspase-3s and caspase-3 was significantly higher in these cell lines than in normal mammary tissues (P = 0.0370 for SK-BR-3, P = 0.0101 for HBL-100, and P = 0.0040 for UACC-3199; Fig. 1B).

To determine the effect of caspase-3 transcripts on induced apoptosis, we compared cyclophosphamide effects on caspase-3–deficient MCF-7 and caspase-3–proficient HBL100 cell lines transfected with full-length cDNA of each transcript. In a first set of experiments, we checked the expression of the two transcripts by quantitative reverse transcription-PCR. The results show that transfection restores the expression of both transcripts in MCF-7 cells, and provokes 3- to 7-fold increase of caspase-3 and 40- to 43-fold increase of caspase-3s expressions in HBL100 cells. Treatment of caspase-3–transfected cells with cyclophosphamide induces apoptosis 1.8-fold in MCF-7 (P = 0.0005, Student's t test) and 2.3-fold in HBL100 (P < 0.0001) compared with the control vector cells (Fig. 1C and D, respectively). Caspase-3s–transfected cells also has this ability, but compared with caspase-3 presence, its increased apoptosis after treatment remains less efficient, namely in HBL100 cells (P < 0.0001) that contain a normal CASP-3 gene. Interestingly, overexpression of caspase-3s in caspase-3–transfected MCF-7 and HBL100 cells reveals a strong decrease (P < 0.0001 for both cell lines) in induced apoptosis levels compared with matched caspase-3–transfected cells (Fig. 1C and D).

Expression of caspase-3 transcripts in breast carcinomas and adjacent normal tissues. Coexpression of caspase-3 and caspase-3s was detected in the 130 examined breast carcinomas and adjacent normal tissues. However, mean expression levels of caspase-3 was significantly (P = 0.0396, two-sample Wilcoxon rank-sum test) more important in carcinomas than in adjacent normal tissues (Fig. 2A), whereas no differences (P = 0.5287) was detected between the two tissues for caspase-3s (Fig. 2B).

Relationship with standard prognostic factors and patient survival outcome. The mean expression levels of the caspase-3 and its splice variant, caspase-3s, did not differ according to the patient characteristics. No relation was also found for the ratio caspase-3s/caspase-3 expression levels (Table 1).

Cox univariate analysis realized according to overall and disease-free survivals of the 130 patients showed no significant relation between either the two transcripts or their ratio expression levels and survival outcome. In multivariate analysis, only ages >50 years and positive nodal metastases are significantly associated with both poorest overall and disease-free survivals. Estrogen receptor status was also associated with poorest overall survival (Table 2). No relationship was also found among the 90 locally advanced patients (data not shown).

Relationship with response to chemotherapy. The pathologic response to neoadjuvant chemotherapy was classified for the 90 locally advanced breast carcinomas as described previously (29). No responders corresponded to patients with residual tumor evidently modified by treatment as well as to patients with histologically unmodified tumor. Responders included patients with no evidence of residual tumor in the breast or axillary lymph nodes and patients with only residual in situ carcinoma. Thus, no pathologic response was observed for 60 carcinomas and complete pathologic response for 30 cases. Caspase-3 mean expression increased from nonresponder (3.69 ± 0.86) to responder (5.30 ± 0.23) cases (P < 0.0001, Mann-Whitney test). In contrast, caspase-3s mean expression decreased from nonresponder (0.11 ± 0.09) to responder (0.02 ± 0.01) cases (P = 0.0248). These results suggested that carcinomas with high levels of caspase-3 and weak levels of caspase-3s could be more sensitive to treatment. As illustrated in Fig. 3, caspase-3s/caspase-3 ratio expression showed a significant chemoresistance-dependent increase (P = 0.01). Logistic univariate analysis realized according to pathologic response revealed that only caspase-3s (odds ratio, 0.16; 95% confidence interval, 0.03-0.66; P = 0.012) is associated with pathologic response, whereas no association was found for caspase-3 (odds ratio, 1.83; 95% confidence interval, 0.66-5.10; P = 0.245). According to logistic multivariate analysis, increased caspase-3s expression is significantly (P = 0.026) indicative of no response to chemotherapy (Table 3).

The coexpression of caspase-3 and its splice variant, caspase-3s, was reported in diverse tumor cell lines (kidney, HeLa, and colon) and normal human tissues (heart, brain placenta, and lung; ref. 15). Our results show, for the first time, the coexpression of the two transcripts in breast tumor cell lines as well as in both breast carcinomas and corresponding normal tissues.

In the first step of our analysis, we determined the role of each transcript expression as well as of their coexpression on induced apoptosis in tumor cell lines. The results indicate that the presence of caspase-3 significantly increase apoptosis in response to cyclophosphamide, one of the drugs used in breast carcinoma treatment, and that overexpression of caspase-3s variant in caspase-3–transfected cells strongly inhibits induced apoptosis.

In pretreatment of locally advanced carcinomas, basal expression levels of caspase-3 was significantly higher in carcinomas compared with normal tissues, whereas caspase-3s levels did not differ between the two tissues. Increased expression of caspase-3 transcript may indicate a potential for apoptosis in breast carcinomas. This result is consistent with the fact that up-regulation of caspase-3 and increase in apoptosis rates were detected simultaneously at high levels in breast carcinomas compared with adjacent normal tissues (21–24). Because the short splice variant acts as a dominant-negative regulator of the caspase-3 transcript (15), its expression in breast carcinomas permit the generation of antiapoptotic activities, thus explaining their ability to tolerate high levels of caspase-3. These results agree with the antagonistic actions between the two transcripts and suggest that their expression might be monitored by an inverse regulation.

Next, we investigated the prognostic effect of the two transcript expression levels and found no significant relationship with either patient's clinical variables or patient's overall and disease-free survival. This result is in contradiction with only one study showing that immunohistochemical expression of caspase-3 has a negative influence on breast cancer patient's overall survival (30). However, others have found no significant association with patient's clinical variables and survival outcome by immunohistochemistry, Western blot, or mRNA analysis (21, 23).

The second aim of our study was to investigate whether caspases-3 expression has predictive effect. At present, there was no study on the role of caspase-3 and its splice variant in the response to chemotherapy in breast carcinomas. Using immunohistochemistry, no relationship was found between caspase-3 active form expression and clinical response to chemotherapy (31). However, lack of or down-regulation of caspase-3 expression, both at transcriptional and protein levels, was reported to constitute a possible mechanism to chemoresistance in breast carcinomas (16). Thus, we hypothesize that overexpression of the caspase-3s antiapoptotic variant could promote cell survival and chemoresistance in breast tumor cells. Consistently, our results show that the achievement of a complete response seems to be dependent on the presence of caspase-3, but require a decrease in caspase-3s expression. In addition, logistic univariate and multivariate analysis show that no response to treatment is related to caspase-3s increased expression and not to decrease of caspase-3 one. Therefore, locally advanced breast cancer patients with high levels of caspase-3 expression and weak levels of caspase-3s expression could benefit from neoadjuvant cyclophosphamide-containing chemotherapy.

Although caspase-3s is expressed at low levels that are not sufficient to completely suppress apoptosis, imbalance of the two transcript ratios may render the cells resistant to apoptosis (15, 16). Consistent with that, increased caspase-3s/caspase-3 ratio expression was significantly associated with no response to chemotherapy in the examined population. Taken together, our results show that although caspase-3 expression has no prognostic effects in breast carcinomas, it might be of interest for chemotherapy response determination. Thus, expression of caspase-3s/caspase-3 ratio might be used as a predictive marker because their expression levels may define a subset of locally advanced breast cancer patients who are more likely to benefit from neoadjuvant cyclophosphamide-containing chemotherapy.