Loss of Retinoblastoma Tumor Suppressor Protein Makes Human Breast Cancer Cells More Sensitive to Antimetabolite Exposure

The retinoblastoma tumor-suppressor protein, pRB, plays a key role in mechanisms controlling cell cycle progression (1). pRB is likely to be the major factor controlling the transition from the G₁ to the S phase. pRB interacts with a family of transcriptional regulators termed the E2Fs, which control the expression of those genes whose products are important for entry and passage throughout the S phase. In its hypophosphorylated form, pRB is bound to E2Fs, thus preventing them from activating the E2F target genes (2, 3). In its hyperphosphorylated form, pRB leaves the E2Fs free to activate the target genes involved in the synthesis of DNA such as thymidylate synthase, dihydrofolate reductase, thymidine kinase, ribonucleotide reductase, and DNA polymerase α (4, 5). Phosphorylation of pRB is triggered in the early G₁ phase by the cyclin D–cyclin-dependent protein kinase (CDK)-4 and CDK-6 complexes and is completed, at the end of the G₁ phase, by cyclin E–CDK-2 complexes. The activities of the CDKs are in turn constrained by the CDK inhibitors: CDK-4 and CDK-6 are inhibited mainly by p16INK4a, whereas CDK-2 is negatively regulated by p21Cip1 and p27 (6). The components of the regulatory machinery that controls G₁-S phase transition behave as tumor suppressors or proto-oncogenes and are frequently altered in cancer cells. RB1 mutation or deletion, INK4a mutation, deletion or gene silencing, and cyclin D1 or CDK4 overexpression characterize many human cancers (7). These changes, causing either RB1 loss or pRB hyperphosphorylation, render the major control mechanism of the G₁-S phase checkpoint out of order. Inactivation of the RB tumor-suppressor pathways is associated with tumorigenesis and characterizes a large fraction of many types of cancers (8, 9). These cancers, from the clinical point of view, are generally more aggressive than those with a normally functioning RB pathway (10). This greater aggressiveness might be explained by a series of biological characteristics of these types of cancers. RB inactivation has been shown to cause chromosome instability by compromising the accuracy of the mitosis with consequent genetic changes facilitating tumor progression (11). The absence or functional inactivation of RB was found to be associated with an up-regulation of ribosome biogenesis (12–14), which enhances the cell proliferation rate (15, 16). The pRB status has been shown to influence the response to tamoxifen and DNA-damaging agent exposure in human breast cancer cell line and xenograft models. Functional loss of RB increased the cell sensitivity to DNA-damaging agents but not to tamoxifen treatment, which resulted in continued proliferation and xenograft tumor growth (17). Furthermore, a greater sensitivity to drugs targeting the thymidylate biosynthesis pathway, such as 5-fluorouracil (5-FU) and methotrexate, has been shown in mouse embryonic fibroblasts with loss of functional RB protein (18).

The latter observations may have an obvious clinical importance because hormonal therapy and chemotherapy are currently used in many cancer treatment protocols, and predicting the efficacy of therapy may determine the choice of more appropriate therapy strategies and improve patients' clinical outcome. To gain further information on the importance of the RB status in the response of cancer cells to drugs that target the thymidylate biosynthesis pathway, we carried out a prospective study on a large series of primary invasive breast cancers in patients who received standard adjuvant systemic chemotherapy including 5-FU and methotrexate and whose pRB expression was defined by immunohistochemistry. The predictive value of pRB expression was also investigated in the whole patient series and in patients who exclusively received endocrine therapy. We observed that patients with cancers characterized by the absence of pRB expression and who received systemic chemotherapy had a better disease-free survival (DFS) than those whose cancers showed pRB positivity. With a view to explaining the different clinical behavior according to the status of the tumor suppressor protein, we investigated whether a similar different sensitivity might characterize pRB-deficient and pRB-proficient cancer cells exposed to 5-FU and methotrexate treatment. For this purpose, we studied the effect of 5-FU and methotrexate treatment on cell cycle progression and cell population growth of the human cancer cell lines, the MCF-7 and HCT-116 cells, silenced for RB1 by specific RNA interference. We found that 5-FU and methotrexate, used at doses and time exposures derived from the evaluation of the interstitial pharmacokinetics of the drugs in human breast cancers in vivo (19), strongly reduced the growth rate of RB1 silenced but not of control MCF-7 and HCT-116 cells.

Patients. We studied a total of 518 consecutive patients who underwent surgical resection for primary invasive breast carcinoma at the Department of Surgery, University of Bologna, between 1991 and 1995. Patients' age ranged from 25 to 89 years, with an average (±SD) of 60 (±12.9) years (median value, 61 years). Tumors were histologically classified and staged according to the WHO and the Unio Internationale Contra Cancrum tumor-node-metastasis systems, respectively. Histologic grading (G) was done in ductal carcinomas according to Elston and Ellis (20). Due to patient age, axillary dissection was not done in 7 patients (1.3%): In the remaining 511 cases, axillary lymph node metastases were reported as absent (N₀) or present (N₊). Estrogen receptor (ER) and progesterone receptor (PR) status; Ki67 antigen expression; and p53, HER2, and RB status were assessed on histologic sections by standard immunohistochemistry, as reported below. All immunohistochemical analyses were done at the time of diagnosis. Patients were then regularly followed up every 6 mo for a median observation time of 109 mo (range 4-142 mo). The main clinical and histopathologic characteristics of the studied population are summarized in Table 1.

The present study was approved by the Senior Staff Committee, the board, which, at the time of patient enrollment, regulated noninterventional studies and was comparable with an institutional review board.

Adjuvant treatments. Three hundred and forty-two patients underwent mastectomy and 176 patients underwent conservative breast surgery. One hundred and forty-five received six cycles of the cyclophosphamide, methotrexate, and 5-FU chemotherapy regimen that was given on days 1 and 8 of each treatment cycle. The dose of cyclophosphamide and fluorouracil was 600 mg/m² of body surface area, and the dose of methotrexate was 40 mg/m². Each of the three drugs was repeated every 28 d. Two hundred and thirty-one patients who did not receive systemic chemotherapy received adjuvant endocrine therapy alone (tamoxifen, 20 mg daily, for at least 2 y). Forty-nine patients received radiotherapy only and 93 patients did not receive any kind of adjuvant therapy.

Immunohistochemical assessment. From each case, one block of formalin-fixed, paraffin-embedded tissue was selected, including a representative tumor area. Four-micrometer-thin serial sections were cut, collected on 3-ethoxy-aminoethyl-silane–treated slides, and allowed to dry overnight at 37°C. Sections were then processed for immunohistochemistry according to a streptavidin-biotin-peroxidase complex protocol combined with microwave-based antigen retrieval pretreatment in citrate buffer solution (pH 6.0), and highlighted using a peroxidase/3,3′-diaminobenzidine enzymatic reaction. pRB immunostaining was assessed using two different monoclonal antibodies (mAb): clone G3-245 (BioGenex Laboratories), which specifically recognizes the phosphorylated pRb form (ppRB), and clone 1F8/Rb1 (Neomarkers, Lab Vision), which identifies all forms of pRB (phosphorylated as well as unphosphorylated or underphosphorylated). The following mAbs were also used: anti-p53 mAb (clone BP53-12.1), anti-Ki67 (clone MIB-1), anti-HER2 internal domain (clone CB11), anti-ER (clone 1D5), and anti-progesterone receptor (anti-PR; clone 1A6), all from BioGenex Laboratories. All mAbs were applied overnight at room temperature at the predetermined optimal concentrations. Negative controls were stained along with test sections; control sections received the same treatment, except for application of the primary antibody.

Nuclear immunostaining of ER, PR, Ki67, p53, and pRB was assessed by image cytometry using the Cytometrica program (C&V), as previously detailed (21); it was expressed as the percentage of labeled nuclear area over the total neoplastic nuclear area in the section [labeling index (LI)]. For each case, at least 2,000 cells were evaluated.

Cells and drug treatment. The human breast cancer cell line MCF-7 (American Type Culture Collection) was maintained in RPMI 1640 (Euroclone) supplemented with 10% fetal bovine serum (Euroclone) at 37°C with 5% CO₂. The human colon cancer cell line HCT-116 (American Type Culture Collection) was maintained in DMEM (high glucose) supplemented with 10% fetal bovine serum (Euroclone). 5-FU (Fluorouracile Teva Pharma B.V. Olanda) was used at 10 μg/mL and methotrexate (Metotrexato Mayne-Mayne Pharma) was used at 0.05 μg/mL; both drugs were diluted directly from stock solutions and mixed in RPMI and in DMEM with 10% fetal bovine serum. Cells were left for 1 h at 37°C in the presence of the drug cocktail, then washed extensively with warm PBS and fed with fresh medium.

5-FU and methotrexate concentrations and time administration were chosen according to Muller et al. (19).

RB1 silencing. Asynchronously growing MCF-7 and HCT-116 cells were subjected to RNA interference to silence RB1 expression. Cells were transfected with RB1 siRNAs (Stealth RNA Select-Invitrogen) and scrambled sequences (Invitrogen) were used as a negative control. Transfections were done using Lipofectamine 2000 following procedures recommended by the manufacturer (Invitrogen) at a concentration of 40 nmol/L for RB1 and relative scrambled control sequences. The concentration of RB1 siRNAs used resulted to be lowest one capable to reduce the RB1 mRNA level to at least the 80% of control for a duration of 120 h.

Evaluation of RB1 silencing by real-time reverse transcription-PCR analysis. Total RNA was extracted from cells 48 and 120 h after siRNA transfection using Trizol reagent (Invitrogen). For each sample, 2 μg of total RNA were reverse-transcribed using a High-Capacity cDNA Archive Kit (Applied Biosystems), following the manufacturer's instructions. The cDNA was subjected to real-time PCR analysis in a Gene Amp 7000 Sequence Detection System (Applied Biosystems) using the TaqMan approach. Cycling conditions were as follows: 50°C for 2 min, 95°C for 10 min, 45 cycles at 95°C for 15 s, and 60°C for 1 min. For each sample, three replicates were analyzed. Sets of primers and fluorogenic probes specific for RB1 mRNA were purchased from Applied Biosystems (Assay on Demand). The relative amount of the target gene in the cells transfected with the specific siRNAs compared with that of scrambled sequences of transfected cells was evaluated by the ΔΔC_t method (22), using human β-glucuronidase as an endogenous control (Applied Biosystems).

Immunocytochemical and Western blot analysis of pRB expression. The expression of pRB was also evaluated by immunocytochemical and Western blot analysis in cells 48 h after transfection. For immunostaining, cells were first washed in PBS, fixed, and permeabilized for 4 min with 2% paraformaldehyde added with 1% Triton X-100 diluted in PBS, and incubated with anti-pRB antibody (clone 1F8; Lab Vision Corporation) diluted 1:200 in PBS containing 1% bovine serum albumin and processed according to a streptavidin-biotin-peroxidase complex protocol. The streptavidin-peroxidase complex was visualized using diaminobenzidine. For Western blot analysis, proteins were extracted in 0.1 mol/L KH₂PO₄ (pH 7.5); 1% Igepal was added with a complete protease inhibitor cocktail (Roche Diagnostics) and 0.1 mmol/L β-glycerophosphate lysis buffer. After 30 min, insoluble fractions were eliminated by centrifugation at 13,000 × g for 30 min and supernatants were kept for analysis. All steps were done at 4°C. Twenty micrograms of proteins in Laemmli buffer were loaded in each lane. Protein samples were electrophoresed in 10% or 12% SDS-polyacrylamide gels and electrotransferred to cellulose nitrate membranes (Hybond C Extra, Amersham). Filters were then saturated with 5% nonfat dry milk in TBS [20 mmol/L Tris-HCl, 137 mmol/L NaCl (pH 7.6)]–0.1% Tween 20 (Sigma; TBS-T) for 1 h at room temperature. Nitrocellulose membranes were then rinsed in TBS-T and incubated overnight at 4°C with primary antibodies in 3.5% bovine serum albumin TBS-T. The following primary mouse monoclonal antibodies were used: anti-pRB diluted 1:200 (clone 1F8; Lab Vision Corporation) and anti–β-actin (Sigma Chemical Company) diluted 1:4,000. Membranes were washed 1× 10 min and 2× 5 min in TBS-T to remove unbound antibody, and were incubated for 1 h in the presence of horseradish peroxidase–labeled secondary antibody (dilution 1:10,000 in 5% milk TBS-T). After several washings, the horseradish peroxidase activity was detected using an enhanced chemiluminescence kit and was revealed on Hyperfilm enhanced chemiluminescence films (Amersham). The intensity of the bands was evaluated with the densitometric software GelPro analyzer 3.0 (Media Cybernetics).

Evaluation of cell population growth. MCF-7 and HCT-116 cells silenced for RB1 and transfected with scrambled sequences were used 48 h after the end of the transfection procedure. Cells were treated with 5-FU and methotrexate for 1 h. Treated and untreated cells were formalin fixed for the growth assay 72 h after the end of drug treatment. Quantitative evaluation was carried out in triplicate, using the crystal violet assay (23). After formalin fixation, cells were washed with distilled water, then stained with 0.1% crystal violet in 20% methanol for 30 min at room temperature. After extensive washing, crystal violet was resolubilized in 10% acetic acid for 15 min at room temperature and quantified spectrophotometrically at 595 nm as a relative measure of the cell number.

Evaluation of cell death rate. MCF-7 and HCT-116 cells either silenced for RB1 or transfected with scrambled sequences were treated with 5-FU and methotrexate for 1 h. Twenty-four hours after the end of drug treatment, the floating cells in the medium of each flask were transferred to centrifuge tubes. After detachment of the adherent cells with trypsin, the cells were mixed with the corresponding floating cells before centrifugation. The cells were then stained with 0.4% trypan blue, and the number of trypan blue–positive and trypan blue–negative cells were counted on a hemocytometer by light microscopy. The experiments were carried out in triplicate.

Cell cycle progression analysis by dual-parameter flow cytometry. To define the effect of 5-FU and methotrexate treatment on cell cycle progression, the MCF-7 cell line was used. Dual-parameter flow cytometry for the simultaneous evaluation of DNA content and bromodeoxyuridine (BrdUrd) incorporation was done.

Asynchronously growing MCF-7 cells were either silenced for RB1 expression or transfected with scrambled sequences. Seventy-two hours after the end of silencing procedure, BrdUrd was added at a final concentration of 20 μmol/L for 1 h, and then removed and fresh medium was added. Twelve hours later, cells were treated with 5-FU and methotrexate at doses of 10 and 0.05 μg/mL for 1 h. Cells were harvested 12 and 24 h later. Untreated cells were used as control. Cells were collected by centrifugation and fixed in 70% alcohol. Dual-parameter flow cytometry was done by a direct labeling of incorporated BrdUrd by FITC monoclonal antibody followed by propidium iodide–DNA counterstaining (24). Cytofluorimetric analyses were carried out in triplicate. Measurements were done by means of a Partec PAS II flow cytometer equipped with dual excitation system (argon ion laser and HBO 100 W arc lamp). The 488-nm blue line of the laser has been used to excite propidium iodide intercalated into the DNA and the FITC bound to BrdUrd. A preliminary instrument alignment and control has always been set up (with rat thymocytes stained with propidium iodide) to assure best instrumental analytic performances. Immediately before measurement, each sample has been filtered by “Filcons” 100 (ConsulTS) to remove cell clusters. For a sample measurement, a minimum of 20,000 events was acquired. The green (BrdUrd-FITC) and red (DNA-propidium iodide) fluorescence emission bands were collected, converted, and stored as DNA distribution values (histogram) or dual-parameter correlated dot plots by means of a dedicated computer integrated into the instrument. Data were elaborated and plotted thanks to the “Flow Max” software installed in the computer. Cell cycle analyses and the relative statistical data (coefficient of variation of the DNA distributions) were done by means of a dedicated software.

Effect of drug treatment on p53 activation and DNA double-strand breaks accumulation. MCF-7 cells silenced for RB1 and transfected with scrambled sequences were used 48 h after the end of the transfection procedure. Cells were treated for 1 h with the 5-FU and methotrexate and harvested 6, 12, and 24 h after the end of treatment, along with an untreated control sample for every condition. The experiments were conducted in triplicate. Proteins were extracted for Western blot analysis as described above. Anti-p53 mAb (clone BP53-12.1), anti-p21 mAb (clone SX118, Dako), and anti–phospho-H2AX histone diluted 1:1,000 (Cell Signaling Technology) were used. Membranes were then incubated for 1 h in the presence of horseradish peroxidase–labeled secondary antibody (diluted 1:5,000). On the same membranes, β-actin was evaluated as an internal control and densitometric analysis was carried out as for pRB expression.

Statistical analysis. Differences between groups were evaluated by Student's t test. Univariate analysis for DFS was done according to the Kaplan-Meier and Cox proportional hazards regression analysis. Multivariate analysis was done by applying the Cox proportional hazards regression model. All statistics were done using the SPSS statistical software package (Statistical Package for Social Science, SPSS, Inc.). P < 0.05 was regarded as statistically significant.

Immunohistochemical definition of RB status in tumor samples. In the present study, we first assessed the pRB phosphorylation level by evaluating the ppRB LI in all tumors considered using an anti-pRb mAb that specifically recognizes the ppRB form (clone G3-245). ppRB LI ranged from 0% to 91.2%, with a mean (±SD) value of 14.54% (±13.13%). Because in our series 40 cases (7.7%) showed a very low positivity for phosphorylated pRB (ppRB LI <1%), these cases were assumed to include two kinds of tumors: (a) tumors in which pRB was present but phosphorylated only in a few cells and (b) tumors in which both the pRB forms were absent, very likely due to RB1 deletion. To differentiate between these two groups, the 40 cases were investigated for the presence of total pRB, using a specific mAb (clone 1F8/Rb1) that recognize both the phosphorylated and the unphosphorylated or underphosphorylated pRB forms. Nine cases showed positive immunostaining in some cancer cells, whereas the remaining 31 cases showed no immunostaining in the cancer cell population. These latter cases were definitively regarded as RB1 deleted, and were included in the RB negative (−) group. The clinical and histopathologic characteristics of this group are reported in Table 1. The remaining 487 cases were included in the RB positive (+) group. Figure 1 shows two breast carcinomas characterized by a different pRB expression. Both cases were immunostained with the mAb that recognizes the total pRB form (clone 1F8/Rb1). A great number of positive nuclei are present within the cancer cell population of the case reported in Fig. 1A (included in the RB+ group), whereas all cancer cells from the case reported in Fig. 1B (scored in the RB− group) are negative for pRB.

Prognostic value of pRB expression and phosphorylation. After a mean follow-up time of 109 months (95% confidence interval, 104-113 months), the 11-year DFS estimate for the whole series of patients was 64.7%. Among the 518 patients, 93 patients did not receive any kind of adjuvant therapy, 49 patients received radiotherapy only, 231 patients endocrine therapy only, and 145 patients received chemotherapy. Table 2 reports the prognostic effect (univariate DFS analysis) of the pRB protein expression and phosphorylation in the whole series of patients, in patients treated by endocrine therapy alone, and those who received chemotherapy. In the whole series, the RB protein expression (RB+ or RB−) did not show any significant correlation with prognosis, whereas it became a significant predictor of DFS both in patients treated with endocrine therapy alone and in those treated with chemotherapy. It is worth noting that, whereas in patients who received tamoxifen as the sole adjuvant therapy the absence of pRB expression represented a negative prognostic index, it was associated with a better clinical outcome in patients treated with chemotherapy. To evaluate the relationship between the pRB phosphorylation and the patient clinical outcome, the ppRB LI variable was dichotomized using the cutoff point of 25% according to Derenzini et al. (16). The ppRB variable was significantly associated with DFS in the whole series, whereas it did not reach the significant value of P < 0.05 in patients receiving either endocrine therapy alone or chemotherapy.

The absence of pRB expression is the only predictive factor of good clinical outcome in patients treated with adjuvant chemotherapy. The results reported above indicated that the lack of pRB and not its inactivation by phosphorylation represented a predictive variable of DFS in patients who received endocrine therapy alone or chemotherapy. Therefore, we have further investigated the relationship between pRB expression and the clinical outcome in these two groups of patients, considering the possibility that the significant predictive effect of pRB found for both hormone-treated and chemotherapy-treated patients might be related to other clinical and histopathologic variables associated with the clinical outcome that can confound the results of the statistical analysis. Node status; tumor size; histologic grade; ER, PR, Ki67, and p53 LI; and HER2 status are well established tumor-related factors that might influence the clinical outcome of the patients treated with adjuvant therapy (25, 26). We compared the relative predictive value of these variables with that of pRB status in a multivariate analysis. The multivariate DFS analysis indicated that, whereas for patient treated with endocrine therapy the pRB status was no longer an independent predictive factor, the absence of pRB expression resulted to be the only significant variable predicting the clinical outcome in patients treated with chemotherapy (Table 3). Furthermore, because the number of RB− patients treated with chemotherapy was low (n = 16), we did a second DFS analysis comparing the population of patients with RB− cancer with a population of patients with RB+ cancer exhibiting the same characteristics. As reported in Table 4, the RB− cancers were all characterized by high histologic grade (G3), high Ki67-LI (>30%), and absence of ER. The RB− cancers were then matched with RB+ cancers according to these three variables. A match with a ratio of 1:2 was conducted according to Streiner and Norman (27). DFS analysis indicated that, also in this second data set, the absence of pRB expression remained a highly predictive factor of a better clinical outcome (hazard ratio, 6.09; 95% confidence interval, 1.40-26.46; P = 0.015). In Fig. 2, the DFS curves relative to RB− and RB+ patients are also reported.

5-FU and methotrexate treatment hindered cell population growth of RB1− silenced but not of control MCF-7 and HCT-116 cells. To ascertain whether the better prognosis of pRB-deficient tumors treated with adjuvant chemotherapy might be the consequence of a higher sensitivity of pRB-deficient cells to the drugs used, we treated asynchronously growing MCF-7 and HCT-116 cells, either silenced for RB1 or transfected with the scrambled sequences, with 5-FU and methotrexate at doses of 10 and 0.05 μg/mL for 1 hour. The dose of 5-FU and methotrexate used and the time of drug exposure were defined considering the pharmacokinetics of the drugs found in the interstitium of breast cancers of patients who received 5-FU and methotrexate chemotherapy according to the standard regimen (19). Cyclophosphamide was not considered because it requires metabolic changes occurring in the liver to be active (28). The effect of the RNA interference procedure on the level of RB1 mRNA was checked by real-time reverse transcription-PCR, and its effect on the expression of pRB was checked by immunocytochemistry and Western blot analysis. Cells transfected with scrambled sequences were used as controls and are so defined hereafter. In Fig. 3, the results relative to the effect of RB1 silencing on MCF-7 cells are reported; similar data were obtained using the HCT-116 cell line (not shown). We observed that 48 hours after the RNA interference procedure, a strong reduction in RB1 mRNA expression occurred, which was maintained at a very low level up to 120 hours (Fig. 3A). Immunocytochemical analysis for pRB expression revealed that, as early as 48 hours after the RB1 interference procedure, the intensity of the immunostaining was markedly reduced in comparison with control samples (Fig. 3B). Western blot analysis confirmed the reduction of pRB expression (Fig. 3C). At this point, we investigated the long-term effect of 5-FU and methotrexate treatment on the cell population growth in control and RB1-silenced MCF-7 and HCT-116 cells. We treated cells with 5-FU and methotrexate at doses of 10 and 0.05 μg/mL, respectively, for 1 hour and measured the expansion of the cell population after 3 days. The drug treatment started 48 hours after the RB1 silencing procedure was completed. In Fig. 4, a representation of the results obtained is shown. Seventy-two hours after the end of 5-FU and methotrexate treatment, the cell population growth of both MCF-7 and HCT-116 cells silenced for RB1 was significantly hindered. On the contrary, regarding the control cells, the 5-FU and methotrexate treatment induced a not significant reduction in the MCF-7 cell population growth (P = 0.313) and no reduction at all in the HCT-116 cells. To investigate the reason for the reduced growth rate of RB1-silenced MCF-7 and HCT-116 cells after drug treatment, we also evaluated the cell death rate in these cells and in control cells 24 hours after the end of 5-FU and methotrexate exposure. We found that the drug treatment was responsible for a significantly greater mortality in RB1-silenced MCF-7 and HCT-116 cells than in control cells (Fig. 4).

5-FU and methotrexate treatment caused a cell cycle arrest in control but not in RB1-silenced cells. To obtain information on the cause of the higher sensitivity of RB1-silenced cells to 5-FU and methotrexate treatment, we evaluated the effect of the drug exposure on cell cycle progression of control and RB1-silenced asynchronously growing MCF-7 cells. We carried out a dual-parameter flow cytometry analysis for DNA content and incorporated BrdUrd evaluation. For this purpose, both control and RB1-silenced cells, 72 hours after the end of the silencing procedure, were labeled with BrdUrd for 1 hour. Twelve hours later, when most of the labeled cells were passed to the G₁ phase, the cells were either immediately harvested (control cells) or treated with 5-FU and methotrexate at doses of 10 and 0.05 μg/mL for 1 hour and harvested 12 and 24 hours later for dual-parameter flow cytometry analysis. As shown in Fig. 5A and B, the control cells were mainly located in the G₀-G₁ region. Twelve hours after the exposure to 5-FU and methotrexate, the BrdUrd-labeled RB1-silenced cells seemed to move to the S phase and, 24 hours after the end of drug treatment, were accumulated in the G₂-M region (Fig. 5A, T1 and T2). On the other hand, at the same time, the BrdUrd-labeled, drug-treated control cells were prevalently confined to the early S-phase region and only a limited aliquot was able to reach the G₂-M compartment, without any accumulation in the G₂-M phase compartment (Fig. 5B, T1 and T2). These results indicated that 5-FU and methotrexate treatment caused an arrest of cell cycle progression in control cells but not in RB1-silenced cells. The arrest of cell cycle progression in control cells was removed 36 hours after the end of drug treatment.

The p53/p21 pathway was normally activated in RB1-silenced cells treated with 5-FU and methotrexate. Because the flow cytometry results indicated a temporary arrest of cell cycle progression in control but not in RB1-silenced MCF-7 cells after 5-FU and methotrexate treatment, we investigated whether in RB1-silenced cells the p53/p21 pathway, which is involved in the genotoxic-induced arrest of cell cycle progression (7), was hindered. We therefore used a time course Western blot analysis to evaluate the expression of p53 and p21 after 1-hour treatment with 5-FU and methotrexate in control and RB1-silenced MCF-7 cells. We found that in both control and RB1-silenced cells, the amount of p53 was greatly increased 6 hours after the drug treatment and progressively decreased thereafter (Fig. 6A). The expression of p21 reflected the p53 time course.

RB1-silenced cells accumulated DNA double-strand breaks. We also investigated whether the higher sensitivity of RB1-silenced cells to drug exposure might be the consequence of their reduced capacity for repairing the drug-induced DNA changes in comparison with control cells. For this purpose, we carried out a Western blot analysis with anti–phospho-H2AX antibody to reveal the accumulation of DNA double-strand breaks in drug-treated and untreated control and RB1-silenced MCF-7 cells. In agreement with the findings by Pickering and Kowalik (29), we observed that RB1 silencing caused untreated cells markedly to accumulate phosphorylated (γ) H2AX, thus suggesting a failure to repair the endogenously arising double-strand breaks promptly enough. The level of γ-H2AX seemed not to be increased after drug treatment. Control cells showed a very low level of γ-H2AX, which was not modified by drug exposure (Fig. 6B).

The present results showed that 5-FU and methotrexate treatment was much more effective on breast cancer cells lacking pRB expression than in those expressing the tumor-suppressor protein. This different sensitivity to the drugs was shown both in a series of human primary breast cancers and in human breast cancer cells cultured in vitro.

The absence but not inactivation of pRB predicts the clinical outcome of patients treated with tamoxifen or cyclophosphamide, methotrexate, and 5-FU adjuvant therapy. First, we have evaluated, in a univariate analysis for the DFS, the predictive value of both the expression of pRB and the degree of its phosphorylation in patients with breast cancers who were treated either with the standard chemotherapy regimen, which includes 5-FU and methotrexate, or with hormonal therapy. In fact, there is evidence that from the functional point of view, hyperphosphorylation abolishes the tumor suppressor activity of pRB (7). Thus, regarding the biological behavior of cancer cells, both the lack of phosphorylation and the hyperphosphorylation of pRB might have similar effects. Regarding the relationship between pRB expression and patient clinical outcome, we found that this pRB variable was not a significant prognostic variable in the whole series of patients. However, among the patients who received chemotherapy, those whose cancers lacked pRB had a better prognosis than those expressing the tumor suppressor protein. Chemotherapy extended the DFS in the former group of patients significantly longer than in the latter group. On the other hand, in the group of patients who received hormonal therapy, we found that the absence of pRB expression was associated with a worse prognosis. Regarding the relationship between the level of pRB phosphorylation and prognosis, we found that the level of pRB phosphorylation strongly correlated with the clinical outcome in the whole series of patients: The higher the pRB-LI, the shorter the DFS. However, among the patients who received chemotherapy, no significant difference in DFS rates was observed whatever the level of pRB phosphorylation, as well as among patients who received hormone therapy alone.

Lack of pRB expression was the only independent factor predicting a good clinical outcome in patients treated with adjuvant chemotherapy. According to the results reported above, we focused our attention on the relevance of the expression of pRB as a predictive variable in patients treated with adjuvant chemotherapy or hormone therapy alone. For this reason, we carried out a multivariate analysis for DFS, including, in addition to the pRB expression variable, all those tumor-related factors known to have a predictive value in breast cancers such as node status; tumor size; histologic grade; ER, PR, KI67, and p53 LI; and HER2 status (25, 26). We found that pRB expression resulted to be the only independent factor associated with the prognosis in patients treated with chemotherapy, the group of patients with RB− cancers having a better clinical outcome than those with RB+ cancer. On the other hand, in patients treated with hormone therapy, the pRB variable was not an independent predictor of DFS.

Because the number of breast cancers lacking pRB was small (n = 16), to validate the significant association between pRB expression and prognosis the 16 RB− tumors were matched with 32 RB+ tumors according those variables that characterized all the RB− tumors (high histologic grade, high proliferation rate, and absence of ER). Also, in this data set, patients with tumors lacking pRB expression had a significantly better clinical outcome than patients with RB+ tumors.

Therefore, even if the number of breast cancers lacking pRB expression is only a small fraction of total breast cancers, altogether these data indicated that the only reason for the good clinical outcome of patients treated with adjuvant chemotherapy is the absence of pRB expression from their cancers.

RB1-silenced MCF-7 and HCT-116 cells are more sensitive to 5-FU and methotrexate exposure than those expressing pRB. To ascertain the mechanism at the basis of the enhanced sensitivity of pRB-negative tumors to antimetabolite action, we analyzed the effects of 5-FU and methotrexate treatment on cell proliferation rate and cell cycle progression in MCF-7 and HCT-116 cells, either silenced for RB1 or transfected with scrambled sequences. The cytotoxicity mechanism of 5-FU, a drug widely used in breast and colon cancer chemotherapy, has been ascribed to the incorporation of fluoronucleotides into DNA and RNA and to the inhibition of the nucleotide synthetic enzyme thymidylate synthase (30). In breast cancer treatment, 5-FU is associated with methotrexate, an antifolate inhibitor of dihydrofolate reductase, which enhances the antitumor activity of 5-FU by facilitating the conversion of 5-FU to fUMP, the active metabolite of the drug (30). We found that a 1-hour treatment with 5-FU and methotrexate caused a marked reduction of the RB1-silenced MCF-7 and HCT-116 cell population growth, but not of control cells. Moreover, a greater death rate was observed in drug-treated RB1-silenced cells than in control cells.

The greater sensitivity of pRB-deficient cells to 5-FU and methotrexate exposure is due to the absence of a DNA damage checkpoint and DNA repair mechanisms. The cause of the different effects of 5-FU and methotrexate on control and RB1-silenced cells was investigated by evaluating the changes induced by the drugs on cell cycle progression of MCF-7 cells. Analysis of the cytofluorimetric results indicated that 1-hour drug treatment caused an arrest of cell cycle progression in control cells but not in RB1-silenced MCF-7 cells. These data, obtained using 5-FU and methotrexate at doses and time exposures derived from the evaluation of the interstitial pharmacokinetics of the drugs in human breast cancers in vivo (19), were consistent with the available evidence indicating that several DNA damage inducers used in human tumor chemotherapy inhibit G₁- and S-phase progression in RB-proficient but not in RB-deficient cells (31, 32). Specifically, it has been shown that RB-proficient cells exposed to 5-FU failed to accumulate in any phase of the cell cycle, indicating that the drug is responsible for the arrest in all phases of the cell cycle (33).

The arrest of cell cycle progression observed in control MCF-7 cells after 5-FU and methotrexate treatment was transitory: In fact, 36 hours after the end of drug treatment, the cytometry profiles of control MCF-7 cells showed an accumulation of BrdUrd-labeled cells in the G₂-M phase. In control cells, the drug-induced arrest of cell cycle progression may reduce incorporation of fluorinated nucleotides and allow DNA damage caused by fluorinated nucleotide incorporation to be repaired. This was not the case for the pRB-deficient cells in which the absence of a DNA damage checkpoint allowed fluorinated nucleotides to be incorporated into DNA with the consequent progressive accumulation of DNA damage and reduction of cell viability (34). In this regard, it is worth noting that control cells did not show accumulation of double-strand breaks, thus indicating a normal DNA repairing activity, whereas RB1-silenced cells exhibited elevated levels of γ-H2AX, indicative of defects in the DNA repair machinery.

In other words, pRB-proficient cells may be more resistant to antimetabolite exposure than pRB-deficient cells because they have the time for repairing the 5-FU–induced damage by possessing functioning cell cycle checkpoint and DNA repair mechanisms. This repair would be impossible for cells lacking pRB in which the DNA damaging agents do not induce arrest of cell cycle progression and DNA repair mechanisms are hindered. The importance of having the time for DNA repairing in determining cancer cell sensitivity to chemotherapy agents was also previously stressed by Wang et al. (35) who showed that 5-FU–resistant human colon and breast cancer cells were characterized by a slower growth rate, higher proportion of G₁, and lower proportion of S-phase cells than the parental cell lines, and suggested that this phenotype may protect resistant cells from cell death by allowing time to repair the 5-FU–induced damage.

Taken together, these data indicate that the absence of pRB expression in human breast cancers is predictive of a good clinical outcome for patients treated with adjuvant chemotherapy. This seems to be due to the higher sensitivity of pRB-deficient cells to 5-FU and methotrexate exposure than RB-proficient cells, as a consequence of the absence of a DNA damage checkpoint in the pRB-deficient cells. Our results also suggest that systemic chemotherapy should be considered to represent the first choice adjuvant treatment for patients with RB− cancers.