Phospholipase Cγ1 (PLCγ1) is highly expressed in several tumors. We have previously reported that both stable and inducible PLCγ1 down-regulation resulted in an almost complete inhibition of breast cancer-derived experimental lung metastasis formation. The aim of our study is to evaluate the association between the expression of PLCγ1 and of PLCγ1 phosphorylated at Tyr1253 (PLCγ1-pY1253) and at Tyr783 (PLCγ1-pY783) with the clinical outcome of patients with node negative, T1/T2 breast cancers. The study groups consisted of 292 (training set) and 122 (validation set) patients presenting with primary unilateral breast carcinoma (T1-T2), with no evidence of nodal involvement and distant metastases. PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 protein expression were assessed by immunohistochemistry on tissue microarrays and the results correlated with the clinical data using Kaplan–Meier curves and multivariate Cox regression analysis. Tumor cells while expressing variable proportions of cytoplasmic PLCγ1, express PLCγ1-pY1253 and PLCγ1-pY783 predominantly in the nucleus. High expression of PLCγ1, and of its activated forms, is associated with a worse clinical outcome in terms of incidence of distant metastases, and not of local relapse in T1-T2, N0 breast cancer patients undergone adjuvant chemotherapy. PLCγ1 over-expression appears to be a reliable predictive surrogate marker of development of metastases. Thus, targeting PLCγ1 pathways might represent a potential therapeutic approach for the prevention of metastatic disease in breast cancer.
Breast cancer is the most frequent cancer in women, accounting for more than 200,000 new cases/year in the US and 350,000 in Europe.1, 2 Recently, declines in mortality from breast cancer were reported, likely as a result of public awareness, implementation of screening programs, which favor the increased detection of early stage, node-negative (N0) tumors at low risk of recurrence and advances in the adjuvant treatments.1, 3 Nevertheless, only about 70–80% of these tumors are cured by local or regional treatment alone.4 Adjuvant systemic therapy significantly reduces the risk of both disease recurrence and associated death from breast cancer.4 Based on the risk-group identification provided by well established prognostic factors (such as tumor size and grade, steroid hormone receptor status, age or menopausal status) 85–90% of patients without nodal involvement are currently receiving some kind of adjuvant treatment, in spite of tumor recurrence in only one-third of the cases.5, 6 Besides the economic constraints which health care systems are increasingly facing, it should be emphasized that adjuvant treatments are not without associated risks. In view of this, the search of biomarkers endowed with high discriminatory power to identify node negative patients for whom adjuvant therapy is mandatory.
Metastasis, the ability of cancer cells to spread from a primary site and form tumors at distant sites, is the main cause of death associated with cancers. Lipid signaling in disease is an emerging field of investigation and phosphoinositides are the most extensively studied lipids involved in cell signaling pathways, leading to cell differentiation, proliferation or apoptosis. Activation of the enzyme phospholipase Cγ1 (PLCγ1) is thought to play a critical role in both cytoskeletal changes and migration associated with the metastatic process.8, 9 Activation of PLCγ1 by phosphorylation can occur downstream of many tyrosine kinases receptor including epidermal growth factor receptor, vascular endothelial growth factor receptor-2, c-MET, platelet-derived growth factor receptor and certain integrins. Activation induces hydrolysis of phosphatidylinositol-4,5-biphosphate to form the second messengers diacylglycerol and inositol-1,4,5-triphosphate, which in turn activate a number of signaling pathways.10, 11 The tyrosine kinase receptors phosphorylate PLCγ1 on three tyrosine residues: Y771, Y783 and Y1253. Phosphorylated PLCγ1 exhibits increased enzymatic activity and phosphorylation at Y783 has been reported to be necessary for PLCγ1 activation in vitro and in vivo.12–14
PLCγ1 is highly expressed in several tumors, including breast carcinomas8, 15–17 in which the enzyme has been shown to be required for epidermal growth factor induced migration of breast cancer cells.18 Furthermore, both stable and inducible PLCγ1 down-regulation resulted in an almost complete inhibition of breast cancer-derived experimental lung metastasis formation.19 Experimental lung metastases of breast and prostate cancer were also significantly reduced by a dominant negative fragment of PLCγ.20
To establish the significance of PLCγ1 in breast cancer, we examined by immunohistochemistry (IHC) the expression levels of PLCγ1 and of PLCγ1 phosphorylated at Tyr1253 (PLCγ1-pY1253) and Tyr783 (PLCγ1-pY783) in a cohort of patients with T1/T2, N0 breast cancers. The study hypothesis was that high expression levels of the protein would be associated with a significantly shorter disease-free survival (DFS).
Material and Methods
Eligible patients were extracted from 691 cases consecutively diagnosed between 1985 and 2000 at the Regina Elena National Cancer Institute, Rome, Italy and presenting with primary unilateral breast carcinoma. From the original series, only N0 patients (n = 379) and among them only those with T1/T2 tumors (n = 357) were tentatively included into the study. pN0 cases were re-evaluated by lymph node step-sectioning and cytokeratin expression analysis as reported.21 Minimal lymph node deposits were present in 49 of pN0 patients, which were then excluded from further analysis. Archival material was not available in 16 cases. Thus, the final number of evaluable patients decreased to 292.
An independent validation set consisted of 122 (T1/2; N0) breast cancer patients who underwent initial surgery at the same Institute, between 2001 and 2003.
Patients and tumor characteristics for both sets are summarized in Table 1. The study was reviewed and approved by the ethics committee of the “Regina Elena” National Cancer Institute and written informed consent was obtained from all patients.
Table 1. Patients and tumor characteristics (training and validation sets)
All patients of training and validation sets received radiation therapy. In the training set, 96 patients received hormonal therapy and 120 patients were treated with adjuvant chemotherapy (followed or not by hormonal therapy). In the validation set, 64 patients received hormonal therapy and 52 patients were treated with adjuvant chemotherapy.
Patients with HER-2-positive tumors did not receive trastuzumab, because it was unavailable in the study period. The median follow-up was of 91 months (range 6–298 months) and 59 months (range 2–123 months) for training and validation sets, respectively. Follow-up data were obtained from institutional records or by the referring physicians. During follow-up, 35 patients (12.0%) in the training set and 15 patients (12.3%) in the validation set developed a local recurrence. A distant recurrence was observed in 49 (16.8%) and 16 (13.1%) cases in the training and validation set, respectively.
Tissue microarrays (TMA) were constructed by removing 2-mm diameter cores of histologically confirmed invasive breast carcinoma areas from each original paraffin block and re-embedding these cores into gridded paraffin blocks, using a precision instrument (MTA, Beecher Instuments, WI). TMA sections were incubated with the anti-PLCγ1 mouse monoclonal antibody (sc-7290) and with rabbit polyclonal PLCγ1-pY1253 (sc-22141-R) and anti-PLCγ1-pY783 (sc-12943-R) antibodies (Santa Cruz Biotechnology, Santa Cruz, CA). Although validated by company, the specificity of the antibodies has been further tested by IHC as reported in Supporting Information Figure S1. The anti-mouse and the anti-rabbit EnVision kits (Dako, Glostrup, Denmark) were used for signal amplification, as appropriate. In control sections, the specific primary antibody was omitted or replaced with nonimmune serum or isotype-matched immunoglobulins. Immunohistochemical analysis was done by two pathologists (M.P., R.L.) by consensus without knowledge of the clinicopathologic informations.
For confocal laser-scanning microscopy (LSM 510 META microscope, Zeiss, Jena, Germany), after antigen retrieval, sections were incubated overnight at 4°C with the primary antibodies. Then, Alexa Fluor® 488 conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR) and Alexa Fluor® 546 conjugated goat anti-rabbit IgG (Molecular Probes) were used (60 min incubation at 1:150 dilution) for signal detection, as appropriate. Cell nuclei were counterstained with DRAQ5® (BioStatus, Shepshed, Leicestershire, UK).
Pathologic tumor size and tumor grade were dichotomized according to the St. Gallen criteria22 for the definition of risk categories (T ≤ 2 cm vs. T > 2 cm; Grade 1 vs. Grade 2–3). Estrogen (ER) and progesterone (PR) receptors were classified as positive if ≥ 10% of positive staining cells were counted. A specimen was considered with high levels of Ki-67 expression if ≥ 14% of tumor nuclei stained positively.23 HER-2 membranous staining was scored according to Herceptest (Dako) and classified as positive if the intensity was scored 3+, with more than 30% of cells showing complete membrane staining,24 or if the intensity was scored 2+ in presents of an amplification of the HER-2 gene as assessed by fluorescent in situ hybridization. To dichotomize PLCγ1 and PLCγ1-pY1253 protein expression, a cut-off value corresponding to the 50th percentile in the training set was chosen (75 and 61% of positive cells, respectively). A cut-off value of 59%, corresponding to the 75th percentile in the training set, was chosen for PLCγ1-pY783 (Supporting Information Fig. S2). For analytical purposes, patients of the training and validation sets were also split in two groups: chemotherapy-treated (n = 120 and n = 52, respectively) and hormonal-treated group (n = 96 and n = 64, respectively). The relationships between PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 expression and clinicopathological parameters were assessed by Pearson's χ2 or Fisher's exact test, as appropriate. DFS was defined as the time from surgery to the first of the following events: tumor recurrence at local site or at distant sites. Local (LRFS) and distant relapse-free survivals (DRFS) were the times from surgery to the occurrence of relapse at local and distant sites, respectively. Kaplan–Meier plots were used to illustrate the survival in specified cohorts and the log-rank test to test for equality of survival curves. The association of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 expression with outcome, adjusted for other prognostic factors, was tested by Cox's proportional hazards model. The following covariates were included in the multivariate DFS models: tumor size, tumor grade, ER and PR status, Ki-67 status, HER-2 status, PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 status. Appropriateness of the proportional hazard assumption was assessed by plotting the log cumulative hazard functions over time and checking for parallelism. SPSS Version 15.0 (SPSS, Chicago, IL) was used throughout and p < 0.05 was considered statistically significant.
Expression of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 proteins in breast tissues
In non-neoplastic breast specimens, PLCγ1 was immunohistochemically expressed in terminal duct lobular units (TDLU). A homogenously weak expression was observed in the cytoplasm of luminal epithelia, but not in myoepithelia. The majority of cells in both epithelial and myoepithelial layers showed a specific nuclear immunoreactivity for PLCγ1-pY1253. On the contrary, TDLU cells did not express PLCγ1-pY783 (Figs. 1a and 1c). Nipple epithelium and lactiferous ducts were negative for PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 staining (data not shown).
In breast tumors, a diffuse immunoreactivity for nonphosphorylated PLCγ1 was found in the cytoplasm of neoplastic cells, whereas positivity for PLCγ1-pY1253 and PLCγ1-pY783 was predominantly nuclear, with < 5% of the cases showing a membrane staining. The percentage of positive cells on cell membrane ranged from 5 to 92 (Supporting Information Fig. S3). Examples of high and low expression of PLCγ1 in breast cancers are shown from figure 1d to figure 1i. Confocal microscopy analysis of paraffin-embedded tumors confirmed the prevalent nuclear localization of activated PLCγ1 forms (Fig. 1l and 1m).
High PLCγ1 expression (PLCγ1High) was inversely correlated with PR status (p = 0.018). PLCγ1-pY1253 and PLCγ1-pY783 expressing tumors did not differ significantly for the distribution of clinicopathological variables evaluated (Supporting Information Table 1).
Expression of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 proteins and DFS
Forty-nine out of 146 patients (34%) with PLCγ1High and 35 out of 146 (24%) patients with low PLCγ1 (PLCγ1low) expressing tumors had a disease relapse. Seventeen out of 146 (12%) with PLCγ1High and 18 out of 146 patients (12%) with PLCγ1Low developed a local recurrence. Distant metastases were observed in 32 out of 146 (22%) with PLCγ1High and 17 out of 146 (12%) patients with PLCγ1Low. Fifty out of 146 patients (34%) with PLCγ1-pY1253High and 34 out of 146 patients (23%) with PLCγ1-pY1253Low expressing tumors had a disease relapse. Eighteen out of 146 (12%) with PLCγ1-pY1253High and 17 out of 146 patients (12%) with PLCγ1-pY1253Low developed a local recurrence. Distant metastases were observed in 32 out of 146 (22%) with PLCγ1-pY1253High and 17 out of 146 (12%) patients with PLCγ1-pY1253Low. Twenty-nine out of 73 patients (40%) harboring PLCγ1-pY783High tumors and 55 out of 219 patients (25%) with PLCγ1-pY783Low tumors had a tumor relapse. A local recurrence was observed for 12 out of 73 (16%) PLCγ1-pY783High and 23 out of 219 (11%) PLCγ1-pY783Low tumors. Twenty-three percent (17 out of 73) of subjects with PLCγ1-pY783High and 15% (32 out of 219) of those with PLCγ1-pY783Low tumors developed distant metastases.
Univariate analyses showed a significant association of high expressions of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 with lower DFS rates (p = 0.030, p = 0.031 and p = 0.026, respectively). In particular, tumors over-expressing PLCγ1 or its phosphorylated forms were associated with a significantly higher incidence of distant relapse (p = 0.010, p = 0.008 and p = 0.030), while no significant association with local recurrence was found (Fig. 2). Multivariate analyses of DFS adjusted for the other prognostic factors showed an independent prognostic significance of PLCγ1 (HR1.6; 95% CI, 1.0–2.5; p = 0.048), PLCγ1-pY1253 (HR1.8; 95% CI, 1.1–2.9; p = 0.013), PLCγ1-pY783 (HR1.9; 95% CI, 1.2–3.0; p = 0.011) and tumor grade. However, the prognostic significance of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 was limited to DRFS (Table 2).
Table 2. Multivariate analyses in the training set (n = 292)
To explore the possible predictive value of PLCγ, PLCγ1-pY1253 and PLCγ1-pY783 expression, analyses of DFS were conducted on the subset of 216 patients of the training set receiving systemic adjuvant treatment (chemotherapy or hormonal). At univariate analyses, high expressions of PLCγ1-pY1253 and PLCγ1-pY783 were significantly associated with higher DRFS rates (p = 0.030 and p = 0.001) only in patients treated with chemotherapy (Fig. 3). Multivariate analyses revealed that high expression of PLCγ1-pY1253 (HR3.0; 95% CI, 1.1–8.3; p = 0.031) and PLCγ1-pY783 (HR4.4; 95% CI, 1.7–11.4; p = 0.002) were independent variables influencing the DRFS (Table 3).
Table 3. Training set: multivariate analysis in patients treated with chemotherapy (n = 120)
To confirm the prognostic value of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783, we conducted a validation study consisting of an independent set of 122 patients. PLCγ1 expression was positively correlated with tumor grade (p < 0.001). PLCγ1-pY1253 and PLCγ1-pY783 expressing tumors did not differ significantly for the distribution of clinicopathological variables evaluated (Supporting Information Table 2).
Survival analyses showed a statistically significant association of high expressions of PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 with lower DFS rates (p = 0.014, p = 0.033 and p = 0.021, respectively) and higher incidence of distant relapse (p = 0.011, p = 0.020 and p = 0.020, respectively), while no significant association with local recurrence was found (Supporting Information Fig. S4).
Multivariate analyses of DFS showed an independent prognostic significance of PLCγ1 (HR2.6; 95% CI, 1.2–5.7; p = 0.020), PLCγ1-pY1253 (HR2.4; 95% CI, 1.1–5.4; p = 0.033) and PLCγ1-pY783 (HR2.4; 95% CI, 1.0–5.7; p = 0.042). No significant association between PLCγ1, PLCγ1-pY1253 and PLCγ1-pY783 status and local recurrence were found. In the analyses of distant recurrences, a significant association with PLCγ1 (HR3.9; 95% CI, 1.2–12.7; p = 0.021), PLCγ1-pY1253 (HR3.3; 95% CI, 1.2–9.5; p = 0.025) and PLCγ1-pY783 (HR3.4; 95% CI, 1.1–10.3; p = 0.035) status were found (Supporting Information Table 3).
Survival analyses conducted on the subset of 116 patients receiving systemic adjuvant treatment (chemotherapy or hormonal) showed that high expressions of PLCγ1-pY1253 and PLCγ1-pY783 were significantly associated with higher DRFS rates (p = 0.003 and p = 0.033) only in patients treated with chemotherapy (Supporting Information Fig. S5).
At multivariate analyses only PLCγ1-pY1253 status remained an independent predictor of DRFS (HR4.1; 95% CI, 1.3–13.9; p = 0.036) while PLCγ1-pY783 showed a trend toward statistical significance (HR4.4; 95% CI, 0.8–10.3; p = 0.098; Supporting Information Table 4).
A prerequisite for individualized therapy in breast cancer is the identification of N0 patients at high or low risk of relapse. This could avoid unnecessary morbidity to the latter patients while providing benefits to the formers. Here, we evaluated the prognostic role of phosphorylated PLCγ1 expression in T1/T2, N0 breast cancer patients. Increased levels of tyrosine phosphorylated PLCγ1 has been previously described in breast tumors but no relationship with the disease stage or outcome has been investigated.15
Of the three-phosphorylation sites of PLCγ1, Tyr-771 and Tyr-783 are located between the X and Y catalytic domains whereas Tyr-1253 is localized near the COOH terminus.25 Our analyses showed a predominant if not exclusive nuclear localization of both pY1253 and pY783 PLCγ1 in breast cancer tumors. PLCγ1, which lacks nuclear localization signal, predominantly localizes in the cytoplasm. However, nuclear localization has been described in tumorigenic and in highly proliferating cell lines, but not in primary embryo skin or lung fibroblasts.26, 27 Activated PLCγ1 translocates to the nucleus, where it acts as a guanidine exchange factor for the nuclear GTPase PI3K enhancer, which subsequently activates PI3K.26, 27 Although the classical phosphoinositide cycle operates at the plasma membrane level, other phosphoinositide cycles have been described, with one of them operating within the nucleus.26, 28, 29 Our findings show an association between nuclear PLCγ1 and breast cancer progression. Indeed, high expression levels of activated nuclear forms were predictive of worse DFS, because of a significantly higher incidence of distant metastases. By multivariate analyses, over-expression of activated nuclear forms, in addition to not activated PLCγ1, were the only independent predictors of distant relapse risk in both the training and validation sets. This finding is not unprecedented since down regulation of PLCγ1 expression resulted in an almost complete inhibition of lung metastasis formation as well as in metastasis regression in an experimental model of breast cancer metastatic disease.19, 20
Yet, the mechanism by which PLCγ1 promotes migration and metastatization is still far from being fully defined. It is known that tumor cell migratory and invasive properties require rearrangement of the cytoskeleton to form migratory structures such as lamellipodia and filopodia.9, 30 Compelling evidence is present for the critical role of PLCγ1 in cytoskeletal changes required for the acquisition of the metastatic phenotype.8, 9 Dephosphorylation of PLCγ1 on residue Y783 prevents PLCγ1 activation, thus blocking PLCγ1-activated remodeling of the cytoskeleton and migration of gliomas cell.31 PLCγ1 can contribute to metastatization by activating directly,32 or indirectly RAC1,19, 33 thus inducing migration-supporting cellular structures.
It is important to underline that in this study activated PLCγ1 over-expression was found to be a significant negative prognostic factor for distant metastases only in patients who received adjuvant chemotherapy. In this context, it should be recalled that the presence of cancer cells with stem cell-like properties is known to occur in breast cancer cell lines endowed with metastatic capacity and in primary invasive breast tumors. This feature has been shown to correlate with higher risk of distant metastases and poor clinical outcome.34, 35 In addition, clinical observations and in vitro studies have shown that residual breast cancer cells surviving chemotherapy display stem-like cell features.36, 37 All together these observations strongly lead to the working hypothesis that cancer cells over-expressing activated PLCγ1 could be selected/resistant to chemotherapy and could belong to the compartment of stem-like/tumor initiating cells.
In summary, our study demonstrates a highly significant association between high expression of activated PLCγ1 and risk of metastatic relapse in T1/T2, N0 breast cancer patients treated with chemotherapy. This represents the first clinical pathological correlation supporting the notion that PLCγ1 may play a prometastatic role in human breast cancer. Thus, the measurement of the activated enzyme may represent a novel reliable predictive surrogate marker of distant relapse. Similarly, targeting PLCγ1 pathways might emerge as a new therapeutic approach for the prevention of metastatic progression.