c-Jun NH2-terminal kinase (JNK) is characterized as a stress-activated protein kinase (SAPK), which regulates apoptosis following cellular stress.1, 2, 3 It is originally identified by its ability to phosphorylate the N-terminal transactivation domain of c-Jun upon UV irradiation and oncoprotein expression.4 The key evidence that relates JNK to apoptosis comes from the observation that jnk1−/−jnk2−/− mice are resistant to apoptosis induced by UV irradiation, anisomycin or MMS.5 Furthermore, activation of JNK is involved in the BRCA1-facilitated apoptosis in cells derived from osteosarcoma, breast cancer and ovarian cancer.6, 7 However, activation of JNK is also observed in oncogene-expressing cells,8, 9, 10, 11 and activated JNK can transform NIH 3T3 fibroblasts in vitro12 and is sufficient for ex vivo fibrosarcoma development in nude mice.13 Although active forms of JNK have been associated with primary glial tumors,14, 15 their in vivo roles in human cancers, such as breast cancer, are undefined. A recent report indicated an imperfect association between the increased immunoreactivity of activated c-Jun, a downstream effecter of p-JNK, and the poorer survival in breast cancer (p = 0.061).16 However, the clinical information about the role of p-JNK in breast cancer is lacking.
JNK and another SAPK, p38, as well as extracellular signal-regulated protein kinase (ERK) comprise MAPK family.17, 18 In breast cancer, hyperphosphorylation and overexpression of ERK1/2 has been observed in tissue specimens19, 20, 21 and cell lines.22, 23 The patients with the increased ERK1/2 activity in breast tumor are associated with a reduced disease recurrence-free survival rate.21 Activation of p38 occurs following exposure to cytokines, heat shock or high osmolarity, and plays either a positive or a negative role in the proliferation of cancer cells.24, 25, 26, 27, 28, 29 Only few publications provide evidence for the oncogenic potential of p38. In breast cancer, phosphorylated p38 is present in 17–20% of breast carcinomas and is associated with a poor outcome in node-positive breast carcinoma.30, 31 Intriguingly, some in vitro studies on MAPK signaling pathways revealed a high degree of crosstalk. Prolonged activation of ERK can stimulate JNK activity through the induction of heparin-binding epidermal growth factor (hb-EGF).32 Activation of JNK in turn negatively regulates the signal propagation through the cytoplasmic cascade. In addition, activation of JNK and p38, which leads to concurrent inhibition of ERK, is critical for induction of apoptosis in rat PC-12 pheochromocytoma cells.33 Moreover, activated JNK and p38 are both able to increase the stability of VEGF mRNA and thus support tumor growth.34 There is little information available whether the crosstalk/correlation between the activated MAPKs is present in human cancers, in vivo. If it indeed exists in human cancers, the crosstalk might affect the processes of tumor initiation and/or progression and possibly contribute to the survival outcome of cancer patients.
To elucidate the roles of phosphorylated JNK and its possible correlation with other MAPK family members (p-p38 and p-ERK) in breast cancer, we analyzed their expression patterns in the paired cancer and noncancer tissues from breast infiltrating ductal carcinoma (IDC) cases by immunoblotting. Herein, we reported that aberrant p-JNK expression was closely associated the overall survival of IDC and the positive correlation between p-JNK and p-p38 might play a role in the carcinogenesis of breast cancer.
Material and Methods
Tissue samples and cell culture
Sixty-eight patients with confirmed breast IDC were enrolled in this study. The paired cancer and noncancer breast tissue specimens were obtained from the same patients who underwent surgical treatment at the Department of Surgery, Kaohsiung Medical University Hospital, during the period from 2001 to 2004. The specimens were frozen immediately in liquid N2 and stored at a −80°C freezer until further analysis. None of the patients had undergone radiotherapy or chemotherapy before operation. The ethics committee approved the study and informed consent was obtained from each patient. The clinicopathological features of the patients were recorded (Table I). The histological types of the primary tumor were determined according to a system based on a modification of the WHO classification. The staging of the breast cancer was defined according to the TNM system. MCF7 human breast cells were obtained form ATCC and were grown in DMEM/F12 medium supplemented with 10% FBS, 100 U/ml penicillin and 100 mg/ml streptomycin.
Table I. Clinicopathological Characteristics and the Expression Profiles of P-JNK1/2, P-ERK1/2 and P-P38 in Breast Infiltrating Ductal Carcinoma Cases
N: normal breast tissue; C: breast cancer tissues; NE: nonexpression.
Not determined in a small numbers of cases.
LN metastasis, n
ER status, n
PR status, n
Her2/Neu status, n
C > N
C = N
C < N
C > N
C = N
C < N
C > N
C < N
The frozen breast tissues were ground up and dissolved in EBC buffer (50 mM Tris (pH 7.6), 120 mM NaCl, 0.5% Nonidet P-40, 1 mM EDTA, 1 mM 2-mercaptoethanol, 50 mM NaF, 1 mM Na3VO4). MCF-7 breast cancer cells, grown in charcoal-stripped DMEM medium, were served as internal control in immunoblotting analysis. The detailed procedures for immunoblotting analysis were followed accordingly.35 Antibodies (anti-p-ERK1/2, anti-p-JNK1/2 and anti-p-p38) were purchased from Cell Signaling Technology (Beverly, MA). For comparison of the expression levels of p-ERK1/2, p-JNK1/2 and p-p38 in cancer tissues versus matched noncancer breast tissues, the AlphaImager™ 1220 documentation and analysis system was used (Alpha Innotech Corporation, San Leandro, CA). C > N was defined as a more than 50% higher expression of activated MAPK (normalized by β-actin) in cancer tissue than in paired noncancer tissue. C < N was defined as a more than 50% higher expression of activated MAPK (normalized by β-actin) in noncancer tissue than in paired cancer tissue. N = C was defined as a less than 50% difference in activated MAPK expression (normalized by β-actin) between the cancer tissue and the paired noncancer tissue.
The detailed protocol for immunohistochemical study was followed accordingly.35 In brief, the tissue samples were fixed with 10% buffered formalin, then dissected, dehydrated and coated with wax. The samples were sliced to a thickness of 4 μm, and then either stained with hematoxylin–eosin or incubated with the primary antibody (antiestrogen receptor, antiprogesterone receptor and anti-Her2/Neu were purchased from DAKO; anti-p-ERK1/2, anti-p-JNK1/2 and anti-p-p38 antibody were purchased from Cell Signaling Technology), followed by Universal LAB + kit/HRP (DAKO), or counterstained with hematoxylin. The results were captured by the Nikon E-800M microscope, and then processed by Kodak MGDS330 and Adobe Photoshop 6.0.
Notably, the paraffin-embedded human lung carcinoma tissues were used as the control for the p-JNK1/2 and p-p38 staining, and the paraffin-embedded human breast carcinoma tissues were used as the control for the p-ERK1/2 staining according to the manufacturer (Cell Signaling Technology). Only the nuclear staining in tumor cells (about 1,000 cells in 3–4 high power fields) was calculated. The percentage of positive cells >50% of tumor cells was regarded as high expression, while the percentage of positive cells <10% of tumor cells was regarded as low expression. Accordingly, the controls of p-JNK1/2, p-p38 and p-ERK1/2, which were identified as high and low expression in lung and breast carcinoma, respectively, were applied along with each staining run. Her2/Neu was scored by the widely accepted criteria that assessed the intensity and completeness of membrane staining.36 In this study, scores 0 and 1+ were considered normal (i.e., negative for overexpression), and 2+ and 3+ were considered positive for Her2/Neu overexpression according to the HerceptTest system scale (DAKO 5204). The paraffin-embedded human breast carcinoma tissues previously scored as positive or negative for Her2/Neu overexpression were used as the control of Her2/Neu staining in each staining run.
All of the statistical analyses were performed using the SPSS 10.0 statistical package for PC (SPSS, Chicago, IL). Groups with the different expression levels of activated MAPK family members according to dominant expression pattern [the high expression group (C > N) and the low expression group (N = C plus C < N) for p-ERK1/2; the low expression group (C < N) and the high expression group (N = C plus C > N) for p-JNK; and the expression group versus the nonexpression group for p-p38] were correlated with tumor stage, tumor grade, lymph nodes (LN) status, tumor size, ER status, PR status and Her2/Neu status by two-tailed Student t-test, Pearson's and Spearman's rank correlation tests, and were further confirmed by Fisher's exact test. Survival curve was generated by Kaplan–Meier survival curves, and the significance of the difference between curves was analyzed by log-rank test. Furthermore, Cox regression model was used for the multivariate analyses on overall survival. Two-tailed p ≤ 0.05 was considered statistically significant.
Expression profiles of phosphorylated JNK (p-JNK) in the paired cancer and noncancer breast tissues
The clinicopathological characteristics of the patients with breast IDC are shown in Table I. To investigate the expression profiles of p-JNK in breast IDC tissues, immunoblotting analysis was applied (Fig. 1a and Table I). As shown in Table I, 48.5% of the cancer tissues had a lower p-JNK1/2 expression, when compared to the nearby noncancer tissues (Table I, C < N). While p-p38 was not detected in 36.8% of the cases (neither in cancer tissues nor the matched noncancer tissues), p-p38 expression was increased in 48.5% (C > N) and decreased in 14.7% (C < N) of breast cancer tissues (Table I). Further examination by immunohistochemistry showed the percentage of positive tumor cells with an intense nuclear p-ERK1/2 and p-p38 staining was corresponding to the immunoblotting results (data not shown) and was predominantly seen in the matched breast cancer lesions (Figs. 2a and 2e) but not the neighboring normal mammary duct (Figs. 2b and 2f). Similarly, the percentage of positive tumor cells with a nuclear p-JNK1/2 staining was also corresponding to the immunoblotting results (data not shown) and was predominantly observed in the normal mammary duct (Fig. 2d) but not the matched breast cancer lesions (Fig. 2c). There was a weak to absent immunoreactivity of p-JNK1/2, p-ERK1/2 and p-p38 in the neighboring stroma cells, but a moderate to strong immunoreactivity of those phosphoproteins in the peripheral lymphocytes.
Correlation of p-JNK expression with clinicopathological characteristics, p-ERK1/2 and p-p38
The expression patterns of activated MAPKs in breast cancer tissues were further correlated to clinicopathological characteristics, such as tumor stage, tumor grade, lymph node status, tumor size, age at diagnosis, ER status, PR status and Her2/Neu status. We found that decreased p-JNK1/2 expression in breast cancer tissues was correlated significantly with increased tumor grade and decreased patient age at diagnosis (p = 0.030 and 0.029, respectively) (Table II and Fig. 3a). The expression of p-p38 in breast cancer tissues was correlated significantly with decreased tumor grading and negative Her2/Neu status (p = 0.004 and 0.023, respectively), while increased p-ERK1/2 expression in breast cancer tissues was correlated significantly with increased tumor stage and size (p = 0.005 and 0.029, respectively) (Table II and Fig. 3b). To analyze the correlation among these MAPKs themselves, we found that p-JNK1/2 expression was positively correlated with p-p38 expression (Table III, p = 0.002), but not p-ERK1/2 expression, in breast cancer tissues as compared to the matched noncancer breast tissues. And, there was no significant correlation between p-p38 expression and p-ERK1/2 expression (data not shown).
Table II. Correlation of Phosphorylated MAPK Family Members with Clinicopathological Characteristics in Breast Infiltrating Ductal Carcinoma Cases
NE: nonexpression; E: expression (C > N plus C < N).
The expression of p-ERK1/2, p-JNK1/2 and p-p38, determined by immunoblotting, were further correlated with the overall survival of the patients, using Kaplan–Meier survival curves. Low p-JNK1/2 expression group (C < N) was correlated significantly with a better overall survival than high p-JNK1/2 expression group (C ≥ N) (Fig. 4b, p = 0.004). Although there was no significant difference in overall survival between the different expression groups of p-ERK1/2 and p-p38 (Figs. 4a and 4c), low p-JNK1/2 expression in the low p-ERK1/2 expression group, but not in the high p-ERK1/2 expression group, was associated significantly with a better overall survival (p = 0.008) (Fig. 5a). In the p-p38 expression group, but not the p-p38 nonexpression group, low p-JNK1/2 expression was significantly associated with a better overall survival rate (p = 0.007) (Fig. 5b). Accordingly, we found that p-JNK1/2 expression was an independent determinant for the overall survival of breast IDC, after adjustment for p-ERK1/2 expression, p-p38 expression, estrogen receptor status, progesterone receptor status, Her2/neu status, tumor size, tumor grade and LN metastasis, using COX regression model (odds ratio = 5.721, 95% confidence interval = 1.55–21.03, p = 0.009) (Table IV).
Table IV. Cox Regression Multivariate Analysis of Overall Survival for Breast Infiltrating Ductal Carcinoma Cases
Odds ratio (OR)
CI: confidence interval.
In this study, we noticed that p-JNK1/2 expression was persistent in both cancerous and noncancerous breast tissues (Fig. 1). While p-JNK1/2 expression was predominantly decreased in breast cancer tissues as compared to the matched noncancer tissues (Table I), its expression was correlated significantly with poorly differentiated tumors and decreased age at diagnosis (Table II and Fig. 3a), and thus suggested that the patients with the decreased p-JNK1/2 expression in the cancer lesions tend to be with earlier onset and poorer prognosis. It has been demonstrated that one mechanism for the tumor suppressor activity of BRCA1 is through its activation of JNK in breast cancer cells7 and it is presumably possible that the decreased p-JNK expression in cancer tissues, which leads to the lower JNK activity and the lower tumor suppressor function of JNK, may play a role in the development of breast cancer.
With the observation that the decreased p-JNK1/2 expression in the cancer lesions was correlated significantly with higher tumor grade and decreased age at diagnosis (Table II and Fig. 3a), it was puzzling to see that the lower p-JNK1/2 expression in breast cancer tissues was associated with a better overall survival (Fig. 4b). One example similar to this ambivalent observation for p-JNK is the case of Ras, which has either positive or negative effects on the regulation of apoptosis depending on cell context and multiplex signaling.37, 38
Alternatively, p-JNK1/2 may be merely an effecter of other signaling pathways rather than an independent player during the breast carcinogenesis. A recent report claims that mitogen-activated protein kinase kinase 4 (MKK4), a major kinase for JNK,39 has pro-oncogenic activity, instead of tumor suppressor activity, in breast tumors.40 Therefore, the sustained activation of JNK may be also a reflection of upstream signaling cascades, although in vivo evidence is lacking. Previous reports demonstrated that an elevated p-JNK expression plays a crucial role in the erbB2 (Her2/Neu)-facilitated expression of PEA3, cyclin D1 and Pin1 in breast cancer cells.41, 42, 43 However, our result does not support the idea that JNK works as a direct downstream effecter of erbB2 signaling pathway. First, no significant correlation between erbB2 and p-JNK1/2 was observed in this study (Table II). Meanwhile, the lower p-JNK1/2 expression was associated with a better overall survival in both erbB2-positive and erbB2-negative groups (p = 0.043 and 0.022, respectively) (data not shown). Second, our result that the high p-JNK1/2 expression was associated with a poor survival in the erbB2-positive group (p = 0.043 and data not shown) was inconsistent with the work by Guerra-Vladusic et al. who showed that the increased activation of JNK by heregulin (HRG) has a growth-inhibitory effect on breast cancer cells overexpressing erbB2.44 Taken together, our results implied that JNK1/2 might not be a direct in vivo effecter of erbB2-related signaling in breast IDC.
Other than a constitutive activation of JNK by autophosphorylation15 or other signaling pathways, an imbalanced activation of MAPK members may be one of the mechanisms for carcinogenesis. Some reports suggest that the fate of cell growth or death depends on the balance of the competing signal transduction pathways involving different MAPK family members.32, 33, 39, 45, 46 Accordingly, we found that p-JNK1/2 expression in breast cancer tissues was positively correlated with p-p38 expression, but not with p-ERK1/2 expression (Table III). We also observed that high p-JNK1/2 expresser in the p-p38 expression group was associated significantly with a lower overall survival (Figs. 5a and 5b). This observation was supported by the report that JNK as well as p38 are both able to increase the stability of the VEGF mRNA34 and promote tumor growth. Activation of both JNK and p38 may increase tumor angiogenesis and growth, thus leading to a poor outcome in breast cancer. Interestingly, we found that the patients with low p-JNK1/2 and p-ERK1/2 expression (n = 15) were all alive through the period of follow-up (Fig. 5c). It is possible that reduced p-JNK1/2 and p-ERK1/2 expression may improve the cure of breast cancer, since these 2 MAP kinases have been involved in the production of drug-resistance.47, 48 Taken together, our results provided information that different expression patterns of the MAPK family members might play a critical role in breast cancer.
In conclusion, we provided the information that an imbalance among the MAPK family members may play a role in the carcinogenesis of breast cancer and the p-JNK1/2 status may serve as an independent prognostic factor for breast IDC. Further studies are required to address the question whether the above observation is restricted to the IDC or also applied to breast cancer of other histology types.
This work was supported by grants to S.S.F.Y. (NHRI-EX93-9306BI) and to M.F.H. (NSC93-2314-B-037-082-).