Abstract
Background/Aim: Heat-shock protein 90 (HSP90) is a molecular chaperone that stabilizes many client proteins in normal and cancer cells, and approximately 3% of intracellular HSP90 is located in the nucleus. HSP90 is also targeted for the treatment of advanced non-small cell lung cancer (NSCLC). This study aimed to elucidate the clinical role of nuclear HSP90 levels in tissues from patients with NSCLC. Materials and Methods: Nuclear and total HSP90 levels were assessed using immunohistochemistry. Results: Multivariate logistic regression showed that the total HSP90 level was independently positively associated with age (p=0.041) and tumor histology of squamous type (p=0.007). By contrast, nuclear HSP90 level was independently positively associated with higher performance score (p=0.011), ever-smoking history (p=0.006) and presence of lymph node (p=0.036) or distant (p<0.001) metastasis, but not age or tumor histology. The level of nuclear HSP90, but not total HSP90, was negatively correlated with patient survival time (p<0.001). Conclusion: Nuclear accumulation of HSP90 might be a predictor of metastasis and survival in patients with NSCLC.
Lung cancer is an important cause of cancer death in both males and females worldwide (1). Adenocarcinoma and squamous cell carcinoma (SCC) are the major histological types of lung cancer and constitute about 80% of all non-small cell lung cancer (NSCLC) (1). Prognosis of patients with lung cancer can be predicted by classic clinical features such as age, gender, tumor histology, cancer stage at diagnosis, and surgical treatment (2, 3). Although significant improvements have been made in diagnostic tests, surgical techniques and therapeutic drugs, the 5-year survival rate remains alarmingly low at 15% (4). Tumor recurrence and metastasis are the primary reasons for such a low rate of survival. Although targeted therapies such as epidermal growth factor receptor inhibitors are very effective in NSCLC patients with specific genotypes, tumor relapse still occurs approximately 2 years after therapy because of acquired resistance (5-7). The overall survival of patients with stage I/II cancer is about 65%; however, nearly 40% will develop recurrence or metastasis after surgical resection (8).
Heat-shock protein 90 (HSP90) is a molecular chaperone existing in two isoforms: stress-induced HSP90α and constitutive HSP90β. HSP90 stabilizes many client proteins such as oncoproteins in the post-translational assembly of oligomeric complexes via an ATP-driven chaperone cycle regulated by a number of co-chaperones, such as HSP70 and p23 in normal and in cancer cells (9, 10). HSP90 clients are involved in tumorigenesis and the interactions of HSP90 with client proteins occur mainly in the cytoplasm (9, 10). Recently, Liu et al. found that overexpression of cytoplasmic HSP90 correlated with favorable overall survival (11). Furthermore, Biaoxue et al. demonstrated that up-regulation of HSP90β in the cytoplasm correlated with poor prognosis and lymphatic metastasis in patients with NSCLC (12). These findings imply that HSP90 plays a prognostic role in NSCLC.
Studies have shown that HSP90 protein expression occurs not only in the cytoplasm but also in the nucleus and in extracellular spaces (13-17). Approximately 3% of the intracellular pool of HSP90 is located in the nucleus (17). Human cell models have shown that the physiological interactions of HSP90 and steroid receptors occur in the nuclei of target cells (16, 18). Furthermore, Holowina et al. found that nuclear accumulation of HSP90 protein correlated withdecreased response to glucocorticoid drugs in patients with chronic pulmonary obstructive diseases (15). Since the physiological function of HSP90 in the cytoplasm differs from that in the nucleus, the clinical role of HSP90 may also differ depending on where it is expressed. However, no clinical studies have investigated the association between nuclear HSP90 and NSCLC as far as we are aware.
To elucidate the clinical role of total and nuclear HSP90 in NSCLC, we conducted immunohistochemical studies to detect differences in expression of HSP90 in tissue microarrays created from lung tissue samples from patients with and without NSCLC.
Materials and Methods
Study participants and tissue microarray (TMA). Lung tumor tissues were obtained from 190 patients who were treated for NSCLC at the Chung Shan Medical University Hospital during the period 1998-2008. TMAs were constructed using paraffin tissue blocks obtained from the Department of Pathology. All tissue samples had been obtained during surgery or by core biopsy, and each tissue core was 1.5 mm or 2 mm in diameter. Cytoplasmic and nuclear HSP90 levels were detected by immunohistochemistry. Twenty-five non-neoplastic lung tissue specimens were obtained from patients with spontaneous pneumothorax. This study was approved by the Institutional Review Board of the China Medical University Hospital (number: CMUH103-REC2-067).
Clinical features. Patient characteristics and clinical features at diagnosis, such as gender, age, performance score, smoking history, tumor histology and cancer stage, were verified either from the hospital medical records or by telephone interview. Performance score, also referred to as the Eastern Cooperative Oncology Group (ECOG) score, ranged from 0 to 5, with 0 denoting perfect health and 5, death. Smoking history was recorded as “ever” or “never”. “Never” smoking history was defined in patients who had never smoked previously. Individuals who currently smoked, or had ever smoked, were defined as “ever” smokers. For this study, only patients with SCC and those with AD diagnosed according to the World Health Organization classification criteria were enrolled (19). Patients with other tumor histological types were excluded because of the small number of cases. The samples were divided into stages I to IV according to the Tumor–Node–Metastasis (TNM) criteria outlined in the Cancer Staging Manual from the American Joint Committee on Cancer (20). At the time of our analysis in October 2012, a total of 114 out of the 190 patients with NSCLC had died. The follow-up period (mean=855 days) was defined as the time from diagnosis to death or last contact with the patient.
Immunohistochemistry. TMA sections (4 μm-thick) were dewaxed with xylene, rehydrated with serially decreasing concentrations of alcohol, and bathed in a phosphate buffer. Antigen retrieval was accomplished in a Tris-EDTA buffer (pH 9.0) for 20 min using a pressure cooker (Cell Marque Corp, Hot Springs, AZ, USA). Primary specific antibody to HSP90 (1:100 dilution; Leica Biosystems, Newcastle upon Tyne, UK) was used. Secondary antibody conjugated to a peroxidase-labeled polymer was purchased from Leica Biosystems. The sections were counterstained with hematoxylin (Muto Pure Chemicals, Tokyo, Japan). Mouse serum and phosphate buffer were used as negative controls.
Assessment of HSP90 immunoreactivity. The total immunoreactivity of HSP90-positive cells in each tissue core was assessed by the H-scoring system via imaging software (Aperio Technologies, Inc., Vista, CA, USA) to obtain an objective evaluation and avoid subjective interpretation. The grades of staining intensity were four-tiered as follows: 0, negative staining; 1, weak staining; 2, medium staining; and 3, strong staining. The H-score of each tissue core represented the sum of the mean value multiplied by the proportion of positively staining cells in each grade. The H-scores ranged from 0 to 300. This assessment included the staining intensity of cytoplasmic and nuclear immunoreactivity, hence the nuclear index was counted manually. Nuclear HSP90 positivity was defined as nuclear HSP90-positive cells exhibiting more intense staining in the nucleus than in the stroma or cytoplasm. The percentage of nuclear-positive cells among the tumor cells (at least 300 cells) was calculated by a pathologist.
The mean H-score of 25 normal bronchial/bronchiolar epithelia was 251.3, with a standard deviation (SD) of 36.7, and was considered the basal level of HSP90. A cut-off point, derived by subtracting the standard deviation from the mean, was used to categorize NSCLC samples into cancer with low- and high-expression levels of total HSP90. In addition, tumor tissues in which more than 5% of cells displayed positive nuclear staining were defined as high nuclear HSP90-expressing cancer; conversely, tumor tissues in which fewer than 5% of cells displayed positive nuclear staining were defined as low nuclear HSP90-expressing.
Statistical analysis. Descriptive statistics were used to calculate the frequencies of all clinical features, including age, gender, performance score, smoking history, tumor histology, TNM cancer stage at diagnosis, and status of metastasis. The Pearson chi-square test was performed to examine the associations between total or nuclear HSP90 levels and clinical features. Multivariate stepwise logistic regression was used to determine the independent factors associated with HSP90 level. The Kaplan–Meier plot and log-rank test were used to assess the association between levels of total and nuclear HSP90 and survival. Multivariate Cox regression was used to identify the independent prognostic factors. A two-sided p-value of less than 0.05 was considered significant. All the statistical operations were performed using the SPSS statistical software package (Version 14.0, SPSS, Inc., Chicago, IL, USA.).
Results
Study participants. The 190 patients with NSCLC comprised of 117 men and 73 women, with a median age of 67 years. The ECOG score in most patients was 0 or 1 and the majority of patients were “ever” smokers. Adenocarcinoma was the predominant type of NSCLC (64%) and most patients had advanced lung cancer, including 68 patients with distant metastasis. In samples from the 25 patients with non-neoplastic lung disease, the mean H-score for total HSP90 immunoreactivity was 251.3±36.7. No nuclear HSP90 immunoreactivity was observed in the 25 samples of normal bronchial/bronchiolar epithelia. In samples from the 190 patients with NSCLC, the mean total HSP90 H-score was 182.3±49.2 and the nuclear HSP90 level was 9.0±15.5%. The distribution of total and nuclear HSP90 levels stratified by clinical variables is shown in Table I.
HSP90 immunoreactivity in non-neoplastic lung and NSCLC tissues. HSP90-positive cells were brown in color. The acellular stroma was HSP90-negative. The normal bronchial and bronchiolar epithelia exhibited strong cytoplasmic HSP90 staining (Figure 1); no nuclear HSP90 staining was identified. Alveolar type I cells revealed negative staining and alveolar macrophages weak cytoplasmic staining. Furthermore, lung carcinomatous cells displayed either cytoplasmic or nuclear staining, or both (Figure 2).
Relationship between total/nuclear HSP90 level and clinical features in NSCLC. Of the 190 samples of NSCLC tissue, 62 samples had high total levels of HSP90 and 84 samples had high nuclear HSP90 level. High levels of HSP90 expression in cytoplasm and in the nucleus were seen in 30 samples.
Total, but not nuclear, HSP90 level was positively associated with male gender (p=0.030, Pearson Chi-square test) and SCC (p=0.016, Pearson chi-square test). Nuclear HSP90 level was associated with higher ECOG score (p<0.001, Pearson chi-square test), smoking history (p=0.015, Pearson Chi-square test), increasing TNM stage (p<0.001, Chi-square for linear-by-linear association) and lymph nodal/distant metastasis (p<0.001, Pearson chi-square test) (Table II).
Independent factors associated with total/nuclear HSP90 level. Age <65 years (p=0.041) and SCC (p=0.007) were independently positively associated with total HSP90 level in NSCLC. By contrast, higher ECOG score (p=0.011), smoking history (p=0.006), lymph nodal (p=0.036) or distant (p<0.001) metastasis, and SCC were independently associated with high nuclear HSP90 level (Table III). A high level of HSP90 expression in the nucleus may, therefore, be a predictor of metastasis in patients with early-stage NSCLC.
Nuclear accumulation of HSP90 positively correlates with patient survival. The log-rank test was used to estimate the effect of total or nuclear expression of HSP90 on survival of patients with NSCLC. The median overall survival time of patients with NSCLC was 770±178 days [95% confidence interval (CI)=520-1020 days]. Total HSP90 level was not associated with survival (p=0.288, log-rank test; Figure 3A). By contrast, nuclear HSP90 level was negatively associated with survival. The median survival time was 1025±230 days (95% CI=754-1656 days) for patients with low nuclear HSP90-expressing cancer and 358±93 days (95% CI=176-540 days) for patients with high nuclear HSP90-expressing cancer (p<0.001, log-rank test; Figure 3B). Cox proportional hazard model assessment revealed that the nuclear HSP90 level (hazard ratio=0.631, p=0.047), but not total HSP90 level (hazard ratio=1.012, p=0.958) was associated with death due to cancer (Table IV).
Discussion
To the best of our knowledge, this is the first study to investigate whether HSP90 expression in the nucleus is a prognostic marker in patients with NSCLC. We found that the level of total HSP90 was associated with patient age and tumor type and that the level of nuclear HSP90 correlated with smoking status, higher ECOG score, metastasis and poor survival. Accordingly, the clinical role of total HSP90 level was greatly different from that of the nuclear HSP90 level in NSCLC. These findings imply that nuclear, but not total, accumulation of HSP90 might be a predictor of tumor metastasis and patient survival.
HSP90 exhibits distinct activities in the cytoplasm and nucleus. Evidence has shown that nuclear HSP90 modulates the activity of steroid hormone receptors such as glucocorticoid receptor (GR) (21). Steroid hormone receptors shuttle between the cytoplasm and the nucleus, and demonstrate distinct regulatory or transcriptional activities in each compartment; accordingly, HSP90 participates in these processes (22). Cytoplasmic HSP90 maintains the GR in a state capable of binding hormone, and assembles hormone, GR and motor proteins for translocating the GR complex from the cytoplasm to the nucleus, where GR-specific genes are up-regulated (16, 18). Furthermore, in lymphoma models, HSP90 binds to B-cell lymphoma 6 protein (BCL6) and the complex in the nucleus can inhibit expression of tumor-suppressor genes, such as p53, and is involved in lymphoma formation (21). These findings suggest that HSP90 may have different activities in the cytoplasm and nucleus of normal or cancer cells, and further give a possible explanation for our present data showing that total HSP90 and nuclear HSP90 levels clearly differed by clinical features for NSCLC.
Nuclear HSP90 participates in chromatin remodeling, which is mediated by its binding to or dissociation from histones; additionally, the open chromatin conformation was explored at the promoter region of HSP90B gene, which was mediated by the action of the chromatin remodeling factors called the Swi/Snf complex (9). Hyperacetylation of nuclear HSP90α and extracellular HSP90 has been reported to increase tumor cell invasion in cell models of breast cancer (22), melanoma (23) and pancreatic cancer (24). Recently, Chen et al. revealed that an HSP90 inhibitor called PBT-1 inhibited growth and metastasis in lung adenocarcinoma cells (25). Therefore, increased nuclear HSP90 expression of HSP90 may be related to metastasis in patients with NSCLC.
Liu et al. found that HSP90 protein levels were higher in lung cancer tissues than in tissues from normal lungs or benign lung tumor tissues. They also found that a high level of HSP90 expression in lung cancer tissues correlated with favorable overall survival (11). Biaozue et al. found that levels of constitutive HSP90β were higher in lung cancer tissues than in normal lung tissues, and that such overexpression was correlated with poor survival and lymphatic metastasis in patients with NSCLC (12). In our study, most lung cancer tissues presented low levels of total HSP90 protein when compared with normal respiratory epithelia. In addition, there was no significant difference in survival or metastasis status by total HSP90 level. The differences in results between our study and the studies performed by Liu et al. (11). and Biaozue et al. (12). might be due to differences in control tissues used. In our study, control tissues comprised normal respiratory epithelia, whereas in the studies by Liu et al. and Biaozue et al., the control tissues comprised normal lung parenchyma. NSCLC has been suggested to be derived from bronchial or bronchiolar epithelia (26) and HSP90 regulates the proper folding of client proteins in many tissues (9, 27). Thus, we believe that it is reasonable to use normal bronchial/bronchiolar epithelium as a control tissue.
Kamal et al. found that the HSP90 protein level was higher in cancer cells than in normal cells, leading to the high tumor selectivity of HSP90 inhibitor 17-allylaminogeldanamycin (27); however, in their study, normal human respiratory epithelial cells or normal lung tissues were not used. To the best of our knowledge, no studies have investigated the expression of HSP90 in normal bronchial/brochiolar epithelia, lung parenchyma and lung cancer cells. Watterson et al. found that air-pollution particulate matter induced significant up-regulation of HSP90 in normal bronchial cells in response to the increased number of unfolded proteins induced by endoplasmic reticulum stress (28). In our findings, the total HSP90 level was lower in more than half of all lung cancer samples than in normal respiratory epithelia. This finding is possibly due to increased constitutive or stress-induced expression of HSP90 in respiratory epithelium.
Conclusively, nuclear accumulation of HSP90 may be a negative predictor of metastasis and survival in patients with NSCLC. Furthermore, the nuclear HSP90 immunohistochemical level may be used as a chemotherapeutic marker for HSP90 inhibitors.
Acknowledgements
This work was partially supported by research grant MOST 103-2314-B-039-036 from the National Science Council, Taiwan, ROC, and DMR-103-50 from China Medical University Hospital.
Footnotes
Conflicts of Interest
No financial or personal relationship disclosures are required for any of the Authors.
- Received February 25, 2016.
- Revision received April 11, 2016.
- Accepted April 13, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved