Abstract
Background: The prognostic significance of S100A as potential biomarker for breast cancer was reported; however, this finding has recently been challenged. Here, the aim was to assess whether S100A4 could also be a prognostic biomarker of lung cancer. Materials and Methods: A specific high-titer anti-S100A4 monoclonal antibody was developed. The utility and specificity of this antibody was validated by immunostaining experiments. The antibody was tested against a newly developed high-density tissue microarray including 400 lung cancer tissues to examine the clinico-pathological and prognostic significance of S100A4 in lung cancer. Results: The staining of S100A4 was significantly associated with patients' poor prognosis in lung squamous cell carcinoma but not lung adenocarcinoma. Conclusion: S100A4 seems to be a prognostic biomarker of lung squamous cell carcinoma (5-year survival rate of 38.5% versus 7.4%, p<0.01), but not of adenocarcinoma.
The current trend in cancer treatment is shifting from uniform protocols to personally customized methods. The prediction of prognosis or therapy response has become central to the efficiency of customized cancer therapies. Prognosis is an important factor used to decide the extent of surgery. Immunohistochemistry is an excellent technique to link basic research data to therapy and is critical for the identification of specific proteins overexpressed in malignant tissues. Nevertheless, the prognostic significance of biomarkers identified using immunohistochemical analysis needs to be carefully validated by judging the specificity and sensitivity of the immunostaining protocol. Only after rigorous testing, can one determine if these proteins can be considered as suitable candidates as prognostic/predictive markers.
Lung cancer has been reported to have the highest worldwide incidence of all types of cancer (1). Considering its high incidence of occurrence and the large range of therapeutic approaches provided to lung cancer patients, it has become clear that there exists a pressing need for additional prognostic/predictive markers (2, 3).
The S100A4 protein, a member of the S100 family of calcium-binding proteins, has been reported as a candidate prognostic marker of cancer. S100A4 protein is a small-molecular-weight (12 kDa) protein that has two domains: an S100 domain and a calcium-binding domain (EF domain) (4). This protein does not have a known enzymatic activity; however, its binding to several known proteins is indicative of its function. S100A4 binds to actin, to non-muscle myosin, and to the tumor suppressor p53 protein (5-8). This interaction may increase the cell motility and modulate the function of p53 (9).
Expression of S100A4 in breast cancer tissues has been associated with a worsened disease prognosis. For example, one large cohort study that analyzed 349 stage I and II breast cancer patients reported that 80% of S100A4-negative patients (versus only 11% of S100A4-positive patients) survive more than 19 years (10). A similar prognostic trend has been observed in several other tumor types (11-13); however, discrepant results have also been reported (14). Consequently, insufficient data are available to readily use this biomarker for general clinical use. Additional studies are therefore necessary to evaluate the prognostic significance of S100A4 expression and its usefulness as a prognostic marker.
The clinical significance of S100A4 in lung cancer has not been reported. This paper investigates whether S100A4 could be useful as a prognostic maker in non-small cell lung cancer using immunohistochemical staining of a high-density tissue microarray (TMA). A monoclonal antibody that specifically reacts against the carboxyl-terminal residues of the human S100A4 protein was developed. This antibody reacts exclusively with the S100A4 protein and does not cross-react with the other members of the S100 family of proteins. The antibody was then applied to a high-density TMA, which has recently been recognized as a powerful high-throughput tool in surveying many tumor tissues under same conditions (15). The TMA used here for immunostaining contained 400 lung cancer cases. The staining results obtained for the tumor tissues were correlated with the clinical data of patients, including survival rates.
Materials and Methods
Preparation of the S100A4 antigen and production of a monoclonal antibody. The DNA corresponding to the entire open reading frame of the human S100A4 protein was amplified by PCR using the cDNA of HEK293 cells as a template. The primers used were 5′-GCGGGTACCATGGCGTGACCTCTGGAG-3′ and 5′-GCGCT CGAGTTTCTTCCTGGGCTGCTTATG-3′.
The PCR products were extracted from the agarose gel and purified. After digestion of the products with Kpn1/Sal1, the resulting fragment was inserted into the Kpn1/Sal1 site of the pFlag-Omp vector (Sigma, St. Louis, MO, USA). The S100A4 protein was expressed in Escherichia coli and purified using an anti-Flag M2 Affinity gel (Sigma) according to the manufacturer's protocol. The purified S100A4 protein was used as the immunogen to produce monoclonal antibodies in the mouse. The monoclonal antibodies were obtained via a procedure described previously (16). Here, the mouse myeloma cell line P3U1 was used.
Recombinant protein expression and Western blotting. Using methods similar to those employed in the preparation of the S100A4 protein, S100 family proteins and various fragments of S100A4 protein were cloned into the pGEX vector (GE Healthcare, Buckinghamshire, UK). The recombinant proteins were then expressed in E. coli and purified using glutathione-Sepharose 4B resin (GE Healthcare), according to the manufacturer's protocol.
Western blotting was performed as described previously (17), using a mouse monoclonal anti-S100A4 antibody or a goat polyclonal anti-GST antibody (GE Healthcare) as the primary antibody, and a horseradish peroxidase (HRP)-conjugated rabbit anti-mouse (Zymed, CA, USA) or rabbit anti-goat IgG (Zymed) as the secondary antibody, respectively. The peroxidase reaction was detected using the ECL kit (GE Healthcare) and X-ray film (Fujifilm, Tokyo, Japan).
Clinical samples and tissue microarray (TMA). A TMA containing 400 lung cancer tissue samples and 100 normal lung tissues was constructed as previously described (18). In short, 400 lung cancer cases were selected randomly from tissue archives of the Laboratory of Pathology of Toyama University Hospital, Japan. The most representative tumor areas and normal lung areas from the same block were selected based on matching hematoxylin and eosin-stained slides. Cylindrical plugs of tissue (0.6 mm in diameter) were cored from the selected region of the donor block and transplanted directly into the recipient block using coordinated wells. Four-micron-thick sections were cut using an adhesive-coated tape and transferred on to a glass slide. When the sections were completely dry, the adhesive tape was removed using xylene in a coplin jar. The slides were then immunohistochemically stained as described above. The S100A4 signal was scored by two independent observers (SK and JF), including a trained pathologist (JF). Tissue cores of the TMA slide containing only a small number of tumor cells, necrotic changes, or uncertain histology were excluded from the analysis. All cores of the TMA were validated using immunostaining against vimentin, which is a member of the intermediate filament family of proteins and should be present ubiquitously in all mesenchymal cells. Cores with weak or no vimentin staining were excluded from the analysis (to exclude the possibility of insufficient antigen preservation).
All materials were used in accordance with the protocols approved by the Review Boards of the Ethics Committee of Toyama University. (No19-18)
Immunostaining of cell lines and TMA. Immunostaining and Western blotting analyses were performed to evaluate S100A4 immunostaining accuracy, using three cell lines (WI38 and MDA-MB453 obtained from RIKEN, Tsukuba, Japan; MDA-MB231 from ATCC, Virginia, USA). For immunofluorescent staining, the cells were cultured with MEM (Sigma) or L-15 (Sigma) supplemented with 10% fetal bovine serum. After washing, the cells were fixed with 4% formaldehyde for 30 min, treated with 0.1% Triton® X-100, blocked with 1% bovine serum albumin, then treated with Mab-51D12 followed by the treatment with FITC-labeled anti-mouse IgG (Zymed). The cells were observed under the fluorescent microscope (Olympus, Tokyo, Japan). Further, a formalin-fixed, paraffin-embedded cell block was constructed as previously described (19). Sections from a TMA containing 50 tissue types were mounted on the same glass slide, and 12 immunostaining protocols and 3 titrations of the antibody were examined to select the optimal protocol (see examples in Figure 3).
Immunohistochemical staining. Four-micron-thick TMA sections were used in all analyses. The sections were deparaffinized with xylene and rehydrated through graded alcohol incubations into water. They were then processed following the optimal staining protocol, as described above. In short, heat-induced epitope retrieval was performed using a Decloakig Chamber (DAKO, Kyoto, Japan), in which tissues were heated to 125°C then cooled to 90°C in Tris/EDTA buffer at pH 9 (Target Retrieval Solution; DAKO). After heat-induced epitope retrieval treatment, endogenous peroxidases were blocked with Peroxidase-Blocking Solution (DAKO) for 5 min. The sections were then incubated with anti-S100A4 antibody (1:50 dilution) for 60 min, followed by incubation with EnVision+ Dual Link system, according to the manufacturer's recommendations (DAKO). The reaction products were visualized with DAB+ (DAKO). The nuclei were lightly counterstained with Mayer's hematoxylin solution. All procedures were carried out at room temperature.
Scoring of the TMA staining and statistical analyses. The S100A4 staining of cancer cells was scored using a method described previously (18). Briefly, the distribution score (DS), which reflects the distribution of the positive signal was scored as 0 (0% of nuclei with positive staining), 1 (1-50% of nuclei stained) and 2 (51-100% of nuclei stained), to assess the percentage of tumor cells positive for S100A4 among all tumor cells. Staining intensity (IS) was scored as 0 (no signal), 1 (weak signal), 2 (moderate signal) or 3 (marked signal). The total score (TS) was obtained by adding DS and IS. Throughout this study, TS0, TS1 and TS2 were regarded as negative, whereas TS3, TS4 and TS5 were regarded as positive.
The specific reactivity of Mab-51D12 to S100A4 protein in S100 family. The eight GST-tagged S100 family proteins were expressed in E. coli, and purified by use of glutathione-Sepharose 4B resin. Each purified protein was subjected to SDS-polyacrylamide gel and Western blotted with anti-GST, Mab-47E3 and Mab-51D12 followed by HRP-labeled secondary antibody. The reacted bands were visualized by ECL method. The molecular weights of S100 family were detected by anti-GST antibody between 36 kDa and 38 kDa as GST (25 kDa)-fused proteins. Both Mab-47E3 and Mab-51D12 react exclusively with S100A4 protein.
Comparisons of clinicopathological factors (i.e. gender, age at diagnosis, histological type, pathological stage, differentiation, tumor stage, pathological stage, lymph node metastasis and metastasis) between positive- and negative-staining groups were performed using Chi-squared tests. The log-rank test was used to compare survival distributions between the positive- and negative-staining groups, and Kaplan-Meier curves were plotted for the two groups. The clinical factors were accounted as being reasonable by fitting Cox's proportional hazards models.
Results and Discussion
Specificity of the anti-S100A4 monoclonal antibody. Eleven of the produced monoclonal antibodies reacted with the S100A4 protein (12 kDa) in the breast cancer cell line MDA-MB231. Two of the antibodies (Mab-47E3 and Mab-51D12) were selected and their specific reactions against S100A4 were verified. The S100 family comprises approximately 19 members (20). Among them, seven members have both the S100 and EF domains and exhibit high sequence similarity with the S100A4 protein; therefore, whether these antibodies reacted exclusively with S100A4 protein was assessed. The seven S100 family members (S100A1, S100A5, S100A6, S100A7, S100A15, S100B and S100Z) were expressed in E. coli using the pGEX vector and were purified using glutathione-Sepharose beads. The expressions of these proteins were checked by anti-GST antibody (Figure 1). All of these proteins show molecular weights between 36 and 38 kDa including GST-tag (MW ~25 kDa). Figure 1 shows that Mab-47E3 and Mab-51D12 reacted exclusively with the S100A4 protein by Western blotting. The reactive epitopes were then verified using various lengths of dissected recombinant S100A4 protein. Figure 2a shows the Western blotting of the fragments of the S100A protein using the two monoclonal antibodies. All peptides used here contained a GST tag at the N-terminus, which was used for the detection of each peptide in the gel (membrane). Figure 2b shows a schematic diagram of the S100A4 peptide fragments and the reactivity of these antibodies. The Mab-47E3 clone reacted with a peptide located 20 amino acid residues from the N-terminus, which includes the S100 domain. The Mab-51D12 clone reacted with a peptide located 20 amino acid residues from the C-terminus. This region does not contain the S100 or the calcium-binding (EF) domains.
Even with the highest titration (1:10) and multiple trials of staining protocol, Mab-47E3 antibody reacted less frequently with S100A4 protein than the Mab-51D12 antibody did in TMA (data not shown). Furthermore, the Mab-51D12 antibody reacts with the non-structural carboxyl terminus and thus was considered to be more suitable for immunohistochemistry than the Mab-47E3 antibody that reacts with the S100 domain.
Validation of S100A staining using cell lines and TMA. Immunostaining and Western blotting analyses were performed to evaluate S100A4 immunostaining accuracy, using three cell lines: human diploid fibroblast WI38 and breast cancer epithelial cell lines MDA-MB231 and MDA-MB453. The S100A4 protein was detected only in MDA-MB231 cells by western blotting, but not other cell lines (Figure 3a). By immunofluorescent staining of the cells (Figure 3b, c), the MDA-MB231 cells stained positively, but WI38 cells were not stained. Figure 3d-g shows the staining of paraffin-sections of the cell blocks of those cell lines and tissues from TMA by the optimal protocol used here. The MDA-MB231 cells were stained positively but MDA-MB453 were not, as expected from the results of western blotting and fluorescent staining. Figure 3f and g show representative positive and negative cases of SCC, respectively. In Figure 3g, internal positive staining was observed in the interstitium.
Staining characteristics of S100A4 in lung cancer. A total of 42.2% of all tumors analyzed were homogenously stained with S100A4 antibody. The expression of S100A4 in carcinoma cells was granular and relatively low when compared with that seen in stromal non-tumor cells. As described above, such strong staining signals observed in surrounding stromal tissues, especially prominent in lymphocytes, myofibroblasts and macrophages, were found in all cases (Figure 3f and g).
Characterization of the TMA. A high-density TMA containing 400 lung cancer tissues was constructed. The clinical characteristics of these patients are summarized in Table I. Overall, 284 cases [181 adenocarcinoma (AD) and 103 squamous cell carcinoma (SCC) samples] were scorable, and clinical information was available for 268 cases (175 AD and 93 SCC samples). Age at diagnosis was available for all cases, with an average of 65.3 years (range, 35-97 years). Gender was available for 266 cases, which included 84 female and 182 male patients. TNM and pathological stage data were available for all cases. Follow-up time or time to death was available for 200 cases, with a median follow-up of 2.7 years. Ninety-six patients were alive at the most recent follow-up. A correlation between lymph node metastasis and survival length was found in AD but not in SCC.
Correlation between S100A4 expression and patient survival. The Mab-51D12 antibody was used to stain the TMA, as this antibody, which reacts with the non-structural C-terminus of S100A4, was thought to be more specific and stronger than the Mab-47E3 antibody, which reacts with the structural domain. The S100A4 staining of the carcinoma samples of the TMA was assessed by microscopic observation, as described in Materials and Methods. Table II provides a summary of the staining of the TMA with S100A4. A significant correlation was detected between S100A4 staining and metastasis to lymph nodes (pN): the data show that the S100A4-expressing SCC samples had a decreased rate of metastasis to lymph nodes. Although metastasis of SCC to lymph nodes is not prominent in this cohort, this result was not expected, based on the argument that S100A4 protein positively regulates metastasis. This result provides evidence that the S100A4 protein may not play a role in metastasis for particular cancer types.
The clinical history of all samples was investigated and a survival curve was plotted for each case of the positive or negative staining groups, as shown in Figure 4. These data revealed a significant negative correlation (p=0.0009) between survival length and S100A4 positive staining in SCC, but not in AD. These results suggest a poor prognosis for SCC patients whose lung cancer tumor tissues express S100A4. These data indicate that the detection of S100A4 expression may be a promising SCC prognostic marker.
The S100A4 protein has been shown to modulate the expression of p53-responsive genes by binding to the tumor suppressor protein p53. This function could represent a mechanism for the aggressive proliferation of tumor cells. In addition, S100A4 promotes the motility of cells by binding to actin or non-muscle myosin. This increase in cell motility also represents an advantage of proliferation. These functions of the S100A4 protein seem to concur with the inverse correlation observed between its expression levels and the survival of SCC patients. The adverse effect of increased cell motility does not directly imply an increase in metastasis of the tumor to lymph nodes. Many other factors, including expression of the laminin receptor at the cell surface and an increase in MMP9 activity, are thought to be crucial to the process of metastasis; therefore, it was not surprising to find an inverse correlation between the expression levels of S100A4 and lymph node metastasis. It is possible that the S100A4 protein contributes mainly to the increase in the aggressiveness of the tumor (at the proliferative level), and does not affect metastasis in lung SCC, in which metastasis is not very prominent.
Recently three papers have been published describing the correlation between survival rate and S100A4 high expression in lung AD (21-23). In one of them, there was no significant difference between survival rate and S100A4 high expression (21). The data shown here confirm their results. In a second paper, the authors showed a weak correlation between high expression of S100A4 and survival rate in lung AD (22). However, in the presented data, significance in prognosis was not detected between them. The reason for this discrepancy may lie in the detection rate of high S100A4 expression. In their paper the S100A4-positive cases were only 20%, and in contrast, this study contains 50% S100A4-positive cases. This discrepancy could also have resulted from the different antibodies used in both studies, and also from the different techniques used to evaluate and score the immunostaining results. As indicated, a very sensitive as well as specific antibody was selected for the staining procedure. A final paper showed poor prognosis in S100A4-positive cases compared to S100A4-negative cases in non-small cell lung carcinoma (23). Although the statistical significance of their data is not strong (p<0.05), their results are consistent with those presented here.
Clinicopathological parameters.
For the clinical use of S100A4, the immunostaining method is thought to provide more accurate prognostic information than the Western blotting method, as tumor and normal cells coexist in the same tissue samples where normal stromal cells are strongly positive.
Summary of the distribution of S100A4 positive and negative patients.
Finally, Cox's proportional hazards model was used to confirm whether S100A4 expression was correlated with the prognosis of patients. After adjustment for gender, age and cancer stage, it was found that age (p=0.0098), cancer stage (p=0.0242) and S100A4 expression (p=0.0022) were independently associated with SCC patients' survival. In contrast, in AD cases, only gender (p=0.0412) and cancer stage (p=0.0002) showed a significant correlation with survival (no correlation was found between S100A4 expression and survival in these patients).
Nevertheless, these data is based upon a retrospective study and a relatively small sized cohort from a single institute. For more practical use, the significance of S100A4 in SCC needs to be confirmed with additional cohort studies.
The detection of epitope of S100A4 by Mab-47E3 and Mab-51D12. A) S100A peptides having various lengths were expressed in E. coli as GST-fusion proteins. Each peptide was Western blotted as in Figure 1. Mab-51D12 reacts only with complete form of S100A protein. On the other hand, Mab-47E3 reacts with all peptides except EF domain. B) Schematic diagram of each S100A4 peptide. From the results shown in Figure 2a, Mab-47E3 and Mab-51D12 seem to react with 4 to 20 and 82 to 101 amino acid residues from the N-terminus of S100A4 protein, respectively.
Immunostaining of cell lines and the tissues of the tissue microarray (TMA). a) MDA-MB231, MDA-MB453 and WI38 cells were Western blotted by use of Mab-51D12. b) (1) MDA-MB231 cells without staining, (2) MDA-MB231 cells stained with Mab-51D12 showing positive cytoplasmic signals. c) (1) WI38 cells without (1) and with (2) staining with Mab-51D12. WI38 cells are not stained with Mab-51D12. d) MDA-MB231 cells in the paraffin-embedded cell line block are positively stained with Mab-51D12, while MDA-MB453 cells are not stained with Mab-51D12. f) Representative tissue cores of SCC from TMA showing diffuse positive staining with Mab-51D12. g) Representative case of SCC from TMA stained negatively with Mab-51D12. Note internal positive staining in the stromal tissues.
Kaplan-Meier survival curves demonstrating the relationship between patient survival and protein expression of S100A4. a) S100A4 staining in SCC patient shows significantly poorer prognosis (p=0.009), while staining in AD patient (b) does not show prognostic significance.
Acknowledgements
We would like to express our gratitude for the strong support from the members of the Laboratory of Pathology, Toyama University Hospital. We are also grateful to Dr. Stephen P. Henry, M.D. Anderson Cancer Center, for kindly editing the paper. This work was partly funded by the Japanese Ministry of Health, Labor and Welfare.
Footnotes
- Received December 23, 2008.
- Revision received February 24, 2009.
- Accepted March 23, 2009.
- Copyright© 2009 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
