Abstract
Background/Aim: Protein connexin 43 (Cx43), a part of intercellular gap junctions, is frequently down-regulated in tumors. The aim of the study was to compare Cx43 expression in primary colorectal tumors of patients with stage III and IV disease. Patients and Methods: Immunohistochemical expression of Cx43 was analyzed in 50 colorectal adenocarcinomas from surgically-treated patients of stage pT3N1-2 without metastases (M0) and 50 specimens of the same pTN stage from patients with synchronous liver metastases (M1). Association of Cx43 expression with clinicopathological factors and tumor site was also analyzed. Results: There was no significant difference in Cx43 expression between M0 and M1 tumor specimens (p=0.817), as well as in Cx43 expression between colonic and rectal tumors (p=0.116), respectively. Stromal expression of Cx43 was higher in M1 than in M0 tumors (p=0.004). Conclusion: Stromal Cx43 expression is possible indicator of metastatic potential of colorectal adenocarcinoma.
Colorectal cancer (CRC) is nowadays considered a heterogenous, multifactorial disease rather than a single malignancy. To improve clinical management and treatment results, the focus of CRC research is shifting from a clinical perspective towards understanding the molecular basis of these tumors. Thus, a wide range of moleculogenetic factors, that could indicate tumor aggressiveness, predict individual risk of relapse and serve as a possible target for specific anticancer therapy, are currently under investigation.
Connexins (Cxs) are a family of 21 related, but distinct transmembrane proteins forming intercellular channels (gap junctions) (1). Gap junctions have an essential role in intercellular communication, adhesion, cell proliferation and cell differentiation by enabling direct exchange of small molecules (<1 kD) and subsequently biologic signaling between cells (2). Dysregulation of Cxs expression is thought to have a role in carcinogenesis but also in tumor invasion and metastasis (3, 4).
The best studied connexin isoform is connexin 43 (Cx43), known as a downstream target of β-catenin, a key component of Wnt signaling pathway (5). Cx43 is ubiquitously expressed in human tissues. It regulates cell growth and differentiation not only via gap junction-mediated intercellular communication but also by interaction with apoptotic factors or direct control of gene transcription in the cell nucleus (6). Cx43 is frequently down-regulated in tumors (4, 7). It has been shown that down-regulation of Cx43 can result in more aggressive growth of breast and lung tumors (8, 9). Conversely, forced expression of Cx43 in cells of various tumor types can restore normal cell growth and differentiation (10). This finding suggests the possible role of Cx43 as a tumor suppressor and a potential target in cancer prevention and therapy (11).
In normal colon, epithelial cells in the crypts of colon mucosa express Cx43 as a membrane protein (12). It has been shown that, during colorectal carcinogenesis, gradual loss of Cx43 expression occurs, as well as the shift of its localization from cell membrane to intracellular compartments, mainly cytoplasm or cell nucleus (13). Aberrant expression of Cx43 in CRC is the consequence of abnormal phosphorylation of Cx43 molecule, which occurs as the result of gene mutations, epigenetic or post-transcriptional events (14). Aberrant Cx43 expression correlates with loss of expression of cell adhesion molecules E-cadherin and β-catenin and has a possible role in loss of gap junctions-mediated intercellular communication, as well as in dediferentiation and invasion and, probably, metastatic potential of tumor cells (15).
Cx43 expression in primary colorectal carcinomas in patients with stage III and IV disease.
It has recently been suggested that Cx43 acts as a tumor suppressor in CRC and that loss of Cx43 expression can be considered a poor prognostic factor in stage I and II CRC (16). However, there is an evidence of correlation of higher level of Cx43 expression with more advanced stage of CRC (17). These findings warrant further investigation of Cx43 expression as a potential prognostic biomarker in CRC. To date, no direct comparison of Cx43 expression in primary colorectal tumors in patients with stage III and IV disease has been reported.
Patients and Methods
Patients and materials. This retrospective study was performed on tumor specimens of 100 patients, median age 64.0 years (interquartile range=58.0-70.0) who underwent surgical resection of primary colon and rectal cancer in the University Hospital Center Sestre milosrdnice between January 2006 and December 2013. Fifty patients had radiographically or intraoperatively confirmed synchronous liver metastases (M1) and 50 patients had no distant metastases at the time of diagnosis (M0). All resected primary tumors were adenocarcinomas and were pathologically staged as T3, N1-2 according to the American Joint Committee of Cancer (AJCC) TNM classification.
Clinical data were collected from each patient's medical records and included information on age, gender, localization of primary tumor (colon/rectum) and presence or absence of synchronous distant metastases. Information about T- and N- stage and other pathohistological prognostic factors of resected tumors (grade of differentiation, presence or absence of lymphovascular and/or perineural invasion and extranodal tumor deposits) were retrieved from pathological reports. The study was conducted in accordance with the ethical standards laid down in the Declaration of Helsinki and its amendments and was approved by the Ethics Committee of the University Hospital Center Sestre milosrdnice.
Methods. Tumor specimens were previously fixed in 10% buffered formalin, embedded in paraffin, cut at 5-μm thickness and routinely stained with hematoxylin and eosin. The diagnosis of adenocarcinoma was histologically confirmed in all cases.
Deparaffinization and immunohistochemical staining were performed following the Microwave Streptavidin ImmunoPeroxidase (MSIP) protocol. Connexin 43 immunohistochemical stains were performed with a polyclonal goat IgG antibody to Cx43 (sc- 6560, dilution 1:200; Santa Cruz Biotechnology Inc., Dallas, TX, USA). Immunoreactivity was visualized by using EnVision FLEX-PTL system on a Dako Tech-Mate™ Horizon automated immunostainer (DAKO, Copenhagen, Denmark).
Cx43 immunostaining was evaluated by two independent pathologists. Immunohistochemical staining was analyzed according to cellular localization of Cx43 (nuclear/cytoplasmic). The identification of areas with the highest staining intensity was done on the whole mount of the chosen slide under low (x40) magnification and Cx43 expression was evaluated under high magnification (×400).
Percentage of positively stained cells was assessed according to a 3-point scale: 0 (<10% positively stained cells), 1+ (10-50% positively stained cells) and 2+ (>50% positively stained cells). For the purpose of statistical analysis, the results of Cx43 immunostaining were expressed as Cx43-positive (level of expression 1+ and 2+) and Cx43-negative (level of expression 0).
Statistical analysis. Statistical analysis was performed using the Kolmogorov- Smirnov test, Kruskal- Wallis test and χ2 test. Correlation between clinicopathological factors and Cx43 expression was analyzed using the Kendall tau_b coefficients of correlation. The levels of statistical significance were set at least at p<0.05. Computations were performed using data analysis software system IBM SPSS Statistics, version 21.0 (www.spss.com).
Clinicopathological characteristics of patients with stage III and IV colorectal carcinoma.
Results
Cx43 expression. The results of the Cx43 expression analysis in primary colorectal tumors of patients without and with synchronous liver metastases are summarized in Table I. Nuclear expression of Cx43 was not observed in any tumor specimen, thus this parameter was not further analyzed. Positive Cx43 immunostaining was confirmed in 25 (25%) tumor specimens and was exclusively cytoplasmic, while 75% of specimens were Cx43-negative (Figures 1a-c). We observed positive Cx43 staining of peritumoral stromal cells in 60 (60%) specimens and, therefore, we decided to include this parameter in statistical analyses (Figures 2a-c).
Cytoplasmic Cx43 staining was positive in 12 (24%) M0 and 13 (26%) M1 tumor specimens, respectively. Strong cytoplasmic Cx43 immunoreactivity was detected only in 2 (4%) tumor samples from patients with synchronous metastases, while the distribution of negative and intermediate expression was equal between M0 and M1 group. Expression of Cx43 in peritumoral stroma was found in 23 (46%) M0 and 37 (74%) M1 tumor specimens, respectively. In M1 group, 7 tissue specimens (14%) showed strong and 30 specimens (60%) moderate Cx43 expression. In contrast, in M0 group, strong Cx43 expression in peritumoral stroma was detected in only 3 (6%) specimens, 20 specimens (40%) showed moderate expression, while over 50% of tumors showed no Cx43 expression. Patients with synchronous liver metastases had significantly different stromal Cx43 expression compared with patients without synchronous distant metastases (p=0.004), while there was no significant difference between groups regarding cytoplasmic Cx43 expression (p=0.817).
Cx43 expression in colon vs. rectal cancer. The analysis of colon and rectal cancer specimens showed no significant difference regarding localization of primary tumor, neither for cytoplasmic (p=0.116), nor stromal expression of Cx43 (p=0.392). Cytoplasmic expression was shown in 13 (20%) colon cancer specimens and 12 (34.3%) rectal cancer specimens, while stromal positivity was detected in 41 (63.1%) colon and 19 (54.3%) rectal cancer specimens.
Patients' characteristics. Clinicopathological features of patients in M0 and M1 group are shown in Table II. No significant correlation was found between M status and patients' gender (p=0.673), grade of tumor differentiation (p=0.900), presence of perineural invasion (p=0.790) and lymphovascular invasion (p=0.398). Patients with synchronous liver metastases had significantly higher N-stage (p=0.024) and higher incidence of extranodal tumor deposits (p=0.049).
Correlation of clinicopathological prognostic factors with Cx43 expression. Neither cytoplasmic nor stromal expression of Cx43 correlated significantly with known clinicopathological prognostic factors of CRC: N-stage (p=0.906; p=0.558), grade of differentiation (p=0.058; p=0.428), lymphovascular invasion (p=0.223; p=0.058), perineural invasion (p=0.442; p=0.232) and presence of extranodal tumor deposits (p=0.258; p=0.568), respectively. However, cytoplasmic Cx43 expression was associated with a trend towards a significantly positive correlation with stromal expression of Cx43 (p=0.059).
Analysis of correlation coefficients confirmed a clear association between stromal Cx43 expression and presence of synchronous metastases (p=0.004) and lymphovascular invasion (p=0.029), as well as correlation between cytoplasmic Cx43 expression and higher tumor grade (p=0.022).
Discussion
Aberrant expression of connexins may contribute to the loss of intercellular communication via gap junctions. Previous studies have shown that loss of Cx43 expression and a shift from its membranous to intercellular localization frequently occurs during colorectal carcinogenesis as a result of Cx43 down-regulation (13). Various authors have reported the correlation of Cx43 expression and known clinicopathological prognostic factors of CRC, suggesting the possible role of Cx43 as a potential biomarker of biological aggressiveness of CRC (16-18). Sirnes et al. recently provided the first and robust clinical evidence of the tumor suppressor role of Cx43, as well as of its prognostic value. The authors analyzed colorectal adenocarcinomas from 674 patients and showed aberrant expression of Cx43 in tumor cells (diffusely in cytoplasm, nucleus or intracellular vesicles) and peritumoral stroma. Loss of Cx43 expression was associated with shorter overall survival (OS) (p=0.017) and progression-free survival (PFS) (p=0.017) in the whole population of patients, as well as in patients with stage I and II disease, while there was no significant correlation between Cx43 expression and OS/PFS for patients with stage III and IV CRC. To our knowledge, there are no published data on possible correlation of Cx43 expression in primary colorectal tumors of patients with stage III and IV disease, which was the aim of our work.
Cx43 expression in epithelial cells of colon adenocarcinoma. In most colon adenocarcinoma specimens no Cx43 immunostaining or weak staining (0) was found in tumor cells (a, ×200). When detected, moderate (1+) and strong (2+) Cx43 immunoreactivity was intracytoplasmic and diffusely distributed (b, ×200 and c, ×200).
Expression of Cx43 in peritumoral stromal cells: negative (0) (a, ×200), moderate (1+) and strong (2+) immunohistochemical staining (b, ×200; c, ×200).
In our study, no nuclear expression of Cx43 was shown. Expression of Cx43 in tumor cells nuclei is thought to have a role in regulation of gene transcription. In the above mentioned study, Sirnes et al. found nuclear Cx43 positivity in 58% of tumor samples of patients with all stages of CRC; however, data about distribution of Cx43 nuclear positivity between stage III and IV specimens are lacking. Considering the same methodology of tumor tissue preparation and immunostaining, in our work, the absence of detection of nuclear Cx43 immunostaining could be explained by the relatively low number of patients included in the present study.
Aberrantly accumulated connexins in the cytoplasm of tumor cells lose their physiological functions as transmembrane proteins and can act as signaling molecules in control of tumor growth (19). Many authors have reported about aberrant Cx43 expression in various cytoplasmic compartments or complete loss of its expression in colorectal cancer cells but with no clear association of Cx43 expression with disease stage, mainly due to low number of included patients (13, 17, 18). Our results showed no significant difference in cytoplasmic Cx43 expression between M0 (stage III) and M1 (stage IV) group. The incidence of Cx43 positivity in our series (24% in M0 group and 26% in M1 group, respectively) partially coincide with series of Ismail et al., but is clearly opposite to results of Sirnes et al. who found 89% Cx43-positive cases among 295 analyzed stage III and IV tumors (16). This can be possibly explained by the low number of patients included in our study and differences in interpretation of results of immunostaining. The authors of the aforementioned study used combined scoring system (5-graded scale of percentage of positively stained cells and 4-graded scale of staining intensities), while in our work we used a simplified 3-graded scale of percentage of positively stained cells with no evaluation of staining intensity.
During our work we unexpectedly noticed a high incidence of Cx43 immunostaining of peritumoral stromal cells (60% of all tumor samples). In their microenvironment, tumor cells interact with non-tumor cells (mainly lymphocytes and stromal cells). Paracrine signals between tumor cells and stromal cells are mediated by circulation of nucleotides (NAD+ and ATP) through gap junctions and result in increase of intracellular levels of Ca2+ (20). Homeostasis of Ca2+ is extremely important in the control of cell growth; the increase of intracellular level of Ca2+ can result in evasion of apoptosis, cell migration and promotion of neoangiogenesis (21, 22). Moreover, transmembranous transport of nucleotides has an important role in activation of antitumoral immune response mediated by peritumoral lymphocytes (23). Han et al. first described the expression of Cx43 in stromal cells close to the neoplastic epithelium in colorectal cancer, while Husøy et al. reported about peritumoral stromal expression of Cx43 in mice with multiple intestinal neoplasia (24). They showed that Cx43 expression in stromal myofibroblasts coexists with higher expression of COX-2, which is also highly expressed in human and animal adenomas of the colon. Considering the fact that myofibroblasts produce signaling molecules, that promote migration and invasion of epithelial cells, and that stromal expression of Cx43 is also reported in human breast cancer, Han et al. have assumed that peritumoral stromal Cx43 expression could have a role in epithelial-mesenchymal transition during colorectal carcinogenesis. Our results showed significantly higher stromal Cx43 expression in patients with synchronous liver metastases than in patients without metastases, suggesting the possible role of stromal Cx43 positivity as a biomarker of metastatic potential of CRC.
Expression of Cx43 in colon versus rectal tumors was investigated only by Kanczuga-Koda et al. (13). In our work, as in the above mentioned study, no significant difference in Cx43 expression was found between tumors localized in colon and rectum. These findings are concordant with the results of the whole-genome analysis in CRC performed by the Cancer Genome Atlas Network Group, which showed very similar, nearly identical patterns of genomic alterations in adenocarcinomas localized in colon and rectum (excluding hypermutated tumors localized in the right colon) that may be considered biologically identical tumors (25).
Many authors analyzed correlation of Cx43 expression with known pathological prognostic factors (13, 16, 18). Loss of Cx43 expression was frequently found to be associated with more advanced tumor stages. Our results show a trend towards statistically significant correlation of higher cytoplasmic Cx43 expression and higher tumor grade but, due to the low sample size, no conclusion can be drawn based upon these data. A similar trend was also shown regarding stromal Cx43 positivity and presence of lymphovascular invasion. This fact could indicate that stromal expression of Cx43, as well as lymphovascular invasion, is the possible indicator of biological aggressiveness of tumor, based on the role of Cx43 in tumor angioinvasion process, as has been experimentally proven on breast cancer cell lines (26). Increased expression of Cx43 has also been reported in lymph node metastases compared to primary tumors of breast and skin, suggesting a potential role of Cx43 in lymphatic tumor spread (27, 28).
It seems that, in colorectal carcinogenesis and in earlier stages of CRC, Cx43 acts as a tumor suppressor, but in later stages of the disease it has a role in tumor promotion. Our work is, to our knowledge, the first report of possible association of cytoplasmic and stromal Cx43 expression in CRC and warrants further prospective studies. Clear correlation of stromal Cx43 expression in primary tumors and presence of synchronous metastases, as shown in our work, could, if confirmed on larger number of patients, serve as the basis for further studies for the possible role of stromal Cx43 expression as an indicator of metastatic potential in stage II and III patients who are candidates for adjuvant systemic treatment.
Footnotes
Financial Support
None.
Conflicts of Interest
The Authors report no conflicts of interest.
- Received February 16, 2016.
- Revision received March 24, 2016.
- Accepted March 28, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved