Abstract
Background/Aim: Chloride intracellular channel protein (CLIC1), E- and P-cadherin (Ecad, Pcad) are certified factors of aggressivity, but they have not been studied in breast cancer to date. The aim was to study CLIC1, Ecad and Pcad impact on breast cancer in terms of defining new high-risk subgroups. Materials and Methods: Ninety-seven breast cancer biopsies were immunohistochemically evaluated for CLIC1, Ecad and Pcad expression related to molecular subtypes. CLIC1 expression was assessed in both tumor cells (CLIC1T) and blood vessels (CLIC1V). Results: For 23% of Luminal A cases, both cadherins and CLIC1V were positive. Luminal B/HER2 subtype, had two specific phenotypes: Ecad–/Pcad–/CLIC1T–/CLIC1V+ and Ecad+/Pcad–/CLIC1T–/CLIC1V+. All TNBC cases were clustered into two subgroups: 60% were Ecad+/Pcad+/CLIC1T+/CLIC1V+) while 40% were Ecad+/Pcad+/CLIC1T+/CLIC1V–). Conclusion: CLIC1, Ecad and Pcad association stratifies molecular types of breast cancer in subgroups that may explain different response to therapy and different aggressiveness previously observed by other authors within the same molecular subtype.
E-cadherin (Ecad) and P-cadherin (Pcad) are well known adhesion molecules with a certified role in promoting tumor invasion, metastasis and stemness potential. Ecad is expressed in luminal cells while Pcad is restricted to the basal cells of the normal mammary gland epithelium. This well defined expression in the normal epithelial cells of the breast suggests a differential involvement in breast malignant pathology. Ecad is incriminated as having a major role in the epithelial mesenchymal transition and Pcad is responsible for increasing breast cancer aggressiveness by its expression in cancer stem cells (1) and in predicting an unfavorable prognosis in axillary metastatic cases (2).
It becomes more obvious that Ecad and Pcad should be included in a future molecular classification of breast cancer. Breast cancer molecular heterogeneity, different response to chemotherapy, early development of resistance to therapy observed for some cases, and the long-term unpredictable prognosis are major reasons suggesting that new prognostic and therapeutic factor validation is needed.
Chloride intracellular channel protein 1 (CLIC1) is known to influence tumor agressiveness, metastasis and cancer cells stem potential in glioblastomas (3, 4) and recently it was shown that the CLIC1 gene together with MARPE1 and SERPINA3 genes significantly stratified overall survival for patients with triple-negative breast cancer (5). There is no evidence about CLIC1 expression in other molecular subtypes of breast cancer, except for triple-negative breast cancer, as we previously mentioned.
CLIC1 is a soluble protein found in the nucleus and cytoplasm of normal cells. In pathological conditions (including malignancies) it has the ability to enter the cell membrane and form active channels (6). CLIC1 membrane insertion induces tumor proliferation and metastasis, but the mechanism of this process is not yet completely understood. Ecad is responsible for similar actions during tumorigenesis, its membrane persistence being responsible of survival of tumor cells with the highest metastatic potential (7). Pcad is accepted as one of breast cancer stem cells (CSC) marker (8) as CLIC1 is related with CSC (especially metastatic CSC) from other cancer types (9, 10).
Currently, there are no data regarding the differential expression of CLIC1 and no correlation with other markers of tumor progression, metastasis and stem like potential in molecular subtypes of breast cancer. Based on previously described evidence, we aimed to study the influence of CLIC1, Ecad and Pcad association on molecular subtypes of breast cancer and to find if this association define new risk subgroups inside each molecular type. The identification of new markers is urgently needed in breast cancer because of a high heterogeneity related to tumor progression, response to therapy and relapse.
Materials and Methods
Breast cancer tissue specimen. A total number of 97 formalin-fixed paraffin embedded (FFPE) biopsies were retrospectively selected from the archive of the Department of Pathology from County Hospital Arad, Romania. The biopsies were harvested between 2011-2017 by open surgery from women with breast cancer and were included in the present study. Routine hematoxylin and eosin stain was performed to establish histopathological diagnosis and to select slides for immunohistochemistry, done on tissue microarray (TMA) FFPE blocks.
Tissue microarray technique. The technique was performed by using automated TMA Grand Master system (provided by 3DHistech, Budapest, Hungary) by selecting four areas, 2 milimeters in diameter each (2 from the middle and two from the periphery of the tumor) from each donor block and automatically transferred into paraffin recipient block. Each final recipient block included five different cases with four cores for each tumor case.
Immunohistochemistry. After the evaluation of the specimens stained with the haematoxylin and eosin and vimentin (clone V9), immunohistochemistry was perfomed on all TMA tumor blocks by using the following antibodies: CLIC 1, Ecad, Pcad, ER, PR, Ki67, Her2. The primary antibodies used were: mouse monoclonal anti human CLIC1 antibody (clone 356.1, dilution 1:2,000, Santa Cruz Biotechnology, Heidelberg, Germany), monoclonal mouse anti-human E-cadherin antibody (clone 36B5, ready to use, Leica Biosystems, Newcastle Upon Tyne, UK), monoclonal mouse anti-human P-cadherin antibody (clone 56C1, Labvision, Fremont, CA, USA), ER (clone 6F11, ready to use, Leica Biosystems, Newcastle Upon Tyne, UK), PR (clone 16, ready to use, Leica Biosystems, Newcastle Upon Tyne, UK), Ki 67 (clone MM1, ready to use, Leica Biosystems, Newcastle Upon Tyne, UK), HER2/neu (clone CB11, Novocastra, Newcastle Upon Tyne, UK). The incubation time was 30 min, at room temperature. The visualisation was performed with Bond Polymer Refine Detection System (Leica Biosystems, Newcastle Upon Tyne, UK). DAB was used as chromogen for 10 min and the counterstain with hematoxylin was the final step. All steps were performed in a fully automated controlled procedure with the BOND MAX autostainer (Leica Biosystems, Newcastle Upon Tyne, UK).
The first step in breast cancer cases stratification was based on immunohistochemical expression of ER, PR, Ki67 and HER2 to define molecular subtypes (Luminal A, Luminal B, Luminal B/HER2, TNBC). The second stratification was applied by the assessement of the relationship between CLIC1, Ecad and Pcad expression inside each molecular subtype in order to define high risk subgroups for each molecular subtype. CLIC1 expression was noted in both tumor and endothelial cells, while Ecad and Pcad expression was assessed for both membrane and cytoplasmic pattern.
Results
We evaluated CLIC1 expression in tumor cells, intratumor and peritumor blood vessels. About 84.31% out of total cases had CLIC1T+ pattern while less than 54% showed CLIC1V+ pattern.
CLIC1 had a cytoplasmic-dotted pattern distributed inside tumor cells and for some tumor cells we observed a membranar pattern related to the cytoplasmic one (Figure 1a). CLIC1 expresssion was heterogeneous (Figure 1b) or homogeneous (Figure 1c).
No nuclear expression has been found as in other malignanices. Pre-existing peritumor blood vessels were CLIC1– (CLIC1V–) while small-sized vessels, as well as those with a morphology suggesting intense angiogenic activity, were strongly CLIC1+ (CLIC1V+), with cytoplasmic distribution, in more than 50% of breast cancer cases. Subsequently, we wanted to see if there are significant changes between CLIC1T and CLIC1V expression depending on the molecular types of breast cancer. These data were also correlated with Ecad and Pcad expression for identifing different subclasses of breast tumors with a higher agressiveness.
CLIC1 was expressed not only in tumor cells but also in the tumor blood vessels within the tumor areas and also in those with peritumoral distribution (Figure 2). Stromal fibroblasts were positive for CLIC1. In a limited number of cases CLIC1+ was observed at the nuclear level in endothelial cells. CLIC1V positivity was much better visible for CLIC1T negative cases (Figure 3).
We considered the study of cadherins and CLIC1 expression by a separate evaluation of the Ecad and Pcad co-expression with CLIC1T, as well as with CLIC1V and then we found useful the combination of each cadherin and the pair CLIC1T/CLIC1V.
We defined 4 different immunophenotypes related with Ecad and CLIC1T association. Ecad+/CLIC1T+ phenotype was present in all molecular subtypes of breast cancer in various proportions, but for luminal B and TNBC this phenotype was found in all cases (100%). The Ecad–/CLIC1T+, as well as the Ecad–/CLIC1T– phenotypes were characteristic for Luminal A and Luminal B/HER2 mixed types respectively. In contrast, the Ecad–/CLIC1T+ phenotype was present in Luminal A, HER2 and Luminal B/HER2 subtypes.
Ecad/CLIC1V phenotype stratified TNBC cases in two subgroups, 40% of them not having CLIC1-positive blood vessels. The triple association between Ecad, CLIC1T and CLIC1V defined 8 phenotypes (Figure 4).
By applying the panel enlarged by three markers, two phenotypes specific to certain molecular classes were identified. Thus, the phenotype in which all three markers were negative was specific for Luminal A subtype whereas the phenotype in which all three markers were positive characterized Luminal B subtype. It should be noted that for the molecular classes known as having an aggressive and unpredictable evolution malignancy phenotypes based on Ecad expression were grouped in the Ecad + area, subgroup differentiation being achieved by CLIC1 expression heterogeneity in tumor cells and/or blood vessels.
P cadh/CLIC1 was evaluated in the same manner as that applied to E cadh/CLIC1. Initially, it was started with the double association between Pcad and CLIC1T, followed by the association with CLIC1V and subsequently by combining them in a triple association. This time the major interest was focused on the association of Pcad with CLIC1T, given that both markers are incriminated on the one hand as factors favoring rapid tumor progression and metastasis and, on the other hand, as factors characterizing cancer stem cells with high potential for local, regional and distant dissemination or for developing resistance to therapy, even if this therapy is a personalized one. Pcad+/CLIC1T+ phenotype characterized all triple-negative cases. Also, this association allowed the stratification of the HER2+ subgroup into two subtypes depending on the presence or absence of CLIC1 expression in tumor cells. Three-quarters of the cases were found to be positive for both markers in the tumor cells. At the same time, in case of Luminal B subtype, two subgroups were defined, more than half being in the category in which the two markers are positive. Pcad/CLIC1V association maintained the stratification of Luminal B cases and divided TNBC subtype in Pcad+/CLIC1V+ and Pcad+/CLIC1V– phenotypes (Figure 5a and b).
Triple association of Ecad/Pcad/CLIC1T impact on molecular classification of breast cancer had also been studied. Ecad+/Pcad–/CLIC1T– was specific for Luminal B/HER2 subtype whereas the type Ecad–/Pcad+/CLIC1+ was identified only for Luminal A cases, an aspect with possible involvement in tumor progression and metastasis (Figure 6a and b). The highest heterogeneity of this triple association was observed for Luminal A subtype.
The association between Ecad, Pcad CLIC1T and CLIC1V defined by this quadruple association, 16 phenotypes. The overall evaluation of the 16 phenotypes is shown in Figure 7a and b.
Luminal A and Luminal B/HER2 showed the highest heterogeneity of case distribution amongst these 16 phenotypes. Despite this heterogeneity for Luminal A cases, we identified two specific subtypes, which were not present in the other molecular forms of breast cancer. Thus, in about 23% of Luminal A cases, the phenotype with both cadherins positive was specific and CLIC1V is positive, CLIC1T being negative. In a significantly reduced percentage, we identified a subgroup with Pcad+ and CLIC1T+, the other factors being negative. The third particular phenotype specific for Luminal A subtype was one in which all four factors were negative.
For Luminal B/HER2 subtype, there were also two specific phenotypes: Ecad–/Pcad–/CLIC1T–/CLIC1V+ and Ecad+/Pcad–/CLIC1T–/CLIC1V+. Another important peculiarity observed when applying such a stratification was that, for the known molecular classes with an increased risk of metastasis as well as with an early development of resistance to therapy, most cases were located in the Ecad+/Pcad+ subgroup, CLIC1T and CLIC1V contributed to the differentiation of subgroups within each molecular phenotype.
HER2, TNBC and to a lesser extent Luminal B/HER2 presented this particularity. All TNBC cases were clustered into two subgroups, respectively those belonging to the group with both cadherins positive, 60% having both CLIC1T+ and CLIC1V+, also. (Ecad+/Pcad+/CLIC1T+/CLIC1V+) while 40% did not show CLIC1V+ (Ecad+/Pcad+/CLIC1T+/CLIC1V–).
For HER2 subtype the distribution was similar but, compared to the TNBC subtype, another subtype is added where nor CLICT, neither CLIC1V were positive. This phenotype was not found in other molecular subtypes and thus we considered as being specific for HER2 subtype.
Based on the expression heterogeneity derived by analyzing the phenotypes described above, we found useful to study the cytoplasmic expression of Ecad with CLIC1T and CLIC1V as it is well known the fact that the cytoplasmic expression of E cadherin correlated with loss of membrane expression are factors that favor mesenchymal epithelial transition correlated with increased tumor invasion and metastasis.
Inside Ecad+/CLIC1T+ group, 23.07% out of cases co-expressed cytoplasmic Ecad (EcadC) and CLIC1T. At this time, this subgroup can be considered as a risk subgroup for Luminal A subtype, characterized by the presence of tumor cells with an increased risk of mesenchymal epithelial transition, as well as local and distant metastasis. The percentage of EcadC+/CLIC1T+ cases was significantly increased for HER2 + subtype, where more than half of the cases showed such phenotype. The highest percentage of EcadC+/CLIC1T+ was observed in TNBC cases (60%).
Analysis of CLIC1V and EcadC expression revealed an extremely heterogeneous distribution between molecular forms of breast cancer.
Interpretation of triple association in EcadC+ tumors was compared with the overall expression of Ecad in triple association with CLIC1T and CLIC1V. Thus, for Luminal A, 50% of cases from Ecad+/CLIC1T+/CLIC1V+ group, were EcadC+/CLIC1T+/CLIC1V+ and 66.6% out of Ecad+/CLIC1T+/CLIC1V– cases had EcadC+/CLIC1T+/CLIC1V– pheotype.
HER2+ cases showed a more sensitive stratification by quantifying EcadC compared to its global expression. Approximately 75% of HER2 + cases with global expression of Ecad had EcadC+/CLIC1T+/CLIC1V+ phenotype, and 80% of those with Ecad+/CLIC1T+/CLIC1V– phenotype were EcadC+/CLIC1T+/CLIC1V–.
Discussion
Prognostic markers useful to identify cases with increased potential for progression and metastasis are extremely limited and nonspecific in malignant tumors, including breast cancer. There is currently no specific marker for tumor progression and metastasis, specifically for tumors of different origins and, for this reason, stratification of mammary malignancies within the same molecular class becomes mandatory.
Current studies describe a series of versatile markers that loose or modify their expression depending on tumor status, but their role in tumor progression and metastasis is not yet elucidated. From these, the markers we chose for our study are extremely controversial. Furthermore, the combination of Ecad, Pcad and CLIC1 for stratification of malignant mammary tumors within the same molecular class is not known.
It is widely accepted, at this time, that the loss of membranous expression of Ecad is followed by increased invasion and metastasis of tumor cells. In a recent publication it has been shown that, in breast cancer, this hypothesis is fragile because, in many cases of metastatic breast cancer, tumor cells strongly express cytoplasmic and/or membranous Ecad (11). The soluble form of Ecad observed in high amounts in ascites from ovarian cancer has been shown to be intensely angiogenic by releasing it as exosomes (11). This indicates that the loss of Ecad intercellular junctions and its release into the tumor microenvironment support tumor progression and metastasis not only through intercellular disruption, but also by stimulating the tumor microenvironment that supports these processes such as the cooption of new blood vessels in the tumor environment. Activation of endothelial cells by Ecad is accomplished by the interaction between V-cadherin and exosomes of Ecad, a process that induces a number of complex intracellular mechanisms involved in initiating the stages of tumor angiogenesis. In our study, CLIC1 was expressed in both tumor cells and endothelial cells and, in a limited number of cases in tumor-associated fibroblasts. All these aspects suggest the release of CLIC1 from tumor cells in the most likely form of exosomes that interact with endothelial cells from peritumoral and/or intratumoral vessels.
If the ligand on the endothelial cell is known for soluble form of Ecad, for soluble CLIC1 it is less studied. CLIC1 is incriminated in increased invasion and metastasis in gastric tumors or in glioblastomas (12, 13) and moreover, CLIC1 appears to be overexpressed in cancer stem cells (13). Djuric et al. demonstrated that stratification of glioblastomas based on CLIC1 and PLOD3 overexpression (marker incriminated in EMT) defined two groups of patients with different survival: those in whom CLIC1 was poorly expressed at 3-year survival rate was 10-19% compared to those who had an increased expression of CLIC1, in which survival over the same time frame was 0-6% (13).
Also, in CSCs, overexpression of Pcad has also been shown to be involved in tumor metastasis. We found useful to associate Pcad expression with CLIC1 to define a class of tumors with a high content of stem cells amongst breast cancer molecular subtypes. Thus, within each molecular group, we identified a subgroup co-expressing Pcad and CLIC1. Inside each breast cancer molecular subtype we identified Pcad/CLIC1+ cases, but the percentage of such cases was dominant for HER2 and TNBC types, in which this phenotype was significantly increased (75% and 100% respectively).
Our results support the use of Pcad/CLIC1 association in stratifying the risk of metastasis in HER2 and TNBC breast cancer subtypes and moreover, this combination can be used with a predictive role for the further development of the resistance to therapy given that both markers are incriminated in achieving of this process.
CLIC1 is responsible for developing resistance to therapy in gastric cancers and choriocarcinoma (12, 14, 15). It is well known that Pcad has a suppressive role for invasion and metastasis from malignant melanoma but, in breast cancer, it is a promoter of the two processes mentioned above (16). The versatility of this mechanism of action may be influenced by the coexistence of other factors with stimulatory role, of which CLIC1 is a potential candidate.
Recently, Xi et al. published a portfolio of Pcad in breast cancer (16). Ribeiro and his team pointed to SRC factor-dependent Pcad activation in basal-like breast cancers, SRC being a well-known factor involved in tumor angiogenesis and tumor cell proliferation (17). The same team studied the effects of Pcad on mesenchymal epithelial transition as a factor favouring the intermediate state of metastasis in breast cancer (17). Based on the preliminary observations made by the Ribeiro’s team, we found it useful to associate Ecad and Pcad with CLIC1 in vessels and tumor cells. On the other hand, CLIC1 is also involved in the development of resistance to therapy in ovarian cancer, glioblastoma and breast cancer (15, 18-21).
Conclusion
Based on our findings we may conclude that Ecad, Pcad and CLIC1 association may stratify cases within the molecular subtypes as distinct subclasses with a different behaviour in tumor progression, metastasis and early development of resistance to therapy.
Ecad, Pcad, CLIC1T and CLIC1V association defined several subgroups inside Luminal B, HER2+, and TNBC breast cancer types. The global evaluation of Ecad/CLIC1T coexpression showed that in TNBC we had 100% Ecad+/CLIC1T+ cases. Differential quantification of cytoplasmic Ecad coexpression with CLIC1T stratified TNBC cases into two distinct subgroups Of the two groups, we consider that the subgroup with cytoplasmic expression of Ecad and CLIC1T+ (EcadC+/CLIC1T+) represents a subgroup of risk for TNBC with a high metastasis potential and possibly with increased resistance to therapy.
The association of CLIC1V expression with the Ecad+/CLIC1T+ phenotype defined the Ecad+/CLIC1T+/CLIC1V+ phenotype. In this subtype 89.90% of the cases had cytoplasmic expression of Ecad, so that EcadC+/CLIC1T+/CLIC1V+ can be considered a risk subgroup.
The evaluation of Pcad/CLIC1T demonstrated a similar percentage to that of the E cad/CLIC1T association for the phenotype in which both markers are positive in the HER2 + and TNBC molecular subtypes. In case of Pcad, the subgroup Pcad+/CLIC1T+ was of interest. The lowest percentage of this phenotype was recorded in Luminal A, and the highest was observed for TNBC subtype. In contrast, Pcad–/CLIC1T+ phenotype was present in more than 55% of the cases of Luminal A whereas in TNBC and HER2 subtypes this phenotype was absent. Ecad/Pcad/CLIC1T association defined a phenotype specific only for Luminal A, Ecad–/Pcad+/CLIC1T+ observed in a small percentage of cases (7.69%). This group may be considered a high-risk group for breast cancer Luminal A subtype.
Ecad/Pcad/CLIC1T/CLIC1V phenotype certified the usefulness of this panel in stratifying breast tumors within the same molecular subtype.
Acknowledgements
The Authors are grateful to Victor Babes University of Medicine and Pharmacy Timisoara Romania for logistic and financial support of this paper. Also, many thanks to Adriana Meche for providing us tumor specimens and also to Emanuel Ciprian Onica and Patricia Berzava for technical assistance.
Footnotes
Authors’ Contributions
MR proposed the study highlighting that CLIC1 expression was not study before in molecular types of breast cancer. ARC performed immunohistochemistry. MR and AMC evaluated immunohistochemical expression of CLIC1, Ecad and Pcad, defined molecular subclasses and revised the final version of the manuscript. SC and SS performed statistics and data analysis.
Conflicts of Interest
The Authors have no conflicts of interest to declare regarding the present study.
- Received October 31, 2020.
- Revision received January 3, 2021.
- Accepted January 11, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.