Immunohistochemistry was applied to examine the level of BRMS1 protein in a panel of benign nevi and primary and metastatic melanomas in order to evaluate the impact of altered expression on clinical outcome. In accordance with previous studies demonstrating higher mRNA and/or nuclear BRMS1 protein expression in normal or benign tissues as compared to malignant tumors [4, 12, 29], BRMS1 was highly expressed in the nucleus of benign nevi whereas only a minor fraction of primary melanomas or metastases showed a similar expression profile. Our findings also partially agree with the study by Li et al.  who observed a decline in nuclear BRMS1 expression from dysplastic nevi to primary melanomas. A possible explanation for the more dramatic decline in our study may be that, in the previous study, malignant lesions were compared to dysplastic nevi while benign nevi of compound and intradermal subtypes where utilized in our investigations. Thus, it may be speculated that loss of nuclear BRMS1 expression is an early event, distinguishing common and dysplastic nevi.
Although BRMS1 has been recognized as a mainly nuclear protein, Rivera et al.  recently showed that BRMS1 contains both nuclear import and export signals, and implied that nuclear-cytoplasmic shuttling may represent a novel mechanism for altering the activity or function of BRMS1. The data in this study and a recent immunocytochemistry report in breast cancer  are consistent with that hypothesis. During melanocyte transformation and progression, BRMS1 appears to relocalize from the nucleus to the cytoplasm. Additionally, cytoplasmic staining intensity appeared to have a more pronounced impact than number of immunoreactive cells, suggesting that the level of BRMS1 protein has to reach a threshold level to have an effect on cellular behavior. Although there was no obvious difference in the percentage of tumor cells expressing BRMS1 in the cytoplasm, there was a clear decline in staining intensity in metastases as compared to nevi and primary tumors. Our findings are, however, in contrast to the study by Li et al.,  who observed predominantly nuclear BRMS1 expression in benign and malignant melanocytic cells. One explanation for this discrepancy may be the use of different antigen retrieval and detection methods.
Frolova et al.  reported that cytoplasmic BRMS1 was associated with increased proliferation in estrogen receptor-negative breast cancers. In contrast, the melanoma samples used here show that cytoplasmic expression of BRMS1 is inversely associated with markers of proliferation, cyclin D3, cyclin A, Ki67, p21Waf1/Cip1. When coupled with our previously demonstrated associations between cell cycle regulators, numbers of mitosis and disease progression [21, 22, 24], the data presented further strengthen the hypothesis that cytoplasmic BRMS1 is associated with a less aggressive melanoma phenotype (i.e., thinner tumors, less ulcerated, longer survival).
ERK1/2, when sequestered in the cytoplasm, has been suggested to prevent transcription of pro-survival and proliferative proteins as well as enhance the activity of pro-apoptotic cytoplasmic proteins . Furthermore, whereas Gayer et al.  recently suggested that excluding ERK1/2 from the nucleus inhibits proliferation, Jovanovic et al. observed that the presence of activated ERK1/2 in the cytoplasm of melanomas was associated with better prognosis. In line with this, we demonstrated a strong positive association between BRMS1 and activated ERK1/2 when both were localized in the cytoplasm. Moreover, we also observed a strong inverse association between cytoplasmic BRMS1 expression and accumulation of activated p-Akt in the nucleus. Prior studies observed a negative association between BRMS1 and activation of PI3-kinase/Akt signaling [28, 33], which is consistent with the well documented evidence that PI3-kinase/Akt signaling affects numerous steps of the metastatic cascade, including proliferation, apoptosis, migration and invasion . In thyroid cancer, nuclear localization of activated Akt was associated with tumor invasion and metastasis . Together these results suggest that cytoplasmic BRMS1 may at least partly negatively regulate melanoma progression and metastasis through sequestering of activated ERK1/2 in the cytoplasm and by preventing accumulation of nuclear active Akt.
In contrast to the results obtained in another melanoma cohort , as well as in breast  and NSCL cancers  our study showed that high nuclear expression of BRMS1 was associated with more aggressive tumors and shorter disease-free survival. The higher nuclear expression of BRMS1 found in benign nevi than in primary melanomas or metastases seems contradictory to the association to prolonged disease-free survival for patients having low nuclear tumor expression of BRMS1. As yet, we are unable to explain this phenomenon, but benign nevi are commonly terminal lesions as opposed to melanomas. Thus, the molecular events regulating these processes might differ. In this regard it has been shown that mutated B-Raf plays a role in inducing senescence in melanocytes, whereas in melanomas it contributes to oncogenesis . Furthermore, in a recent study we showed that expression of fatty acid binding protein 7 (FABP7) is higher in benign nevi than in melanomas, still FABP7 was suggested to contribute to disease progression, most likely by increasing tumor cell invasion . Interestingly, we observed a strong positive association between nuclear BRMS1 and FABP7 expression, suggesting that nuclear BRMS1 may, in fact, increase the invasive potential. In support of this, we showed that down-regulation of BRMS1 in two metastatic melanoma cell lines, expressing predominantly nuclear BRMS1, reduced the invasive ability.
Although it has not yet been verified at the protein level, the mRNA for various splice variants that are differently expressed in metastatic and non-metastatic breast cancer cell lines have been identified [18, 37]. Thus, it may be speculated that benign nevi and primary and metastatic melanomas express different BRMS1 variants with different biological functions.