In this report, we found that SnoN is expressed in the normal human esophageal, ovarian, pancreatic and breast tissues predominantly in the cytoplasm of both the epithelium and stroma. The intracellular localization of SnoN did not alter significantly in tumor samples, but its expression is downregulated in all four types of tumor epithelia at lower malignant grades, but gradually recovered in higher grade tumors and markedly upregulated in the stroma of esophageal and ovarian cancer tissues. These results suggest that SnoN expression may vary at different stages of tumor progression and that SnoN may play different roles at different stages of malignancy. In particular, the initial downregulation of SnoN in the tumor epithelium is consistent with its anti-tumorigenic activity. The later upregulation of SnoN in the stroma microenvironment in esophageal and ovarian cancers suggests that it may play a role in promoting tumorigenesis in an indirect manner. Thus, the pro-oncogenic activity of SnoN in human cancer may be non-tumor cell autonomous and through affecting the tumor microenvironment.
Our results differ somewhat from those published previously, in particularly those that reported an increase in SnoN levels in cancer tissues 
. Although the reason for this difference is not known at this time, several possible causes could be speculated. First, the difference could potentially be due to the variations among different samples. Indeed we found that even normal breast tissue samples from different individuals showed a certain level of heterogeneity in term of SnoN levels, possibly due to the different conditions (tissue damage, stress, etc) or treatment procedures that the tissues have experienced. It is therefore quite possible that some cancer samples may also show variations in SnoN levels as a result of the diverse and complex tissue environment under which the tumors develop. However, the overall trend in all four cancer types with over hundreds of tumor samples in our hands is a consistent reduction of SnoN in the tumor epithelium. The level of SnoN may also change depending on the malignant stages of tumors and vary from tissue to tissue. While in esophageal and ovarian cancer, SnoN is downregulated in low grades adenocarcinoma but again elevated in grade III tumors, grade III breast cancer did not show any elevation even when compared to low grade tumors. Similarly, while high SnoN expression is found in stroma of high grade esophageal and ovarian cancer, pancreatic and breast cancer did not show a strong elevation in stroma SnoN expression. Thus, SnoN may play different roles in the progression of different cancer types and possibly through different mechanisms. Another reason may be related to the methods used to measure SnoN expression. Since SnoN expression is upreguated in the stroma while decreased in the tumor epithelium, it is possible that by measuring overall SnoN levels by western blotting, one might detect an increase in the overall expression of SnoN. Finally, the difference could be due to the different antibodies used in the staining. The commercial anti-SnoN antibodies commonly used before showed a high background and sometimes detected nonspecific signals in our hands (data not shown). The anti-SnoN C-terminal peptide antibody has been carefully characterized, and antigen competition experiments demonstrated the specificity for the SnoN signal () 
. Moreover, using this antibody, we have detected dynamic changes in SnoN expression during normal tissue morphogenesis 
and malignant progression, providing additional support that the signals we detect are specific.
Our study has revealed a previously unappreciated role of SnoN in the tumor microenvironment. Especially in esophageal and ovarian tissues, multiple stromal cell types showed high levels of SnoN expression, including smooth muscle cells, blood cells, fibroblasts and infiltrating lymphocytes. Interestingly, the cancers of esophagus and ovary tend to harbor 3q26 amplification and show SnoN copy number increases, and they also display a stronger increase in stroma SnoN levels. Thus, the high SnoN expression in the stroma may be related to gene amplification, and this in turn may enhance malignant progression in these tissues. More and more studies have indicated that tumor-associated stromal cells and infiltrating lymphocytes promote epithelial tumorigenesis, invasion and metastasis, and cancer inflammation 
. In addition, infiltrating B- and T-cell lymphocytes can regulate solid tumor development through their abilities to regulate myeloid cell function in solid tumors 
. Interestingly, TGF-β is a well-known regulator of the tumor microenvironment by modulating the proliferation of fibroblasts, secretion of chemokines and cyokines that control inflammation, recruitment of bone-marrow-derived myeloid cells and host immunosurveillance 
. As a negative regulator of TGF-β signaling, it is conceivable that SnoN may also modulate the tumor microenvironment and tumor-host interaction. Our results clearly indicate that this is an important area to focus on in future studies.
High levels of SnoN have been shown to directly stabilize and activate p53. Thus, tissues or cells that show elevated SnoN expression (either due to gene amplification or other means) may cease proliferation due to p53-mediated senescence or cell cycle arrest. Malignant cells have to overcome this barrier in order to maintain their survival and expansion and may do so by either inactivating/deleting SnoN itself or any downstream steps in the SnoN-p53 pathway. Indeed, our analysis of CLE cell lines has indicated a significant correlation between increased SnoN copy number and p53 inactivation, supporting the idea that the tumor suppressive SnoN-p53 pathway must be inactivated to allow malignant progression. A previous study of human breast cancer tissue samples also note a correlation between SnoN upregulation and elevated p53 expression, which is indicative of inativating mutation of p53 
. However, in human cancer tissue samples that we have tested, no correlation between SnoN expression and p53 status can be detected. This is not entirely surprising given that these tumor tissues have already downregulated SnoN expression and therefore had no pressing need for additionally inactivating p53.
In summary, we have found that SnoN is expressed in both epithelial and stromal cells in normal human esophageal, ovarian, pancreatic and breast tissues. Unlike what has been depicted in the current models, SnoN expression is downregulated at the protein level in the tumor epithelium in low grade tumors in all four tissues, consistent with its anti-tumorigenic activity. In contrast, SnoN expression is markedly upregulated in tumor stroma in some cancer types, especially at late stages of malignancy. Giving the importance of tumor stroma in influencing malignant progression, this result promopted us to speculate that SnoN may also regulate tumorigenesis through modulating tumor microenvironment.