In the present paper we first demonstrated that human breast proliferative disorders such as FH, FIB and DCIS, express FST and FLRG and confirmed previous findings describing FST and FLRG expression in NB and IDC . In the NB and in all the breast proliferative disorders analyzed, FST and FLRG were localized in the epithelial and in the stromal cells at mRNA and protein levels. These expressional patterns are similar to the one we [2, 28] and others  previously described for activin in human normal and pathological breast tissues. Here we also found an increased expression of FST in stromal cells of FIB compared to NB and FLRG up regulation in IDC.
FIB are benign tumors of the breast typically composed of stromal and epithelial cells  and we previously reported that activin βA was down regulated in this breast disease . Taken together, the findings that demonstrate down regulation of activin βA and up regulation of FST in FIB suggest that activin anti-proliferative effects may be weakened, thus favoring the cellular events that lead to the establishment of the FIB lesion. Additionally, in a mouse model, Krneta and co-workers (2006)  showed that although FST-expressing R30C tumors displayed increased angiogenesis, they were highly susceptible to undergo serum starvation-induced apoptosis, suggesting a role for FST in limiting tumor progression. Moreover, despite some controversy about the importance of hormonal effects in the development of FIB, some reports documented increased estradiol levels in the serum of women carrying FIB lesions . In this connection, FST mRNA transcripts were decreased in the mammary gland of rats after ovariectomy, showing that FST expression is regulated by estrogen in the mammary gland .
Regarding FLRG, in the present study, we found that its expression was increased in IDC at protein level as previously reported  and that such increased expression was also present at mRNA level, implying that the tumor itself is the source of increased FLRG protein expression. Concerning activin's role in breast cancer, previously we reported an increased transcript profile of activin βA subunit and an augmented activin A concentration in homogenates of breast carcinoma . On the other hand, in high grade breast cancer, an impairment of the activin signal transduction system has been linked to oncogenic progression, since a reduced expression of activin βB and its receptors, as much as alterations of smad signaling have been characterized . All in all, together with the findings that FLRG contributes to tumor cell proliferation through antagonizing activin effects , we assume that the increased activin A expression reported in breast carcinoma is counteracted by an increased FLRG expression that prevents activin from binding its receptors and thus reducing smad signaling and activin anti-proliferative effects. Additionally, we also analyzed FLRG pattern of expression in pre-neoplastic diseases and in DCIS, a lesion where cancer cells do not infiltrate the adjacent stromal tissue and which has been pointed to be a precursor of IDC . FLRG was only over-expressed in the IDC, suggesting that FLRG is strongly correlated with tumor progression and breast malignancy.
As for the cellular localization, the typical nuclear staining of FLRG already described in other cell types  was present in the normal and pathologic mammary glands evaluated here. We also observed FLRG staining in the cytoplasm, which recapitulates in vivo observations in human endometrium [21, 25]. In the current study, FST staining was stronger in the cytoplasm, suggesting that the localization pattern of this protein in mammary gland epithelial cells is the same already described for other tissues [22, 33]. Yet, we also noted nuclear staining of FST, which is rather atypical but has been described before in spermatogenic cells .
It is interesting to note that FST expression was increased in the stromal cells of FIB, while FLRG was up regulated in IDC, indicating that the two activin binding proteins may play diverse roles in tumor progression. Although their overall gene structure is quite similar, FST and FLRG seem to regulate activin pathway in different ways . Indeed, in human liver tumor specimens, FST expression was increased in about 60% while FLRG transcript profile remained unchanged . Conversely, in endometrial carcinoma, FST expression is unchanged while FLRG expression is down-regulated . FLRG was also found to be down-regulated in ovarian endometriosis, while FST expression was found to be up-regulated . Collectively, these finds demonstrate a number of differences between FST and FLRG, further suggesting that they may not be complete functional homologues.
Finally, a further mechanism to be hypothesized is that activin may have, similarly to TGFβ, a dual role on cancer progression. In early phase, TGF-β inhibits growth of cancer cells by cytostasis, differentiation and apoptosis. In later phase, instead, it works to promote cancer progression and metastasis by several mechanisms including evasion from immunological recognition, production of growth factors, differentiation into an invasive phenotype, metastatic dissemination as much as the establishment and dispersion of metastatic colonies [36, 37]. Although whether activins work similar to TGF-β is not currently known, despite the anti-proliferative actions of activin, follistatin seems to act as an inhibitor of cancer metastasis .