The expression of different formin proteins in human tissues has not been systematically investigated. As the protein family has 15 members, the proteins are expected to possess individual and different functions and expression patterns, reflecting the demands for cytoskeletal regulation in various cell types. Here, we set out to investigate how widely FMNL2 is expressed by using immunohistochemistry. An antibody was raised for this purpose. After validation by Western and Northern blotting, as well as a peptide competition and siRNA experiments, we conclude that this antibody is specific and suitable for immunohistochemistry.
FMNL2 mRNA has been reported to be expressed in many normal tissues [6
]. Using a large set of information on FMNL2 transcripts, we confirmed that FMNL2 mRNA indeed can be detected in virtually all tissues [7
]. Highest expression was seen in the central nervous system, followed by the peripheral nervous system and several epithelial tissues. A consistent result was achieved by immunohistochemical analysis of 26 tissues. In the central nervous system, expression was wide, as strong expression was seen both in neurons and supporting glial cells. Strong expression was, however, not exclusive for nervous tissues. Moderate or strong expression was seen in most epithelia and lymphatic tissues. Especially gastric, intestinal, mammary and endocrine epithelia expressed FMNL2 strongly. Connective tissue components including fibroblasts, endothelium and adipose cells expressed little or no FMNL2. It will be interesting to study whether FMNL2 expression changes in invasive cancer, when epithelial cells lose their polarity and cell-to-cell contacts.
In the hematopoietic system, lymphocytes, myeloid precursors and macrophages displayed moderate to strong FMNL2 expression. Erythroid precursors, as well as mature red blood cells were negative. Consistent with this, no FMNL2 was detected by Western blotting in the K562 erythroleukemia cells. Such cell-type specific expression in cells differentiating from a common precursor may have clinical implications. Another DRF, Diaphanous 1 (mDia1), has been linked to bone marrow disease. mDia1 knockout mice develop age-dependent myeloproliferative defects mimicking human myeloproliferative and myelodysplastic syndromes [10
]. FMNL2 could also be involved in diseases with defects in myelopoiesis. However, the phenotype of an FMNL2 knockout mouse has not been described.
One of the epithelia expressing FMNL2 was colon. The expression was relatively high both as analyzed by mRNA signal and immunohistochemistry. Our findings are distinct from the results reported in a study linking increased FMNL2 expression to metastatic activity in colorectal cancer. In that study, normal colorectal epithelium was reported to be devoid of FMNL2 or express it only weakly [11
]. The difference may be due to several causes. The previously reported antibody may have a lower sensitivity than ours, resulting in positive immunoreactivity only in tissues with exceptionally high FMNL2 expression. Also, the antibodies may have different isoform-specifity. The antibody used for the colorectal cancer metastasis study was directed toward an antigen in FMNL2 isoform 1 that shares 91% identity with FMNL2 isoform 2. It is possible that the antibody has weaker or no affinity to FMNL2 isoform 2. In contrast, the antibody used in our study was raised towards a sequence common to both isoforms. The colorectal metastasis study illustrated the relative change of FMNL2 expression in normal colon and colorectal adenocarcinoma, especially metastatic disease. At this point, we conclude that normal colorectal mucosa does express FMNL2, but do not compare the level of expression to cancer. Thus, our results do not necessarily contradict the conclusions made. Our preliminary transcription analysis indicates that FMNL2 expression is up-regulated in several forms of cancer, and future studies will be directed to find the ones where this difference has biological and medical relevance.
Uncontrolled proliferation, combined with invasive and metastatic potential, is a hallmark of malignancy. Rho GTPases have been shown to mediate cytoskeletal changes essential for malignant cell behavior in vitro, and evidence of in vivo relevance is mounting. Rho GTPases activate effector molecules in different cellular locations. Effectors, in turn, mediate cytoskeletal changes [12
]. Identifying Rho GTPase effectors is crucial for understanding actin remodeling in normal and malignant cells. FMNL2 contains a GTPase binding domain. Recently, FMNL2 was identified as a specific RhoC effector [13
]. In the same study, RhoC and FMNL2 were, by using RNA interference, found to be essential for the invasive motility of rounded human melanoma cells. FMNL2 silencing did not affect the invasive capacity of breast carcinoma cells or fibrosarcoma cells, which display a more mesenchymal elongated morphology. In our study, confocal microscopy of WM164 melanoma cells revealed a co-localization of F-actin and FMNL2 at the tips of cell protrusions. A431 squamous cell carcinoma cells and MKN-1 gastric carcinoma cells had a different, unspiked morphology and in these cells FMNL2 and F-actin were not co-localized. These results are in agreement with the finding that FMNL2 has cell-specific functions and that it has a special role in protrusions of invading melanoma cells.
As Rho GTPase expression levels change in malignancies, changes in effectors are to be expected. Without doubt, several other formins will in the near future be identified as effectors specific Rho GTPases. Understanding the cascades behind malignant cell behavior in different malignancies can ultimately offer targets for novel therapies.