Primary carcinomas are considered to be unorganized organs that are composed of various cell types, including cancer cells, fibroblasts and other mesenchymal cells, and cells related to immunity and vasculature. The tumor-stroma microenvironment leads to fibroblast activation and paracrine signaling between fibroblasts and cancer cells 
. In nemosis, activated fibroblasts start to produce proteins involved in inflammation, proteolysis and cancer progression and at the same time downregulate the expression of cytoskeletal proteins 
The objective of this work was to investigate the nemosis response of patient-matched normal and cancer-associated fibroblasts, and to study the expression pattern of CAF markers and their behaviour in nemosis. Only one of the normal fibroblast strains (FB-43) and one of the CAFs (FB-74) induced COX-2 in nemosis. This is in contrast with previously published results, where COX-2 induction has been considered a hallmark feature of nemosis. However, in those studies fibroblasts have been from neonatal origin and here we have used fibroblasts obtained from adults. This conflicting result therefore indicates that COX-2 should not be solely used to measure nemosis response, but other markers, such as the profile of secreted proteins, should be investigated as well.
The other key feature of nemosis is the time-dependent degradation of cytoskeleton. On protein level three of the fibroblast strains expressed α-SMA, surprisingly also the normal skin fibroblasts FB-74. The other normal fibroblast strain FB-43 did not express α-SMA at the protein level. However, when measuring the mRNA levels with the more sensitive Q-PCR method, all four fibroblast populations showed α-SMA expression and this was downregulated in nemosis. The time-dependent downregulation on both protein and mRNA level is in line with previous results on nemosis, indicating the decomposition of the cytoskeleton. Interestingly the CAFs started to regain the α-SMA mRNA expression at 72 h, the difference was significant when compared to their normal counterparts. In contrast to our results, a study by Shannon et al. 
showed that normal skin fibroblasts, but not oral fibroblasts expressed α-SMA. However, more recent results showed, in line with our results, that normal oral fibroblasts express α-SMA and this expression increased when these cells were cultured in conditioned medium obtained from OSCC cells 
. These contradicting results might come partly from the method that was used to measure the α-SMA expression; in the first one immunoblotting was used, in the second the method was slightly more sensitive immunohistochemistry.
We investigated also the mRNA levels of two other CAF markers, FSP1 and FAP. In nemosis FSP1 levels decreased in FB-43, FB-74 and CAF-74 spheroids, but increased in CAF-43 cells. The third investigated CAF marker FAP was upregulated in nemosis, more in CAFs than in normal fibroblasts, the difference was significant with the 43 fibroblast strains. With all three CAF markers the nemosis response followed the pattern of decreased expression of cytoskeletal genes (α-SMA and FSP1) and increase in proteolytic gene expression (FAP). Clearly different response was seen with CAF-43 cells where, instead of downregulation of FSP1, the levels increased in nemosis.The heterogeneity of fibroblasts becomes evident when looking at the basal levels of the CAF marker expression; CAF-43 cells had lower levels of all three markers, CAF-74 had less α-SMA, slightly more FSP1 and over 10-fold more FAP. These results also emphasize that α-SMA, the most commonly used CAF / myofibroblast marker, should not be used solely to define activated fibroblasts.
Another hallmark of nemosis is the induction of growth factors. It has been shown that oral fibroblasts produce significantly more FGF7 and HGF/SF when compared to skin fibroblasts 
. These two growth factors, together with VEGF, are known to be important in wound repair and cancer progression 
. The basal expression of VEGF, HGF/SF and FGF7 mRNA was lower in CAFs than in normal fibroblasts, and this is in contrast to previous results 
. However, the growth rate of these cells was slower than that of their normal counterparts, which might reflect their decreased production of growth factors. The SA-β-gal activity of the CAFs supports this theory, indicating that these cells are senescent. The need for these growth factors to be secreted by fibroblasts could be reduced in the CAFs since the tumor cells themselves, along with infiltrated macrophages and endothelial cells, are capable to produce these factors. As expected, VEGF, HGF/SF and FGF7 mRNAs were upregulated in fibroblast nemosis, and the level of induction varied between fibroblast populations. VEGF induction was highest in CAF-74 spheroids, HGF/SF in CAF-43 spheroids and FGF7 in FB-43 spheroids.
Based on these results it seems that the capability of normal and cancer-associated fibroblasts to produce these growth factors in nemosis is somewhat related to the extent they are needed in cancer progression. The dependence of tumors on stromal fibroblasts, and particularly on the growth factors they produce, decreases in the course of tumor progression. Epithelial cells require FGF7 to break the epithelial polarization. FGF7 is only expressed by stromal cells and its receptor FGFR2IIb only by epithelial cells, indicating the role FGF7 in the beginning of tumor progression 
. Of the studied fibroblast populations the FB-43 cells, which appear to be most normal of the studied strains (based of induction of COX-2 and lack of α-SMA), had the highest FGF7 induction in nemosis. HGF/SF is required for the migration / scattering of the epithelial cells from the initial break point. Nemotic CAF-43 cells produced more HGF/SF than the other three cell strains. Supporting this Kankuri et al. 
have shown that HGF/SF produced by fibroblast spheroids directly promotes cancer cell invasion. Also another study has shown that oral fibroblasts drive invasion of OSCC cells by increasing secretion of HGF/SF 
. VEGF is required later in the tumor progression when the cancer cell mass extends the point where it can no longer grow without oxygen supply. VEGF, secreted by fibroblasts, induces angiogenesis by recruiting endothelial cells to form new blood vessels 
. CAF-74 cells, which are senescent, have by far the highest level of VEGF in nemosis.
It has been well established that CAFs, but not normal fibroblasts, are capable to promote tumor progression 
. More recent results have shown that initially the normal fibroblasts inhibit the growth of cancer cells 
, and our present results concur with that notion. We show here that normal fibroblasts indeed inhibit the colony formation of recurrent SCC cells, but curiously this was not seen with primary tumor cells. The CAFs seem to be able to influence only the primary SCC cells and not the recurrent cells. The CAFs produced lower levels of growth factors, and it could be that for this reason they are capable to influence the more responsive primary SCCs, but the less sensitive recurrent cells do not respond to this lower amount of secreted growth factors.
The observed spontaneous spheroid formation of FB-43 and CAF-43 in monolayer cultures is in line with the results from Kankuri et al. 
, where spontaneous clustering of fibroblasts, i.e. nemosis, could be achieved by adding tumor cell-derived conditioned medium to a fibroblast monolayer. However, we did not find this with the fibroblast strains from the other SCC patient. This might be partly due to the induction of tumor suppressor p53 in the 74A and 74B SCC cells. Nonetheless, the fibroblasts did grow faster under the influence of 74B SCC cells; this was also true with the CAF-74 cells that seem to be in a state of stress-induced senescence. Further more, we did not see anchorage-independent growth of the fibroblasts, conflicting with the results obtained with prostate- and prostate carcinoma associated fibroblasts 
. Possible explanations for this are individual variations and the origin of the fibroblasts. It is worth noting that in the soft-agarose experiments the SCC cells and fibroblasts were not in direct contact but separated by a solid layer of agarose, and the cultures were not replenished by fresh medium. Therefore the paracrine signaling between these two cell types must be mediated by soluble factors.
In conclusion, this study clearly demonstrates that fibroblasts obtained from different individuals vary in gene expression and behavior and that the expression of CAF markers differs between normal fibroblasts and CAFs in nemosis. Both normal and cancer-associated fibroblasts modulate tumor cells, normal fibroblasts by inhibiting the growth of invasive SCC cells and CAFs by further enhancing the growth of primary SCC cells. Nemosis, an in vitro model of fibroblast activation, may have its in vivo counterpart in cancer-associated fibroblasts and is a valuable tool in studying the variations between fibroblasts obtained from different individuals. Nemosis response, particularly of the CAF markers α-SMA and FAP, could therefore be used as a prognostic marker to predict the stromal reaction of tumors.