Integrin expression
Previous studies have identified a linkage between the expression of β1 and αv integrins and breast cancer [4, 6]. In addition, cell agonists such as PMA that activate protein kinase C and induces phosphorylation of pERK, promote integrin-mediated cell adhesion, focal adhesion formation and cell signaling in many cell types including cancer cells [19, 20]. Therefore, we first identified an optimal concentration of PMA that induced pERK formation (Figure 1) and then assessed the relative levels of these integrins expressed by adhered breast cancer cells and Hek-293 cells using flow cytometry of untreated (Figure 2A) and PMA treated cells (Figure 2B). To determine the optimal concentration of PMA to use, MDA-MB-435 cells were stimulated with different concentrations of PMA (50 to 200 nM) and then the level of pERK was determined by western blot analysis (Figure 1). Results indicated that 150 nM PMA produced the highest levels of pERK, in agreement with our previous studies using similar concentrations of PMA as an activator of cell adhesion in other cell lines [18, 19]. Therefore, 150 nM PMA was used as the PMA stimulus in the remaining experiments.
To maintain the integrity of the surface expression of integrins on cell adhered to FN, all cells washes and incubations were performed at 4°C prior to their analysis by flow cytometry. We consistently found that the non-breast cancer cell line, Hek-293, generally expressed lower integrin levels as compared to the three breast cancer lines (Figure 2A). Hek-293 expressed very low levels of β3, β5, αvβ3, αvβ5 and αvβ6, but higher levels of β1 and αv. All three breast cancer cell lines expressed high levels of β1 and αv, and they also expressed higher levels of β5 and αvβ5 in comparison to Hek-293. MDA-MB-435 integrin expression distinguished this cell line from all others as they consistently expressed higher levels of integrins and they were the only cell line to express high levels of β3 and αvβ3.
Next, the effect of short-term PMA stimulation on integrin expression in the cancer and Hek-293 cells was evaluated (Figure 2B). The results obtained for PMA treated cells were nearly identical to those of mock DMSO treated cells and untreated cells (panel A). Integrin expression remained unchanged or was only slightly altered by PMA treatment. These results are consistent with previous findings that short-term PMA treatment does not enhance integrin expression [20], rather it activates integrins [18]. In addition, we determined that short-term suspension or adhesion of cells in the presence or absence of PMA did not affect integrin expression (experiments performed in duplicate or triplicate). For example, expression of αvβ3 in MDA-MB-231 suspension cells treated with DMSO or PMA was 9.7% and 9.9%, respectively, and expression of αvβ3 in two hour adhered MDA-MB-231 cells was 2.5% and 2.8%. Furthermore, the expression of αvβ3 in MDA-MB-435 suspension cells treated with DMSO or PMA was 99.1% and 98.2%, respectively, and expression of αvβ3 in two hour adhered MDA-MB-435 cells was 98.4% and 98.8%.
Adhesion of breast cancer cell lines
Cell adhesion plays a vital in the survivability and progression of a cancer as engagement of integrins with the ECM prevents some cancers from undergoing apoptosis while it induces cell proliferation in others. In metastatic cancers, cell adhesion undergoes rapid regulatory changes that allow the cancer cell to disengage from the ECM, migrate and then reengage with the ECM at its secondary metastatic site. In addition, short-term exposure of cells to cell agonists such as PMA, results in increased αv integrin-mediated cell adhesion and spreading onto ECM proteins [19, 20]. Therefore, we assessed the capacity of 150 nM PMA to influence the adherence of the breast cancer cells to ECM proteins (Figure 3). We used FN, Fg and VN as ligands with differing specificity for αv integrins and collagen as a non-αv integrin ligand. In general, the adhesion of unstimulated cells, cells incubated in media alone, was markedly greater than we previously reported for GM1500 or M21 cancer cells [18, 19], with 20 to 40% of the total cells adhering within one hour. The majority of cells that adhered within one hour were firmly attached and cell spreading (formation of lamellipodia and filopodia) was readily detected (data not shown). Unstimulated MDA-MB-435 (Figure 3A) and MDA-MB-231 (Figure 3B) cell adhered highest to FN, while MCF7 (Figure 3C) and Hek-293 cells (Figure 3D) had equal preference for FN, Fg and VN. MDA-MB-231 showed the lowest nonspecific binding to BSA, and MCF7 cells were the only cell line that adhered well to collagen. However, in contrast to our previous studies using αvβ3-expressing GM1500 cancer cells [18], PMA treatment did not upregulate cell adhesion. Increasing the PMA treatment and adhesion time to four hours also showed no PMA effect (data not shown). The adhesion of mock treated cells, incubated with the same concentration of DMSO as was present in the PMA samples, were also similar to that of unstimulated cells (data not shown). Therefore, we tested the hypothesis that the non-PMA treated cells were already near maximal levels of adhesion which negated any further increase with PMA treatment. Using GM1500 cells, we observed that less than 5% of the non-treated cells adhered to Fg, and the cell adhesion increased two to four-fold following PMA treatment (data not shown). These results led us to conclude that the breast cancer and Hek-293 cells expressed an integrin co-receptor or a non-integrin adhesion receptor that upregulated or directly facilitated cell adhesion. To determine to what extent the adhesion was mediated by integrins, the cells were allowed to adhere to FN for one and two hours in the absence and presence of αv and β1 functional-blocking antibodies. The adhesion of MDA-MB-435, MDA-MB-231, MCF7 and Hek-293 cell after one hour was inhibited 79.1% ± 8.8; 79.8% ± 8.4; 42.3% ± 24.5; 80.7% ± 8.7 (mean ± stdev), respectively by the addition of both antibodies (n = 5 in duplicate experiments). At two hours the adhesion was inhibited 82.5% ± 7.25; 75.4% ± 11.4; 64.5% ± 14.7; and, 90.2% ± 4.9, respectively. Thus, MDA-MB-435, MDA-MB-231 and Hek-293 cell adhesion was highly integrin-mediated, while only two-thirds of MCF7 adhesion was integrin-mediated. This led us to speculate that the increase in adhesive capacity of these cell lines was a result of increased integrin activation through the action of either a co-receptor or upregulated signaling through intracellular pathways.
Agonist-induced signaling
Cells continuously respond to their extracellular environment and cues provided by ECM proteins, growth factors, cytokines and other cell agonists can invoke varying responses within different cell types. Thus, some of the heterogeneity of breast cancer could be a result of varying responses by different breast cancer cells. Therefore, we determined if all the breast cancer cells responded in a similar manner to a cell agonist. Furthermore, as integrins are responsible for transmitting signals from the environment to the cell, we also determined if the high adhesion of unstimulated breast cancer cells resulted in upregulated intracellular signaling. We therefore allowed the cells to adhere overnight onto FN-coated plates and then measured the levels of integrin signaling molecules before and for various times after treatment with 150 nM PMA. MEK levels were unchanged by PMA treatment in MCF7 and Hek-293 cells, and only decreased in MDA-MB-435 and MDA-MB-231 cells after two hours of treatment (Figure 4A). However, marked changes occurred in the levels of activated pMEK (S217/S221). In MDA-MB-435 cells, pMEK levels in untreated and PMA treated cells remained high until 2 hours of PMA treatment and then decreased, while in MDA-MB-231 cells pMEK levels remained higher and unaltered by PMA treatment. The pattern of pMEK expression in MCF7 cells was markedly different from the metastatic cells. All non-PMA treated MCF7 cells (lanes 1-3) containing undetectable levels of pMEK, and only a weak transient signal was detected following PMA treatment. The pattern of pMEK expression in Hek-293 was similar to that of MCF7 cells. Furthermore, regardless of the differences in pMEK levels following PMA treatment, high pMEK levels in adhered MDA435 and MDA231 cells separated these metastatic cells from the non-metastatic MCF7 and Hek293 cells.
PMA treatment had no effect on the high levels of ERK present in each cell line (Figure 4B). In contrast, the levels of activated pERK were very low in most of the non-treated cells and PMA treatment resulted in differential upregulation of pERK. The levels of pERK in MDA-MB-435 cells transiently increased in a biphasic response to PMA, reaching maxima at 30 min and two hours. In MDA-MB-231 cells, pERK levels never reached a maximum, while pERK levels in MCF7 cells increased between 30 min and two hours. There was high and sustained induction of activated pERK in Hek-293 cells following PMA treatment (Figure 4B). Thus, there was heterogeneity in MAPK pathway signaling by adhered breast cancer cells in the absence and presence of PMA.
The Src pathway was investigated in the cells by evaluating their levels of c-Src, activated Src [pSrc(Y416)] and deactivated Src [pSrc(Y527)]. The levels of c-Src remained unchanged in MCF7 and Hek-293 cells, while they decreased after two hours of PMA treatment in the metastatic MDA-MB-435 and MDA-MB-231 cells (Figure 4C). PMA induced activation of Src in MDA-MB-435 cells, with pSrc(Y416) levels reaching at maxima at two hours. There was minimal induction of pSrc(Y416) in MDA-MB-231, MCF7 and Hek-293 cells. In addition, all cells grown in media containing 10% fetal calf serum that supports cell proliferation (lane one) contained higher levels of activated pSrc(Y416) than when grown in 1% fetal calf serum (lanes 2-7). This cell proliferation effect was not observed for any of the other signaling proteins examined. To confirm that these cell lines expressed low levels of activated pSrc in 1% fetal calf serum, we also measured the level of pSrc(Y416) in αIIbβ3-expressing Chinese hamster ovary (CHO) cells adhered to Fg (Figure 4D). Here, pSrc(Y416) levels were readily detected and upregulated. The levels of deactivated pSrc(Y527) in MDA-MB-435 and MDA-MB-231 cells also reached a maximum at two hours, while they increased in MCF7 cells after two hours. In contrast to the cancer cells, Hek-293 cells expressed high and unaltered levels of deactivated Src.
FAK levels remained unchanged in all cell lines, except after two hours of treatment in MDA-MB-435 cells (Figure 4E). The levels of activated pFAK also remained unaltered in MCF7 and Hek-293 cells but did transiently increase at two hours in MDA-MB-435 cells, which correlated with maximal levels of pERK and pSrc(Y416). MDA-MB-231 pFAK levels increased after one hour which correlated only with their pERK levels. Therefore, we observed heterogeneity in MAPK and Src signaling by the breast cancer cells.
Immunocytochemistry
Integrin signaling is complex as it not only governed by the binding of an ECM ligand but it is also regulated by the recruitment and interaction of integrin-associated proteins with integrin clusters and the formation of integrin-based structures, such as focal adhesions. As adhered breast cancer cells differed in their signaling (Figure 4), we investigated if these differences in signaling were due to changes in integrin-based structures. Therefore, experiments were performed to determine whether the differences were due to changes in the subcellular distribution of F-actin stress fibers or the formation of focal adhesions when the cells were allowed to attach to and spread on ECM ligands (Figure 5). The cells were plated onto coverslips coated with collagen, Fg, FN or VN, and allowed to adhere overnight. Cells were fixed, permeabilized, and stained for F-actin and focal adhesions. F-actin stress fibers were easy to identify and major differences in the distribution and organization of F-actin fibers were observed (Figure 5A). In MDA-MB-435 cells adhered to the four ECM ligands, many bundles of stress fibers spanning the core of the cells were observed, and adherence to FN and VN induced the greatest formation of stress fibers. In MDA-MB-231 cells, F-actin was mainly present at the perimeter of the cell and localized to membrane protrusions resembling filopodia. When grown on FN and VN, MDA-MB-231 cells contained more and denser clustering of the protrusions than MDA-MB-435 cells. The distribution of F-actin in MCF7 was condensed and localized to the leading edge of spreading cells. In contrast, Hek-293 cells were almost devoid of stress fibers.
Vinculin is a prominent component of focal adhesions and it induces integrin clustering and focal adhesion formation through interactions with talin, an actin-integrin linkage protein [21]. Therefore, focal adhesions were visualized using vinculin staining (Figure 5B). Compared to the three other cell lines, MDA-MB-435 adhered to the four ECM ligands show enhanced focal adhesion formation, which correlated with the presence of strong stress fibers. Some focal adhesions were found distributed at the periphery of MCF7 cells, while only FN induced the formation of a few focal adhesions in MDA-MB-231 cells. No focal adhesions were detected in Hek-293 cells.
The staining pattern with anti-talin was similar to that of vinculin (Figure 5C). As talin is reported to be both an integrin-linkage protein and an integrin activator [22], its recruitment to focal adhesions also serves as a mechanism for focal integrin activation and signaling. In MDA-MB-435 and MCF7 cells adhered to any of the ligands, talin staining revealed a diffuse distribution of talin within the cytoplasm and a strong recruitment of talin to focal adhesions localized to lamellipodia and filopodia. In MDA-MB-231 cells adhered to collagen, Fg and VN, very few focal adhesions were detected using talin staining. However, a dot-like distribution pattern resembling focal complexes was observed in MDA-MB-231 cells adhered to FN. Hek-293 cells did not form any focal adhesions and cell spreading was much higher on FN than on the other ligands. Therefore we observed that MDA-MB-435 cells expressed the highest level and organization of actin-integrin linkage structures and focal adhesions. The higher level of focal adhesions in the MDA-MB-435 cells is consistent with our observation that this cell line had the strongest correlation between PMA-induced activation of pFAK, pSrc(416) and pERK (Figure 4). Furthermore, our MDA-MB-435 data is consistent with previous findings that higher expression levels of integrin αvβ3, are associated with well-developed focal adhesions and thicker stress fibers in primary breast cancer cells compared with the normal breast epithelial cells [23]. Finally, we also observed that a two hour treatment of cells with PMA induced stress fiber perturbations in all cell lines, loss of focal adhesions in MDA-MB-435 cells and induced some MCF7 cells into apoptosis (data not shown).
uPAR and VEGFR expression
Integrin signaling is a dynamic process, being influenced by a number of factors including the cross-talk with other cell surface receptors, such as uPAR and VEGFR. These two receptors are also implicated in breast cancer tumor progression and invasiveness. Signaling by uPAR requires interactions with integrin or other co-receptor as it lacks a transmembrane and an intracellular domain [14]. uPAR also contributes to breast cancer development by directly supporting cell adhesion to VN, and by coordinating ECM proteolysis and remodeling through activation of plasmin and breakage of integrin-ECM linkages that allow for cell migration and metastasis [14]. The interaction of VEGFR with integrins, such as αvβ3, αvβ5 and α5β1, is involved in cancer-induced angiogenesis that facilitates the growth and progression of breast cancers [11]. Therefore, the levels of uPAR and VEGFR expressed by the cell lines were determined (Figure 6).
The breast cancer and Hek-293 cells all expressed uPAR, with MCF7 expressing slightly higher levels of uPAR than MDA-MB-231 and MDA-MB-435 cells (Figure 6). As all cells, and in particular MCF7 cells, adhered well in the absence of an agonist (Figure 3), we questioned whether uPAR may have been involved in the upregulated adhesion. To address this question we also determined the levels of uPAR in GM1500 cells which we demonstrated had low adherence in the absence of a cell agonist. However, we found that uPAR levels in GM1500 cells were similar to those of MDA-MB-231 and Hek-293 cells (Figure 6). This led us to conclude that the levels of uPAR expressed in MDA-MB-231 and Hek-293 cells were insufficient to upregulate cell adherence. In contrast to uPAR expression, VEGFR expression varied greatly between the cell lines (Figure 6). MCF7 cells expressed greater than 10-fold more VEGFR compared to MDA-MB-435 and GM1500 cells, while MDA-MB-231 and Hek-293 cells expressed low to moderate amounts, respectively. In addition, we determined that all cell lines produced very low amounts of VEGF (data not shown). Thus, MCF7 cells were readily distinguished from the metastatic cells based upon their expression of VEGFR.
Adhesion-induced differential signaling
During the adherence of a cell to the ECM, integrins interact with a number of matrix and cellular proteins that result in the activation of signaling pathways resulting in changes in cellular function and biology. As the breast cancer cells used in this study differed in their capacity to form focal adhesions, we explored the possibility that part of the heterogeneity of breast cancer was due to variations in adhesion-induced signaling through MAPK and Src pathways by different breast cancers. In looking at the Src pathway, we discovered that Src was highly deactivated in all cell lines and that the level of pSrc(Y527) and c-Src were unchanged by adherence to ECM proteins (data not shown). Therefore, we focused our attention on the MAPK pathway by first ascertaining if there was constitutive signaling from integrins through to ERK by measuring the levels of pFAK, pMEK, and pERK in non-adherent suspension cells (Figure 7A). All cancer cells contained activated pFAK, pMEK, and pERK in suspension, with MDA-MB-231 cells expressing much greater levels of pFAK and pMEK. Hek-293 suspension cells contained very low levels of pMEK and pERK, and typical of a nonadherent cell, they contained undetectable levels of pFAK.
As MDA-MB-231 suspension cells expressed the highest levels of pFAK and pMEK, but MDA-MB-435 expressed the highest levels pERK, we further investigated the differences in their regulation of MAPK pathway using adhered cells (Figure 7B). Adhered MDA-MB-231 cells contained higher levels of pFAK compared to MDA-MB-435 cells, but only MDA-MB-435 cells exhibited a slight but reproducible adhesion-dependent increase in pFAK. This result was consistent with MDA-MB-435 cells containing more focal adhesions than MDA-MB-231 cells (Figure 5). Adhesion of MCF7 cells to ECM ligands resulted in only small changes in pFAK, while Hek-293 cells contained no pFAK (Figure 7B). The absence of activated pFAK in Hek-293 cells was consistent with this cell line containing no focal adhesions. The levels of pMEK and pERK in non-metastatic MCF7 cells clearly distinguished this cell line from the metastatic MDA-MB-435 and MDA-MB-231 cells. Adhered MCF7 cells contained nearly undetectable levels of pMEK and pERK, while MDA-MB-435 and MDA-MB-231 cells contained high levels of both these proteins. Most adhered Hek-293 cells contained low but detectable levels of pMEK and pERK, and pERK levels increased following adhesion.
Adhesion-induced changes in pMEK and pERK levels also distinguished MDA-MB-435 from MDA-MB-231 cells (Figure 7B). There was an adhesion-dependent increase in pMEK levels in MDA-MB-435 cells, but not in MDA-MB-231 cells. It also appeared that there was constitutive activation of pMEK in MDA-MB-231 cells, as the level of pMEK in suspension cells were similar to those found in adhered MDA-MB-231 and MDA-MB-435 cells. However, once again, high pMEK levels in adhered metastatic MDA435 and MDA231 cells separated these cells from non-metastatic MCF7 and Hek293 cells. The effects of adhesion on the level of pERK in MDA-MB-435 and MDA-MB-231 cells contrasted those of pMEK. Here we observed an adhesion-dependent increase in pERK levels in MDA-MB-231 cells, but not in MDA-MB-435 cells. These differences were not due to changes in total FAK, MEK or ERK levels which remained unaltered (data not shown). As ERK is immediately downstream from MEK, we speculate that the differences in pERK levels were due to differences in the regulation of pERK-related phosphatase activity within these cells. In MDA-MB-231 cells, we propose that adhesion suppresses phosphatase activity allowing for pERK levels to increase, while in MDA-MB-435 cells, either adhesion increases phosphatase activity or pERK levels in suspension cells are already at maximal. Whatever explanation is correct, there were differences in MAPK signaling between MDA-MB-435 and MDA-MB-231 cells and a marked reduction in MAPK signaling by MCF7 cells. We also noted that there are likely other non-integrin receptors involved in cell adhesion-induced signaling as adhesion to BSA resulted in increased pFAK, pMEK and pERK levels in some cell lines.
We also examined the effect of cell adhesion on Bcl2 and pErb2 levels. Bcl2 is an important regulator of apoptosis and Bcl2 itself is regulated by integrin signaling. pErbB2 is involved in signal pathways leading to cell growth and differentiation which are two cellular processes regulated by integrin signaling. Therefore, we determined the effect of cell adhesion on Bcl2 and pErb2 levels to identify any correlations in changes in their levels to that of pMEK, pERK or pFAK. Bcl2 levels were unaffected by cell adhesion, and similar to the levels of phosphorylated kinases, no major differences in Bcl2 levels were found in cells adhered to FN versus Fg or collagen. MDA-MB-435 expressed the highest levels Bcl2, but expressed the lowest level of activated pErbB2. MDA-MB-231 and MCF7 cells expressed similar amounts of pErbB2 while Hek-293 cells expressed the lowest, in agreement with pErbB2 being a prognostic marker for some breast cancers.