The ability of colon tumor cells to invade adjacent structures and to metastasize at distant sites is the major cause of unfavorable outcome for colon cancer patients. However, the molecular mechanisms involved in these different steps remain largely unknown. Actin cytoskeletal reorganization is crucial for many different aspects of tumor progression, notably cell motility, cell adhesion, morphogenesis or cytokinesis, and is controlled to a great extent by the Rho family of GTPases (Rho, Rac and Cdc42).1, 2 Remodeling of the actin cytoskeleton in response to extracellular stimuli acting through cell surface receptors is mediated by multiple actin-binding proteins that depolymerize/polymerize, nucleate, cap, sever or cross-link actin filaments.3, 4
The actin depolymerizing factor (ADF)/cofilin family of proteins so far comprises 3 members: cofilin-1, a muscle-specific isoform cofilin-2 and destrin (another name of ADF in humans). These proteins are essential for enhancing the turnover rate of actin through their ability to sever actin filaments thereby creating new free barbed ends and to depolymerize filaments at their pointed ends (reviewed in 3, 5–7). ADF/cofilin proteins are mainly inactivated through phosphorylation on Ser3 by LIM kinase (LIMK) and TES kinase (TESK).8 Specific phosphatase slingshot (SSH) and chronophin are involved in the reactivation of ADF/cofilin proteins by dephosphorylation.9, 10, 11
Transient activation of cofilin-1 plays a key role in free barbed end generation at the leading edge and in lamellipodium formation required for efficient EGF chemotaxis of mammary cancer cells.7, 12, 13 Moreover, a balance among cofilin-1, LIMK and SSH activities is required for the directionality of cancer cell migration.14, 15 In colon cancer, the role of cofilin-1 in cell migration is poorly documented. An increase of cofilin-1 phosphorylation was observed in K-RasG13D-deleted HCT116 colon cancer cells for which the rate of wound closure was decreased, compared to K-RasG13D expressing cells.16 Moreover, the expression of the active nonphosphorylated cofilin-1 in the human colon cancer SW620 cell line was positively correlated with the expression of the adhesion molecule CD44,17 known to be linked to the invasion of colon cancer cells.18
It is noteworthy that most studies focused only on the role of cofilin-1 in cell motility although destrin is coexpressed with cofilin-1 in many mammalian cells.5, 19 Destrin and cofilin-1 share about 70% amino acid sequence identity, and both proteins can rescue lethality of yeast cofilin null mutants.20 However, even though these 2 proteins have similar biochemical properties, they are not functionally identical. Indeed, destrin is more efficient to sequester actin monomers and to depolymerize F-actin than cofilin-1, and their activity is differentially regulated by pH variations.19, 21, 22, 23, 24 Increasing the monomeric actin pool in a murine myoblast cell line through expression of a mutant β-actin or with latrunculin A decreases ADF but not cofilin-1 expression, suggesting that their expression is differentially regulated within the same cell.25 Cofilin-1 is usually considered ubiquitous whereas destrin expression seems specific of epithelia and endothelia,19 suggesting that they may behave differently according to cell type and situations. Functional differences between these 2 proteins thus deserve investigation when both are expressed by a given cell type.
Because cofilin-1 involvement in colon cancer cell invasion is poorly documented, and that of destrin remains completely unknown, we investigated in a comparative manner the role of these 2 proteins in processes leading to colon tumor cell invasion, and we asked whether these 2 proteins are involved in the same biological functions. For this purpose, we used small interfering RNAs (siRNAs) to specifically reduce either cofilin-1 or destrin expression in the human colon cancer Isreco1 cell line. Our data indicate that cofilin-1 and destrin play a similar role in cell adhesion to extracellular matrix (ECM) components, but that only destrin is essential for Isreco1 cell migration on collagen I, and for cell invasion upon stimulation with the proinvasive neuroendocrine peptide bombesin (BBS). A relationship between destrin, but not cofilin-1, and p130Cas is suggested to partly explain the differential involvement of these 2 proteins in Isreco1 colon cancer cell migration and invasion.
Collagen I, collagen IV and Matrigel were obtained from Becton-Dickinson (Le Pont-de-Claix, France). Laminin I and bombesin (BBS) were purchased from Sigma (St Quentin Fallavier, France). Rabbit polyclonal anti-cofilin-1 antibody and human recombinant cofilin-1 were from Cytoskeleton (Tebu, le Perray en Yvelines, France). Human recombinant destrin was a kind gift from Dr. J. Bamburg (Colorado State University, Fort Collins, CO). Rabbit polyclonal anti-destrin/ADF and mouse monoclonal anti-α-tubulin antibodies were from Sigma. Rabbit polyclonal anti-FAK antibody was from Biosource (Clinisciences, Montrouge, France). Mouse monoclonal anti-p130Cas antibody was from Upstate Biotechnology (Euromedex, Mundolsheim, France). The antibodies directed against the phosphorylated forms of cofilin-1 (pSer3), FAK (pTyr397) and p130Cas (pTyr165) were from Santa Cruz, Biosource and Cell Signaling (Ozyme, Saint Quentin en Yvelines, France), respectively. The horseradish peroxydase-conjugated antibodies against rabbit or mouse IgG, and FITC-conjugated donkey anti-mouse IgG were purchased from Jackson ImmunoResearch (Interchim, Montluçon, France). Thirteen human colorectal cancer cell lines, obtained from Dr. R. Hamelin (INSERM U762, Paris, France), were used in our study. All cell lines were well-characterized in terms of microsatellite instability, LOH status and mutations of most common oncogenes and tumor suppressor genes.26 The cells were grown in Dulbelcco's modified Eagle's medium (DMEM, Invitrogen, Cergy-Pontoise, France) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine and antibiotics (100 IU/ml penicillin + 50 μM streptomycin) in a humidified CO2:air (5:95%) incubator at 37°C.
Cofilin-1- and destrin-specific small interfering RNA (siRNA) duplexes were designated using the algorithm described by Elbashir et al.27 and were synthesized by Eurogentec (Seraing, Belgium) and Ambion (Ambion/Applied biosystems, Courtaboeuf, France). The following oligonucleotides were used: (i) cofilin-1 sense 5′-GGUGUUCAACGACAUGAAG-3′, cofilin-1 antisense 5′-CUUCAUGUCGUUGAACACC-3′; (ii) destrin : sequence (a) sense 5′-GUAGCUGAUGAAGUAUGUC-3′, sequence (a) antisense 5′-GACAUACUUCAUCAGCUAC-3′; sequence (b) sense 5′-GCUAGGACACCUGUGGUAU, sequence (b) antisense 5′-AUACCACAGGUGUCCUAGC (when not specified, destrin siRNA refers to destrin sequence (a)); (iii) non silencing control sense 5′-UUCUCCGAACGUGUCACGU-3′, non silencing control antisense 5′-ACGUGACACGUUCGGAGAA-3′ (Qiagen, Courtaboeuf, France). A BLAST search confirmed that the selected oligonucleotide sequences would not interfere with other genes. Isreco1 (3 × 105/60 mm dish), HCT116 (5 × 105/60 mm dish) or LS174T (5 × 105/60 mm dish) cells were transfected with the siRNA duplexes by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's recommendations. Transfection of a fluorescent siRNA duplex, followed by flow cytometry analysis, resulted in a mean transfection rate of 70–80% for all cell lines tested.
Cell staining and fluorescence microscopy
Seventy-two hours after transfection with cofilin-1, destrin or control siRNA, Isreco1 cells were detached with trypsin/EDTA, and kept in suspension in DMEM-0.1% BSA for at least 1 hr at 37°C. About 3 × 104 cells/well were then replated onto collagen I-precoated 4-well chamber slide. After incubation for 1 hr at 37°C, cells were fixed for at least 10 min in a fresh solution of 4% paraformaldehyde in PBS, and permeabilized for 5 min in a solution of 0.2% Triton-X100 in PBS. To visualize the actin cytoskeleton, F-actin was stained with FITC-conjugated phalloidin (Invitrogen) in PBS for 30 min at room temperature, according to the manufacturer's recommendations. To visualize adhesion structures, cells were incubated with anti-paxillin antibody (1:250 dilution in PBS-10% FCS) for 1 hr at room temperature and then with FITC-conjugated donkey anti-mouse IgG (1:100 dilution in PBS-10% FCS) for 30 min at room temperature.
Seventy-two hours after siRNA transfection, cells were suspended in DMEM-0.1% BSA and replated onto 10 μg/ml collagen I-, IV- or laminin 1-precoated 96-well cell culture dishes at the concentration of 5 × 104 cells/well. After incubation for 1 hr at 37°C, cells were washed twice with PBS, and adherent cells were fixed in a solution of 4% paraformaldehyde for 30 min and stained with 1% borax and 1% methylene blue. After solubilization with 1% SDS, absorbance (630 nm) was measured.
Migration and invasion assays
Experiments were carried out as previously described.28 Briefly, 110 μl of siRNA-treated cell suspension were seeded into the upper chamber of a 12-well chemotaxis chamber (Neuroprobe, Gaithersburg, MD) at a final concentration of 1 × 106 cells/ml. For migration assays, both upper and lower compartments were filled with DMEM containing 0.1% BSA (Isreco1 and HCT116 cells) or 2.5% FCS (LS174T cells) and were separated with a polycarbonate membrane (12-μm pore size) precoated for 4 hr with 100 μg/ml type I collagen. Cells were then incubated for 3 hr at 37°C in 5% CO2 humidified conditions. For invasion assays, lower wells were filled with DMEM-0.1% BSA with or without BBS (1 nM). The polycarbonate membrane was precoated for 2 hr with 370 μg/ml Matrigel, and Isreco1 cells were incubated for 17 hr at 37°C. Cells were then fixed with 4% paraformaldehyde, and stained with 1% borax and 1% methylene blue. Cells of the upper surface of the filter were removed with a cotton swab, and those underneath were quantified.
Western blot analysis
Cells were lysed in cold lysis buffer containing 50 mM HEPES (pH 7.5), 1 % Triton X-100, 150 mM NaCl, 2 mM Na3VO4, 100 mM NaF, 100 U/ml aprotinin, 20 μM leupeptin, 5 mM benzamidine, 2 μM pepstatin and 0.2 mg/ml phenylmethylsulfonylfluoride (PMSF). Cell extracts were clarified (14,000g, 15 min, 4°C) and aliquots of lysates were diluted in 2X SDS-PAGE sample buffer (125 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 10% 2-mercaptoethanol and 0.16% bromophenol blue), boiled for 5 min and resolved on SDS-polyacrylamide gels. After electrophoresis, proteins were transferred onto nitrocellulose membranes that were blocked using 5% (w/v) non fat dried milk in Tris-buffered saline containing 0.1% Tween 20, and exposed to the primary antibodies overnight at 4°C. After incubation with appropriate secondary antibodies conjugated to horseradish peroxidase, blots were revealed using the ECL method (Covalab, Lyon, France). Signals were quantified using ImageQuant software (Molecular Dynamics).
Measurement of destrin phosphorylation by isoelectric focusing
Cells were serum-starved for 24 hr and treated with 10 nM BBS for 0–60 min. Cells were then washed twice with cold PBS and lysed in 20 mM Tris buffer (pH 7.5) containing 1 mM EDTA, 1 mM EGTA, 1% Triton X-100 and protease inhibitors. Equal amounts of proteins (25 μg) were mixed with 2X glycerol 50% and separated on a precast pH 3–10 gel according to the manufacturer's protocol (Bio-Rad, Marnes la Coquette, France). Proteins were denatured and negatively charged in 250 mM Tris-HCl pH 8.0, 3% 2-mercaptoethanol and 2% SDS for 10 min at room temperature, then transferred onto nitrocellulose membrane, and immunobloted with anti-destrin antibody, as described earlier. Because destrin and phospho-destrin have 2 different isoelectric points, phosphorylated protein appears slightly lower on the gel than the unphosphorylated form.
Results were calculated as the mean ± SEM. Data were analyzed using Student's t test, and differences between 2 means with a value of p < 0.05 were regarded as significant.
Cofilin-1 and destrin are coexpressed in human colon carcinoma cell lines
ADF/cofilin family proteins have been found in numerous non tumor cells, but their expression in colon cancer cells was not known. We therefore examined the level of cofilin-1 and destrin in various human colon cancer cell lines by Western blot experiments using 2 polyclonal antibodies for which specificity towards cofilin-1 and destrin, respectively, has been verified (Fig. 1b). As shown in Figure 1a, cofilin-1 and destrin were coexpressed in all tested cell lines. No correlation was observed between cofilin-1 or destrin expression, and microsatellite instability or p53 mutations. To quantitatively determine the relative levels of expression of both proteins in the different cell lines, we first calculated the amount of cofilin-1 and destrin in Isreco1 cells, a well-characterized model of invasive colon cancer28 (Fig. 1b). By comparison with known amounts of recombinant human cofilin-1 and destrin, we found that cofilin-1 was expressed at a 5-fold greater level than destrin in Isreco1 cells, accounting for nearly 1.2% of the Triton X-100 soluble protein fraction (Figs. 1b and 1c). By extrapolation from the values calculated in Isreco1 cells, we then estimated the ratio of destrin over the total pool of cofilin-1 + destrin in all the cell lines (Fig. 1d). We found that: (i) cofilin-1 was predominantly expressed in all the cell lines tested and (ii) Isreco1 cells had the highest proportion of destrin, corresponding to ∼17% of cofilin-1 + destrin total pool. This prompted us to choose the Isreco1 cell line as a favorable model to investigate the respective roles of cofilin-1 and destrin.
siRNA transfection of Isreco1 cells results in an efficient and specific reduction of cofilin-1 and destrin expression
To establish the possible role of cofilin-1 and destrin in the invasive phenotype of human colon cancer Isreco1 cells, we used specific siRNAs to silence the expression of either cofilin-1 or destrin. Transfection with increasing amounts of cofilin-1 siRNA (0.1 to 100 nM) indicated that optimal cofilin-1 protein depletion at 48 hr after transfection was obtained with 10 nM siRNA (Fig. 2a). Cofilin-1 expression analysis of cell lysates from different time points after transfection (from 24 to 120 hr) with 10 nM cofilin-1 siRNA revealed a massive and maximal reduction of cofilin-1 expression 72 hr after transfection (90 ± 3% silencing, Fig. 2b). Cofilin-1 siRNA was designed to specifically target cofilin-1 mRNA, excluding any cross-reaction with destrin mRNA, so that destrin protein expression remained unmodified under cofilin-1 siRNA treatment (Fig. 2b). Similarly, 10 nM destrin siRNA efficiently repressed destrin expression, with a maximal effect at 72 hr (86 ± 2% silencing), while cofilin-1 expression was unaffected by destrin siRNA transfection (Figs. 2c and 2d). Transfection with control nonsilencing siRNA affected neither cofilin-1 nor destrin expression. As a control to validate the functional repercussion of cofilin-1 and destrin siRNA, we looked at multinucleated cells since proteins of the ADF/cofilin family have been shown to be involved in cytokinesis, notably at the contractile ring for proper cell cleavage.29, 30 As expected, the number of multinucleated cells increased 2- to 3-fold both in cofilin-1 and destrin siRNA-treated cells (13% of total cell population compared to 5% when cells were transfected with control siRNA, p < 0.05) (Fig. 2e), an observation consistent with a cytokinesis defect and in line with previous reports.30
Cofilin-1 and destrin are important for the migratory and adhesive morphology of cells
To further analyze the impact of reduction of cofilin-1 or destrin expression on cell morphology, Isreco1 cells transfected with control, cofilin-1 or destrin siRNA were plated onto collagen I-coated dishes and stained with FITC-phalloidin or paxillin antibody to visualize the actin cytoskeleton and focal adhesion structures, respectively. A significant proportion of control cells (55 ± 3%) spontaneously exhibited a keratocyte-like, polarized, migratory morphology defined by the extension of a broad lamellipodium at 1 edge, and small adhesion structures at the periphery (Fig. 3a). Interestingly, the percentage of polarized cells strongly decreased in cofilin-1 or destrin siRNA-treated Isreco1 cells (24% ± 2% and 27% ± 3%, respectively, p < 0.05, Fig. 3b). Those cells were less polarized, with larger paxillin-containing adhesion located both at the cell periphery and at the base of protrusions. These observations suggested that both cofilin-1 and destrin could be involved in localized lamellipodium protrusion and focal adhesion turnover in Isreco1 cells.
Reduction of cofilin-1 or destrin expression enhances cell adhesion to ECM components
Since the morphology of cofilin-1- or destrin-depleted Isreco1 cells suggested a more adherent phenotype than control cells, we investigated the role of cofilin-1 and destrin in Isreco1 cell adhesion. Control, cofilin-1 or destrin siRNA-treated Isreco1 cells were seeded onto collagen I coated dishes, and adherent cells were quantified after 1 hr (Fig. 3c). We found that the number of adherent cells was similarly greater when cofilin-1 or destrin expression was decreased in Isreco1 cells, compared to control cells (170% ± 5% and 173% ± 5%, respectively, p < 0.05). An enhanced cell adhesion was also observed on collagen IV and laminin 1 when cofilin-1 or destrin were silenced (Fig. 3c). These data suggested that both cofilin-1 and destrin played a major role in the control of Isreco1 cell adhesion to the ECM.
Destrin, but not cofilin-1, is involved in Isreco1 cell migration and invasion
Remodeling of the actin cytoskeleton by actin-binding proteins is requisite for cell motility,3, 31 which largely depends on the ability of a cell to extend a pseudopod in the direction of migration, to form transient contacts with the substrate at the leading edge, and to pull the cell body forward, resulting in net cell translocation. Since our data indicated that polarized lamellipodium formation and cell adhesion were cofilin-1 and destrin-dependent in Isreco1 cells, we then investigated the effect of the reduction of cofilin-1 or destrin expression on cell migration. For this purpose, we measured the migration of cofilin-1 and destrin siRNA-treated Isreco1 cells across type I collagen-coated filter using a Neuroprobe chamber system as previously described.28 Surprisingly, the number of Isreco1 cells reaching the lower side of the filter was not modified by cofilin-1 depletion, compared to control cells (Fig. 4a). A different cofilin-1 siRNA that was equally effective in inhibiting cofilin-1 expression also did not affect Isreco1 cell migration (data not shown). In contrast, siRNA-induced destrin silencing significantly reduced Isreco1 cell migration, by about 30% with the sequence previously used in Figures 2 and 3 (dest(a), migration representing 67 ± 3% of control level, p < 0.05) or by about 15% when using a different specific sequence (dest(b), 84 ± 1% of control, p < 0.05) (Fig. 4a). Interestingly, the defect in Isreco1 cell migration closely paralleled the level of destrin silencing obtained after siRNA treatment, with the greatest effect observed with dest(a) siRNA (Fig. 4b). As sequence (a) appeared to be the most efficient, only this siRNA was used in subsequent experiments to reduce destrin expression. To determine whether the observed differential effect of cofilin-1 and destrin depletion was specific to the process of cell migration on collagen I or not, we measured the invasion of cofilin-1 and destrin siRNA-treated cells using our previously validated system of BBS-induced Isreco1 cell invasion through Matrigel.28 The number of invading cells was similar between control and cofilin siRNA-treated cells, in both basal and BBS-stimulated conditions (Fig. 4c). On the contrary, the basal invasion of destrin-depleted cells was slightly but significantly reduced compared to control siRNA-treated cells (82 ± 4% of control, p < 0.05). The BBS stimulatory effect was also reduced in destrin siRNA-treated cells by ∼50% (p < 0.05) since invasion in the absence or presence of BBS was 82% and 106% of control, respectively, in destrin siRNA-treated cells (i.e. a 1.3-fold induction compared to 1.6 in control siRNA-treated cells).
Impairment of cell migration and invasion cannot be imputed to an effect of destrin siRNA on cell viability since we verified that the number of metabolically active cells (as measured by MTT assay) was not different between control, cofilin-1 and destrin siRNA-treated cells under the conditions used for those assays (data not shown). Hence, these results pointed to a functional role of destrin in the migration and invasion of colon cancer Isreco1 cells, and indicated that cofilin-1 was not a central regulator of these processes. This suggested that cofilin-1 and destrin may be differentially involved in signaling pathways leading to migration and invasion of Isreco1 cells.
To further analyze the role of destrin in the process of tumor cell invasion, we measured the impact of destrin or cofilin-1 silencing on the migratory capacities of 2 other colon cancer cell lines, chosen on the basis of their relative amount of destrin that was either moderate (HCT116) or low (LS174T) compared to Isreco1 cells (see Fig. 1d and table Fig. 4d). Destrin depletion slightly but significantly reduced HCT116 cell migration (86 ± 3% of control, p < 0.05) but did not affect LS174T cell migration (92 ± 10% of control, p = 0.6) (Fig. 4d, grey bars). As far as cofilin-1 is concerned, its depletion led to a significant reduction of HCT116 cell migration (75 ± 3% of control, p < 0.05), and exerted an even greater inhibitory effect on LS174T cell migration (67 ± 6% of control, p < 0.05) (Fig. 4d, black bars). The efficiency of destrin and cofilin-1 silencing was verified by Western blot (Fig. 4e). These results showed a positive correlation between the destrin/cofilin-1 ratio in a given cell line and the involvement of destrin vs. cofilin-1 in cell migration. This suggested that, in some cases, destrin could play a predominant role in the invasive properties of colon cancer cells, despite an expression level much lower than cofilin-1. This also confirmed the interest of the Isreco1 cell line as a valuable model to address this question.
Bombesin rapidly dephosphorylates cofilin-1 and destrin
Since ADF/cofilin family proteins are commonly regulated by phosphorylation/dephosphorylation on Ser3 in several cell types, and activation of particular surface receptors can rapidly induce cofilin dephosphorylation,5, 32, 33 we determined whether BBS was able to modulate cofilin-1 and destrin phosphorylation levels in Isreco1 cells. The level of phospho-cofilin-1 was analyzed by Western blot using an antibody specific to the phosphorylated form of cofilin-1 (Fig. 5a). Cofilin-1 was rapidly dephosphorylated 1 min after BBS treatment (10 nM) and its phosphorylation level did not return to basal even after 1 hr. In these conditions, the total level of cofilin-1 protein was not modified. No phospho-cofilin-1 or cofilin-1 signals were detected in cells transfected with cofilin-1 siRNA. Because of the lack of commercially available antibody specific to the phosphorylated form of destrin, we used isoelectric focusing (IEF) gels to separate the unphosphorylated form of destrin from the phosphorylated inactive form. Figure 5b shows that destrin was mostly present in a non phosphorylated form that represented over 75% of total destrin in nonstimulated Isreco1 cells. BBS treatment decreased the phospho-destrin level in 2 transient waves, a first decrease at 1 min, and a second one between 15 and 30 min after treatment. Phospho-destrin level returned to basal at 60 min. Between these 2 activation peaks, the phosphorylation level measured after 5 min of BBS stimulation was similar to that of nonstimulated cells. As a control, we did not detect any phospho-destrin or destrin in cells transfected with destrin siRNA. These results showed that BBS treatment of Isreco1 cells induced differentially sustained changes in the phosphorylation levels of cofilin-1 and destrin.
Destrin is required for adhesion-stimulated p130Cas phosphorylation
In an attempt to identify mechanisms explaining the differential effects observed in the migratory properties of cofilin-1 and destrin siRNA-treated Isreco1 cells, we examined the activation of the adapter protein p130Cas. Indeed, p130Cas is known to be localized in focal adhesions and to regulate cell migration through its phosphorylation by focal adhesion proteins such as the focal adhesion kinase (FAK).34, 35 Tyrosine phosphorylation levels of p130Cas and FAK in control, cofilin-1 or destrin siRNA-treated Isreco1 cells were therefore determined under the conditions used for the migration assay. Following cell adhesion to collagen I for 15 min, total lysates were prepared and analyzed by Western blot with specific antibodies. As expected for control cells, p130Cas was phosphorylated on Tyr165 after adhesion to collagen I (Fig. 6a, upper panels). Interestingly, adhesion-induced phosphorylation of p130Cas was decreased by 56% (p < 0.05) in cells with reduced destrin expression whereas cofilin-1 siRNA had no effect (Fig. 6a). Phosphorylation of FAK on Tyr397 was also induced by cell adhesion but neither destrin nor cofilin-1 siRNA significantly modified FAK phosphorylation level (P-FAK/FAK ratios representing 1.3 ± 0.2, p = 0.15 and 1.2 ± 0.1, p = 0.2 for destrin and cofilin-1 siRNA, respectively, normalized to control siRNA) (Fig. 6a, lower panels). Since p130Cas has been shown to be involved in cell migration in other cell models,36, 37 our results support the idea that the effect of destrin silencing on Isreco1 cell migration could be exerted through a defect in p130Cas activation following integrin engagement to collagen I. We also determined p130Cas phosphorylation level of siRNA-transfected cells after adhesion to Matrigel (Fig. 6b). Similar to what obtained on collagen I, only silencing of destrin decreased p130Cas phosphorylation induced after Isreco1 cell adhesion for 30 min to Matrigel (no significant reduction of phospho-p130Cas level was observed at 60 min, p = 0.15) whereas p130Cas phosphorylation level in cofilin-1 siRNA-treated cells was comparable to that of control cells.
The mechanisms involved in colon cancer cell invasion, a critical step of tumor progression leading ultimately to metastasis dissemination, are only partly understood. An important feature of malignant tumor cells is their ability to modulate their adhesion and motility in a dynamic manner to migrate and invade the local environment. The reorganization of actin cytoskeleton stands central in these different aspects of tumor cell invasion. However, the role of actin-binding proteins directly regulating actin dynamics in colon cancer cells has been poorly investigated. In the present study, we examined the relative contribution of cofilin-1 and destrin, 2 members of the ADF/cofilin family, in adhesion, migration and invasion of human colon cancer Isreco1 cells.
Coexpression of cofilin-1 and destrin in many cell types may indicate that, despite their close sequence similarity, these proteins could have complementary and/or distinct functions in a given cell type. Among the rare studies addressing this issue and specifically comparing the roles of destrin vs. cofilin-1, some previously suggested functional differences between the 2 proteins. Indeed, destrin has been shown to be a more potent factor than cofilin-1 for depolymerizing actin filaments in cosedimentation assays and increasing actin filament turnover in measurements of treadmilling rate of actin filaments.19, 22, 23 Moreover, differences in the relative abundance and expression patterns of cofilin-1 and destrin mRNA in adult mouse tissues19, 22 may reflect distinct roles of the corresponding protein products.
Interestingly, our results underline distinct contributions of cofilin-1 and destrin in the colon cancer Isreco1 cell migration and invasion. This indicates that, when the 2 family members are coexpressed in the same cell type, destrin can be a predominant regulator of several actin-dependent processes despite a level of expression dramatically lower than that of cofilin-1, supporting the idea that these 2 proteins are not entirely functionally redundant. Also, differently from the situation in NIH3T3 fibroblasts where individual cofilin-1 or ADF silencing increased the expression of the remaining family member,30 no enhancement of destrin expression was observed in cofilin-1 siRNA-treated Isreco1 cells. This suggests that destrin did not compensate for the loss of cofilin-1 in our model. Since destrin can be more efficient than cofilin-1 to induce actin filament disassembly,19, 23 the effect of destrin depletion could also be due to a greater alteration of the β-actin monomer pool than that obtained with cofilin-1 siRNA treatment. Indeed, previous works have shown that increasing the monomeric actin pool triggers a negative feedback regulation of β-actin mRNA stability38 whose localization near the leading edge is known to be required for cell polarity and directional movement of crawling cells.39, 40
On the other hand, our data are in agreement with a similar role of cofilin-1 and destrin in cell adhesion since silencing of either protein led to an equivalent alteration of paxillin-containing adhesion structures and to a similar increase of Isreco1 cell adhesion to various matrix components (collagen I, collagen IV and laminin 1). Our results are consistent with previous reports showing a positive correlation between increased level of the inactive phosphorylated form of cofilin-1 and enhanced adhesion of colon cancer HCT116 cells16 as well as formation of actin stress fibers and enlarged focal contacts upon cofilin-1 and ADF silencing in NIH3T3 fibroblasts.30
The effects of cofilin-1 or destrin depletion in Isreco1 cells are virtually indistinguishable as far as adhesion is concerned, indicating that enhanced cell adhesion cannot be used alone to predict a migratory defect in this system. Interestingly, we here show that tyrosine phosphorylation of the adapter protein p130Cas in response to adhesion on collagen I or Matrigel was decreased specifically in destrin siRNA-treated Isreco1 cells for which migratory and invasive capacities were impaired. It is demonstrated that p130Cas, known to be present in focal adhesions, plays a role in the sequential turnover of focal adhesions and thus regulates cell migration.34, 35, 41 Hence, p130Cas(−/−) fibroblasts contain disorganized short actin filaments and increased number of focal adhesions, and display reduced spreading and migration on fibronectin.42, 43 A role of p130Cas in cell migration through its association with Crk has also been demonstrated. Indeed, phosphorylation of p130Cas following integrin engagement is correlated with enhanced migration in many reports36, 44, 45, 46 and results in the formation of p130Cas-Crk complexes that lead to Rac1 activation through recruitment of the guanine exchange factor DOCK180, resulting in membrane ruffling and efficient cell migration.37, 44, 47 Therefore, our results suggest that destrin, unlike cofilin-1, is engaged in a pathway regulating adhesion-dependent phosphorylation of p130Cas on Tyr165 and cell migration and invasion. Obviously, the direct involvement of p130Cas in these processes remains to be confirmed in our model and further experiments are required to answer this question. Tyrosine phosphorylation of p130Cas can be regulated by FAK following integrin engagement to ECM.34, 35 However, we here show that the activating phosphorylation of FAK is independent of cofilin-1 and destrin in Isreco1 cells, suggesting that reduction of destrin expression decreased p130Cas phosphorylation and Isreco1 cell migration through FAK-independent mechanisms. Investigating other p130Cas-binding protein kinases such as Src as well as p130Cas-Crk complex formation could provide new information on the destrin-dependent signaling pathway(s) to p130Cas and cell motility.
A recent report underlined the importance of the precise spatiotemporal regulation of cofilin-1 activity by both negative (LIMK1) and positive (SSH1) regulators for Jurkat T cell chemotaxis in response to stromal cell-derived factor-1α.15 Moreover, studies aiming at defining the gene expression pattern of the invasive fraction of tumor cells have shown that genes regulating positively and negatively the ADF/cofilin pathway are upregulated in this cell population (reviewed in Ref.48). This finding is in agreement with the essential transient activation of cofilin-1 by EGF for efficient chemotaxis of mammary cancer cells.12, 13, 49 In our model, the distinct phosphorylation kinetic profiles of cofilin-1 and destrin observed upon stimulation with BBS (sustained for cofilin-1 and more transient with at least 2 waves of dephosphorylaton in the case of destrin) could reflect a differential mode of activation of the proteins. This could therefore be a reason why siRNA-mediated silencing of cofilin-1 or destrin in Isreco1 cells has distinct functional repercussions on BBS-stimulated cell invasion. Nevertheless, we cannot exclude that phosphorylation-independent regulatory mechanisms such as binding to phosphatidylinositol (4,5)-biphosphate or protein–protein interactions involving 14-3-3, AIP1 or CAP1,7 could take part in the differential implication of cofilin-1 and destrin in migration and invasion processes of Isreco1 cells.
In conclusion, our results reveal that destrin, a member of the ADF/cofilin family, is an important regulator of different aspects of the invasive properties of the human colon cancer Isreco1 cell line. Cofilin-1 appears to be involved in only a subset of them, such as cell adhesion. Since coexpression of cofilin-1 and destrin is a general trend observed in many other colon cancer cell lines, these results suggest that these 2 proteins may play distinct roles in the invasive behavior of colon cancer.
Authors thank Dr. R. Hamelin (INSERM U762, Paris, France) for providing the tumor cell lines, Dr. M. Billaud (CNRS UMR 5201, Lyon, France) for his help in siRNA design and Dr. J Bamburg (Colorado State University, Fort Collins, USA) for providing the destrin standard. The authors acknowledge the CeCIL (Centre Commun d'Imagerie Laennec) for providing part of the technical devices used in this work.