Cell migration is a complex process requiring density and rigidity sensing of the microenvironment to adapt cell migratory speed through focal adhesion and actin cytoskeleton regulation. ICAP-1 (also known as ITGB1BP1), a β1 integrin partner, is essential for ensuring integrin activation cycle and focal adhesion formation. We show that ICAP-1 is monoubiquitylated by Smurf1, preventing ICAP-1 binding to β1 integrin. The non-ubiquitylatable form of ICAP-1 modifies β1 integrin focal adhesion organization and interferes with fibronectin density sensing. ICAP-1 is also required for adapting cell migration in response to substrate stiffness in a β1-integrin-independent manner. ICAP-1 monoubiquitylation regulates rigidity sensing by increasing MRCKα (also known as CDC42BPA)-dependent cell contractility through myosin phosphorylation independently of substrate rigidity. We provide evidence that ICAP-1 monoubiquitylation helps in switching from ROCK2-mediated to MRCKα-mediated cell contractility. ICAP-1 monoubiquitylation serves as a molecular switch to coordinate extracellular matrix density and rigidity sensing thus acting as a crucial modulator of cell migration and mechanosensing.
Motile cells continuously sample in space and time the heterogeneity in the composition and stiffness of their extracellular matrix (ECM) through integrin-mediated focal adhesions (FAs) (Moore et al., 2010). As a mechanical link between ECM and actin stress fibers, integrins are crucial for force transmission and signal transduction (Moore et al., 2010). FA assembly, growth and maintenance depend on actomyosin traction forces, which adapt to the substrate elasticity (Burridge and Wittchen, 2013). In spite of alternative pathways involving MRCK (which has two isoforms, MRCKα and MRCKβ, also known as CDC42BPA and CDC42BPB, respectively), MLCK (also known as MYLK) or mDia (Burridge and Wittchen, 2013; Chen et al., 2014; Jégou et al., 2013; Totsukawa et al., 2004), a key event is the modulation of cellular contractility through myosin-based contractility and ROCK (which has two isoforms, ROCK1 and ROCK2) activity. However, signaling pathways underlying FA-mediated rigidity sensing and the mechano-response are not fully understood.
ICAP-1 (also known as ITGB1BP1), a negative regulator of β1 integrin, enables the cell to sense ECM density to adapt its adhesive and migratory responses (Millon-Frémillon et al., 2008) and to control fibronectin (FN) remodeling (Brunner et al., 2011; Faurobert et al., 2013). ICAP-1 specifically binds to the cytoplasmic tail of β1 integrin, maintaining the integrin in its inactivated form by competing with the two activators named Kindlin and talin (Brunner et al., 2011; Millon-Frémillon et al., 2008; Montanez et al., 2008; Ye et al., 2014). ICAP-1 also binds to ROCK1 (Peter et al., 2006). Thanks to these interactions, ICAP-1 may be a good candidate for regulating myosin-based contractility and cellular response to ECM stiffness. Tunable post-translational modifications may control ICAP-1 functions enabling the cell to adapt its migratory response. As ubiquitylation is emerging as important for cell migration dynamics and cell contractility (Carvallo et al., 2010; Sahai et al., 2007; Schaefer et al., 2012; Su et al., 2013; Wang et al., 2003), we addressed whether ubiquitylation may control ICAP-1 functions, enabling the cell to adapt its migratory response. Here, we show that ICAP-1 is monoubiquitylated by SMAD ubiquityl regulatory factor 1 (Smurf1). This monoubiquitylation impairs ICAP-1 binding to β1 integrin and is involved in ECM density and rigidity sensing as well as in coordination of the dynamics of adhesion sites and contractile machinery. ICAP-1 monoubiquitylation plays an important role in the responses of migrating cells to mechanical inputs in a β1 integrin-independent manner by promoting the switch from a ROCK2-mediated to an MRCKα-mediated contractility pathway.
ICAP-1 is monoubiquitylated by Smurf1 at the β1 integrin-binding site
To investigate ICAP-1 ubiquitylation, we performed nickel-bead pulldown experiments on Chinese Hamster Ovary (CHO) cells transfected with ICAP-1 either in the presence or absence of co-transfection with His-tagged ubiquitin. The proteasome inhibitor MG132 was added to prevent proteasomal degradation of any ubiquitylated ICAP-1. When expressed alone, ICAP-1 appeared on a western blot an apparent molecular mass that was slightly greater than 20 kDa, whereas co-transfection with His-tagged ubiquitin and pulldown on nickel beads resulted in isolation of ICAP-1 with higher molecular mass forms, with a band above 35 kDa (Fig. 1A), showing that ICAP-1 is indeed ubiquitylated. This band above 35 kDa most likely corresponds to ICAP-1 monoubiquitylation. HA-tagged ubiquitin (HA–Ubi) was also coexpressed with ICAP-1 fused to Flag and our results show that ICAP-1–Flag can be recognized by both anti-Flag and anti-HA antibodies after immunoprecipitation with anti-Flag antibodies (Fig. 1B), confirming that ICAP-1 can be ubiquitylated. Furthermore, to identify which lysine residue is monoubiquitylated, we analyzed whether truncated forms of ICAP-1 could be monoubiquitylated (Fig. 1C). We determined that the monoubiquitylation site was located in the binding domain for β1 integrin. The point mutation of either one of the two lysine residues present in this domain identified lysine K158 as the site of monoubiquitylation, as its replacement with arginine led to the absence of the 35 kDa band (Fig. 1D) without changing the ICAP-1 polyubiquitylation states (Fig. 1D). The non-ubiquitylable K158R mutant was even less stable than wild-type (WT) ICAP-1, suggesting that the monoubiquitylated form of ICAP-1 is not targeted for proteasomal degradation but rather may have a signaling function (Fig. S1A,B). Because Smurf1 catalyzes the ubiquitylation of the integrin activator talin (Huang et al., 2009), we hypothesized that Smurf1 could be responsible for ICAP-1 monoubiquitylation. To test this hypothesis, Smurf1 was silenced by small interfering RNA (siRNA); there was a high efficiency in reducing Smurf1 transcript and protein levels without affecting ICAP-1 expression (Fig. 1E). ICAP1 monoubiquitylation was blocked when Smurf1 was knocked down, suggesting that Smurf1 is necessary for promoting ICAP-1 monoubiquitylation (Fig. 1E,F). A pulldown assay shows that purified recombinant Smurf1–GST is able to bind to exogenously expressed ICAP-1 in CHO cells, in contrast to the null interaction with GST alone (50-fold less) or with the weak binding to GST fused to Smurf2 (10-fold less) (Fig. S1C). Smurf2 had been chosen as a control because overlapping but distinct substrate and regulator specificity has been observed between Smurf1 and Smurf2 (Lu et al., 2008, 2011). The co-immunoprecipitation between Smurf1–Myc and ICAP-1–Flag expressed in CHO cells confirms that Smurf1 and ICAP-1 belong to the same complex (Fig. S1D). A direct interaction between Smurf1 and ICAP-1 was demonstrated by an ELISA assay using purified recombinant GST–Smurf1 and purified recombinant ICAP-1–His (Fig. S1E). Taken together, our results indicate that Smurf1 is responsible for ICAP-1 monoubiquitylation.
The monoubiquitylation of ICAP-1 prevents binding to β1 integrin and regulates β1 integrin-dependent adhesion
According to structure predictions and crystallographic data (Chang et al., 2002; Liu et al., 2013), the monoubiquitylation site is located in the β1 integrin-binding domain of ICAP-1 facing the isoleucine residue important for the binding to β1 integrin (Fig. 2A). As this monoubiquitylation could interfere with the interaction between ICAP-1 and β1 integrin, we used two classical methods to produce an ubiquitylated form of a protein (Torrino et al., 2011; Visvikis et al., 2008), first by co-transfecting ICAP-1 with His-tagged ubiquitin and second by creating a chimera made of ubiquitin fused to the C-terminal tail of ICAP-1 (ICAP-1–Ubi) (Fig. 2B). We tested the ability of WT, non-ubiquitylatable (K158R) and monoubiquitylated ICAP-1 (endogenous ubiquitylation or chimera) to interact with the cytoplasmic domain of either β1 integrin or β3 integrin fused with GST or with GST alone by pulldown assay (Fig. S2A) or by ELISA assay (Fig. 2C). As previously reported (Millon-Frémillon et al., 2008), we confirmed that ICAP-1 specifically interacts with the cytoplasmic domain of β1 integrin (Fig. 2C; Fig. S2A). Furthermore, the non-ubiquitylated K158R mutant retained the ability to interact with the cytoplasmic domain of β1 integrin, whereas both ubiquitylated forms of ICAP-1 (His-tagged and chimeric) lost the capacity to interact with the cytoplasmic domain of β1 integrin (Fig. 2C; Fig. S2A). These results show that ICAP-1 monoubiquitylation prevents the interaction of ICAP-1 with β1 integrin.
Next, we investigated whether the monoubiquitylation of ICAP-1 could affect FA organization by rescuing ICAP-1-deficient osteoblast cells with a similar stable expression of WT ICAP-1, non-ubiquitylatable ICAP-1 K158R and of the chimeric ubiquitylated form. All osteoblast cell lines were able to spread onto FN and develop FAs containing β1 integrins, as revealed by 9EG7 antibody staining for activated β1 integrin (Fig. 2D). Like ICAP-1-deficient cells, cells expressing the ubiquitylated form of ICAP-1 displayed more numerous β1 integrin-containing FAs compared with cells expressing the WT form (Fig. 2D–F) because of the inability of the monoubiquitylated ICAP-1 to inhibit β1 integrin. Conversely, cells expressing the non-ubiquitylatable ICAP-1 K158R mutant displayed fewer, smaller and more-punctate adhesion sites (Fig. 2D–F) compared with those of WT ICAP-1, likely due to its ability to interact with β1 integrin and thus inhibit the assembly of larger FAs (Bouvard et al., 2007; Millon-Frémillon et al., 2008).
As Smurf1 is responsible for ICAP-1 monoubiquitylation, we investigated whether the formation of β1 integrin-containing FAs was dependent on Smurf1 activity. As expected, the deletion of Smurf1 led to a decrease in the number and area of β1 integrin-containing FA (Fig. S2B,C,D) phenocopying the non-ubiquitylatable ICAP-1 K158R phenotype (Fig. 2D–F). Conversely, the ubiquitylated ICAP-1 was able to bypass the destructive effect of Smurf1 deletion on β1 integrin-containing FAs (Fig. S2B,C,D). Thus, Smurf-1-mediated ICAP-1 monoubiquitylation plays a crucial role in the organization of β1 integrin-containing FA by preventing or disrupting the ICAP-1–β1-integrin interaction.
ICAP-1 monoubiquitylation is a signal coordinating FN density sensing with rigidity sensing
We wondered whether ICAP-1 monoubiquitylation was involved in FN density and rigidity sensing. To test an effect on FN density sensing, single-cell tracking of sparse cells was performed to monitor the migration speed of ICAP-1-deficient osteoblast or mouse embryonic fibroblast (MEF) cells expressing WT ICAP-1, K158R ICAP-1 or ICAP-1–Ubi in the presence of increasing concentrations of FN. As expected (Discher et al., 2005; Engler et al., 2006; Raab et al., 2012), WT ICAP-1-expressing osteoblasts (Fig. 3A; Movies 1,2) or MEFs (Fig. S3A) displayed faster migration rates with increasing FN density. While the migratory speed of the cells expressing the ubiquitylated ICAP-1 form depended on ECM density, like ICAP-1 null cells, the cells expressing the non-ubiquitylatable K158R mutant maintained the same migration speed whatever the density of FN coating (Fig. 3A; Fig. S3A, Movies 3,4). Moreover, the ability to adapt their migration response to ECM density was lost in cells treated with siRNA against Smurf1 but was rescued in cells co-expressing the monoubiquitylated ICAP-1 showing that the Smurf1-dependent monoubiquitylation of ICAP-1 is necessary for cells to sense and respond to FN density (Fig. S3B).
To explore the possibility that the inability of the K158R mutant to adapt its migration speed to FN density could be due to a greater capacity to lock β1 integrin in its inactivated form than with WT ICAP-1, we analyzed the response of cells treated with β1 integrin-blocking antibodies to increasing FN density. We showed that these cells were unable to sense the density of FN or adapt their migratory behavior (Fig. 3B), confirming the requirement for β1 integrin activation for the adaptation of the cell migration rate to the FN density. Additionally, cells co-expressing a β1 integrin mutant that lacks ICAP-1 binding (β1 V787T) with the ICAP-1 K158R mutant or in the context of silenced Smurf1 were still able to adapt their migration speed to the FN density (Fig. 3B; Fig. S3C). Therefore, the unresponsiveness of cells to the FN density is most likely due to the inhibitory interaction between the non-ubiquitylatable ICAP-1 and β1 integrin. Overall, ICAP-1 monoubiquitylation by Smurf1 is required to release ICAP-1 inhibitory effect on β1 integrin in order to permit the adaptation of cell migration to ECM density.
We next evaluated the effects of ICAP-1 monoubiquitylation on the ECM rigidity sensitivity. Osteoblast cells (Fig. 3C) or MEF cells (Fig. S3D) infected with ICAP-1 WT, ICAP-1 K158R and ICAP-1–Ubi were plated onto FN-conjugated elastomeric polyacrylamide (PAA) gels with increasing Young's modulus (E) and monitored for cell migration. As expected, the WT ICAP-1 cells moved more quickly on stiffer gels than they did on softer gels (40% increase on the stiffer substrate) (Fig. 3C; Fig. S3D, Movies 5,6). Cells expressing ICAP-1 K158R still responded to the increase in matrix rigidity, whereas cells expressing the monoubiquitylated ICAP-1 displayed a constant migration velocity that was independent of the stiffness of the substrate, like ICAP-1-deficient cells (Fig. 3C; Fig. S3D, Movies 7,8). However, the migration speed of ICAP-1−/− cells was slightly but significantly higher as compared to that of ICAP-1–Ubi cells. This suggests that ICAP-1 monoubiquitylation also controlled the capacity of cells to adapt their velocity to ECM rigidity. As monoubiquitylation prevents ICAP-1 and β1 integrin interaction, we then investigated whether rigidity sensing was dependent on ICAP-1 and β1 integrin interaction. Cells expressing the β1 integrin V787T mutant that are unable to interact with ICAP-1 still adapt their velocity in response to the external rigidity (Fig. 3D) whereas ICAP-1 deficiency led to insensitiveness to substrate stiffness (Fig. 3C). Thus, the presence of ICAP-1 is required even though ICAP-1 interaction with β1 integrin is dispensable for rigidity sensing. Monoubiquitylation of ICAP-1 is a signal that allows the sensing of matrix density and rigidity by decoupling the inhibitory role of ICAP-1 on β1 integrin from an unexpected role that is independent of its interaction with β1 integrin.
The monoubiquitylation of ICAP-1 increases cell contractility
As rigidity sensing is associated with cell contractility, we sought to determine whether the monoubiquitylated form of ICAP-1 might interfere with cell contractility. First, as a contractility marker, we analyzed the phosphorylation state of myosin light chain (pMLC) by western blotting lysates from WT, and ICAP-1–Ubi and ICAP-1-deficient cells plated onto FN-coated plastic or elastomeric PAA gels with a Young's modulus (E) of 4 or 50 kPa (Fig. S4A). As expected, the level of pMLC in total cell lysates of cells expressing ICAP-1 WT increased with the substrate rigidity. ICAP-1-deficient cells displayed the same behavior as ICAP-1 WT cells. In contrast, cells expressing the monoubiquitylated ICAP-1 showed a constant level of pMLC independently of the rigidity of the substrate. This loss of pMLC regulation is correlated with the inability of ICAP-1–Ubi cells to adapt their velocity to ECM rigidity (Fig. 3C). In addition, an increase of pMLC staining along the stress fibers in ICAP-1–Ubi cells was noted (Fig. 4A). To investigate whether the monoubiquitylated ICAP-1 is involved in the genesis and modulation of forces applied to the substratum, traction force microscopy (TFM) was used. Traction forces generated by the cells were twice as high in ICAP-1–Ubi cells as compared to the WT cells and ICAP-1-deficient cells (Fig. 4B). Therefore, the monoubiquitylation of ICAP-1 increases cell contractility by forcing the phosphorylation of myosin independently of the substrate rigidity.
The monoubiquitylation of ICAP-1 drives MRCKα-mediated cell contractility
Cell contractility relies on the balance between ROCK, MLCK and mDia activities to control elongation and organization of actin filament (Burridge and Wittchen, 2013). To explore the contractility pathways potentially affected by ICAP-1–Ubi, a pharmacological approach was used by testing ROCK, MLCK and mDia inhibitors (Y27632, ML7 and SmifH2, respectively) on the migration of osteoblasts adhered to 4 kPa gels coated with 5 µg/ml of FN. Like WT cells, ICAP-1–Ubi cells migrated slower upon MLCK and mDia inhibition (Fig. S4B). As previously described (Totsukawa et al., 2000), WT cells migrate faster upon ROCK inhibition. In contrast, ICAP-1–Ubi cells were insensitive to Y27632 treatment since no change in migratory speed response was observed as compared with the WT cells (Fig. S4B). This insensitivity to ROCK inhibition in ICAP-1 Ubi cells is not due to the loss of the interaction between ICAP-1–Ubi and β1 integrin since cells expressing the V787T mutant of β1 integrin, which is unable to interact with ICAP-1, are still sensitive to ROCK inhibition (Fig. S4C). Thus, ICAP-1–Ubi cell migration is independent of ROCK-controlled contractility, suggesting an alternative contractile pathway for ICAP-1–Ubi cells.
Besides regulating ROCK1 (Peter et al., 2006), ICAP-1 has been shown to inhibit Cdc42 and Rac1 (Degani et al., 2002), which are involved in the regulation of MRCK. Therefore, we sought to assess whether ICAP-1 could regulate MRCK-dependent cell contractility (Leung et al., 1998). To test this hypothesis, we used a siRNA strategy to knockdown ROCK1, ROCK2, MRCKα and MRCKβ (Fig. 4C,D). The WT ICAP-1 cells moved more quickly on stiffer gels than they did on softer gels whatever the siRNA used except in conditions of ROCK2 deletion suggesting that WT cells adapt their migratory behavior through a ROCK2-dependent contractility and this behavior is independent of ROCK1, MRCKα and MRCKβ (Fig. 4C). In contrast, only MRCKα silencing in ICAP-1–Ubi cells led to an increase in the cell migration speed when rigidity of the substrate was increased (Fig. 4D). Thus, the cell contractility mode imposed by ICAP-1 monoubiquitylation is dependent on MRCKα and is independent of ROCK1, ROCK2 and MRCKβ. To confirm the involvement of MRCKα in the monoubiquitylated ICAP-1-dependent phosphorylation of myosin, we tested the effect of siRNA against MRCKα or ROCK2 on the decoration of stress fibers by T18/S19 phosphorylated MLC (ppMLC) (Fig. 4E). Whereas the siRNA against ROCK2 decreased the level of ppMLC in WT cells, the depletion of MRCKα significantly reduced the level of ppMLC in cells infected with ICAP-1–Ubi. Thus, ICAP-1 monoubiquitylation favors the phosphorylation of myosin II that is dependent on the activity of MRCKα whereas ROCK2 activity is responsible for the phosphorylation of myosin II in WT cells. Taken together, these results show that ICAP-1 monoubiquitylation allows the switch from ROCK2-mediated to MRCKα-mediated cell contractility.
Our data show that monoubiquitylation of ICAP-1, a protein that associates with integrin cytoplasmic domains, by Smurf1 is involved in regulating the balance between adhesion and contractility. ICAP-1 monoubiquitylation inhibits its binding to β1 integrin, subsequently regulating the number and organization of β1 integrin-containing FAs. ICAP-1 and its monoubiquitylated form may be crucial mediators involved in the balance between ROCK2 and MRCKα activities in order to adapt cell contractility to the variability of ECM stiffness. Our results show that these two functions of ICAP-1 are integrated by the cell to sense both matrix density and rigidity.
Smurf1 as a node to control focal adhesion dynamics and cell contractility
In addition to its ability to ubiquitylate talin (Huang et al., 2009), Smurf1 was a good candidate for ICAP-1 monoubiquitylation because Smurf1 associates with the cerebral cavernous malformations (CCM) complex (Crose et al., 2009), which interacts with ICAP-1 (Hilder et al., 2007). Smurf1 also possesses an NPxY motif that might be able to interact with ICAP-1 phosphotyrosine-binding (PTB) domain. Smurf1 is also involved in cell polarity and cell migration (Sahai et al., 2007; Wang et al., 2003). We demonstrated that the monoubiquitylation of ICAP-1 by Smurf1 is not involved in ICAP-1 degradation via the proteasome, but rather, regulates the assembly and organization of FAs by modulating the ICAP-1–β1-integrin interaction. The ICAP-1–β1-integrin interface is likely disrupted upon ICAP-1 monoubiquitylation since K158 is in close vicinity to the I138 residue known to be important for the β1 integrin interaction (Chang et al., 2002; Liu et al., 2013).
In addition to their canonical roles in cell growth and differentiation mediated through TGF signaling (Zhu et al., 1999), accumulating evidence indicates that Smurfs play key roles in regulating cell adhesion and migration. Smurf1 is localized in lamellipodia and filopodia, with a fraction of Smurf1 in FAs (Huang et al., 2009; Wang et al., 2003). Smurf1 ubiquitylates molecules involved in both cell adhesion and contractility. Smurf1 controls talin head degradation, and subsequently adhesion stability and cell migration (Huang et al., 2009). RhoA ubiquitylation by Smurf1 causes its degradation at the leading edge of migrating cells and promotes lamellipodium formation (Sahai et al., 2007; Wang et al., 2003). Our data demonstrate that Smurf1 is a node controlling both FA dynamics and cell contractility through a common target, ICAP-1. ICAP-1 monoubiquitylation not only regulates the number and organization of β1 integrin-containing FAs but also inhibits ROCK signaling and promotes the MRCK signaling pathway. Therefore, we add another piece of evidence showing that the RhoA–ROCK pathway is inhibited by Smurf1, and we demonstrate for the first time that Smurf1 controls a switch from a ROCK-dependent to a MRCK-dependent cell contractility.
The monoubiquitylation of ICAP-1 as a switch from ROCK2-mediated to MRCKα-mediated contractility
In addition to its role in the β1 integrin activation cycle (Millon-Frémillon et al., 2008), ICAP-1 interferes with small GTPase signaling and cell contractility by putting a cap on RhoA activation (Faurobert et al., 2013) and inhibiting Rac1 and Cdc42 (Degani et al., 2002). So far, how ICAP-1 can regulate both RhoA–ROCK signaling and the Cdc42 and Rac1 pathway was unclear. It has been described that a cooperation between RhoA–ROCK and Cdc42 or Rac1–MRCK signaling can control cell contractility cell polarity, morphology and morphogenesis (Gally et al., 2009; Unbekandt and Olson, 2014; Wilkinson et al., 2005). Their respective contribution might depend on ECM rigidity. ICAP-1, independently of its interaction with β1 integrin, could act as a sensor of ECM rigidity differently modulating the activity of each enzyme depending on the substrate stiffness. It could act by playing on the level of activation of RhoA, Rac1 or Cdc42 and by directly modulating the activity of ROCK2 and MRCKα. Thus, we propose that ICAP-1 monoubiquitylation by Smurf1 is a key event leading to a switch from ROCK2-mediated to MRCKα-mediated cell contractility. ICAP-1 and its monoubiquitylated form regulate ROCK2- and MRCKα-dependent MLC phosphorylation independently of interaction with β1 integrin. This is in line with previous studies, which do not attribute a major role of β1 integrin to ECM rigidity sensing (Jiang et al., 2006). Taken together, our results show that ICAP-1 contributes to an elaborate signaling network responsible for maintaining cell tensional homeostasis, going from the dynamics of cell adhesion to the adaptation of contractile actomyosin machinery. ICAP-1 may function in β1 integrin-dependent and -independent pathways to orchestrate both the chemo and mechanical regulation of cell migration. These two pathways might regulate distinct signaling cascades through a switch operated by Smurf1 to adapt the cellular migratory response (Fig. 5). ICAP-1 is essential in rigidity sensing and its monoubiquitylation might be crucial for the adaptation of cells to a local variation of ECM stiffness in tissues or a change of ECM composition during development or in pathological situations. ICAP-1 monoubiquitylation would allow the cell to adapt its the contractility depending on substrate stiffness by controlling the balance between ROCK2-and MRCKα-mediated cell contractility. In future studies, it will be important to identify the factors that are regulated by ICAP-1 independently of its interaction with β1 integrin in order to develop a more complete understanding of the functions of ICAP-1 in mechanosensing.
MATERIALS AND METHODS
The plasmids pCMVFlag-Smurf1 WT, pGEX4T1-Smurf1 WT, pGEX4T1-Smurf2 WT, pRK5-Myc-Smurf1 and pRK5-HA-Ubiquitin-WT were obtained from Addgene (Cambridge, MA; numbers 11752, 13502, 13504, 13676 and 17608). pGEX4T1 plasmids containing the β1 or β3 integrin cytoplasmic domain, as well as pCLMFG retroviral vectors containing WT β1 integrin or the V787T mutant, have been previously described (Brunner et al., 2011). The pSG5-ubiquitin-His vector was a kind gift from Saadi Khochbin (U823 INSERM-UJF, Grenoble, France). The full-length cDNA of WT human ICAP-1 was subcloned into the EcoRI and BamHI sites of the pBabe-puro retroviral vector (pBabe-ICAP-1 WT). The K158R substitution was introduced into the ICAP-1 cDNA via site-directed mutagenesis (pBabe-ICAP-1 K158R). The Myc tag was inserted at the 3′ end of the ICAP-1 or ubiquitin cDNA using PCR. The Myc-tagged ICAP-1 cDNA was subcloned between the BamHI and EcoRI sites of the pcDNA3.1 expression vector and mutated to generate the K158R mutant. The cDNA of Myc-tagged ubiquitin was amplified and inserted at the 3′ end of the ICAP-1 cDNA, between the EcoRI and XhoI sites of the pcDNA3.1 vector (pcDNA3.1-ICAP-1-myc, pcDNA3.1-ICAP-1 K158R-myc and pcDNA3.1-ICAP-1-Ubi-myc). The ICAP-1-Ubi-myc cDNA was subcloned into the pBabe-puro, between the BamHI and SalI sites (pBabe-ICAP-1-Ubi-myc).
Cell culture, transfection and antibodies
Immortalized osteoblasts were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Life Technologies, Cergy Pontoise, France), CHO cells and HeLa cells were grown in αMEM (PAA) at 37°C in a humidified, 5% CO2 chamber. All media are supplemented with 10% fetal calf serum (FCS; Invitrogen), 100 U/ml penicillin and 100 µg/ml streptomycin. Immortalized osteoblasts from icap-1−/−; Itgb1flox/flox mice were generated as previously described (Bouvard et al., 2007). These cells were treated with or without adenoCre viruses obtained from the gene transfer vector core (University of Iowa) to generate β1 integrin-null cells. The ICAP-1-null cells were incubated with or without retroviral particles to obtain rescued cells expressing ICAP-1 WT, ICAP-1 K158R or the ICAP-1–Ubi chimera. The cells were selected with 1 mg/ml puromycin to produce cell populations with heterogeneous ICAP-1 expression levels. β1 integrin-null cells that had already been rescued with ICAP-1 were again infected with retrovirus to obtain double-rescued cells expressing ICAP-1 (WT or mutant) and WT β1 integrin or the V787T mutant. For all experiments, cells were trypsinized and washed in PBS before plating in DMEM containing 4% FN-free FCS for 3 h. Osteoblasts (90×104 cells) were transfected with 25 pmol siRNA and 6 µl Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer's instructions. The cells were used 2 days after transfection. SMARTpool siRNA (Dharmacon Research Inc., Lafayette, LA) was used against appropriate proteins, along with the control siRNA sequence 5′-AGGUAGUGUAAUCGCCUUG-3′. HeLa cells were transfected with control or Smurf1 siRNA SMARTpool siRNA (Dharmacon Research Inc.) using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer's instructions; two rounds of transfection were performed. ICAP-1 and His-tagged ubiquitin were overexpressed using Fugene (BD Biosciences, Le Pont de Claix, France) according to the manufacturer's instructions. CHO cells were transfected with ExGen (EUROMEDEX, Souffelweyersheim, France) following the manufacturer's instructions using pcDNA3.1-ICAP-1-myc, pcDNA3.1-ICAP-1K158R-myc or pcDNA3.1-ICAP-1-Ubi-myc. CHO cells were cotransfected with pcDNA3.1-ICAP-1-myc or pcDNA3.1-ICAP-1 K158R-myc and pSG5-ubiquitin-His. After 24 h, the transfected cells were incubated with the proteasome inhibitor MG132 (20 µM) for 4 h. The antibodies used in this study were the following: rat anti-β1 integrin 9EG7 (1:100; BD Biosciences, 553715), donkey anti-rabbit-IgG conjugated to HRP (1:12,000; Jackson ImmunoResearch, UK, 711-036-152), goat anti-rat-IgG conjugated to Alexa Fluor 488 (1:1000; Invitrogen, A-11006), mouse anti-actin (1:1000; Sigma-Aldrich, Saint Quentin Fallavier, France, A2066), mouse anti-Smurf1 (1:1000; Santa Cruz Biotechnology, Heidelberg, Germany, Sc-100616) rabbit anti-T18/S19 MLC [1:1000 (western blotting) or 1:100 (immunofluoresence); Cell Signaling Technology, Leiden, The Netherlands], and rabbit anti-ICAP-1 (1:1000; Millon-Frémillon et al., 2008).
Purification of His-tagged ubiquitylated proteins
Transfected CHO cells were lysed in phosphate-buffered saline (PBS) containing 10% glycerol, 0.3% NP40, 5 mM NEM, 10 mM NaF, phosphatase inhibitor cocktails 2 and 3 (Sigma-Aldrich), and a protease inhibitor cocktail (cOmplete, EDTA-free, Roche, Meylan, France). After centrifugation (15,000 g for 20 min), the supernatants were incubated with Talon Metal Affinity resin (Clontech, Saint Germain en Laye, France) for 2 h. After three washes, the proteins were eluted in Laemmli buffer and analyzed by western blotting (3% of the total lysate is used for the input track).
GST–Smurf1 and GST–Smurf2 were expressed in E. coli (BL21 DE3 RIL) as previously described (Wang et al., 2006). Transfected CHO cells were lysed in buffer containing 50 mM Tris-HCl pH 7.7, 150 MG132, protease inhibitor cocktail and phosphatase inhibitor cocktails 2 and 3. The supernatants were incubated for 3 h with GST–Smurf1-, GST–Smurf2- or GST-coupled glutathione–Sepharose beads. After five washes in lysis buffer, the samples were eluted in Laemmli buffer and analyzed by western blotting (3% of the total lysate is used for the input track). GST–β1-integrin and GST–β3-integrin were expressed in E. coli (BL21 DE3 RIL), and pulldown experiments with supernatants from transfected CHO cells were performed as previously described (Brunner et al., 2011).
ICAP-1 protein lifetime measurement
Transfected CHO cells were incubated with 100 µg/ml cycloheximide (Sigma-Aldrich) with or without 20 µM MG132. Cells were lysed in RIPA buffer at the indicated times, and the protein concentration was measured using the BCA assay. Total proteins (20 µg) were separated by SDS-PAGE and immunoblotted as below.
Transfected CHO cells were lysed in lysis buffer (50 mM NaCl, 10 mM Pipes, 150 mM sucrose, 50 mM NaF, 40 mM Na4P2O7·10H2O, 1 mM Na3VO4, pH 6.8, 0.5% Triton X-100, 0.1% sodium deoxycholate, and protease inhibitor cocktail). The supernatants were incubated for 1 h with anti-Flag M2 magnetic beads (Sigma-Aldrich). After four washes with lysis buffer, the samples were eluted in lysis buffer containing 100 µg/ml Flag peptide (Sigma-Aldrich) and analyzed by western blotting (3% of the total lysate is used for the input track).
The interaction between recombinant ICAP-1 and ICAP-1–Ubi was analyzed using a solid-phase assay. Briefly, a 96-well tray (MaxiSorp, Nunc) was coated with either ICAP-1-His or ICAP-1–Ubi–His (40 µg/ml) for 16 h at 4°C and blocked with 3% BSA in PBS for 1 h at room temperature. Increasing concentrations of GST, the GST–β1-integrin cytoplasmic domain or GST–Smurf1 were added for 1 h. After three washes in PBS with 0.1% Tween20, detection of bound proteins was performed by using the antibodies directed against β1 integrin cytoplasmic domain or Smurf1. Nonspecific binding to BSA-coated wells was subtracted from the results as background.
pMLC western blot analysis
Cells were plated on plastic or on PAA gels with controlled rigidities of 50 kPa or 4 kPa (Cell Guidance System, Cambridge, UK) coated with 1 µg/cm2 (5 µg/ml) of FN. The next day cells were lysed in Laemmli buffer and analyzed by western blotting. Immunoblots were visualized using the ECL system (Biorad) and Chemidoc imaging system (Biorad).
Traction force microscopy
The PAA substrates were prepared on two-well LabTek slides (Thermo Fischer Scientific, Ulm, Germany) using 8% acrylamide mixed with appropriate percentage of bis-acrylamide and 10 mM HEPES (pH 8.5) gels. After two Sulfo-SANPAH (Thermo Fischer Scientific, Ulm, Germany) activations, the gels were coated with 5 µg/ml FN (1 µg/cm²) at 4°C overnight. We used a concentration of 0.15% of bis-acrylamide to create gels with controlled rigidities of 5 kPa. Cells were plated at an approximate density of 2×104 cells per cm2 for 3-4 h and images were acquired on an iMIC Andromeda (FEI, Gräfelfing, Germany) microscope at 40x magnification. Force calculations were performed as previously described (Tseng et al., 2011).
Random migration analysis
Cells were plated on a 12-well plate containing a PAA substrate (Cell Guidance System) or on an 8-well LabTek slide coated with various FN concentrations at an approximate density of 1.2×105 per cm2 for 3 h in CO2-independent DMEM containing 4% FN-free FCS. The cells were maintained at 37°C and imaged on an inverted microscope (Zeiss Axiovert 200) equipped with a motorized stage, cooled CCD camera (CoolSnap HQ2, Roper Scientific) and a 10× objective (EC Plan-Neofluar) for live-cell imaging for 5 h at a frequency of 1 image every 4 min. Inhibitors were added as indicated to the medium 10 min prior to the initiation of image acquisition and maintained throughout the migration assay at a final concentration of 10 µM for Y27632 (Calbiochem), 5 µM for ML7 (Calbiochem) and 2 µM for SmifH2 (Calbiochem). Cell velocity was obtained using the manual tracking plug-in in ImageJ software. A total of 150–300 cells were analyzed from at least five different locations in each experiment, and results were collected from three independent experiments.
Cells were plated at an approximate density of 2×104 cells per cm2 for 2.5 h in 24-well plates on slides coated with 0.6 µg/cm2 (2 µg/ml) or 1.5 µg/cm2 (5 µg/ml) of FN in DMEM containing 5% FN-depleted serum; the cells were then fixed and immunostained as previously described (Millon-Fremillon et al., 2008). For the focal adhesion analysis, images were acquired on an Axio Imager (Zeiss) microscope at with a 63× objective. We analyzed the β1 integrin staining of 30–40 cells from two independent experiments using a thresholding method and the particle analyzer in ImageJ. Particles larger than 0.5 µm2 were analyzed. Internal focal adhesions are defined as a FA that was more than 3 µm distal to the plasma membrane. For the ppMLC-decorated stress fibers, images were acquired on an iMIC Andromeda (FEI) microscope at with a 40× objective. We analyzed the phosphorylation of Thr18 and/or Ser19 on the light myosin chain in 90–100 cells from three independent experiments by using the ‘Unsharp mask’ and the particle analyzer plug-in in ImageJ software. Objects bigger than 0.5 µm2 were analyzed.
All data sets were analyzed with R (http://www.R-project.org/). We used an ANOVA-2 analysis and Tukey's HSD post-hoc test when necessary. Results are mean±s.e.m. Significance is indicated with asterisks (*P<0.05, **P<0.005, ***P<0.0005).
We thank Daniel Bouvard for osteoblast production, the Institut Albert Bonniot Cell Imaging facility, and Jacques Mazzega and Alexei Grichine for their assistance with the microscopy studies. We are indebted to O. Destaing and Richard Demetz for scientific discussions, Rachel Ramchurn for manuscript reading, to Saadi Khochbin for biological tools.
C.A.-R., A.-P.B., and E.P. designed and analyzed the experiments. A.-P.B., M.R.-K., A.-S.R., A.K., E.F., H.-N.F., I.B.-R., S.M.-D., C.O. and M.B. helped with the experimental design and the procedures, performed the experiments, and analyzed the data. C.A.-R., A.P.-B. and E.P. wrote the manuscript. All of the authors provided detailed comments. C.A.-R. initiated the project.
This work was funded by Fondation ARC pour la Recherche sur le Cancer and Institut National Du Cancer (INCA). The C.A.-R. team is supported by la Ligue Contre le Cancer (Equipe labellisée Ligue 2014). M.R.-K., A.K., and H.-N.F. were supported by Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche fellowships.
The authors declare no competing or financial interests.