Summary
The microtubule (MT) cytoskeleton is essential for many cellular processes, including cell polarity and migration. Cortical platforms, formed by a subset of MT plus-end-tracking proteins, such as CLASP2, and non-MT binding proteins such as LL5β, attach distal ends of MTs to the cell cortex. However, the mechanisms involved in organizing these platforms have not yet been described in detail. Here we show that 4.1R, a FERM-domain-containing protein, interacts and colocalizes with cortical CLASP2 and is required for the correct number and dynamics of CLASP2 cortical platforms. Protein 4.1R also controls binding of CLASP2 to MTs at the cell edge by locally altering GSK3 activity. Furthermore, in 4.1R-knockdown cells MT plus-ends were maintained for longer in the vicinity of cell edges, but instead of being tethered to the cell cortex, MTs continued to grow, bending at cell margins and losing their radial distribution. Our results suggest a previously unidentified role for the scaffolding protein 4.1R in locally controlling CLASP2 behavior, CLASP2 cortical platform turnover and GSK3 activity, enabling correct MT organization and dynamics essential for cell polarity.
Introduction
The microtubule (MT) cytoskeleton is essential for many cellular functions including cell polarity and migration. MTs are polarized filaments that switch between phases of polymerization and depolymerization (Howard and Hyman, 2003), a behavior known as dynamic instability (Mitchison and Kirschner, 1984). A diverse group of proteins called MT plus-end-tracking proteins (+TIPs) bind to growing MT plus-ends and regulate MT dynamics and interactions of MTs with other structures (Galjart, 2010). Some +TIPs, including cytoplasmic linker-associated protein-2 (CLASP2), are also part of ‘cortical’ or ‘peripheral’ platforms, which are located near the plasma membrane and, in motile cells, are strongly enriched at the leading edge. A major function of these cortical platforms, which also contain non-MT-binding proteins such as LL5β and ELKS (also known as ERC1), is to attach MTs to the cell cortex (Lansbergen et al., 2006). CLASP2 acts as a bridging protein between MT distal ends and cortical platforms, ensuring MT capture and attachment, and thereby regulating asymmetric MT organization, cell polarity and migration (Akhmanova et al., 2001; Lansbergen et al., 2006; Mimori-Kiyosue et al., 2005). A major aim in this field of research is to determine the molecular mechanisms involved in organizing cortical platforms.
Protein 4.1R is the founding member of the large band 4.1 superfamily, all of whose members contain a highly conserved region known as the ‘FERM domain’ (Chishti et al., 1998) that binds membrane proteins and lipids (Bennett and Baines, 2001). Protein 4.1R was originally identified as an 80-kDa multifunctional protein of the membrane skeleton of human red blood cells. In these cells, protein 4.1R stabilizes the spectrin–actin network and mediates the attachment of the underlying cytoskeleton to the overlying lipid bilayer through interactions with integral membrane proteins and lipids (Conboy, 1993). It was subsequently discovered, however, that nucleated cells express multiple isoforms of protein 4.1R (Anderson et al., 1988) and that the complex expression pattern of 4.1R in these cells is mainly due to the extensive alternative splicing of the 4.1R-encoding pre-mRNA (Conboy, 1999; Tang et al., 1990). Subsequent studies showed that, in nucleated cells, protein 4.1R isoforms localize at multiple subcellular sites including the nucleus (de Cárcer et al., 1995), the centrosome (Krauss et al., 1997; Pérez-Ferreiro et al., 2004), MTs (Huang et al., 2004; Pérez-Ferreiro et al., 2001) and the endoplasmic reticulum (Luque et al., 1999). A role for 4.1R in organizing nuclear architecture was established from studies showing interactions between 4.1R and components of the splicing machinery (Lallena et al., 1998). Further analyses determined that 4.1R is essential for proper nuclear assembly (Krauss et al., 2002) and centrosome–nucleus association (Meyer et al., 2011). In addition, a role for 4.1R as a linker between the actin cytoskeleton and components of tight junctions (Mattagajasingh et al., 2000) and adherens junctions (Yang et al., 2009) has been reported. Recently, we showed that 4.1R selectively accumulates at the leading edge of migrating cells and that its depletion affects cell motility (Ruiz-Sáenz et al., 2011). All these findings indicate that 4.1R is not only an adaptor protein linking transmembrane proteins and the actin cytoskeleton but also a multifunctional protein acting at different subcellular sites.
In this study, we show that protein 4.1R associates with CLASP2 independently of MTs, and that protein 4.1R is required for the correct localization and dynamics of CLASP2 in cortical platforms as well as for the organization, dynamics and attachment of MTs to the cell cortex. Our data indicate that protein 4.1R controls CLASP2 binding to MT plus-ends by locally affecting GSK3 activity. As a result, upon reduction of 4.1R expression, cells lose polarity, the association of CLASP2 with the MT lattice increases, and growing MTs that approach the cell periphery are not captured properly, but continue to grow, bending and curling at cell margins and losing their radial distribution. Our results suggest a key role of the scaffolding protein 4.1R in establishing MT network asymmetry at the leading edge, thereby ensuring the correct MT organization and dynamics essential for cell polarity.
Results
Protein 4.1R is involved in MT organization
Migrating cells are characterized by the repositioning of the centrosome and of a large population of MTs toward the leading edge (Gundersen and Bulinski, 1988; Schmoranzer et al., 2009). We have recently shown that protein 4.1R selectively accumulates at the leading edge and is necessary for cell migration (Ruiz-Sáenz et al., 2011). It is therefore plausible that 4.1R has a role in MT organization at the cell edge. To explore this possibility, we adopted a loss-of-function strategy in human epithelial ECV304 cells using small interfering RNAs (siRNAs) specific to 4.1R (Fig. 1A), either used individually (siRNA1, siRNA2) or as a pool (siRNA_pool, or si4.1R). The same siRNAs were previously employed to characterize the role of 4.1R in cell migration (Ruiz-Sáenz et al., 2011). Immunofluorescence analysis of control migratory cells showed that MTs formed a dense radial array oriented roughly perpendicular to the cell edge (Fig. 1B). Unlike control cells, 4.1R-knockdown (KD) cells exhibited an abnormal MT organization with many MTs oriented more parallel to the membrane, often deviating less than 25° from the cell edge (Fig. 1C,D).
We used CLIP-170, the prototype +TIP (Perez et al., 1999), to mark growing MT plus-ends in control and 4.1R-KD cells (Fig. 1E,F). CLIP-170 was associated with MT plus-ends in the margin of the 4.1R-KD cells, indicating that many of the disoriented MTs were growing. MT disorganization upon 4.1R KD was also associated with an altered orientation of the centrosome in wound-healing assays (Fig. 1G–I). Collectively, these results suggest a role for 4.1R in MT organization and cell polarization.
Protein 4.1R is involved in the regulation of MT dynamic instability
Next, we examined the effect of 4.1R silencing on MT behavior near the cell edge. Time-lapse videomicroscopic analysis using the +TIP EB3 fused to mCherry indicated that the MT growth rate was higher in 4.1R-KD cells than in control cells (Fig. 2A–C). To investigate what aspects of MT behavior were affected upon 4.1R depletion, we also analyzed MT dynamics in ECV304 cells that stably expressed mCherry-α-tubulin (Fig. 2D; Table 1; supplementary material Movie 1). Dynamic instability is known to be more prevalent at the cell periphery, where it acts to preserve the radial orientation of MTs (Komarova et al., 2002). Control cells exhibited this typical MT behavior with many MTs growing perpendicularly toward the cell periphery (Fig. 2D, upper panels) and, upon reaching the cell margin, undergoing episodes of pausing (typically lasting 10–20 seconds), followed by bouts of rapid shortening, and regrowth. In contrast, in 4.1R-KD cells, MTs in the vicinity of cell margins exhibited up to 4.5-fold fewer episodes of catastrophe (catastrophe rate is given as the number of transitions per second from a growth or a pause, to a shrinking state), 2.4-fold longer periods of growth and less time pausing and shortening (Table 1). Importantly, instead of tethering transiently at the cell edge, these MTs continued to grow, bending and curling at margins (Fig. 2D, lower panels). As a result, dynamic MT ends were maintained for longer in the vicinity of the cell margin in 4.1R-KD cells than in control cells.
MT dynamic instability was analyzed in control and 4.1R-KD cells stably expressing mCherry–α-tubulin. MT life history plots are represented as the percentage of time spent in growth, shrinkage or pause. Catastrophe rate is given as the number of transitions per second from a growth or a pause, to a shrinking state. Rescue rate corresponds to the frequency per second at which MTs transitioned from a shrinking state to a growth state. Data were pooled from three independent experiments; means ± s.d. *P<0.05; **P<0.01; ns, non-significant.
Protein 4.1R interacts with CLASP2
To explain how 4.1R might affect MT architecture and dynamics, we screened for 4.1R interaction partners. We performed 4.1R pull-down assays using cells co-expressing 4.1R tagged with a biotinylation sequence and GFP (bio-GFP–4.1R) and the Escherichia coli biotin ligase BirA (Driegen et al., 2005). Biotinylated 4.1R and interacting proteins were isolated from cell lysates with magnetic beads coupled to streptavidin. Beads were washed, bound proteins were boiled off the beads and separated by SDS-PAGE, and 4.1R-interacting proteins were subsequently identified by mass spectrometry (data not shown). The experimental set-up was validated by identifying known binding partners of 4.1R such as NuMA (Mattagajasingh et al., 1999), U2AF35 (Lallena et al., 1998) and spectrin (Ungewickell et al., 1979). Of the new proteins identified by this analysis, we focused on CLASP2, because it selectively regulates MT dynamics at the cell edge (Akhmanova et al., 2001; Drabek et al., 2006; Mimori-Kiyosue et al., 2005), and is a component of cortical platforms (Lansbergen et al., 2006).
The interaction between 4.1R and CLASP2 was verified by co-immunoprecipitation assays in cells expressing bio-GFP–4.1R (Fig. 3A), or GFP–CLASP2 (Fig. 3C, left panels) and in untransfected cells (Fig. 3B). As described before, ECV304 cells contain several 4.1R isoforms (Ruiz-Sáenz et al., 2011), which are mainly produced by alternative splicing (Conboy, 1999; Tang et al., 1990). Although several 4.1R isoforms were found in the cell extract, only the ∼80-kDa 4.1R species was detected in the GFP–CLASP2 immunoprecipitates (Fig. 3C) even in cells treated with high concentrations of nocodazole, which induces MT depolymerization (Fig. 3D). The ∼80-kDa 4.1R species was not detected in immunoprecipitates from control cells expressing GFP (Fig. 3C, right panels). Our results indicate that the 4.1R-CLASP2 association occurs independently of polymerized MTs.
To identify the 4.1R region involved in CLASP2 association, we performed pull-down experiments using GST fusions of the full-length 80-kDa isoform of 4.1R and fragments of this isoform (referred to as FERM, CORE and Cter; Fig. 3E). Another isoform of 4.1R, which contains a truncated N-terminal FERM domain and is called 4.1R60, was also included in the pull-down assays (Fig. 3E–H). Endogenous CLASP2 bound to the FERM and the CORE regions, but not to the Cter fragment of protein 4.1R (Fig. 3G,H). Neither EB1, the core +TIP, nor the dynein intermediate chain (DIC) bound to any of the 4.1R fragments used (Fig. 3G). By contrast, ELKS, a protein that associates with CLASP2 at peripheral platforms (Lansbergen et al., 2006), also bound to 4.1R through the CORE region (data not shown). We conclude that 4.1R associates with CLASP2 through its FERM and CORE domains, probably in the region common to both. Protein 4.1R60 seems to bind more weakly to CLASP2 than the full-length 80-kDa isoform of 4.1R or the CORE region. One possible explanation for this is that 4.1R proteins can adopt different conformations (Pérez-Ferreiro et al., 2006) and that, in the case of 4.1R60, a region that is important for the interaction may be hindered. The CORE region of 4.1R is also involved in binding to the scaffolding protein IQGAP1 (Ruiz-Sáenz et al., 2011), which is also a CLASP2-binding partner (Watanabe et al., 2009). Because the 4.1R–CLASP2 association can occur in IQGAP1-KD cells (supplementary material Fig. S1), IQGAP1 is not essential for the 4.1R–CLASP2 interaction.
To examine a potential overlap in intracellular distribution of 4.1R and CLASP2 we performed immunofluorescence analysis in ECV304 cells. We observed that 4.1R distribution at the cell periphery resembled that of CLASP2 (Fig. 3I,J). The cortical distribution of 4.1R and CLASP2 was further examined in nocodazole-treated cells as these maintain peripheral LL5β-positive platforms in the absence of MTs (Lansbergen et al., 2006). 4.1R and CLASP2 accumulations were observed at sites with LL5β accumulation (Fig. 3K,L). By contrast, protein 4.1R did not accumulate at focal adhesions as shown by double staining of 4.1R and vinculin (Fig. 3M; supplementary material Fig. S2). Taken together our data suggest that protein 4.1R colocalizes with the CLASP2 population present at cortical platforms. Thus, in addition to CLASP2, LL5β and ELKS which are known to be important players in cortical platforms (Drabek et al., 2006; Lansbergen et al., 2006), these specialized cortical structures also contain 4.1R. Specifically, the ∼80-kDa isoform of 4.1R is involved in the association with CLASP2 through the C-terminal half of the FERM domain, suggesting that it is this isoform that forms part of the cortical platforms.
4.1R depletion increases CLASP2 binding to MTs at the cell cortex
We next examined the distribution of endogenous CLASP2 at MT distal ends in control and 4.1R-KD ECV304 cells. Immunofluorescence analysis of cells transfected with siRNAs specific for 4.1R showed CLASP2 staining associated with MT plus-ends, as in control cells. Strikingly, unlike control cells, 4.1R-KD cells also showed a much more extended CLASP2 localization along MTs at the cell edge (Fig. 4A). Ectopically expressed GFP–CLASP2 is a useful tool for studying the dynamic behavior of CLASP2 in living cells (Drabek et al., 2006). We first examined GFP–CLASP2 behavior at MT distal ends in control and 4.1R-KD cells. In control cells, GFP–CLASP2 accumulated at MT plus-ends in the cell body and at the cell margin (Fig. 4B; supplementary material Movie 2). Indeed, in all regions of control cells, the maximal association of CLASP2 with MT plus-ends occurred proximally to the very tip and, on average, decayed exponentially with increasing distance and a half length of ∼1 µm (Fig. 4C). However, at the cell edge of 4.1R-KD cells, GFP–CLASP2 not only tracked MT plus-ends, but also remained associated for several micrometers along the MT lattice (Fig. 4B,C). By contrast, the distribution of GFP–CLASP2 at MT plus-ends in the cell body was similar to that in control cells (Fig. 4B,C; supplementary material Movie 2). The role of 4.1R was specific to CLASP2, because 4.1R KD did not change the distribution of EB3–mCherry (Fig. 4D,E; supplementary material Movie 3). Thus, our results indicate that 4.1R depletion induces an increase of CLASP2 binding to the MT lattice at the cell cortex.
Binding of CLASP2 to MT distal ends is regulated by glycogen synthase kinase-3β (GSK3β) (Akhmanova et al., 2001). Inhibition of GSK3β through phosphorylation on serine 9 induces CLASP2 association with MT plus-ends and the MT lattice (Kumar et al., 2009; Wittmann and Waterman-Storer, 2005). Because we observed increased MT lattice binding of CLASP2 in 4.1R-KD cells (Fig. 4C; see also Fig. 5G), we decided to analyze the levels of GSK3 serine phosphorylation after 4.1R KD. Immunoblot analysis showed higher levels of phosphorylation of GSK3α and GSK3β [on Ser residues 21 (pS21) and 9 (pS9), respectively] relative to control cells (Fig. 5A–C), indicating that 4.1R depletion results in the inactivation of these kinases. Interestingly, overexpression of constitutively active GSK3β (S9A) in 4.1R-KD cells resulted in a reduced association of GFP–CLASP2 with MT plus-ends and lattice, even in comparison to non-treated ECV304 cells (Fig. 5F,G). Taken together our data suggest that GSK3 inhibition underlies the increased binding of CLASP2 to MT plus-ends and lattice at the cell cortex of 4.1R KD cells (Fig. 5D–G). Thus, our results suggest that 4.1R controls CLASP2 association with the MT lattice at the cell periphery through the local regulation of GSK3β activity.
Protein 4.1R is required for CLASP2 localization and dynamics at cortical platforms
We next investigated whether 4.1R is involved in the correct localization of CLASP2 at cortical platforms. Indeed, 4.1R KD led to a halving of the number of CLASP2 platforms per square micrometer, as measured both for exogenous GFP–CLASP2 and for endogenous CLASP2 (Fig. 6A–D). Cortical CLASP2 is involved in the attachment of distal MT ends to the cell cortex, so a decrease in the number of CLASP2 platforms might contribute to the altered MT architecture found at the edges of 4.1R-KD cells (Fig. 1). Indeed, upon CLASP depletion in ECV304 cells (Fig. 6E,F), MTs displayed a non-radial array with a reduced angle between the MTs and the cell margin compared with control cells (Fig. 6E,G). Thus, the depletion of either 4.1R or CLASP results in the loss of MT radial orientation, albeit to different extents. The CLASP-dependent phenotype of MT orientation along the cell margin was previously described in HeLa cells (Mimori-Kiyosue et al., 2005). An additional effect in these cells was the loss of MT density (Mimori-Kiyosue et al., 2005) which we also observed in CLASP-KD ECV304 cells (data not shown).
To determine whether 4.1R silencing also affects the dynamic behavior of CLASP2 in cortical platforms, we performed fluorescence recovery after photobleaching (FRAP) experiments in control and 4.1R-KD cells transiently expressing GFP–CLASP2 (Fig. 6H; supplementary material Movie 4). We bleached GFP–CLASP2 in peripheral platforms at cell edges (Fig. 6H). Consistent with FRAP results previously obtained in HeLa cells (Drabek et al., 2006), we observed a relatively immobile GFP–CLASP2 fraction (∼27%) near the cell edge in control cells (Fig. 6I). Recovery of the mobile fraction (∼73%) could be fitted using a single-exponential model (Fig. 6I). Remarkably, in 4.1R-KD cells the immobile GFP–CLASP2 fraction was significantly reduced (to ∼8%) and the recovery of the mobile fraction (∼92%) could be best fitted using a double-exponential model (Fig. 6J; supplementary material Fig. S3). Data point distribution in the residual plots of control and 4.1R-KD cells agrees with the exponential models chosen (Fig. 6I,J; upper panels). Single FRAP profiles from independent experiments are also shown (supplementary material Fig. S3). In 4.1R-KD cells, the fluorescence recovery of GFP–CLASP2 in cortical platforms suggests the presence of two different binding states with distinctive binding affinities (fast and slow components; Fig. 6J). The rate of recovery of the slow component (0.42±0.24/second) was not significantly different from the rate of recovery of the mobile GFP–CLASP2 fraction in control cells (0.70±0.05/second; P = 0.118). However, the rate of recovery of the fast component (4.95±1.25/second), which accounts for ∼50% of all molecules, was seven times faster than that of control cells (P<0.005). Because GFP–CLASP2 accumulates at dynamic MT ends as well as in cortical platforms (i.e. dots), and these structures often contact each other, it is possible that in some single frames of movies a dot resembles a plus end. To unequivocally show that dynamic MT ends contact and pass cortical platforms, we imaged, using TIRF microscopy, cells co-expressing EB3–mCherry and GFP–CLASP2 and observed that dynamic comets often contacted and passed cortical platforms (supplementary material Fig. S4; Movie 5). In addition, cortical platforms that often appeared as regularly shaped dots were also observed after nocodazole treatment (supplementary material Movie 6). These data indicate that the dots do not represent MT tips of vertically oriented MTs in the TIRF plane. Cortical platforms with dotted appearance have also been reported in other cell types (Drabek et al., 2006; Hotta et al., 2010).
In conclusion, the experiments shown in Fig. 6H–J suggest that 4.1R depletion results in an increase of the fraction of mobile GFP–CLASP2 molecules in cortical platforms, and in a more dynamic exchange of a significant proportion of GFP–CLASP2 molecules.
Discussion
MT organization and dynamics are essential for the establishment and maintenance of cell polarity. In polarized cells, MT asymmetry depends on several different mechanisms such as the repositioning of the centrosome and the localization of specific proteins that attach MT arrays in limited areas of the cell cortex (Sugioka and Sawa, 2012). CLASPs are +TIPs that act as local regulators of MT dynamic instability at the cell periphery (Akhmanova et al., 2001; Mimori-Kiyosue et al., 2005). CLASPs are also part of cortical platforms, which are protein assemblies at the cell cortex that also contain non-MT-binding proteins such as LL5β and ELKS (Lansbergen et al., 2006) and that serve to attach MTs to the cell cortex. A major challenge is to elucidate the mechanisms, and identify all the proteins, controlling the organization of cortical platforms. The present study shows that protein 4.1R interacts with CLASP2 independently of polymerized MTs, and colocalizes with CLASP2 at cortical platforms at the cell edge. In agreement with this, protein 4.1R is required for the correct localization and dynamic behavior of CLASP2 in cortical platforms as well as for the organization, dynamics and attachment of MTs to the cell cortex. We also show that protein 4.1R is required for centrosome repositioning in migrating cells. Hence, 4.1R emerges as a key regulator of asymmetrical MT organization in polarized cells.
4.1R is a multifunctional protein that plays structural roles, organizing membrane protein domains and/or linking membranes to internal actin cytoskeletal and nucleoskeletal networks (Hung et al., 2000; Kontrogianni-Konstantopoulos et al., 2000; Lallena et al., 1998; Mattagajasingh et al., 2000; Nunomura et al., 1997). Our data reveal that downregulation of 4.1R also leads to marked disorganization of the microtubular architecture and altered MT dynamics. The increase in MT growth speed at the cell cortex observed in 4.1R-KD cells could be caused by a higher tubulin/MT ratio in these cells. Moreover, MTs growing close to the cortex in 4.1R-KD cells undergo fewer episodes of catastrophe, longer periods of growth and less time pausing and shortening, compared with control MTs. Although a defective MT nucleation from the centrosome could account for some of these effects, no reduction in MT nucleation has been observed in 4.1R-KD cells (Krauss et al., 2008). We therefore assume that the effects on the dynamic behavior of the MT network at the cell edge that we describe here are specifically due to a lack of 4.1R at the cell cortex. Several studies showed that +TIP proteins regulate MT dynamic instability (Galjart, 2010). Specifically, CLASPs contribute to MT stabilization and rescue and also decrease MT catastrophe frequency (Al-Bassam et al., 2010; Mimori-Kiyosue et al., 2005). We find that 4.1R depletion induces binding of CLASP2 to MT lattices at the cell edge. This could contribute to an increase in MT growth persistence and a decrease in catastrophe frequency.
How does 4.1R regulate CLASP2 interaction with MTs? As with many other MT-associated proteins, binding of CLASP2 to MTs is negatively regulated by phosphorylation (Wittmann and Waterman-Storer, 2005) (Kumar et al., 2009). CLASP2 is directly phosphorylated by GSK3β at several sites in the domain required for MT plus-end tracking (Kumar et al., 2009) and CLASP2 hyperphosphorylation by GSK3β impairs its interaction with MTs (Akhmanova et al., 2001; Kumar et al., 2009). GSK3 is a constitutively active kinase that is inactivated by serine phosphorylation by various kinases in numerous signaling cascades (Etienne-Manneville and Hall, 2003; Yoshimura et al., 2006). Extensive binding of CLASP2 to the MT lattice through GSK3β inactivation has been reported at the leading edge of migrating Ptk1 epithelial cells (Wittmann and Waterman-Storer, 2005). We observed that 4.1R depletion leads to an increase in GSK3 serine phosphorylation, and hence a low level of GSK3 kinase activity, concomitant with CLASP2–MT lattice binding at the cell cortex. Consistent with a role of GSK3 inactivation at the cell edge in 4.1R-KD cells, expression of constitutively active GSK3β abolished this CLASP2–MT lattice binding. Our results therefore suggest that 4.1R controls CLASP2 association with the MT lattice at the cell periphery through the local regulation of GSK3 activity.
At the periphery of control cells, the majority of CLASP2 molecules (∼70%) are mobile, whereas a fraction of CLASP2 molecules (∼30%) is immobile. After 4.1R KD two (approximately equally large) mobile CLASP2 populations were distinguished. We speculate that 4.1R depletion renders all cortical CLASP2 molecules more mobile, i.e. the immobile CLASP2 fraction that is present in control cells becomes mobile after 4.1R depletion, whereas the already mobile CLASP2 fraction becomes even more mobile after 4.1R depletion. Our observations suggest that 4.1R is essential for maintaining the number of CLASP2-positive platforms and for keeping CLASP2 inside cortical platforms. Thus, 4.1R appears to be involved in the regulation of MT dynamics at the cell edge in two distinct manners: by controlling cortical platform turnover and CLASP2 dynamics within cortical platforms, and by reducing GSK3 activity locally.
Because GSK3 also regulates other MT-associated proteins and +TIPs, an altered activity of these proteins, in addition to CLASP2, might contribute to the persistent MT growth that is observed at the edge of 4.1R-KD cells.
We recently showed that 4.1R binds and recruits IQGAP1, a MT–microfilament scaffolding protein, to the leading edge (Ruiz-Sáenz et al., 2011). It is worth noting that, although IQGAP1 and CLASP2 interact (Watanabe et al., 2009), we have shown in this study that IQGAP1 is not essential for the association between 4.1R and CLASP2. Therefore, 4.1R may form different complexes with IQGAP1 and CLASP2. Phosphorylation of CLASP2 by GSK3β regulates its interaction with IQGAP1 (Watanabe et al., 2009). Given that 4.1R is required for proper GSK3 activity at the cell edge, 4.1R might indirectly modulate the formation of CLASP2–IQGAP1 complexes at this cell site.
Our results can be summarized in a model of 4.1R function at the cell cortex (Fig. 7). At the periphery of normal cells, 4.1R interacts with CLASP2 and other cortical proteins, thereby contributing to the correct localization of CLASP2 at cortical platforms and to the maintenance of the number and complexity of these peripheral platforms. 4.1R may also restrict GSK3 inactivation to these regions. When growing MTs reach these platforms, they become anchored to the cell edge by virtue of CLASP2 tethering: CLASP2 can bind to the incoming MT lattice and to cortical proteins such as LL5β, ELKS and IQGAP1. The 4.1R–CLASP2 complex could be involved in MT alignment and anchoring to the cell cortex, thus contributing to local MT stabilization and membrane protrusion during cell polarization and migration. However, in cells lacking 4.1R, the localization of CLASP2 and of cortical platforms is disturbed, and MT organization and attachment to the cell margin are compromised. We propose that as a result of the reduced number of cortical platforms and dynamic alterations of the remaining platforms, growing MTs that approach the cell periphery are not captured properly but continue to grow. In addition, 4.1R depletion deregulates GSK3, causing increased binding of CLASP2 to the MT lattice. Our data show, for the first time, that this enhanced binding of CLASP2 to the MT lattice is not sufficient to cause MT capture, although it might stimulate MT growth persistence, and hence the bending of MTs at the cell edge of 4.1R-KD cells.
In conclusion, our study demonstrates a previously unknown role for the scaffolding protein 4.1R that, by properly organizing cortical platforms and regulating GSK3 activity, controls the MT organization and dynamics essential for cell polarity and migration. We propose that protein 4.1R enables the partitioning of the leading edge of motile cells into functional domains including cortical platforms, and that this organizing activity is essential for the regional regulation of GSK3 activity and MT dynamics.
Materials and Methods
Cells and transfections
ECV304 cells were grown in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum at 37°C in a 95% air/5% CO2 atmosphere. siRNA and plasmid transfections in ECV304 cells were performed using Lipofectamine 2000 according to the manufacturer’s instructions (Invitrogen). Cells were processed 72 hours after transfection. HEK293T cells were grown and plasmids were transfected as described previously (Drabek et al., 2006; van Haren et al., 2009). siRNA duplexes targeting mRNA sequences to human 4.1R were used as described (Ruiz-Sáenz et al., 2011). siRNAs targeting CLASP2 have also been described previously (Mimori-Kiyosue et al., 2005). Cells treated with only Lipofectamine 2000 or transfected with a control siRNA were used as controls in all experiments.
Antibodies
Affinity-purified polyclonal anti-4.1R (10b) and mouse monoclonal (mAb) anti-4.1R (M21) antibodies have been previously described (Correas et al., 1986; Ruiz-Sáenz et al., 2011). The mouse mAb to EB1 was obtained from BD Transduction Laboratories, the mouse mAb to GSK3α/β was from Invitrogen and the mouse mAbs to γ-tubulin (GTU88), dynein (intermediate chain) and α-tubulin (DM1A) were from Sigma-Aldrich. The rat mAb to tyrosinated tubulin (YL12) was from Abcam, and antibodies to CLASP2 (2358 and 12H2) have been described previously (Akhmanova et al., 2001; Maffini et al., 2009). The rabbit mAb to GAPDH (14C10) and the rabbit polyclonal antibody to pGSK3α/β (S21/9) were from Cell Signaling Technology. The rabbit polyclonal antibody to CLIP-170 (2360) has been described previously (Coquelle et al., 2002). The hamster polyclonal antibody to LL5β (Mimori-Kiyosue et al., 2005) was provided by the KAN Research Institute (Japan). Horseradish-peroxidase-labeled secondary antibodies were from Southern Biotechnology Associates and Jackson ImmunoResearch Laboratories, Inc. Secondary antibodies conjugated to Alexa Fluor 488, 594 or 647 were from Invitrogen.
Expression constructs
We used expression vectors for GFP-CLASP2γ (Akhmanova et al., 2001) and EB3-mCherry (Stepanova et al., 2003). The mCherry-α-tubulin was a gift from Dr R. Tsien (UCSD, San Diego, USA) and the mRFP-GSK3β(S9A) was a gift from Dr T. Wittmann (UCSF, San Francisco, USA; Addgene plasmid 24371). The construct for expression of the E. coli biotin ligase BirA was a gift from Dr D. Meijer (Erasmus MC, Rotterdam, The Netherlands). The pbio-GFP-4.1R construct was generated by a PCR-based strategy using pEGFP-C1 with a sequence encoding a biotinylatable tag for BirA upstream of the GFP coding sequence (Driegen et al., 2005).
Immunofluorescence analysis
Cells were either fixed with 10% formalin (37% formaldehyde solution; Sigma) and permeabilized with 0.02% Triton X-100, or fixed with methanol at −20°C (for CLASP2 and γ-tubulin staining) or with methanol at −20°C and 10% formalin at room temperature (for CLIP-170 staining). Cells were blocked and stained with the indicated antibodies followed by the appropriate secondary antibodies conjugated with Alexa Fluor 488 (excitation at 488 nm and emission collected at 505–530 nm), Alexa Fluor 594 (excitation at 543 nm and emission at 585–615 nm) or Alexa Fluor 647 (excitation at 633 nm and emission from 650 nm) and processed as described previously (de Cárcer et al., 1995). Controls to assess labeling specificity included incubations with control primary antibodies or omission of the primary antibodies. Fluorescence was examined using a confocal laser-scanning microscope LSM 510 in conjunction with an inverted microscope Axiovert 200M (Zeiss). LSM images were converted to TIFF format.
Microscopic and image analysis
Live imaging experiments and fluorescence-based analyses were performed as described (Drabek et al., 2006) (van Haren et al., 2009). Live-cell imaging of ECV304 cells transfected with GFP-CLASP2 was performed on an inverted Nikon Eclipse Ti-E (Nikon) research microscope with the Perfect Focus System (Nikon), equipped with a Nikon CFI Apo TIRF 100× 1.49 NA oil objective (Nikon) and QuantEM 512SC EMCCD camera (Roper Scientific). The latter was connected to the microscope by a Dualview (DV2, Roper) to enable simultaneous imaging of red and green fluorescence. Cells were analyzed at 37°C, using a stage-top incubator (model INUG2E-ZILCS, Tokai Hit). For imaging GFP–CLASP2 associations with MT plus-ends and for measuring MT dynamics in ECV304 cells stably expressing mCherry–α-tubulin, we used a spinning disk microscope similar to the set-up described above, equipped with a CSU-X1-A1 spinning disc confocal scanner unit (Yokogawa). Microscopes were controlled by MetaMorph 7.5 software (Molecular Devices).
Imaging was performed at 1 frame/second for measurement of MT dynamics and at 10 frames/second for fluorescence recovery after photobleaching (FRAP) assays, which were essentially carried out as described previously (Lee et al., 2010). Briefly, we bleached a circular region of interest (ROI) and measured fluorescence inside the bleached zone and in small ROIs outside of the bleached zone to correct for photobleaching during image acquisition. Data were analyzed with Prism (GraphPad). Curves were first normalized, and then values were averaged. Curves were fitted with the single-exponential equation: for control cells and with the double-exponential equation: for 4.1R-KD cells with the following supporting formulas: span1 = (plateau−y0)percent1×0.01; span2 = (plateau−y0)(100−percent1)×0.01. Correlation coefficients obtained in the curve fittings ranged from 0.96 to 0.97. The half-life of recovery was calculated as ln(2)/krecovery.
Colocalization analysis was performed using the Intensity Correlation Analysis plugin in Fiji (http://www.uhnresearch.ca/facilities/wcif/software/Plugins/ICA.html). The ROI was selected by the threshold masking function. Pearson’s correlation coefficient and Manders’ colocalization coefficients for channel 1 (protein labeled in red) and channel 2 (protein labeled in green) are shown in Fig. 3 (Bolte and Cordelieres, 2006). Images were prepared using ImageJ and Adobe Photoshop.
Pull-down and western blot analyses
For pull-down assays with GST proteins, recombinant proteins GFT–4.1R80, GST–4.1R60, GST–FERM, GST–CORE and GST–Cter and ECV304 cell extracts were prepared as described previously (Ruiz-Sáenz et al., 2011). The procedure used for the pull-down assays has been described elsewhere (Pérez-Ferreiro et al., 2006). Protein samples were separated by SDS-PAGE and transferred to PVDF (Millipore) in Tris-glycine-methanol buffer. Membranes were processed and developed as described by de Cárcer et al. (de Cárcer et al., 1995).
Streptavidin pull-down assays and mass spectrometry
For pull-down assays with streptavidin-coated beads, HEK293T cells transiently transfected with BirA and bio-GFP-4.1R constructs were processed as described previously (van Haren et al., 2009). One-dimensional SDS-PAGE gel lanes were cut into 2 mm slices using an automatic gel slicer and subjected to in-gel reduction with dithiothreitol, alkylation with iodoacetamide and digestion with trypsin (Promega, sequencing grade), essentially as described by van den Berg et al. (van den Berg et al., 2010). Nanoflow LC-MS/MS was performed on an 1100 series capillary LC system (Agilent Technologies) coupled to an LTQ-Orbitrap mass spectrometer (Thermo), operating in positive mode and equipped with a nanospray source. Peptide mixtures were trapped on a ReproSil C18 reversed phase column (Dr Maisch GmbH; column dimensions 1.5 cm×100 µm, packed in-house) at a flow rate of 8 µl/minute. Peptide separation was performed on ReproSil C18 reversed-phase columns (Dr Maisch GmbH; column dimensions 15 cm×50 µm, packed in-house) using a linear gradient from 0 to 80% B [A = 0.1% formic acid; B = 80% (v/v) acetonitrile, 0.1% formic acid] for 70 minutes and at a constant flow rate of 200 nl/minute using a splitter. The column eluent was directly sprayed into the ESI source of the mass spectrometer. Mass spectra were acquired in continuum mode; peptides were fragmented in data-dependent mode by CID. Peak lists were automatically created from raw data files using the Mascot Distiller software (version 2.3; MatrixScience). The Mascot search algorithm (version 2.2, MatrixScience) was used for searching against the UniProt database (release IPI_human_20091218.fasta). Peptide tolerance was set to 10 ppm and fragment ion tolerance was set to 0.8 Da. A maximum of two missed cleavages by trypsin were allowed, and carbamidomethylated cysteine and oxidized methionine were set as fixed and variable modifications, respectively. The Mascot score cut-off value for a positive protein hit was set to 65. Individual peptide MS/MS spectra with Mascot scores below 25 were checked manually and interpreted as valid identifications or discarded.
Statistical analysis
Data are expressed as means ± standard deviation (s.d.). An unpaired Student’s t-test was used to establish the statistical significance of differences between the means (*P<0.05; **P<0.01; ***P<0.001). Data from at least three independent experiments were processed.
Acknowledgements
We thank V. Serrano and L. Fernández for their technical expertise (CBMSO, Madrid). The expert technical advice of the personnel of the Optical and Confocal Microscopy Facilities (CBMSO and CNIC, Madrid) and the Bioinformatics Facility (CBMSO, Madrid) is gratefully acknowledged. A. R.-S. is a recipient of a fellowship from the Ministerio de Economía y Competitividad (MINECO), Spain.
Author contributions
A.R.-S. performed most of the experiments; J.H., L.S. and L.R. performed and analyzed some of the experiments; A.R.-S., J.H., N.G. and I.C. designed and analyzed most of the experiments; J.D., J.M. and M.A.A. designed and interpreted some of the experiments; A.R.-S., J.H., L.S., N.G. and I.C. participated in the writing of the manuscript; all the coauthors corrected the manuscript.
Funding
This work was supported by the Ministerio de Ciencia e Innovación [grant numbers BFU2011-22859 to I.C., CSD2009-00016 and BFU2012-32532 to M.A.A., SAF2011-22624 to J.M.]; the Comunidad de Madrid [grant number S2010/BMD-2305 to I.C.]; the Netherlands Organization for Scientific Research [grant number 91208002 to N.G.]; and the Cancer Genomics Centre of the Netherlands [grant number 050-060-206 to N.G.].