|Home | About | Journals | Submit | Contact Us | Français|
Eukaryotic cell migration proceeds by cycles of protrusion, adhesion and contraction, regulated by actin polymerization, focal adhesion assembly, and matrix degradation. However, mechanisms coordinating these processes remain largely unknown. Here we show that local regulation of thymosin-β4 (Tβ4) binding to actin monomer (G-actin) coordinates actin polymerization with metalloproteinase synthesis to promote endothelial cell motility. In particular and quite unexpectedly, FRET analysis reveals diminished interaction between Tβ4 and G-actin at the cell leading edge despite their co-localization there. Profilin-dependent dissociation of G-actin–Tβ4 complexes simultaneously liberates actin for filament assembly and facilitates Tβ4 binding to integrin-linked kinase (ILK) in the lamellipodia. Tβ4–ILK complexes then recruit and activate Akt2, resulting in matrix metalloproteinase-2 production. Thus, the actin-Tβ4 complex constitutes a latent coordinating center for cell migratory behavior, allowing profilin to initiate a cascade of events at the leading edge that couples actin polymerization to matrix degradation.
Adherent cell locomotion is a highly integrated process, initiated by forward extension of lamellipodia, and proceeding by repeated cycles of protrusion, adhesion, and contraction (Lauffenburger and Horwitz, 1996; Pollard and Borisy, 2003). The individual cellular processes in turn require orchestration of events at the molecular level, e.g., lamellipodial extension demands precise coordination of actin polymerization, matrix degradation, and assembly/disassembly of focal adhesions. Thymosin β4 (Tβ4) is a candidate regulator of cell protrusion because it is a multi-functional, “moonlighting” protein with important roles in distinct protrusion-related processes: actin polymerization and matrix metalloproteinase (MMP) expression. Interestingly, Tβ4 has apparently opposing roles in these processes. As the major sequestering protein of monomeric, globular actin (G-actin), Tβ4 is considered to have an anti-migratory function because it inhibits actin polymerization in vitro and in vivo, and induces cytoskeletal disorganization (Safer et al., 1991; Sanders et al., 1992). In contrast, activation of integrin-linked kinase (ILK) and Akt by Tβ4 is considered pro-migratory, because these signals induce MMP synthesis (Bock-Marquette et al., 2004). Consistent with the latter findings, overexpression of Tβ4, or treatment with Tβ4 peptide, promotes cell survival, migration, and angiogenesis (Arboleda et al., 2003; Bock-Marquette et al., 2004; Malinda et al., 1997; Smart et al., 2007). The magnitude of these opposing activities of Tβ4 may provide a critical determinant of cell migration, but the integration of signals and coordination of activities has not been resolved.
Tβ4 sequesters G-actin, rendering it unavailable for actin filament (F-actin) generation (Goldstein et al., 2005). Polarized elongation of F-actin in lamellipodia of migrating cells is well-established, and is subject to spatial and temporal coordination by multiple signaling molecules, e.g., rho GTPases and PI-3 kinase/Akt, and by actin-binding proteins, e.g., Arp 2/3 and cofilin (Lauffenburger and Horwitz, 1996; Ridley et al., 2003). Although G-actin availability is rate-determining for F-actin elongation, little is known about its spatial regulation during cell movement. Passive diffusion into the lamellipodia may provide sufficient G-actin for filament formation (Pollard et al., 2000). However, preferential G-actin accumulation in protruding zones of the leading edge of migrating cells suggests the possibility of active G-actin polarization during migration (Cao et al., 1993).
In addition to its intracellular actin-sequestering activity, extracellular administration of Tβ4 induces MMP expression in several cell types, including endothelial cells (EC) (Cierniewski et al., 2007; Philp et al., 2006; Qiu et al., 2007). MMP-mediated matrix degradation at the cell surface, particularly at the leading edge, is essential for migration of adherent cells. MMP-2 and membrane-type MMPs are critical for EC migration and for the angiogenic switch to an invasive phenotype (Haas and Madri, 1999; Malik and Kakar, 2006). The mechanism by which Tβ4 stimulates MMP expression is unclear, but it may be mediated by ILK (Bock-Marquette et al., 2004). The interaction of Tβ4 with ILK activates Akt, which promotes cell migration and survival. Moreover, ILK expression is up-regulated by Tβ4, and ILK induces MMP-2 expression in epithelial cells and increases cell movement during the epithelial to mesenchymal transition (Arboleda et al., 2003; Huang et al., 2007).
Profilin (PFN) is a potential mediator of Tβ4-mediated migration events, because it releases G-actin from Tβ4 and facilitates actin polymerization (Pantaloni and Carlier, 1993). In addition, PFN interacts with multiple pro-migratory proteins, including Rac/Rho, ROCK, and proline-rich vasodilator-stimulated phosphoprotein (VASP), exhibiting characteristics of a protein interaction platform that facilitates convergence of these signals (Witke, 2004). Preferential localization of PFN at the protrusive edge in fibroblasts (Buss et al., 1992) suggests PFN regulates local signal activation events and actin dynamics during cell migration, consistent with a role for PFN in modulating the potential effects of Tβ4 on G-actin availability and metalloproteinase synthesis.
Here we show the interaction between Tβ4 and G-actin is polarized, with diminished interaction at the leading edge of migrating EC. Polarized dissociation of the actin–Tβ4 complex increases the flux of actin from the Tβ4-bound form to the F-actin pool and actin polymerization at the leading edge. Moreover, polarized release of Tβ4 in the cell front permits Tβ4 binding to lamellipodial ILK to facilitate Akt2-ILK interaction and Akt2 phosphorylation, inducing MMP-2 expression. Finally, PFN is required for that polarization and consequent actin polymerization and protease synthesis. These findings establish a mechanism by which spatially-restricted release of G-actin coordinates Tβ4 functions to enhance polarized actin filament formation and metalloproteinase synthesis during EC migration.
G-actin availability in protrusive lamellipodia is rate-limiting for actin filament formation; however, little is known about its spatial distribution during cell movement. G-actin accumulated preferentially at the leading edge of EC induced to migrate by wound injury (Figure 1A), consistent with its appearance in lamellipodial extensions during cell spreading (Cao et al., 1993). Because G-actin available for filament formation depends on local concentration and on the absence of inhibitory binding partners, we examined the distribution of Tβ4, an abundant cytoplasmic protein that sequesters G-actin. Tβ4 (but not Tβ10 used as a control) co-localized with G-actin at the cell leading edge. This result was unexpected because it suggests low availability of G-actin for filament formation. Co-localization does not necessarily indicate physical association, and therefore Förster resonance energy transfer (FRET) was used to examine the spatial distribution of actin–Tβ4 complex in migrating EC. FRET was detected by acceptor photobleaching using enhanced green fluorescent protein (GFP)-actin as donor and Discosoma sp. red fluorescent protein (DsRed)-Tβ4 as acceptor. To restrict analysis to monomeric G-actin, cells were transfected with a construct expressing GFP-actin containing a G13R mutation in the nucleotide-binding pocket that prevents polymerization and filament formation in cells (Posern et al., 2002). Fusion protein interaction was confirmed by co-immunoprecipitation (Supplementary Figure S1A). DsRed-Tβ4 was photobleached near the center or the lamellipodia of migrating EC, and the interaction between Tβ4 and actinG13R determined as FRET (EFRET) efficiency to GFP-actinG13R. FRET was optimized with Tβ4 fusion proteins with varying linker lengths, and a maximum EFRET of 23% was detected using DsRed-GPVAT-Tβ4 (Supplementary Figure S1B). EFRET between Tβ4 and actinG13R in the lamellipodia of migrating cells was about 10%, less than half of that observed near the cell center (Figure 1B, left). In quiescent cells, EFRET was similar in the edge and near the center, and both were comparable to that in the center of rapidly migrating cells. Similar results were obtained in EC expressing a different non-polymerizable actin mutant (actinR62D, not shown). In a positive control experiment, an EFRET of 34% was detected in EC transfected with pGFP-DsRed with an intrinsic short linker; in a negative control, an EFRET of <3% was detected in EC co-transfected with pDsRed and pGFP (Figure 1B, right).
To more precisely locate the actin–Tβ4 interaction, the spatial distribution of EFRET was measured by sensitized emission FRET, which depends on simultaneous detection of decreased GFP emission and increased DsRed emission. The results verified the polarized interaction with a low EFRET of about 3% in the lamellipodia of migrating cells despite co-localization of actinG13R-GFP and DsRed-Tβ4 in this region (Figure 1C, top panels). An EFRET of up to 25% was observed in the cell body. In control experiments, co-transfection of migrating EC with plasmids expressing GFP and DsRed showed uniformly weak interaction throughout the cell (EFRET ≈ 4%), and transfection with GFP-DsRed-expressing plasmid exhibited high-level FRET (EFRET ≈27%) interaction. In addition, actinG13R-GFP accumulated at the leading edge of migrating EC (Figure 1C) confirming G-actin localization detected with DNase I. This conformation is important because DNase I also binds pointed ends of actin filaments. However, specificity for G-actin is enhanced in the leading edge by capping of F-actin by gelsolin, actinin, and Apr2/3 that prevents DNase I binding. Taken together, these results show that despite co-localization of G-actin and Tβ4 in migrating cells, their interaction is polarized, characterized by a large amount of G-actin, free of Tβ4, in lamellipodia.
We examined whether reduction of Tβ4 expression by a short-hairpin RNA (shRNA) increases G-actin availability and facilitates actin polymerization in migrating cells. ECs were transfected with a bicistronic vector expressing Tβ4 shRNA, and separately expressing GFP for detection of transfected cells. Tβ4 expression was reduced by about 60% and 90% in transiently and stably transfected EC, respectively (Figure 2A). shRNA specificity was shown by the lack of inhibition of Tβ10 expression (Figure 2A). Transient knock-down of Tβ4 using the shRNA increased the amount of G-actin and F-actin in lamellipodia of migrating EC compared to control cells transfected with shRNA targeted to luciferase (Figure 2B). This partial knock-down also increased wound-induced EC migration by about 30 to 40% compared to controls, consistent with the observed increase in available G-actin and F-actin in the cell front (Figure 2C and Supplementary Movie1).
The influence of Tβ4 on G-actin availability and cell migration was investigated further in cells over-expressing actin with point mutations at Tβ4 binding sites. Lys11 and Leu17 in Tβ4 are critical for binding to actin (Carlier et al., 1996) via interaction with cognate actin sites, Glu167 and Asp24 (Irobi et al., 2004; Safer et al., 1997). His-tagged, actin mutants containing D24A, E167A, or both, were co-expressed with Flag-tagged Tβ4. Substantially less Tβ4 bound actin containing either single or double mutations (Figure 2D). The E167A mutation, but not D24A, slightly suppressed actin binding to profilin-1, and neither affected the interaction with cofilin (Supplementary Figure S2A). We examined the influence of reduced interaction with Tβ4 on intracellular dynamics of actin during cell migration. Actin polymerization was determined by fluorescence recovery after photobleaching (FRAP). The calculated lateral diffusion coefficient (D) of actin-GFP reflects the actin polymerization rate in vivo, not the diffusion rate of actin monomer (McDonald et al., 2006; McGrath et al., 1998; Vasanji et al., 2004). ActinD24A,E167A exhibited an enhanced polymerization rate compared to wild-type as indicated by a 1.6-fold increase in D at leading edge; there was no difference in the central area of the cell (Figure 2E). The influence of actin mutants on cell migration was measured in cells transfected with a biscistronic vector expressing GFP. Overexpression of actin carrying single mutation D24A and E167A, or both, markedly enhanced wound-induced cell migration, as measured by the number of migrating cells and by migration distance (Figure 2F and Supplementary Figure S2B,C). In control experiments, normal polymerization of the actin mutants was shown by the filamentous morphology of actinD24A,E167A-GFP chimeras, and overexpression of the fusion proteins did not alter Tβ4 localization (Supplementary Figure S3). Together, these data indicate that G-actin availability, mediated by polarized dissociation of the actin–Tβ4 complex, is likewise polarized, facilitating actin polymerization at the cell leading edge and enhancing cell migration.
Polarized dissociation of the actin–Tβ4 complex may influence other Tβ4-mediated migratory events including metalloproteinase synthesis, necessary for extracellular matrix degradation and cell migration (Goldstein et al., 2005). The specific function of Tβ4 was shown in EC stably transfected with shRNA targeted against Tβ4 which markedly reduced MMP-2 activity as measured by gelatin zymography (Figure 3A, top), and confirmed by immunoblot (Figure 3A, bottom). Transfection of cells with human Tβ4 cDNA increased intracellular Tβ4 by about 30%, and MMP-2 by about 50% (Figure 3B). Transfection with Tβ4 mutant K11P, but not L17A, further stimulated MMP-2 synthesis; both mutants bind actin with reduced affinity (Carlier et al., 1996). The activity of Tβ4 was further tested by a rescue experiment. Transfection of cells with cDNA encoding wild-type or mutant human Tβ4 effectively restored Tβ4 expression in bovine EC stably expressing Tβ4 shRNA-1; the presence of two mismatched nucleotides between human and bovine Tβ4 likely contributed to the bypass of shRNA inhibition. MMP-2 synthesis was partially restored by over-expression of wild-type Tβ4, but completely restored by the K11P mutant indicating the actin–Tβ4 interaction is inhibitory (Figure 3B). The L17A mutation did not restore MMP-2 expression, suggesting an intact LKKTET domain is required for Tβ4-mediated protease synthesis, consistent with the reported role of this domain in Tβ4-induced EC migration and angiogenesis (Philp et al., 2003; Philp et al., 2006).
To evaluate the influence of Tβ4 on cell migration, confluent EC stably expressing shRNA targeted against Tβ4 were induced to migrate by razor-wound. Near-complete Tβ4 depletion inhibited cell migration by about 70% (Figure 3C; note that levels of Tβ4 in these experiments are much lower than that in Figure 2C). Over-expression of Tβ4, and especially Tβ4K11P, partially restored migration. The influence of Tβ4 on in vitro angiogenesis was investigated by tube formation in matrigel, a process dependent on both matrix degradation and cell motility. Overexpression of Tβ4 promoted tube formation, and again the Tβ4K11P mutant exhibited greater stimulatory activity (Figure 3D, top). Measurement of cell migration by razor-wound gave essentially identical results (Figure 3D, bottom). Taken together, these data reveal a stimulatory role of Tβ4 on MMP-2 synthesis and EC migration. The apparent contradiction with the stimulatory effect of Tβ4 partial knockdown on migration (Figure 2C) suggests a complex relationship between Tβ4 level and cell migration. Possibly, anti-migratory sequestration of G-actin and pro-migratory stimulation of MMP-2 expression exhibit dissimilar dependences on Tβ4 concentration or temporal expression (see below).
An interaction between Tβ4 and ILK has been proposed as a promoter of cell migration (Bock-Marquette et al., 2004). Thus, dissociation of the actin–Tβ4 complex may promote an interaction between newly available Tβ4 and ILK to increase MMP-2 expression and cell motility. To test this hypothesis, the interaction of Tβ4 with ILK in the presence or absence of G-actin was determined in vitro by surface plasmon resonance (SPR). Chimeric ILK-myc/His expressed in insect cells bound with moderate affinity (Kd = 18.7 μM) to Tβ4 immobilized on the sensor chip (Figure 4A). Less tight interaction was observed for bacterially-expressed GST-ILK (Kd > 0.2 mM), suggesting that a high-affinity interaction may require a post-translational modification. We investigated whether Tβ4 binds at or near the ILK kinase domain (ILK-KD). Tβ4 binding to immobilized GST-ILK-KD was shown by SPR (Figure 4B). Preincubation of Tβ4 with G-actin markedly suppressed its interaction with ILK-KD suggesting it successfully competes with ILK-KD for Tβ4 binding; indeed, G-actin binds Tβ4 (Kd = 0.7 μM (Yu et al., 1993)) with a higher affinity than the ILK–Tβ4 interaction. In a control SPR experiment, G-actin did not inhibit ILK-KD binding to Tβ4K11P-His or Tβ4L17A-His (Figure 4C), indicating that G-actin binding to Tβ4, not to ILK-KD, inhibits the Tβ4 and ILK-KD interaction (Figure 4B).
We determined whether the interaction of Tβ4 with ILK subsequently affects its interaction with Akt1/2, as both may be kinase targets of ILK. SPR experiments showed that Akt2 (Figure 4D) and Akt1 (not shown) interact with GST-ILK-KD. Addition of Tβ4 stimulated Akt2 interaction with GST-ILK-KD, primarily by increasing the binding “on-rate”. However, Tβ4 (at 10 μM) did not increase Akt1 binding to GST-ILK-KD (not shown), indicating highly selective stimulation. Furthermore, direct interaction between Tβ4 and Akt2 was not observed (not shown), suggesting that Tβ4 binding to ILK facilitates Akt2 recruitment to an ILK-KD–Tβ4 complex.
We investigated the influence of Tβ4 and ILK on Akt2 phosphorylation at Ser474, a reported ILK target (Troussard et al., 2003). GST-Akt2 (dephosphorylated) was incubated in vitro with pre-activated ILK-myc/His, and Akt2 phosphorylation at Ser474 detected by immunoblot with anti-phospho-Akt antibody. ILK by itself induced low-level phosphorylation of GST-Akt2 (Figure 4E). Tβ4 substantially increased ILK-mediated Akt2 phosphorylation, and G-actin prevented this increase. Maximal phosphorylation of Akt2 was detected at the lowest level of Tβ4, but the “on-rate” of Tβ4 binding to Akt2 is dose-responsive in the SPR assay, likely due to different assay time-scales. Pre-activation of ILK-myc/His by incubating it with lysate from migrating cells increased Tβ4–ILK-induced Akt2 phosphorylation, suggesting that post-translational modifications of ILK may be one of several possible mechanisms leading to Akt2 activation; PI(3)K inhibitor LY294002 attenuated the cell lysate-induced Akt2 phosphorylation, indicating a role in ILK activation (Supplementary Figure S4). Together, these data suggest the Tβ4 interaction with ILK facilitates Akt2 binding to ILK-KD, causing Akt2 phosphorylation. Moreover, G-actin prevents Akt2 activation by blocking the interaction between Tβ4 and ILK.
We determined whether a reduction in Tβ4 binding to G-actin in cells results in a reciprocal increase in Tβ4 binding to ILK and its binding partner, PINCH. Flag-Tβ4 bearing K11P or L17A mutations that prevent its interaction with G-actin were expressed in EC. Both mutations markedly reduced Tβ4 interaction with actin (Figure 4F) confirming previous results (Carlier et al., 1996). Decreased binding to actin was accompanied by increased binding of Tβ4 to both ILK and PINCH. Immunofluorescence confocal scanning microscopy (at a focal plane near the substratum plasma membrane) revealed nearly uniform cytoplasmic distribution of ILK and PINCH in quiescent cells, as well as co-localization at apparent focal adhesions. However, ILK, and to a lesser extent PINCH, preferentially accumulates at the forward edge of migrating EC, indicating co-localization with actin-free Tβ4, and thus an opportunity for interaction in the lamellipodia (Figure 4G). In control experiments we found that knock-down of Tβ4, or overexpression of wild-type or mutant Tβ4, did not alter ILK localization in migrating cells (Supplementary Figures S5, S6, S7).
We examined whether the ILK–Akt2 interaction is required for Tβ4-induced metalloproteinase expression and cell migration. Over-expression of Tβ4 substantially increased MMP-2 expression (Figure 5A). The induction was inhibited by PI(3)K inhibitors wortmannin and LY294002 (Figure 5A), and by siRNA-mediated knockdown of Akt2 (Figure 5B), indicating a specific requirement for Akt2 activity. Knockdown of ILK likewise suppressed Tβ4-induced MMP-2 expression (Figure 5C). Tβ4 markedly increased phosphorylation of Akt2 at Ser474 in an ILK-dependent manner; only low-level induction of Akt1 phosphorylation at Ser473 was observed indicating target specificity (Figure 5D). Finally, Akt2 knockdown inhibited both wound-induced cell migration and chemotaxis (Figure 5E). Together, these data show that productive interaction between ILK and Akt2 is required for Tβ4-induced MMP-2 synthesis and EC migration.
We determined whether ILK–Akt2-induced MMP-2 expression depends on mammalian target of rapamycin (mTOR), a major Akt substrate. Tβ4 overexpression moderately increased PKB/Akt-dependent mTOR activation, as determined by phosphorylation at Ser2448 (Supplementary Figure S8A). mTOR siRNA attenuated Tβ4-induced MMP-2 expression, indicating a partial dependence of Tβ4-inducible MMP-2 expression on mTOR (Supplementary Figure S8B,C). In summary, Tβ4 dissociation from G-actin complex increases formation of a ternary Tβ4–ILK–Akt2 complex, leading to Akt2 phosphorylation, mTOR activation, and consequent MMP-2 expression to promote cell motility.
Overexpression of Tβ4 carrying the K11P mutation at the actin-binding site induced a small increase in Akt2 phosphorylation (Supplementary Figure S9A), consistent with the increased restoration of MMP-2 expression in Tβ4-depleted cells (Figure 3B), and with the regulatory role of G-actin in Tβ4-inducible Akt2 phosphorylation in vitro (Figure 4D). Despite the increase in actin polymerization and cell motility (Figure 2), transient knockdown of Tβ4 did not affect Akt phosphorylation or MMP-2 expression (Supplementary Figure S9B). However, stable depletion of Tβ4 inhibited both MMP-2 expression and cell motility (Figure 3). These results suggest differential sensitivity of ILK–Akt activation and MMP-2 induction to decreases in cellular Tβ4 level.
PFN dissociates G-actin from Tβ4, and freed G-actin, possibly in association with PFN, facilitates actin polymerization (Pantaloni and Carlier, 1993). Thus, PFN could play a significant role in polarized dissociation of the actin–Tβ4 complex in migrating cells. PFN co-localized with G-actin at the leading edge of migrating EC, but distributed uniformly in quiescent ECs (not shown), consistent with the preferential localization at protrusive edges of fibroblasts (Buss et al., 1992). PFN was efficiently knocked down by siRNA targeting PFN-1, or both PFN-1 and PFN-2 (Figure 6A). PFN-1 knockdown depolarized the actin–Tβ4 interaction in migrating EC, as indicated by uniform EFRET in siRNA-treated EC (Figure 6B), suggesting PFN increases G-actin availability at the leading edge. PFN-1 knockdown cells remained polarized during random cell migration, but the fraction of polarized cells was reduced (Supplementary Figure S10). Consistent with a previous report (Ding et al., 2006), PFN-1 siRNA did not exhibit compensatory up-regulation of PFN-2 (not shown). The effect of PFN-1 knockdown on actin polymerization was measured in cells transfected with a biscistronic vector co-expressing GFP-actin and shRNA targeting PFN-1. PFN-1 knockdown almost completely blocked formation of lamellipodial F-actin (Figure 6C). These data establish an essential role of PFN-1 in the spatially restricted dissociation of the actin–Tβ4 complex, and consequent polarization of G-actin during cell migration.
PFN-mediated release of Tβ4 from its complex with G-actin can influence a second pro-migratory pathway, i.e., it can facilitate the interaction of Tβ4 with ILK and thereby increase metalloprotease synthesis. Indeed, PFN-1 knockdown induced a phenotypic switch characterized by a 70% increase in the actin–Tβ4 interaction and about a 60% decrease in the interaction of Tβ4 with ILK (and with PINCH to a lesser extent) (Figure 7A). The diminished Tβ4–ILK interaction is expected to decrease Akt2 activation, and consequently inhibit MMP-2 expression. In fact, PFN-1 knockdown markedly reduced MMP-2 expression in migrating cells to nearly the level observed in quiescent cells, and decreased MMP-2 activity (Figure 7B), consistent with the observed decrease in Akt2 phosphorylation at Ser474 (Supplementary Figure S10B). The somewhat greater effectiveness of the PFN knockdown (which depletes both PFN-1 and PFN-2) compared to the PFN-1 knockdown suggests a possible role for PFN-2 in Tβ4-induced MMP-2 synthesis. Over-expression of Tβ4, particularly the actin-blind Tβ4K11P mutant, significantly restored MMP-2 expression in PFN-1 knockdown cells (Figure 7C). PFN knockdown inhibited wound-induced EC migration by about 50% to 60% (Figure 7D and Supplementary Figure S10C). Migration was partially rescued by co-overexpression of Tβ4K11P and actinD24A,E167A, but not by either protein by itself (Figure 7E). The incomplete restoration may be due to disrupted activity of other motility-related, PFN-interacting proteins, e.g., Rac/Rho, ROCK, VASP, and WASP, required for actin cross-linking, branching, and bundling (Witke, 2004). Together, these results suggest PFN activates a dual function, pro-migratory switch in which the release of Tβ4 from G-actin promotes both actin polymerization and MMP-2 expression.
Our results show that dissociation of the actin–Tβ4 complex is spatially restricted to the leading edge of migrating EC. PFN-dependent, polarized dissociation increases the flux of actin from the Tβ4-bound form to the F-actin compartment at the leading edge. Concurrently, the release of Tβ4 from G-actin facilitates formation of an ILK–Tβ4–Akt ternary complex that increases Akt phosphorylation (Figure 7F). The resultant activation of Akt2 induces MMP-2 synthesis to facilitate matrix degradation, and Akt1 activation promotes actin polymerization at the leading edge.
Cell locomotion is driven by polarized F-actin growth that is rate-determined by the availability, i.e., the time-integrated amount of G-actin available for polymerization, in the leading edge. PFN increases G-actin availability by increasing the flux from the Tβ4–G-actin pool, but decreases G-actin concentration by promoting addition to F-actin filaments. The high intracellular ratio of Tβ4:G-actin (e.g., 0.4 mM:0.1 mM in human platelets) suggests that most cytosolic G-actin is sequestered by Tβ4 (Huff et al., 2001; Safer et al., 1991). Our experiments reveal that substantial G-actin is generated at the leading edge by the polarized dissociation of G-actin from Tβ4. The consequences of this interaction were shown by experiments in which the interaction was inhibited by Tβ4 knockdown or by mutation of actin at its Tβ4-binding site. In both cases we observed increased G-actin availability, lamellipodial polymerization of actin, and cell motility. These results also confirm the important role of regulated G-actin availability in cell motility. A contrasting result was reported in tumor cells in which Tβ4 knockdown decreased cell migration (Kobayashi et al., 2002). The different response may be due to a greater dependence on Tβ4-inducible protease expression for tumor cell movement in a matrix-rich environment.
ILK phosphorylates PKB/Akt (Delcommenne et al., 1998) and binds PINCH, paxillin, and parvin (Wu and Dedhar, 2001), all important components regulating cell migration (Goldstein et al., 2005). ILK is critical for EC migration and angiogenesis (Friedrich et al., 2004; Tan et al., 2004; Vouret-Craviari et al., 2004). The mechanism by which Tβ4 stimulates MMP expression is unclear (Bock-Marquette et al., 2004). Because Tβ4 co-immunoprecipitates with ILK–PINCH, activation of Akt in the ILK–Akt complex could be responsible for MMP expression and increased cell migration. Our studies show an interaction of Tβ4 with the kinase domain of ILK, and that formation of the ternary Tβ4–ILK–Akt2 complex increases Akt2 phosphorylation with a consequent increase in MMP-2 expression and cell migration. These Tβ4-mediated processes are regulated by G-actin binding to Tβ4: dissociation of actin–Tβ4 complex promotes Tβ4–ILK interaction and MMP-2 expression. Thus these results provide a molecular mechanism for Tβ4-promoted protease synthesis and cell migration.
The Akt family consists of three isoforms: Akt1, Akt2, and Akt3. The role of Akt2 in cell migration is controversial: Akt2 knockout elevates Rac and Pak1 activities to increase fibroblast cell motility (Philp et al., 2006), but our knockdown studies indicate that Tβ4-inducible Akt2 activation induces protease synthesis to facilitate EC migration. Consistent with our findings, transient knockdown or over-expression studies show a stimulatory role of Akt2 in cell migration and invasion in multiple cell types (Arboleda et al., 2003; Cheng et al., 2007; Irie et al., 2005; Sithanandam et al., 2005). Thus, Akt2 may have dual roles in induction of MMP and inhibition of migratory signaling including Rac. Akt1 is critical for actin polymerization and cell motility, particularly for EC migration (Ackah et al., 2005; Chen et al., 2005) and in vivo angiogenesis (Ackah et al., 2005; Phung et al., 2006). Our data show that Tβ4 stimulates stronger ILK-mediated activation of Akt2 than that of Akt1. We speculate that weaker Akt1 induction results in a tighter localization of active Akt1 to the leading edge (not shown), enhancing local actin polymerization.
There is evidence that ILK is a kinase-inactive “pseudokinase” (Boudeau et al., 2006). However, ILK phosphorylation of Akt1 at Ser473 has been shown (Delcommenne et al., 1998), and recently confirmed (Joshi et al., 2007; White et al., 2006). Our experiments suggest that ILK phosphorylates Akt2 at Ser474, that Akt2 is a better substrate than Akt1, and that a post-translational modification to ILK is required for its activity. Our results may help to explain the absence of detectable ILK-mediated Akt1 phosphorylation in some conditions.
Cell protrusion requires spatial and temporal coordination of actin polymerization, matrix degradation, and assembly/disassembly of focal adhesions. Several mechanisms may coordinate actin polymerization and cell adhesion during cell migration, including biphasic, fibronectin-dependent Rac activation (Cox et al., 2001), Rho-mediated feedback (Coluccio and Geeves, 1999), and myosin II-regulated mechanical responses (Giannone et al., 2007; Gupton and Waterman-Storer, 2006). Our results indicate that regulated dissociation of Tβ4–G-actin coordinates motility-related processes: locally released G-actin and Tβ4 induce actin polymerization and ILK–Akt2-mediated MMP-2 synthesis, respectively. The high intracellular ratio of Tβ4-to-G-actin presents an argument against a regulatory role for actin–Tβ4 complex dissociation since excess free Tβ4 would be expected to bind ILK constitutively. However, we show that alteration of intracellular Tβ4 level by siRNA-mediated silencing or by cDNA over-expression, modulates MMP-2 expression and cell motility. These results provide compelling evidence that the amount of available Tβ4 is not in excess in non-stimulated cells, and that PFN-mediated release of Tβ4 from G-actin can indeed cause biologically significant downstream consequences. Possibly, the apparent inconsistency is explained by low-affinity binding of excess Tβ4 to F-actin (Ballweber et al., 2002; Carlier et al., 1996), particularly at the leading edge where F-actin is highly enriched. Thus, sequestration of Tβ4 by both actin forms may prevent interaction with ILK in quiescent cells, until specifically released from G-actin at the cell front during migration. Our experiments are consistent with previous reports of exogenous Tβ4 enhancing MMP-2 expression and cell migration (Philp et al., 2006); Moreover, our results suggest that interaction of Tβ4 with cell surface receptors is not necessary for its activity.
Tβ4 binding to G-actin negatively regulates G-actin availability and consequent actin polymerization, whereas Tβ4 binding to ILK enhances Akt2-dependent MMP-2 expression and extracellular matrix degradation. These Tβ4-dependent events have opposing effects on cell migration that results in a complex, multi-phasic dependence of motility on intracellular Tβ4 (Supplementary Figure S11). Our data are consistent with a model in which Tβ4 overexpression results in an increase in MMP-2 expression that more than compensates for decreased G-actin availability, resulting in increased cell movement. Paradoxically, Tβ4 knockdown results in an increase in G-actin availability that more than compensates for decreased MMP-2 expression, again resulting in increased cell movement. However, deep depletion of Tβ4 almost completely suppresses MMP-2 expression and matrix degradation forming a migration barrier accompanied by cortical F-actin enrichment (not shown) even in the presence of a high level of available G-actin.
In summary, our results show that PFN induces polarized dissociation of actin–Tβ4 complex resulting in G-actin availability at the leading edge of a motile EC. The spatially-restricted dissociation process enhances cell movement by simultaneously promoting actin polymerization and inducing formation of a ternary Tβ4–ILK–Akt2 complex which induces MMP-2 synthesis. These findings provide a new mechanism for coordination of actin polymerization and matrix degradation during cell migration.
Bovine ECs isolated from adult bovine aortas were cultured in DME/Ham’s F-12 medium (GIBCO, Gaithersburg, MD) containing 5% fetal bovine serum (FBS) at 37°C in a humidified atmosphere of air containing 5% CO2. Human umbilical vein ECs were maintained in MCDB medium with 10% FBS. EC migration was quantitated by “razor-wound” method (Ghosh et al., 2002).
pEGFP-actin and pEGFP-C1 were from Clontech (Mountain View, CA). Wild-type and mutant actin cDNAs were amplified by PCR as KpnI/XhoI fragments, and cloned into pcDNA 3.1/myc-His (Invitrogen, Carlsbad, CA). GFP cDNA, amplified from pEGFP-C1, was cloned into pTandem-1 (Novagen, San Diego, CA) immediately after the IRES, and actin cDNAs as NcoI/XhoI fragments were cloned into the plasmids before IRES with a polyHis tag. The Tβ4 gene coding region was amplified from a bovine cDNA pool (BioChain, Hayward, CA) and cloned into pDsRed-N1 (gift from Dr. Yang Guo, Cleveland Clinic) with respective linker sequences, or into pcDNA 3.1 with or without FLAG-tag. GFP cDNAs with linker sequences were cloned into pDsRed-N1 to express GFP-DsRed chimeras. All cloned plasmids were verified by enzyme digestion and sequencing. ECs at 40% confluence were transfected with plasmids using lipofectin (Invitrogen) in serum-free Opti-MEM medium (Invitrogen) overnight.
Actin and Tβ4 mutants were generated by PCR using GeneTailor Site-Directed Mutagenesis System (Invitrogen). In brief, pEGFP-actin (Clontech, Mountain View, CA) was used as a template to generate G13R, R62D, and L267D mutations for inhibiting actin polymerization, and D24A and E167A for disrupting interaction of actin and Tβ4. pcDNA-Tβ4 (with or without FLAG tag) was used as a template to generate K11P and L17A mutations. All clones were verified by double enzyme digestion and sequencing.
ILK cDNA (OriGene, Rockville, MD) was cloned into pcDNA-myc/his as BamHI/XhoI fragment or pET41-GST (Novagen) as NcoI/XhoI fragment. ILK-myc/his was in vitro translated from pcDNA-ILK vector with insect EasyXpress kit (Qiagen, Valencia, CA), and purified with MagneHis system (Promega). pET-GST-ILK was transformed into Rosetta-gemi2 bacteria (Novagen), and protein expression induced with 0.2 mM IPTG at 30 °C for 5 h with chloramphenicol, streptomycin, tetracycline and kanamycin. Soluble protein was extracted with CelLytic B cell lysis reagent (Sigma), and GST-ILK purified with B-PER GST purification kit (Pierce, Rockford, IL). Insect GST-Akt1 and GST-Akt2 were obtained from Cell Signaling Technology, and Akt1-His and Akt2-His from Millipore (Billerica, MA). Tβ4 cDNA was cloned into pTriEx-his (Novagen) as a NcoI/XhoI fragment, and expressed in BL21 bacteria by 0.2 mM IPTG induction at 30 °C for 3 h. Tβ4-His was purified using B-PER His purification kit (Pierce).
Experiments were conducted on BIAcore 3000 instrument (GE Health, Uppsala Sweden) with a running buffer of 10 mM Hepes (pH 7.4), 150 mM NaCl, 0.005% Surfactant P20 (for high concentration of actin: 5 mM phosphate buffer) at 20 μl/min, 25°C. Purified Tβ4, ILK-myc/his, or ILK-KD was immobilized on CM5 sensor chip (GE Health) at about 1,200 resonance units. Proteins dialyzed in running buffer at 4°C were loaded onto the chip for 1 min, followed by injection of running buffer. A pulse of 5 μl of 10 mM NaOH and 1 M NaCl was injected to regenerate the surface.
ECs were fixed with PBS containing 3.7% paraformaldehyde, permeabilized with 0.1% Triton X-100 for 5 min, and stained with anti-Tβ4 (BioDesign, Saco, ME or Calbiochem, San Diego, CA), anti-ILK (Cell Signaling Technology, Danvers, MA), or anti-PFN-1 (Cell Signaling Technology) antibodies, followed by incubation with Alexa Fluor 488- or 568-conjugated IgG (Molecular Probes, Eugene, OR). For visualization of G-actin or F-actin, cells were labeled with Alexa Fluor 488-, or 597-conjugated DNase I and Alexa Fluor 350-, 488-, or 568-phalloidin (Molecular Probes), respectively, and examined by confocal scanning using a TCS-SP microscope (Leica, Heidelberg, Germany). To avoid thickness artifacts, single-plane confocal images were obtained with a small, 1.0 Airy pinhole. Images were analyzed with ImagePro software (Media Cybernetics, Carlsbad, CA), and mean fluorescent intensity was quantified in 5–8 randomly selected 2 × 2 μM rectangles in each of 12 to 15 cells.
ECs were transfected with a plasmid encoding GFP-LinkerGPVAT-DsRed (as positive control), or co-transfected with plasmids expression GFP and DsRed separately (as negative control), or plasmids expressing DsRed-tagged Tβ4 plus GFP-actinG13R. Two FRET-based methods were used:
Acceptor Photobleaching ECs were fixed with 3.7% paraformaldehyde, and fluorescence images of fixed cells were taken in both donor (GFP) and acceptor (DsRed) channels before and after photobleaching. Acceptor photobleaching was performed with a 568-nm krypton laser at maximal power for a time sufficient to provide 90% photobleaching. FRET efficiency was calculated as increased mean GFP fluorescence intensity in the photobleached area.
Sensitized Emission Transfected ECs were fixed and serial images acquired in three channels: GFP, DsRed, and FRET. Removal of spectral bleed-through, correction for differences in fluorophore expression levels, and background subtraction were performed with PFRET software (CircuSoft, Hockessin, DE). The results of FRET efficiency were displayed with a color spectrum.
Molecular mobility of GFP-actin was evaluated by FRAP as described (Vasanji et al., 2004). In brief, ECs were transfected with pEGFP-actin and subjected to wound-induced migration. Medium was replaced with phenol red-free medium containing 25 mM Hepes. Living cells were photobleached in 2-μm-diameter circles using a 488-nm argon laser at maximal power. Recovery was monitored by repetitive scanning of bleached areas at 20% power. Fluorescence recovery curves were fit by non-linear regression and expressed as lateral diffusion coefficients, D.
Duplex oligonucleotides encoding Tβ4 shRNA, PFN-1 shRNA, or control luciferase shRNA (Invitrogen) were annealed as a BamH I/Hind III fragment and cloned into pRNAT-U6-GFP/Neo (GenScript, Piscataway, NJ). Cells were transiently transfected with FuGene HD (Roche, Indianapolis, IN) and used 24 h after transfection. Stably transfected EC lines were selected with 600 μg/ml neomycin, and colonies from a single cell maintained with 400 μg/ml neomycin. For Tβ4 reconstitution assay, stable Tβ4-depleted ECs were transfected with pcDNA (Vector), pcDNA-Tβ4WT, pcDNA-Tβ4K11P, or pcDNA-Tβ4L17A with FuGene HD overnight in medium containing 5% serum. For PFN rescue experiments, cells were transfected with pRNAT-Luc, or pRNAT-PFN-1 plus pcDNA (vector), pcDNA-Tβ4 or pcDNA-actin with or without mutations. Cells were cultured for another 24 h, and then replaced with serum-free medium for protease zymography, MMP-2 immunoblot assay, and for wound-induced migration assay.
Student’s t test was used to calculate statistical significance with p < 0.05 representing a statistically significant difference.
Detailed methods for wound-induced cell migration assay, immunoprecipitation, immunoblot, optimization of FRET measurement, in vitro angiogenesis assay, siRNA treatment, MMP zymography, chemotaxis assay, protein pretreatment, and oligonucleotide primer sequences are in the Supplementary Experimental Procedures.
We are grateful to Alan Levine, Tom Egelhoff, and Sanjay Pimplikar for helpful suggestions. We thank Amit Vasanji and Judith Drazba for technical assistance in image analysis. This work was supported by National Institutes of Health grants P01 HL29582, P01 HL76491, and R01 HL075255 (to P.L.F.), and by an American Heart Association Scientist Development Grant (to P.K.G.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.