CapG (gCap39) is a ubiquitous gelsolin-family actin modulating protein involved in cell signalling, receptor-mediated membrane ruffling, phagocytosis and motility. CapG is the only gelsolin-related actin binding protein that localizes constitutively to both nucleus and cytoplasm. Structurally related proteins like severin and fragmin are cytoplasmic because they contain a nuclear export sequence that is absent in CapG. Increased CapG expression has been reported in some cancers but a causal role for CapG in tumour development, including invasion and metastasis, has not been explored. We show that moderate expression of green fluorescent protein-tagged CapG (CapG-EGFP) in epithelial cells induces invasion into collagen type I and precultured chick heart fragments. Nuclear export sequence-tagged CapG-EGFP fails to induce invasion, whereas point mutations in the nuclear export sequence permitting nuclear re-entry restore cellular invasion. Nuclear import of CapG is energy-dependent and requires the cytosolic receptor importin β but not importin α. Nuclear CapG does not possess intrinsic transactivation activity but suppresses VP16 transactivation of a luciferase reporter gene in a dose-dependent manner. Furthermore, invasion requires signalling through the Ras-phosphoinositide 3-kinase pathway and Cdc42 or RhoA, but not Rac1. We show for the first time active nuclear import of an actin binding protein, and our findings point to a role for nuclear CapG in eliciting invasion, possibly through interfering with the cellular transcription machinery.
Genomic and proteomic aspects of cancer (or tumour) development (including invasion and metastasis) have received much attention (Hahn and Weinberg, 2002; Friedl and Wolf, 2003; Mareel and Leroy, 2003) but some putative invasion pathways have remained unexplored. There is evidence that actin binding proteins (ABPs) play a role in cancer invasion and metastasis. For example, expression of gelsolin, an F-actin severing protein, is downregulated in carcinomas of the breast (Asch et al., 1999), bladder (Tanaka et al., 1999) and prostate (Lee et al., 1999), but upregulated in non-small cell lung cancers (Shieh et al., 1999) and urothelial carcinomas (Rao et al., 2002). Overexpression of gelsolin induces invasion of cells in collagen type I and precultured chick heart fragments (De Corte et al., 2002). Upregulation of thymosin G-actin sequestering proteins correlates with tumorigenicity of malignant mouse fibrosarcoma cells (Kobayashi et al., 2002) and metastasis (Bao et al., 1996; Weterman et al., 1993; Bao et al., 1998). In tumour cell lines, cofilin is detected predominantly in the nucleus, in contrast to normal cells, suggesting that changes other than the expression level of ABPs are associated with cell transformation (Samstag et al., 1996; Nebl et al., 1996). Although a causal relationship between modulation of ABP expression and several aspects of tumour (or cancer) development has been established, the underpinning molecular circuitry remains, however, largely unexplored.
This study focuses on CapG (Mbh1 or gCap39), an omnipresent nuclear-cytoplasmic gelsolin-related actin filament capping protein (Prendergast and Ziff, 1991; Onoda and Yin, 1993; Witke et al., 2001) devoid of F-actin severing or actin nucleating activities, but regulated by calcium and phosphoinositides such as phosphatidylinositol-4,5-bisphosphate (PIP2) (Johnston et al., 1990; Yu et al., 1990). We have previously shown that CapG nuclear localization is due, in part, to lack of a nuclear export sequence that is present in structurally related ABPs (Van Impe et al., 2003). Increased expression of CapG is associated with ocular melanoma (Van Ginkel et al., 1998) and elevated expression of the CapG transcript (eightfold) was also detected in glioblastomas as compared with normal brain tissue (Lal et al., 1999). Melanoma and glioblastoma are two examples of aggressive cancers killing the patients by metastasis and local invasion, respectively.
Here we studied CapG nuclear import and, using well-established in vitro models, we investigated invasion of CapG-overexpressing cells into collagen type I. Collagen type I is the major element of the stromal extracellular matrix in which successfully invasive cells need to penetrate and survive. We were able to discriminate between cytoplasmic and nuclear CapG in promoting cell invasion. This is the first study detailing, in part, the molecular pathway employed by an ABP to enter the nucleus.
Materials and Methods
Texas Red-X-phalloidin, Alexa 488 and Alexa 594-conjugated goat anti-mouse IgG and Alexa 488-conjugated goat anti-rabbit were from Molecular Probes (Eugene, OR). Anti-importin-β (mAb3E9) was obtained from Affinity BioReagents (Golden, CO); other antibodies were used as described by De Corte et al. (De Corte et al., 2002). Restriction enzymes were from New England Biolabs (Beverly, MA). Pfu turbo polymerase and Quikchange site-directed mutagenesis kit were from Stratagene (La Jolla, CA). 4,6-diamidino-2-phenylindole (DAPI) was obtained from Sigma (St Louis, USA). C3 transferase was kindly provided by C. Gespach (INSERM U482, France). Wortmannin and PD98059 were from Calbiochem (Darmstadt, Germany). LY294002 was obtained from Biomol Research Laboratories (Devon, UK). Aprotinin was obtained from ICN Pharmaceuticals (Costa Mesa, CA); galardin and a structural analogue were kindly provided by Kjeld Danø (The Finsen Laboratory, Copenhagen, Denmark). Recombinant TIMP-1 was obtained from R&D systems (Abingdon, UK). Cyclic decapeptide (CTTHWGFTLC) was synthesized as described by Wielockx et al. (Wielockx et al., 2001).
MDCK-AZ (Vleminckx et al., 1991), HeLa and HEK293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Invitrogen), supplemented with 10% fetal bovine serum (FBS, Invitrogen), 0.05% L-glutamine, 100 U/ml penicillin and 0.1 mg/ml streptomycin. HCT8/E-11 cells were grown in RPMI 1640 medium with 10% fetal bovine serum (Invitrogen, Merelbeke, Belgium), 1 mM sodium pyruvate, 100 μg/ml streptomycin, 100 IU/ml penicillin, and transiently transfected using LipofectaMINE reagent (Invitrogen). MDCK-AZ cells were stably transfected with pEGFP-CapG using LipofectAMINE reagent (Invitrogen), and selected in medium with 1 mg/ml G418 (Duchefa, The Netherlands) or 4 μg/ml blasticidin S. HEK293T cells were transfected using calcium phosphate.
SV40 nuclear localization signal peptide synthesis and coupling
The SV40 large T antigen peptide corresponding to the nuclear localization signal (NLS) (CGGGPKKKRKVED) was synthesized using Fmoc (fluoren-9-ylmethoxycarbonyl) chemistry with an Applied Biosystems 431A automated peptide synthesizer (Foster City, CA) and purified by reverse-phase HPLC. The peptide's mass was confirmed by MALDI-mass spectrometry. The NLS peptide was coupled onto TRITC-BSA (Adam et al., 1990) (Sigma). TRITC-BSA and SV40 NLS-TRITC-BSA were dialysed against 10 mM HEPES, pH 7.3, 110 mM potassium acetate.
Nuclear import assays
Permeabilization and import reactions were carried out as described previously (Adam et al., 1990). Briefly, MDCK-AZ cells were grown on glass coverslips for 48 hours. Cells were washed in transport buffer (20 mM Hepes pH 7.3, 110 mM potassium acetate, 5 mM sodium acetate, 2 mM magnesium acetate, 2 mM DTT, 0.5 mM EGTA and protease inhibitors) and then treated with digitonin (20 μg/ml) for 5 minutes on ice. After permeabilization, the cells were washed twice with transport buffer and inverted on top of 50 μl of import reaction containing 200 nM of BSA-TRITC or NLS-BSA-TRITC or 4.5 μM Myc-CapG-V5-His6, untreated rabbit reticulocyte lysate (Promega) as a source of cytosol and an energy-regenerating mixture (1 mM ATP, 5 mM creatine phosphate and 20 U/ml creatine phosphokinase) and incubated at 30°C for 30 minutes. Cells were washed twice in transport-buffer, fixed in 3% paraformaldehyde for 20 minutes at room temperature, and permeabilized with 0.1% Triton X-100 in PBS. Cells were then blocked in PBS/1% BSA for 10 minutes at room temperature and incubated with primary antibody for 1 hour at 37°C [anti-V5 antibody (Invitrogen) or affinity-purified anti-CapG antibody]. Cells were washed in PBS, incubated with secondary antibody (Alexa 488-conjugated goat anti-mouse, Alexa 488-conjugated goat anti-rabbit) and stained with DAPI for 30 minutes at room temperature. Following immunostaining, samples were analysed using a Zeiss Axioplan I epifluorescence microscope (×40 objective) equipped with an Axiocam cooled CCD camera and processed using KS100 software (Zeiss, Göttingen, Germany). For importin-β inhibition experiments, digitonin-permeabilized cells were incubated for 30 minutes at room temperature in transport buffer containing 50 μg/ml anti-importin-β, coverslips were then rinsed extensively in transport buffer and subsequently incubated with the transport mix. Importin-β was visualized using Alexa-594 conjugated goat-anti-mouse antibody. GTP-γ-S was used at a concentration of 2 mM in transport mix and cells were incubated with apyrase (100 U/ml) for 5 minutes at 30°C before the transport reaction. For competition experiments, 6.2 mM SV40 NLS peptide was used.
Invasion and cellular aggregation assays
Gels were prepared in a six-well plate from a collagen type I solution (Upstate Biotechnology, Lake Placid, NY) and incubated for at least 1 hour at 37°C to allow gelification. Cells (1 ×105) were incubated on top of the gels for 24 hours at 37°C. Cells inside the gel were scored with a phase contrast microscope controlled by a computer program (Bracke et al., 2001a). Invasive and superficial cells were counted in 12 fields of 0.157 mm2. The invasion index is the percentage of cells invading the gel over the total number of cells counted. DHD-FIB rat colon myofibroblasts (Dimanche-Boitrel et al., 1994) were used as a positive control. Experiments were performed in triplicate. Mean values and standard deviations were calculated. Chick heart invasion assays (Bracke et al., 2001b) were performed as described. Briefly, a drop of suspension cells was confronted with 9-day-old precultured embryonic chick-heart fragments. The tissue cultures were incubated for 7 days on an enriched semi-solid agar medium, fixed in Bouin-Hollande's solution and embedded in paraffin. Consecutive sections were stained with haematoxilin-eosin or immunohistochemically with antiserum against MDCK cells. Cell aggregation assays were done as described previously (Boterberg et al., 2001).
The expression plasmids pGal4 and pGal4-VP16 were described previously (Schmitz and Baeuerle, 1991; Schmitz et al., 1995). The p(Gal4)250hu.IL6P-luc+ reporter plasmid was described in Plaisance et al. (Plaisance et al., 1997). The pUT651 construct expressing β-galactosidase was obtained from Eurogentec. The cDNA encoding CapG was cloned into pGal4, downstream of the Gal4-DBD, and the coding sequence for the NLS of the SV40 large T-antigen was subsequently cloned upstream of Gal4-CapG (pNLS-Gal4-CapG).
Transient transfection and luciferase assay
Exponentially growing HEK393T cells were seeded in six-well plates at a density of 3×105 cells/well. After 24 hours cells were transfected using calcium phosphate with 2 μg of luciferase reporter plasmid, 50 ng pUT651, and a constant amount (450 ng) of pGal4 vector (either empty or combined with different amounts of pGal4-VP16 or pNLS-Gal4-CapG). Forty-eight hours after transfection, cells were resuspended and seeded in a black 96-well plate (Costar). After another 24 hours, luciferase activity from triplicate samples was determined. The cells were lysed (lysis buffer: 25 mM Tris, pH 7.8; 2 mM EDTA; 2 mM DTT; 10% glycerol; 1% Triton X-100) for 15 minutes at room temperature and 35 μl of luciferase substrate buffer (20 mM Tricine; 1.07 mM (MgCO3)4Mg(OH)2.5H2O; 2.67 mM MgSO4.7H2O; 0.1 mM EDTA; 33.3 mM DTT; 270 μM coenzyme A; 470 μM luciferin; 530 μM ATP) was added per 50 μl lysate. Light emission was measured for 5 seconds in a TopCount Chemiluminescence Counter (Packard). Luciferase activity, expressed in arbitrary light units, was corrected for protein concentration by normalization to β-galactosidase. β-Galactosidase protein levels were quantified with a chemiluminescent reporter assay Galacto-Star kit (TROPIX, Bedford, MA) and a Topcount Chemiluminescence Counter (Packard).
Lysis of cells and western blot analysis was performed as described (De Corte et al., 2002). Proteins were visualized by enhanced chemiluminescence detection (ECL, Amersham Pharmacia Biotech). Point mutations were created with the Quikchange site directed mutagenesis kit using pEGFP-N1-CapG (Van Impe et al., 2003) as a template. All constructs were verified by sequencing. The PAK1 constructs PAK1 83-149 and PAK1 K299R were kindly provided by J. Chernoff (Fox Chase Cancer Center, Philadelphia, Pennsylvania). Other cDNA constructs that were used are described by De Corte et al. (De Corte et al., 2002) and Van Impe et al. (Van Impe et al., 2003). Myc-CapG-V5-His6 was purified as described by Van Impe et al. (Van Impe et al., 2003). Affinity purification of anti-CapG antibodies was performed according to standard procedures.
Moderate expression of CapG-EGFP induces invasion into collagen type I and organ cultures
Epifluorescence microscopy of MDCK-AZ cells expressing CapG-EGFP in a stable manner (MDCK-CapG-EGFP, or MDCK-C, cells) showed cytoplasmic and nuclear staining of the fusion protein (Fig. 1A), similar to the endogenous protein (Fig. 1B). MDCK-C cells, MDCK-AZ and control EGFP-transfected MDCK-AZ cells (MDCK-E) showed an F-actin pattern with distinct stress fibres and cortical actin filaments (Fig. 1A',B',D'), indicating that CapG expression does not promote drastic changes in the organization of the cellular microfilament system. Staining for endogenous CapG in HeLa cells also revealed a nuclear and cytoplasmic localization of the actin modulating protein (Fig. 1E). Expression levels in seven independent MDCK-C clones varied between 0.77 (MDCK-C cl1) and 1.34 (MDCK-C clC) (Fig. 1F), relative to the endogenous protein. In contrast to MDCK-AZ or MDCK-E cells (Fig. 1G, 1-3), MDCK-C clones invaded a collagen type I matrix (Fig. 1G, 5-11). These results indicate that moderate elevation of CapG triggers cellular invasion. MDCK-C cells were confronted also with embryonic chick heart fragments. MDCK-E cells formed a distinct epithelial cell layer surrounding the embryonic chick heart tissue (Fig. 2A,B,arrowheads), whereas MDCK-C cells invaded the heart tissue fragments (Fig. 2C,D, arrowheads).
We examined the effect of CapG-EGFP expression on E-cadherin-dependent cell aggregation by slow and fast aggregation assays. In the absence of DECMA-1 antibody, MDCK-E cells aggregated as dense clusters (Fig. 2E), but these cells showed reduced cell aggregation in the presence of DECMA-1 (Fig. 2F). Interestingly, aggregating MDCK-C cells were phenotypically similar to DECMA-1-treated MDCK-E cells (Fig. 2G,H). These findings may suggest that E-cadherin-mediated cell-cell adhesion is affected in MDCK-C cells, but co-immunoprecipitation experiments revealed an intact E-cadherin/β-catenin/α-catenin complex in these cells, indicating that the cell-cell adhesion complex is not disturbed (data not shown). Changes in the particle diameter of MDCK-C cells was investigated in more detail by fast aggregation assays. In the absence of DECMA-1 antibody, the particle diameter of MDCK-E cells was larger than the particle diameter of MDCK-C cells (Fig. 2I, compare -○-with -▪-). MDCK-E cells treated with DECMA-1 antibody (Fig. 2I -▵-) showed a drastic reduction in particle diameter to a size comparable to MDCK-C cells.
Nuclear import of CapG requires energy and importin-β
ABPs such as zyxin (Nix and Beckerle, 1997), cofilin (Bamburg, 1999) and a myosin I isoform (Pestic-Dragovich et al., 2000) shuttle between nucleus and cytoplasm. However, no information is available about the molecular mechanism of ABP nuclear import at large, except that several representatives either contain a nuclear localization signal (NLS) or a nuclear export signal (NES). In this respect CapG nuclear import represents a conundrum because CapG does not harbour a typical NLS although its PIP2-binding region contains several positively charged amino acids in close proximity (see below). We employed in vitro nuclear import assays (Adam et al., 1990) using Myc-CapG-V5-His6-tagged recombinant protein as substrate (CapG-V5) to study CapG nuclear import. Recombinant Myc-CapG-V5-His6 displays F-actin capping activity (Van Impe et al., 2003). Western blot analysis showed that purified recombinant Myc-CapG-V5-His6 was not contaminated by degradation products (Fig. 3A). CapG did not accumulate in the nucleus of permeabilized MDCK-AZ cells (in the absence or presence of ATP and GTP) when rabbit reticulocyte lysate (RRL) was excluded from the import mixture (Fig. 3B,C). Similarly, addition of CapG-V5 or the SV40 large T-antigen NLS coupled to TRITC-BSA (NLS-TRITC-BSA) did not result in nuclear enrichment in the presence of GTP-γ-S and RRL (Fig. 3D,E). In the presence of RRL and energy, CapG-V5 or NLS-TRITC-BSA accumulated in the nucleus, indicating that cytosolic factor(s) are required for CapG nuclear import (Fig. 3F,G). No nuclear staining was observed with monoclonal anti-V5 antibody alone (Fig. 3H); TRITC-BSA did not accumulate in the nucleus either (Fig. 3I). Interestingly, anti-importin-β antibodies prevented nuclear import of CapG (Fig. 3J1-J3) as well as of NLS-TRITC-BSA (Fig. 3K). Finally, the SV40 large T-antigen peptide inhibited the nuclear import of NLS-TRITC-BSA (not shown), but had no effect on the nuclear accumulation of CapG (Fig. 3L), indicating that the cargo is actively transported to the nucleus, and that transport requires importin-β, but not importin-α. These data are summarized in Table 1.
Nuclear CapG promotes invasion
Nuclear entry of CapG is due, in part, to the lack of a Rev-like nuclear export sequence present in structurally related ABPs like severin and fragminP (Fig. 4A, upper panel). Tagging the NES of fragminP (FrgP) onto CapG prevents nuclear accumulation of CapG (Van Impe et al., 2003). To discriminate between nuclear and cytoplasmic CapG in promoting cellular invasion, CapG-EGFP and FrgPM1-L27(NES)-CapG-EGFP were transfected in HEK293T cells and analysed for collagen invasion. Strikingly, expression of NES-CapG-EGFP did not promote collagen invasion (Fig. 4B, lane 3). Mutation of Ile9 or Ile14 to Ala has no effect on export activity (Van Impe et al., 2003) and this is mirrored by the failure of these fusion proteins to induce invasion (Fig. 4B, lanes 4-5). By contrast, mutation of either Leu17, Leu21, or Leu27 to Ala, all of which prevent export, each induced invasion (Fig. 4B, lanes 6-8), further suggesting that nuclear CapG is responsible for eliciting the invasive phenotype.
CapG represses transcription activation of Gal4-VP16
The ability of CapG to modulate transcriptional activity was investigated by reporter assays. For this purpose we used a transactivation assay based on the transcriptional activation of a Gal4-responsive luciferase reporter gene in the presence of fusion proteins with the Gal4-DNA binding domain (Gal4-DBD). As a positive control we used the transactivation domain of the viral protein VP16 fused to Gal4-DBD. VP16 is a prototypic acidic herpes simplex transcriptional activator, which is often used to evaluate the possible effect of a transcription regulator. In addition, actin is a component of the mammalian BAF complex (termed Swi/Snf in Saccharomyces cerevisiae), a multiprotein complex involved in modulating chromosome condensation (Olave et al., 2002). The BAF complex binds PIP2 micelles (Rando et al., 2002), which stabilizes its association with actin filaments, and is involved in transcription activation of VP16. As CapG is a nuclear, PIP2 regulated ABP, we analysed the modulatory role of CapG on the transactivation activity of VP16.
Analysis of the subcellular distribution of Gal4-DBD-CapG following transfection in Hek293T revealed that the fusion protein was present in the cytoplasm (data not shown). Insertion of the NLS of the SV40 large T-antigen upstream of Gal4-DBD-CapG promoted cytoplasmic as well as nuclear staining (data not shown). Co-transfections in HEK293T cells of Gal4-DBD, Gal4-DBD-VP16 or NLS-Gal4-DBD-CapG with the reporter construct revealed that luciferase values of NLS-Gal4-DBD-CapG were comparable to that of the empty Gal4 vector (data not shown), suggesting that CapG does not harbour intrinsic transactivation activity. We also studied the potential of CapG to modulate transactivation activity of VP16. Co-transfection of NLS-Gal4-DBD-CapG in combination with Gal4-VP16 showed that increasing amounts of NLS-Gal4-DBD-CapG decreased the luciferase activity of VP16 in a dose-dependent manner (Fig. 5). These results suggest that CapG expression can modulate the transactivation activity of VP16 and acts as a repressor of transcriptional activity.
Ras and phosphoinositide 3-kinase participate in the signal transduction pathway leading to CapG-induced collagen invasion
Co-transfection experiments with dominant-negative signalling polypeptides revealed that gelsolin-induced invasion requires Ras, phosphoinositide 3-kinase (PI3K) and Rac activity (De Corte et al., 2002). To unravel the signal transduction pathway accompanying CapG invasion we transiently expressed CapG-EGFP in HEK293T cells (293C-EGFP) or HCT-8/E11 colon carcinoma cells (Fig. 6A,B, lane 2). Co-expression with dominant-negative Rac1N17, RhoAN19 or Cdc42N17 resulted in inhibition of CapG-triggered invasion by RhoAN19 and Cdc42N17 (Fig. 6A,B, lanes 4-5), but not by Rac1N17 (Fig. 6A,B, lane 3). C3 transferase, an inhibitor of RhoA, also blocked MDCK-C invasion (data not shown). Thus, invasion mediated by CapG involves signalling activities distinct from gelsolin (De Corte et al., 2002). Further analysis revealed that collagen invasion by CapG requires Ras activity because RasGAP inhibits invasion (Fig. 6C, lane 5). Three independent lines of evidence also support a role for PI3K in CapG invasion. First, expression of the p85 subunit of PI3K lacking the regulatory SH2 domain (p85ΔiSH2N) (Hara et al., 1994), inhibited invasion (Fig. 6C, lane 6). Second, inhibitors of PI3K such as wortmannin and LY294002 also strongly inhibited invasion (data not shown). Third, co-expression with RasV12 domain effector mutants (Rodriguez-Viciana et al., 1997) showed that RasV12S35 or RasV12E38, which can activate Raf-1 but not PI3K, blocked CapG invasion (Fig. 6C, lanes 8,9). By contrast, RasV12C40, which allows activation of PI3K but not of Raf-1, did not block invasion (Fig. 6C, lane 10). CapG-induced invasion was also counteracted by dominant-negative mitogen-activated protein kinase kinase (MEK) (Fig. 6C, lane 7) or PD98059 (data not shown), a pharmacological inhibitor of MEK, indicating that invasion mediated by CapG requires signalling through the MAPK pathway. A schematic overview is shown in Fig. 7.
We investigated the role of gelatinases in CapG invasion by using inhibitors of matrix metalloproteinases (MMP). Invasion of MDCK-C cells was blocked in the presence of aprotinin, galardin, tissue inhibitor of MMP 1 (TIMP-1), or a cyclic decapeptide that inhibits MMP-2 and MMP-9 (Koivunen et al., 1999) (Fig. 6F), indicating that proteolytic degradation of the collagen gel is required for these cells in order to migrate through the polymeric matrix.
In recent years, it has become apparent that many ABPs, including canonical cytoplasmic members, temporarily or constitutively reside in the nucleus. Representatives include zyxin (Nix and Beckerle, 1997), supervillin (Ting et al., 2002), cofilin (Samstag et al., 1996), a myosin I isoform (Pestic-Dragovich et al., 2000), protein 4.1 (Krauss et al., 2003), profilin (Skare et al., 2003; Stuven et al., 2003) (reviewed by Shumaker et al., 2003), myopodin (Weins et al., 2001; Van Impe et al., 2003) and gelsolin (Nishimura et al., 2003). The molecular weight of these nuclear ABPs ranges from very small, such as the 13 kDa actin sequestering protein profilin, to relatively large, like the 100 kDa C-terminal fragment of filamin A (Loy et al., 2003) or the 100 kDa N-terminal fragment of supervillin (Wulfkuhle et al., 1999). This broad range in molecular weight suggests that nuclear transport can occur in a passive manner, for some small members (<40 kDa), whereas other representatives require active, energy-dependent, transport. At present however, the transport mechanism of nuclear ABPs in general remains elusive. Nuclear ABPs perform various functions, ranging from control of nucleus assembly to modulation of transcriptional regulation. Interestingly, several nuclear ABPs, including supervillin (Ting et al., 2002), gelsolin (Nishimura et al., 2003) and filamin A fragment (Loy et al., 2003) modulate transcriptional activity of the androgen receptor, a male steroid hormone receptor frequently mutated in prostate cancer.
While it is important to understand the role of ABPs in the nucleus, it is also important to recognize how nuclear ABPs migrate in and out of the nucleus and how their nucleo-cytoplasmic transport is regulated. The occurrence of a (predicted) NLS does not always correspond with a functional shuttling sequence. For instance, supervillin (Pestonjamasp et al., 1997) has been shown to contain functional NLSs (Wulfkuhle et al., 1999), but two predicted nuclear localization sequences in the actin bundling protein myopodin do not appear to alter the subcellular localization when mutated (Weins et al., 2001). In this study we report that shuttling of a constitutive nuclear-cytoplasmic ABP, CapG, requires active (energy-dependent) transport. Although the molecular weight of wild-type CapG (40 kDa) is still in agreement with potential passive transport, this is much less probable for CapG-EGFP (67 kDa). The requirement of importin-β for CapG nuclear import is relatively unique because, usually, NLS-containing cargos interact with importin-α, which is subsequently bound by importin-β. This complex then docks onto the nuclear pore complex. Beta-catenin nuclear entry occurs through yet another pathway, namely through direct interaction with the nuclear pore complex, and docking is not inhibited by an NLS peptide but by importins/karyopherins, indicating that they both interact with common nuclear pore components (Fagotto et al., 1998). In CapG we did not identify a conspicuous monopartite or bipartite NLS, although the 137-KKLYQVKGKK-146 PIP2-binding region harbours several basic residues in close proximity. Mutation of these amino acids, either separately or in combination, did not result in changes in subcellular localization of CapG, arguing against a role for this sequence as an NLS (data not shown). CapG in vitro import assays in the presence of the SV40 large T antigen NLS peptide GGGPKKKRKVED, at concentrations that inhibited NLS-TRITC-BSA import, were ineffective towards CapG, implying that CapG import probably does not require importin-α. Although it is uncertain at present which region in CapG constitutes a nuclear targeting sequence, it is probable that the N-terminal region is involved in protein-protein interactions controlling the subcellular localization of CapG. Evidence in support of this hypothesis is based on the observation that N-terminal fusion of CapG to the DNA binding domain of Gal4 results in cytoplasmic localization of the fusion protein, and the fusion protein is prevented from accumulating in the nucleus. This fusion protein localized to the nucleus only after insertion of the SV40 large T antigen nuclear localization sequence. By contrast, C-terminal fusion of fluorescent proteins to CapG does not impair nuclear transport. Importantly, tagging short sequences to the N-terminus of CapG (i.e. a myc tag), does not affect localization of the fusion protein nor its F-actin capping properties (Van Impe et al., 2003).
Collagen invasion of cells, induced by the overexpression of CapG, is linked to a signal transduction pathway that is distinct from that of gelsolin (De Corte et al., 2002), suggesting that the activity of these ABPs is controlled via separate upstream regulators. Both gelsolin and CapG promote invasion into chick heart fragments and both affect cell aggregation, as evidenced by slow and fast aggregation assays. However, although moderately elevated nuclear CapG levels elicit an invasive phenotype in a Cdc42- and RhoA-dependent manner, gelsolin-induced invasion requires the small GTP-ase Rac. Whether or not CapG is also implicated in the invasion of human cancers, including melanoma and glioblastoma, in vivo remains to be elucidated. Interestingly, inhibition of CapG nuclear accumulation through coupling with the fragmin NES inhibited invasion, whereas point mutants in the nuclear export sequence of fragmin that abolish nuclear export activity (allowing nuclear re-entry) restored collagen invasion. Thus, nuclear CapG appears to be responsible for the invasive phenotype, and this might be linked with its activity as a modulator of transcription. The observation that potential transcriptional regulation by CapG is controlled by upstream effectors such as MAP kinase or PI3K is not unprecedented. Indeed, other transcription regulators that are modulated by MAP kinases and PI3K have been documented in the literature. For example, CREB, a leucine zipper transcription factor, is phosphorylated on activation of MAP kinase signalling cascades. This allows CREB to recruit the transcriptional adapter CBP (CREB binding protein) and promote transcription activation of Fos (Shaywitz and Greenberg, 1999). PI3K/Akt is known to play a critical role in prostate cancer cell growth and survival. Recent studies have shown that the effect of PI3K/Akt in prostate cells is mediated through androgen signalling. The PI3K/Akt signal induces phosphorylation and inactivation of GSK3-β, resulting in increased nuclear levels of β-catenin. Consequently, increased β-catenin elevates androgen receptor activity to stimulate prostate cell growth and survival (Sharma et al., 2002).
Nuclear CapG interaction partners have not yet been identified but, on the basis of transcriptional inhibition of Gal4-VP16-induced luciferase activity by CapG, we propose that CapG can negatively regulate transcription activity. Nuclear actin has been implicated in chromatin remodelling and gene activation via a family of nuclear proteins called Brg or hBrm associated factors (BAFs) (Olave et al., 2002). Actin has also been reported to associate with members of the A/B-group of heterogeneous nuclear ribonucleoproteins (hnRNP) (Percipalle et al., 2002). Binding of CapG to actin in the BAF complex could regulate these complexes, hence influencing transcription. An alternative hypothesis is that changes in the cytoplasmic actin status are involved in nuclear ABP shuttling, similar to dynamic G-actin–F-actin turnover that regulates serum response factor (SRF) activity (Miralles et al., 2003).
In conclusion, we have shown that the F-actin capping protein CapG triggers invasion of cells into collagen type I and chick heart fragments. Intriguingly, its major biological function most probably does not contribute to invasion, for two reasons: because overexpression of cytoplasmic CapG (`NES-CapG-EGFP') does not induce collagen invasion and because nuclear actin is thought to exist only as short oligomers, or even monomers (Pederson and Aebi, 2002). Collectively, our data therefore point to a possible role for CapG as a tumour promoter, and suggest that CapG could modulate invasive properties of cells during tumorigenesis.
We thank W. Vanden Berghe (Department of Molecular Biology, Ghent University, Ghent, Belgium) for help with transactivation assays, and Mark Goethals for peptide synthesis. This work was supported by the `Belgische Federatie tegen Kanker', the Concerted Actions Program of Ghent University (GOA), the Fund for Scientific Research-Flanders (FWO-Vlaanderen) and by Fortis Bank Verzekeringen. KVI is supported by the concerted research actions of Ghent University (GOA). VDC is a Postdoctoral Fellow of the Fund for Scientific Research-Flanders (Belgium) (F.W.O.).
↵* Authors contributed equally to this work
- Accepted July 13, 2004.
- © The Company of Biologists Limited 2004