The flightless I protein colocalizes with actin- and microtubule-based structures in motile Swiss 3T3 fibroblasts: evidence for the involvement of PI 3-kinase and Ras-related small GTPases

The Drosophila melanogaster flightless I (fliI) gene encodes a 1,256 amino acid protein with a predicted molecular mass of 143,672 Da (Campbell et al., 1993). It has highly conserved homologues in Caenorhabditis elegans (49% amino acid sequence identity to Drosophila), mouse (95% identity to human) (Campbell et al., 2000) and human (58% identity to Drosophila) (Campbell et al., 1993; Campbell et al., 1997). Northern analysis of human FLII mRNA expression patterns showed the highest level in skeletal muscle, but the gene was expressed at varying levels in all tissues examined (Campbell et al., 1997). Viable mutations of the D. melanogaster fliI locus cause ultrastructural defects in the indirect flight muscles (Deak et al., 1982; Miklos and de Couet, 1990), with frayed and disorganized myofibrils and Z bands that appear abnormal or absent. Null alleles of fliI are homozygous lethal (Perrimon et al., 1989). In contrast to normal embryonic development (Miller and Kiehart, 1995), mutant embryos have an irregular cytoskeleton and arrangement of nuclei at the cell cortex, followed by poorly coordinated membrane invaginations (Kajava et al., 1995; Straub et al., 1996). These developmental events require an intact cytoskeleton. The phenotypes of moderate and severe fliI mutants indicate a possible role for the fliI protein in regulating actin cytoskeletal rearrangements.

A 380 amino acid leucine-rich repeat region (LRR) (Kobe and Deisenhofer, 1995a), characterizes the N-terminal domain of the fliI protein. This motif is thought to play a role in specific protein-protein interactions, that in some cases are responsible for a role in signal transduction (Kobe and Deisenhofer, 1995b). This amphipathic beta sheet/alpha helix motif (Kobe and Deisenhofer, 1994; Kobe and Deisenhofer, 1995b) was first observed in the α₂-glycoprotein (Takahashi et al., 1985) and has since been identified in hormone receptors (McFarland et al., 1989), certain enzymes (Tan et al., 1990), tyrosine kinase receptors (Martin-Zanca et al., 1989) and molecules responsible for cell adhesion (Hashimoto et al., 1988). A variety of LRR functions have been characterized (Kobe and Deisenhofer, 1994), including the LRR region in yeast adenylate cyclase which is known to interact with Ras (Suzuki et al., 1990).

C-terminal to the LRR region lie two large duplicated domains each characterized by three tandemly repeated sequences of between 125 and 150 amino acids (Campbell et al., 1993; Way and Weeds, 1988). This region shows significant homology to members of the gelsolin-family of actin-binding proteins (ABPs) (Campbell et al., 1997; Kwiatkowski et al., 1986; Hartwig and Kwiatkowski, 1991). The gelsolin family members are ABPs that can bind monomeric actin subunits promoting polymerization (Janmey et al., 1985; Yin, 1986; Kwiatkowski et al., 1989). Members of this family can cap (Lind et al., 1987) and/or sever actin filaments (Kwiatkowski et al., 1989; Arora and McCulloch, 1996; Chaponnier et al., 1986). ABPs are important for actin cytoskeletal rearrangements and are regulated by micromolar concentrations of calcium (Janmey et al., 1985; Pope et al., 1995; Yin et al., 1988; Young et al., 1994) and phosphoinositide binding (Janmey and Stossel, 1989; Lassing and Lindberg, 1985; Lu et al., 1996).

The cytoskeletal network plays important roles in the maintenance of cell shape, the transport and anchoring of cellular components involved in cellular adhesion, and migration. While significant progress has been made in identifying the biochemical signals that are involved in regulating the cytoskeleton, the relationship between them and the impact on cellular functions are not well defined. Recent evidence has shown that members of the Ras-related GTPase family are responsible for the regulation of a variety of actin-based structures (Allen et al., 1997; Nobes and Hall, 1995; Tapon and Hall, 1997; Rodriguez-Viciana et al., 1997). In addition to this, signaling pathways involved in the regulation of proteins that influence cytoskeletal dynamics are increasingly seen to involve phosphoinositide 3-kinase (PI 3-kinase) (Derman et al., 1997; Rodriguez-Viciana et al., 1997; Johanson et al., 1999; Hill et al., 2000).

The work presented here describes the development of flightless I-specific antibodies and the identification and examination of the subcellular distribution of the mouse flightless I protein (Fliih) protein in quiescent and serum-stimulated Swiss 3T3 fibroblasts. We also examine potential interacting proteins and regulatory molecules for Fliih. The results of our experiments imply a role for flightless I in the regulation of cytoskeletal rearrangements involved in cytokinesis and cell migration.

The cDNA sequences of flightless I and homologues have been determined (Campbell et al., 1993) (D. melanogaster GenBank accession no. U01182, C. elegans GenBank U01183 and Homo sapiens GenBank U01184; U80184). Peptide L corresponding to a sequence (CKLEHLSVSHN, amino acids 57-66) within the LRR (L) domain of human FLII was synthesized (Biomolecular Resource Facility, JCSMR), for subsequent production of anti-peptide antibodies. An antibody against peptide G within the gelsolin-like domain (CSHFKRKFIIH, amino acids 1032-1041) has been previously described (Davy et al., 2000). Peptide L was conjugated to keyhole limpet hemocyanin (Sigma), using the protocol described by Goldsmith and coworkers (Goldsmith et al., 1987). The N-terminal cysteine was added to the peptide for coupling purposes. A New Zealand White rabbit was injected subcutaneously and serum was collected following clot retraction and stored at −70°C. The rabbit was boosted at regular intervals and further samples were collected.

The FliL antibodies were affinity-purified using 1 ml, NHS-activated, HiTrap Affinity columns (Pharmacia, Sweden) coupled with peptide L, according to the manufacturer’s instructions. For peptide blocking, 50 μg of peptide was incubated with 50 μl of the antibody for one hour at room temperature. This preparation was then used in the same manner as antibodies that were not blocked.

The previously reported human FLII cDNA (phfli2) (Campbell et al., 1993) was almost full length, but was missing a part of the ATG initiation codon. A human EST clone, Image Consortium clone 135082 (GenBank accession no. R33910), was previously identified as containing the ATG initiation codon and 37 bp of 5′ untranslated sequence (Campbell et al., 1997). However, this EST clone contained an insert of only 800 bp. Clone 135082 plasmid DNA was digested with XhoI and end repaired with T4 DNA polymerase. The DNA was digested with BspEI and the 300 bp XhoI-BspEI fragment was gel purified. Plasmid phfli2 was digested with EcoRV and BspEI, dephosphorylated with calf intestinal alkaline phosphatase, and the 6.8 kb fragment was gel purified. These two fragments were ligated together to yield phfli2FL. The insert in phfli2FL was sequenced from both ends by dye primer sequencing (ABI Prism) using the –21 M13 forward and reverse primers, confirming the expected structure. The 5′ sequence of the phfli2FL cDNA up to and across the BspEI site was identical to the previously determined cDNA and genomic sequences (Campbell et al., 1993; Campbell et al., 1997).

The transcription/translation of phfli2FL plasmid DNA was achieved using the TNT T7 Quick Coupled Translation/Transcription System (Promega). [³⁵S]methionine was used to produce a radioactive form of protein that can, when appropriately prepared, be visualized using autoradiography. A 5μl aliquot from each reaction was added to SDS-PAGE sample buffer (300 mM Tris, 50 mM DTT, 15% glycerol, 2% SDS, bromophenol blue) plus 15 mg/ml dithiothreitol (DTT), or 1 ml of lysis buffer containing inhibitors (see Cell Harvest methods). Samples containing lysis buffer and 5 μl of transcription/translation reaction were heated for 30 minutes at 70°C and precleared by incubation with Protein A-Sepharose beads only for one hour at 4°C. FliL or FliG antibodies (1:50) were added and samples were incubated overnight at 4°C on a rotating mixer. Protein A-Sepharose beads were hydrated with Tris buffered saline (TBS: 50 mM Tris-HCl, pH 7.4, 0.2 M NaCl) and 5 mg (w/v) was added to each sample. The samples were mixed for 1 hour at 4°C. The beads were pelleted by centrifugation for 45 seconds at 17,400 g, and the supernatant discarded. The Protein A-Sepharose-antibody-antigen pellets were rinsed 3 times with 1 ml of TBS. SDS-PAGE sample buffer was added to each pellet and samples were analyzed by SDS-PAGE. The SDS-PAGE gel was dried prior to autoradiography.

Bacterial expression of flightless I

Two primers were prepared that enabled amplification of the human FLII coding region for insertion into an E. coli expression vector. Primer HDC128 (5′-AGA GCG GCC ATA TGG AGG CCA CCG GGG TGC TGC CG-3′) was designed to reconstruct the translation initiation codon and add an overlapping NdeI site. Primer HDC129 (5′-GCC AGC ATC GAT TAG GCC AGG GCC TTG CAG AAG GCG-3′) was designed to match the 3′ end of the coding region and add a flanking ClaI site. Each primer (10 pmol) was used in a 50 μl reaction with 1 ng phfli2 cDNA (Campbell et al., 1993) as template in a Perkin-Elmer 9600 PCR machine using the Expand Long Template PCR system (Boehringer Mannheim) mixture of Taq and Pwo DNA polymerases in Expand buffer 1 (50 mM Tris-HCl, pH 9.2 (25°C), 14 mM (NH₄)₂SO₄ 1.75 mM MgCl₂). TaqStart antibody (Clontech) was used to obtain a hot start. After 1 minute at 94°C, 25 cycles of 94°C (30 seconds), and 68°C (3 minutes), were carried out. The reaction product was digested with NdeI and ClaI, purified by gel electrophoresis and the 3.8 kb fragment was ligated between the NdeI and ClaI sites of pETMCSI. This vector is a derivative of the pET T7 promoter vectors (Studier et al., 1990) and was a generous gift from Dr Nick Dixon, Research School of Chemistry, ANU. For the expression, 6 independent isolates of this construct were transformed into BL21(DE3)/pLysS. Transformants were then grown at 37°C to an OD₅₉₅ of 0.5 and induced by addition of 1 mM isopropyl-beta-D-thiogalactopyranoside (IPTG). Cultures were incubated at 37°C for 2 hours after induction and cells harvested by centrifugation. Control cultures were treated identically except that IPTG was not added. The number of cells was normalized by OD measurements taken when cells were harvested. Samples were analyzed by SDS-PAGE and western transfer.

Swiss 3T3 fibroblasts (Commonwealth Serum Laboratories, Australia) were cultured and harvested as described previously (Crouch and Simson, 1997).

Lysates were fractionated as described previously by Franze-Fernandez and Pogo (Franze-Fernandez and Pogo, 1971) into the nuclear (N), cytoskeletal (C) or membrane/cytosol (M) fractions. The resulting fractions were used for immunoprecipitation procedures.

Appropriate antibodies were added to cell lysates and incubated overnight on a rotating mixer. Protein A-Sepharose beads were added and samples mixed for 1 hour. Beads were pelleted by centrifugation for 45 seconds at 17,400 g, and the supernatant discarded. Protein A-Sepharose-antibody-antigen pellets were rinsed 3 times with 1 ml Tris buffered saline (50 mM Tris-HCl, pH 7.4, 0.2 M NaCl).

Proteins were electrophoresed on 7% SDS-polyacrylamide gels and western transferred to nitrocellulose membrane (Schleicher and Schuell, BA-85, Germany). All samples for electrophoresis required the addition of SDS-PAGE sample buffer plus DTT, and were heated for 5 minutes at 100°C prior to loading. The blot was incubated with the appropriate antibody overnight at 4°C. Detection of antigens after the primary antibody incubation was performed using horseradish peroxidase-labeled secondary antibodies followed by enhanced chemiluminescent detection (ECL; Amersham, UK).

The human FLII coding region was fused in frame to the C terminus of the coding region for the enhanced green fluorescent protein (EGFP) as follows. The plasmid phfli2FL containing the reconstructed full-length human FLII cDNA was digested with NcoI and end-repaired. Following BamHI digestion, the 4.1 kb fragment containing the FLII coding region was purified by gel electrophoresis. The vector pEGFP-C1 (Clontech) was digested with SalI, end-repaired, digested with BamHI, purified and ligated with the FLII fragment to yield pEGFP-C1-FLII. The structure of the construct was verified by restriction analysis and end-sequencing using the EGFP-C sequencing primer (Clontech).

Swiss 3T3 fibroblasts were seeded in a 12-well plate, each well containing a coverslip, at 2×10³ cells/ml, maintained in 10% FCS/DMEM containing antibiotics (penicillin/streptomycin), and allowed to attach overnight. The medium was then changed to 10% FCS/DMEM without antibiotics and allowed to equilibrate for 4 hours. The pEGFP-C1-FLII construct was incubated in the presence of FuGene (Boehringer Mannheim), according to the manufacturer’s instructions, and this mix was then added to the cultured cells. The empty pEGFP-C1 vector was used as the control. Cells were allowed to incubate without a change in medium for a further 48 hours. Cells were subsequently fixed with 2% paraformaldehyde, mounted onto slides and EGFP-fluorescence visualized with the FITC wavelength (Ex: 494 nm; Em: 520 nm) using confocal microscopy.

Cells were seeded into 12-well plates, as above. Fibroblasts were seeded at 7.5×10³ cells/ml, maintained in 10% FCS/DMEM and allowed to attach overnight. The medium was then replaced with 0% FCS/DMEM and the incubation continued for a further 24 hours. Cells were subsequently activated with 10% FCS with or without preincubation with test agents, LY294002 (25 μM, Biomol, USA) or rapamycin (100 nM, ICN, USA), as indicated. Cells were fixed with 2% paraformaldehyde for 15 minutes at room temperature, and then washed 5 times with cold PBS. Cells were permeabilized with PBS/1.0% BSA/0.1% SDS for 15 minutes followed by blocking for 1 hour with PBS/1.0% BSA. The appropriate antibody (1:100) was diluted in PBS/1.0% BSA and added to each well and incubation continued at 4°C overnight. Wells were rinsed 5 times with cold PBS then incubated with FITC-labeled secondary antibodies (1:100) (Jackson Immunoresearch, USA) in PBS/1.0% BSA for 1 hour at room temperature. Fibroblasts were also double-labeled in some instances with two antibodies. In these cases, Texas Red-labeled secondary antibody (Jackson Immunoresearch, USA) was used in conjunction with the FITC-labeled secondary antibody. There was no crossover of fluorescence between the FITC (Ex: 494 nm; Em: 520 nm) and Texas Red channels (Ex: 596 nm; Em: 618 nm). Cellular fluorescence was visualized and recorded with a Leica (Germany) confocal microscope. Commercial antibodies used were against gelsolin and Ras (Transduction Laboratories), RhoA, Cdc42, and Rac1 (Santa Cruz). Staining of filamentous actin was achieved with the use of Texas Red Phalloidin (Molecular Probes). Phalloidin was dried under nitrogen and redissolved in PBS/1.0% BSA. Phalloidin (25 U/ml) was added to each well and the same protocol as for secondary antibodies was followed.

A plasmid containing the complete coding region of human FLII was used for in vitro-coupled transcription/translation reactions containing the T7 RNA polymerase for the transcription step. The construct, named phfli2FL, was verified by restriction analysis and end-sequencing. In phfli2FL, the T7 RNA polymerase promoter in the pBluescript SK-vector is located upstream from the 5′ end of the FLII cDNA. One labeled protein present following completion of the reaction had a molecular mass of 145 kDa (Fig. 1A, lane1), and was specifically recognized by the anti-peptide antibodies, FliL (Fig. 1A, lane 2) and FliG (not shown). The immunoprecipitation of this protein was blocked when antibodies were incubated with peptide L, prior to use (Fig. 1A, lane 3). The 145 kDa protein was absent in samples that contained the product of a luciferase-encoding control cDNA (Fig. 1A, lane 4) when probed with the FliL antibody (Fig. 1A, lane 5), or samples that did not contain cDNA (not shown). Thus, the FliL and FliG antibodies were specific for the 145 kDa protein encoded by FLII cDNA.

E. coli cells transformed with a T7 promoter expression vector containing FLII cDNA were grown in liquid culture and induced with IPTG for 2 hours, or left uninduced, prior to lysis. Cell lysates were prepared from six independent E. coli clones and analyzed by 7% SDS-PAGE. Western blots were immunoblotted (IB) with the FliL antibody (Fig. 1B) or FliG (not shown) antibodies. A 145 kDa protein was identified in samples that had been induced with IPTG (Fig. 1B, lanes 2, 4 and 6), but not in lysates from cells that were not induced (Fig. 1B, lanes 1, 3 and 5). The 145 kDa protein was also absent when antibodies were blocked with the appropriate peptide (not shown). The lower molecular mass bands may be degradation products or are possibly truncated forms of FLII that can occur when translation is disrupted during the reaction. The identity of the nonspecific 110 kDa protein is unknown, but this was present in both induced and non-induced cells and was recognized non-specifically by the secondary antibody.

The FliG antibody identified a 145 kDa protein in whole cell lysate from Swiss 3T3 fibroblasts (Fliih) (Fig. 1C, lanes 1 and 2) that was specifically blocked with peptide G (Fig. 1C, lanes 3 and 4). Similar results were seen when the FliL antibody was used (not shown). The increase in the amount of protein in stimulated cells (Fig. 1C, lane 2) reflects protein synthesis following the long stimulation time. FliL or FliG antibodies were also used to immunoprecipitate (IP) protein from Swiss 3T3 mouse fibroblasts. The FliL antibody immunoprecipitated a protein with a molecular mass of approximately 145 kDa that was recognized following western transfer by both the FliL (Fig. 2) and FliG antibodies (not shown). A protein of identical molecular mass, immunoprecipitated by FliG antibodies, was recognized by both antibodies following western transfer (not shown).

In quiescent cells, the FliL/FliG-immunoreactive protein was predominantly found in the nuclear (Fig. 2, lane 1) and membrane/cytosol (Fig. 2, lane 3) fractions, with a lesser amount associated with the cytoskeleton (Fig. 2, lane 5). Stimulation with 10% FCS induced a marked decrease in the amount of protein accumulated in the nucleus (Fig. 2, lane 2), and a concurrent increase in Fliih associated with the cytoskeleton (Fig. 2, lane 6). There was little change in the amount of Fliih in the membrane/cytosol fraction (Fig. 2, lane 4) following stimulation. The evidence leads us to conclude that the rabbit anti-peptide FliL and FliG antibodies specifically recognize the translated human phfli2FL product, the human FLII protein expressed in bacterial cells and the endogenous murine Fliih protein in cultured Swiss 3T3 fibroblasts. In the latter, Fliih appeared to undergo a stimulus-induced subcellular translocation from the nucleus to the cytoskeleton.

Swiss 3T3 fibroblasts were fixed with 2% paraformaldehyde and subsequently incubated with affinity-purified FliL (Fig. 3) or FliG (not shown) antibodies followed by FITC-labeled secondary anti-rabbit antibodies. Quiescent Swiss 3T3 fibroblasts were flat and circular in shape with no evidence of a polarized morphology (Fig. 3A) or structures normally associated with motile fibroblasts. Unstimulated cells accumulated Fliih predominantly in the nuclear (N) and perinuclear regions. Fliih could be seen extending in a filamentous arrangement from the nuclear/perinuclear region into the cytoplasm of the cell. We consistently noted pockets of increased fluorescence within the nucleus in regions reminiscent of nucleoli. The significance of this nuclear localization is unknown.

To examine the localization of Fliih following stimulation, quiescent cells were activated with 10% FCS, prior to immunohistochemistry. Fliih remained localized to the nuclear/perinuclear region in cell populations that were confluent, and therefore non-migratory (not shown). In migratory cells, Fliih localized to discrete bands posterior to the plasma membrane (Fig. 3B). We believe these structures to be actin arcs (see Fig. 4G). Fliih also localized to the actin-rich plasma membrane underlying the leading edge (Fig. 3B). We also found Fliih localized to membrane ruffles, actin-rich structures that form on the dorsal surface of the lamellipodia in motile cells (Fig. 3C). Similar results were seen when the FliG antibody (not shown) was used.

In view of the likely actin-binding properties of the Fliih protein (Liu and Yin, 1998), fibroblasts were simultaneously labeled with Texas Red-phalloidin (Fig. 4A,D,G) and FliL (Fig. 4B,E,H) to examine colocalization sites between the polymerized actin-based network and Fliih. Phalloidin labeling of unstimulated cells revealed short actin filaments (Fig. 4A) in a disorganized fashion throughout the cytoplasm of the cell. Fliih predominantly localized to the nuclear/perinuclear region (N) with evidence of a filamentous network extending into the cytoplasm (Fig. 4B). The Fliih-based network did not extensively colocalize with phalloidin-defined filaments in unstimulated cells, a result evident when images were merged using Adobe Photoshop (Fig. 4C), although there was some evidence for colocalization on particular filaments (compare Fig. 4A-C).

Following stimulation with serum, phalloidin-defined stress fibers can be seen extending across the cell (Fig. 4D). The Fliih protein translocated from the nuclear/perinuclear region toward the periphery of the cell (Fig. 4E). The Fliih-defined filaments were perpendicular to the actin-based stress fiber network. Despite some apparent colocalization between Fliih with actin-based networks, represented by the population of yellow filaments seen when the images were merged (Fig. 4F), we believe that the Fliih-based network is not coincident with stress fibers. We do note, however, discrete points of apparent colocalization between Fliih-defined filaments and the terminal ends of stress fibers (Fig. 4F). The actin-rich leading edge (Fig. 4G), the actin arc and membrane ruffles are structures normally associated with migration. Phalloidin-loaded cells simultaneously labeled with the FliL antibody (Fig. 4H) revealed discrete regions of colocalization (Fig. 4I) at actin arcs, membrane ruffles and the leading edge of migrating cells. FliG antibodies gave identical staining patterns to FliL antibodies for all conditions, and the specificities of both the antibodies was further established by showing that preincubation of the antibodies with their cognate peptides blocked the signal (not shown).

We have constructed the vector pEGFP-C1-FLII by fusing the human FLII cDNA in frame to the C terminus of the coding region for the enhanced green fluorescent protein (EGFP). The construct prepared for this work contains the entire coding region for human FLII including the ATG initiation codon. This is fused in frame to the 3′ end of the coding region for EGFP via a 15 amino acid region encoded by the polylinker of pEGFP-C1. The human cytomegalovirus immediate early promoter drives expression. The plasmid also contains the neo gene under the control of the SV40 early promoter. We used this vector to transiently transfect Swiss 3T3 fibroblasts and subsequently examined the localization of EGFP-tagged FLII protein. Fibroblasts were transiently transfected with the EGFP-C1 vector (not shown) or with the EGFP-C1-FLII construct (Fig. 5). Cells were labeled with Texas Red phalloidin (Fig. 5A and D) and the localization of the GFP-C1-FLII protein (Fig. 5B and E) was examined using the FITC channel. We show a phalloidin-labeled cell that is not transfected and subsequently fails to emit an FITC-sensitive signal (Fig. 5A and B, arrow). In transfected cells, we see EGFP-C1-FLII localizing to the peripheral membrane (Fig. 5B) and membrane ruffles (Fig. 5E) of migrating cells. When images were merged, yellow regions reveal sites of colocalization between the GFP-C1-FLII protein and actin-rich regions of motile cells (Fig. 5C and F). GFP alone did not localize with F-actin, but rather showed a mainly nuclear localization (not shown).

The failure of Fliih-defined filaments to colocalize with actin filaments led us to examine the possibility that translocation from the nuclear/perinuclear region is mediated by the β-tubulin-based network. We simultaneously labeled activated fibroblasts with β-tubulin (Fig. 6A) and FliL (Fig. 6B) antibodies and examined the resulting fluorescent images for regions of colocalization (Fig. 6C). It appeared that Fliih colocalized with a population of microtubules (yellow filaments) extending into a polarized region of the cell following serum stimulation.

We also examined other β-tubulin-based structures including the mitotic spindle, a structure involved in chromosome separation and cell division. In Fig. 6D, the β-tubulin-based mitotic spindle is evident in a dividing cell. Two intensely fluorescent regions at opposing poles of the mitotic spindle could be seen when the localization of Fliih was visualized (Fig. 6E). When images were merged (Fig. 6F), these appeared to be centrosomes at the spindle poles. Fliih was also found concentrated at the midbody of telophase cells. Fliih was absent from the actin-rich contractile ring (Fig. 6G) but was enriched on the microtubular midbody alone (Fig. 6H). Higher magnification of this midbody region more clearly shows the localization of Fliih (Fig. 6I) to these β-tubulin-based structures (not shown) in Swiss 3T3 fibroblasts. At this stage, most of the Fliih was again localized to the nuclear region (Fig. 6H).

Quiescent and stimulated cells were fixed then incubated overnight with an antibody directed to gelsolin and subsequently incubated with secondary Texas-Red labeled anti-mouse antibodies. Gelsolin localized to short, fragmented actin filaments in quiescent fibroblasts (Fig. 7A). The association of gelsolin with fragmented F-actin decreased after activation with 10% FCS, yielding a more punctate distribution throughout the cytoplasm (Fig. 7B).

Similar results were seen when fibroblasts were stimulated with thrombin (not shown). Swiss 3T3 fibroblasts were simultaneously labeled with Fliih and Cdc42 antibodies and subsequently incubated with Texas-Red and FITC-secondary antibodies. Cdc42 localized predominantly to the perinuclear region in serum-starved cells and translocated in a filamentous arrangement when cells were stimulated for 1-4 hours with 10% FCS (not shown). In cells activated with 10% FCS for 16-18 hours, Cdc42 (Fig. 8A,D) and Fliih (Fig. 8B,E) antibodies colocalized (Fig. 8C,F) at the actin arc and the leading edge of motile cells.

Fibroblasts were also double-labeled with Ras (Fig. 8G) and Fliih (Fig. 8H) antibodies, and subsequent confocal images merged (Fig. 8I). Ras was seen in a diffuse punctate distribution throughout the cytoplasm of the cell, with regions of increased fluorescence localized at actin arcs and the leading edge of migrating cells. When the images were merged it was clear that Fliih and Ras colocalized at an actin arc and the leading edge. Sites of Fliih enrichment within the nucleus were clearly evident in Fig. 8B and Fig. 8K.

Swiss 3T3 fibroblasts were simultaneously labeled with RhoA (Fig. 8J) and Fliih (Fig. 8K) antibodies. RhoA localized to the perinuclear region in quiescent cells and did not colocalize with filamentous Fliih in the early stages of activation (not shown). Following long-term serum stimulation, RhoA predominantly localized in a punctate distribution throughout the cytoplasm of the cell. However, regions of enrichment were evident on actin arcs and the leading edge of migrating cells. When images were merged (Fig. 8L), RhoA and Fliih colocalized on the actin arc and the actin-rich leading edge of motile fibroblasts. RhoA was also seen localized to membrane ruffles associated with lamellipodia (not shown). In contrast, RhoB and Rac1 showed no evidence of colocalization with Fliih-defined filaments, on actin arcs or at the leading edge of migrating cells (not shown) following similar serum activation times.

Fibroblasts were pretreated with rapamycin or LY294002 for 30 minutes prior to activation with 10% FCS. Cells were labeled with the FliL antibody and subsequently exposed to secondary antibodies and fluorescence visualized using confocal microscopy. The translocation of Fliih to the leading edge was not inhibited by rapamycin (Fig. 9A), but LY294002 markedly inhibited the translocation of Fliih to the leading edge of migrating cells (Fig. 9B). We noted an almost complete absence of actin arcs in cells treated with rapamycin or LY294002, and the development of lengthy tails in both rapamycin- and LY294002-treated cells.

The characterization of the Drosophila melanogaster fliI protein, and conserved homologues, suggests that this protein may play a role in linking the cytoskeleton to intracellular proteins involved in signal transduction (Liu and Yin, 1998; Goshima et al., 1999; Campbell et al., 1993; Claudianos and Campbell, 1995). To investigate the role of flightless I, we have produced two anti-peptide antibodies that recognize defined regions of the human FLII and mouse Fliih sequences. FliL and FliG (Davy et al., 2000) antibodies recognize a 145 kDa protein as measured by western blotting and immunoprecipitation methods. Fliih antibodies specifically react with a 145 kDa FLII protein synthesized in vitro from cDNA or prepared by bacterial expression in E. coli. A protein of identical size is also specifically detected in mouse cell extracts. All of the evidence strongly indicates that this is the Fliih protein. In support of this, the 145 kDa immunoreactive protein is reduced by 50% in liver extracts from heterozygous Fliih mutant mice generated by gene targeting (H. Campbell et al., unpublished results). The development of flightless I-specific antibodies provides the opportunity to characterize the intracellular response to different cellular conditions and to investigate the localization of Fliih in Swiss 3T3 fibroblasts.

In quiescent Swiss 3T3 fibroblasts, Fliih predominantly localizes to the nuclear/perinuclear region, as is the case for some other actin-binding proteins (Pope et al., 1998; Wulfkuhle et al., 1999; Matsuzaki et al., 1988). The nuclear staining is punctuated by dense foci of Fliih protein, which may represent the accumulation of Fliih within nucleoli (Shaw and Jordan, 1995; Wulfkuhle et al., 1999; Matsuzaki et al., 1988). Proteins larger than 45 kDa require a suitable nuclear localization signal (Garcia-Bustos et al., 1991; Newmeyer and Forbes, 1988; Nigg et al., 1991) to target them to the nucleus and in some cases a nuclear export signal (Dingwall and Laskey, 1992; Goldfarb, 1991) is required to exit the nucleus. Examination of the human and mouse flightless I sequences reveal putative nuclear localization (e.g. ¹⁰³⁵KRKFIIHRG-KRK¹⁰⁴⁶) and export (¹⁵⁰LTDLLYLDL¹⁵⁸) signals that are highly conserved between species for all known sequences. Fliih can be induced to translocate out of the nucleus when cells are stimulated, although it remains unclear whether the putative flightless I nuclear localization and export signals are functional.

The flightless I protein features both an actin-binding and leucine-rich-repeat domain (Claudianos and Campbell, 1995; Campbell et al., 1993). This has led to the concept that flightless I may be an intermediate protein directly linking the cytoskeleton to signaling proteins such as Ras (Claudianos and Campbell, 1995). In support of this hypothesis, it has been shown that flightless I is able to bind actin in vitro (Liu and Yin, 1998), and others have found that the C. elegans form of expressed protein interacts with Ras both in vitro and in vivo in the yeast system (Goshima et al., 1999). We have demonstrated the colocalization of the flightless I protein with actin in Drosophila and mouse embryos using an immunohistochemical approach (Davy et al., 2000). The present study has identified Fliih localizing to actin-based structures associated with motility, such as actin arcs (Heath and Holifield, 1993; Soranno and Bell, 1982) and membrane ruffles (Cheresh et al., 1999). Fliih also localizes to the actin-rich cortical layer at the leading edge of migrating cells. This localization was confirmed when cells were transfected with pEGFP-C1-FLII.

The leading edge of polarized, motile cells is a functionally distinct region, the most conspicuous feature being lamellipodia (Small, 1994). Polymerization of growing actin filaments (Schafer et al., 1998; Cooper and Schafer, 2000; Mikhailov and Gundersen, 1998; Mitchison and Cramer, 1996) creates a propulsive force that pushes forward lamellae of motile cells (Mallavarapu, 1999). Subsequent rearrangements of the plasma membrane enable the formation of new cell-substrate contacts (Nabi, 1999) at the leading edge necessary for traction (Cheresh et al., 1999; Drubin and Nelson, 1996). Like gelsolin, flightless I is able to sever filaments, at least in the case of the C. elegans protein (Goshima et al., 1999), and this may account functionally for the localization of Fliih to cytoskeletal structures. Whereas gelsolin links to fragmented actin filaments in unstimulated cells and is displaced by growth factor stimulation, Fliih is induced to migrate to highly specific cytoskeletal structures within lamellae under the same conditions. This suggests that despite sequence homology, the actin-binding and-severing activities of flightless I and other gelsolin-family members target different actin-based structures in the cell. In summary, Fliih localizes to cytoskeletal-based structures associated with migrating cells, is critically positioned to play a role in cytoskeletal regulation and may do so by binding to and severing actin filaments.

Microtubules are believed to lend structural support and have a role in the directed movement of many cell types (Conrad et al., 1989). Another less appreciated role is their function as a physical anchor and long-range transport substrate for key mediators of protein expression (Knowles et al., 1996; Bassell and Singer, 1997; Bassell et al., 1994). After translocation from the nucleus, Fliih localizes to actin-based structures but does not appear to translocate via the actin-based network. It is possible that the movement of Fliih may relate to the translocation of factors via microtubules to the actin cytoskeleton. There is precedent for this hypothesis, the best described examples being the localization of β-actin mRNA to the cell periphery of motile cells (Kislauskis et al., 1993; Latham et al., 1994) and the localization of proteins required for myofibrillar repair (Russell and Dix, 1992) and genesis (Morris and Fulton, 1994). Studies have also suggested that the presence of microtubules in developing indirect flight muscles in Drosophila melanogaster may provide factors for the formation of myofibrils (Fernandes et al., 1991; Reedy and Beall, 1993). Since myofibrillar structure is severely disrupted in flightless I mutants a role for flightless I in the development of these structures is possible. Additionally, the recent identification of flightless I-binding partners that have homology to transcription factors (Liu and Yin, 1998; Fong and de Couet, 1999) or may be involved in binding dsRNA (Wilson et al., 1998), supports the suggestion that Fliih may be part of a ribonucleoprotein complex responsible for the delivery of factors via microtubules, and the subsequent attachment to actin-based structures.

We also find that flightless I localizes to β-tubulin-based structures known to be involved in cell division and cytokinesis. Fliih localizes to centrosomes, structures that radiate a microtubular network forming the mitotic spindle which is responsible for the regulation of chromosome separation during cell division (Field et al., 1999) and discrete accumulations within the midbody, the final remnant that connects daughter cells (Glotzer, 1997; Mandato et al., 2000). The Fliih protein is distributed throughout the cytoplasm during early stages of cell separation, but is clearly localized to the nucleus when cytokinesis is nearing completion. The presence of Fliih at these dynamic sites supports the view that it may be required for cytokinesis. Such a role could explain the embryonic lethality observed with homozygous Fliih knockout embryos (H. Campbell et al., unpublished results).

The phenotype of Drosophila melanogaster fliI null mutants (Straub et al., 1996) supports the suggestion that Fliih plays a role in cytokinesis. Cellularization is a specialized form of cytokinesis (Miller and Kiehart, 1995; Loncar and Singer, 1995; Miller, 1995) involving microtubule-(Warn and Warn, 1986; Foe and Alberts, 1983) and actin-(Warn and McGrath, 1983; Miller and Kiehart, 1995; Warn and Robert-Nicoud, 1990) based structures for successful completion. We have shown that flightless I localizes to cellularization structures in D. melanogaster syncytial embryos (Davy et al., 2000), which lends support to the view that Fliih may also be involved in cytokinetic processes in Swiss 3T3 fibroblasts.

Cytoskeletal rearrangements that result in the formation of specialized migratory structures are known to be regulated, in part, by members of the small GTPase family, including Ras (Bar Sagi and Feramisco, 1986), RhoA (Ridley and Hall, 1992) and Cdc42 (Nobes and Hall, 1995). The involvement of different GTPases in cytoskeletal regulation has also been described in other cell types (Ziman et al., 1993; Eaton et al., 1995; Chen et al., 1996; Faix et al., 1998; Johnson, 1999). A hierarchical relationship is believed to exist between the small GTPases in the signaling control of cytoskeletal behavior in Swiss 3T3 fibroblasts (Ridley and Hall, 1992), although recent evidence indicates a more complex situation (Lim et al., 1996; Aspenström, 1999). We found that Rac1 and RhoB were diffusely distributed predominantly throughout the cytoplasm of Swiss 3T3 fibroblasts and did not colocalize with Fliih. When cells were stimulated with 10% FCS, Ras, Cdc42 and RhoA were induced to colocalize with Fliih on actin arcs and the leading edge of lamellae. There is evidence of GTPase activity modulating the behavior of proteins homologous to flightless I via a direct interaction (Azuma et al., 1998), however, the data emerging concerning the flightless I protein is less clear. Although Goshima et al. (Goshima et al., 1999) show that an expressed form of C. elegans flightless I is able to bind Ras, as predicted (Claudianos and Campbell, 1995), Liu and Yin have failed to show binding between flightless I and Ras or Ras-related GTPases using the two-hybrid system (Liu and Yin, 1998). Classical GTPase cascades include Rac (Ma et al., 1998a), a protein that does not appear to colocalize with Fliih. The failure to observe colocalization between flightless I and Rac may be due to an interaction occurring at time points that were not examined in this study. Alternatively, any PI 3-kinase-mediated regulation of Fliih may be via a Rac-independent pathway (Lim et al., 1996; Ma et al., 1998b). It is unclear as to whether the colocalization between Fliih and GTPase proteins is representative of a direct interaction, although this remains a significant possibility. The effect of these signaling molecules on Fliih need not be a direct one, and indeed these proteins may lie upstream or downstream and link to Fliih via intermediary proteins. In support of a common functional role, microinjection of dominant negative Rho or constitutively active Cdc42 into Drosophila melanogaster embryos results in similar effects to those seen in flightless I null mutants (Crawford et al., 1998). This includes disruption of actomyosin furrow canal formation during Drosophila cellularization and embryonic lethality prior to gastrulation. This evidence, together with the colocalization between Fliih and GTPase proteins described here, suggests that flightless I, Ras, RhoA and/or Cdc42 may be functionally linked in the control of cytoskeletal rearrangements.

Phosphoinositides, particularly phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5-triphosphate (Singh et al., 1996; Yin et al., 1988), can regulate gelsolin and other ABPs (Yin et al., 1981; Young et al., 1994; Janmey and Stossel, 1987). Cell locomotion and changes in cell shape are generally accompanied by synthesis and hydrolysis of inositol phospholipids and a rapid, but transient increase in the intracellular calcium levels (Janmey, 1994). Fliih localization to the leading edge was absent when PI 3-kinase was inhibited by pretreatment with LY294002. Our results also show that exposure to LY294002 inhibits the translocation of Fliih into the cytoplasm following stimulation. The involvement of PI 3-kinase-regulated pathways mediating nuclear localization and export has previously been described (Klingenberg et al., 2000; Belkowski et al., 1999). GTPase proteins can lie upstream or downstream of phosphoinositide activity (Rodriguez-Viciana et al., 1994; Reif et al., 1996), and the inhibition of Fliih translocation may be due to a direct inhibition of PI 3-kinase or GTPase activity, or may be an indirect effect due to the disruption of actin filaments (Johanson et al., 1999). These results suggest that Fliih may be a direct or indirect biochemical target of phosphorylated phosphoinositides or GTPase proteins, and we plan to investigate any direct physical link or regulation of Fliih function by PI 3-kinase-phosphorylated products.

Work that describes the involvement of p70 S6 kinase in the regulation of actin cytoskeletal rearrangements (Crouch, 1997; Berven et al., 1999) led to speculation about the effect rapamycin may have on the flightless I protein in Swiss 3T3 fibroblasts. Treatment with this inhibitor did not appear to affect the translocation of Fliih to the leading edge and therefore Fliih does not appear to play a role in rapamycin-sensitive cytoskeletal rearrangements. It should be noted that cells treated with rapamycin and LY294002 failed to form actin arcs and appeared to form elongated tails, a phenomenon previously described (Berven et al., 1999).

In summary, we have identified a 145 kDa protein in Swiss 3T3 fibroblasts and shown that Fliih localizes to a variety of actin- and tubulinbased structures. Fliih colocalizes with Ras, RhoA and Cdc42 implicating them in a common functional pathway, and this anterior localization is inhibited by LY294002. Future work will involve investigating the functional significance of this inhibition, the colocalization between Fliih and GTPase proteins and any role this may play in the transduction of a Fliih-mediated signal. We will also investigate the potential role of PI 3-kinase in Fliih-mediated regulation of cytoskeletal rearrangements. The actin severing activity and ability to colocalize and/or interact with important signal transduction molecules highlights the possibility of an important role for flightless I in cell migration and proliferation.

We thank Professors Ian Young and Ian Hendry for helpful discussions throughout the course of this work, and Dr Klaus Matthaei for advice concerning the construction of the EGFP-C1-FLII construct. We thank Katharina Heydon for advice on the use of the confocal microscope, Dr Nick Dixon for the generous gift of the pET vector and valuable advice, and Dr Peter Milburn and Cameron McCrae of the ANU Biomolecular Resource Facility for running the sequencers. We also thank Professor David Jans for helpful discussions about nuclear localization and import sequences.