Research ArticleCell Biology

The Tyrosine Kinase Fer Is a Downstream Target of the PLD-PA Pathway that Regulates Cell Migration

See allHide authors and affiliations

Science Signaling  08 Sep 2009:
Vol. 2, Issue 87, pp. ra52
DOI: 10.1126/scisignal.2000393


Phosphatidic acid (PA), which can be produced by phospholipase D (PLD), is involved in various signaling events, such as cell proliferation, survival, and migration. However, the molecular mechanisms that link PA to cell migration are largely unknown. Here, we show that PA binds to the tyrosine kinase Fer and enhances its ability to phosphorylate cortactin, a protein that promotes actin polymerization. We found that a previously unknown lipid-binding module in Fer adjacent to the F-BAR [Fes-Cdc42–interacting protein 4 (CIP4) homology (FCH) and bin-amphiphysin-Rvs] domain mediated PA binding. We refer to this lipid-binding domain as the FX (F-BAR extension) domain. Overexpression of Fer enhanced lamellipodia formation and cell migration in a manner dependent on PLD activity and the PA-FX interaction. Thus, the PLD-PA pathway promotes cell migration through Fer-induced enhancement of actin polymerization.


Cell migration, which is driven by robust actin polymerization, is initiated and maintained by coordinated interactions between the plasma membrane and actin regulators (1); such interactions are partially mediated by conserved domain species that specifically bind to acidic phospholipids, including phosphatidic acid (PA) (2, 3). Among the various cellular functions controlled by PA, cell proliferation and survival signaling are mediated by specific interactions with mammalian Son of Sevenless (mSOS) (4) and mammalian target of rapamycin (mTOR) (5), respectively. Moreover, the phospholipase D (PLD)–PA pathway plays an indispensable role in cell migration (68). In addition, dedicator of cytokinesis 2 (DOCK2), a guanine nucleotide exchange factor (GEF) for Rac, binds PA through a polybasic region to mediate neutrophil chemotaxis (9). However, other protein targets that interact with this signaling phospholipid to promote cell migration are largely unknown.

The cytosolic tyrosine kinase Fer and its hematopoietic homolog Fes contain an F-BAR domain (10) (fig. S1A), a lipid-binding and membrane-bending module conserved among various proteins involved in actin regulation (1113). Fer plays an essential role in cell migration through tyrosine phosphorylation of cortactin, an activator of Arp (actin-related protein) 2-3–mediated actin polymerization (14, 15). Here, we show that Fer is involved in a PLD-PA signaling pathway that mediates cell migration.


Fer tyrosine kinase contains a PA-binding domain

To gain insight into the role of Fer in membrane dynamics and the actin cytoskeleton, we first assessed the lipid-binding activity of Fer by means of a cosedimentation assay with brain liposomes. The F-BAR domain of Fer exhibited surprisingly weak membrane-binding ability, whereas the same region of Cdc42-interacting protein 4 (CIP4), another F-BAR protein that can tubulate membrane (12), binds to liposomes with high affinity (Fig. 1A). Furthermore, full membrane-binding activity was observed when the construct was extended to the C-terminal side, and the extended region (referred to as the FX domain for F-BAR extension) (fig. S1A) retained affinity for the membrane (Fig. 1A). A secondary-structure prediction generated by the Jpred program (16) indicated that the FX domain is composed of two long helices and two strands (fig. S1B). The predicted C-terminal strands are essential for membrane binding because binding was abolished by deletion of these strands (FX ΔC), as well as simultaneous substitutions of positively charged residues in this region (Arg417, Arg425, and His426) with alanine residues (FX AAA) (Fig. 1B).

Fig. 1

The identification of the PA-binding domain in Fer. (A) Each recombinant protein (5 to 10 μg) was incubated with (+) or without (−) brain liposomes (1 mg/ml), and then centrifuged to separate the unbound (supernatant, s) and bound fractions (pellet, p). (B) Wild-type FX domain or the FX domain with deletions (FX ΔC) or point mutations (FX AAA) were subjected to the same assay as described in (A). (C) Lipid specificity was assessed with liposomes composed of either 80% PE and 20% PC (first bar) or 70% PE, 20% PC, and 10% each of the phospholipids PE, PC, PA, PS, and PI. Error bars indicate SEM (n = 3 independent experiments).

Next, we examined the lipid specificity of the region encompassing the F-BAR and FX domains and found that this region bound preferentially to PA compared to other acidic phospholipids (Fig. 1C and fig. S2). The F-BAR–FX unit resembles the BAR-PH or PX-BAR units in APPL (adaptor protein containing PH domain, PTB domain, and leucine zipper motif) proteins and sorting nexins (SNX) that are thought to simultaneously recognize both the membrane curvature and the lipid species (17, 18) (fig. S1C). Indeed, the F-BAR–FX unit is another example of a functional unit because the preferential binding to PA is determined by the FX domain (Fig. 1C and fig. S2). We previously showed that the Fer F-BAR domain has low membrane-bending activity (13). Thus, the F-BAR domain of Fer might serve as a curvature sensor and not a curvature driver and may function in combination with the adjacent FX domain.

Fer is activated by PA

The findings that Fer preferentially binds to PA prompted us to examine whether the kinase activity of Fer is regulated by this signaling phospholipid. A full-length recombinant Fer protein bearing a C-terminal FLAG tag was obtained by purifying the expressed protein from transfected COS-7 cells with a FLAG monoclonal antibody conjugated to agarose beads. An in vitro kinase reaction was carried out with cortactin, a promoter of actin polymerization, as a substrate in the presence of brain liposomes. As the liposome concentration was increased, Fer-induced phosphorylation of cortactin increased up to 3.5-fold (Fig. 2, A and B). However, when the liposome concentration was further increased, Fer was activated to a lesser degree or even inhibited (Fig. 2, A and B). This biphasic activation is also exhibited by membrane-activated and polymerized enzymes, such as dynamin guanosine triphosphatase (GTPase) (19). The inclusion of PA in the phosphatidylethanolamine (PE)- and phosphatidylcholine (PC)-based liposomes activated wild-type Fer, but not the Fer mutant that cannot bind lipids because of alanine substitutions in the FX domain (Fer AAA) (Fig. 2C). Thus, Fer is a PA-dependent tyrosine kinase that is activated by interaction between the FX domain and the signaling phospholipid.

Fig. 2

PA-mediated activation of Fer. (A) An in vitro kinase assay was carried out with GFP-Fer-FLAG as the kinase and GST-cortactin as the substrate in the presence or absence of ATP and increasing concentrations of brain liposomes (0.00125, 0.0025, 0.005, 0.0125, 0.025, 0.05, 0.125, 0.25, 0.5, and 1 mg/ml). Phosphorylation of cortactin was evaluated by Western blotting with phosphotyrosine (pTyr) antibody (clone 4G10). Substrate abundance in each assay was verified by immunoblotting for GST. (B) Quantification of (A). Tyrosine phosphorylation was standardized by substrate amounts. Data are presented as arbitrary units. Error bars indicate SEM (n = 3 independent experiments). (C) The same assay as in (B) was performed with controlled compositions of liposomes (0.02 mg/ml) as described in Fig. 1D. The inclusion of PA significantly activated wild-type Fer (white bars) but not the FX mutant (Fer AAA; black bars). Error bars indicate SEM (n = 4 independent experiments). **P < 0.01.

Fer induces lamellipodia formation in a manner dependent on PA-FX interaction

Our in vitro experiments suggested that the Fer-cortactin axis of the actin regulatory pathway is activated by PA. Tyrosine phosphorylation of cortactin has been implicated in actin-based processes such as cell migration and cancer metastasis (20, 21). Overexpression of Fer tagged with green fluorescent protein (GFP-Fer) in COS-7 cells promoted robust lamellipodia formation, which can be recognized by the presence of arc-shaped membrane areas, as well as actin polymerization at the plasma membrane (Fig. 3A). In addition, tyrosine phosphorylation of cortactin, as well as that of Vav2, a GEF for Rac, was enhanced by overexpression of Fer (figs. S3 and S4B). Consistently, lamellipodia formation required Fer catalytic activity and was blocked by a dominant-negative form of Rac (figs. S4A and S5), indicating that Fer acts upstream of this small GTPase. Membrane curvature generation or sensing may also be involved in this process because a GFP-Fer mutant lacking the F-BAR domain (ΔF-BAR) did not promote lamellipodia formation (fig. S6). Moreover, the PA-Fer interaction was necessary because both GFP–Fer ΔFX and GFP–Fer AAA did not promote lamellipodia formation (Fig. 3, B and C). Furthermore, PLD activity was required for Fer-induced lamellipodia formation because treatment with 1-butanol, but not tert-butanol, significantly reduced the formation of lamellar membranes, with many structures resembling filopodia (relatively sharp membrane projections enriched with actin filaments) remaining in Fer-overexpressing cells (Fig. 3D).

Fig. 3

Lamellipodia formation by Fer overexpression. (A) COS-7 cells were transfected with GFP-Fer (green) and stained with rhodamine-phalloidin (red). A section of lamellipodia is magnified in the inset and shows the colocalization of Fer with the F-actin structure. (B and C) Deletion (GFP–Fer ΔFX) or mutation (GFP–Fer AAA) of the FX domain failed to induce lamellipodia formation. Error bars indicate SEM (n = 5 independent experiments). At least 50 cells were examined for each condition. **P < 0.01. (D) COS-7 cells overexpressing GFP-Fer were treated with 0.5% 1-butanol (1-BuOH) or tert-butanol (t-BuOH) for 30 min, fixed, and stained with rhodamine-phalloidin. Percentages of cells forming lamellipodia: 50.0 ± 3.8% (control), 6.3 ± 1.6% (1-BuOH), and 52.9 ± 3.3% (t-BuOH) (n = 3 independent experiments). Differences between control and 1-BuOH conditions, as well as between 1-BuOH and t-BuOH conditions, were statistically significant (P < 0.01). At least 50 cells were examined for each treatment. Scale bars, 10 μm (A, B, and D) or 2 μm [inset in (A)].

Epidermal growth factor (EGF) can promote PLD-mediated production of PA (4, 22). EGF stimulation of COS-7 cells expressing GFP-Fer recruited Fer to the plasma membrane and caused concomitant membrane extensions (fig. S7A). EGF-induced membrane recruitment of GFP-Fer was effectively blocked by 1-butanol treatment, indicating that PLD activity is required (fig. S7A). Although expression of GFP-Fer in COS-7 cells did not induce lamellipodia formation, the F-BAR–FX unit was also recruited to small membrane ruffles that may be precursors of extended lamellipodia (fig. S7B). These data suggest that Fer is recruited to the plasma membrane through the F-BAR–FX unit upon PA production.

Consistent with the findings of Sangrar et al. that reactive oxygen species are potent activators of Fer during cell migration (14), endogenous Fer was activated by hydrogen peroxide, and this activation was blocked by pretreatment with 1-butanol (fig. S8). Thus, activation of Fer in the promotion of lamellipodia formation requires PLD activity.

Fer tyrosine kinase is essential for cell migration downstream of the PLD-PA signaling pathway

Next, we investigated the relevance of Fer-induced lamellipodia formation, which requires both PA-Fer interaction and PLD activity, to cell migration regulated by PLD-PA signaling. Knockdown of Fer significantly suppressed the migration of NRK52E cells (Fig. 4, A and B, and fig. S9), consistent with previous findings obtained for knock-in MEF cells expressing a kinase-dead form of Fer (14). The suppression was alleviated by the exogenous expression of wild-type Fer, but not by the expression of Fer AAA (Fig. 4B), suggesting that the PA-Fer interaction is essential for cell migration. The use of inhibitors suggested that PA production required for cell migration is mediated by PLD and not by the activity of class I diacylglycerol kinases (fig. S10A). Consistent with previous studies (6, 7), knockdown of PLD1 or PLD2 suppressed migration with additional suppression with knockdown of both PLD1 and PLD2, indicating nonredundancy between both isoforms (fig. S10B). Intriguingly, the triple knockdown of both PLD isoforms and Fer did not produce any further inhibition of migration (fig. S10B); this suggests that Fer and PLD are present in the same branch of a signaling pathway that regulates cell migration.

Fig. 4

The Fer-PA interaction is necessary for cell migration. (A) NRK52E cells transfected with Fer siRNAs were subjected to a Transwell assay in which cells were seeded on a culture insert with small pores (8 μm) through which they were allowed to migrate for 18 hours. Migrated cells at the bottom side of the chamber were fixed and stained with rhodamine-phalloidin. Scale bar, 100 μm. (B) Knocked-down cells were transfected with GFP-Fer constructs to recover the function of Fer in cell migration; however, this was prevented by mutation of the FX domain. Error bars indicate SEM (n = 3 independent experiments). **P < 0.01. (C) Fer and PLD2 synergistically promote cell migration. Cells were transfected with the indicated constructs and subjected to the Transwell migration assay for 8 hours. PLD2 KR, a lipase-dead form with a Lys758→Arg (K758R) mutation. Error bars indicate SEM (n = 3 independent experiments). **P < 0.01.

Conversely, GFP-Fer–overexpressing cells migrated in higher numbers compared to GFP-transfected cells through a porous membrane support over an 8-hour period (Fig. 4C). Similarly, the enhancement of migration due to the overexpression of hemagglutinin-tagged PLD2 (HA-PLD2) was dependent on endogenous Fer [HA-PLD2 + Fer siRNA (small interfering RNA); Fig. 4C]. Moreover, simultaneous expression of GFP-Fer and HA-PLD2 increased the number of migrated cells, and this synergistic activation was abolished when wild-type PLD2 was replaced with a lipase-dead form of PLD2 (PLD2 KR; Fig. 4C). These results suggest that Fer acts downstream of PLD, likely the PLD2 isoform, to mediate cell migration.


Our results collectively indicate that Fer tyrosine kinase is a key player downstream of the PLD-PA pathway that promotes actin polymerization and cell migration. Tyrosine phosphorylation of cortactin by PA-activated Fer results in a phospho-cortactin gradient, with the highest concentration at the plasma membrane. This could promote actin polymerization, which in turn generates protrusive force against the membrane during cell migration (23). The mechanism underlying the generation or sensing of membrane curvature by F-BAR domains in cell migration remains unclear. Furthermore, several studies have shown that membrane-bending proteins, including BAR or F-BAR domain–containing proteins and dynamin GTPases, are involved in cell migration (2427); this indicates that curved membranes may play a role in migration. Notably, PA, which has a smaller head group compared to those of other phospholipids, has been suggested to induce negative curvature on the inner leaflet of the lipid bilayer (27, 28). Thus, PA could facilitate the recruitment of the FX domain to the edge of protruding membranes that retain highly negative curvatures. The combination of specific phospholipid distribution and membrane curvature could thus be required to achieve the dynamic complexity of cellular membranes.

Materials and Methods

Antibodies and reagents

Phosphotyrosine (clone 4G10) and actin monoclonal antibodies were purchased from Millipore Corp. (Billerica, MA). Cortactin (phospho 421) polyclonal antibody was purchased from Abcam Inc. (Cambridge, MA). GFP polyclonal antibody was from MBL (Nagoya, Japan). Cortactin, Vav2, and glutathione S-transferase (GST) polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Fer polyclonal antibodies were raised by immunizing rabbits with recombinant Fer (amino acids 1 to 445) that included both the F-BAR and the FX domains. The specific antibody was purified from the raised antisera with antigen protein coupled to HiTrap NHS-activated HP Columns (GE Healthcare Life Sciences, Piscataway, NJ). 1-Butanol was purchased from Nakalai Tesque Inc. (Kyoto, Japan) and tert-butanol was from Cambridge Isotope Laboratories Inc. (Andover, MA). The DGK inhibitor R59949 (29) was from Calbiochem (Gibbstown, NJ). All fluorescent reagents (rhodamine, Alexa Fluor 647-phalloidin, and Alexa Fluor 594-conjugated goat anti-rabbit or anti-mouse secondary antibodies) were purchased from Molecular Probes (Eugene, OR). The bovine brain lipid extract used in the preparation of the brain liposome was from Sigma (St. Louis, MO). Purified phospholipids [PI (phosphatidylinositol), PA, PS (phosphatidylserine), PE, and PC] were from Avanti Polar Lipids (Alabaster, AL). All phosphorylated phosphoinositides [PI(3)P, PI(4)P, PI(5)P, PI(3,4)P2, PI(3,5)P2, PI(4,5)P2, and PI(3,4,5)P3] were from Cell Signals Inc. (Columbus, OH).

DNA constructs

The GST and GFP fusion constructs were created by ligating the respective complementary DNA (cDNA) into pGEX6P-1 (Amersham Biosciences, Piscataway, NJ) and pEGFP-C1 (Clontech, Mountain View, CA). Full-length Fer cDNA was obtained by reverse transcription polymerase chain reaction (RT-PCR) using human spleen cDNA (Clontech) with specific primers that amplify three regions of the entire cDNA. The fragments thus obtained were ligated with Eco RV and Cla I to assemble the full-length Fer cDNA; this was then ligated into the vector at the Bam HI and Sal I sites. The FLAG tag was subsequently inserted into the 3′ Sal I site to obtain the Fer recombinant protein. The Fer F-BAR (amino acids 1 to 300), FX (amino acids 270 to 445), and F-BAR–FX domains (amino acids 1 to 445) were amplified by RT-PCR with specific primer sets. All deletion and point mutations were introduced with the PrimeSTAR MAX DNA polymerase (Takara Bio Inc., Shiga, Japan) according to the manufacturer’s instructions. Wild-type PLD2 and Lys758→Arg (K758R) cDNAs were provided by M. Frohman (State University of New York at Stony Brook, NY) (30). Vav2 with Myc tags, and the dominant (or constitutively) active (RacDA; Gly12→Val or G12V) and dominant-negative (RacDN; Thr17→Gln or T17N) forms of Rac1 were provided by Y. Takai (Kobe University, Japan). GST-cortactin was provided by S. J. Park (our laboratory).

Cell culture and transfection

COS-7 and NRK52E cells (JCRB, Osaka, Japan) were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum at 37°C in the presence of 5% CO2. Transfection was carried out with Lipofectamine LTX with PLUS reagent (Invitrogen, Carlsbad, CA), and the cells were fixed for immunofluorescence after 16 to 24 hours.

Protein purification and recombinant proteins

With the exception of the full-length Fer protein, all the recombinant proteins were obtained from bacterial expression systems, according to the manufacturer’s instructions. The GST tag of the proteins used in the liposome binding assays was always removed by means of on-bead cleavage with PreScission proteases (GE Healthcare, Piscataway, NJ), and the released protein was dialyzed in assay buffer [25 mM Hepes-NaOH (pH 7.4), 100 mM NaCl, and 5 mM EDTA]. The GFP-Fer-FLAG protein was obtained by transfecting the expression vector into COS-7 cells and lysing the cells in a radioimmunoprecipitation assay buffer [50 mM tris-HCl (pH 7.4), 150 mM NaCl, 5 mM EDTA, 1% NP-40] followed by purification with FLAG monoclonal antibodies conjugated to agarose beads (Sigma), according to the manufacturer’s instructions.

Liposome binding assay

The liposomes (1 mg/ml) were prepared in 100 μl of liposome buffer [25 mM Hepes-NaOH (pH 7.4) and 100 mM NaCl] and incubated with 10 μg of protein for 15 min at room temperature. The liposomes were sedimented by centrifugation at 100,000g for 20 min. The unbound fraction in the supernatant was transferred to another tube, and the bound protein that was cosedimented with the liposomes in the pellet was dissolved in 100 μl of assay buffer; both fractions were separated by SDS–polyacrylamide gel electrophoresis and visualized by Coomassie staining. The protein bands were quantified with ImageJ software.

In vitro kinase assay

GFP-Fer-FLAG protein (5 ng) and GST-cortactin (0.1 μg) were incubated either in the absence or in the presence of increasing concentrations of brain liposomes; the reaction was then initiated by the addition of 1 mM adenosine 5′-triphosphate (ATP)–Mg at 30°C. The reaction was stopped after 30 min by the addition of SDS sample buffer, and tyrosine phosphorylation was detected by immunoblotting with phosphotyrosine antibodies. One arbitrary unit was defined as the intensity of lane 2 in one of the triplicate experiments. Intensity values were standardized by the substrate amounts.

Immunofluorescence, microscopy, and image analysis

COS-7 cells were grown on coverslips and fixed with 3.7% paraformaldehyde. After permeabilization with 0.2% Triton X-100 for 5 min, the cells were sequentially incubated with the first and second antibodies for 1 hour each. The coverslips were mounted in PermaFluor (Thermo, Pittsburgh, PA) and observed under a FluoView 1000-D confocal microscope equipped with 473-, 568-, and 633-nm diode lasers (Olympus, Tokyo, Japan). The images were processed with Adobe Photoshop. Lamellipodia were considered to be arc-shaped membrane areas with F-actin structure gradually decreasing in density with increasing distance from the edge. Quantification of lamellipodia was performed as previously described (31). Cells with more than 50% of their perimeter occupied by lamellipodia were considered to have formed lamellipodia.

siRNA protein knockdown and migration assay

Fer siRNAs to specifically target the rat Fer messenger RNA (mRNA) sequence were designed with the BLOCK-iT RNAi Designer (Invitrogen; and synthesized by Invitrogen Corp. The target sequences were as follows: Fer siRNA #1, CAGCAACGACUGGAUAAUAUGAGAA; Fer siRNA #2, UGAUCAAGGACAAGCAGCAAGUGAA; and Fer control siRNA, UUCAUUAUCCUACUGAGUUGCUCUG. For knockdown of PLD1 or PLD2, validated siRNAs were purchased from Invitrogen. siRNAs were transfected into NRK52E cells with Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions, and abundance after knockdown was assessed after 48 hours by Western blotting or RT-PCR. On day 3, the cells were trypsinized and 2 × 104 cells were seeded onto the top of the cell culture inserts with 8-μm pores (BD Falcon, Franklin Lakes, NJ). The cells were fixed after 18 hours and stained with rhodamine-phalloidin, unless stated otherwise. The cells at the top and bottom sides of the chamber were counted, and the migration index was calculated as the proportion of migrated cells (at the bottom) to the nonmigrated cells (at the top). At least 300 cells were examined for each condition.


Statistical analysis was performed with analysis of variance (ANOVA) followed by the two-tailed multiple t test with Bonferroni correction (Fig. 2C) and χ2 test followed by multiple Fisher’s exact test or χ2 test with Bonferroni correction (Figs. 3C and 4, A to C, and fig. S10, A and B). The differences were considered to be significant if P < 0.05.


We are grateful to M. Frohman and Y. Takai for plasmids and H. Hiroaki and A. Shimada for discussion. We thank all the members of the Takenawa-Itoh laboratory for their helpful comments and I. Itoh for providing continuous support. This study was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Culture, Sports, Science, and Technology of Japan to T.I.; a Grant-in-Aid for Creative Scientific Research from the Japan Society for the Promotion of Science (JSPS) to T.T.; a Global COE (Centers of Excellence) Program from the JSPS to T.I., J.H., and T.T.; and the Nakajima Foundation to T.I.

Supplementary Materials

Fig. S1. The F-BAR–FX unit and other known units for membrane curvature and specific lipid interactions.

Fig. S2. The lipid specificity of Fer.

Fig. S3. Cortactin tyrosine phosphorylation by Fer.

Fig. S4. Fer tyrosine kinase activity is required for lamellipodia formation.

Fig. S5. Fer-induced lamellipodia formation is dependent on Rac.

Fig. S6. The F-BAR domain is essential for Fer-induced lamellipodia formation.

Fig. S7. Fer is translocated to the plasma membrane through the F-BAR–FX unit upon EGF stimulation.

Fig. S8. PLD activity is necessary for Fer activation.

Fig. S9. Fer knockdown.

Fig. S10. PLD and Fer are present in the same signaling pathway that results in cell migration.

References and Notes

View Abstract

Navigate This Article