Phosphatidic acid generated by PLD2 promotes the plasma membrane recruitment of IQGAP1 and neointima formation

Abstract

Very little is known about how lipid signaling regulates intima hyperplasia after vascular injury. Herein, we report that deletion and pharmacological inhibition of phospholipase D (PLD)2, which generates the signaling lipid phosphatidic acid (PA), reduced neointimal formation in the mouse carotid artery ligation model. PLD2 deficiency inhibits migration of vascular smooth muscle cells (VSMCs) into the intima in mice as well as migration and formation of membrane ruffles in primary VSMCs. PA specifically binds to the IQ motif–containing guanosine triphosphatase–activating protein 1 (IQGAP1) scaffold protein. The binding between PA and IQGAP is required for the plasma membrane recruitment of IQGAP1. Similar to PLD2 inhibition, knockdown of IQGAP1 blocks ruffle formation and migration in VSMCs, which are rescued by expression of the exogenous IQGAP1 but not the PA binding–deficient mutant. These data reveal that the PLD2-PA-IQGAP1 pathway plays an important role in VSMC migration and injury-induced vascular remodeling, and implicate PLD2 as a candidate target for therapeutic interventions.—Wang, Z., Cai, M., Tay, L. W. R., Zhang, F., Wu, P., Huynh, A., Cao, X., Di Paolo, G., Peng, J., Milewicz, D. M., Du, G. Phosphatidic acid generated by PLD2 promotes the plasma membrane recruitment of IQGAP1 and neointima formation.

The migration of vascular smooth muscle cells (VSMCs) is critical for both physiologic processes, such as angiogenesis and blood vessel development, and for pathologic processes, such as atherosclerosis and postangioplasty restenosis (1–3). One of the major characteristics of these lesions is an increased number of VSMCs in the neointima of the blood vessels due to proliferation and migration of VSMCs from the medial layer of the artery. Constriction of the blood vessels can eventually lead to severe clinical problems, such as thrombosis or embolus formation (1, 4–6). Growing evidence indicates that inhibition of VSMC migration can reduce the size of neointimal lesions in response to vascular injury (3).

Migration is a dynamic and highly coordinated process that requires the remodeling of the actin cytoskeleton beneath the plasma membrane (7–9). This process involves the plasma membrane recruitment and coordinated activation of a diverse array of proteins through a group of scaffold proteins. One such scaffold protein is the IQ motif containing guanosine triphosphatase (GTPase)–activating protein 1 (IQGAP1) (10–12). IQGAP1 is an evolutionarily conserved multidomain protein that can mediate the membrane binding of a diverse set of target proteins. Recruitment of IQGAP1 to the plasma membrane in response to specific stimuli promotes migration through increasing actin remodeling mediated via Rac family small GTPase 1 (RAC1), cell division cycle protein 42 (CDC42), and their effectors, as well as through increased microtubule dynamics via microtubule plus end regulators, cytoplasmic linker protein 170 (CLIP170) and adenomatous polyposis coli (APC) (10–12). IQGAP1 knockout mice exhibit impaired neointimal formation in a wire injury model due to reduced cellular migration (13). Therefore, there is significant interest in delineating the precise molecular mechanisms underlying the recruitment and activation of IQGAP1.

Lipids play an important role in recruiting cytosolic signaling proteins to the plasma membrane for their activation (14, 15). Increasing evidence has implicated the signaling lipid phosphatidic acid (PA) in cell migration (16–19). A major pool of signaling PA results from the hydrolysis of phosphatidylcholine (PC) by phospholipase D (PLD), which contains 2 family members, PLD1 and PLD2 (16, 17). PLD is activated in VSMCs in response to platelet-derived growth factor (PDGF) and angiotensin II stimulation (2, 20–23). However, the functional assignment of PLD in previous studies relied on nonspecific PLD inhibitors, such as primary alcohols. These inhibitors not only fail to differentiate the action of PLD isoforms but also generate many false-negative results (24). Moreover, although the activation of PLD in the cardiovascular system has been documented, the pathophysiological roles of PLD and the molecular mechanisms of PLD signaling in the vascular system remain unknown.

In the current study, we present biochemical and genetic evidence to support a novel mechanism of the regulation of VSMC migration during vascular remodeling. We show that deletion of Pld2 and inhibition of PLD activity by a small molecule inhibitor block neointima formation and vessel occlusion in an experimental mouse model of a vascular injury. PLD2-generated PA directly binds to and recruits IQGAP1 to the plasma membrane. Blockade of the PLD2-IQGAP signaling impairs actin cytoskeletal remodeling and migration of VSMCs. These data, for the first time, demonstrate that PLD2 contributes to pathologic vascular remodeling and implicate PLD2 as a new therapeutic target for vascular diseases involving cell migration.

MATERIALS AND METHODS

Common reagents and antibodies

PLD inhibitor 5-fluoro-2-indolyl des-chlorohalopemide (FIPI) (24) was obtained from Cayman Chemicals (Ann Arbor, MI, USA), and PLD1 inhibitor (VU0359595) and PLD2 inhibitor (VU0285655-1) (25) were obtained from Avanti Polar Lipids (Alabaster, AL, USA). Protease inhibitor cocktail (05892970001), PDGF-BB (P3201), and mouse anti–α tubulin antibody (T5168), were obtained from MilliporeSigma (Burlington, MA, USA). Rabbit anti–α smooth muscle actin (α-SMA) (AB5694), glutathione S-transferase (GST) (AB111947), and CD68 antibodies (AB125212) were obtained from Abcam (Cambridge, MA, USA). Rabbit anti-cleaved caspase-3 antibody (9661) was obtained from Cell Signaling Technology (Danvers, MA, USA). Mouse anti-6-His antibody (8904-1) was from Takara Bio (Kusatsu, Japan). Mouse anti–proliferating cell nuclear antigen (PCNA) (SC-56) and rabbit anti-IQGAP1 antibodies (SC-10792) were obtained from Santa Cruz Biotechnology (Dallas, TX, USA). Goat anti-rabbit IgG conjugated with Alexa Fluor 488 (A-11034), 594 (A-11032), and 680 (A-21076), Alexa Fluor 488–phalloidin (A-12379), and DAPI (D-1306) were obtained from Thermo Fisher Scientific (Waltham, MA, USA). Goat anti-mouse IgG conjugated with IRDye 800CW (92632210) was obtained from Rockland Immunochemicals (Limerick, PA, USA).

Animal and carotid artery ligation

All mouse experiments were performed in accordance with Association for Assessment and Accreditation of Laboratory Animal Care International guidelines and with University of Texas Health Science Center Institutional Animal Care and Use Committee approval. The generation of Pld2 knockout mice was previously described in Oliveria et al. (26). Pld2 on a mixed C57BL/6J-129/SvJ background was backcrossed for 20 generations to the FVB/N background, which showed significant neointimal hyperplasia after arterial injury (27). The knockout allele has always been maintained in a hemizygous background to avoid potential functional compensation. The carotid artery ligation was performed on both male and female wild-type (WT) and Pld2^−/− mice at the age of 10 wk. In brief, after anesthesia, the left common carotid artery was dissected free of connective tissues and ligated with 5-0 sutures proximal from the carotid bifurcation. The cut was closed with autoclips. The mice were euthanized and perfused with formalin at 2 or 4 wk after the ligation surgery.

For FIPI treatment, mice were treated with vehicle control or PLD inhibitor FIPI (3 mg/kg body weight) through intraperitoneal injection immediately before artery ligation. Thereafter, mice were treated twice daily with control or daily sets of 1 mg/kg body weight FIPI in the morning and 3 mg/kg FIPI 8 h later, until the animals were euthanized. This dosage has been established in previous studies (24, 28).

Histologic analyses

The ligated left common carotid artery and the control right artery were collected and prepared in paraffin-embedded blocks. At the area of 0.5 mm distally from the ligation site, the cross-sections were collected with 5-μm thickness for the following analysis of morphology and histology. Hematoxylin and eosin (H&E) and immunohistochemical staining were performed using standard protocol. Briefly, paraffin sections were deparaffinized in xylene and rehydrated in decreasing concentration of ethanol. Antigen retrieval was performed by boiling the sections in 10 mM sodium citrate (pH 6.0) for 10 min after rehydration. When the sections were cooled down, endogenous peroxidase was quenched by incubation of the slides with 3% H₂O₂ for 30 min, blocked in 5% normal goat serum in PBS for 30 min at room temperature. Slides were incubated with desired primary antibodies overnight at 4°C, followed by appropriate biotinylated secondary antibodies, and visualized with avidin-biotin complex (ABC) staining from Vectastain (Vector Laboratories, Burlingame, CA, USA). The medial area was calculated by subtracting the area defined by the internal elastic lamina from the area defined by the external elastic lamina, and the neointima area was calculated as the difference between the area inside the internal elastic lamina and the luminal area. Intima and media areas were measured by ImageJ (National Institutes of Health, Bethesda, MD, USA). The number of immunohistochemically stained cells positive to an antibody was blind counted.

Cell culture and lentiviral transduction

Primary VSMCs were isolated from mouse superior mesenteric artery as previously reported in Golovina et al. (29) and cultured in a smooth muscle basal medium (SmBM) BulletKit from Lonza (Basel, Switzerland). Before PDGF or Ang II stimulation, VSMCs were starved overnight in SmBM that did not contain growth factors or serum. For exogenous PA stimulation, 1,2-dilauroyl-sn-glycero-3-phosphate from Echelon (Salt Lake City, UT, USA) was dried by nitrogen gas and resuspended in culture medium to make 10 mM stock, and then sonicated briefly with low to medium power. Growth factor– starved cells were stimulated with 100 μM 1,2-dilauroyl-sn-glycero-3-phosphate for 15 min. Small hairpin RNA (shRNA) knockdown of gene expression and exogenous cDNA expression in VSMCs were mediated by lentiviral delivery. Viruses were generated in TLA-293T cells from GE Healthcare (Waukesha, WI, USA) by cotransfection of 4 plasmids including the lentiviral vector carrying desired shRNAs or cDNAs, pMDLg-pRRE, pRSV-Rev, and pMD2.G, using Lipofectamine and Plus reagent from Thermo Fisher Scientific. At 48 and 72 h posttransfection, virus-containing supernatants were collected and concentrated by centrifugation at 10,000 g for 4 h at 4°C with 10% sucrose [50 mM Tris HCl (pH 7.4), 100 mM NaCl, 0.5 mM EDTA, and 10% sucrose] (30). The infected cells were used for experiments 2–3 d postinfection.

Subcellular fractionation

The membrane and cytosol fractions were isolated using a modified method reported before in Del Pozo et al. (31). Briefly, after incubating with 0.5 ml of a hypotonic lysis buffer [20 mM HEPES (pH 7.4), 10 mM KCl, 2 mM MgCl₂, 1 mM EDTA, 1 mM EGTA, and ×1 protease inhibitor cocktail] on ice for 15 min, cells were scraped and passed through a 27-gauge needle 10 times. Homogenates were then incubated on ice for 20 min and centrifuged at 700 g for 3 min to remove nuclei and intact cells. The supernatants were spun at 100,000 g for 30 min at 4°C. The cytosol‐containing supernatant was collected and the crude membrane pellet was gently washed once with the lysis buffer. After being resuspended in 1× SDS-PAGE loading buffer, proteins were resolved by 1× SDS-PAGE and detected by Western immunoblotting.

Migration assay

VSMCs were detached by trypsin digestion and plated at 1 × 10⁴ cells per well in SmBM (without growth factors) in the presence or absence of PLD inhibitors in the upper chamber of a 24-well transwell with 8-µm pore size polycarbonate filters from Greiner Bio-One (Kremsmünster, Austria). PDGF (10 ng/ml) or angiotensin II (100 nM) was added as a chemoattractant in the lower chamber to allow cell migration to the bottom chamber for 4 h in the incubator (37°C, 5% CO₂). After removing the nonmigrated cells left on the upper chamber with a cotton swab, filters were fixed with cold methanol and cut; then, cell nuclei were stained with DAPI and the filters were mounted on the slides. About 2–5% of the plated cells migrated to the bottom chamber under the basal condition. The numbers of migrated cells from 6 random fields were counted using a Nikon (Tokyo, Japan) fluorescence microscope.

Western blotting and immunofluorescent microscopy

Western blotting using the Odyssey infrared imaging system from Li-Cor Biosciences (Lincoln, NE, USA) and immunofluorescent staining and confocal microscopy procedures for cultured cells were previously described in Roach et al. and Zhang et al. (32, 33). Nuclei and F-actin were stained by DAPI and Alexa Fluor 488-phalloidin, respectively. The images were captured using a Nikon A1 confocal microscope and processed using Nikon NIS-Elements software.

Plasmid construction

To generate the 5 mutants that disrupt the potential PA binding sites, GST-tagged IQGAP-C was subjected to PCR site–directed mutagenesis to change all the arginine and lysine residues to alanine. The mutations were confirmed by DNA sequencing. The IQGAP1 shRNA was cloned into the pLKO.2 vector with the targeting sequence of 5′-CCTCAGTTTGTACCTGTTCAA-3′. The control luciferase shRNA has been described before in Zhang et al. and He et al. (33, 34). To generate lentiviral expression constructs, the protein coding sequence of human IQGAP1 was amplified from green fluorescent protein–tagged IQGAP1 by PCR using primers IQGAP1 forward and IQGAP1 reverse and ligated to the XhoI and BamHI sites of pCDH-CMV-MCS-EF1-puro (pCDH) that was engineered to contain an N-terminal Myc tag. The sequences of oligos used for cloning are shown in Supplemental Table S1.

Bacterial protein expression, purification, and endotoxin removal

Recombinant proteins were expressed in the Escherichia coli Rosetta (DE3) strain from MilliporeSigma. Cells were cultured with 100 μg/ml ampicillin at 37°C until the optical density at a wavelength of 600 reached 0.6, then induced for protein expression with 1 mM isopropyl β-D-1-thiogalactopyranoside at 18°C overnight. GST- or His-tagged proteins were purified using glutathione agarose (Thermo Fisher Scientific) and HisPur Cobalt Resin (Thermo Fisher Scientific), respectively. After elution, proteins were dialyzed with the liposome binding buffer [150 mM NaCl and 20 mM Tris (pH 7.4)] for the subsequent assays. The purified proteins were examined by SDS-PAGE and Coomassie blue staining for purity. The concentrations of endotoxins in purified proteins were measured by a ToxinSensor Chromogenic LAL Endotoxin Assay Kit from GenScript (L00350C; Piscataway, NJ, USA) using the supplied endotoxin standard. To remove endotoxins, purified proteins were incubated with polymyxin B immobilized on Separopore 4B-CL from Bio-world (20181092-1; Dublin, OH, USA) at 4°C for 4 h.

Protein-lipid overlay assay

A lipid strip was prepared by spotting 100 pM of each lipid on a nitrocellulose membrane (0.2 μm; Bio-Rad, Hercules, CA, USA). 1,2-didecanoyl-sn-glycero-3-phosphorylated-l-serine (PS), PC (1,2-dioleoyl-sn-glycero-3-phosphocholine), PA (1,2-dioleoyl-sn-glycero-3-phosphate), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (PE), and phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2] were obtained from Avanti Polar Lipids. Phosphatidylinositol 3, 4, 5-trisphosphate was obtained from Matreya (State College, PA, USA). D-(+)-sn-1-O-oleoyl-glyceryl-3-phosphate, sphingosine 1-phosphate, phosphatidylinositol diC16, phosphatidylinositol 3-phosphate diC16, phosphatidylinositol 3,4-bisphosphate diC16, phosphatidylinositol 3,5-bisphosphate diC16, phosphatidylinositol 4-phosphate diC16, phosphatidylinositol 5-phosphate diC16, and 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphocholine were obtained from Echelon. The strips were blocked with 5% fatty acid–free bovine serum albumin in Tris-buffered saline with Tween for 2 h, followed by incubation with 1 μg/ml GST-IQGAP1 proteins in blocking buffer overnight. The lipid strips were then incubated with a rabbit anti-GST antibody for 1 h and a horseradish peroxidase–conjugated goat anti-rabbit IgG for 1 h. The strips were washed with Tris-buffered saline with Tween for 5 min, 3 times each. The lipid-bound protein was visualized with exposure to X-ray film using an ECL Kit from Thermo Fisher Scientific.

Liposome preparation and binding

Sucrose-loaded liposomes were prepared as previously described in Roach et al. and Tay et al. (32, 35). PC, PE, and one of the following lipids, PA, PS, PI(4,5)P2, or phosphatidylinositol 3, 4, 5-trisphosphate were mixed at a molar ratio of 6:3:1. 0.4-μg purified GST-tagged proteins were mixed and incubated with liposomes at 4°C for 30 min. After centrifugation at 70,000 g for 30 min, the supernatants were carefully removed. The proteins in pellets were detected by Western blot.

GST-C1 and His-C2 binding in the presence of liposomes

The binding assay was performed in the liposome pulldown buffer by adding 200 nM of His-tagged C1 (aa 956-1274 of IQGAP1) and 1 μM of GST-tagged C2 (aa 1275-1657 of IQGAP1) bound to glutathione beads in the presence or absence of 50 μM liposomes. After incubation at 25°C for 30 min, unbound proteins were washed out, and the protein complex was analyzed by Western blot.

Statistics

The statistical differences were evaluated between the control and each of the treatments using an unpaired, 2-tailed Student’s t test. All data are shown as means ± sd.

RESULTS

Decreased neointima formation after carotid artery ligation in Pld2^−/− and PLD inhibitor–treated mice

To determine the role of PLD2 in the injury-induced vascular remodeling, we performed common carotid artery ligation in WT and Pld2^−/− mice (36). In this model, flow cessation causes VSMCs to migrate into the intima, where their subsequent proliferation leads to the formation of neointima and a narrowing of the lumen. No neointima formation or luminal narrowing was observed in the unligated right common carotid arteries of WT and Pld2^−/− mice (Fig. 1A). In contrast, ligation of the left common carotid artery led to significant neointima growth 2 and 4 wk after ligation in WT female mice but little to no increase in Pld2^−/− female mice (Fig. 1A, B). Morphometric analysis of ligated arteries revealed that the average intima area was significantly smaller in Pld2^−/− than in WT mice, whereas no difference in medial area was found between genotypes. Consequently, the intima-to-media ratio was significantly decreased in Pld2^−/− mice compared with that in WT (Fig. 1A, B). Similarly, neointima formation was also significantly inhibited in Pld2^−/− male mice compared with those in WT male mice 4 wk after carotid artery ligation (Fig. 1C, D). Because PLD2 deficiency in females showed stronger inhibition of neointima formation, all of our experiments were performed in female mice thereafter.

Figure 1.

Reduced neointima formation in PLD2-deficiency mice following carotid artery ligation. A) Representative H&E staining of cross-sections of unligated and ligated female carotid arteries in female WT and Pld2^−/− mice. B) Quantification of intimal and medial size in A, which measures the area between the inner elastin layer and the distance between the inner and outer elastin layers, respectively (Week 2: WT, n = 9; Pld2^−/−, n = 10. Week 4: WT, n = 12; Pld2^−/−, n = 9). C) Representative H&E staining of cross-sections of ligated carotid arteries in male WT and Pld2^−/− mice. D) Quantification of intimal and medial size in C (WT, n = 6; Pld2^−/−, n = 6). E) Representative H&E staining of cross-sections of ligated carotid arteries in female WT mice treated with or without the PLD inhibitor FIPI. F) Quantification of intimal and medial size in E (WT, n = 6; Pld2^−/−, n = 6). NS, not significant. Scale bars, 100 μm. *P < 0.05, **P < 0.01, ***P < 0.001.

To evaluate the potential of inhibiting PLD2 in the treatment of vascular diseases related to VSMC migration, we treated mice subjected to carotid artery ligation with a broad PLD inhibitor, FIPI, which has been successfully used in mice in several previous studies (24, 28), for 4 wk. Similar to Pld2^−/− mice, intraperitoneal injection of FIPI significantly inhibited neointima formation. The average intima area was significantly smaller in FIPI-treated mice than in vehicle-treated control mice, whereas no difference in the media area was found between FIPI-treated and control mice (Fig. 1E, F).

PLD2 ablation has no effect on VSMC proliferation and death

The accumulation of VSMC in the intima of injured arteries is mainly caused by the migration and proliferation of VSMCs (3, 36). To determine whether inhibition of neointima formation in Pld2^−/− artery was due to a reduction in VSMC proliferation, we performed immunohistochemical staining for a proliferation marker, PCNA, in the neointima and media at 1, 2, and 4 wk after carotid ligation injury. There were no differences in the percentages of PCNA-positive neointimal and medial VSMCs between WT and Pld2^−/− mice across all time points (Supplemental Fig. S1A, B). Consistent with this in vivo data, there was no difference in the number of viable cells between WT and Pld2^−/− primary VSMCs in culture (Supplemental Fig. S1C). Additionally, there was no difference in the apoptosis marker–cleaved caspase-3 and the macrophage marker CD68 immunostaining between the 2 groups (Supplemental Figs. S2 and S3), suggesting PLD2 deficiency does not affect the apoptosis of VSMCs or recruitment of macrophages.

PLD2 deficiency impairs VMSC migration

The same ratio of PCNA-positive cells in injured arteries between WT and Pld2^−/− mice suggested that the reduction in neointima formation in Pld2^−/− mice might be due to a decrease in the migration of VSMCs. To identify the VSMCs in neointima, we performed immunohistochemical staining for α-SMA at 4 d and 1 wk after carotid ligation injury (Fig. 2A, B). PLD2 deficiency significantly reduced the appearance of VSMCs in the intima at both time points. Because VSMCs do not replicate in the first few days after injury (37–39), the accumulation of VSMCs in the intima was most likely caused by the migration of VSMCs from media. To confirm the role of PLD2 in VSMC migration, we performed a transwell assay in vitro. Primary VSMCs from Pld2^−/− mice showed impaired migration in response to the stimulation of either angiotensin II or PDGF, 2 potent stimuli for SMC migration (Fig. 2C). Consistently, blockade of PLD2 activity using a PLD2-specific inhibitor also reduced migratory responses to the same stimuli (Fig. 2D). In contrast, inhibition of PLD1 by a PLD1-specific inhibitor had no effect on VSMC migration (Fig. 2E), suggesting that only PLD2 mediates VSMC migration. Together, these in vivo and in vitro data suggest that the inhibition of neointima formation in Pld2^−/− mice is mainly caused by the reduction of VSMC migration but not VSMC proliferation, apoptosis, or macrophage recruitment.

Figure 2.

PLD2 inhibition reduces the migration of VSMCs. A) α-SMA staining of ligated carotid arteries at early dates after ligation. I, intima; M, media. Scale bars, 100 μm. B) Quantification of α-SMA–positive (SMA⁺) VSMCs in A (Day 4: WT, n = 5; Pld2^−/−, n = 6. Week 1: WT, n = 6; Pld2^−/−, n = 8). C) Reduced migration in Pld2^−/− VSMCs. Cell migration was measured for 4 h using transwells in the presence or absence of 10 ng/ml PDGF or 100 nM Ang II (n = 3). D) PLD2-specific inhibitor (PLD2i) VU0285655-1 (2 μM) impaired VSMC migration (n = 3). E) PLD1-specific inhibitor (PLD1i) VU0359595 (2 μM) had no effect on VSMC migration (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001.

PLD2 deficiency reduces actin cytoskeletal remodeling

Cell migration is powered by active actin cytoskeletal reorganization. There were no differences in cell morphology and F-actin staining between WT and Pld2^−/− VSMCs in the basal condition (Fig. 3). PDGF stimulation in WT VSMCs lead to the formation of an F-actin–rich structure, dorsal membrane ruffle, which promotes cell migration (8, 9, 40) (Fig. 3A, B, and Supplemental Fig. S4). In contrast, the formation of membrane ruffles in Pld2^−/− VSMCs was significantly lower than that in WT VSMCs (Fig. 3A, B). To rule out that the changes in actin structures were caused by secondary alterations after long-term removal of PLD2 in Pld2^−/− VSMCs, we pretreated WT VSMCs with PLD2 inhibitor for 30 min and then performed PDGF stimulation. As in Pld2^−/− VSMCs, PLD2 inhibitor also impaired the formation of membrane ruffles (Fig. 3C, D).

Figure 3.

PLD2 regulates actin cytoskeletal reorganization in VSMCs. A) The formation of membrane ruffles induced by PDGF is reduced in Pld2^−/− VSMCs. Scale bars, 20 μm. B) Quantification of membrane ruffles in A [n = 3 independent experiments (≥200 cells were quantified in each independent experiment)]. C) The formation of membrane ruffles induced by PDGF is inhibited by PLD2 inhibitor. Scale bars, 20 μm. D) Quantification of membrane ruffles in C [n = 3 independent experiments (≥200 cells were quantified in each independent experiment)]. Membrane ruffles are marked by arrowheads. F-actin was stained with Alexa Fluor 488-phalloidin. Total cells quantified: Pld2^+/+, n = 135; Pld2^−/−, n = 117; DMSO, n = 126; PLD2 inhibitor (PLD2i), n = 116. *P < 0.05, **P < 0.01, ***P < 0.001.

PA binds to the C terminus of IQGAP1

IQGAP1 is a multidomain protein that plays a critical role in regulating cytoskeletal dynamics, including membrane ruffle (10–12). The membrane recruitment of IQGAP1 from cytosol is required for its activation (10–12). We recently identified IQGAP1 as one of the top hits in a PA binding protein screening using liposome pulldown followed by mass spectrometry (41). To confirm the direct and specific binding of PA, we purified 4 GST-tagged IQGAP1 fragments that cover most of the IQGAP1 protein [N (aa 1–216), M (aa 521–914), Ras–GTPase-activating protein related domain (GRD) (aa 998–1271), and C (aa 1418–1657)] from E. coli and tested their binding to a variety of phospholipids spotted on nitrocellulose membranes (Fig. 4A). Fragments N, M, and GRD showed no detectable binding to any lipids on the lipid strips. In contrast, the C fragment (IQGAP1-C) showed binding to PA and, to a lesser extent, PI(4,5)P2 and PE (Fig. 4A). Similarly, the binding of IQGAP1-C to PA is also strongest in a liposome pulldown experiment (Fig. 4B, C). Proteins purified from bacteria are known to contaminate with endotoxin. Because lipid A, the active component of bacterial endotoxin, structurally resembles PA (42, 43), it is likely that the binding between PA and IQGAP1-C may be interfered by lipid A in our binding assay. To evaluate the contribution of endotoxin, we removed endotoxin from IQGAP1-C, then performed a liposome pulldown assay. Our result shows that the removal of endotoxin does not affect the binding between PA and IQGAP1-C (Supplemental Fig. S5).

Figure 4.

IQGAP1 binds to PA directly and specifically. A) Lipid strip binding assay. Top, domain structure of IQGAP1and GST fusion constructs used for the experiment. CHD, calponin homology domain; IQ-Repeat, IQGAP-specific repeat motif; WW, domain with 2 conserved Trp (W) residues; IQ, IQ motif. Bottom, the purified C terminus of IQGAP1 (IQGAP1-C) specifically bound to PA, and to a lesser extent to PI(4,5)P₂, PE, and PS. Protein bound to lipid strip was detected using a GST antibody. LPA, D-(+)-sn-1-O-oleoyl-glyceryl-3-phosphate; LPC, 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphocholine; PI(3)P, phosphatidylinositol 3-phosphate diC16; PI(4)P, phosphatidylinositol 4-phosphate diC16; PI(5)P, phosphatidylinositol 5-phosphate diC16; S1P, sphingosine 1-phosphate; PI(3,4)P2, phosphatidylinositol 3,4-bisphosphate diC16; PI(3,5)P2, phosphatidylinositol 3,5-bisphosphate diC16; PI(3,4,5)P3, phosphatidylinositol 3,4,5-trisphosphate. B) IQGAP1-C specifically bound to PA in a liposome pulldown assay. IQGAP1-C bound to liposomes containing indicated phospholipids was detected by Western blot using a GST antibody. C) Quantification of the binding of IQGAP1-C to liposomes in B (n = 3). D) Identification of the PA binding site, A3, on IQGAP1. Left, the candidate PA binding sites on IQGAP1-C and corresponding alanine mutants. Right, Western blot analysis of PA-binding ability of IQGAP1-C WT and mutants in a liposome pulldown assay of serially diluted liposomes. E) Quantification of the binding of IQGAP1-C proteins to PA liposomes in D (n = 3). ***P < 0.001.

Because PA binding sites are often composed of a stretch of positively charged amino acids (44, 45), PA might bind to IQGAP1-C through 1 of the 5 regions rich in lysine and arginine. We mutated all lysine and arginine residues to alanine (A) in these 5 sites, expressed and purified the resulting mutants (A1, A2, A3, A4, and A5) in E. coli, and tested their PA-binding ability using a liposome pulldown assay (Fig. 4D). One of the mutants, A3, significantly lost its binding to PA liposomes (Fig. 4D, E). The A5 site was previously identified as PI(4,5)P2 binding (46). Interestingly, mutations in this site did not reduce the PA binding of IQGAP1 (Fig. 4D, E), suggesting that IQGAP1 binds to PA and PI(4,5)P2 through independent sites. Together, these data show that IQGAP1 directly and selectively binds to PA through the A3 site and that the A3 mutation suffices to disrupt the binding between IQGAP1 and PA.

PA binding does not relieve the autoinhibition of IQGAP1

The intramolecular interaction between the GRD and the Ras–GTPase-activating protein C terminus (RGCT) domains keeps IQGAP1 in an inactive conformation (12). Ras homologous (Rho) GTPase binding to the GRD, PI(4,5)P2 binding to the A5 site, or phosphorylation of Ser1443 relieves autoinhibition and activates IQGAP1 (46, 47). The PA binding site we identified is in the RGCT domains and is close to Ser1443 (Fig. 4D), suggesting that PA binding may open the inactive conformation. Using a similar assay reported before in Choi et al. and Grohmanova et al. (46, 47), the binding between His-C2 (containing RGCT) to immobilized GST-C1 (containing GRD) was tested in the presence or absence of liposomes containing different phospholipids (Fig. 5A). In the absence of liposomes, His-C2 bound to GST-C1. However, the binding was impaired in the presence of PI(4,5)P2 liposomes as reported before in Choi et al. (46) (Fig. 5B). In contrast, the addition of PA and other liposomes did not affect the binding (Fig. 5B). This result indicates that PA binding does not disrupt the intramolecular autoinhibition of IQGAP1. Introduction of the A3 mutations in the C2 (PA binding deficiency) did not change the effect of PI(4,5)P2 on the binding between His-C1 and GST-C2 (Fig. 5B), further supporting that IQGAP1 binds to PA and PI(4,5)P2 through independent binding sites.

Figure 5.

PI(4,5)P2 but not PA binding relieves the intramolecular autoinhibition of IQGAP1. A) Diagram of the experiment demonstrating lipid-binding disruption of the intramolecular interaction between the GRD and the RGCT. B) His-C2 WT or A3 mutant was incubated with GST-C1 immobilized on glutathione beads in the absence or presence of liposomes containing the indicated lipids. GST-C1 and His-C2 proteins bound to GST-C1 were detected by Western blot using an anti-GST and anti‐His antibody, respectively.

PLD2-generated PA mediates the membrane recruitment of IQGAP1 and is required for IQGAP1-regulated VSMC migration

The specificity of PA regulation of IQGAP1 is further supported by localization studies in VSMCs. In the basal condition, IQGAP1 was localized in the cytosol. Upon PDGF stimulation, IQGAP1 was translocated to plasma membranes, including cell peripheral lamellipodia and dorsal ruffles. In contrast, IQGAP1 remained in the cytosol in PLD2 knockout (Fig. 6A) and PLD2 inhibitor–treated cells (Supplemental Fig. S6). To confirm PLD2 regulation of IQGAP1 localization, we performed subcellular fractionation and found that PLD2 deficiency led to a reduction in PDGF-promoted IQGAP1 in the membrane fraction assay (Fig. 6B). Addition of exogenous PA to VSMCs in the absence of PDGF failed to recruit IQGAP1 to plasma membranes (Supplemental Fig. S7), suggesting that PLD2-generated PA needs to work with other signals to promote IQGAP1 plasma membrane association.

Figure 6.

PLD2 is required for the plasma membrane localization of IQGAP1. A) PDGF-promoted plasma membrane localization (peripheral and membrane ruffles) of IQGAP1 was disrupted in Pld2^−/− VSMCs. Scale bars, 20 μm. Cell peripherals and membrane ruffles are marked by arrows and arrowheads, respectively. B) IQGAP1 is reduced in the membrane fraction in PDGF-stimulated Pld2^−/− VSMCs. Membrane and cytosol fractions were isolated 7 min after 10 ng/ml PDGF stimulation.

To directly evaluate PA regulation of IQGAP1, we introduced the A3 mutations to the full-length human IQGAP cDNA to generate an IQGAP1 PA binding–deficient mutant (IQGAP1-PA⁻). Using lentiviral delivery, we knocked down the endogenous IQGAP1 in mouse VSMCs using an shRNA, then re-expressed the WT IQGAP1 and IQGAP1-PA⁻ mutant (Fig. 7A). Like the endogenous IQGAP1, both exogenously expressed IQGAP1 and IQGAP1-PA⁻ mutant were localized to the cytosol in basal condition (Supplemental Fig. S8). PDGF stimulation recruited both endogenous and exogenous WT IQGAP1s but not IQGAP1-PA⁻ to cell peripherals and membrane ruffles (Fig. 7B). Consistent with the membrane localization, only re-expression of WT IQGAP1 was able to restore the defects in the formation of membrane ruffles in PDGF-stimulated VSMCs (Fig. 7B, C). Similarly, whereas the exogenous IQGAP1 WT completely restored the migration defect in VSMCs expressing the IQGAP1 shRNA, the IQGAP1-PA⁻ mutant failed to do so (Fig. 7D). Together, these data support a model that the binding of PLD2-generated PA to IQGAP1 is critical to its membrane targeting, actin remodeling, and migration in VSMCs.

Figure 7.

PA binding is required for IQGAP1 plasma membrane localization, actin reorganization, and migration in VSMCs. A) Western blot showed the knocking down of endogenous IQGAP1 by an shRNA and re-expression of the exogenous IQGAP1 and IQGAP1-PA⁻ mutant in VSMCs. B) IQGAP1 and F-actin staining in the control luciferase (Luc) and IQGAP1 knockdown and re-expression VSMCs. Scale bars, 20 μm. C) Quantification of membrane ruffles in B [n = 3 independent experiments (at least 180 cells were quantified in each independent experiment)]. D) Migration in the control and VSMCs in which IQGAP1 was knocked down and re-expressed (n = 3). Cell peripherals and membrane ruffles are marked by arrows and arrowheads, respectively. ***P < 0.001.

DISCUSSION

Vascular wound repair in response to arterial injury involves migration and proliferation of VSMCs. The complex mechanisms regulating this process are not fully understood. Up to now, there has been no report on the role of PLD2 in vascular remodeling. In the present study, we investigated the mechanism through which PLD2 regulates vascular remodeling in a well-established carotid artery ligation model in Pld2^−/−- and FIPI-treated mice. We found that PLD2 is essential for intimal hyperplasia (Fig. 1). The results of this study reveal an important role of PLD2 in mediating VSMC migration and neointima formation in response to injury and stress. Pld2 knockout mice show no detectable developmental abnormality (26), suggesting that PLD2 is a potential target for vascular dysfunction. Importantly, we did not find any discernable toxicity in mice in our study, supporting the safety and effectiveness of FIPI in previous studies (28, 48). Furthermore, halopemide, from which FIPI was derived, has been used clinically as a psychotropic agent (49, 50).

We did not find a statistically significant difference in the rate of VSMC proliferation between WT and Pld2^−/− arteries and primary VSMCs (Supplemental Fig. S1). Additionally, we did not observe changes in apoptosis and macrophages between injured WT and Pld2^−/− arteries (Supplemental Figs. S2 and S3). The major differences we found were fewer intimal α-SMA–positive cells in Pld2^−/− mice than in WT mice 4 d and 1 wk after ligation (Fig. 2A, B) and reduced migratory capability of PLD2-inhibited VSMCs in vitro (Fig. 2C, D). These results indicate that PLD2 regulates injury-induced neointima formation primarily through VSMC migration. The lack of effect of PLD1 inhibitors on VSMC migration also indicates that this process is specifically regulated by PLD2 but not PLD1.

Membrane ruffles on the plasma membrane are critical for cell migration (8, 9, 40). We showed that the number of membrane ruffles was significantly inhibited in Pld2^−/− or PLD2 inhibitor–treated VSMCs (Fig. 3), suggesting that the impairment of migration in PLD2-deficient VSMCs is due to the reduced formation of membrane ruffles. Our finding is consistent with several previous studies showing that inhibition of VSMC migration without affecting VSMC proliferation is sufficient to prevent pathologic vascular remodeling (3, 39, 51), indicating that therapeutic disruption of vascular signaling that leads to uncontrolled migration may benefit the treatment of vascular diseases, such as restenosis after angioplasty. However, our results cannot fully rule out the role of PLD2 in VSMC proliferation in vivo. The number of VSMCs in WT neointima is higher than in Pld2 knockout mice. It is likely that there is a transient increase in the proliferation rate of VSMCs at a certain time point after injury, which is not reflected by the measurement at the examined time. This transient increase in proliferation might not be directly related to the primary PLD2 or other migration-regulatory proteins. For example, VSMCs migrated into intima may increase their proliferation when more nutrients are available from the blood.

The plasma membrane recruitment of IQGAP1 is critical for its regulation of cytoskeletal reorganization and migration (10–12). IQGAP1 was reported to bind to PI(4,5)P2, and this binding is required for the activation but not the membrane recruitment of IQGAP1 (46, 52). We find that PA binds to the C terminus of IQGAP1 at a site different from the PI(4,5)P2-binding site (Fig. 4). In contrast with PI(4,5)P2, PA binding is critical to the plasma membrane recruitment of IQGAP but does not seem to directly contribute to the relief of the autoinhibition of IQGAP1 that is required for its activation (Fig. 5). We showed that PLD2-generated PA is required for the recruitment of IQGAP1 to the plasma membrane (Fig. 6 and Supplemental Figs. S4 and S6), and disruption of the PA binding site impaired IQGAP1 plasma membrane recruitment and inhibited membrane ruffle formation and cell migration (Fig. 7). Furthermore, the reduced neointimal phenotype of Pld2^−/− mice after carotid artery ligation is similar to that of Iqgap1^−/− mice (13), supporting the in vivo role of PLD2-generated PA regulation of IQGAP1.The nonoverlapping PA- and PI(4,5)P2-binding sites suggest that these 2 phospholipids function in a coordinated manner. We therefore propose this model of IQGAP1 activation: upon migratory stimulations, PLD2-generated PA works with PI(4,5)P2 and other regulators to recruit IQGAP1 to the plasma membrane where it can be further activated (Fig. 8). Functionally, the bindings to both PA and PI(4,5)P2 are important for the actin-remodeling activity of IQGAP1. Interestingly, it has been proposed that there is a potential feed-forward regulation of PLD (i.e., PA) and type I phosphatidylinositol-4-phosphate 5-kinase (PIPKI) [i.e., PI(4,5)P2] (32, 53). PA directly binds and activates PIPKIs, the major PI(4,5)P2-generating enzymes (32, 35, 53, 54). Conversely, PI(4,5)P2 interaction is also essential to the activation of PLD2 (22, 55). This positive feed-forward signaling might allow local generation of high levels of both PA and PI(4,5)P2 for a rapid activation of IQGAP1 at specific subcellular locations upon stimulations. PLD2 was also reported to have a PA generation–independent role by functioning as a guanine exchange factor for Rac family small GTPase 2 (RAC2) (56). PLD2 may also signal through other PA effectors to regulate proliferation and migration (16, 57, 58). It is likely that IQGAP1-independent PLD2 functions contribute to the observed VSMC phenotypes reported in this study. However, we found that the PLD2 inhibitor largely suppressed migration, IQGAP plasma membrane recruitment, and membrane ruffle in VSMCs, and Pld2 and Iqgap1 knockouts showed similar VSMC migration and neointima formation phenotypes. These results support that the PLD2-PA-IQGAP1 pathway plays a major role in the observed phenotypes in these cells. In the future, it would be important to directly test the involvement of this pathway in vascular injury using a knock-in mouse model in which the PA binding site on IQGAP1 is mutated, similar to what has been done for actin reorganization and migration in VSMCs in the current study (Fig. 7).

Figure 8.

A model of PA regulation of IQGAP1. Upon activation of membrane receptors, IQGAP1 is phosphorylated and recruited to the plasma membrane from the cytosol by direct binding to PLD2-generated PA, and activated Rac family small GTPase 1 (Rac1) and cell division cydle protein 42 (Cdc42). PI(4,5)P2 on the plasma membrane then binds to IQGAP1 and relieves its intramolecular autoinhibition between the GRD and RGCT domains, leading to activation of its downstream signals, including actin polymerization. CHD, calponin homology domain; IQ-Repeat, IQGAP-specific repeat motif; P, phosphorylation; WW, domain with 2 conserved Trp (W) residues; IQ, IQ motif.

In summary, our study reveals a new role for the signaling lipid PA in VSMC function. PLD2-generated PA recruits IQGAP1 to the plasma membrane for its activation. Inhibition of this process impairs VSMC migration and neointima formation after vascular injury. Our findings support the notion that the PLD2-PA-IQGAP1 signaling pathway functions as a crucial regulator of injury-induced vascular remodeling and might be a therapeutic target for vascular diseases.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

α-SMA

α smooth muscle actin

FIPI

5-fluoro-2-indolyl des-chlorohalopemide

GRD

Ras GTPase–activating protein related domain

GST

glutathione S-transferase

GTPase

guanosine triphosphatase

H&E

hematoxylin and eosin

IQGAP1

IQ motif–containing GTPase-activating protein 1

IQGAP1-C

aa 1418–1657 fragment of IQGAP1

IQGAP1-PA⁻

IQGAP1 PA binding–deficient mutant

PA

phosphatidic acid

PC

phosphatidylcholine

PCNA

proliferating cell nuclear antigen

PDGF

platelet-derived growth factor

PE

1,2-dioleoyl-sn-glycero-3-phosphoethanolamine

PI(4,5)P2

phosphatidylinositol 4,5-bisphosphate

PIPKI

type I phosphatidylinositol-4-phosphate 5-kinase

PLD

phospholipase D

PS

1,2-didecanoyl-sn-glycero-3-phosphorylated-l-serine

RGCT

Ras–GTPase-activating protein C terminus

shRNA

small hairpin RNA

SmBM

smooth muscle basal medium

VSMC

vascular smooth muscle cell

WT

wild type

AUTHOR CONTRIBUTIONS

Z. Wang and G. Du designed research; Z. Wang, M. Cai, L. W. R. Tay, F. Zhang, P. Wu, A. Huynh, and X. Cao performed experiments; G. Di Paolo generated Pld2 knockout mice; J. Peng performed the mass spectrometry experiment and data analysis; Z. Wang, D. M. Milewicz, and G. Du analyzed the data; Z. Wang and G. Du wrote the manuscript; and G. Di Paolo and D. M. Milewicz edited the manuscript.