Oxidized LDL/CD36 interaction induces loss of cell polarity and inhibits macrophage locomotion.

Cell polarization is essential for migration and the exploratory function of leukocytes. However, the mechanism by which cells maintain polarity or how cells revert to the immobilized state by gaining cellular symmetry is not clear. Previously we showed that interaction between oxidized low-density lipoprotein (oxLDL) and CD36 inhibits macrophage migration; in the current study we tested the hypothesis that oxLDL/CD36-induced inhibition of migration is the result of intracellular signals that regulate cell polarity. Live cell imaging of macrophages showed that oxLDL actuated retraction of macrophage front end lamellipodia and induced loss of cell polarity. Cd36 null and macrophages null for Vav, a guanine nucleotide exchange factor (GEF), did not show this effect. These findings were caused by Rac-mediated inhibition of nonmuscle myosin II, a cell polarity determinant. OxLDL induced dephosphorylation of myosin regulatory light chain (MRLC) by increasing the activity of Rac. Six-thioguanine triphosphate (6-thio-GTP), which inhibits Vav-mediated activation of Rac, abrogated the effect of oxLDL. Activation of the Vav-Rac-myosin II pathway by oxidant stress may induce trapping of macrophages at sites of chronic inflammation such as atherosclerotic plaque.

INTRODUCTION

Cell polarization is a prerequisite for migration (Lauffenburger and Horwitz, 1996) and is mediated by interlinked molecular pathways (Ridley ). This intrinsically self-reinforcing mechanism results in generation of the protrusive front and retraction of the rear end, and thus maintains a certain degree of persistent directional cell movement even in the absence of external directionality cues (Zigmond ; Pankov ). For leukocytes, exploratory locomotion driven by spontaneous breakage of cellular symmetry enables cells to sense the surrounding environment and is critical for the propagation of immune and inflammatory responses (Parent and Devreotes, 1999; Vicente-Manzanares and Sánchez-Madrid, 2004). Mechanisms underlying spontaneous cell polarization, however, are not clearly defined. The ability of cells to spontaneously generate asymmetry is linked to mechanisms by which cells maintain symmetry and remain stationary. Therefore elucidating the process in which cells lose polarity may help identify key mediators of the polarization process.

Rho family small-molecular-weight guanine triphosphatases (GTPases) are known to play a role in cell polarization by regulating cytoskeletal dynamics (Nobes and Hall, 1999; Wittmann and Waterman-Storer, 2001; Raftopoulou and Hall, 2004). Rac generates protrusive force at the leading edge by regulating polymerization of actin, and RhoA is involved in retraction of the rear end. Cytoskeletal regulation by Rho GTPases is complex because of a network of interacting molecules including guanine-nucleotide exchange factors (GEFs), GTPase-activating proteins, scaffold proteins including Wiskott-Aldrich syndrome proteins (WASPs), WASP family verprolin-homologous proteins, and phosphoinositide-3 kinase (Charest and Firtel, 2007). Interplay among different Rho GTPases can also produce compound effects on the cytoskeleton (Sander ; Yamaguchi ). Cells expressing a dominant-negative Rac1 have defects in pseudopodia protrusion and generation of F-actin–rich leading edges, resulting in poor motility (Chung ). Significantly, cells expressing constitutively active forms of Rac are also defective in motility, suggesting the importance of balanced regulation of interlinked pathways (Chung ; Dumontier ). Nonmuscle myosin II is regulated by Rho GTPases and also mediates cytoskeletal dynamics (Zhao and Manser, 2005). The activity of nonmuscle myosin II depends on the reversible phosphorylation of the myosin regulatory light chain (MRLC) on Ser-19 (Somlyo and Somlyo, 2003) and is further increased by additional phosphorylation of Thr-18 of MRLC in the presence of phosphorylated Ser-19 (Ikebe ). RhoA activates nonmuscle myosin II by inhibiting myosin phosphatase (MP), thereby increasing phosphorylation of MRLC (Amano ; Kimura ; Kawano ). Rac promotes contractility by activating p21-activated kinase (PAK), which directly phosphorylates MRLC (Chew ). However, there are contradictory findings about the effect of Rac on myosin II, because Rac signaling is also implicated in the negative regulation of myosin by promoting actin–myosin disassembly (van Leeuwen ). In addition, activated PAK phosphorylates and inhibits myosin light chain kinase (MLCK) (Sanders ; Goeckeler ). To date, all studies on the roles of Rho GTPases in cell polarization have used dominant negative mutants or introduced constitutively active forms that may have driven these contradictory effects. No studies have been published showing that these pathways can be perturbed endogenously to drive pathological processes.

Our lab has a long-standing interest in CD36, a transmembrane glycoprotein receptor expressed in a variety of cells including monocytes and macrophages. CD36 promotes atherosclerosis by mediating oxidized low-density lipoprotein (oxLDL) uptake in macrophages leading to the formation of lipid-laden foam cells (Febbraio , 2004; Rahaman ; Guy ; Kuchibhotla ). CD36 also inhibits macrophage migration (Park ) and thus contributes to foam cell accumulation in the vascular intima, leading to development of atherosclerotic plaque. We and others have hypothesized that strategies to promote lipid-laden macrophage egress from the vessel wall may be useful to prevent or reverse atherosclerosis, and indeed several studies have shown that regressed atherosclerotic plaque is characterized by the disappearance of foam cells (Daoud ; Llodra et al., 2004) associated with their emigration to regional lymph nodes (Llodra et al., 2004).

In the current study we tested the hypothesis that oxLDL/CD36 interactions inhibit migration as a result of intracellular signals that regulate cell polarity. We showed that Rac1 activation by Vav family GEFs induced retraction of front end lamellipodia and loss of cell polarity in macrophages and found that interaction of oxLDL with CD36 triggered this pathway, leading to inactivation of nonmuscle myosin II through inhibition of phosphorylation of MRLC. These studies suggest that endogenous “danger signals” generated by oxidant stress can inhibit macrophage motility and provide a mechanistic explanation for altered migratory function of macrophages within atherosclerotic plaque and other inflammatory milieu containing oxLDL. The CD36–nonmuscle myosin II pathway could thus be a target for development of novel strategies to promote regression of atherosclerosis. Our data also describe a novel integrated paradigm for the mechanism of cell polarization that is modulated by a link between the Vav/Rac pathway and nonmuscle myosin II.

RESULTS

OxLDL inhibits murine macrophage locomotion by inducing loss of cell polarity

Live cell imaging showed that resident peritoneal macrophages from wild-type C57BL/6 (wt), cd36 null, and vav1 null mice plated on serum-coated glass coverslips made protrusions and then spontaneously polarized. Polarized macrophages protruded broad lamellipodia on their front ends and started to move by retracting their rear ends, leaving retraction fibers at the rear (Figure 1A; Supplemental Video 1). After the addition of NO2LDL, a form of oxLDL modified by a myeloperoxidase (MPO)-nitrite system that is a specific ligand for CD36 (Podrez ), wt macrophages retracted their front end lamellipodia and generated retraction fibers around the front end, thus losing their polarity as well as their ability to advance (Figure 1A; Supplemental Video 2). Macrophages from cd36 null mice did not show these changes and thus maintained the ability to migrate in the presence of NO2LDL (Figure 1B; Supplemental Video 3). Similarly, macrophages from mice null for Vav1, a GEF recently shown to be a downstream effector of CD36 (Wilkinson , Rahaman ), did not show lamellipodial retraction in response to NO2LDL (Figure 1C; Supplemental Video 4).

FIGURE 1:

OxLDL induces retraction of lamellipodia and loss of cell polarity. (A) Resident peritoneal macrophages from wt mice were plated onto a serum-coated, glass bottom dish and allowed to spontaneously polarize. Time-lapse images were taken every 15 s for 1 h before and after the addition of NO2LDL (50 μg/ml). Solid arrows indicate the front end lamellipodia, and dashed arrows indicate the rear end. Macrophages from cd36 null mice (B) and vav1 null mice (C) were tested as described in (A). Data are representative of five separate experiments analyzing 10–15 cells for each cell type. White scale bar = 10 μm.

Quantitative analysis of the live cell imaging studies was performed using several different parameters. NO2LDL increased the number of retraction fibers per cell by 1.5-fold in wt macrophages but not in cd36 null or vav1 null cells (Figure 2, A and B). Dynamic movement of the macrophage membrane, assessed by measuring ruffle area, was decreased by NO2LDL in wt but not cd36 null macrophages (Figure 2, A and C; Supplemental Videos 5 and 6). NO2LDL-induced changes were limited to the cellular front; ruffle area was not changed in the rear (Supplemental Figure S1). The response in vav1 null cells was intermediate (Figure 2C). Macrophage velocity, measured as travel distance in 1 h, was decreased by NO2LDL in wt but not cd36 null or vav1 null cells (Figure 2D). Thioglycollate-elicited macrophages behaved similarly to resident macrophages in this system (Supplemental Figure S2, A and B). In all studies, NO2(–)LDL, a control LDL that was exposed to all the components of the MPO system except the oxidant, had no effect (Figure 2, E and F). These studies, in sum, showed that NO2LDL inhibited directional cell movement in macrophages via a CD36-Vav–dependent mechanism.

FIGURE 2:

OxLDL induced retraction fiber formation around lamellipodia and decreased ruffle formation of macrophages. (A) Images from the time-lapse microscopy described in Figure 1 were analyzed with Image-Pro software (Media Cybernetics). Green or pink indicator lines were used to mark protrusions (top panels). The area in green is the newly formed protrusion from the prior cell margin imaged 15 s earlier (bottom panels). White scale bar = 10 μm. (B) Retraction fiber counts and (C) ruffle area were compared among wt, cd36 null, and vav1 null macrophages. (D) Velocity measured as travel distance in 1 h was compared among wt, cd36 null, and vav1 null macrophages. (E and F) Wt macrophages were treated with NO2LDL or NO2(–)LDL at 50 μg/ml as in Figure 1, and retraction fiber count (E) and velocity (F) were measured. (A–D) Data are representative of five separate experiments analyzing 10–15 cells for each cell type. (E and F) Data are representative of three separate experiments analyzing 9–12 cells for each treatment.

OxLDL-induced inhibition of macrophage migration depends on CD36 and Vav family GEFs

We performed scratch wound closure assays combined with time-lapse microscopy to assess the effect of oxLDL-induced loss of polarity on macrophage migration. As shown in the representative image in Figure 3A, after 19 h, wt cells migrated into and completely filled the scratched cell-free space. As reported previously, migration of vav1 null macrophages was slower than wt under basal conditions (Wells ; Spurrell ), and, as we reported previously (Park ), NO2LDL treatment inhibited macrophage migration of wt but not cd36 null cells by 50% (Figure 3, A and B). NO2LDL treatment had significantly less impact on migration of vav1 null macrophages compared with wt (Figure 3C). Because macrophages also express Vav3 (Sindrilaru ), we tested Vav1,3 double-null macrophages and found that, like cd36 null cells, Vav1,3 double-null macrophages were not inhibited by NO2LDL (Figure 3D). The bar graphs in Figure 3E show quantitative data from multiple migration experiments.

FIGURE 3:

CD36-dependent inhibition of macrophage migration by oxLDL requires Vav family GEFs. Macrophages from wt (A), cd36 null (B), vav1 null (C), or vav1,3 double-null (D) mice were plated onto a glass bottom dish. After 18 h, the confluent cell layer was scratched and treated with NO2LDL or control LDL at 50 μg/ml. Macrophages migrating into the free space were counted after 19 h (right panels). (E) Quantitative analysis of migrated cells. Open bars are untreated macrophages, and filled bars are NO2LDL-treated macrophages. Lines indicate SD. Data are representative of three separate experiments. In every experiment, three randomly chosen fields were recorded by time-lapse microscopy. (F) Macrophages from wt, cd36 null, and vav1 null mice were added to the upper chamber of the transwell with or without NO2LDL (50 μg/ml) and were allowed to migrate through the porous membrane into the lower chamber containing medium with MCP-1. Migrated cells on the lower side of the membrane stained with DAPI were counted under a fluorescence microscope (100× magnification) and compared.

We also performed a modified Boyden chamber migration assay to see whether this effect of oxLDL inhibits chemoattractant-directed migration of macrophages. We placed murine macrophages with or without NO2LDL onto the upper chamber and allowed migration toward the lower chamber containing monocyte chemotaxis protein-1 (MCP-1). Macrophage migration was facilitated by 1.4-fold when MCP-1 at 20 ng/ml was placed in the lower chamber. NO2LDL treatment inhibited MCP-1–directed migration of wt macrophages but not that of cd36 null cells and vav1 null cells (Figure 3F).

OxLDL induces MRLC dephosphorylation

To evaluate mechanisms by which NO2LDL induced lamellipodial retraction and loss of cell polarity, we determined the effect of NO2LDL on activity of nonmuscle myosin II, a cell polarity determinant that is required to generate lamellipodial traction force (Phillips ; Vicente-Manzanares ). Western blot assays to detect activating phosphorylation of T18/S19 in MRLC (Ikebe and Hartshorne, 1985; Ikebe ) showed that NO2LDL treatment decreased the levels of phosphorylation by 60% in wt cells but not in macrophages from cd36 null or vav1 null mice (Figure 4A; n = 8, p < 0.05). NO2LDL also induced a 60% decrease in phosphorylation of MRLC in human peripheral blood monocyte–derived macrophages. This decrease was blocked by an inhibitory anti-CD36 monoclonal antibody and was not observed using NO2(–)LDL control (Figure 4, B and C).

FIGURE 4:

OxLDL-CD36 interaction inhibits nonmuscle myosin II activity by dephosphorylating MRLC. (A) Wt, cd36 null, and vav1 null macrophages were incubated with NO2LDL at 50 μg/ml for the indicated times and lysed. The lysates were analyzed by Western blot to detect MRLC using an antibody specific for the Thr-18/Ser-19–phosphorylated form. Anti-α-tubulin was used as a loading control. The bar graph shows the fold changes compared with untreated cells (* p < 0.05). Data are representative of eight experiments. (B) Human peripheral monocyte–derived macrophages were treated with an isotype control IgG or inhibitory anti-CD36 monoclonal antibody. After incubating with NO2LDL, macrophages were analyzed as in (A). (C) Phosphorylated MRLC was quantified in macrophages treated with NO2LDL, NO2(–)LDL, or medium alone. (D) Myosin IIa (left) and IIb (right) expression was detected in wt macrophages, COS-7 cells, and HeLa cells by Western blot. Data are representative of three (B and C) or two (D) experiments.

Nonmuscle myosin II has three different heavy chain isoforms, IIa, IIb, and IIc, and leukocytes are generally known to express myosin IIa (Simons ). To determine which isoform is most affected by NO2LDL, we assessed specific isoform expression by Western blot and found that macrophages expressed high levels of nonmuscle myosin IIa and low levels of IIb (Figure 4D). No differences were observed comparing wt to cd36 null cells. We thus concluded that the cytoskeletal changes derived from myosin inactivation in macrophages were likely due to perturbed function of myosin IIa.

The small-molecular-weight G protein, Rac, is activated by oxLDL

To evaluate the mechanism by which MRLC dephosphorylation was induced by NO2LDL, we used enzyme-linked immunosorbent assays (ELISAs) to detect the active GTP-bound forms of Rac and RhoA, and showed a dynamic increase in GTP-bound Rac in wt, but not in cd36 null or vav1 null macrophages after exposure to NO2LDL (Figure 5A). GTP-bound RhoA was not affected by NO2LDL (Figure 5B), nor was the level of phosphorylated myosin binding subunit (Supplemental Figure S3), a downstream substrate for active RhoA (Ito ). We also performed immunoprecipitations of macrophage cell lysates with anti-Rac antibody to assess the physical association of Vav and Rac. Figure 5C shows that anti-Rac antibody coprecipitated Vav and that NO2LDL treatment increased the amount of Vav coprecipitated by twofold within 2 min (n = 3, p = 0.18).

FIGURE 5:

OxLDL-CD36 interaction induces Vav-Rac interaction and Rac activation. (A) GTP-bound Rac was measured by ELISA in wt, cd36 null, and vav1 null macrophages treated with NO2LDL at 50 μg/ml for the indicated times. (B) GTP-bound Rho was measured by ELISA in wt and Cd36 null macrophages treated as in (A). Data are representative of five (A) or three (B) separate experiments. (C) Rac1 was immunoprecipitated from wt macrophages exposed to NO2LDL for the indicated times. Immunoprecipitates were then analyzed by Western blot with anti-Vav (top) and anti-Rac1 (bottom) monoclonal antibodies. Bar graph shows quantitative analysis of the scanned blots (n = 3).

OxLDL inhibits myosin activity through Rac-mediated inhibition of MLCK

We used a pharmacological approach to confirm that NO2LDL inactivated nonmuscle myosin II by activating Rac. As shown in the model in Figure 6A, calyculin-A is a myosin activator that increases MRLC phosphorylation by inhibiting MP (Kato ; Takai ; Ito ), thus bypassing the effect of myosin inhibitors such as Y27632 that act upstream of the phosphatase by inhibiting Rho and/or Rho kinase (ROK) (Shabir et al., 2004). Calyculin A, however, does not block myosin inhibitors such as ML-7 that act through the MLCK pathway (Fazal ). We thus incubated macrophages from wt mice with NO2LDL in the presence or absence of calyculin-A. Western blot (Figure 6B; Supplemental Figure S4) showed that calyculin-A did not block the MRLC dephosphorylating effect of NO2LDL (black arrow). As expected, it did block the effect of Y27632 (gray arrow) but not ML-7 (white arrow). Calyculin-A did not affect the level of phosphorylated MRLC by itself (Supplemental Figure S4). These results suggest that the activity of NO2LDL is most likely mediated by inhibition of the MLCK pathway, as would be expected of an agent that activates Rac. MLCK is known to be inactivated by phosphorylation of its residues of Ser-439 and Ser-991 by PAK, a downstream effector of Rac (Goeckeler ; Lei ). Western blots showed that macrophages indeed had higher levels of phosphorylated (Ser-439) MLCK after exposure to NO2LDL (Figure 6C), confirming that an NO2LDL-induced signaling pathway inhibits myosin light chain phosphorylation through inactivation of MLCK.

FIGURE 6:

OxLDL-induced dephosphorylation of MRLC is not blocked by calyculin-A. (A) Model of the MRLC phosphorylation/dephosphorylation cycles showing points where NO2LDL and the pharmacologic agents calyculin-A, Y27632, and ML-7 act. Y27632 inhibits myosin through inhibition of Rho and ROK, whereas ML-7 works through MLCK. Calyculin-A inhibits MP and thus reverses the effect of myosin inhibitors that function through Rho or ROK, but not through MLCK. (B) Wt macrophages were treated with NO2LDL at 50 μg/ml followed by treatment with 3 nM calyculin-A, and cell lysates were then examined by Western blot to detect phosphorylated MRLC (Ser-19). To demonstrate that the calyculin-A functioned as predicted, cells were also incubated with 2 μM Y27632 or 12 μM ML-7 and 3 nM calyculin-A and examined as in (B). Data are representative of three experiments. (C) Macrophages from wt mice were treated with NO2LDL as above and examined by Western blot using an antibody specific for the S439-phosphorylated form of MLCK. Data are representative of two experiments.

Blocking oxLDL-induced Rac activation by 6-thioguanine triphosphate (6-thio-GTP) or RNA interference (RNAi) results in nonmuscle myosin II inhibition

We next evaluated whether inhibition of Rac activation blocked the inhibitory effect of NO2LDL on MRLC phosphorylation. Six-thio-GTP, a metabolite of azathioprine, is an inhibitor that blocks Vav binding to Rac (Poppe ). Macrophages pretreated with 6-thio-GTP did not show an increase in GTP-bound Rac after NO2LDL exposure (Figure 7A). Furthermore, Western blots for phosphorylated (Ser-19) MRLC showed that NO2LDL failed to inhibit phosphorylation of MRLC in 6-thio-GTP pretreated macrophages (Figure 7, B and C). The inhibitory effect of 6-thio-GTP was time dependent; 4-h incubation incompletely blocked the effect of NO2LDL; however, 16 h of incubation completely blocked it (Figure 7C). Sixteen-hour incubation with 6-thio-GTP did not affect cell viability; this finding was verified by trypan blue exclusion and Annexin V staining. Macrophages were also incubated with NSC23766, a Rac inhibitor that does not influence Vav but inhibits binding of alternative GEFs, including Tiam-1 and Trio (Gao ). Unlike 6-thio-GTP, NSC23766 did not block the effect of NO2LDL on Rac activation (Figure 7A) or block the inhibitory effect of NO2LDL on MRLC phosphorylation, further supporting a key role for Vav in NO2LDL-induced inhibition of nonmuscle myosin II (Figure 7B).

FIGURE 7:

Effects of oxLDL on Rac, MRLC, cell polarity, and migration are blocked by 6-thio-GTP. (A) Wt macrophages were incubated with 5 μM 6-thio-GTP or 50 μM NSC23766 and treated with NO2LDL at 50 μg/ml for 20 min. GTP-bound Rac was measured as described in Figure 5. Data are representative of three experiments. (B) Macrophages were treated with 5 μM 6-thio-GTP, 50 μM NSC 23766, or vehicle for 16 h and then exposed to NO2LDL. Phosphorylated MRLC (S19) was detected by Western blot. Data are representative of two experiments. (C) Macrophages were incubated with 6-thio-GTP for the indicated times and then analyzed for phosphorylated MRLC (S19) as in (C). Data are representative of three experiments. (D) Wt macrophages were incubated with 6-thio-GTP or DMSO (for untreated control) and treated with NO2LDL. Time-lapse images were taken as described in Figure 1. Solid arrows indicate the front end, and dashed arrows indicate the rear end of cells. White scale bar = 10 μm. (E) Retraction fiber counts, ruffle area, and average velocity were calculated as in Figure 2 after exposure to NO2LDL in the presence or absence of 5 μM 6-thio-GTP. (D and E) Data are representative of three separate experiments analyzing 12 cells. (F) Confluent cell layers of macrophages were pretreated with 6-thio-GTP, scratched, and exposed to NO2LDL. Bar graphs show migrated cell numbers determined as in Figure 3. Data are representative of three separate experiments, and three randomly chosen fields were recorded by time-lapse microscopy.

We also confirmed the role of Rac in NO2LDL-mediated inhibition of MRLC phosphorylation by using RNAi to knock down Rac1 expression in macrophages. The Rac1 RNAi down-regulated Rac1 protein expression by 54% (Supplemental Figure S5A) compared with a scrambled sequence control RNA. Western blot for phosphorylated (Ser-19) MRLC showed that NO2LDL decreased the level of phosphorylated (Ser-19) MRLC in control macrophages, whereas it did not in the cells treated with the RNAi (Supplemental Figure S5B). Thus we conclude that Rac1 was required for inhibition of phosphorylation (Ser-19) of MRLC by NO2LDL.

Six-thio-GTP blocks the effect of oxLDL on cell polarity and restores macrophage migration

Having shown that 6-thio-GTP blocks the biochemical signaling effects of NO2LDL on Rac activation and MRLC phosphorylation, we next used live cell imaging to evaluate its effects on NO2LDL-induced loss of cell polarity. Six-thio-GTP did not influence spontaneous polarization or locomotion of macrophages (Figure 7D), but it blocked the effects of NO2LDL on lamellipodial retraction and retraction fiber formation (Figure 7, D and E; Supplemental Videos 7 and 8). Dynamic movement of membrane and cell velocity were similarly unaffected by NO2LDL when macrophages were treated with 6-thio-GTP (Figure 7E). The scratch wound migration assay also showed that migration of 6-thio-GTP–treated macrophages was not inhibited by NO2LDL (Figure 7F). Six-thio-GTP–treated macrophages thus maintained cell polarity and migrating ability despite the presence of NO2LDL. These studies show that NO2LDL-induced cytoskeletal changes can be reversed by the blockade of the Vav-Rac interaction.

MLCK inhibition by ML-7 caused loss of cell polarity, similar to the effect of oxLDL

We hypothesized that any pharmacological myosin inhibitor that replicated the signaling effect of NO2LDL would also induce loss of macrophage cell polarity. ML-7, an MLCK inhibitor that, similar to NO2LDL, caused decreased phosphorylation of MRLC also induced retraction of front end lamellipodia and loss of polarity when incubated with macrophages (Figures 6B and 8A). However, y27632, a myosin inhibitor that acts by inhibiting ROK, did not induce loss of cell polarity at a concentration that inhibited phosphorylation of MRLC (Figures 6B and 8B). These experiments confirm that inhibition of nonmuscle myosin II by inhibiting MLCK causes loss of macrophage cell polarity.

FIGURE 8:

MLCK inhibition by ML-7 caused loss of cell polarity, similar to the effect of oxLDL. (A) Resident peritoneal macrophages from wt mice were plated onto a serum-coated, glass bottom dish and allowed to spontaneously polarize. Time-lapse images were taken before and after the addition of 12 μM ML-7. (B) Time-lapse images were taken before and after the addition of 2 μM y27632 onto wt macrophages as described in (A). (A and B) Solid arrows indicate the front end lamellipodia, and dashed arrows indicate the rear end. Left, representative images taken 40 min after the addition of ML-7 or y27632. White scale bar = 10 μm.

DISCUSSION

Dysfunctional macrophage migration leading to loss of ability to migrate out of inflammatory microenvironments contributes to the pathogenesis of many important chronic diseases, including obesity and atherosclerosis (Ross, 1999; Glass and Witzum, 2001; Lumeng ). These conditions are in contrast to acute inflammation, which resolves in part through inflammatory cell egress. Adipose tissue macrophages in obese/diabetic subjects and macrophages in atherosclerotic plaque share common features. In both settings, macrophages have high intracellular lipid content, and their accumulation is proportionate to the extent of disease (Ross, 1999; Glass and Witzum, 2001; Weisberg ; Lumeng ; Zeyda and Stulnig, 2007). Importantly, conditions that reverse disease burden are associated with a reduction in the number of macrophages (Weisberg ; Clement ; Cancello ). The molecular mechanisms mediating macrophage “trapping” are incompletely understood. Elegant studies from Randolph and colleagues showed that transplantation of atherosclerotic arterial vessel segments from hyperlipidemic mice into normal mice led to migration of lipid-laden macrophages from plaque to regional lymph nodes and regression of plaque, suggesting that cues from the disease tissue microenvironment undoubtedly play a role in trapping (Llodra ). These studies also suggest that targeting the mechanisms responsible for macrophage trapping may lead to a new therapeutic strategy that induces reversal of arterial inflammation.

On the basis of older studies showing that oxLDL inhibits macrophage migration in vitro and that specific forms of oxLDL known to be ligands for the scavenger receptor CD36 are present in obese adipose tissue and advanced atheromatous plaque (Quinn ; Nicholson ), we hypothesized that interaction of oxLDL with CD36 triggers a signaling cascade that inhibits migration. In support of this hypothesis, we recently demonstrated that oxLDL-mediated inhibition of macrophage migration in vitro and in vivo was dependent on CD36 expression (Park ). We defined some of the mechanisms underlying this phenomenon by showing that CD36 signals through src-family kinases and focal adhesion kinase (FAK) to increase actin polymerization and that CD36-dependent generation of intracellular oxidant stress leads to oxidative inactivation of the protein tyrosine phosphatase responsible for terminating FAK activity. The loss of ability to coordinate actin assembly and disassembly led to enhanced cell spreading and inability to migrate (Park ).

We now extend these studies in an important new direction by using live cell imaging to define a previously unknown macrophage signaling pathway involved in regulating cell polarity that is triggered by oxLDL interaction with CD36. Cellular movement starts with the establishment of protrusive forces for membrane extension and traction forces for contraction. (Lauffenburger & Horwitz, 1996). The process of cell polarization has been studied for decades, but the relationship between all participating molecules and their specific roles remains incompletely understood. Small-molecular-weight G proteins including Rho, Rac, and Cdc42 are known to participate (Nobes and Hall, 1999; Wittmann and Waterman-Storer, 2001), and there is abundant evidence that active Rac initiates and maintains directional front end protrusion (Ridley and Hall, 1992; Ridley ; Kraynov ). In contrast, Rho is more active at the sides and the rear of the cell, antagonizing the function of Rac (Kraynov ; Wong ). Our data suggest that this simplified view of cell polarity may in fact be incomplete. We showed that activated Rac, induced by oxLDL binding to CD36, breaks cellular asymmetry at the front end by inducing lamellipodial retraction. This effect was driven by Vav-dependent activation of Rac1 with subsequent inhibition of nonmuscle myosin II. Whereas previous studies showed that reduction in myosin-mediated actomyosin contractility induced by RhoA inhibition enhanced retraction of the trailing edge (Omelchenko ; Wong ), our data showed that Rac1-dependent inhibition of myosin II by oxLDL/CD36 induced retraction at the front end.

Our experiments showed that oxLDL inhibits both random migration and chemotaxis of macrophages by modulating cell polarity, the prerequisite for cellular motility. Asymmetric morphology with defined leading and trailing edges is generated by polarized intracellular signaling that orients protrusion of the front end lamellipodia, integrin-mediated adhesion of the extended membranes, and detachment of the rear end. This sequence of steps, known as the cell motility cycle, occurs in response to a variety of factors. However, it is not clear how this basic motility mechanism is coupled to a steering mechanism that directs cells toward a certain environmental cue and maintains directionally persistent migration. Directional migration has two components: intrinsic cell directionality of migration and external regulation. Cells undergo directed migration when an asymmetric guidance cue such as chemical gradient is presented. The mechanism by which cells maintain directionally persistent migration toward the cues has not been clearly defined. One explanatory mechanism is the “compass model,” which proposes that directional sensing couples cellular locomotion by generating new pseudopods (lamellipods) toward the cues (Arrieumerlou and Meyer, 2005). However, a study by Andrew and Insall (2007) contradicts this model by showing evidence that pseudopod generation is controlled independently of chemotactic signaling and directional migration occurs by selecting the most accurately directing preexisting pseudopod. Thus cells bifurcate their pseudopods by suppressing the lateral pseudopods. We have shown that oxLDL/CD36-mediated signaling induced retraction of protrusive lamellipodia and thus inhibited both random and chemotactic migration. In this regard, our data are consistent with the latter model. Because the directing pseudopod is driven from the preexisting pseudopods, inhibition of the random protrusion is predicted to inhibit directional selection of pseudopods for chemotactic movement.

Recent studies have revealed that myosin IIa and IIb have distinct functions (Cai ; Chen, 2007). We now show that myosin IIa is the major isoform in macrophages and that it is a critical determinant of cell polarity. Remarkably, retraction of preformed protrusions was induced in polarized macrophages by inactivating myosin II, suggesting that myosin IIa functions in maintaining lamellipodial protrusion and conservation of polarity. Activity of myosin II is regulated by the coordinated activities of Rho and Rac (Zhao and Manser, 2005). Whereas Rho activates myosin II by inhibiting MP (Amano ) and/or activating ROK (Kimura ; Kawano ), the effects of Rac on myosin activity are not clearly defined. Rac activates PAK (del Pozo ), but two opposing views on the effect of PAK on MRLC phosphorylation have been published. Some studies report that PAK can phosphorylate MRLC at Ser-19 (Chew ), whereas others report that PAK inhibits MLCK activity and decreases phosphorylation of MRLC (Sanders ; Goeckeler ). It has also been shown that Rac can activate MLCK through a cascade involving extracellular signal-regulated kinase (Robinson and Cobb, 1997). Some of these confounding data may have resulted from the use of overexpression of constitutively active or dominant negative Rac or PAK mutants to probe the system. Our studies, however, took advantage of a natural endogenous receptor-ligand system to modulate the pathway. They showed that oxLDL decreases the activity of myosin II by Rac-mediated inhibition of MRLC phosphorylation by MLCK. Previous studies showing that MLCK regulates MRLC phosphorylation at the cellular periphery while ROK functions at the center of cells (Totsukawa , 2004) are consistent with the live cell imaging demonstrating that oxLDL-induced retraction is localized to the lamellipodial edges (Figure 1; Supplemental Video 2; Supplemental Figure S1). This finding is also supported by our experiments with ML-7, an MLCK inhibitor and y27632, a ROCK inhibitor. ML-7 induced retraction of lamellipodia and loss of macrophage cell polarity as NO2LDL, but y27632 did not show the effect at a concentration that decreased MRLC phosphorylation (Figures 6B and 8).

Studies from our lab and others showed that Vav family members are downstream effectors of CD36 signaling (Wilkinson ; Rahaman ). Vav null macrophages have previously been shown to have defects in spreading (Spurrell ) and adhesion-induced Rac and Rho activation (Bhavsar ), and the cellular responses to oxLDL-induced Vav activation observed in our studies are opposite to these effects, consistent with a critical role for Vav in mediating oxLDL-CD36 cytoskeletal signaling. We found that 6-thio-GTP, a pharmacologic agent that blocks binding of Vav to Rac (Poppe ), abrogated oxLDL effects on cell polarity and MRLC phosphorylation. The relevance of Vav to these processes was confirmed by showing that genetic ablation of Vav1 and Vav3 in macrophages phenocopied CD36 deficiency with regard to polarity and migration responses to oxLDL. Individual Vav family proteins are known to have both distinct and overlapping functions (Hornstein ; Sindrilaru ), and indeed we found that Vav3 partially compensates for deficiency of Vav1 in macrophages: vav1 null cells had delayed and partial response to oxLDL in terms of dynamic membrane movement and cell migration. These effects were blocked by concomitant vav3 deletion.

Our data provide additional mechanistic support for the proatherogenic role of macrophage CD36 (Febbraio , 2004; Guy ; Kuchibhotla ) and support a model by which oxLDL/CD36 interactions activate Rac through src-family kinase activation of Vav. Activated Rac then inhibits nonmuscle myosin II by inhibiting phosphorylation of MRLC. Inactivated myosin II cannot generate tension or traction force on lamellipodia resulting in lamellipodial retraction. Lamellipodial retraction leads to a loss of cell polarity, an essential requirement for macrophage migration. Blockade of any step in this process—including CD36 deletion, Vav deletion, or Rac inhibition—prevents the effect of oxLDL and restores macrophage migration. These studies suggest novel ways to promote mobilization of lipid-laden macrophages and induce regression of atherosclerosis. We also suggest that macrophage trapping as a common phenomenon of oxLDL-enriched environments may be a shared target for the treatment of atherosclerosis and adipose tissue inflammation, two major components of metabolic syndrome.

MATERIALS AND METHODS

Reagents and antibodies

LDL prepared from human plasma by density gradient ultracentrifugation (Hatch, 1968) was oxidatively modified by incubation in a buffer containing 50 mM sodium phosphate (pH 7.0) and 100 μM diethylene triamine pentaacetate (DTPA) with 30 nM MPO, 100 μg of glucose, glucose oxidase at 20 ng/ml (grade II; Boehringer Mannheim Biochemicals, Penzberg, Germany), and 0.5 mM NaNO2 at 37ºC for 8 h (NO2LDL) (Podrez ). The oxidation reaction was terminated by the addition of 40 μM butylated hydroxyl-toluene and 300 nM catalase to the reaction mixture. We also prepared a control LDL termed NO2(–)LDL, by incubating LDL with all the components mentioned above but NaNO2. Antibodies for phosphorylated MRLC (T18/S19) and phosphorylated MRLC (S19) were purchased from Cell Signaling Technology (Danvers, MA); antibodies for actin and α-tubulin were obtained from Santa Cruz Biotechnology (Santa Cruz, CA); 6-thio-GTP was purchased from Jena Bioscience GmbH (Jena, Germany). COS-7 and HeLa cell lysates, anti-myosin IIa and anti-myosin IIb antibodies, y27632, ML-7, and NSC23766 were generously provided by Thomas Egelhoff, Department of Cell Biology, Cleveland Clinic.

Animals and cells

Vav1 null mice were provided by J. Rivera (National Institutes of Health [NIH], Bethesda, MD), and Vav1/3 double-null mice were obtained from W. Swat (Washington University School of Medicine, St. Louis, MO). Background-matched mice were used as controls. Resident peritoneal macrophages were collected by lavage and selected by removing unbound cells 30 min after plating onto surfaces devoid of matrix proteins. Thioglycollate-elicited peritoneal macrophages were collected by lavage 4 d after intraperitoneal injection of thioglycollate. Macrophages were cultured in RPMI containing 10% FBS. Human monocytes were isolated from peripheral blood by Ficoll-Hypaque centrifugation and were cultured in RPMI containing human AB serum (10%) for 7 d to allow for macrophage differentiation.

Live cell imaging

Live cell imaging of single cells was performed using TIRF (total internal reflection fluorescence) microscopy (Leica AM TIRF MC System equipped with HCX Plan Apo 100×/1.46 NA Objective Lens; Leica Microsystems, Buffalo Grove, IL). Mouse peritoneal macrophages were plated on a serum-coated glass bottom dish and visualized by transmitted light differential interference contrast imaging. An ImageEM C9100-13 EMCCD camera (Hamamatsu, Shizuoka, Japan) captured an image every 15 s for 1 h before the addition of NO2LDL and for another 1 h after the addition of NO2LDL. The same methods were used for evaluating the effects of ML-7 and y27632. Live cell imaging for cell migration assays was performed using a Leica DMIRB inverted microscope with a 100× objective lens (Leica Microsystems).

Cell migration assay

Scratch wound closure migration assay

Peritoneal macrophages from wt, cd36 null, vav1 null, or vav1,3 double-null mice were plated onto glass bottom 6-well or 12-well plates. After 18 h, confluent monolayers were scratched using a pipette tip and rinsed with PBS. RPMI medium was added with or without NO2LDL. Live cell imaging was used to record the migration of macrophages for 18 h. Images were taken every 5 min and from three randomly chosen locations of each well.

Modified Boyden chamber migration assay

Chemoattractant-directed migration of mouse peritoneal macrophages was measured in a modified Boyden chamber migration assay using Transwell inserts with a 5-μm porous membrane (Corning, Corning, NY). Cells (1.5 × 105) were loaded into the migration chamber with or without lipoproteins including NO2(–)LDL and NO2LDL. Medium containing MCP-1 at 20 ng/ml was placed in some of the lower chambers. After allowing cell migration for 16 h, cells were removed from the upper side of membranes, and nuclei of migratory cells on the lower side of the membrane were stained with 4′,6-diamidino-2-phenylindole (DAPI). The number of migratory cells was counted by fluorescence microscopy (100× magnification).

Image analysis

Microscopic images were analyzed by Image-Pro Plus software (Media Cybernetics, Bethesda, MD). Retraction fibers around the front end lamellipodia were marked in color and counted, as were the ruffle areas as parameters to evaluate dynamic membrane movement. Ruffle area was defined as instant protrusive area from the prior cellular margin taken 15 s earlier. The protrusive areas were counted every 15 s for 1 h before and after the addition of NO2LDL. Pictures from live cell imaging were merged and video clips generated using the software.

Western blot analysis

Mouse peritoneal macrophages incubated with NO2LDL or NO2(–)LDL at 50 μg/ml for the indicated times were lysed with sample buffer containing 4% SDS after treating the cells with 10% TCA. Lysates were separated by SDS–PAGE and transferred to PVDF membranes (Millipore, Billerica, MA). Membranes were probed with antibodies against phosphorylated MRLC, actin, or α-tubulin for normalization. Band intensities were quantified by ImageJ (http://rsbweb.nih.gov/ij) and Gel-Pro Analyzer (Media Cybernetics). Immunoblotting for nonmuscle myosin IIa and IIb was performed as described earlier in text using anti-myosin IIa and anti-myosin IIb antibodies. COS-7 and HeLa cell lysates served as positive or negative controls for myosin IIa or IIb.

Immunoprecipitation

Mouse peritoneal macrophages were lysed in buffer consisting of 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, and 1 mM sodium orthovanadate. Cell lysates were added to protein A/G sepharose beads (Santa Cruz Biotechnology), conjugated with anti-Rac antibody (abcam, Cambridge, MA), and incubated overnight at 4ºC. The sepharose beads were rinsed, resuspended in 2X Laemmli sample buffer (Bio-Rad, Hercules, CA), and heated at 100ºC. After centrifugation, the supernatants were loaded onto SDS–PAGE gels and transferred to PVDF membrane. Immunoblotting for Vav was performed, and ExactaCruz C (Santa Cruz Biotechnology) was used for detection.

Rac and RhoA activity assay

GTP-bound Rac and RhoA were detected by a G-LISA Rac activation assay kit and a G-LISA RhoA activation assay kit from Cytoskeleton (Denver, CO). For these assays, mouse peritoneal macrophages treated with or without NO2LDL were lysed. Cell lysate protein (5–10 μg) was applied to a 96-well plate coated with the p21 binding domain of PAK or the Rho binding domain of Rhotekin, both of which are known to specifically bind to GTP-bound Rac and RhoA, respectively (Burbelo ). After a 30-min incubation, the plates were rinsed and anti-Rac or anti-RhoA antibody was added. Horseradish peroxidase-conjugated secondary antibody was added 1 h later, and the level of antibody binding was determined by a colorimetric method.