Transforming growth factor-β1 regulates macrophage migration via RhoA

Transforming growth factor (TGF)–β regulates diverse cellular functions, including tissue differentiation, cell proliferation, and cell migration. Monocytes/macrophages, in particular, secrete TGF-β, which in turn stimulates numerous responses: production of a variety of cytokines, including interleukin-1α (IL-1α) and -β (IL-1β), tumor necrosis factor (TNF)–α, platelet-derived growth factor (PDGF)–BB, and basic fibroblast growth factor (bFGF); recruitment of monocytes to sites of injury or inflammation; phagocytic activity (by up-regulating the expression of cell-surface FcγRIII); and the expression of several integrin receptors on monocytes, including leukocyte function–associated antigen-1 (LFA-1: integrin αLβ2), α3β1, and α5β1, thereby increasing their cell-cell and cell-matrix interactions.¹  These observation indicate a proinflammatory function for TGF-β on monocytes.²  In contrast to its activating effects on peripheral blood monocytes, TGF-β reduces the host response to a variety of inflammatory stimuli and is a potent immunosuppressive, anti-inflammatory, and macrophage deactivating agent.³  Resting monocytes express high levels of TGF-β type 1 and 2 receptors, whereas receptor levels decline as cells mature and are then activated by agents such as lipopolysaccharide (LPS) and interferon-γ (IFN-γ).¹ 

The functional complex of TGF-β1 receptors at the cell surface is composed of 2 type 2 (TβRII) and 2 type 1 (TβRI) transmembrane Ser/Thr kinase receptors.⁴  Receptor-activated Smads (R-smads: Smad1, Smad2, Smad3, Smad5, and Smad 8), which are phosphorylated by type 1 receptors, are released from the receptor complex to form a heterotrimeric complex of 2 R-Smads and a common Smad4 (Co-Smad); the complex then translocates to the nucleus, where it regulates transcription. The structurally distinct Smads, Smad6 and Smad7, act as inhibitory Smads (I-Smads) by competing with R-Smads for receptors.⁵  The expression of I-Smads is strongly regulated by extracellular signals, and the induction of Smad6 and Smad7 expression by TGF-β1 reveals an inhibitory feedback mechanism for ligand-induced signaling.⁶  In addition to the R-Smad/Co-Smad activation pathway, TGF-β can activate the extracellular signal-regulated kinase (ERK), c-Jun N–terminal kinase (JNK), and p38MAPK pathways, the last 2 of which are activated via TGF-β–activated kinase 1 (TAK1).⁴ 

Rho GTPases regulate the actin cytoskeleton, cell polarity, gene expression, microtubule dynamics, and vesicular trafficking.⁷  Regulation of the nucleotide-bound state of RhoGTPases, alternative cycling between active GTP- and inactive GDP-bound states, is accomplished by the action of 3 major classes of proteins: guanine nucleotide exchange factors (GEFs), guanine nucleotide dissociation inhibitors (GDIs), and GTPase activating proteins (GAPs).⁷  Recently, a number of studies have demonstrated that TGF-β1 activates Rho GTPases. It induces lamellipodia and stress fibers in human prostate carcinoma cells, which requires activities of Cdc42 and RhoA.⁸  In skeletal muscle cells, TGF-β activates Rac1 and Cdc42 but inactivates RhoA.⁹  In human lens epithelial cells, it activates both RhoA and Rac1.¹⁰  Furthermore, TGF-β1 induces epithelial to mesenchymal transdifferentiation via a RhoA-dependent mechanism.¹¹  Thus, TGF-β1 activates different GTPases depending on the cell type. On the other hand, Rho GTPases regulate the phagocytosis of apoptotic cells,^12,13  which leads to the resolution of inflammation and is mediated by TGF-β1.¹⁴  From these facts, it can be inferred that TGF-β1 signaling is closely linked with Rho GTPase signaling in inflammatory cells.

In particular, TGF-β recruits monocytes to the site of injury or inflammation through several mechanisms. Therefore, it acts as a chemoattractant for monocytes.²  On the other hand, RhoGTPases play key roles in regulating the cellular responses required for cell migration.¹⁵  Rho and Rac regulate the polymerization of actin to produce stress fibers and lamellipodia, respectively. In addition to stress fibers, Rho regulates the assembly of focal adhesion complexes. Cdc 42 triggers the formation of the actin-based structure found at the cell periphery, filopodia, which includes localization of lamellipodial activity to the leading edge.¹⁶  Leukocyte migration also may be induced by various chemokines. The chemokines are classified mainly into CC and CXC families on the basis of location of N-terminal cysteine residues. In inflammation, the CXC or α-chemokines, including interleukin-8 (IL-8), act mainly on neutrophils, and the CC or β-chemokines, including macrophage inflammatory protein (MIP)–1α and monocyte chemoattractant protein-1 (MCP-1), act mainly on monocytes, lymphocytes, and eosinophils.^17,18 

In this study, we found that RhoA was activated and inactivated in sequence in response to TGF-β1 and showed that its activity is relevant to cell migration both directly and indirectly via the production of chemokines, including MIP-1α and MCP-1. We focused on the mechanism of RhoA inactivation in response to long-term exposure to TGF-β1 and demonstrated that RhoA is in part inactivated as a result of phosphorylation by protein kinase A (PKA) and in part by activation of p190 Rho GTPase activating protein (p190RhoGAP).

PD98059, SB203580, phenylmethylsulfonyl fluoride (PMSF), Triton X-100, glutathione (GSH), formyl Met-Leu-Phe (fMLP), isopropyl-β-d-thiogalactoside (IPTG), LPS, IFN-γ, H89, cycloheximide (CHX), actinomycin D (ActD), cytocalasin D (CytD), ML-7, and Y-27632 were purchased from Sigma Chemical (St Louis, MO). SP600125 was obtained from Tocris (Bristol, United Kingdom). Protein A–agarose bead was from Pierce (Rockford, IL). GSH-Sepharose 4B bead was from Amersham Biosciences (Freiburg, Germany), and Ni-NTA His Bind resin was from Novagen (Madison, WI). Anti-Rac1 antibody was purchased from Upstate Biotechnology (Waltham, MA), and anti-RhoA, anti-Cdc42, anti-Rap1A, anti-RhoGDI, and anti-Smad3 antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA); anti–phospho-ERK, anti–phospho-JNK, anti–phospho-p38 MAPK, anti-ERK, anti–p38 MAPK, and anti-JNK antibodies were from Cell Signaling Technology (Beverly, MA).

Mouse macrophage Raw 264 cells were cultured in Dulbecco modified Eagle medium-F12 (DMEM-F12) containing 5% fetal bovine serum (FBS) and antibiotics (100 units/mL streptomycin and 100 units/mL penicillin: Cambrex Bio Science, Walkersville, MD) at 37°C in 5% CO₂. HL-60 cells (human promyelocytic leukemia cell) were cultured in RPMI 1640 containing 10% FBS and antibiotics. P815 (mastocytoma cells) and 293 (human kidney epithelial cells) cells were cultured in DMEM containing 10% FBS and antibiotics.

To introduce C3 exoenzyme efficiently into cells, we fused the HIV-1 Tat transduction domain (amino acid residues 49-57, KKKRRQRRR) to the N-terminus of the C3 exoenzyme containing a His tag (Tat-C3).^19,20  Purification of Tat-C3 was performed following a previously described method.^19,20  For the transduction of Tat-C3, macrophage cells were grown to confluence in 6-well plates for 4 to 6 hours. The culture medium was replaced with 1 mL serum-free medium, and Raw 264 cells were incubated with 10 μg/mL Tat-C3 for 30 minutes at 37°C.^20,21 

Assays for migration were performed in transwell cell culture chambers with polycarbonate filters (24-well, 8-μm pore, Corning Costar, Cambridge, MA). The filters were coated with 1 mg/mL fibrinogen for 6 hours and dried, and cells in 200 μL DMEM-12 were added to the upper reservoir. One milliliter of DMEM-12 with or without chemotaxis reagents (1 μg/mL LPS, 50 U/mL IFN-γ, 100 nM fMLP, and 5 ng/mL TGF-β1) was placed in the lower reservoir. After 6 hours of migration, cells remaining on the upper membrane were scraped off, and cells that had migrated to the lower membrane were stained by the Giemsa staining methods (Sigma) and counted under an Axiovert 200 microscope (Zeiss, Göttingen, Germany) with a 40×/0.10 NA Achiroplan objective, and photographed with a Photometrics CoolSNAP camera (Roper Scientific, Tucson, AZ). Images were processed with IPLab 3.65a software (IPLab, Guernsey, United Kingdom) and Adobe Photoshop 7.0 (Adobe Systems, Beaverton, OR). Reagents with effects on cytoskeleton rearrangement (CytD, ML-7 and Y-27632), MAPK inhibitors (PD98059, SB203580, and SP600125), H89 (an inhibitor of PKA), CHX, or ActD were preincubated with cells before TGF-β1 was added to the cells. Cells pretreated with reagents, including TGF-β1, were washed, resuspended in 200 μL DMEM-12, and then loaded in the upper reservoir.

Total RNA was extracted from macrophages using RNA plus RNA extraction solution (Quantum, Quebec, Canada). Reverse transcription (RT) of mRNA and the polymerase chain reaction (PCR) were carried out with the Superscript II system (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Primer sequences were as follows: MIP-1α, 5′-GCCCTTGCTGTTCTTCTCTGA-3′ (sense), 5′-GGCAATCAGTTCCAGGTCAGT-3′ (antisense)²² ; glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-AACGGATTTGGCCGTATT-3′ (sense), 5′-ACTGTGGTCATGAGCCCTT-3′ (antisense). The GAPDH primer pair was used as an internal control. PCR products were separated on 1.5% agarose gel, stained with ethidium bromide, and photographed.

MIP-1α and MCP-1 secreted from Raw 264 cells in response to TGF-β1 were quantitatively determined using enzyme-linked immunosorbent assay (ELISA) technique following manufacturer's instruction (R&D Systems, Minneapolis, MN, and BD Bioscience Pharmingen, San Jose, CA, respectively).

Macrophages were washed in phosphate-buffered saline (PBS) and lysed in lysis buffer containing 20 mM Tris-HCl, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA (ethylene glycol tetraacetic acid), 1 mM dithiothreitol, 0.5% deoxycholate, 0.1% SDS, 1.5 mM MgCl₂, 150 mM NaCl, pH 8.0. The lysates were centrifuged at 17 000g for 30 minutes at 4°C, and the supernatants were incubated overnight with 2 μg anti-RhoGDI or anti-RhoA antibody. Protein A–conjugated agarose beads (Pierce) were added for 2 hours. The beads were washed 3 times with 1 mL lysis buffer. Bound proteins were eluted by boiling in 2 × Laemmli sample buffer and subjected to Western blot analysis using anti-RhoA, anti-RhoGDI, or anti–phospho-Ser antibodies.

Ninety percent confluent Raw 264 cells in 6-well dishes (Corning) were transiently cotransfected with 4 μg pcDNA3 plasmid constructs encoding Smad3 (wild-type [WT] Smad3 and the dominant-negative [DN] Smad3) and WT Smad4 (5 μg), RhoA (WT, constitutive active [CA]-G14V, DN-T19N), Cdc42 (WT, CA-G12V, and DN-T17N), or p190RhoGAP (WT, and DN: lacking the GTP-binding domain²³ ), using lipofectamine 2000 transfection reagent. After 24 hours, the cells were incubated with 5 ng/mL TGF-β for the indicated time periods, resuspended, and lysed for IP and Western blot assays. In some cases, the transfected cells also were used for assay of migration.

We washed 2 × 10⁶ cells in 100-mm plates in ice-cold PBS, harvested them, and lysed them in lysis buffer (50 mM Tris-HCl, pH 7.2, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 500 mM NaCl, 10 mM MgCl₂, 5 μg/mL each of leupeptin and aprotinin, and 1 mM PMSF). After centrifugation (16 000g, 15 minutes, 4°C), aliquots of the supernatant were incubated with the GST-Rho binding domain of Rhotekin (GST-RBD), the GST-GTPase binding domain of p21-activated kinase-1 (PAK-1) (GST-PBD), or His-Ral guanine nucleotide dissociation stimulator (GDS). These had been preincubated for 1 hour with 50 μg of GSH-Sepharose beads in the case of GST-fusion proteins or Ni-NTA His-Bind resins for the His-fusion protein. RhoA-GTP also was detected using EZ-Detect Rho activation kit containing GST-RBD (Pierce) following manufacturer's instructions. The beads were incubated with the cell lysates and washed, and the proteins on the beads were run on SDS-PAGE. RhoA, Rac1, Cdc42, and Rap1 were assayed by Western blotting.^20,24  Alteration of proteins detected by Western blot was quantified by densitometer.

Generally, the experiments were performed in triplicate, each experiment was carried out with duplicate or triplicate samples. Data are presented as mean ± SE. Comparison between 2 groups was tested by t test using GraphPad Prizm program (GraphPad Software, San Diego, CA). Difference between 2 groups was statistically significant if P values were less than .05.

To address the effect of TGF-β1 on chemotaxis of Raw 264 cells, we examined migration of the cells in a transwell chamber assay. In the presence of chemotaxis-inducing agents such as LPS, IFN-γ, and TGF-β1, in the bottom chamber, pretreatment with TGF-β1 for 1 hour stimulated migration, whereas 24 hours' pretreatment with TGF-β1 dramatically reduced migration (Figure 1, Figure 2A). This indicates that TGF-β1 has 2 successive effects on the chemotaxis of macrophages: first stimulation, then inhibition. Although fMLP did not enhance migration of the Raw 264 cells pretreated with TGF-β1 for 1 hour, 24 hours' TGF-β1 treatment dramatically reduced the stimulation of migration by fMLP. Pretreatment with TGF-β1 for 1 hour or 24 hours, however, did not significantly affect chemotaxis of other cell types such as 293 cell (human kidney epithelial cell), HL-60 (human promyelocytic leukemia cell) as neutrophil-like cells, and P815 (mastocytoma cell) as mast cells (data not shown).

As cell movement requires dynamic reorganization of the actin cytoskeleton, and this is regulated by RhoGTPases,⁷  we tested whether the migration of cells in response to TGF-β1 requires Rho GTPases. Blockade of endogenous RhoA by 10 μg/mL Tat-C3, which completely modifies the Arg 41 residue of RhoA with ADP-ribosyl group of NAD⁺,¹⁵  markedly inhibited cell migration, whereas 0.001 to 0.1 μg/mL Tat-C3, which modified up to about 50% of RhoA molecules, stimulated migration (Figure 2A,B). The doublet band appears to be unmodified RhoA (lower band) and RhoA modified by Tat-C3 (upper band). In addition, Y-27632 (an inhibitor of ROCK [RhoA-activated kinase]), cytocalasin D (an inhibitor of filamentous actin formation), and ML-7 (an inhibitor of myosin light chain kinase, MLCK) blocked TGF-β1–stimulated cell migration (Figure 2C). These suggest that induction of cell migration by TGF-β1 for 1 hour requires actin-myosin assembly, which is regulated by Rho, ROCK, MLCK, and filamentous actin formation.

To confirm the involvement in cell migration of chemokines secreted by the cells upon exposure to TGF-β1, protein synthesis induced by TGF-β1 was blocked by the addition of CHX and ActD. As shown in Figure 3A, these agents reduced the stimulation of cell migration by 1-hour treatment with TGF-β1. These observations indicate that both TGF-β1 itself, and, secondarily, chemokines released from the cells by TGF-β1 treatment, may contribute to the stimulation of cell migration. MIP-1α and MCP-1 have been described as strong chemotactic factors for monocytes.^17,18  Indeed, we also observed that MIP-1α induced migration of Raw 264 cells (data not shown). We therefore tested whether TGF-β1 induces cell migration by increasing expression of chemokines such as MIP-1α and MCP-1. RAW 264 cells were stimulated with 5 ng/mL TGF-β1 or 1 μg/mL LPS for the indicated times, and MIP-1α mRNA level was determined by RT-PCR. Notably, exposure to TGF-β1 for 1 to 3 hours stimulated MIP-1α gene expression (Figure 3B,C).

Remarkably, H89 reversed both the inhibition of cell migration (Figure 3A) and the production of MIP-1α mRNA caused by exposure to TGF-β1 for 24 hours (Figure 3C), suggesting PKA is implicated in both processes. Although PD98059 abrogated cell migration induced by TGF-β1 (Figure S1A, available at the Blood website; see the Supplemental Figures link at the top of the online article), it did not block the induction of MIP-1α mRNA by TGF-β1 (Figure 3C), suggesting that ERK is probably essential for cell migration induced by TGF-β1 in accordance with previous literature,²⁵  but the involvement of ERK in cell migration is not relevant to the expression of MIP-1α, although ERK also is implicated in c-fos induction via Elk-1 phosphorylation.²⁶ 

In addition, protein levels of MIP-1α and MCP-1 secreted from Raw 264 cells in response to TGF-β1 were measured by ELISA: TGF-β1 increased MIP-1α and MCP-1 with a maximum at 6 to 12 hours (Figure 3D). To confirm that secreted chemokines such as MIP-1α and MCP-1 in response to TGF-β1 play a pivotal role in cell migration, the effect of neutralizing the chemokines with specific antibodies was examined: anti–MIP-1α and MCP-1 antibodies significantly reduced cell migration, suggesting that MIP-1α and MCP-1 actually induce migration of Raw 264 cells in response to TGF-β1 (Figure 3E).

The levels of MIP-1α mRNA induced by TGF-β1 was lowered by Tat-C3 (Figure 4A), suggesting that RhoA also is essential for transcription of these genes in response to TGF-β1. To confirm the requirement for RhoA for TGF-β1–induced MIP-1α transcription, Raw 264 cells were transiently transfected with expression plasmids encoding wild-type (WT)–RhoA, constitutively active (CA)–RhoA (RhoA G14V), and dominant-negative (DN)–RhoA (RhoA T19N) prior to stimulation by TGF-β1. DN-RhoA completely blocked the TGF-β1–induced increase of MIP-1α mRNA, whereas WT-RhoA and CA-RhoA did not affect it (Figure 4B). Furthermore, DN-RhoA reduced a level of MIP-1α protein secreted from Raw 264 cells in response to TGF-β1 (Figure 4C). This result also provides strong evidence that RhoA is required for MIP-1α expression in response to short-term exposure to TGF-β1. Cdc42, a member of the Rho GTPase subfamily, also was examined to determine whether it participates in regulation of MIP-1α expression. In contrast, DN-Cdc42 did not block MIP-1α protein expression in Raw 264 cells in response to TGF-β1 (Figure 4C), suggesting that Cdc42 is not relevant to expression of MIP-1α.

The level of MIP-1α mRNA was increased by exposure to LPS for 6 to 12 hours (Figure 3A) in accord with previous findings.²⁷  Exposure to LPS plus TGF-β1 for 6 hours, however, decreased MIP-1α mRNA levels compared with LPS alone (Figure S2). Thus, TGF-β1 appeared to inhibit induction of MIP-1α expression by LPS. Tat-C3 also reduced expression of MIP-1α in response to LPS (Figure S2), indicating that activated RhoA also is required for transcription of these genes induced by LPS. Remarkably, H89, an inhibitor of PKA, reversed the inhibition of MIP-1α expression by TGF-β1 (Figure S2), suggesting that PKA is required for the inhibition of MIP-1α expression. However, H-89 had no effect on the expression of MIP-1α in response to short-term exposure to TGF-β1 (Figure 4A).

To see whether RhoGTPases are activated by TGF-β1 in macrophage cells, we performed pull-down assays using GST-fusion of the RhoGTPases binding domains of various effectors of RhoGTPases. Cells were treated with TGF-β1 for various times, and GST-PAK-CRIB and GST-Rhotekin-RBD were incubated with the cell lysates to isolate GTP-loaded Rac1, Cdc42, and RhoA of the cells. Rac1 was found to be in the GTP-bound form in the absence of stimulation, and TGF-β1 did not alter its level except marginally at 30 minutes. In contrast, the GTP-bound forms of RhoA and Cdc42 increased from 10 to 60 minutes and were strikingly reduced after 12 hours (Figure 5, Figure 6A).

GTPase activating proteins (GAPs) function by stimulating GTP hydrolyzing activity of their substrate GTPases, thereby inactivating them.^7,22  Among the numerous RhoGAPs, p190RhoGAP accounts for a substantial fraction of the total Rho inhibitory activity in cultured cells.²²  Thus, p190RhoGAP appears to be a strong candidate for the RhoA inactivation upon long-term exposure to TGF-β1, and we tried to verify this assumption. Indeed, WT p190RhoGAP abrogated cell migration induced by TGF-β1. DN p190RhoGAP, however, significantly increased migration of the cells pretreated with TGF-β1 for 24 hours compared with that of control cells (Figure 6B). As shown in Figure 6A, TGF-β1 activated RhoA by 30 minutes and thereafter inactivated it. When WT p190RhoGAP was overexpressed, TGF-β1 did not activate RhoA, probably because p190RhoGAP accelerates the hydrolysis of GTP bound to RhoA. However, in the presence of overexpressed DN p190RhoGAP, which lacks the GTP-binding domain²²  and therefore lacks GAP activity, the active GTP-bound state of RhoA was sustained, except that it slightly decreased at 24 hours. This suggests that long-term treatment with TGF-β1 may inactivate RhoA via p190RhoGAP.

PKA inhibits RhoA activation and facilitates RhoA-RhoGDI complex formation²⁸  during neurite outgrowth²⁹  and inhibition of sustained smooth muscle contraction.³⁰  In addition, TGF-β–induced Smad3/Smad4 complexes directly activate PKA.³¹  Therefore, we asked whether TGF-β1 induces RhoA phosphorylation and so inactivates RhoA in macrophages. To measure phosphorylation of RhoA by PKA or in response to TGF-β1, cells were treated with 100 μM forskolin to activate adenylate cyclase or with 5 ng/mL TGF-β1, and phosphorylation of RhoA was measured by Western blotting with anti-RhoA antibody after immunoprecipitation of phospho-Ser proteins from cell lysates with anti–phospho-Ser antibody. Forskolin-mediated phosphorylation of RhoA was evident after 20 minutes and sustained for 24 hours. TGF-β1 also caused sustained phosphorylation of RhoA (Figure 7A). To confirm that this phosphorylation of RhoA was mediated by PKA, macrophages were treated with H89 in the absence and presence of TGF-β1. As shown in Figure 7B, both TGF-β1 and forskolin induced phosphorylation of RhoA, and H89 prevented it, suggesting that RhoA is phosphorylated by PKA in response to TGF-β1. To confirm the involvement of PKA in the inactivation of RhoA in response to TGF-β1, H89 was administered to macrophages in the presence of TGF-β1. This reversed the inactivation of RhoA in response to TGF-β1 for 12 to 24 hours, indicating that PKA inactivates RhoA (Figure 7C) and that phosphorylation of RhoA by PKA may be engaged in a regulatory step in TGF-β1 signaling. H89, however, could not reverse Cdc42 inactivation in response to TGF-β1 treatment for 12 to 24 hours (data not shown).

To see whether exposure to TGF-β1 stimulates interaction between RhoA and RhoGDI, we performed coimmunoprecipitation experiments using anti-RhoGDI antibody. The amount of RhoA bound to RhoGDI was higher in cells stimulated with TGF-β1 for 24 hours compared with control cells (Figure 7D), suggesting that RhoA was inactivated by TGF-β1 treatment for 24 hours. However, the stimulation of RhoA-RhoGDI complex formation by TGF-β1 treatment for 24 hours was reduced by H89, indicating that PKA was responsible for RhoA-RhoGDI complex formation.

We have shown that short-term treatment with TGF-β1 stimulated the migration of macrophages and raised a level of GTP-RhoA and mRNA and protein levels of MIP-1α, whereas long-term exposure decreased both effects. Furthermore, long-term treatment of TGF-β1 suppressed the mRNA level of MIP-1α induced by LPS, which is known to cause inflammation (Figure S2). These results are in accord with previous evidence that TGF-β1 functions subsequently as a stimulatory and an inhibitory cytokine.^1,2 

Although TGF-β1 is a potent chemoattractant for monocytes and macrophages,^1,2  the role of activation of RhoA and Cdc42 in response to TGF-β1 is not clear. To demonstrate that TGF-β1 controls macrophage migration via RhoA, we provided evidence that Tat-C3 blocked cell migration (Figure 2) and expression of MIP-1α (Figure 3), that DN-RhoA reduced mRNA and secreted protein levels of MIP-1α, and that TGF-β1 treatment regulated RhoA activity (Figure 5). Thus, it is likely that RhoA is essential for the chemokine production leading to macrophage activation and migration in response to TGF-β1 signaling. Nevertheless, CA-RhoA inhibited cell migration, whereas WT-RhoA and DN-RhoA did not (data not shown). The reason DN-RhoA did not reduce cell migration may be that DN-RhoAcould not completely block endogenous RhoA, probably because of partial transfection: transfection efficiency was about 30% to 40% in Raw 264 cells. Indeed, RhoA inhibited by 0.1 μg/mL Tat-C3 up to 50% did not cause the reduction of cell migration (Figure 2). CA-RhoA may interfere with cell migration because it induces tighter confocal adhesion,³²  and it may persist for a long time bound to its effector proteins, preventing cycling between the GTP- and GDP-bound states of RhoA and inhibiting cell migration.

Reversibility of focal complex formation/adhesion is important for cell migration: a high level of integrin-mediated adhesion inhibits cell migration because of the strength of the attachment to the extracellular matrix, and this correlates with high levels of Rho activity.^15,33  However, contraction of the bodies of moving cells is dependent on actomyosin contractility and is regulated by RhoA.³⁴  Therefore, reducing RhoA activity has opposing effects: it promotes migration by lowering adhesion, but decreases migration by inhibiting cell body contraction.¹⁵ 

In less-adherent cells lacking focal adhesion, such as macrophages, neutrophils, and various cancer cell lines, RhoA does not affect adhesion but induces cell body contraction, and in these cases RhoA is clearly required for cell migration.¹⁵  DN-Cdc42 could not reduce secreted protein levels of MIP-1α (Figure 4C). This indicates that Cdc42 may be relatively less important than RhoA to stimulate cell migration in response to TGF-β1 in macrophages, although short-term treatment of TGF-β1 activates Cdc42 (Figure 5). In macrophages, Rho and Rac are required for the process of cell migration, whereas Cdc42 is essential for cells to respond to a gradient of chemoattractant such as colony stimulatory factor-1 (CSF-1) but not essential for cell locomotion.³⁵ 

Short-term treatment with TGF-β1 increased MIP-1α and MCP-1 as well as cell migration. The MIP-1α and MCP-1 produced in response to TGF-β1 could act as secondary chemoattractants (Figure 3E). Here, we attempt to explain the signal pathway from TGF-β1 ligand to MIP-1α, MCP-1 expression via RhoA and NF-κB using previously reported documents.

Notably, the activity of the transcription factor NF-κB can be modulated by members of the Rho family small GTPases,³⁶  and RhoA seems to activate NF-κB via a MEKK1-independent mechanism.³⁷  Inactive NF-κB binds to IκB, and NF-κB becomes active and is translocated into the nucleus to play as a transcription factor when IκB is phosphorylated and degraded. We have observed that TGF-β1 induces phosphorylation of IκB and concomitantly reduces its concentration, probably via degradation (data not shown), suggesting that TGF-β1 activates NF-κB. The activated NF-κB may stimulate serum response factor (SRF)–dependent transcription and in turn induce the transcription of c-fos, a component of AP-1 with the aid of ternary complex factors (TCFs).^38,39  Indeed, it has been demonstrated that RhoA potentiates AP-1 transcription in T cells.⁴⁰  Since the promoter of MIP-1α contains 5 AP-1 binding sites⁴¹  and the promoter of MCP-1 contains NF-κB, AP-1, and SP-1 binding sites,⁴²  AP-1 seems to be important for the transcription of MIP-1α, and AP-1 and NF-κB are likely to be responsible for transcription of MCP-1. Therefore, TGF-β1 probably induces transcription of MIP-1α and MCP-1 by the following routes: TGF-β1 → RhoA → NF-κB → SRF → c-fos (AP-1) → MIP-1α and MCP-1, and TGF-β1 → RhoA → NF-κB → MCP-1. Hence, it is likely that the activation of RhoA in the first 60 minutes of exposure to TGF-β1 induces the production of chemokines such as MIP-1α and MCP-1 as well, directly stimulating cell migration. Recently, it was documented that TGF-β1 stimulates MCP-1 expression,⁴³  which is regulated by Rho and Rho-kinase.⁴⁴  Indeed, it also was reported that MIP-1α and MCP-1 are induced by several stimuli via NF-κB activation,⁴⁵  and TGF-β1 transiently activates NF-κB through TAK1 and IκB kinase (IKK).⁴⁶ 

Although ERK stimulates TCFs such as Elk-1, which together with SRF induces transcription of c-fos,³⁹  ERK activated in response to TGF-β1 (Figure S1B) was not involved in the induction of MIP-1α (Figure 3C). Here it should be noted that activated ERK during TGF-β1–induced cell migration is translocated into focal contact,⁴⁷  although a detail regulatory mechanism of ERK in cell migration is not clear. Furthermore, Rho and ROCK are essential upstream components of ERK1/2, in that pretreatment of Raw 264 cells with Tat-C3 and Y-27632 for 30 minutes inhibited TGF-β1–stimulated ERK1/2 (Figure S1B).

The assumption that TGF-β1 abrogates cell migration by inducing expression of new proteins such as integrin receptors and extracellular matrix proteins so causing excessively tight binding of the cells^1,2  can be excluded, in that CHX and ActD blocked the induction of new genes but could not reverse the reduced cell migration in response to long-term exposure to TGF-β1 (Figure 3A). TGF-β1 is known to be an immunosuppressive cytokine with many effects on immune responses.⁴⁸  Furthermore, administration of TGF-β1 induces Smad 6 and Smad 7 (I-Smads) that in turn negatively regulate TGF-β signaling by interacting with TβRI^49,50  and recruiting the E3 ligases Smurf1 and Smurf2, which direct ubiquitin-dependent degradation of TGF-β receptor–Smad7 complexes.^51,52 

The basis of the inhibitory action of TGF-β1 on cell migration (Figure 1), however, has not been elucidated. Here, we showed that inhibition of RhoA is essential for its inhibitory action on cell migration (Figure 2) and that one of the direct causes of the inhibition of cell migration may be inactivation of the RhoA previously activated early upon short-term exposure to TGF-β1. Thus, the mechanism by which RhoA is inactivated during long-term exposure to TGF-β1 needs to be elucidated.

P190RhoAGAP inactivates RhoA. P190RhoGAP may be involved in a mechanism that may be responsible for RhoA inactivation (Figure 6A). DN p190RhoGAP, which lacks a GTP-binding domain, did not inactivate RhoA in response to TGF-β1. If the phosphorylation of RhoA was primarily responsible for RhoA inactivation, the GTP-bound state of RhoA would be reduced even in the presence of DN p190RhoGAP. However, the opposite was the case. Thus, this second mechanism may be the main pathway involved. In particular, DN RhoGAP significantly increased migration of the cells pretreated with TGF-β1 for 24 hours compared with that of control cells (Figure 6B). This also provides strong evidence that the activity of RhoA regulates cell migration induced by TGF-β1. Although p190RhoGAP can be regulated via phosphorylation by Src and protein kinase C,⁵³  it is elusive how p190RhoGAP is activated in response to long-term exposure to TGF-β1.

Phosphorylation of RhoA by PKA inactivates RhoA. The inactivation of RhoA by TGF-β1 was reversed by H89, suggesting that PKA inactivates RhoA (Figure 7C). Indeed, TGF-β1 phosphorylated RhoA (Figure 7A), and as a result the complex of RhoA-RhoGDI might increase during long-term exposure to TGF-β1. H89 also reversed the inhibition of cell migration resulting from long-term exposure to TGF-β1 (Figure 3A). Indeed, the view that PKA is involved in the regulation of cell migration is reinforced by the evidence that PKA inhibits expression of MIP-1α (Figure 3C).^27,54  It is not clear whether the activation of PKA in response to TGF-β1 is dependent on cAMP.

It has been independently reported that PKA phosphorylates RhoA and stabilizes the RhoA-RhoGDI complexes⁵⁵  implicated in protection against endothelial barrier dysfunction⁵⁶  and in activation of neurite outgrowth of PC12 cells.²⁹  Recently, the molecular mechanism of PKA activation in response to TGF-β1 has been clarified: TGF-β1 treatment leads to the formation of a complex between Smad3/4 proteins and the regulatory subunit of PKA, resulting in release of the PKA catalytic subunit.³¹  Contrary to our expectation that Smad3/4 would activate PKA and lead to inactivation of RhoA, overexpressed Smad3/4 activated RhoA even in response to long-term administration of TGF-β1 (data not shown). Overexpressed Smad3/4 may compete with and overcome the effect of the I-Smads, resulting in normal transmission of the TGF-β1 signal. Thereafter, TGF-β1 may induce the expression of Net1, a GEF for RhoA, via the overexpressed Smad3, resulting in activation of RhoA.⁵⁷  PKA also is reported to inhibit AKAP-Lbc, one of the Rho-GEFs, resulting in RhoA inactivation.⁵⁸ 

In conclusion, we propose that macrophage migration in response to TGF-β1 is essentially regulated by RhoA activity and that RhoA is eventually inactivated via phosphorylation by PKA and the action of p190RhoGAP (Figure 8). It is to be answered how TGF-β1 allows these factors to inactivate RhoA.

We greatly thank Dr J. L. Bos of the University Medical Center Utrecht for providing His-tagged RalGDS-RBD plasmid DNA. We thank Dr S. O. Huh and Dr D. K. Song from the Department of Pharmacology of Hallym University for providing P815 and HL-60 cells, respectively. We also are grateful to Dr J. Settleman of Harvard Medical School for providing the p190RhoGAP cDNAs and to Dr R. Derynck of the University of California at San Francisco for Smad3 and Smad4 cDNAs. In addition, we thank Dr R. Yu of the University of Ulsan, Korea, for an aid to measurement of cytokines.