Report

Role of Rac1 and Oxygen Radicals in Collagenase-1 Expression Induced by Cell Shape Change

See allHide authors and affiliations

Science  08 May 1998:
Vol. 280, Issue 5365, pp. 898-902
DOI: 10.1126/science.280.5365.898

Abstract

Integrin-mediated reorganization of cell shape leads to an altered cellular phenotype. Disruption of the actin cytoskeleton, initiated by binding of soluble antibody to α5β1 integrin, led to increased expression of the collagenase-1 gene in rabbit synovial fibroblasts. Activation of the guanosine triphosphate–binding protein Rac1, which was downstream of the integrin, was necessary for this process, and expression of activated Rac1 was sufficient to increase expression of collagenase-1. Rac1 activation generated reactive oxygen species that were essential for nuclear factor kappa B–dependent transcriptional regulation of interleukin-1α, which, in an autocrine manner, induced collagenase-1 gene expression. Remodeling of the extracellular matrix and consequent alterations of integrin-mediated adhesion and cytoarchitecture are central to development, wound healing, inflammation, and malignant disease. The resulting activation of Rac1 may lead to altered gene regulation and alterations in cellular morphogenesis, migration, and invasion.

Modifications of cell shape are crucial for tissue morphogenesis, cell migration, and invasion. These alterations in cell morphology are thought to rely on the organization of the actin cytoskeleton and the modulation of cell adhesion. Changes in cell morphology lead to specific signaling from cell adhesion receptors and a consequent change of gene expression (1), including genes encoding the matrix metalloproteinase (MMP) family, which includes collagenase-1 (CL-1) and stromelysin-1 (2, 3). The organization and dynamics of the various structures that constitute the actin cytoskeleton are controlled by members of the Rho family of small guanosine triphosphate (GTP)–binding proteins (4). Lysophosphatidic acid–induced formation of stress fiber is governed by Rho, growth factor–stimulated extension of lamellipodia and ruffling are regulated by Rac, and filopodia formation is controlled by Cdc42. We have studied the signal transduction cascade initiated by alteration of integrin-controlled adhesion that leads to increased expression of the gene encoding CL-1 in rabbit synovial fibroblasts (RSFs).

Expression of CL-1 is increased when RSFs adhere and spread on specific integrin ligands or when the cells round and disrupt the actin cytoskeleton (5). These processes probably involve two distinct pathways, because they differ in integrin specificity and kinetics. Ligation of α5β1 integrin with an adhesion-perturbing monoclonal antibody (mAb) to α5β1 in solution (5) reduced the organized actin microfilament cytoskeleton and induced CL-1 expression in RSFs (Fig. 1A). We tested whether Rho family guanosine triphosphatases (GTPases) transduce the signals from integrin ligation that lead to CL-1 expression by transfecting RSFs with constructs encoding the dominant interfering mutants Rac1N17, RhoAN19, and Cdc42N17 (6-8). Increased expression of CL-1 caused by addition of soluble mAb to α5, which inhibits substrate adhesion of cells and causes cell rounding, was abrogated by transient expression of Rac1N17 but not by expression of RhoAN19 or Cdc42N17 (Fig. 1B). In contrast, enhanced expression of CL-1 caused by spreading of the cells via α5β1 on the same mAb bound to the culture dishes as a substrate was independent of Rac1 (Fig. 1B). These data indicate that α5 integrin–mediated spreading and rounding have distinct signaling pathways. Treatment of cells with cytochalasin D (CD) rounds cells directly by disrupting the actin-dependent cytoarchitecture and increases expression of CL-1 (2). This effect of CD was not inhibited by expression of Rac1N17 (9), indicating that Rac1 either acts upstream of shape changes in the signaling pathway or does not participate in the effect of CD. Rac1N17 alone did not alter expression of CL-1. Expression of a dominant negative mitogen-activated protein (MAP) kinase kinase (MEK) construct (6), which interferes with activation of the MAP kinase pathway, had no effect on basal expression of CL-1 or on CL-1 expression induced by antibody to α5 (anti-α5) in solution (9).

Figure 1

Effect of actin reorganization and Rho GTPase on CL-1 expression. (A) Actin reorganization and expression of CL-1. RSFs were cultured on fibronectin (FN) substrates (3, 10) in serum-free medium for 18 hours in the absence (a and c) or presence (b and d) of function-perturbing BIIG2 anti-α5 (4 μg/ml). The cultures were stained simultaneously for fibrillar actin with fluorescein-phalloidin (a and b) and CL-1 with biotinylated mAb to CL-1 (c and d) followed by Texas Red–labeled streptavidin (3). (B) Requirement of Rac1 for cell shape–dependent expression of CL-1. RSFs transiently expressing β-Gal (control) or dominant interfering mutants RacN17, RhoN19, or CdcN17 (6, 8) were cultured on FN with anti-α5 in solution (4 μg/ml) or on immobilized anti-α5 (3, 8). CL-1 was detected by immunoblotting of medium (30 μl) after incubation for 18 hours with 1.5 × 104 transfected cells (5). (C) Reorganization of actin in RSFs expressing activated Rho GTPases. RSF were stained with mAb to Myc and with Texas Red–labeled antibody to mouse IgG to detect the Myc epitope tag of the transfected GTPases (6, 7) (a, c, e, and g) or with fluorescein-phalloidin for actin (b, d, f, and h). (D) Analysis by slot immunoblot (3) of CL-1 in medium conditioned for 24 hours by RSFs transiently transfected with members of the Ras superfamily. Data are expressed as fold induction relative to control (FN) and are the mean ± SEM of three separate experiments. (E) Transcriptional regulation of CL-1 by Rac1. RSFs were transfected with activated GTPases and the hCL-1/Luc construct. The transfected RSFs were plated onto wells coated with FN for 18 hours, and the luciferase reporter activity was determined (12). Data are expressed as fold induction compared with induction of control β-Gal–transfected cells (FN) and are the mean ± SEM of the results of six experiments. (D andE) Values showing statistical significance in comparison with controls (P < 0.05, analysis of variance) are indicated by an asterisk. Scale bar in (A) and (C), 20 μm.

If Rac1 contributes to increased expression of CL-1 in response to binding of soluble anti-α5, then constitutively activated forms of Rac1 alone should increase expression of CL-1. Transient expression of Rac1V12 reduced stress fibers in >90% of cells, stimulated formation of actin-rich lamellipodia and pinocytosis, and caused the formation of some multinucleated cells (Fig. 1C). In contrast, RhoAV14 increased formation of stress fiber in >90% of cells. Cells expressing Cdc42V12 had many filopodia and complete dissolution of stress fibers. At equivalent amounts of expression, Rac1V12 was effective in increasing CL-1 expression, whereas RhoAV14 and Cdc42V12 had little effect (Fig.1D). Oncogenic RasV12 (6) or Raf-1 (10) did not increase expression of CL-1 (Fig. 1E).

We used a promoter-luciferase construct containing the −517- to +63-bp segment of the human collagenase promoter (hCL-1/Luc), which includes an AP1 site, a PEA3 site, a transforming growth factor–β (TGF-β) inhibitory element (TIE), and two AP2 sites (11,12) to determine whether these GTPases altered transcription of the gene encoding CL-1. Transfection of Rac1V12 with the hCL-1/Luc construct increased expression of the luciferase reporter, similar to what was observed for endogenous CL-1 production, whereas expression of RhoAV14 or Cdc42V12 did not increase expression of luciferase significantly (f in Fig 1C). In cells transfected with Rac1N17, CD increased expression of luciferase 21-fold (9).

In phagocytic cells, Rac constitutes part of the multimolecular β-nicotinamide adenine dinucleotide phosphate (NADH)–oxidase complex that generates the reactive oxygen species (ROS) superoxide free radical (O2) and its dismutation product H2O2(13). The enzymes that generate ROS in fibroblasts have not been identified. However, RSFs treated with anti-α5 in solution or with CD, or transfected with Rac1V12, produced H2O2 (Fig. 2A). CD, but not anti-α5, effectively induced H2O2production in cells transfected with Rac1N17 (Fig. 2A). Scavengers of oxygen free radicals N-acetyll-cysteine (NAC) or Tiron (1,2-dihydroxybenzene-3,5-disulfonate) abrogated anti-α5–and Rac1V12-mediated expression of the hCL-1/Luc reporter, whereas inhibition of reactive nitrogen species by 7-nitroindazole (Fig. 2, B and C) andN G-nitro-l-arginine methyl ester (9) had no effect. NAC also inhibited secretion of CL-1 induced by soluble anti-α5 and CD (9).

Figure 2

Actin reorganization–dependent induction of hCL-1/Luc expression mediated by production of ROS. (A) ROS production by RSFs. Control cells transiently expressing Rac1N17 or Rac1V12 were plated on FN in the presence of soluble anti-α5 or CD as indicated and assessed for H2O2 production (21). (B and C) CL-1 promoter transcription, measured as luciferase expression, was induced by soluble anti-α5 in (B) control cells or (C) cells transiently expressing Rac1V12 treated as indicated with 50 mM NAC or 7-nitroindazole (7-NI). Data are expressed as fold induction in comparison with control cells plated on FN alone and are the mean ± SEM of four experiments. (D) Cell shape–dependent translocation of NF-κB into nuclei of RSFs (22) and inhibition by NAC. Rac1V12-transfected (a) or control RSFs (b through f) were plated on cover slips coated with FN (a through e) or immobilized anti-α5 (f) for 3 hours and then treated with 50 μM H2O2 for 3 hours (c) or incubated with 50 mM NAC (for 1 hour) (d) before treatment with 50 μM H2O2 for 3 hours, or treated with anti-α5 (4 μg/ml) in solution for 3 hours (e) or left untreated for 3 hours (a and b). Arrows, cytoplasmic NF-κB (p65); arrowheads, nuclear NF-κB (p65). Scale bar, 20 μm.

The nuclear factor kappa B (NF-κB)/Rel family of transcription factors, which are activated by stimulants such as H2O2 (14), mediated Rac1-induced expression of CL-1. Cytoplasmic NF-κB exists in an inactive complex consisting of DNA-binding components (p50/RelA) bound to an inhibitory component of the IκBα family (14). Treatment of RSFs with anti-α5 in solution or with 50 μM H2O2, but not with immobilized anti-α5, induced rapid nuclear translocation of NF-κB that was blocked by incubation of the cells with 50 mM NAC (Fig. 2C), which is consistent with a role for NF-κB downstream of ROS in regulating expression of CL-1. Cells expressing Rac1V12 that constitutively expressed CL-1 also showed a nuclear localization of NF-κB (Fig. 2C). Expression of IκBαM, a transdominant-negative mutant of IκBα (15), inhibited activation and translocation of NF-κB into the nucleus (16) (Fig. 3, A and B) and repressed expression of both endogenous CL-1 (9) and hCL-1/Luc (Fig. 3, C and D) in cells treated with soluble anti-α5 or transfected with Rac1V12. Nuclear extracts prepared from Rac1V12 transfectants showed constitutive activation of NF-κB, which was increased by soluble anti-α5 (Fig. 3B).

Figure 3

Requirement for activation of NF-κB during cell shape–dependent induction of CL-1 expression by anti-α5. (A) Protein immunoblot of IκBα from control RSFs or RSFs transfected with Rac1V12, with and without cotransfection of IκBαM (23). IκBαM (upper band) was less abundant than the endogenous IκB protein. (B) Enhanced nuclear translocation of NF-κB after integrin ligation and expression of Rac1V12. RSFs were transfected with vector control (lanes 1 through 6) or Rac1V12 (lanes 7 through 13) or IκBαM as indicated. Cells were plated onto FN-coated plates for 18 hours, then incubated with anti-α5 (4 μg/ml) in solution for 0 (Unstim), 15, 30, and 60 min; nuclear extracts from these cells were prepared and subjected to electromobility shift analysis (19). Unlabeled (“cold”) NF-κB oligonucleotides 100-fold in excess decreased specific DNA-protein binding (lanes 6 and 12). Antibody to p50 induced a supershift of the probe (lane 13). Also shown is expression of the hCL-1/Luc reporter in (C) control and (D) Rac1V12-transfected cells. Cells were plated on anti-α5 (20 μg/ml) (Sub.), or on FN (30 μg/ml) with anti-α5 (4 μg/ml) added in solution (Sol.), and transfected with IκBαM as indicated. Data are expressed as fold induction relative to control on FN and are the mean ± SEM of the results of three experiments.

Although NF-κB can transactivate genes directly, the hCL-1/Luc construct has no NF-κB/Rel consensus sequence, indicating that NF-κB-mediated induction of CL-1 expression is indirect. Because interleukin-1 (IL-1), which is an NF-κB–regulated protein, can act in an autocrine manner to induce CL-1 in primary fibroblasts (17), we investigated the role of IL-1α in integrin-mediated expression of CL-1. Treatment of RSFs with anti-α5 in solution stimulated synthesis of IL-1α mRNA (9), secretion of the protein (Fig.4, A and B), and expression of the hCL-1/Luc reporter (Fig. 3, C and D). Expression of Rac1V12 caused a 28-fold increase in IL-1 secretion, which was increased further by treatment with soluble anti-α5. The ROS inhibitor NAC blocked the IL-1 expression induced by soluble anti-α5 and Rac1V12 (Fig. 4, A and B). These data indicate that increased expression of the IL-1 gene is a consequence of integrin- and Rac1-mediated alteration of cell shape and of ROS generated in this pathway. IL-1 expression was required for shape-dependent expression of CL-1. IL-1 receptor antagonist (IL-1RA), a competitive inhibitor for IL-1 interaction with its receptor, suppressed CL-1 expression induced by anti-α5 in solution, Rac1V12 (Fig. 4C), or CD (9, 17). These data indicate that an IL-1–controlled autocrine loop mediates the induction of CL-1 caused by changes in cell shape.

Figure 4

Cell shape–dependent induction of IL-1 expression and IL-1–dependent CL-1 expression. Expression of IL-1 (24) in (A) control and (B) Rac1V12-transfected cells. Cells were exposed to anti-α5 or 50 mM NAC as indicated. Data are expressed as means of two separate experiments done in duplicate. (C) Control and Rac1V12-transfected cells were plated onto FN-coated wells and were left untreated or were treated with anti-α5 in solution (4 μg/ml) in the presence or absence of IL-1RA (10 ng/ml). Medium from 5 × 104cells cultured for 18 hours was analyzed for CL-1 by immunoblotting (3). Data are expressed as n-fold induction in comparison with the amount of CL-1 induced from control cells plated on FN and are shown as the mean ± SEM of three separate experiments.

Our results indicate that the Rac GTPase is an essential component in the signaling cascade initiated by integrin-mediated cell rounding that leads to the expression of CL-1. MMPs are crucial in cell migration and tumor invasion (18, 19). Rac1 has been implicated in invasion of lymphoma and epithelial cells and is essential for growth factor–induced migration of fibroblasts (4,19). Although Ras is important in some types of integrin signaling (20), and acts upstream of Rac1 in growth factor–mediated directed migration in mammary epithelial cells and Rat1 fibroblasts (19), we showed a distinct signaling pathway through Rac1 that is functional and separate from a Ras-mediated signaling cascade. Interference with α5 integrin–mediated cell adhesion activated Rac1, which then caused generation of ROS, induced activation of NF-κB, and led to induction of IL-1, an autocrine inducer of CL-1 expression. This pathway is distinct from that activated by α5 integrin–dependent spreading, which also leads to expression of CL-1 and requires rapid activation of AP1 (3), but does not require Rho-family GTPases, ROS, NF-κB activation, or IL-1 expression. Thus, exposure of cells to a soluble integrin ligand produces cell rounding, whereas exposure to an insoluble integrin ligand induces spreading. In fibroblasts, these distinct changes in cell shape trigger diverse signaling pathways, which happen to impinge on the gene encoding CL-1. Regulation of MMPs through integrin-mediated disruption of the actin cytoskeleton and alteration of cell form may influence developmental, migratory, and invasive properties of cells. Modifications in the extracellular matrix composition change both integrin-mediated signaling (20) and proteolysis (18) and, through these mechanisms, are strong determinants of cell behavior in normal and pathological settings.

  • * Present address: Baxter Healthcare Corp., Round Lake, IL 60073, USA.

  • To whom correspondence should be addressed. E-mail: zena{at}itsa.ucsf.edu

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article