View metadata, citation and similar papers at core.ac.uk brought to you by CORE providedORIGINAL by Elsevier ARTICLE- Publisher Connector

Cross-Talk Between RhoGTPases and Stress Activated Kinases for Matrix Metalloproteinase-9 Induction in Response to Keratinocytes Injury

Laurent Turchi, Anne-Amandine Chassot, Isabelle Bourget,Ã Christine Baldescchi, Jean Paul Ortonne, Guerrino Meneguzzi, Emmanuel Lemichez,w and Gilles Ponzio INSERM U 385, ÃINSERM U 364, and wINSERM U 452 Nice, France

Cell migration and extracellular matrix remodeling are soluble factor, secreted by wounded normal human two essential processes of wound healing, regulated by keratinocytes. We also demonstrate that the Rho family extracellular metalloproteinases such as matrix metal- of small GTPases, p38[MAPK] and JNK together play a loproteinase-2 (Gelatinase A) and matrix metallopro- key part in the signaling pathways controlling the sti- teinase-9 (Gelatinase B). Expression of matrix metallo- mulation of matrix metalloproteinase-9 in wounded proteinase-9 is deregulated in numerous wound healing cells. We provide lines of evidence indicating that in pathologies. To date the mechanisms regulating matrix wounded keratinocytes, upregulation of matrix metal- metalloproteinase-9 during normal wound healing are loproteinase-9 depends on two distinct pathways. The poorly documented. Using both primary cultures of ¢rst involves Rac1 and/or Cdc42 that control normal human keratinocytes and a wounding device the activation of p38[MAPK]. The second depends on especially designed to dissect the molecular events dur- RhoA activation that is required for stimulation of ing the healing process in vitro, we show that matrix JNK. Key words: JNK/keratinocytes wound healing/matrix metalloproteinase-9 is stimulated by injury in normal metalloproteinase-9, p38[MAPK]/Rho GTPases. J Invest Der- human keratinocytes. This upregulation results from matol 121:1291 ^1300, 2003 the mechanical stress created by injury and not from a

atrix metalloproteinases (MMP) are a family of proteolytic cleavage mediated by other proteinases (Werb, 1997; extracellular endoproteinases that degrade extra- John and Tuszynski, 2001). MMPs display a conserved Zn2 þ cellular matrix (ECM) proteins, such as lami- binding catalytic site and can be inhibited by speci¢c physiologic nin, ¢bronectin, and collagen. These enzymes inhibitors (Sternlicht and Werb, 2001). are involved in several physiologic processes Cutaneous wound healing is a complex process characterized Msuch as development and tissue remodeling but they also contri- by the co-ordinated re-epithelization of the epidermis and bute to pathologic processes including tumorigenesis and arthritis restoration of injured connective tissue. During wound healing (Werb, 1997; Sternlicht and Werb, 2001). the keratinocytes, the endothelial cells, and the ¢broblasts prolif- The di¡erent MMPs have been classi¢ed into ¢ve groups in- erate and migrate into the wound bed (Singer and Clark, 1999). cluding gelatinases (MMP-2 and MMP-9), collagenases (MMP-1, Cell migration and tissue remodeling require the controlled MMP-8, MMP-17, and MMP-13), membrane-type MMP (MT1^ degradation of the ECM and the activation of growth factors MMP), stromelysins (MMP-3, MMP-10, and MMP-7), and un- (Vassalli and Saurat, 1996). These processes are generally mediated classi¢ed MMP (MMP-11,12, 19, 26, etc.) (Werb, 1997).These pro- by serine proteases and MMP family members (Sternlicht and teases, which are either secreted or membrane bound, are Werb, 2001). Speci¢cally it has been shown that MMP-2 (72^68 produced as inactive forms (pro-enzyme) that are activated in kDa Gelatinase A) and MMP-9 (92^82 kDa Gelatinase B) play an the pericellular environment via an autocatalytic cleavage or by essential part during healing of wounded skin. MMP-2, which localizes in the connective tissue ¢broblasts and endothelial cells, is thought to have an important function during granulation and Manuscript received March 20, 2003; revised May 19, 2003; accepted for the early remodeling phases of repair. MMP-9 is stimulated in res- publication July 14, 2003 ponse to injury and is mostly expressed in the migrating keratino- Address correspondence and reprint requests to: Gilles Ponzio INSERM cytes sheet (Salo et al,1994;Munautet al, 1999; Soo et al, 2000). U 385, FaculteŁ de MeŁ decine Avenue de Vallombrose 06107 Nice cedex 02, MMP-9 catalyzes cleavage of type IV collagen and other base- France. Email: [email protected] ment membrane components (Sternlicht and Werb, 2001). Abbreviations: AEBSF, 4-(2-aminoethyl) benzenesulfonyl £uoride Recently, the ¢brinogen (Lelongt et al, 2001), interleukin (IL)-1 MMP, matrix metalloproteinase; JNK, NH2 terminal c-Jun kinase; ECM, extracellular matrix; SAPK, stress activated protein kinases; MAPK, mito- (Schonbeck et al, 1998), transforming growth factor-b (Yu and gen activated protein kinase; ERK, extracellular signal-regulated kinase; Stamenkovic, 2000), and the hemidesmosomal type XVII col- NHK, normal human keratinocytes; GST, glutathione S transferase; PI3K, lagen (Liu et al, 2000) have been also identi¢ed as MMP-9 sub- phosphatidylinositol 3 phosphate kinase; ToxA, toxin A of Clostridium di⁄- strates. In wounded tissues, expression of MMP-9 parallels the cile; exoC3, exoenzyme C3 of Clostridium botulinum. reabsorption of the provisional matrix and assembly of new

0022-202X/03/$15.00 . Copyright r 2003 by The Society for Investigative Dermatology, Inc.

1291 1292 TURCHI ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY structures of the epidermal/dermal junction; then, the expression Culture cell Keratinocytes were seeded on mitomycin C (10 mg per mL) trea- decreases to basal level (Soo et al, 2000). In contrast, £uids from ted 3T3-J2 ¢broblasts feeder layers (2 105 cells per cm2)andgrownin chronic wound persistently contain high levels of MMP-9 that ‘‘Green’’ medium (2/3 Dulbecco minimal Eagle’s medium, 1/3 HamF12 sup- probably contribute to abnormal healing of the lesion (Wysocki plemented with 10% fetal calf serum, HEPES 20 mM, penicillin (1000 U per mL), streptomycin (1000 U per mL), insulin (5 mg per mL), hydrocorti- et al, 1993; Yager et al, 1996). Indeed, excessive proteinase activity sone (0.4 mm M), cholera toxin (10 mM), recombinant human epidermal may degrade growth factors and ECM components required for growth factor (10 mM), triiodothyronine (2 mM), and adenine (18 mM). the restoration of ECM bed that support migration of the cells co- lonizing the wound. Therefore, the control of molecular pathways Immuno£uorescence Con£uent NHK cells were wounded and involved in MMP-9 regulation is an essential requirement for the incubated at 371C/5% CO2. Twenty-four hours after injury, cells were healing process and its deregulation may result in abnormal healing. washed with phosphate-bu¡ered saline, pH 7.4 and ¢xed in 2% The molecular pathways involved in regulation of MMP-9 have paraformaldehyde. Cells were then incubated with monoclonal anti- been widely investigated in variety of physiologic or physiopatho- MMP-9 antibody (1/50) for 1 h at room temperature and with a secondary logic systems; however, the mechanisms of MMP-9 regulation £uorescein isothiocyanate conjugated goat anti-mouse antibody (1/50). For nuclei labeling, cells were incubated for 10 min at room temperature with a during wound healing are poorly documented. In other contexts, 10 ng per mL Hoechst reagent solution. MMP-9 has been shown to be regulated, at least partially, at the transcriptional level. Several consensus DNA sequences for tran- Protein extracts and immunoblotting analysis Cells wounded or scription factors have been identi¢ed within the promoter of not for the indicated times were solubilized at 41Cin500mL RIPA bu¡er MMP-9. Activation of activator protein-1 (AP-1) and nuclear fac- (Tris/HCl 10 mM pH 7.5, NaCl 150 mM, sodium deoxycholate 1%, tor-kB(NF-kB) appears to be essential for the induction of MMP- nonidet P-40 1%, sodium dodecyl sulfate (SDS) 0.1%, 1 mM 9 by IL-1 (Yokoo and Kitamura, 1996) or 12-O-tetradecanoyl- phenylmethylsulfonyl £uoride, 50 mg aprotinin per mL, 50 mg leupeptin phorbol-13-acetate (Simon et al, 2001). Several consensus binding per mL, 1 mg pepstatin A per mL, 20 mM NaF, 1 mM NaVO4,1mMb glycerophosphate). Proteins (100 mg per lane) were resolved on SDS^ sites, including those for Sp1, PEA3, AP-1, Ets, NF-kB, and a reti- polyacrylamide gel electrophoresis and transferred on to Immobilon-P noblastoma binding element, are involved in the transcriptional ac- membrane (Millipore) as described previously (Turchi et al, 2002). The tivation of MMP-9 in H-ras and c-myc transformed rat embryo membrane was saturated with phosphate-bu¡ered saline supplemented cell lines (Gum et al, 1997; Himelstein et al, 1997). In the trans- with 3% bovine serum albumin, 0.5% gelatin, and 0.1% Tween 20, and formed cell systems studied up to date, MMP-9 promoter is driven incubated with either anti-MMP-9 (1/200), anti-phospho/ACTIVE by signal transduction pathways involving stress- and mitogen-ac- p38[MAPK] or JNK (1/2000), phospho-speci¢c anti-CREB (1/1000) or -c- tivated protein kinases (p38[MAPK],JNK,andERK1/2),whichare Jun (1/1000) or anti-pan p38[MAPK] (1/10001) antibodies, for 2 h at room stimulated either by growth factors, mitogen, cytokines, or envir- temperature. The membranes were incubated for 1 h at room temperature onmental changes (Robinson and Cobb, 1997). with horseradish peroxidase-conjugated goat anti-rabbit IgG (1/4000) and blots were revealed using the Amersham (Roche/Amersham, Les ULIS, We have recently developed an in vitro wounding system that al- France) ECL detection system. lows the detection and the quanti¢cation of the molecular events occurring in keratinocytes and ¢broblasts in response to injury. Antibodies Monoclonal anti-MMP-9 antibody was obtained from Using this device we have reported that injury of normal human Research Diagnostic, Inc. (Flanders, NJ) Polyclonal p38[MAPK] (pTPpY) keratinocytes (NHK) stimulated the activation of early signaling antibody was purchased from Promega Corporation (Madison, WI) Anti- events such as tyrosine kinases, MAP kinases (ERK, p38[MAPK], phospho-CREB (Ser133) antibody (used to detect phospho-ATF1) was and JNK), AP-1 binding activity, and late signaling events such as from Biolabs Inc. (Beverly, MA). Anti-Rho, Rac, Cdc42 monoclonal laminin-5, MMP-9, ¢bronectin EDA, IL-8, b4 integrin expression antibodies were purchased from Transduction Laboratories (Lexington, (Turchi et al, 2002). The speci¢c molecular pathways linking these KY). Horseradish peroxidase-conjugated goat anti-rabbit IgG and £uorescein isothiocyanate-conjugated goat anti-rabbit antibodies were early and late events, however, remained to be elucidated. purchased from Dako. Anti-pan p38[MAPK] antibody was a generous gift In this study, we dissect one of these pathways, by demonstrat- of Dr B. DeŁ rijard (Nice, France). ing that the mechanical injury of NHK monolayers stimulates expression of MMP-9 at mRNA and protein levels, through a Rho family of small GTP binding proteins activation (‘‘pull- mechanism not described previously, and involving the Rho fa- down’’ experiments) The activation of cellular Rho, Rac, and Cdc42 mily of guanosine triphosphate (GTP) binding proteins and the was determined by precipitation with the glutathione S transferase (GST) stress kinases JNK and p38[MAPK]. This stimulation likely results protein coupled to the Rho binding domain of Rhotekin (Ren et al, 1999) from the mechanical stress created by the injury and does not in- or to Rac/Cdc42 binding domain of PAK (Manser et al, 1998), according to the procedure described by Herrmann et al (1995). Brie£y, wounded (or volve a soluble factor secreted by wounded NHK. intact) keratinocytes were lyzed in (1) Rho-speci¢c lysis bu¡er (Tris/HCl 50 mM pH 7.2, NaCl 500 mM, MgCl2 10 mM, Triton X-100 1%, sodium deoxycholate 0.5%, SDS 0.1%, AEBSF 100 mM, 1 mg leupeptin per mL, MATERIALS AND METHODS Pepstatine A 1 mM), or (2) Rac/Cdc42 speci¢c lysis bu¡er (HEPES 25 mM pH 7.3, NaCl 150 mM, MgCl2 5 mM, EGTA 0.5 mM, Triton X-100 0.5%, Reagents Trypsin, ethylenediamine tetraacetic acid, HEPES, penicillin, glycerol 4%, 20 mM b-glycerophosphate, 10 mM NaF, AEBSF 100 mM, and streptomycin were purchased from Invitrogen Corp, Carlsbad, CA. leupeptin 1 mg per mL, Pepstatine A 1 mM). Fifteen milligrams of clari¢ed Dulbecco minimal Eagle’s medium, HamF12, and fetal calf serum were protein extracts were incubated with glutathione-Sepharose precoupled with obtained from Hyclone (Northumberland, UK). Insulin, hydrocortisone, GST-Rho-BD or GST-Rac-BD protein (30 mg) for 45 to 60 min at 41C. cholera toxin, human recombinant epidermal growth factor, triiodo- Beads were washed four times with Rac/Cdc42 lysis bu¡er (for Rac and thyronine, adenine, mitomycin, actinomycin D, and gelatin were purchased Cdc42) or with Tris bu¡er (Tris/HCl 50 mM pH 7.2, NaCl 500 mM, from Sigma (Sigma Aldrich, St. Louis, MO). RNAble, RNA preparation MgCl2 10 mM,Triton X-100 1%, AEBSF 100 mM, 1 mg leupeptin per mL, kit, was from Eurobio (Les Ulis, France). LY294002 SB203580, PD 98059, Pepstatine A 1 mM) (for Rho) and denatured in Laemmli bu¡er. GTP- and AG-1478 were purchased from Calbiochem (Calbiochem/France, bound Rho, Rac, and Cdc42 were separated on 12% SDS^polyacrylamide Meudon, France). BB94 is a kind gift of British Biotech (Oxford, UK). gel electrophoresis and measured by western blot using monoclonal Clostridium di⁄cile toxin A and Clostridium botulinum C3 exoenzyme were antibodies rise against Rho (1/250), Rac (1/10001), and Cdc42 (1/2501). generous gifts of Drs I. Just (Hanover, Germany) and P. Boquet (Nice, France), respectively. The cDNA encoding the dominant negative allele of Northern blot analysis Total RNA from NHK were isolated using the JNK1 (JNK1(APF)) was described previously (Derijard et al, 1994; Lin et al, RNABle kit according to the manufacturer’s instructions. Northern blot 1995) and was a generous gift of Dr B. Derijard (Nice, France). analysis were carried out using 20 mg of total RNA following the procedure described previously (Rezzonico et al, 1995). The MMP-9 (458 bp) cDNA probe was obtained by polymerase chain reaction (PCR) Isolation and primary culture of human keratinocytes ampli¢cation from NHK total RNA using speci¢c primers, respectively: Isolation NHK were isolated from healthy neonatal foreskin as described forward: 50 -TGGCCCAGGTGACCGGGGCCCTC-30; reverse: 50 - (Rheinwald and Green, 1975). AAAGGTTAGAGAATCCAAGTTTAT-30. VOL. 121, NO. 6 DECEMBER 2003 MMP-9 REGULATION IN WOUNDED KERATINOCYTES 1293

Gelatin zymography Aliquots (5 mL) of culture medium correspond- double band identifying the inactive form (pro-enzyme 92 kDa) ing to the di¡erent experimental conditions were applied on 10% and the active (82 kDa) enzyme form (Fig 1C). Here we show polyacrylamide gel containing 0.3% gelatin. The gels were incubated in that, in wounded keratinocyte cultures, MMP-9 is mostly 2.5% Triton X-100 for 2 h, then incubated for 18 h at 371Cin100mM expressed in its active form. This observation and the fact that Tris/HCl, pH 7.5, 200 mM NaCl, 1 mM CaCl2, and 1 mM MgCl2. The gels were ¢nally stained with Coomassie Brilliant Blue. MMP-9 was exclusively visualized by immuno£uorescence in keratinocytes located at the edge of the wound (Fig 1B) support the idea that MMP-9 is involved in the degradation of ECM Retroviral infection and isolation of NHK expressing the JNK (APF) dominant negative allele of JNK1 deposed by the cultured keratinocytes and the 3T3-J2 ¢broblasts (Rheinwald and Green, 1975), and consequently that MMP-9 is Plasmid construction The full-length open reading frame of the cDNA active in the healing process. Our data also show a good correla- encoding the dominant negative JNK (APF) allele of JNK (a gift of tion between the variations in the steady-state levels of MMP-9 Dr Derijard, Nice France), was subcloned downstream the mouse Moloney murine leukemia virus (MMLV) long-term repeat (LTR) of Plasmid mRNA, expression of the protein, and enzymatic activity. This pLZRS-Ires-Zeo (Gift of G.P. Nolan, Standford University, California). suggests that upregulation of MMP-9 mainly resulted from en- Plasmids (pLZRS-Ires-Zeo-JNK(APF) were ampli¢ed into Escherichia coli hanced expression of MMP-9 mRNA transcripts consequent to XL10 strain and puri¢ed using QIAquick kit (Qiagen Inc., Chatsworth, CA). mechanical injury of the cell culture. The gelatinolytic activity of the conditioned medium of Retroviral infection Infectious recombinant retrovirus were generated by keratinocyte cultures also revealed that MMP-2 (Gelatinase A) is transfecting the recombinant plasmid pLZRS-Ires-Zeo-JNK (APF) into secreted by unwounded cell cultures. In contrast to MMP-9, the amphotropic Phoenix packaging cells by the calcium phosphate preci- however, expression of MMP-2 was not upregulated following pitation method (Michiels et al, 2000). The supernatant from these cultures injury, con¢rming the observation reported by others (Salo et al, 4 (about 2.5 10 colony forming units (cfu) per mL) were collected after 1994). 48 h and applied for 24 h at 321C to NHK seeded at 3 10 4 cells per cm2 on a feeder of J2-3T3 ¢broblasts. Cells were fed with fresh medium for 2 d MMP-9 induction was not mediated by a soluble secreted and were selected in the presence of 200 mg zeocin per mL. factor MMP-9 is stimulated by chemicals (phorbol myristate acetate), growth factors, and cytokines. Some of which (trans- Reverse transcriptase^PCR Total NHK RNA (2 mg) was reverse- b a transcribed at 371C for 60 min, using Omniscript Kit (Promega, forming growth factor- , IL-1, tumor necrosis factor- )are Madison, WI). Two microliters of each cDNA sample were used for PCR upregulated during wound healing (Salo et al, 1994; Holzheimer using TAQ polymerase (Life Technologies, Inc.). The cDNA encoding and Steinmetz, 2000; Han et al,2001;Turchiet al, 2002). To assess MMP-9 and GAPDH were ampli¢ed using the speci¢c sets of whether induction of MMP-9 could be mediated by a second primers, respectively: forward: 50 -TGGCCCAGGTGACCGGGGCCCTC-30; messenger released by wounded keratinocytes, we tested the reverse: 50 -AAAGGTTAGAGAATCCAAGTTTAT-30,andforward:50 -ATGG- ability of spent medium of injured NHK cultures to upregulate GGAAGGTGAAGGTCGG-3; reverse: 50 -TTACTCCTTGGAGGCCATGT-30. the expression of MMP-9 in exponentially growing and quiescent keratinocytes. The MMP-9 gelatinolytic activity Quantitation of northern and western blot analysis The RNA and present in the conditioned medium of NHK and 3, 6, and 12 h proteins were quantitated using two methods: (1) by densitometric after wounding is presented in Fig 2A, respectively lanes 1, 2, 3, scanning using a computerized microscopic image processor Biocom 500 (Biocom, les Ulis France) comprising a PC/AT-compatible microcomputer, 4. Exposure for 24 h of proliferating NHK to such conditioned a real-time imaging processor, a control monitor, a color high de¢nition medium had no e¡ect on MMP-9 activity (Fig 2B). Indeed, the monitor, and a Panasonic WV-CD50 camera, and (2) Using the NIH activity of MMP-9 in the culture medium of exponentially image software available on web site: http://rsb.info.nih.gov/nih-image/ growing cells (Fig 2B) did not increase compared with the activity present in the conditioned medium of wounded cells (Fig 2A). The same results were obtained when the condi- RESULTS tioned medium was added to con£uent keratinocyte cultures (not shown). To con¢rm these data we performed reverse trans- De novo expression of activated-MMP-9 metalloprotease criptase^PCR experiments using RNA obtained from (1) at the edges of wounded keratinocytes To dissect the keratinocytes before and 12 h after injury, and of (2) 50% molecular pathways involved in the regulation of the metallo- con£uent intact NHK treated for 24 h with the conditioned protease MMP-9 (Gelatinase B) during epidermal wound medium of NHK harvested, 6 and 12 h after wounding. The healing in vitro, we have used NHK and a new technologic results indicate that MMP-9 mRNA was stimulated in NHK 12 approach specially designed for this purpose and already h after injury (Fig 2B, left panel) but not when intact keratin- described (Turchi et al,2002). ocytes were incubated for 24 h with the spent medium of A con£uent monolayer of primary human keratinocytes was wounded cells (Fig 2B, right panel). mechanically injured as described in Materials and Methods. After These results support the data presented in Fig 2(A)and 24 h, the healing cultures were analyzed by immuno£uorescence demonstrate that the conditioned medium of wounded after labeling with Hoechst reagent, which allows the keratinocytes does not contain secreted soluble factors able to identi¢cation of keratinocytes migrating from the edge (dotted stimulate MMP-9 in unwounded cells. This idea is in agreement line) into the wound bed (arrows)(Fig 1A). Interestingly, the with the observation that upregulation of MMP-9 is restricted to wound closure was totally impaired when cells were incubated, the wound edge of injured keratinocyte cell cultures as deter- before injury, with 20 mM of BB94, a wide spectrum metallo- mined by immuno£uorescence studies (Fig 1B). proteinases inhibitors (Davies et al, 1993), indicating that, in our biologic system, MMP are involved in wound repair. p38[MAPK] is involved in MMP-9 induction To characterize Using a speci¢c anti-MMP-9 antibody, we clearly observed further the pathways regulating induction of MMP-9 in that the protease MMP-9 was upregulated in wounded NHK wounded NHK, we tested the e¡ects of known pharmacologic and was speci¢cally expressed in cells forming the migrating inhibitors on the expression of MMP-9 at mRNA levels and front (Fig 1B, arrows). Upregulation of MMP-9 in the wounded gelatinolytic activity of the enzyme. The chosen inhibitors were cultures was further demonstrated at the molecular level by LY 294002, a phosphatidylinositol 3 phosphate kinase (PI3K) gelatinolytic zymography, western and northern blots (Fig 1C). inhibitor, PD 98059, a MEK-1 inhibitor, SB 203590, a p38[MAPK] Results demonstrated that MMP-9 increased between 6 h and inhibitor, and tyrphostin AG1472, an inhibitor of epidermal 24 h after injury of the cell layers (Fig 1C), then was growth factor receptor activity. The results shown in Fig 3(A) downregulated to the basal level after 48 h (not shown). In show that exposure to SB 203590 before wounding inhibited zymography and immunoblots, MMP-9 can be detected as a totally the enzymatic activity of MMP-9, indicating an 1294 TURCHI ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure1. (A) E¡ect of MMP inhibitor on the migration of wounded NHK. Con£uent kera- tinocytes were wounded in the absence (control) or in the presence of the MMP inhibitor BB94 (20 mM) and ¢xed after 24 h in 2% paraformal- dehyde. The cells were visualized by immuno- £uorescence staining by labeling of nuclei with Hoescht reagent. Dotted lines highlight the wound edges and arrows the ‘‘wound bed’’. The histograms on the right part represent a typical wound closure time course. The percentage of wound closure corresponds to the area ¢lled by the ‘‘healing’’ NHK at indicated times compared with the total wound bed area measured immedi- ately after injury. The measurements were made using an inverted microscope ¢tted with a cali- brated ocular grid (B) Immunochemical localiza- tion of MMP-9 in migrating cells. Wounded NHK were maintained in culture for 15 h, ¢xed, then stained using either a nonrelevant serum (NIS) or speci¢c anti-MMP-9 antibody (anti- MMP-9). Arrows indicate the MMP-9 positive cells. (C) Biochemical characterization of MMP- 9 in injured NHK. MMP-9 was measured 3, 6, 12, and 24 h after injury either in the culture medium by zymogram analysis (upper panel) or in cell pro- tein extracts by western blot analysis (middle panel) or in total RNA extracts by northern blot analy- sis (lower panel).‘‘N’’designates the level of MMP- 9 in unwounded NHK. 28S RNA used as load- ing control. The presented experiment is a typical one among ¢ve independent experiments.

involvement of p38[MAPK] in the signaling cascade leading to 203580 speci¢cally inhibited MMP-9 mRNA expression, MMP-9 upregulation in response to injury. In contrast, no whereas LY 294002, PD 98059, and AG 1478 did not (Fig 4A). appreciable e¡ect was observed when NHK were wounded in SB 203580 inhibition was concentration dependent with a half- presence of the other inhibitors. Using anti-phospho/active inhibitory e¡ect at a concentration lower than 4 mM(Fig 4B). speci¢c anti-p38[MAPK] antibodies, we con¢rmed that p38[MAPK] This observation discloses the role of p38[MAPK] as a key was stimulated in NHK, in response to injury (Fig 3B). The intermediate in the cascade of events controlling synthesis of stimulation was rapid and transient with a maximal activity MMP-9 mRNA. Beside this pharmacologic approach, we also detected between 8 and 15 min. We veri¢ed that SB 203580 tried to develop a dominant negative (dn) mutant of p38 e⁄ciently inhibited p38[MAPK] in wounded cells by showing the [MAPK] using a retroviral strategy (see experimental details in inhibition of phosphorylation of ATF1, one of p38[MAPK] the following paragraph). Surprisingly, however, it appeared that physiologic substrates (Fig 3C). We also veri¢ed that, at a the expression of dn form of p38[MAPK] (T180A/Y182F) was concentration that blocked MMP-9 activity, SB 203580 had no lethal and, after three di¡erent attempts, it was impossible to get e¡ect on phosphorylation of c-Jun, which indicates that SB p38dn expressing NHK clones. 203580 does not inhibit JNK a kinase related to p38[MAPK] (Fig 3C). Interestingly, at the same time as MMP-9 inhibition, we MMP-9 expression upon wounding required the NH2 showed that the treatment of wounded cells with SB 203580 but terminal c-Jun kinase (JNK) pathways To examine the not with inhibitors such as LY 294002, PD 98059, or tyrphostin possible involvement of JNK in the signaling cascade leading to AG1472 leads to a marked reduction in the number of cells that MMP-9 induction, we tested the e¡ect of a dominant negative migrate into the wound bed (‘‘wound closure’’) (Fig 3A, inset), allele of JNK1 (JNK1(APF)) on MMP-9 expression and suggesting that the genes controlled by the p38[MAPK] pathway gelatinolytic activity after injury of a con£uent monolayer of (including MMP-9) play a predominant part in NHK wound NHK. JNK1 (APF) corresponded to a JNK1 variant in which closure. alanine and phenylalanine residues respectively substitute To determine the mechanism by which SB 203580 decreases threonine 183 (T183) and tyrosine 185 (Y185). In wild-type cells, the gelatinolytic activity of MMP-9, we analyzed the level of T183 and Y185 are phosphorylated in response to JNKK/SEK1/ MMP-9 mRNA in cell cultures treated with SB 203580. MKK4 activation and are required for activation of JNK Northern blot analysis of wounded cell cultures showed that SB (Derijard et al,1994;Linet al, 1995). The cDNA for JNK1(APF) VOL. 121, NO. 6 DECEMBER 2003 MMP-9 REGULATION IN WOUNDED KERATINOCYTES 1295

In injured cells the induction MMP-9 involved the Rho family of small GTPases Previous studies demonstrated that the activation of the Rho family members, namely the small GTP binding proteins Rho, Rac, and Cdc42, is an essential step for actin reorganization during the healing process (Nobes and Hall, 1999). Because these small GTPases are also involved in the activation of JNK and p38[MAPK] (Bagrodia et al, 1995; Zhang et al, 1995), we investigated their roles in the molecular cascade responsible for MMP-9 stimulation after wounding. First, we inquired whether cell injury induces activation of the small GTP binding proteins RhoA, Rac, and Cdc 42. The extent of Rho GTPases activation was measured by classical pull-down experiments 3 min after injury, using recombinant GST fusion proteins containing (1) the Rho binding domain of rhotekin, to measure RhoA activation (Ren et al, 1999), and (2) the Cdc42 and Rac binding domain of PAK to measure Cdc42 and Rac activities (Manser et al, 1998).We demonstrated, for the ¢rst time, that con£uent NHK cultures wounding resulted in a strong activation of the three GTPases RhoA, Rac, and Cdc42 (Fig 6). To de¢ne the role of these proteins in MMP-9 activation, we analyzed the e¡ect of Rho GTPases-inhibiting toxins on (1) Figure 2. MMP-9 induction was not mediated by a soluble secreted MMP-9 activity and mRNA expression, and (2) on JNK and factor. (A) MMP-9 gelatinolytic activity measured in the conditioned p38[MAPK] activities. The toxins used were toxin A of Clostridium medium of con£uent NHK, 3, 6, and 15 h after wounding (lanes 2^4). Lane di⁄cile (ToxA), a broad Rho GTPases inhibitor inactivating 1 represents the MMP-9 activity in the medium of unwounded NHK. (B) RhoA, Rac, and Cdc42, and Exoenzyme C3 of Clostridium The conditioned medium of NHK unwounded (lane 1) or NHK cultures 3, botulinum (exoC3), a RhoA speci¢c inhibitor. When cells were 6, and 15 h after wounding (respectively lanes 2, 3,4) was added to subcon- preincubated with ToxA (500 nM) and exoC3 (40 mg per mL) £uent NHK. The MMP-9 activity present in the culture medium was mea- before wounding, MMP-9 activity and mRNA expression were sured 15 h later by zymography. Data are representative results from one inhibited by more than 90% (Fig 7A), indicating that the experiment of three. (C)(Left upper panel), analysis by PCR of MMP-9 activation of small GTPases is required for induction of MMP-9. RNA expression in NHK, before wounding (N) and 12 h after injury The e¡ect of ToxA and exoC3 appeared to be speci¢c as these (W).The right upper panel represents the level of MMP-9 detected by reverse toxins did not a¡ect the activity of MMP-2, another transcriptase^PCR using RNA puri¢ed from 50% con£uent NHK incu- metalloprotease highly expressed in NHK (Fig 7A, top panel). bated for 24 h with the conditioned medium of NHK before (1) and 6 h (2) Examination of the e¡ects of Rho GTPases-inhibiting toxins or 12 h (3) after injury. The lower panel represents the level of GAPDH on injury-stimulated JNK and p38[MAPK] activities shows that cDNA shown as a control for the reverse transcriptase^PCR reaction. exposure of NHK to ToxA resulted in inhibition of activation of both kinases (Fig 7B). This suggests that p38[MAPK] and JNK stimulation lays downstream Rho GTPases in the MMP-9 activation cascade triggered by injury. To investigate further the was inserted into the retroviral vector LZRS-IRES-ZEO speci¢city of JNK/p38 activation by Rho GTPases during (Michiels et al, 2000) and transduced into keratinocytes. Stable reparation, we investigated the e¡ect of exoC3 on wounded NHK transduced clones were selected in the presence of zeocin cells. In contrast to ToxA, treatment of NHK with exoC3 and used in wounding assays. Western blot analysis of lysates of speci¢cally inhibited JNK but was ine¡ective on the p38[MAPK] infected cells using a speci¢c anti-JNK antibody showed the activity (Fig 7C) indicating that, in response to injury, activation e⁄cient transduction of JNK in the recipient cells (Fig 5A,top of RhoA is mandatory for stimulation of JNK but not for the panel). The level of JNK1 expression was at least 50-fold higher activation of p38[MAPK]. In addition, the fact that p38[MAPK] was than the level of endogenous JNK measured in NHK infected inhibited by ToxA but not by exoC3 indicates that p38[MAPK] with an empty vector (mock). Analysis of JNK activity in the stimulation is consequent to activation of Rac and/or Cdc42. JNK(APF) expressing cells was assessed by western blot using ToxA and exoC3 are known to cause disorganization of the anti-phospho/active JNK antibodies. The ectopic expression of actin cytoskeleton (Aktories et al, 2000). To determine whether, JNK1(APF) in wounded cultures was found to inhibit by more in our system, MMP-9 resulted from the e¡ect of GTPases on than 80% the phosphorylation of JNK (Fig 5A, bottom panel). actin cytoskeleton, we submitted the wounded cells to a Immuno£uorescence analysis of the transduced cell cultures using treatment by cytochalasin B, an agent that depolymerizes the an anti-JNK antibody suggested that the background activity of actin cytoskeleton. Cells were incubated for 4 h with 1 and 5 JNK was likely due to the presence of a subpopulation of zeocin- mM of cytochalasin B, then wounded. MMP-9 RNA expression resistant cells that did not express JNK(APF) (data not shown). was measured 15 h after injury by northern blot analysis. The As shown in Fig 5(C), the expression of JNK(APF) did not results shown in Fig 7(D) demonstrated that the actin impair the stimulation of p38[MAPK] and ERK1/2 activities in cytoskeleton depolymerization itself had no e¡ect on the the wounded JNK(APF) NHK. induction of MMP-9 transcription. According to these results, The e¡ect of injury on the MMP-9 protein expression and stimulation of MMP-9 expression mediated by injury was gelatinolytic activity was assessed in both JNK(APF) and control independent from the changes in the cellular actin cytoskeleton. keratinocytes. Interestingly, in wounded cells, the level of MMP-9 and its enzymatic activity were strongly inhibited in JNK(APF) expressing keratinocytes as compared with controls DISCUSSION (Fig 5B), which indicates that activation of JNK is required for the stimulation of MMP-9. In view of these results, we Epidermal keratinocytes produce a large spectrum of the protei- concluded that the upregulation of MMP-9, in wounded nases involved in ECM remodeling during wound healing, inva- epidermal keratinocytes, requires the activation of both sion, squamous cell carcinoma, and other pathologic situations of p38[MAPK] and JNK. the skin (Johnsen et al, 1998; Kahari and Saarialho-Kere, 1999; 1296 TURCHI ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 3. E¡ect of kinase inhibitor on injury-stimulated MMP-9 activity during wound healing. (A) Con£uent NHK were incu- bated for 2 h in the absence (Ct) or in the pre- sence of kinase inhibitors: SB203580 (30 mM) (SB), PD98059 20 mM (PD), LY 294002 (10 mM) (LY), and tyrphostin AG1472 (100 nM) then wounded (W) or left intact (N). Fifteen hours after injury, MMP-9 activity was measured in cul- ture medium by zymography. (B) E¡ect of wound healing on p38[MAPK] activity measured by western blot analysis of cell extracts. To p p a n e l . Con£uent human keratinocytes were wounded (W) and harvested after 8, 15, 60, and 180 min. Cells were lyzed and p38[MAPK] activity in cell ex- tract was measured using anti-p38 antibodies to the phosphorylated/active form of p38[MAPK] (p38-p). (N) represents the p38[MAPK] activity in unwounded cells. (C) E¡ect of SB 203580 on p38[MAPK] and JNK activity. Cells were preincu- bated for 2 h with 20 mM of SB203580, then wounded (W) or left intact (N). Thirty minutes after injury, the cells were lyzed, and phosphory- lation of ATF1 and c-Jun was measured using antibodies to the phosphorylated form of ATF 1 (ATF-P) and c-Jun (c-Jun-P). This experiment was performed three times. The inset represents the e¡ect of kinase inhibitors (SB, PD, LY, and AG) on the wound closure determined by mea- suring the surface of the ‘‘wound grooves’’ sponta- neously covered by the keratinocytes 15 h after injury (ct).

Figure 4. E¡ect of pharmacologic inhibitors on MMP-9 mRNA expression. (A) To p p a n e l : Con£uent NHK were preincubated for 2 h in the presence of SB 203580 (30 mM) (SB), PD 95059 (20 mM) (PD), and LY 294002 (10 mM) (LY) then wounded (W) or left intact (N). After 15 h, the cells were lyzed, RNA extracted and the steady-state level of MMP-9 mRNA was mea- sured by northern blot analysis using 32P labeled MMP-9 cDNA. The 28S RNA is the sample loading control. Bottom panel. E¡ect of tyrphostin AG1478 on MMP-9 mRNA expression. Cells were preincubated with AG1472 (100 nM) then wounded and treated as described above. MMP- 9 was measured by northern blot analysis. (B) Dose^response of SB203580 on MMP-9 tran- scription in wounded NHK. Cells were preincu- bated for 2 h at the indicated concentrations of SB 203580. Wounded (W) or unwounded cells (N) were harvested 24 h later. Expression of MMP-9 mRNA was analyzed by northern blot. The histogram represented the relative intensity of each band quanti¢ed using the methods de- scribed in Materials and Methods. Data are repre- sentative of at least three distinct experiments.

Westermarck and Kahari, 1999). MMP-9/gelatinase B appears to et al,1993;Yageret al, 1996) or keloids (Neely et al, 1999). These be particularly important for keratinocyte migration during re- observations strongly suggest that MMP-9 stimulation is a crucial epithelialization of cutaneous wounds (Salo et al, 1994; Okada element in the regulation of the healing process. So far, however, et al, 1997; Madlener et al, 1998) and invasion of tumors (Wester- the mechanisms responsible for MMP-9 regulation have not been marck and Kahari, 1999; Kanayama, 2001). Clinical studies have thoroughly investigated in the context of wound healing. Indeed, demonstrated that MMP-9 is deregulated in leg ulcers (Wysocki the large majority of the studies concerning the mechanisms of VOL. 121, NO. 6 DECEMBER 2003 MMP-9 REGULATION IN WOUNDED KERATINOCYTES 1297

Figure 5. E¡ect of JNK inactivation on MMP-9 expression and enzymatic activity. (A) Analysis of JNK expression and activity in wild-type NHK (mock) and in NHK expressing a dominant nega- tive mutant gene of JNK1 (JNK1(APF)) (see Materi- als and Methods). Wounded or intact wild-type (Mock) and JNK1(APF) human keratinocytes were harvested 15 min after wounding. Cells lysates were prepared then expression of JNK (top panel) and its activity (bottom panel) were measured by Western blot analysis using an anti-pan JNK or an anti-phos- pho/active JNK antibody, respectively. (B) E¡ect of JNK inactivation on MMP-9 expression and gelati- nolytic activity. Con£uent wild-type (mock) and JNK1(APF)) NHK were wounded or not, then left to heal for 15 or 24 h at 371C. The culture media were then collected and cells were lyzed to prepare protein extracts harvested. (Left) MMP-9 protein expression was measured by western blotting (WB) using an anti-MMP-9 antibody. (Right) the gelati- nolytic activity was measured by zymography of cell culture medium 24 h after wounding. (C) E¡ect of JNK inactivation on p38[MAPK] and ERK1/2 activ- ities. Wild-type and JNK(APF) keratinocytes were wounded and cell lysates were separated on SDS^ polyacrylamide gel electrophoresis, then ERK1/2 and p38[MAPK] activities were estimated using speci- ¢c anti-phospho/active ERK1/2 and p38[MAPK] anti- bodies, respectively. Data are from a representative experiment out of three.

Figure 6. E¡ect of wound healing on the activation of small GTP binding proteins. Con£uent NHK were wounded (W) or left intact (N) and harvested 3 min later. Cells were lyzed, and the active GTP-bound form of RhoA, Rac, and Cdc42 were measured by pull-down (Materials and Methods) using recombinant polypeptides corresponding to the Rho binding domain of Rothekin or to Rac/Cdc42 binding domain of PAK. GTP-bound RhoA, Rac and Cdc42 expression was assessed by western blot analysis on total protein extracts (T) and pull-down pellets (P) (left panel). (N) corresponds to the level of Rho, Rac, and Cdc42 in unwounded cells. Right panel: quanti¢cation of western blot experiments presented on left panel. The activation levels of RhoA, Rac, and Cdc42 (GTP-bound) were in the same order of magnitude and represented, respectively, 12-, 5-, and 9-fold the basal value measured before injury.

MMP-9 regulation have been performed on tumor cells in which Here, we show that a mechanical injury of NHK stimulates ex- MMP-9 is constitutively upregulated (Simon et al, 1998, 1999; pression of MMP-9 and that this stimulation speci¢cally occurs Ikebe et al, 1999; Kupferman et al,2000). in the cells situated at the wound edge. As reported in vivo (Pirila To deepen our knowledge on molecular mechanisms regulat- et al, 2001), in our model of wound healing, MMP-9 likely parti- ing MMP-9 in human keratinocytes during wound healing, we cipates in the ECM digestion, an essential step of the wound have used a wounding device that allows the qualitative and healing process. We also present evidence that in wounded quantitative detection of the signaling pathways activated in the NHK, MMP-9 stimulation results from the upregulation of its repair process in vitro (Turchi et al, 2002). By this approach, the mRNA expression, likely by a transcriptional mechanism. enzymatic activity, the mRNA and the protein expression of The ¢rst major question raised by our observations was to MMP-9 can be assayed simultaneously by zymography, northern know whether the stimulation of MMP-9 was the result of the and western blot analysis, which was not possible using the action of soluble factor(s) secreted by wounded cells in response wounding models classically available to study wound healing in to injury or was the consequence of the physical modi¢cation oc- vitro (Dieckgraefe et al, 1997; Goke et al, 1998; Pintucci et al, 1999). curring at the membrane level when the integrity of the con£u- 1298 TURCHI ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 7. E¡ect of ToxA and exoC3 on (1) MMP-9 expression, and (2) JNK and p38[MAPK] activities. (A) Con£uent NHK preincubated for 6 h in the presence of ToxA (500 nM), or 18 h in the presence of exoC3 (40 mM), were wounded (W) and harvested after 15 h. The culture medium was collected and total RNA were puri¢ed from cells. MMP-9 was quanti¢ed by zymography in the culture medium (top panels) or by northern blot analysis of total cellular RNA (bottom panels). (Ct) represents MMP-9 enzymatic activity and mRNA expression in wounded in the absence of toxin. (N) correspond to the level of MMP-9 present in unwounded cells. (B) Con£uent NHK were incubated for 6 h with ToxA wounded (W) and harvested 15 min after wounding. Cells were lyzed and p38[MAPK] and JNK activities were assayed by western blotting using antibodies against the phosphorylated/active forms of p38 (p38- p) and JNK (JNK-p). The bottom panels represent the total expression of p38[MAPK] and JNK estimated by western blotting using anti-pan p38 and anti-pan JNK antibodies, respectively. (N) corresponds to the level of p38[MAPK] and JNK present in unwounded cells. (C) NHKwere incubated for 18 h with exoC3 before injury and analysis of p38[MAPK] and JNK activities as described in (B). The data are from one typical experiment chosen from three distinct experi- ments. The quanti¢cation of JNK and p38 inhibition by ToxA and exoC3, achieved using the methods described in Materials and Methods, indicates that ToxA inhibits 81% of p38 MAPK activity and 75% of JNK activity and that exoC3 inhibits 89% of JNK activity but has no inhibitory e¡ect on p38 MAPK. (D) E¡ect of cytochalasin B on MMP-9 expression. Con£uent NHKwere preincubated for 2 h with 1 mM and 5 mM cytochalasin B (Cyt B) before wounding. The cellular RNA were extracted 15 h after injury and MMP-9 expression was measured by northern blot analysis as described in Materials and Methods. (Ct) represents the MMP-9 mRNA expression in absence of cytochalasin B. ent NHK monolayer was broken. Indeed, several soluble factors, signals that in£uence the cell morphology and activate small including tumor necrosis factor-a, IL-1, and transforming GTPases, stress kinases, and a variety of genes, including trans- growth factor-b are known to induce MMP-9 in keratinocytes forming growth factor-b, collagen, or MMP-9 (Jin et al,2000; when added to cultured cells (Salo et al, 1994; Han et al,2001). Lambert et al, 2001; Swartz et al,2001).Thishypothesisalsoex- Some of them are upregulated in wounded tissues and cell cul- plains why the MMP-9 is stimulated exclusively in keratinocytes tures (Holzheimer and Steinmetz, 2000; Turchi et al,2002)and situated at the wound edge and not in the whole cell population. therefore constitute potential intermediates in injury-mediated The second aim of this study was to gain new insights on the MMP-9 stimulation. It has not been clearly demonstrated, how- molecular cascades underlying MMP-9 stimulation in response to ever, whether the amounts of endogenous factors, produced at cell injury. Using inhibitors of small GTPases family, the SB the wound site, are able to stimulate MMP-9 e⁄ciently. Using 203580, a p38[MAPK] inhibitor, and transfection of keratinocytes conditioned media of wounded NHK cultures, we show that with a dominant negative variant of JNK1, we demonstrate that, MMP-9 stimulation is apparently not mediated by soluble factors in healing keratinocytes, MMP-9 is under the control of a mole- released in response to injury. This conclusion suggests that the cular cascade involving the small GTP binding proteins RhoA, mechanical disruption of wounded keratinocyte monolayers Rac, and Cdc42 and the MAP kinases JNK and p38[MAPK]. may, by itself, directly induce the signaling cascade leading to The MAP kinases, ERK1/2, p38[MAPK], and JNK have been MMP-9 upregulation. This hypothesis is in agreement with nu- involved in the signaling pathways controlling a variety of merous studies demonstrating that mechanical forces generate MMP (Mengshol et al, 2001; Leivonen et al,2002;Reunanen VOL. 121, NO. 6 DECEMBER 2003 MMP-9 REGULATION IN WOUNDED KERATINOCYTES 1299 et al, 2002). To date, the role of MAPK in the regulation of dominant negative variant of JNK, we demonstrate that JNK and MMP-9 has been essentially demonstrated in cancer cells, includ- p38[MAPK] but not ERK1/2 nor PI3 kinase are located down- ing transformed keratinocytes (Francisco Santibanez et al,2002), stream the activation of the Rho GTPases and that these two ki- human squamous carcinoma cell lines (Gum et al, 1997; Simon nases likely constitute the molecular link between Rho proteins et al, 1998), neck carcinoma cells (Simon et al, 1999), and breast and activation of MMP-9. By the mean of ToxA and/or exoC3 cancer cells (Yao et al, 2001), but not in physiologic processes such we also demonstrate that the injury-mediated activation of RhoA as wound healing. In this study we demonstrate that p38[MAPK] preferentially controls the pathways leading to JNK stimulation, and JNK but not ERK1/2 are involved in MMP-9 stimulation whereas p38[MAPK] activation rather depends on the activation of during wound healing of primary human epidermal keratino- Rac and/or Cdc42. Interestingly, the essential role of JNK and cytes. Interestingly, we show that stimulation of MMP-9 requires p38[MAPK] in MMP-9 induction correlates well with the presence the concomitant activation of both JNK and p38[MAPK], because of AP-1 sites in MMP-9 promoter regulatory sequences (Yokoo the inhibition of one of these kinases totally prevents MMP-9 and Kitamura, 1996; Simon et al,2001). upregulation. In conclusion, in this study we identi¢ed the signaling path- In a recent study, Di-Poi et al (2002) showed that the addition ways underlying stimulation of the protease MMP-9, in of tumor necrosis factor-a to cultured mouse keratinocytes sti- wounded human keratinocytes. We demonstrated that MMP-9 mulated MMP-9 via a pathway involving the peroxisome prolif- upregulation results from activation of at least two signaling erator activated receptor-d/b and the PI3 kinase cascade. The pathways. One involves Rac and/or Cdc42 and controls the acti- authors argued that, in this biologic system, the addition of tu- vation of the kinase p38[MAPK], the second requires activation of mor necrosis factor-a mimics the wound healing process (Tan RhoA and JNK. Interestingly, this molecular cascade is speci¢c et al, 2001). In human keratinocytes, using a scratch wounding for the wound healing process and di¡ers from the mechanisms system, we clearly show that the injury-mediated stimulation of involved in MMP-9 upregulation in tumor cells. Because MMP- MMP-9 involves a di¡erent pathway as it is totally insensitive to 9 is deregulated in several pathologies associated with abnormal LY294002, a PI3 kinase inhibitor. wound healing (Ladwig et al, 2002; Lobmann et al,2002),our Similarities exist between tumoral progression and wound results pave the way to the understanding of MMP-9 expression healing (Dvorak, 1986). The fact that MMP-9 plays an active part in abnormal wound healing process in view of the development both in tumoral invasion and wound healing supports this idea. of strategies for the treatment of chronic wounds or keloids. The mechanisms responsible for the overexpression of MMP-9 in tumor cells remain unknown, and identi¢cation of the molecular mechanisms regulating MMP-9 expression during wound heal- Weare very grateful to Drs R. Busca, A. Cures, and B. Mari for skillful reading of the ing may contribute to the understanding of the events associated manuscript. This study was supported by grants of the ‘‘Institut pour la SanteŁet la with the neoplasic transformation in keratinocytes. Recherche MeŁdicale’’ (INSERM) and the ‘‘Association pour la Recherche contre le The identi¢cation of Rho family of GTPases in the signaling Cancer’’ARC grant 5965 and 4500. cascade that controls MMP-9 in healing keratinocytes is the third valuable new information resulting from this study. So far inves- tigations have focused on the contribution of Rho, Rac, and Cdc42 in the control of cell motility and adhesion (Nobes and Hall, 1999; Nguyen et al, 2001). Studies performed on rat embryo ¢broblasts showed that, during wound healing, the signal trans- REFERENCES duction pathways controlled by the four small GTP binding Aktories K, Schmidt G, Just I: Rho GTPases as targets of bacterial protein toxins. proteins Rho, Rac, Cdc42, and Ras, co-operate to promote cell Biol Chem 381:421^426, 2000 movement (Nobes and Hall, 1999). The activation of Rac and Bagrodia S, Derijard B, Davis RJ, Cerione RA: Cdc42 and PAK-mediated signaling Cdc42 induces lamellipodial/¢lopodial protrusion and membrane leads to Jun kinase and p38 mitogen-activated protein kinase activation. J Biol ru¥ing at the leading edge of the wound and thus regulate the Chem 270:27995^27998, 1995 migration stage. Moreover, Rho activity is required to maintain Brock J, Midwinter K, Lewis J, Martin P: Healing of incisional wounds in the em- bryonic chick wing bud: Characterization of the actin purse-string and de- cell^substrate adhesion. Elsewhere, it has been shown that RhoA monstration of a requirement for Rho activation. JCellBiol135:1097^1107,1996 was required during embryonic wound healing for the formation Coso OA, Chiariello M, Yu JC, et al: The small GTP-binding proteins Rac1 and of actin cables (Brock et al, 1996). Using, a wounding model we Cdc42 regulate the activity of the JNK/SAPK signaling pathway. Cell 81:1137^ clearly demonstrate, that in adult NHK, the activities of three 1146, 1995 GTPases are together stimulated by injury and are directly in- Davies B, Brown PD, East N, Crimmin MJ, Balkwill FR: A synthetic matrix me- volved in the molecular cascade linking cell injury to upregula- talloproteinase inhibitor decreases tumor burden and prolongs survival of mice bearing human ovarian carcinoma xenografts. Cancer Res 53:2087^2091, tion of MMP-9. Interestingly our study shows that, in adult 1993 keratinocytes the stimulation of MMP-9 involves Rho activation, Derijard B, Hibi M, Wu IH, et al: JNK1: A protein kinase stimulated by UV light but is independent of the actin cytoskeleton rearrangement as de- and Ha-Ras that binds and phosphorylates the c-Jun activation domain. Cell termined by the insensitivity of the process to cytochalasin B. 76:1025^1037, 1994 This observation is consistent with reports of Sahai and Treisman Dieckgraefe BK, Weems DM, Santoro SA, Alpers DH: ERK and p38 MAP kinase pathways are mediators of intestinal epithelial wound-induced signal transduc- (1998) indicating that Rho A can mediate gene transcription in- tion. Biochem Biophys Res Commun 233:389^394, 1997 dependently from cytoskeletal changes. Accordingly, our results Di-Poi N,Tan NS, Michalik L,Wahli W, Desvergne B: Antiapoptotic role of PPAR- demonstrate that, in wounded cell layers, the Rho family of beta in keratinocytes via transcriptional control of the Akt1 signaling pathway. GTPases exerts a double action by (1) controlling the actin cytos- Mol Cell 10:721^733, 2002 keleton changes, and (2) upregulating MMP-9 expression.Activa- Dvorak HF: Tumors. Wounds that do not heal. Similarities between tumor stroma tion of MMP-9 results in degradation of ECM proteins and thus generation and wound healing. NEnglJMed315:1650^1659, 1986 Francisco Santibanez J, Guerrero J, Quintanilla M, Fabra A, Martinez J:Transforming facilitates cell migration from the wound edge to the wound bed. growth factor-beta1 modulates matrix metalloproteinase-9 production The fourth original point of this study concerns the identi¢ca- through the Ras/MAPK signaling pathway in transformed keratinocytes. Bio- tion of the molecular intermediaries linking injury-stimulated chem Biophys Res Commun 267^273, 2002 Rho GTPases activation to stimulation of MMP-9. Several lines Goke M, Kanai M, Lynch-Devaney K, Podolsky DK: Rapid mitogen-activated pro- of evidence indicate that Rho-related GTPases regulate gene ex- tein kinase activation by transforming growth factor alpha in wounded rat pression through the activation of signaling cascades leading to intestinal epithelial cells. Gastroenterology 114:697^705, 1998 [MAPK] Gum R, Wang H, Lengyel E, Juarez J, Boyd D: Regulation of 92 kDa type IV col- the increase of JNK, p38 , ERK1/2, and PI3 kinase activ- lagenase expression by the jun aminoterminal kinase- and the extracellular ities (Coso et al, 1995; Zhang et al, 1995; Palsson et al, 2000; Murga signal-regulated kinase-dependent signaling cascades. Oncogene 14:1481^1493, et al, 2002). Using pharmacologic inhibitors, ToxA, exoC3, and a 1997 1300 TURCHI ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Han YP,Tuan TL, Hughes M,Wu H, Garner WL: Transforming growth factor-beta- Palsson EM, Popo¡ M,Thelestam M, O’Neill LA: Divergent roles for Ras and Rap and tumor necrosis factor-alpha-mediated induction and proteolytic activation in the activation of p38 mitogen-activated protein kinase by interleukin-1. of MMP-9 in human skin. J Biol Chem 276:22341^22350, 2001 JBiolChem275:7818^7825, 2000 Herrmann C, Martin GA, Wittinghofer A: Quantitative analysis of the complex be- Pintucci G, Steinberg BM, Seghezzi G, et al: Mechanical endothelial damage results tween p21ras and the Ras-binding domain of the human Raf-1 protein kinase. in basic ¢broblast growth factor-mediated activation of extracellular signal- JBiolChem270:2901^2905, 1995 regulated kinases. Surgery 126:422^427, 1999 Himelstein BP, Lee EJ, Sato H, Seiki M, Muschel RJ: Transcriptional activation of Pirila E, Maisi P, Salo T, Koivunen E, Sorsa T: In vivo localization of gelatinases the matrix metalloproteinase-9 gene in an H-ras and v-myc transformed rat (MMP-2 and -9) by in situ zymography with a selective gelatinase inhibitor. embryo cell line. Oncogene 14:1995^1998, 1997 Biochem Biophys Res Commun 287:766^774, 2001 Holzheimer RG, Steinmetz W: Local and systemic concentrations of pro- and Ren XD, Kiosses WB, Schwartz MA: Regulation of the small GTP-binding anti-in£ammatory cytokines in human wounds. Eur J Med Res 5:347^355, protein Rho by cell adhesion and the cytoskeleton. Embo J 18:578^585, 2000 1999 Ikebe T, Shinohara M,Takeuchi H, Beppu M, Kurahara S, Nakamura S, Shirasuna K: Reunanen N, Li SP, Ahonen M, Foschi M, Han J, Kahari VM: Activation of p38 Gelatinolytic activity of matrix metalloproteinase in tumor tissues correlates alpha MAPK enhances collagenase-1 (matrix metalloproteinase(MMP)-1) and with the invasiveness of oral cancer. Clin Exp Metastasis 17:315^323, 1999 stromelysin-1 (MMP-3) expression by mRNA stabilization. J Biol Chem Jin G, Sah RL, Li YS, Lotz M, Shyy JY, Chien S: Biomechanical regulation of 277:32360^32368, 2002 matrix metalloproteinase-9 in cultured chondrocytes. J Orthop Res 18:899^ Rezzonico R, Ponzio G, Loubat A, Lallemand D, Proudfoot A, Rossi B: Two dis- 908, 2000 tinct signalling pathways are involved in the control of the biphasic junB tran- John A, Tuszynski G: The role of matrix metalloproteinases in tumor angiogenesis scription induced by interleukin-6 in the B cell hybridoma 7TD1. J Biol Chem and tumor metastasis. Pathol Oncol Res 7:14^23, 2001 270:1261^1268, 1995 Johnsen M, Lund LR, Romer J, Almholt K, Dano K: Cancer invasion and tissue Rheinwald JG, Green H: Serial cultivation of strains of human epidermal keratino- remodeling: Common themes in proteolytic matrix degradation. Curr Opin cytes: The formation of keratinizing colonies from single cells. Cell 6:331^343, Cell Biol 10:667^671, 1998 1975 Kahari VM, Saarialho-Kere U: Matrix metalloproteinases and their inhibitors in Robinson MJ, Cobb MH: Mitogen-activated protein kinase pathways. Curr Opin tumour growth and invasion. Ann Med 31:34^45, 1999 Cell Biol 9:180^186, 1997 Kanayama H: Matrix metalloproteinases and bladder cancer. J Med Invest 48:31^43, Sahai E, Alberts AS, Treisman R: RhoA e¡ector mutants reveal distinct e¡ector 2001 pathways for cytoskeletal reorganization, SRF activation and transformation. Kupferman ME, Fini ME, Muller WJ, Weber R, Cheng Y, Muschel RJ: Matrix EMBO J 17:1350^1361, 1998 metalloproteinase 9 promoter activity is induced coincident with invasion dur- Salo T, Makela M, Kylmaniemi M, Autio-Harmainen H, Larjava H: Expression of ing tumor progression. Am J Pathol 157:1777^1783, 2000 matrix metalloproteinase-2 and -9 during early human wound healing. Lab Ladwig GP, Robson MC, Liu R, Kuhn MA, Muir DF, Schultz GS: Ratios of acti- Invest 70:176^182, 1994 vated matrix metalloproteinase-9 to tissue inhibitor of matrix metalloprotei- Schonbeck U, Mach F, Libby P: Generation of biologically active IL-1 beta by matrix nase-1 in wound £uids are inversely correlated with healing of pressure ulcers. metalloproteinases: A novel caspase-1-independent pathway of IL-1 beta pro- Wound Repair Regen 10:26^37, 2002 cessing. JImmunol161:3340^3346, 1998 Lambert CA, Colige AC, Munaut C, Lapiere CM, Nusgens BV: Distinct pathways in Simon C, Goepfert H, Boyd D: Inhibition of the p38 mitogen-activated protein the over-expression of matrix metalloproteinases in human ¢broblasts by re- kinase by SB 203580 blocks PMA-induced Mr 92,000 type IV collagenase laxation of mechanical tension. Matrix Biol 20:397^408, 2001 secretion and in vitro invasion. Cancer Res 58:1135^1139, 1998 Leivonen SK, Chantry A, Hakkinen L, Han J, Kahari VM: Smad3 mediates trans- Simon C, Hicks MJ, Nemechek AJ, et al: PD 098059, an inhibitor of ERK1 activa- forming growth factor-b-induced collagenase-3 (MMP-13) expression in tion, attenuates the in vivo invasiveness of head and neck squamous cell carci- human gingival ¢broblasts. Evidence for cross-talk between Smad3 and p38 noma. Br J Cancer 80:1412^1419, 1999 signaling pathways. JBiolChem277:46338^46346, 2002 Simon C, Simon M,Vucelic G, Hicks MJ, Plinkert PK, KoitschevA, Zenner HP:The Lelongt B, Bengatta S, Delauche M, Lund LR, Werb Z, Ronco PM: Matrix metal- p38 SAPK pathway regulates the expression of the MMP-9 collagenase via loproteinase 9 protects mice from anti-glomerular basement membrane ne- AP-1-dependent promoter activation. Exp Cell Res 271:344^355, 2001 phritis through its ¢brinolytic activity. JExpMed193:793^802, 2001 Singer AJ, Clark RA: Cutaneous wound healing. NEnglJMed341:738^746, 1999 Lin A, Minden A, Martinetto H, et al: Identi¢cation of a dual speci¢city kinase that Soo C, Shaw WW, Zhang X, Longaker MT, Howard EW,Ting K: Di¡erential ex- activates the Jun kinases and p38-Mpk2. Science 268:286^290, 1995 pression of matrix metalloproteinases and their tissue-derived inhibitors in cu- Liu Z, Zhou X, Shapiro SD, et al: The serpin alpha1-proteinase inhibitor is a critical taneous wound repair. Plast Reconstr Surg 105:638^647, 2000 substrate for gelatinase B/MMP-9 in vivo. Cell 102:647^655, 2000 Sternlicht MD,Werb Z: How matrix metalloproteinases regulate cell behavior. Annu Lobmann R, Ambrosch A, Schultz G, Waldmann K, Schiweck S, Lehnert H: Rev Cell Dev Biol 17:463^516, 2001 Expression of matrix-metalloproteinases and their inhibitors in the wounds Swartz MA,Tschumperlin DJ, Kamm RD, Drazen JM: Mechanical stress is commu- of diabetic and non-diabetic patients. Diabetologia 45:1011^1016, 2002 nicated between di¡erent cell types to elicit matrix remodeling. Proc Natl Acad Madlener M, Parks WC,Werner S: Matrix metalloproteinases (MMP) and their phy- Sci USA 98:6180^6185, 2001 siological inhibitors (TIMPs) are di¡erentially expressed during excisional skin Tan NS, Michalik L, Noy N, et al: Critical roles of PPAR beta/delta in keratinocyte wound repair. Exp Cell Res 242:201^210, 1998 response to in£ammation. Genes Dev 15:3263^3277, 2001 Manser E, Loo TH, Koh CG, et al: PAK kinases are directly coupled to the PIX Turchi L, Chassot AA, Rezzonico R, et al: Dynamic characterization of the molecu- family of nucleotide exchange factors. Mol Cell 1:183^192, 1998 lar events during in vitro epidermal wound healing. J Invest Dermatol 119:56^63, Mengshol JA, Vincenti MP, Brinckerho¡ CE: IL-1 induces collagenase-3 (MMP-13) 2002 promoter activity in stably transfected chondrocytic cells: Requirement for Vassalli JD, Saurat JH: Cuts and scrapes? Plasmin Heals! Nat Med 2:284^285, 1996 Runx-2 and activation by p38 MAPK and JNK pathways. Nucleic Acids Res Werb Z: ECM and cell surface proteolysis: Regulating cellular ecology. Cell 91:439^ 29:4361^4372, 2001 442, 1997 Michiels F, van der Kammen RA, Janssen L, Nolan G, Collard JG: Expression of Westermarck J, Kahari VM: Regulation of matrix metalloproteinase expression in Rho GTPases using retroviral vectors. Methods Enzymol 325:295^302, 2000 tumor invasion. FA S E B J 13:781^792, 1999 Munaut C, Salonurmi T, Kontusaari S, Reponen P, Morita T, Foidart JM, Tryggva- Wysocki AB, Staiano-Coico L, Grinnell F:Wound £uid from chronic leg ulcers con- son K: Murine matrix metalloproteinase 9 gene. 50 -upstream region contains tains elevated levels of metalloproteinases MMP-2 and MMP-9. JInvestDerma- cis-acting elements for expression in osteoclasts and migrating keratinocytes in tol 101:64^68, 1993 transgenic mice. JBiolChem274:5588^5596, 1999 Yager DR, Zhang LY, Liang HX, Diegelmann RF, Cohen IK: Wound £uids Murga C, Zohar M,Teramoto H, Gutkind JS: Rac1 and RhoG promote cell survival from human pressure ulcers contain elevated matrix metalloproteinase levels by the activation of PI3K and Akt, independently of their ability to stimulate and activity compared to surgical wound £uids. J Invest Dermatol 107:743^748, JNK and NF-kappaB. Oncogene 21:207^216, 2002 1996 Neely AN, Clendening CE, Gardner J, Greenhalgh DG, Warden GD: Gelatinase Yao J, Xiong S, Klos K, Nguyen N, Grijalva R, Li P,Yu D: Multiple signaling path- activity in keloids and hypertrophic scars. Wound Repair Regen 7:166^171, ways involved in activation of matrix metalloproteinase-9 (MMP-9) by here- 1999 gulin-beta1 in human breast cancer cells. Oncogene 20:8066^8074, 2001 Nguyen BP, Ren XD, Schwartz MA, Carter WG: Ligation of integrin alpha 3beta 1 Yokoo T, Kitamura M: Dual regulation of IL-1 beta-mediated matrix metalloprotei- by laminin 5 at the wound edge activates Rho-dependent adhesion of leading nase-9 expression in mesangial cells by NF-kappa B and AP-1. Am J Physiol keratinocytes on collagen. JBiolChem276:43860^43870, 2001 270:F123^F130, 1996 Nobes CD, Hall A: Rho GTPases control polarity, protrusion, and adhesion during Yu Q, Stamenkovic I: Cell surface-localized matrix metalloproteinase-9 proteolyti- cell movement. JCellBiol144:1235^1244, 1999 cally activates TGF-beta and promotes tumor invasion and angiogenesis. Genes Okada A,Tomasetto C, Lutz Y, Bellocq JP, Rio MC, Basset P: Expression of matrix Dev 14:163^176, 2000 metalloproteinases during rat skin wound healing: Evidence that membrane Zhang S, Han J, Sells MA, Cherno¡ J, Knaus UG, Ulevitch RJ, Bokoch GM: Rho type-1 matrix metalloproteinase is a stromal activator of pro-gelatinase A. family GTPases regulate p38 mitogen-activated protein kinase through the J Cell Biol 137:67^77, 1997 downstream mediator Pak1. JBiolChem270:23934^23936, 1995