ORIGINAL ARTICLE

Increased Migration of Murine Keratinocytes Under HypoxiaIs Mediated by Induction of Urokinase Plasminogen Activator

Richard J. Daniel and Richard W. GrovesCenter for Dermatology, Department of Medicine, University College London, UK

One of the key consequences of cutaneous wounding isthe development of tissue hypoxia. Recent data havesuggested that this is a potent stimulus for increasedkeratinocyte migration and hence re-epithelialization,although the mechanisms responsible for this remainunclear. In this study we have investigated the relation-ship between hypoxia, plasminogen activation, and invitro wound healing. Exposure of keratinocyte culturesto hypoxia resulted in upregulation of urokinaseplasminogen activator mRNA and a subsequent in-crease in urokinase plasminogen activator-mediatedplasminogen activation, as determined by indirectchromogenic peptide assay and plasminogen-linkedzymography. Analysis of keratinocyte wound healingin vitro con¢rmed enhanced wound closure in hypoxiccultures compared with normoxic cultures after 16 h.

Pretreatment of normoxic and hypoxic cultures withmitomycin C and cytochalasin B indicated that inthis system wound closure was due to keratinocytemigration rather than proliferation. Addition of thebroad-spectrum serine proteinase inhibitor, p-amino-benzamidine, or the speci¢c urokinase plasminogenactivator inhibitors, amiloride and WX-293, signi¢-cantly reduced wound closure in hypoxic cultures andabrogated the hypoxic enhancement of wound closure.These data indicate a central role for urokinase plasmi-nogen activators in hypoxic keratinocyte migrationand suggest a potential mechanism for enhanced re-epithelialization of wounds under low oxygen tensions.Key words: hypoxia/urokinase/wounding. J Invest Dermatol119:1304 ^1309, 2002

Cutaneous wounding is characterized by a dramaticsequence of cellular changes that result in the transformation of basal keratinocytes to motile cells, aprocess that is essential for re-epithelialization. Thiswound-activated response involves increased levels

of in£ammatory cytokines and growth factors, altered keratinand integrin expression, and induction of proteolytic enzymes,including matrix metalloproteinases (MMP) and plasminogenactivators (reviewed in Martin, 1997). Re-epithelialization, result-ing from increased keratinocyte migration and proliferation, is akey feature of normal cutaneous tissue repair and studies fromtransgenic and knockout mice indicate that an intact plasminogenactivator/plasmin system is required for normal wound healing(Bugge et al, 1996; Romer et al, 1996).

In cutaneous tissue the predominant plasminogen activator isurokinase plasminogen activator (uPA) ( Jensen et al, 1988). uPAis a 45 kDa glycoprotein, initially secreted as an inactive pro-en-zyme, which binds to a speci¢c receptor (uPAR) and is activatedby proteolytic cleavage (Cubellis et al, 1986; Petersen et al, 1988).Receptor binding of uPA is not essential for proteolytic function,as uPA is equally e⁄cient at plasminogen activation whenmembrane-anchored (Lee et al, 1992) or in a nonreceptor-bound

soluble form (Wilhelm et al, 1994; Carmeliet et al, 1998). Onceactivated, uPA converts the ubiquitous zymogen plasminogen toplasmin, a broad range proteolytic enzyme that facilitates matrixdegradation, stimulates growth factor precursors, and activatesseveral prometalloproteinases (Mignatti et al, 1996; Carmelietet al, 1997, 1998).

Throughout the last decade, evidence for involvement of uPAin cellular migration and proliferation has accumulated. Overex-pression of uPA and increased uPA-mediated plasminogen acti-vating activity have been observed in migrating monocytes,endothelial cells, and keratinocytes in culture (Morioka et al,1987; Pepper et al, 1993). In a variety of cell types uPA has beenassociated with promotion of migration either by uPA-mediatedproteolytic activity (Wijnberg et al, 1997), by signal transductionpathways triggered by uPA/uPAR association with integrins(Carriero et al, 1999; Nguyen et al, 1999; Yebra et al, 1999), or by adirect chemotactic e¡ect (Fazioli et al, 1997; Poliakov et al, 1999). Inaddition, in vivo analyses of both mouse and human healingwounds have shown increases in uPA mRNA and plasminogen-activating activity (Grondahl-Hansen et al, 1988; Romer et al, 1991;Schaefer et al, 1994), as well as identifying uPA as a positive regu-lator of epidermal proliferation (Hibino et al, 1999; Jensen andLavker, 1999). These ¢ndings suggest that uPA, and its associationwith uPAR, may play an important part in the migration andextracellular matrix invasion required for normal wound healing.

A major physiologic consequence of vascular disruption andtissue damage in wounded skin is hypoxia (Niinikoski et al, 1972;Chang et al, 1983). In vivo, acute hypoxia has been suggested toregulate enhanced wound re-epithelialization (Winter, 1962;Alvarez et al, 1983; Woodley and Kim, 1992), although long-termhypoxia may contribute to impaired wound repair (Stadelmann

Reprint requests to: Professor Richard W. Groves FRCP, ImperialCollege of Science Technology and Medicine, Department of AcademicDermatology, Chelsea and Westminster Hospital, London, SW10 9NH,U.K. Email: r.groves@ic.ac.uk

Abbreviations: uPA, urokinase plasminogen activator, MMP, matrixmetalloproteinase.

Manuscript received February 1, 2001; revised April 15, 2002; acceptedfor publication May 7, 2002

0022-202X/02/$15.00 � Copyright r 2002 by The Society for Investigative Dermatology, Inc.

1304

et al, 1998). In vitro, human keratinocyte migration is increased un-der hypoxic conditions (O’Toole et al, 1997), and hypoxia isknown to regulate expression of MMP, such as MMP-9, andcomponents of the plasminogen activation/plasmin system, in-cluding uPAR and PAI-1, in a wide variety of cell types (O’Tooleet al, 1997; Pinsky et al, 1998; Graham et al, 1999). At present, how-ever, no data exist concerning the functional consequences of hy-poxia on uPA-mediated plasminogen activation in keratinocytes.

We have therefore investigated the role of hypoxia in theregulation of keratinocyte-derived uPA and, using a series ofspeci¢c inhibitors of uPA activity, have evaluated the role ofuPA in hypoxia-enhanced keratinocyte migration. Our ¢ndingsdemonstrate that hypoxia is a profound regulator of both uPAmRNA and functional protein in keratinocytes, and that hypoxicenhancement of keratinocyte migration is blocked in the presenceof uPA inhibitors. These results suggest that the enhanced kerati-nocyte migration following hypoxia is mediated by hypoxicinduction of uPA.

MATERIALS AND METHODS

Reagents The plasmin-speci¢c chromogenic substrate H-D-valyl-L-leucyl-L-lysine-p-nitroanilide dihydrochloride (S-2251), human plasmino-gen, and plasmin were from Chromogenix (Milan, Italy). Humanurokinase was purchased from Calbiochem (Nottingham, U.K.). Runningbu¡er, renaturing bu¡er, developing bu¡er, and colloidal blue staining kitused for zymographic analysis were all supplied by Novex (Carlsbad, NM).Protease inhibitors, amiloride and p-aminobenzamidine were purchasedfrom Sigma (Poole, U.K.). The selective uPA inhibitor, WX-293 (Sperlet al, 2000) was a kind gift of Dr Katja Wosikowski (Wilex BiotechnologyGmbH, Munich, Germany). The pDB1519 vector (GenBank accession no.X02389) containing the complete murine uPA cDNA was purchased fromthe American Type Culture Collection (ATCC no. 63256). 32P-radiolabelednucleotides were purchased from Amersham Pharmacia (Little Chalfont,U.K.).

Cell culture PAM 212 murine keratinocytes (a kind gift of Dr S. Yuspa,Bethesda, MD) were maintained in RPMI 1640 medium supplementedwith 10% newborn calf serum, glutamine, penicillin/streptomycin, andHEPES (all supplied by Gibco BRL, Paisley, U.K.). At approximately60% con£uence, medium was aspirated, cultures were washed twice andrefreshed with serum-free RPMI 1640 medium. Serum-free conditionswere necessary on account of the plasminogen activating activityconstitutively present in newborn calf serum. After 24 h incubation innormoxic conditions, cultures were washed, refreshed again with serum-free medium, and maintained for an additional 24 h under the sameconditions or made hypoxic by incubation in sealed Billups Rothenburgchambers £ushed with 1% O2, 5% CO2, and 94% N2. Culture £uid O2tensions assessed after 24 h, using an ABL4 analyzer (Radiometer,Copenhagen, Denmark), were 2.7370.12% (mean7SEM).

Keratinocyte viability assays Cell viability was measured by Trypanblue exclusion and lactate dehydrogenase (LDH) release. LDH release wasmeasured using an in vitro toxicology kit (Sigma), according to themanufacturer’s instructions, and calculated as a percentage of total LDHreleased following detergent-mediated disruption of all cells.

Indirect chromogenic peptide assay for uPA activity Plasminogenactivation in supernatants was measured using an indirect chromogenicpeptide assay (Reinartz et al, 1993). Cell supernatants were harvested andtotal protein levels measured using a modi¢ed Bradford protein assay,according to the manufacturer’s instructions (Promega, Southampton,UK). 60 ml samples, containing equal amounts of total protein, wereincubated with 20 ml of S-2251 (2 mg per ml) in 96-well plates for 10min at 371C. Twenty microliters of human plasminogen (0.4 mg per ml)was added to each well, and plates incubated at 371C for the duration of theexperiment. Parallel samples incubated with 0.1 mM amiloride (a speci¢cuPA inhibitor) were assayed to determine the role of uPA in any detectedplasminogen activation. Color change was analyzed every 20 min at405 nm. Rate of change of absorbance at 405 nm against time wascalculated from sample readings (corrected for amiloride) and uPA values(IU per ml) determined by comparison with a human urokinase(Calbiochem) standard curve.

Plasminogen-linked zymography Supernatant samples were equa-lized for total protein prior to separation at 41C on denaturing sodium

dodecyl sulfate^polyacrylamide gel electrophoresis gels, containing 0.2%b-casein (Sigma) and 15 mg human plasminogen per ml. Proteins wererenatured in renaturing bu¡er washes (2� 20 min), and reactionsdeveloped by incubation in developing bu¡er overnight. Gels wereincubated with colloidal blue stain and caseinolytic activity was detectedas clear bands against the blue-stained casein background. Murine urine, arich source of uPA, was used as positive control.

Probes cDNA probes for uPA were prepared by polymerase chainreaction ampli¢cation of a 437 bp fragment from pDB1519, containingmurine uPA cDNA. Primers used were AATGTGCACAGCCAT-CCAGGT for sense, and CCTCAGCTACCTGAGGGCCAT for anti-sense. Probes were labeled using a 32P-deoxycytidine triphosphateoligolabeling kit (Amersham Pharmacia) according to the manufacturer’sinstructions. 32P-uridine triphosphate-labeled anti-sense riboprobes foruPAR were prepared from a murine uPAR cDNA plasmid (a kind gift ofDr F. Lupu, Thrombosis Research Institute, London, U.K.) by reversetranscription using the MAXIscript SP6/T7 kit (Ambion, Austin, TX)according to the manufacturer’s instructions.

RNA analysis Total keratinocyte RNA was extracted using Trizolreagent (Gibco) according to the manufacturer’s instructions. RNA wasresuspended in RNAase-free water, and nucleic acid concentrationdetermined by spectrophotometric analysis at 260 nm. Ten microgramtotal RNA samples were size fractionated on 1% agarose gels containing6.3% formaldehyde. Following transfer to Hybond Nþ membranes, RNAwas ¢xed by ultraviolet cross-linking and probed with the uPA or uPARprobe. After washing under stringent conditions, membranes wereexposed to autoradiography ¢lm. Densitometry was performed followingscanning of autoradiographs, with equal loading determined by analysis of28 S RNA.

In vitro wounding In vitro wounding was performed using a modi¢edversion of the scratch assay described previously (Cha et al, 1996). PAM212 keratinocytes were grown to con£uence in six-well dishes, and after asingle wash in serum-free medium, monolayers were scratched with aplastic pipette tip creating a cell-free area (‘‘wound’’) approximately 2 mmin width. Cultures were washed twice in serum-free medium to removecell debris and a marked area of the wound was photographed underphase-contrast microscopy. Cultures were covered with serum-freemedium, and uPA inhibitors added to relevant wells before incubationunder normoxic or hypoxic conditions. Previous experiments haddemonstrated that these inhibitors, at appropriate concentrations, did nota¡ect keratinocyte viability and did not induce a proliferative response(data not shown). After a speci¢c incubation period the marked area ofthe wound was re-photographed and wound area recovered (mm2) wasdetermined by analysis performed on a Macintosh computer using thepublic domain NIH Image program (developed at the U.S. NationalInstitutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image/).

RESULTS

PAM 212 keratinocyte viability is una¡ected after 24 hculture under hypoxic conditions To determine whetherPAM 212 keratinocytes could be cultured under low oxygenconditions for prolonged periods without signi¢cant loss ofviability, cells were maintained in normoxic and hypoxicconditions for 24 h and assessed for LDH release and TrypanBlue exclusion. Cells cultured under hypoxic conditions showedno variation in cell morphology at the light microscope level (notshown). Trypan Blue exclusion and LDH release analysis showedno signi¢cant di¡erence between normoxic and hypoxic samples(Table I). In addition, no signi¢cant di¡erence was seen in thepH value of conditioned medium from normoxic and hypoxiccultures (Table I).

Upregulation of uPA and uPAR mRNA, and increasedlevels of functional uPA protein are observed in PAM 212keratinocytes cultured under hypoxic conditions Toinvestigate the e¡ect of hypoxia on keratinocyte expression ofuPA and uPAR mRNA, total RNA was prepared from cellscultured under normoxic or hypoxic conditions for 24 h andanalyzed by northern blot analysis. Both uPA and uPARmRNA were easily detected in resting keratinocytes as has

HYPOXIA, UPA, AND KERATINOCYTE MIGRATION 1305VOL. 119, NO. 6 DECEMBER 2002

previously been described (Schaefer et al, 1996), and both signalswere induced approximately 2-fold following 24 h hypoxia(Fig 1). To determine whether there was concomitant inductionof uPA protein, conditioned medium from keratinocytes culturedunder normoxic or hypoxic conditions for 24 h was assayed forplasminogen activating activity. Plasminogen activation wasdetectable in both normoxic and hypoxic samples and couldbe almost entirely inhibited (496% inhibition) by the additionof 0.1 mM amiloride, indicating that this was uPA mediated. Asshown in Fig 2(A), a 4 -fold increase in uPA activity wasobserved in hypoxic compared with normoxic samples.Moreover, plasminogen-linked zymography detected a band of

b-casein degradation corresponding to murine uPA (E45 kDa)that was markedly increased in hypoxic conditioned medium(Fig 2B) con¢rming that the increased plasminogen activatingactivity is uPA mediated.

Hypoxia enhances in vitro wound closure in PAM 212keratinocytes by increased cell migration Under normoxicconditions, wound closure of 0.16970.017 mm2 was observedat 16 h (Fig 3A,C). Under hypoxic conditions, signi¢cantlyincreased wound closure was observed (0.2770.005 mm2)amounting to a 59.872.8% (mean7SEM, po0.01) enhance-ment (Fig 3B,D). To determine the mechanism involved, cellswere treated with cytochalasin B, a potent inhibitor of cellmigration, or mitomycin C, a potent inhibitor of cellularproliferation. Keratinocyte monolayers were wounded asdescribed, and cultured under normoxic and hypoxic conditionsfor 24 h. Inhibition of migration by cytochalasin B almostcompletely prevented wound closure (mean o9% of control)in both normoxic and hypoxic cultures (Fig 4), indicatingthat hypoxic-enhanced wound closure is dependent on cellmigration. Moreover, inhibition of keratinocyte proliferation bymitomycin C had no signi¢cant e¡ect on wound area recoveredover a 24 h period (Fig 4), indicating that cellular proliferation isnot required for enhancement of wound closure under hypoxicconditions. These observations are consistent with previous datademonstrating that, in vitro, primary human keratinocyte woundclosure is dependent on migration and not proliferation(Morioka et al, 1987).

Hypoxia enhances in vitro wound closure by PAM 212keratinocytes as a result of induction of uPA Havingdemonstrated upregulation of functional uPA in hypoxic

Table I. Viability of PAM 212 Keratinocytes cultured undernormoxic and hypoxic conditions

h Normoxia Hypoxia p (Student t-test)

Trypan Blue exclusiona 24 7.071.2 8.170.6 0.41LDH assayb 24 7.271.9 9.671.8 0.24pHc 24 7.370.05 7.470.06 0.34a,bData expressed as percentage cell death (mean7SEM).cData are expressed as pH units7SEM.a,b,cData shown are from triplicate wells in 4 independent experiments.

Figure1. Hypoxia increases uPA and uPAR mRNA levels. RNAwas extracted from PAM 212 keratinocyte cultures maintained in nor-moxic (N) or hypoxic (H) conditions for 24 h. Northern blot analysisshowed an approximately 2-fold increase in both uPA and uPARmRNA in cells cultured under hypoxic conditions. Densitometry data(mean7SEM) are from three independent experiments expressed as apercentage of normoxic control (100%) after correcting for loading.

Figure 2. Hypoxia increases levels of uPA-mediated plasminogenactivation activity in PAM 212 keratinocyte conditioned medium.Conditioned medium from PAM 212 keratinocyte cultures maintained innormoxic (N) and hypoxic (H) conditions for 24 h was assayed for uPAactivity. (A) S-2251 indirect chromogenic peptide assay showed a 4-foldincrease of uPA levels in cells cultured under hypoxic conditions. Datashown (mean7SEM) are from triplicate wells in four independent experi-ments. (B) Plasminogen-linked zymography revealed a marked increase inintensity of a 45 kDa band, corresponding to murine uPA, in hypoxic sam-ples compared with normoxic samples. Data shown is representative ofthree independent experiments.

1306 DANIEL ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

keratinocytes, we were interested to determine whether thisplayed a part in the enhancement of in vitro wound closureobserved under hypoxia. Thus, PAM 212 keratinocyte mono-layers were wounded as described and maintained with orwithout addition of uPA inhibitors. In control cultures, the arearecovered over 16 h was increased by 67.0778.7% (mean7SEM)in hypoxic conditions compared with normoxic conditions.Following addition of amiloride, a selective inhibitor ofurokinase (Vassalli and Belin, 1987), the hypoxia inducedenhancement of wound closure was completely abolished,however, and no signi¢cant di¡erence in wound area recoveredcould be observed between normoxic and hypoxic cultures(Fig 5A). No e¡ect of amiloride was observed on woundclosure under normoxic conditions, suggesting that uPA doesnot play a major part in this situation. Presence of the broad-spectrum proteinase inhibitor, p-aminobenzamidine, however,signi¢cantly reduced wound area recovered in both normoxicand hypoxic cultures (Fig 5B), suggesting that proteases otherthan uPA may be relevant in normoxic wound closure. Finally,we employed WX-293, a recently described highly selectiveinhibitor of uPA activity (Sperl et al, 2000). At 0.2 mM, WX-293completely inhibited hypoxic enhancement of wound closure,whereas at 0.1 mM there was 70% inhibition, suggesting aconcentration-related e¡ect (Fig 5C). Furthermore, the addition

of 0.1 mM and 0.2 mM WX-293 resulted in a 55.372.5% and74.0773.05% inhibition of uPA, respectively, indicating a directcorrelation between uPA inhibition and abrogation of hypoxia-mediated wound closure. Interestingly, like amiloride and incontrast to p-aminobenzamidine, no e¡ect of WX-293 on thewound area recovered in normoxic cultures was observed(Fig 5C). Taken together, these data indicate that the enhancedwound closure seen under hypoxic conditions relies on releaseof functional uPA by keratinocytes, and also suggests thatproteases other than plasminogen activators may play a part inkeratinocyte migration under normoxic conditions.

DISCUSSION

Acute cutaneous wounds are characterized by the rapid develop-ment of hypoxia secondary to interruption of normal vasculatureand the development of a ¢brin clot. Cells local to the woundmust be able to adapt appropriately to this hypoxic environmentsuch that wound healing progresses as rapidly as possible. Thedata presented here suggest that hypoxia is a potent stimulus forkeratinocyte migration and indicate that the likely mediator ofthis response is uPA.

Hypoxic regulation of the plasminogen activator/plasmin sys-tem has been documented in several cell types (Fitzpatrick andGraham, 1998; Graham et al, 1998; Maity and Solomon, 2000),and this study has demonstrated that hypoxia is a potent regulatorof keratinocyte plasminogen activation in vitro. Both uPA anduPAR mRNA were induced after 24 h exposure to hypoxiaand there was a marked induction of uPA-mediated proteolyticactivity in cell culture supernatants following hypoxic stimula-tion. This ¢nding contrasts with the e¡ects of hypoxia in othertissues, such as in lung, trophoblast, and endothelial cells, hypoxiacauses a decrease in soluble plasminogen activation (Pinsky et al,1998). Interestingly however, there is increased cell-associated

Figure 3. Hypoxia enhances in vitro wound closure. PAM 212 kerati-nocyte monolayers were wounded as described, and incubated under nor-moxic (A,C) or hypoxic (B,D) conditions. Wounds were photographedunder phase-contrast microscopy at 0 h (A,B) and 16 h (C,D). Photographsshown are representative of four independent experiments. Under nor-moxic (N) conditions wound closure of 0.16970.017 mm2 was observed.Under hypoxic (H) conditions signi¢cantly increased wound closure of0.2770.005 mm2 was observed amounting to a 59.872.8% (mean7SEM)enhancement in cultures incubated under hypoxic conditions comparedwith normoxic conditions (E). Data shown are from four independent ex-periments. npo0.01 (signi¢cant from normoxia).

Figure 4. In vitro wound closure is dependent on keratinocytemigration. PAM 212 keratinocytes monolayers were treated with mito-mycin C or cytochalasin B for 4 h and wounded as described. Cultureswere then incubated in serum-free RPMI containing cytochalasin B orserum-free RPMI alone (mitomycin C pretreated monolayers), under nor-moxic (N) and hypoxic (H) conditions for 24 h. Image analysis of woundarea recovered (mm2) indicated a marked inhibition in cytochalasin B-trea-ted cultures under both normoxic and hypoxic conditions, whereas no sig-ni¢cant di¡erence was observed between mitomycin C-treated and controlcultures in both normoxic and hypoxic conditions. Data shown(mean7SEM) are from three independent experiments.

HYPOXIA, UPA, AND KERATINOCYTE MIGRATION 1307VOL. 119, NO. 6 DECEMBER 2002

plasminogen activation in these cells, resulting from elevated cellsurface expression of uPAR (Graham et al, 1998, 1999). These dif-ferences between keratinocytes and other cell types likely re£ectthe di¡ering requirements for regulation of plasminogen activa-tion in di¡erent tissues, with di¡use plasminogen activation re-sulting from soluble uPA being appropriate in the setting ofacute cutaneous wounding.

Our data indicate that uPA inhibition by amiloride (Vassalliand Belin, 1987) and WX-293 (Sperl et al, 2000), and broad rangeprotease inhibition by p-aminobenzamidine (Geratz and Cheng,1975), completely blocks hypoxia-mediated enhancement of kera-tinocyte migration, strongly suggesting that the presence of func-tional uPA is required for increased keratinocyte migration underhypoxic conditions. Interestingly, whereas the selective uPA inhi-bitors, amiloride and WX-293, had no e¡ect on wound closureunder normoxic conditions, inhibition of serine proteinase activ-ity by p-aminobenzamidine reduced the wound area recovered innormoxic cultures, suggesting that non-uPA-mediated, broad-spectrum protease activation may play a part in keratinocytemigration under normoxic conditions.

Our observation that hypoxia mediates an increase in keratino-cyte migration by regulation of uPA provides the ¢rst evidence ofa role for the plasminogen activator/plasmin system in hypoxickeratinocyte migration. Much emphasis has been placed on therole of MMP’s in cellular motility (Saarialho-Kere et al, 1993; In-oue et al, 1995; Pilcher et al, 1997; Makela et al, 1999), with migrat-ing keratinocytes overexpressing type I and type IV collagenases(Sarret et al, 1992; Pilcher et al, 1997). Previous studies of migrationof human keratinocytes under hypoxia demonstrated upregula-tion of MMP-9 (O’Toole et al, 1997), although functional evi-dence for a role for this protease in keratinocyte migrationis awaited. Interestingly, several MMP’s, including MMP-2and MMP-9, can be activated by uPA-generated plasmin in vitro

(Mazzieri et al, 1997; Carmeliet et al, 1998). Furthermore, uPA-mediated MMP-9 activation was recently implicated in bronchialepithelial cell migration (Legrand et al, 2001), suggesting that hy-poxic induction of uPA may be a proximal step in the inductionof these proteases by hypoxia. The resulting increase in levels ofboth plasmin and active MMP following hypoxia could encou-rage rapid dissolution of the extracellular matrix thereby facilitat-ing keratinocyte migration and enhancing wound closure.

Although our data suggest that uPA-mediated plasminogen ac-tivation promotes keratinocyte motility in vitro, the exact molecu-lar mechanisms underlying this remain unclear. Recent dataindicate that uPA-mediated cell migration can occur in the ab-sence of uPA-speci¢c proteolytic activity suggesting alternative,nonproteolytic roles for this molecule in the regulation of cellmigration. One possibility is that uPA has a chemotactic e¡ecton hypoxic keratinocytes. The chemotactic properties of uPA onother cell types in culture are well documented and are generallyreliant on uPA/uPAR binding (Resnati et al, 1996; Fazioli et al,1997). Recent data, however, showed uPA was chemotactic forvascular smooth muscle cells after mutation of the catalyticdomain and removal of the growth factor-like domain, used foruPAR binding (Poliakov et al, 1999). These ¢ndings indicate thatthe nonproteolytic chemotactic e¡ect of uPA can function bothin the presence and independently of uPAR.

Thus, although the data presented is limited to a murine kera-tinocyte cell line, necessitating further studies to extend the ob-servations to a human system, and the exact mechanism bywhich uPA regulates hypoxic keratinocyte migration is uncertain,the inhibition of enhanced migration by uPA inhibitors shown inthis study indicates that uPA may play a signi¢cant part in regu-lating keratinocyte migratory responses in a hypoxic environ-ment. Despite long-term hypoxia being an impediment tonormal wound healing, migration of keratinocytes is critical for

Figure 5. Inhibition of uPA activity abrogateshypoxia-mediated enhancement of in vitrowound closure. PAM 212 keratinocyte monolayerswere wounded and incubated under normoxic (N)or hypoxic (H) conditions, with or without the ad-dition of amiloride (A), p-aminobenzamidine (B),or the selective uPA inhibitor, WX-293 (C). Imageanalysis of wound area recovered (mm2) indicatedthat the addition of uPA inhibitors suppressed hy-poxic enhancement of in vitro wound closure to alevel equal to or below that observed in cells cul-tured under normoxic conditions. Data shown(mean7SEM) are from three independent ex-periments. npo0.01 (signi¢cantly from normoxia),ns¼ p40.05 (not signi¢cantly from normoxia).

1308 DANIEL ETAL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

e¡ective wound repair and modulation of tissue responses to theearly stages of hypoxia may prove a powerful means of accelerat-ing wound closure in therapeutic situations.

This study was supported by P¢zer Central Research, Sandwich, UK

REFERENCES

Alvarez OM, Mertz PM, Eaglstein WH: The e¡ect of occlusive dressings on collagensynthesis and re-epithelialization in super¢cial wounds. J Surg Res 35:142^148,1983

Bugge TH, Flick MJ, Danton MJ, et al: Urokinase-type plasminogen activator ise¡ective in ¢brin clearance in the absence of its receptor or tissue-type plasmi-nogen activator. Proc Natl Acad Sci USA 93:5899^5904, 1996

Carmeliet P, Moons L, Lijnen R, et al: Urokinase-generated plasmin is a candidateactivator of matrix metalloproteinases during atherosclerotic aneurysm forma-tion. Nat Genet 17:439^444, 1997

Carmeliet P, Moons L, Dewerchin M, Rosenberg S, Herbert J-M, Lupu F, Collen D:Receptor-independent role of urokinase-type plasminogen activator in peri-cellular plasmin and matrix metalloproteinase proteolysis during vascularwound healing in mice. J Cell Biol 140:233^245, 1998

Carriero MV, Del Vecchio S, Capozzoli M, et al: Urokinase receptor interacts withalpha (v) beta5 vitronectin receptor, promoting urokinase-dependent cell mi-gration in breast cancer. Cancer Res 59:5307^5314, 1999

Cha D, O’Brien P, O’Toole EA, Woodley DT, Hudson LG: Enhanced modulationof keratinocyte motility by TGF alpha relative to EGF. J Invest Dermatol106:590^597, 1996

Chang N, Goodson WHD, Gottrup F, Hunt TK: Direct measurement of woundand tissue oxygen tension in postoperative patients. Ann Surg 197:470^478, 1983

Cubellis MV, Nolli ML, Cassani G, Blasi F: Binding of single-chain prourokinase tothe urokinase receptor of human U937 cells. J Biol Chem 261:15819^15822, 1986

Fazioli F, Resnati M, Sidenius N, Higashimoto Y, Appella E, Blasi F: A urokinase-sensitive region of the human urokinase receptor is responsible for its chemo-tactic activity. EMBO J 16:7279^7286, 1997

Fitzpatrick TE, Graham CH: Stimulation of plasminogen activator inhibitor-1expression in immortalized human trophoblast cells cultured under low levelsof oxygen. Exp Cell Res 245:155^162, 1998

Geratz JD, Cheng MCF: The inhibition of urokinase by aromatic diamidines.ThrombDiath Haemorrh 33:230^243, 1975

Graham CH, Fitzpatrick TE, McCrae KR: Hypoxia stimulates urokinase receptorexpression through a heme protein-dependent pathway. Blood 91:3300^3307,1998

Graham CH, Forsdike J, Fitzgerald CJ, Macdonald Goodfellow S: Hypoxia-mediatedstimulation of carcinoma cell invasiveness via upregulation of urokinase recep-tor expression. Int J Cancer 80:617^623, 1999

Grondahl-Hansen J, Lund LR, Ralfkiaer E, Ottevanger V, Dano K: Urokinaseand tissue-type plasminogen activators in keratinocytes during woundreepithelialization in vivo. J Invest Dermatol 90:790^795, 1988

Hibino T, Matsuda Y, Takahashi T, Goetinck PF: Suppression of keratinocyte prolif-eration by plasminogen activator inhibitor-2. J Invest Dermatol 112:85^90, 1999

Inoue M, Kratz G, Haegerstrand A, Stahle-Backdahl M: Collagenase expression israpidly induced in wound-edge keratinocytes after acute injury in human skin,persists during healing, and stops at re-epithelialization. J Invest Dermatol104:479^483, 1995

Jensen PJ, Lavker RM: Urokinase is a positive regulator of epidermal proliferationin vivo. J Invest Dermatol 112:240^244, 1999

Jensen PJ, Baird J, Morioka S, Lessin S, Lazarus GS: Epidermal plasminogen activatoris abnormal in cutaneous lesions. J Invest Dermatol 90:777^782, 1988

Lee SW, Kahn ML, Dichek DA: Expression of an anchored urokinase in the apicalendothelial cell membrane. Preservation of enzymatic activity and enhance-ment of cell surface plasminogen activation. J Biol Chem 267:13020^13027, 1992

Legrand C, Polette M, Tournier J-M, De Bentzmann S, Huet E, Monteau M,Birembaut P: uPA/Plasmin system-mediated MMP-9 activation is implicatedin bronchial epithelial cell migration. Exp Cell Res 264:326^336, 2001

Maity A, Solomon D: Both increased stability and transcription contribute to theinduction of the urokinase plasminogen activator receptor (uPAR) messageby hypoxia. Exp Cell Res 255:250^257, 2000

Makela M, Larjava H, Pirila E, Maisi P, Salo T, Sorsa T, Uitto VJ: Matrix metallopro-teinase 2 (gelatinase A) is related to migration of keratinocytes. Exp Cell Res251:67^78, 1999

Martin P: Wound healingOaiming for perfect skin regeneration. Science 276:75^81,1997

Mazzieri R, Masiero L, Zanetta L, Monea S, Onisto M, Garbisa S, Mignatti P: Con-trol of type IV collagenase activity by components of the urokinase-plasminsystem: a regulatory mechanism with cell-bound reactants. EMBO J 16:2319^2332, 1997

Mignatti P, Rifkin DB, Welgus HG, Parks WC: Proteinases and tissue remodeling.In: Clark RAF (ed.).The Molecular and Cellular Biology ofWound Repair. PlenumPress, New York, 1996 pp 427^474

Morioka S, Lazarus GS, Baird JL, Jensen PJ: Migrating keratinocytesexpress urokinase-type plasminogen activator. J Invest Dermatol 88:418^423,1987

Nguyen DH, Catling AD, Webb DJ, et al: Myosin light chain kinase functionsdownstream of Ras/ERK to promote migration of urokinase-type plasmino-gen activator-stimulated cells in an integrin-selective manner. J Cell Biol146:149^164, 1999

Niinikoski J, Heughan C, Hunt TK: Oxygen tensions in human wounds. J Surg Res12:77^82, 1972

O’Toole EA, Marinkovich MP, Peavey CL, Amieva MR, Furthmayr H, Mustoe TA,Woodley DT: Hypoxia increases human keratinocyte motility on connectivetissue. J Clin Invest 100:2881^2891, 1997

Pepper MS, Sappino AP, Stocklin R, Montesano R, Orci L, Vassalli J-D: Upregula-tion of urokinase receptor expression on migrating endothelial cells. J Cell Biol122:673^684, 1993

Petersen LC, Lund LR, Nielsen LS, Dano K, Skriver L: One chain urokinase-typeplasminogen activator from human sarcoma cells is a proenzyme with little orno intrinsic activity. J Biol Chem 263:11189^11195, 1988

Pilcher BK, Dumin JA, Sudbeck BD, Krane SM,Welgus HG, Parks WC:The activityof collagenase-1 is required for human keratinocyte migration on a type Icollagen matrix. J Cell Biol 137:1445^1457, 1997

Pinsky DJ, Liao H, Lawson CA, et al: Coordinated induction of plasminogenactivator inhibitor-1 (PAI-1) and inhibition of plasminogen activator geneexpression by hypoxia promotes pulmonary vascular ¢brin deposition. J ClinInvest 102:919^928, 1998

Poliakov AA, Mukhina SA, Traktouev DO, et al: Chemotactic e¡ect of urokinaseplasminogen activator: a major role for mechanisms independent of itsproteolytic or growth factor domains. J Recept Signal Transduct Res 19:939^951,1999

Reinartz J, Batrla R, Boukamp P, Fusenig NE, Kramer MD: Binding and activationof plasminogen at the surface of human keratinocytes. Exp Cell Res 208:197^208, 1993

Resnati M, Guttinger M, Valcamonica S, Sidenius N, Blasi F, Fazioli F: Proteolyticcleavage of the urokinase receptor substitutes for the agonist-induced chemo-tactic e¡ect. EMBO J 15:1572^1582, 1996

Romer J, Bugge TH, Pyke C, Lund LR, Flick MJ, Degen JL, Dano K: Impairedwound healing in mice with a disrupted plasminogen gene. Nat Med 2:287^292, 1996

Romer J, Lund LR, Eriksen J, et al: Di¡erential expression of urokinase-type plasmi-nogen activator and its type-1 inhibitor during healing of mouse skin wounds.J Invest Dermatol 97:803^811, 1991

Saarialho-Kere UK, Kovacs SO, Pentland AP, Olerud JE, Welgus HG, Parks WC:Cell^matrix interactions modulate interstitial collagenase expression byhuman keratinocytes actively involved in wound healing. J Clin Invest92:2858^2866, 1993

Sarret Y, Woodley DT, Goldberg GS, Kronberger A, Wynn KC: Constitutive synth-esis of a 92 kDa keratinocyte derived type IV collagenase is enhanced by type Icollagen and decreased by type IV collagen matrices. J Invest Dermatol 99:836^841, 1992

Schaefer BM, Maier K, Eickho¡ U, Todd RF, Kramer MD: Plasminogen activationin healing human wounds. AmJ Pathol 144:1269^1280, 1994

Schaefer BM, Reinartz J, Bechtel MJ, Inndorf S, Ekkehard L, Kramer MD: Dispase-mediated basal detachment of cultured keratinocytes induces urokinase-type plasminogen activator (uPA) and its receptor (uPA-R, CD87). Exp CellRes 228:246^253, 1996

Sperl S, Jacob U, Arroyo de Prada N, et al: (4 -Aminomethyl) phenylguanidinederivatives as nonpeptidic highly selective inhibitors of human urokinase. ProcNatl Acad Sci USA 97:5113^5118, 2000

Stadelmann WK, Digenis AG,Tobin GR: Impediments to wound healing. AmJ Surg176:39S^47S, 1998

Vassalli J-D, Belin D: Amiloride selectively inhibits the urokinase-type plasminogenactivator. FEBS Lett 214:187^191, 1987

Wijnberg MJ, Quax PH, Nieuwenbroek NM, Verheijen JH: The migration of hu-man smooth muscle cells in vitro is mediated by plasminogen activation andcan be inhibited by alpha2-macroglobulin receptor associated protein.ThrombHaemost 78:880^886, 1997

Wilhelm O,Weidle U, Hohl S, Rettenberger P, Schmitt M, Grae¡ H: Recombinantsoluble urokinase receptor as a scavenger for urokinase-type plasminogen acti-vator (u-PA). Inhibition of proliferation and invasion of human ovarian cancercells. FEBS Lett 337:131^134, 1994

Winter GD: Formation of the scab and the rate of epithelialization ofsuper¢cial wounds in the skin of the young domestic pig. Nature 193:293^294,1962

Woodley DT, Kim YH: A double-blind comparison of adhesive bandageswith the use of uniform suction blister wounds. Arch Dermatol 128:1354^1357,1992

Yebra M, Goretzki L, Pfeifer M, Mueller BM: Urokinase-type plasminogen activa-tor binding to its receptor stimulates tumor cell migration by enhancing integ-rin-mediated signal transduction. Exp Cell Res 250:231^240, 1999

HYPOXIA, UPA, AND KERATINOCYTE MIGRATION 1309VOL. 119, NO. 6 DECEMBER 2002