This woHealth (ert May

Accepte

Reprintgery, UnSuite 12gery.usc

0039-60

� 2010

doi:10.1

Matrix metalloproteinase-9 delayswound healing in a murinewound modelMatthew J. Reiss, MD, Yan-Ping Han, PhD, Edwin Garcia, MD, Mytien Goldberg, MD, Hong Yu, PhD,and Warren L. Garner, MD, Los Angeles, CA

Background. Metalloproteinase-9 (MMP-9) is a type IV collagenase found at elevated levels in chronicwounds. As wounds heal, MMP-9 diminishes. In this study, we investigated whether MMP-9 directlycontributes to chronic wound pathogenesis.Methods. Recombinant proMMP-9 was prepared using immortalized keratinocytes transduced by alentivirus. ProMMP-9 was purified from cell culture media and activated using 4-aminophenylmercuricacetate. Active MMP-9 was then suspended in xanthan gum to a concentration paralleling that foundin human chronic wounds. Two parallel 6-mm punch biopsies were made on the backs of C57BL mice.Wounds were treated daily with MMP-9 or vehicle. Wound areas were measured and tissues examined bydensitometry, real-time RT-PCR, histology, and immunohistochemistry at days 7, 10, and 12.Results. Exogenous MMP-9, at the level found within chronic wounds, delayed wound healing in thisanimal model. By 7 days, wounds in the MMP-9--injected group were 12% larger than control wounds(P = .008). By day 12, wounds in the MMP-9--injected group were 25% larger than those of the controlgroup (P = .03). Histologic examination shows that high levels of active MMP-9--impaired epithelialmigrating tongues (P = .0008). Moreover, consistent with elevated MMP-9, the collagen IV in theleading edge of the epithelial tongue was diminished.Conclusion. MMP-9 appears to directly delay wound healing. Our data suggests that this may occurthrough interference with re-epithelialization. We propose that MMP-9 interferes with the basementmembrane protein structure, which in turn impedes keratinocyte migration, attachment, and thereestablishment of the epidermis. (Surgery 2010;147:295-302.)

From the Division of Plastic Surgery, University of Southern California, Los Angeles, CA

DURING THE LAST 2 DECADES MANY INVESTIGATORS haveilluminated the complex interaction of cells, pro-teins, signaling molecules, and the mechanismswhereby organisms heal wounds. Many of thesediscoveries have improved the ability of cliniciansto prevent failed healing of acute wounds and toimprove the healing of wounds that have becomechronic. Nevertheless, billions of dollars are stillnecessary to treat patients with chronic wounds.Although a considerable effort has been put intothe elucidation of the reasons for the chronic

rk was supported by grants from National Institutes ofGM 050967 to W.L.G and AR051558 to Y.P.H) and Rob-Wright Foundation (Y.P.H).

d for publication October 5, 2009.

requests: Warren L. Garner, MD, Division of Plastic Sur-iversity of Southern California, 1200 N. State Street,-700, Los Angeles, CA 90033. E-mail: wgarner@sur-

.edu.

60/$ - see front matter

Mosby, Inc. All rights reserved.

016/j.surg.2009.10.016

wound environment, failed wound healingremains only partially explained.

A prime candidate for a cause for disorderedwound healing is a particular proteinase, matrixmetalloproteinase type 9 (MMP-9), a type IV col-lagenase. Increased levels of MMP-9 have beenidentified by numerous investigators in manychronic wound types. In an acute surgical wound,MMP-9 is transiently expressed, but quickly returnsto basal levels.1 MMP-9 is seen in decubitous ulcersand venous stasis ulcers.2-5 As Young and Grinellreported in 1994, and as we have observed in ourown lab, MMP-9 is also up-regulated in burnwounds.6 In response to injury the cornea alsoexpresses MMP-9.7 While there is a normalincrease in MMP-9 during acute wound healing,MMPs, and in particular MMP-9, are massively ele-vated in the chronic wound environment8 an addi-tional fivefold.5 We and others have hypothesizedthat it is the prolonged and excessive productionof this protease that leads to disordered woundhealing. The disappearance of MMP-9 is associatedwith resolution of the pathologic state. We have

SURGERY 295

SurgeryFebruary 2010

296 Reiss et al

observed that healed areas of burn wound lose ac-tive MMP-9 expression (unpublished data). Tren-grove et al described the same disappearance ofMMP-9 activity as leg ulcers heal.8 Furthermore,levels of MMP-9 have also been inversely correlatedwith delayed collagen deposition in alveolar bonehealing.9

We propose that MMP-9 is not simply associatedwith these wounds, but is in fact one of the causalfactors contributing to the development of chronicwounds. Using an animal model we applied clin-ically defined, high levels of activated, recombi-nant MMP-9 to the full-thickness wounds. Wefound that this treatment decreased type-IV colla-gen in basement membrane, decreased epithelialmigration, and ultimately resulted in delayedwound healing.

MATERIALS AND METHODS

Animal protocol. Thirty C57-BL mice were usedin this study. The study was approved by theInstitutional Animal Care and Use Committee atthe University of Southern California. All animalsare used in accordance with the policies of theIACUC of the University of Southern Californiaunder protocol 10681. Mice were prepared asdescribed by Galiano et al.10 In brief, C57-BLmice were operated on from age 8--12 weeks. Themouse was initially anesthetized using an intraper-itoneal injection of ketamine and xylazine. Afterthe anesthetic takes effect, the mouse was shaved,and a depilatory agent is applied. After 4 min,the dorsum of the mouse was cleaned usingwarm moist tap water. Sterile surgical prep wasthen performed. Two 6-mm punch biopsies werethen outlined using a punch biopsy tool on dor-sum of the animal. We then excised the skin usinga pair of Iris scissors. Krazy Glue� was applied tothe skin around the wound margin and a 5-mmthick piece of silicone with a 7-mm diameter holewas glued around the wound. This was fixed bythe placement of a series of interrupted 4.0 Pro-lene suture placed circumferentially around thesilicone stent. A gauze pledget was then placedinside the stent (that is, in the wound bed). Finally,an occlusive dressing (Tegaderm�) was appliedwith Krazy Glue to seal the wound, with the telfaunder the occlusive dressing (see Fig 1). Injectionsof bupronorphine were administered for the next48 hr for post operative pain control. Animalswere housed in individual cages.

At the time of surgery, the first treatment wasalso applied. 100 ml of active MMP-9 was deliveredinto the telfa pledget in the wound dressing. Thiswas done using a small needle (29.5 gauge), to

avoid destruction of the dressing. Active drug wasalways delivered to the wound on the right, controlto the wound on the left. Control treatment isexplained in the next section. Treatment andcontrol were delivered daily to both wounds, untilmice were killed.

Protein purification. Conditioned media washarvested from immortalized keratinocytes, trans-duced by lentivrus expressing human MMP-9.11

Control media was harvested from similar, butuntransduced immortalized keratinocytes. Thesemedia was then treated with APMA, a chemicalactivator for MMP-9. After overnight incubation at37�C, the media were purified by passage througha gelatin bead column. Protein was eluted using 6M Urea followed by dialysis against saline. Bovine se-rum albumin (BSA) was then added to aid in main-taining the stability of the metalloproteinase. Thiswas then concentrated using ammonium sulfate,and resuspended in a salt buffer. At the same time,control media from untransduced immortalized ke-ratinocytes was treated the same way (addition ofBSA, ammonium sulfate concentration).

In order to approximate a clinically relevantquantity of MMP-9, conditioned media from over-night organ culture of a burn wound skin werecompared in a series of titrations to the preparedtreatment MMP-9, by zymography (Fig 2, C). Aftertitration against our human samples, both the con-trol and MMP-9 treatment groups were mixed in a1:1 ratio with 0.5% xanthan gum as a carrier, yield-ing a final concentration of 0.25% xanthan gum,and a quantity of MMP-9 similar to that found inunhealed burn wounds.

Western blot. Protein samples were loaded onto a10% SDS-PAGE and electrophoresis was performedat room temperature. Samples were then trans-ferred to PVDF membrane, which was then blockedin 5% nonfat milk, 100 mM NaCl, and 50 mM TrispH 7.5 for 16 hr. Samples were then incubated withanti--MMP-9 antibody (AB19016; Chemicon Inter-national) in 5% milk buffer for 2 hr. Membraneswere then washed and incubated with horseradishperoxidase conjugated second antibody. Antibodieswere the visualized using chemiluminescence(Supersignal West Femto Maximum Sensitivity Sub-strate; Pierce, Rockford, IL).

Gelatin zymography. Recombinant protein andsecreted protein from wounds were mixed withSDS-PAGE sample buffer without reducing agents.The PAGE gel contains 0.5% gelatin. Electropho-resis was performed at 114 V, 4�C for 16 hr. Gelswere then washed in 2% triton X-100. Gels werethen incubated at 37�C for 24 hr in a buffer of 50mM Tris pH 7.5, 100 mM NaCl, and 5 mM CaCl2,

Fig 1. Animal model of cutaneous wound healing. Mice were prepared with a commercial depilatory agent, 6-mmpunch biopsies made, and a silicone stent fixed around the wound with Krazy Glue, and 4.0 prolene suture as describedby Gurtner. A Telfa pledget was then placed in the wound bed and tegaderm was glued to the silicone sheeting and thesurrounding skin, creating a pocket into which we injected MMP-9 (right-side wound) or vehicle (left-side wound).Dressing was reinforced with Koban.

SurgeryVolume 147, Number 2

Reiss et al 297

to allow the development of gelatinolytic activity.Finally, gels were stained with Coomassie BlueR-250 to visualize protein bands.

Wound measurements. On the day of killing notreatment was delivered to the wound. The animalwas killed by intraperitoneal injection of pheno-barbital. After the loss of vital signs (respirationsand cardiac function), wounds were processed.The wound and underlying soft tissue was excisedas a unit and laid flat next to a ruler. Photographsof the wounds were taken using the Alpha Im-ager�, and analyzed using the accompanyingAlphaEase FC Software V. 4.1.0 (Alpha InnotechCorp.). Using this software, wounds were outlinedand the circumscribed area was calculated in pixelsby the software and then compared to standardareas. Standards were derived from circle of knownsize drawn on paper which was then analyzed bythe method described above. These measurementswere not blinded.

Histology. After measurement of the wounds,they were embedded in paraffin. After all sampleswere collected, 5-mm thick pieces of tissue were

used to make a series of slides through the woundbed. The slides were then stained with hematoxylinand eosin for analysis of the epithelial tongue sizeand granulation tissue area. All slides were re-viewed by a blinded investigator. Three to 6 slideswere viewed from each wound, and measurementswere made of epithelial tongue length and area ofgranulation. These measurements were then aver-aged for each wound.

Immunohistochemistry. Frozen sections weresliced at a 5-mm thickness. Collagen IV antibody(MD20451; MDBiosciences, St Paul, MN) or IgGfirst antibody was used in 5% nonfat milk solution.Attached antibody was visualized using a secondantibody conjugated with Cy3. Nuclei were stainedwith DAPI.

Statistics. Wound size and epithelial tonguelength were compared using a paired Studentt test. Control wounds were paired with treatmentwounds by animal. Wound sizes were expressedas a ratio of treated versus control wound. Datawas analyzed at each time point as well as pooledfor all animals.

Fig 2. Preparation of MMP-9. MMP-9 was purified fromthe conditioned medium of immortalized keratinocytetransduced by lentivirus expressing MMP-9. (A) The pu-rified MMP-9 protein was faintly detectable by silver stainand was identified by Western blot analysis. (B) The ge-latinolytic activities of our purified MMP-9 as well as themock control were examined by zymography, which ismuch more sensitive. (C) MMP-9 concentration deliv-ered to mice was determined by titration against condi-tion media from overnight incubation of burnedhuman skin.

Fig 3. Wounds. Twenty-five mice were killed at 3 timepoints: 7, 10, and 12 days. The dorsal skin and underly-ing muscle were excised, placed on a flat sheet of paperand photographed from a set distance. Here, controlwounds are shown in dark gray and MMP-9--exposedwounds in light gray, paired by animal. Inter-animal var-iability is notable, but MMP-9--exposed wounds are con-sistently larger then their paired treatment wound.

SurgeryFebruary 2010

298 Reiss et al

RESULTS

We utilized a full-thickness murine woundmodel to test our hypothesized pathogenicity forexcess activated MMP-9 (aMMP-9) (Fig 1). Paired6-mm dorsal wounds were created and stentedopen to prevent healing by contraction (detailsin methods). We and its originators10 believe thismodel more accurately reflects human woundhealing than a simple punch wound. To mimicthe activated proteinase found in human nonheal-ing wounds, recombinant pro-MMP-9 (pMMP-9)was converted to mature active form with apparentmolecular weight of 82-kDa. The gelatinase waspurified to near homogeneity as shown by silverstaining and zymographic activities (Fig 2, A andB). The amount of MMP-9 loaded to wounds wasevaluated by comparison of its gelatinolytic activi-ties to the MMP-9 from chronic wounds (Fig 2, C).

As hypothesized, the exogenous activated MMP-9 delayed wound healing. Figure 3 shows thewound sizes paired by animal. Overall, the woundsexposed to recombinant human MMP-9 healedmore slowly than those treated with the controlsolution. Although there was significant variabilityin the specific rate of wound closure in the ani-mals as a group, in 19 of 25 animals the wound

exposed to MMP-9 closed more slowly than didthe treatment wound (Fig 3). Because each animalhad both a treatment and control wound, this in-ter-animal variability was normalized by comparingtreated to paired control wounds. After 7 days oftreatment, healing was significantly affected, withsignificantly larger open wounds in the MMP-9--treated group than controls. MMP-9--exposedwounds in this group were 12% larger than theuntreated wound (P < .01). By 12 days this differ-ence was even more dramatic. MMP-9--exposedwounds were 25% larger than their paired normalwounds (P < .05). Animals killed at day 10 showeda similar trend towards larger wounds in the trea-ted group, but did not reach statistical signifi-cance (Fig 4).

Histological analysis of the wounds suggested amechanism for this delay in healing after MMP-9treatment (Fig 5). Wounds were compared by ablinded observer and analyzed in various ways forwound-healing critical activities. We first comparedthe length of the migrating epithelial tongue. Inthe MMP-9--treated wounds, epithelial tongueswere 40% (P < .5) smaller than their counterpartcontrol wounds. Moreover, we observed that thenew epithelium had actually detached from theunderlying tissue in many of our H&E-stainedslides. This finding occurred in 43% (41/96) ofthe slides from our MMP-9--treated woundscompared to 15% (13/88) in our control slides(P < .001). These findings suggest a defect inattachment of the new epithelium. Specificallythis indicates that the MMP-9--mediated degrada-tion of the basement membrane zone decreasesepithelial migration. This finding may explain,at least in part, the mechanism of delayed woundclosure. The analysis of granulation tissue suggests

Fig 4. Exogenous MMP-9 delays wound healing. (A) In order to account for the cross-animal variability, MMP-9 woundswere normalized to treatment wound. Control wounds are pictured in blue, and were set to 1. At day 7 MMP-9--exposedwounds were 12% (P < .01) larger, and by day 12 this difference had extended to 25% (P < .05) larger. Day 10 woundshad a similar trend but the variability was so great that the difference was not significant (error bars are SEM). (B) A day12 mouse demonstrates a dramatically larger MMP-9--exposed wound (right), in contrast to the much smaller controlwound (left).

Fig 5. Excessive MMP-9 provokes type-IV collagen degradation and delays epithlia migration. (A) Epithelial tonguemeasurements were made based on hematoxylin and eosin staining. In those exposed to MMP-9, epithelial tongueshad advanced 62% of the distance measured in the control groups, P < .5. (B) IHC staining for collagen IV at theleading edge of epithelium revealed a clear qualitative difference in the quantity of collagen IV in the basementmembrane.

SurgeryVolume 147, Number 2

Reiss et al 299

further that there is an increase in the amount ofgranulation tissue in the wounds exposed toMMP-9. There is slightly greater than twice asmuch area of granulation tissue (2.1), P < .005.This finding supports our clinical observationthat granulation tissue forms in wounds that arenot healing optimally. When they are analyzed bythickness of granulation tissue, there is no appar-ent difference between wounds.

Staining for collagen IV revealed a paucity ofstaining in the basement membrane at the leadingedge of epithelial cell immigration in MMP-9--treated tissues, while the bright staining was seen

in the control wounds (Fig 5). While all woundsshowed some vascular and lymphatic endothelialcell staining, there were no quantitative or qualita-tive differences appreciated based on lymphaticmarkers of LYVE-1 or CD31 staining.

DISCUSSION

MMP-9 is a type IV collagenase that is transientlyexpressed in the process of normal wound healingbut has been found at elevated levels in chronicwounds. This correlation has been observed by anumber of investigators in diverse types of slowlyor nonhealing wounds. MMP-9 synthesis, activation

SurgeryFebruary 2010

300 Reiss et al

and activity are, in part, regulated by tumornecrosis factor-a (TNF-a).12,13 We have proposedthat the chronic, nonhealing wound result fromthis ongoing inflammatory environment. The on-going release of TNF-a provides the proximatemechanism for excess and continued MMP-9 pro-duction. We propose that normally MMP-9 facili-tates re-epithelialization by degrading thebasement membrane, thereby allowing keratino-cyte migration into the wound and wound closure.In situations where inflammation continues, TNFand subsequently MMP-9 persist, preventing mi-grating keratinocytes from forming new attach-ments to a newly synthesized basementmembrane. This hypothesis suggests a mechanismwhereby the presence of MMP-9 at these highlevels delays the process of normal wound healing.

The murine model we used to test our hypoth-eses is based on the work by Galiano et al at NYU.10

The advantage of this model is that it reduces thedramatic wound contraction usually seen in thenormal process of murine wound healing. Rather,these mice heal by epithelialization and granula-tion, better reflecting normal human wound heal-ing. We agree with those authors who havesuggested that this model more accurately reflectshuman wound healing, which depends on re-epi-thelialization and granulation, with lesser impor-tance of wound contraction. The reader mightreasonably imagine several ways that a collagenasemight affect wound contraction. However, we didnot examine this possibility in the current study.

The murine model has limitations, of course.Individual members of a species heal at differentrates. Mice are not human, so we are creating anartificial ‘‘human wound.’’ Mice are practical,however, and easy to handle. They also providedus with enough surface area to still have pairedtreatment and control wounds.

A second point is the choice of control. Mediafrom the same immortalized keratinocyte cell line,but not transduced, went through the sameextraction process and was then dissolved in thesame xanthan gum carrier that the MMP-9 wasdelivered in. We acknowledge that there areseveral other options for controls. Any collagenaseat excessively levels might be expected to showsimilar results. MMP-9 has been a molecule ofparticular interest and appears to target type IVcollagen and basement membranes. Furthermore,MMP-9 is the only proteinase consistently found innonhealing and chronic wounds. Anotherpotential control would be treatment MMP-9 plusan MMP-9--specific inhibitor. As studies proceed wehope to explore these possibilities in future studies

to determine a specific mechanism for MMP-9action.

One also must ask what role murine MMP-9 isplaying in these wounds. Murine MMP-9 is similarto human and is detected by the same zymographyas human MMP-9. It appears not to be significantin our model. Figure 2, B, shows protein extractedfrom wound dressings on the day of killing. Thisfluid lets us assess ‘‘wound fluid.’’ Note both thepro and the active MMP-9 is seen only in the trea-ted wound, and not in the control. Thus weconclude that native murine MMP-9 was not animportant component of any of the responsesfound in this study.

Between individual mice, there was a fairamount of difference in wound healing. Thereare many possible causes for this variability, includ-ing differences in actual delivered dosage of MMP-9, partial dislodgement of the dressing, loss ofMMP-9 into the telfa of the dressing, and mostimportantly individual differences between mice.In an attempt to minimize the inter-animal differ-ences we created donor and control wounds on theback of each individual animal, giving each woundits own ‘‘internal control.’’ This allowed us tonormalize wounds between animals. To avoid con-fusion the left wound was always the controlwound, and the right always the treatment wound.We acknowledge that this may introduce somebias. It is possible that all the animals favor 1 side---for example, always sleep on their right, or alwaysscratch on the left because they can better see thedoor to the lab from that vantage. This might inturn bias our results. Nevertheless, despite thisvariability, we found that the application of MMP-9clearly delayed wound healing.

This model has the advantage that it allowed usto have both MMP-9--exposed and treatmentwounds on the back of a single animal. In orderto provide a stable reservoir for our collagenaseand vehicle, we used a dressing that preventeddaily analysis of the wound. We chose three timepoints---7, 10, and 12 days---to kill animals andmeasure wound size. In the previous description ofthis model, wounds healed by 12--14 days. Even atthe earliest time point, there is already a significantdelay in the MMP-9--treated group, with woundsbeing 12% larger than wounds in the controlgroup. This delay in healing persists so thatMMP-9--treated wounds were 25% larger than thecontrol wounds by day 12. While MMP-9 alsodelayed healing of wounds in the day 10, the smallnumber of animals made it difficult to find statis-tical significance at this time point, although thetrend was the same. The larger average difference

SurgeryVolume 147, Number 2

Reiss et al 301

seen at day 10 reflects the smaller sample size atthis time point, and the inter-animal variability(see Figs 3 and 4). Our results provide direct evi-dence that MMP-9 plays a causal role in delayingwound healing. Clearly, MMP-9 significantly delayswound healing in this model.

What then is the mechanism of the delay inwound healing? Collagen IV, an essential compo-nent of basement membranes, is one of the majorsubstrates of MMP-9. Hematoxylin and eosin stain-ing of slides suggested some alteration in thearchitecture of the basement membrane. In almosthalf the slides in the MMP-9--treated group we sawthe new epithelium lifting away from the underly-ing tissue bed (43%). In contrast, we seldom sawthis in slides from the normal wounds (13%). Thisfinding suggests a likely mechanism for MMP-9--induced delay of wound healing, that theincreased fragility of basement extra-cellularmatrix (ECM) impairs keratinocyte attachment.We chose, therefore, to stain for collagen IV, whichfits the criteria of both being important in theepidermal-dermal attachments and being a sub-strate of MMP-9. This staining yielded strikingresults. As demonstrated in Fig 4, MMP-9--treatedwounds show almost no staining for collagen IVat the dermal/epidermal junction (Fig 4, A andB). While it seems likely that a certain level ofMMP-9 might be needed in order to release kerat-inocytes from the underlying matrix and allow mi-gration, persistent excess MMP-9 appeared tointerfere with the reattachment of the new epider-mis. Keratinocytes are able to migrate into thewound bed, but then fail to reestablish dermal ep-idermal attachment, and the keratinocytes fall offthe wound---leaving a shorter, and likely more frag-ile epithelial tongue. This is reflected in our epi-thelial tongue measurements, which were 40%shorter in the MMP-9--exposed wounds.

We also observed an increase of granulationtissue in the MMP-9--treated wounds. The mecha-nistic reason for this is unknown. Because granu-lation tissue is primarily newly synthesized bloodvessels, it is worthwhile considering how MMP-9affects this process. There are contradictoryreports about the role MMP-9 plays in revascular-ization of injured tissues; some reports suggestMMP-9 delays angiogenesis, and others that itaccelerates it.14-16 When we looked for differencesin endothelial cell presence within the wound,there was no difference. Furthermore, the differ-ences observed in granulation were in area, notdepth, ie, the larger open wounds in MMP-9--ex-posed mice had a greater area of granulation tis-sue, but not thicker granulation. It may well be

that the greater area of granulation is really a re-flection of persistently open wounds rather thanthe direct impact of MMP-9 on the vascularizationthat is part of granulation tissue production. MMP-9 plays a role in angiogenesis. Because we appliedMMP-9 topically in this model it is difficult to knowhow deep within the wound our topical MMP-9penetrated. We did perform staining for both lym-phatic and blood vessels, but were unable to appre-ciate any differences between treated anduntreated wounds (data not shown); however, be-cause we did not predictably deliver MMP-9 withinthe deeper layers of the wound bed, we hesitate todraw any conclusions.

Taken together these findings lead us toconclude that when MMP-9 is expressed at exces-sive levels, it prevents the reestablishment of thedermal/epidermal junction and thereby limitsepithelial migration and wound closure. This isin contrast to normal wound healing, when tran-sient MMP-9 expression may facilitate keratinocytedetachment and migration into the wound bed.During increased MMP-9 expression in chronicwounds, keratinocytes can begin to migrate intothe wound, but they are unable to re-anchorthemselves to the matrix. This results in the lossof these cells, a shorter epithelial tongue, and adelay or failure of wound closure.

In conclusion, the results presented heredemonstrate that a specific proteinase, MMP-9,may have a direct pathogenic role in impairedwound healing. This likely occurs by interferingwith the reestablishment of a healthy basementmembrane. Although it appears to participatenormal healing, MMP-9 is deregulated andexpressed at excessive levels in delayed woundhealing. The advantages both to the individualpatient with a chronic wound, as well as to thehealth care system as a whole in terms of costare manifold. A number of MMP-9 inhibitors arealready available. Further investigations intothe use of such drugs will be the focus of futureinvestigations.

We thank Chunli Yan for technical support.

REFERENCES

1. Vaalamo M, Weckroth M, Puolakkainen P, Kere J, SaarinenP, Lauharanta J, et al. Patterns of matrix metalloproteinaseand TIMP-1 expression in chronic and normally healing hu-man cutaneous wounds. Br J Dermatol 1996;135:52-9.

2. Weckroth M, Vaheri A, Lauharanta J, Sorsa T, Konttinen YT.Matrix metalloproteinases, gelatinase and collagenase, inchronic leg ulcers. J Invest Dermatol 1996;106:1119-24.

3. Wysocki AB, Staiano-Coico L, Grinnell F. Wound fluidfrom chronic leg ulcers contains elevated levels of

SurgeryFebruary 2010

302 Reiss et al

metalloproteinases MMP-2 and MMP-9. J Invest Dermatol1993;101:64-8.

4. Yager DR, Zhang LY, Liang HX, Diegelmann RF, CohenIK. Wound fluids from human pressure ulcers containelevated matrix metalloproteinase levels and activitycompared to surgical wound fluids. J Invest Dermatol1996;107:743-8.

5. Trengove NJ, Langton SR, Stacey MC. Biochemical analysisof wound fluid from nonhealing and healing chronic leg ul-cers. Wound Repair Regen 1996;4:234-9.

6. Young PK, Grinnell F. Metalloproteinase activation cascadeafter burn injury: a longitudinal analysis of the humanwound environment. J Invest Dermatol 1994;103:660-4.

7. Daniels JT, Geerling G, Alexander RA, Murphy G, Khaw PT,Saarialho-Kere U. Temporal and spatial expression of ma-trix metalloproteinases during wound healing of humancorneal tissue. Exp Eye Res 2003;77:653-64.

8. Trengove NJ, Stacey MC, MacAuley S, Bennett N, Gibson J,Burslem F, et al. Analysis of the acute and chronic woundenvironments: the role of proteases and their inhibitors.Wound Repair Regen 1999;7:442-52.

9. Reinhardt RA, Lee HM, Schmid M, Payne JB, Golub L.Relationship between gelatinases and bone turnover inthe healing bone defect. J Oral Maxillofac Surg 2005;63:1455-60.

10. Galiano RD, Michaels J, Dobryansky M, Levine JP, GurtnerGC. Quantitative and reproducible murine model of exci-sional wound healing. Wound Repair Regen 2004;12:485-92.

11. Han YP, Yan C, Garner WL. Proteolytic activation of matrixmetalloproteinase-9 in skin wound healing is inhibited byalpha-1-antichymotrypsin. J Invest Dermatol 2008;128:2334-42.

12. Han YP, Tuan TL, Hughes M, Wu H, Garner WL. Trans-forming growth factor-beta - and tumor necrosis factor-alpha -mediated induction and proteolytic activation ofMMP-9 in human skin. J Biol Chem 2001;276:22341-50.

13. Han YP, Nien YD, Garner WL. Tumor necrosis factor-alpha-induced proteolytic activation of pro-matrix metalloprotei-nase-9 by human skin is controlled by down-regulatingtissue inhibitor of metalloproteinase-1 and mediated bytissue-associated chymotrypsin-like proteinase. J Biol Chem2002;277:27319-27.

14. Colnot DR, Roos JC, de Bree R, Wilhelm AJ, Kummer JA,Hanft G, et al. Safety, biodistribution, pharmacokinetics,and immunogenicity of 99mTc-labeled humanized mono-clonal antibody BIWA 4 (bivatuzumab) in patients withsquamous cell carcinoma of the head and neck. Cancer Im-munol Immunother 2003;52:576-82.

15. Ulrich D, Lichtenegger F, Unglaub F, Smeets R, Pallua N.Effect of chronic wound exudates and MMP-2/-9 inhibitoron angiogenesis in vitro. Plast Reconstr Surg 2005;116:539-45.

16. Lindsey ML, Escobar GP, Dobrucki LW, Goshorn DK,Bouges S, Mingoia JT, et al. Matrix metalloproteinase-9gene deletion facilitates angiogenesis after myocardial in-farction. Am J Physiol Heart Circ Physiol 2006;290:H232-9.