Expression and role of matrix metalloproteinases and thelaminin-5 γγγγ2-chain in wound healing and cell migration

Emma Pirilä

Helsinki 2003

1

Expression and role of matrix metalloproteinases and the laminin-5 γ2-chain in woundhealing and cell migration

Emma Pirilä

Department of Clinical Veterinary Sciences, Faculty of Veterinary Medicineand

Institute of Dentistry, Faculty of Medicine, Department of Oral and Maxillofacial Diseases, HelsinkiUniversity Central Hospital, University of Helsinki, Helsinki, Finland

andInstitute of Dentistry, Faculty of Medicine,

University of Oulu, Oulu, Finland

Academic Dissertation

To be publicly discussed with the permission of the Faculty of Veterinary Medicine, University ofHelsinki in Auditorium XII at the University of Helsinki, Unioninkatu 34, Helsinki on the 5th of

December at noon.

Helsinki 2003

2

Supervised by:

Professor Päivi Maisi, DVM, PhDDepartment of Clinical Veterinary Sciences, Faculty of Veterinary MedicineUniversity of Helsinki, HelsinkiFinland

and

Professor Timo Sorsa, DDS, PhD, Dipl. Perio.Institute of Dentistry, Faculty of MedicineUniversity of Helsinki, HelsinkiFinland

and

Professor Tuula Salo, DDS, PhDInstitute of Dentistry, Faculty of MedicineUniversity of Oulu, OuluFinland

Reviewed by:

Professor Aarne OikarinenDepartment of Dermatology and Venereology, Faculty of MedicineUniversity of Oulu, OuluFinland

and

Professor of Cell BiologyJuha PeltonenDepartment of Anatomy and Cell Biology, Faculty of MedicineUniversity of Oulu, OuluFinland

Opponent:

Professor Carlos Lopez-OtinDepartemento de Bioquimica y Biologia MolecularFacultad de MedicinaInstituto Universitario de OncologicaUniversidad de Oviedo, OviedoEspaña

ISBN 952-91-6624-9

Available in electronic formatISBN 952-10-1487-3 (URL: http://ethesis.helsinki.fi)

Helsinki 2003Yliopistopaino

3

Dedicated to the memory of

my grandparents Enni & Juho Pirilä

and

Professor Päivi Maisi

4

5

Pirilä, Emma, Expression and role of matrix metalloproteinases and the laminin-5 γγγγ2-chain in wound healing and cell migrationDepartment of Clinical Veterinary Sciences, Faculty of Veterinary Medicine and Institute ofDentistry, Faculty of Medicine, University of Helsinki and Department of Oral MaxillofacialDiseases, Helsinki University Central Hospital, Helsinki, Finland and Institute of Dentistry, Facultyof Medicine, University of Oulu, Oulu, Finland.

Abstract

Inflammation is the host response to resolve any intrusion or attack to the organism. The hostresponse involves removal of the foreign particles or regeneration of open surfaces, extracellularmatrix (ECM) remodelling and cell migration. At the tissue level, inflammation occurs through theco-ordinated action of multiple molecular pathways. The matrix metalloproteinases (MMPs) areendopeptidases which cleave most ECM and basement membrane proteins. MMPs mediateimportant functions during normal physiological remodelling and development, however, MMPactivity is known to be aberrantly upregulated during pathological conditions, including chronicinflammatory diseases and tumor cell invasion. Thus, many attempts have been made to developsynthethic inhibitors against MMP activity. Laminin-5 (Ln-5) is a heterotrimeric glycoprotein part ofthe hemidesomosomal complex which anchors surface epithelia to the underlying stroma. The Ln-5 γ2-chain can be cleaved by MMP-2 and -14, which induces cell migration.In this study, the role and expression of MMPs were studied in chronic inflammatory periodontaldisease tissue representing a typical disease state, where the hosts ability to resolve the primaryinflammation is impaired. In conjunction with this, the effect of inflammatory-related cytokines wasstudied in cultured human oral mucosal keratinocytes and different MMP inhibitors (MMPIs) weretested. Wound healing is known to be delayed due to estrogen-deprivation, which is a typical stateaffecting women past menopause. In this study, MMP expression and activity as well as the Ln-5γ2-chain processing were studied in cutaneous wound healing of estrogen-deficient rats, and theeffect of estrogen or MMPI treatment was additionally assessed. Finally, the direct ability ofdifferent MMPs to process the Ln-5 γ2-chain and subsequently induce carcinoma cell migrationwas studied in vitro.The results demonstrate a role for cytokine regulation of MMP-2 and –9 in human oral mucosalkeratinocytes. MMP-2 and –9 activities could be downregulated in vitro by chemically modifiedtetracyclines as well as in situ by an MMP-2 and –9 specific inhibitor, the CTTHWGFTLC-peptide.Cutaneous wound healing in estrogen-deprived rats was impaired due to significantly reducedcollagen deposition and content, instability of the re-epithelialized wound surface, changes in MMPactivity and expression as well as aberrant Ln-5 γ2-chain processing. These changes could bereversed by treating the animals with estrogen or a chemically modified tetracycline. SeveralMMPs in addition to the previously characterized MMP-2 and –14 were found to process the Ln-5γ2-chain and induce carcinoma cell migration.In summary, the results demonstrate a role for MMPs and the Ln-5 γ2-chain in tissue degradationand remodelling associated with inflammatory processes and in regulating cell migration. The useof MMPIs could be useful in the treatment of both cancer and chronic inflammatory diseases.

6

7

Acknowledgements

This work was carried out at the Institute of Dentistry, University of Helsinki during the years 1998-2003. The head of the department, Jukka H Meurmann is acknowledged for providing the researchfacilities at the Institute of Dentistry. Many thanks to my custos, the dean of Faculty of VeterinaryMedicine, Hannu Saloniemi for support during the final phases of this thesis.My deepest gratitude to my supervisors and mentors Professor Timo Sorsa, DDS, PhD, Dipl.Perio,Professor Tuula Salo, DDS, PhD, and Professor Päivi Maisi, DVM, PhD, for all the things I havelearned from you, for sharing your scientific skills with me and for all the fun we have had together.Thanks to Professor Aarne Oikarinen and Professor Juha Peltonen for critical reading and helpfulcomments of this thesis.Thanks to my collaborator and mentor Professor Nungavarm S Ramamurthy, DVM, PhD,Department of Oral Biology and Pathology, School of Dental Medicine, SUNY at Stony Brook, NewYork, USA for sharing his knowledge in science with me and for all these years of friendship andfruitful collaboration. Thanks to my other collaborators at Stony Brook, Professor Lorne M Golub,DDS, Allan Kucine, DDS and Steve McClain, MD.Thanks to my collaborators at the Scripps Research Institute, La Jolla, USA: Professor VitoQuaranta, PhD and Andrew Sharabi, BSc for your invaluable input to the work in this thesis.Thanks to Dr. Naohiko Koshikawa, PhD, Department of Cancer Cell Research, University ofTokyo, Japan for collaboration and critical reading of this thesis.Many thanks to my collaborators and co-authors Docent Erkki Koivunen, PhD, Marja Mäkelä, DDS,PhD, Professor Veli-Jukko Uitto, DDS, PhD and Professor Hannu Larjava, DDS, PhD.Thanks to all my collaborators and dear friends at the Institute of Dentistry, University of Oulu;Mataleena Parikka, DDS, Tuire Salonurmi, MSc, Merja Ylipalosaari, DDS, Meeri Sutinen, PhD, PiaNyberg, MSc and others. Thanks also to my collaborators Ritva Heljasvaara, PhD and HongminTu, MSc, at the Department of Medical Biochemistry and Molecular Biology, University of Oulu,Thanks to all the people in our research group at the Research Laboratory, Institute of Dentistry,Biomedicum, University of Helsinki. You are all dear friends: Taina Tervahartiala, DDS, PhD,Jaana Wahlgren, DDS, PhD, Pia Heikkilä, DDS, Mathias Stenman, BcDS and Anne Järvensivu,DDS. Thanks for professional assistance, fruitful discussions and extremely entertaining andrelaxing evenings outside the lab. A special thanks to my dear friends Marjo Kivelä-Rajamäki,DDS, Kaiu Prikk, MD, PhD and Heidi Kuula, DDS for all the great fun and support in all kinds ofmatter of life.Thanks to the laboratory staff, Mrs Annikki Siren, Ms Marjatta Kivekäs, Ms Ritva Keva and MrJukka Inkeri for invaluable technical assistance and a special thanks to our laboratory supervisorKirsti Kari, MSc for all help during these years.

8

My deepest gratitude to my mother Eine Pirilä for all the support and love I have got and for alwaysbelieving that I could really do this. I am grateful to my pseudo-grandmother Margit Lundin forbeing my loyal supporter in all matters of life. A million thanks to my godparents for encouragementand support.Thanks to my dear friends Marika Uusitalo, MSc and Reija Paananen, PhD for all these years offriendship. Thanks to all my non-scientific friends, Ms Carina Snäll, Ms Carina Tammilahti, MsLena Juntti and others. A special thanks to my friend Kari Mankinen and his mother Aira Mankinenfor encouragement and support. Many thanks to my dear friend Stina Eklund-Uusitalo, DVM foryour friendship and for letting me ride your horse whenever I felt a need for getting out in thenature.Loving thanks to Timo Korkiamäki, MD, PhD, for sharing his expertise in the human anatomy withme.I am in deep gratitude to my dalmatian dog Olli, for taking care of my daily oxygen-uptake outdoorsand for keeping my mental health at homeostasis.The research work in our laboratory at the Institute of Dentistry, Biomedicum, University of Helsinkihas been supported by grants from the Helsinki University Research Funds, the Helsinki UniversityCentral Hospital EVO (TI020Y0002)-grant, the Finnish Dental Society Apollonia, The Else andWillhelm Stockman Foundation. The work at the Department of Clinical Veterinary Sciences,Faculty of Veterinary Medicine, University of Helsinki was supported by a grant from the FinnishAcademy. I wish to thank the Biomedicum Helsinki Foundation, K.Albin Johansson Foundation andFinska Läkaresällskapet for supporting my thesis with personal grants as well as the MagnusEhrnrooth Foundation and the Helsinki University’s Chancellor Fund for travelling grants.

Helsinki, November 2003

Emma Pirilä

9

Abbreviations

α1-PI α1-proteinase inhibitorα2M α2-macroglobulinAP Adult periodontitisAPMA p-aminophenylmercuric acetateAP-1 Activator protein 1BM Basement membraneC- Carboxy-Ca2+ Calcium-ionCMT Chemically modified tetracyclineCTT CTTHWGFTLC-peptideEB Epidermolysis bullosaECM Extracellular matrixEGF Epidermal growth factorEDTA Ethylenediaminotetraacetic-acidER Estrogen receptorFA Focal adhesionFC Focal contactGCF Gingival crevicular fluidHCC Hepatocellular carcinoma cellsHCl Hydrochloride acidHD HemidesmosomeHMK Human oral mucosal keratinocytesHRT Hormone replacement therapyIC50 Inhibitory concentration on 50% of the measured parameterIL InterleukinkDa kiloDaltonKGF Keratinocyte growth factorLDD Low-dose doxycyclineLn LamininLPS Lipopolysaccharidem milliM MolarityMMP Matrix metalloproteinaseMMPI Matrix metalloproteinase inhibitormRNA messenger RNAMT-MMP Membrane type matrix metalloproteinaseN- Amino-n NanoNHK Normal human keratinocytesOVX OvariectomizedPBS Phosphate-buffered salinePg PlasminogenPI3K Phosphoinositide 3-OH-kinasePMN PolymorphonuclearSCC Squamous cell carcinomaSDS-PAGE Sodium dodecyl sulphate polyacrylamide gel electrophoresisSE Sulcular epitheliumTAT-2 Tumor associated trypsinogen-2TC TetracyclineTGF-β Transforming growth factor βTIMP Tissue inhibitor of metalloproteinasesTNF-α Tumor necrosis factor α

10

X any amino acidY any amino acidZn2+ Zinc-ion

α alphaβ betaγ gammaµ micro [Mu]

Amino acid abbreviations:

A Ala Alanine M Met MethionineC Cys Cysteine N Asn AsparagineD Asp Aspartate P Pro ProlineE Glu Glutamate Q Gln GlutamineF Phe Phenylalanine R Arg ArginineG Gly Glycine S Ser SerineH His Histidine T Thr ThreonineI Ile Isoleucine V Val ValineK Lys Lysine W Trp TryptophanL Leu Leucine Y Tyr Tyrosine(Alberts et al. 1994)

11

List of original publications

I Mäkelä M, Larjava H, Pirilä E, Maisi P, Salo T, Sorsa T & Uitto V-J (1999) Matrixmetalloproteinase 2 (Gelatinase A) is related to migration of keratinocytes. Exp CellRes 251: 67-78.

II Pirilä E, Maisi P, Salo T, Koivunen E & Sorsa T (2001) In vivo localization ofgelatinases (MMP-2 and –9) by in situ zymography with a selective gelatinase inhibitor.Biochem Biophys Res Commun 287: 766-774.

III Pirilä E, Ramamurthy N, Maisi P, McClain S, Kucine A, Wahlgren J, Golub LM, Salo T& Sorsa T (2001) Wound healing in ovariectomized rats: Effects of chemically modifiedtetracycline (CMT-8) and estrogen on matrix metalloproteinases-8, -13 and type Icollagen expression. Curr Med Chem 8: 281-294.

IV Pirilä E, Parikka M, Ramamurthy NS, Maisi P, McClain S, Kucine A, Tervahartiala T,Prikk K, Golub LM, Salo T & Sorsa T (2002) Chemically modified tetracycline (CMT-8)and estrogen promote wound healing in ovariectomized rats: Effects on matrixmetalloproteinase-2, membrane type 1 matrix metalloproteinase, and laminin-5 γ2-chain. Wound Rep Reg 10: 38-51.

V Pirilä E, Sharabi A, Salo T, Quaranta V, Tu H, Heljasvaara R, Koshikawa N, Sorsa T &Maisi P (2003) Matrix metalloproteinases process the laminin-5 γ2-chain and regulateepithelial cell migration. Biochem Biophys Res Commun 303: 1012-1017.

12

13

ContentsAbstractAcknowledgementsAbbreviationsList of original publicationsContents1. Introduction2. Review of the literature………….……….……….……….……….……….……….……….……… 17

2.1. The collagenous extracellular matrix………………………………………………................ 172.2. Matrix metalloproteinases……………………………………………………………………… 17

2.2.1. General characteristics of MMPs…….……….…….... ……….……….……….……… 192.2.2. Collagenases……………………………………………………………………….……… 192.2.3. Gelatinases………………………………………………………………………...……… 212.2.4. Stromelysins……………………………………………………………………………….. 212.2.5. Matrilysins………………………………………………………………………………….. 222.2.6. MT-MMPs…………………………………………………………………………………...222.2.7. The X-files of the MMPs………………………………………………………………….. 232.2.8. Activation mechanisms of MMPs…………………………………………………………23

2.2.8.1. Cell-surface activation by MMP-14………………………………………………… 252.2.9. Transcriptional regulation of MMPs…………………………………………………….. 262.2.10. In vivo functions of MMPs……………………………………………………….……… 272.2.11. In vivo localization of MMPs…………………………………………………….……… 28

2.3. MMP inhibition…………………………………………………………………..……….……… 282.3.1. Physiological inhibition of MMPs………………………………………………...……… 282.3.2. Synthetic MMPIs………………………………………………………………….. ……… 29

2.3.2.1. Batimastat……………………………………………………………………………. 292.3.2.2. Marimastat……………………………………………………………………………. 292.3.2.3. Tetracyclines…………………………………………………………………………. 302.3.2.4. The chemically modified tetracyclines…………………………………......……… 302.3.2.5. The CTTHWGFTLC-peptide………………………………………………………...32

2.3.3. MMPIs in clinical trials……………………………………………………………………..332.4. Laminin-5………………………………………………………………………………………… 33

2.4.1. Structure of the heterotrimeric Ln-5 molecule………………………………………….. 342.4.2. Ligands for Ln-5…………………………………………………………………………… 342.4.3. Processing of the Ln-5 chains…………………………………………………………….352.4.4. In vivo functions of Ln-5…………………………………………………………………...36

2.5. Cell migration……………………………………………………………………………………. 362.5.1. Mechanisms of cell migration……………………………………………………………..362.5.2. Proteolysis and ECM remodelling during cell migration…………………………….....372.5.3. MMPs and tumor cell migration………………………………………………………….. 372.5.4. Ln-5 in physiological cell migration……………………………………………………… 382.5.5. Ln-5 and tumor cell migration……………………………………………………………. 39

2.6. Periodontal tissue……………………………………………………………………….……… 392.7. Skin……………………………………………………………………………………….……… 40

2.7.1. The basement membrane…………………………………………………………………422.7.1.1. The hemidesmosomal complex………………………………………………… 42

2.7.2. Estrogen and skin…………………………………………………………………………. 432.7.2.1. Estrogen and skin collagen content…………………………………….……… 43

2.8. General characteristics of inflammation……………………………………………………….442.8.1. Chronic inflammatory periodontal disease………………………………………………45

2.8.1.1. MMPs in chronic periodontal disease…………………………………..............… 462.8.1.2. MMP inhibition in periodontal disease…………………………….……….……… 46

2.8.2. Cutaneous wound healing………………………………………………………………...462.8.2.1. Clotting………………………………………………………………………………... 47

14

2.8.2.2. The inflammatory response………………………………………………………… 472.8.2.3. Re-epithelialization………………………………………………………………….. 472.8.2.4. Wound contraction……………………………………………………………………472.8.2.5. Growth factors regulating wound healing…………………………………………. 48

2.8.2.5.1. TGF-β and scarring……………………………………………………………..482.8.2.6. MMPs in wound healing…………………………………………………………….. 492.8.2.7. MMP inhibition in wound healing……………………………………………………492.8.2.8. Ln-5 in wound healing………………………………………………………………..50

3. Aims of the study………………………………………………………………………………………514. Materials and Methods………………………………………………………………………………..52

4.1. Animals……………………………………………………………………………………………524.2. Tissue samples…………………………………………………………………………………..524.3. Materials…………………………………………………………………………………………..524.4. Cell lines…………………………………………………………………………………………..534.5. MMP inhibition of cell proliferation……………………………………………………………..534.6. Cell migration assays……………………………………………………………………………53

4.6.1. Radial migration…………………………………………………………………………… 534.6.2. Scratch assay……………………………………………………………………………… 534.6.3. Transwell migration……………………………………………………………………….. 53

4.7. Collagenase and gelatinase activity assays…………………………………………………. 544.8. Zymography………………………………………………………………………………………544.9. Western immunoblotting……………………………………………………………………….. 554.10. In situ zymography……………………………………………………………………………..554.11. Immunohistochemistry…………………………………………………………………………564.12. In situ hybridization……………………………………………………………………………. 564.13. MMP cleavage in vitro………………………………………………………………………… 574.14. N-terminal sequencing…………………………………………………………………………57

5. Results………………………………………………………………………………………………….585.1. MMP inhibition and cell growth…………………………………………………………………585.2. MMP inhibition and gelatinase production by cells………………………………………….. 585.3. MMP inhibition and cell migration……………………………………………………………...58

5.3.1. Radial migration…………………………………………………………………………… 585.3.2. Scratch assay……………………………………………………………………………… 585.3.3. Transwell assay…………………………………………………………………………….59

5.4. Immunolocalization of MMP-2, MMP-9, Ln-5 γ2-chain and CD45-positive cells in inflamed human gingival tissue………………………………………………………………………….. 59

5.5. MMP-2 and MMP-9 mRNA expression in inflamed gingival tissue………………………...595.6. In situ gelatin zymography……………………………………………………………………... 595.7. Immunolocalization of MMP-2 and Ln-5 in epithelial in vitro wounds……………………... 60

5.7.1. Controls for in situ hybridization and immunohistochemistry………………………… 605.8. Effect of estrogen and CMT-8 in rat cutaneous wound healing…………………………….60

5.8.1. Wound collagen content………………………………………………………………….. 605.8.2. Collagenase and gelatinase activity…………………………………………………….. 605.8.3. Expression of collagenases in rat day 7 cutaneous wounds…………………………. 61

5.8.3.1. MMP-8 and MMP-13 protein and mRNA expression……………………………. 615.8.3.2. Western immunoblotting of MMP-8 and MMP-13………………………………... 61

5.8.4. Expression of gelatinases and MMP-14 in rat day 7 cutaneous wounds…………… 615.8.4.1. MMP-2………………………………………………………………………………… 615.8.4.2. MMP-9………………………………………………………………………………… 625.8.4.3. MMP-14………………………………………………………………………………..62

5.8.5. Basement membrane……………………………………………………………………...625.8.5.1. Ln-5 γ2-chain expression…………………………………………………………….625.8.5.2. Ln-5 γ2-chain processing…………………………………………………………….63

5.9. Ln-5 γ2-chain and β-casein processing by MMPs in vitro…………………………………...63

15

5.9.1. N-terminal sequences of MMP cleaved Ln-5 γ2-chain fragments…………………….635.9.2. Cell migration over MMP cleaved Ln-5…………………………………………………. 63

6. Discussion……………………………………………………………………………………………...656.1. MMP inhibition and HMK cell proliferation…………………………………………………….656.2. MMP and Ln-5 γ2-chain expression in HMKs and inflamed gingival tissue……………….656.3. MMP inhibition in HMKs and inflamed periodontal tissue…………………………………...656.4. Cutaneous wound healing………………………………………………………………………656.5. Therapeutic aspects of estrogen and MMPIs in cutaneous wound healing……………….656.6. Cell migration and Ln-5 γ2-chain processing………………………………………………… 696.7. MMP inhibition-last aspects……………………………………………………………………. 72

7. Conclusions…………………………………………………………………………………………….75ReferencesOriginal publications

16

1. Introduction

An acute inflammation is a host response to resolve tissue injury caused by foreign particles,pathogenic attack or mechanical tissue injury. The inability of the host to resolve an acuteinflammation and restore tissue integrity leads to chronic inflammation and host-derived matrixdegradation. Inability to resolve inflammation may be due to nutritional factors, disease-related orhormonally influenced. Many of the molecular mechanisms underlying progression from acute tochronic inflammation are unresolved. Cell migration is a critical feature for many physiologicalprocesses such as wound healing, development, remodelling of the female reproductive organs,bone formation, angiogenesis and neuronal outgrowth and in pathological processes such ascancer and chronic inflammatory diseases.Matrix metalloproteinases (MMPs) are a family of Zn2+-dependent endopeptidases capable ofcleaving most extracellular matrix (ECM) and basement membrane (BM) molecules. MMPsmediate important functions during normal tissue remodelling and development but aberrant MMPactivity has been reported in several tissue destructive pathological conditions such as cancer andchronic inflammatory diseases. MMP inhibitors (MMPIs) have been developed for treatment ofseveral severe human diseases, however, MMP inhibition is often associated with unwanted side-effects due to interference with normal tissue remodelling processes. In addition, the parametersmeasuring MMPI efficacy in the present models of clinical trials in cancer are ill-fitting. As anexception, adult periodontal disease has been successfully treated with a pharmaceutical inhibitingMMPs in the USA.Laminin-5 (Ln-5) is a heterotrimeric glycoprotein part of the hemidesmosome (HD) structure withinthe BM and linking surface epithelia to the underlying connective tissue. Epithelial cells deposit Ln-5 while migrating and MMP-2 and -14 cleavage of the Ln-5 γ2-chain induce cell migration.In women past menopause, estrogen deprivation leads to impaired wound healing due to delayedre-epithelialization, reduced skin collagen synthesis and a prolonged inflammatory response. Whileit is known that elevated MMP expression and activity are associated with estrogen-depriveddelayed wound healing, little is known about the effects of estrogen-deprivation on Ln-5 γ2-chainexpression and processing.Periodontal disease exhibits all typical features of chronic inflammation and resembles woundhealing events. When acute inflammatory gingival inflammation is left untreated, it may progressinto chronic adult periodontitis (AP), leading to soft and hard tissue destruction and eventually toothloss. Periodontal disease and delayed wound healing are typically associated with elevated MMPactivity, reduced expression of tissue inhibitor of metalloproteinases (TIMPs) and excessdegradation of ECM and BM molecules.The present work was conducted to investigate the expression, activity and inhibition of MMPs andthe Ln-5 γ2-chain processing in situations associated with chronic inflammation such as cutaneouswound healing in estrogen-deprived rats, human periodontal disease and cell migration.

17

2. Review of the literature

2.1. The collagenous extracellular matrix

Tissues within an organism are composed of a cell mixture and the surrounding ECM consisting ofinsoluble protein fibrils, mainly collagens and elastins, as well as soluble polymers such asglycosaminoglycans, proteoglycans and cell adhesive glycoproteins. Cells bind to the surroundingECM mainly by transmembrane molecules. ECM supports cell motility within the connectivetissues, regulates cell proliferation, shape and function as well as allows nutrients and chemicalmessengers to freely diffuse (Alberts et al. 1994). Insoluble collagen fibrils make up about 30% ofthe human body’s proteins and approximately half of the human body’s collagens are within thebones while most of the remaining collagen content resides within the skin (Myllyharju andKivirikko 2001). Up to now over 30 different collagen α−chains differing in the primary sequencehave been characterized. The sequence of the various α−chains contains a variable number of theclassical Gly-X-Y repetitive motifs, which form the so-called collagenous domains andnoncollagenous domains of variable length and location. A mature collagen molecule is constitutedby a 300 nm helical rod terminated by very short noncollagenous sequences, the telopeptides.During fibrillogenesis collagen molecules self-assemble into fibrils and subsequently aggregate toform a collagen fiber (Aumailley and Gayraud 1998).The turnover of most adult tissue macromolecules is considered to be very slow. Highly regulateddegradation of ECM molecules serves several distinct functions, namely assembly of the ECM,removal of excess components and remodelling of ECM structure. These three processes are thekey to ECM synthesis and assembly, the physiological remodelling during growth, differentiation,morphogenesis and wound healing (Basbaum and Werb 1996).

2.2. Matrix metalloproteinases

Diffusible proteinases capable of degrading fibrillar collagen were first identified in 1962 frominvoluting tadpole tail (Gross and Lapier 1962). From there on, several families and subclasses ofthese proteinases have been characterized. MMPs are a family of highly conservedendopeptidases dependent on Zn2+-ions for activity. Evolution of MMPs may, however, be moreancient than currently realized. The Bacteroides fragilis-bacteria contain an amino acid sequencewith some identity to an MMP (Massova et al. 1998). MMPs can collectively cleave most ECM andBM macromolecules (Table 1). At present, 25 vertebrate MMPs and 22 human homologues havebeen identified and characterized. MMP numbering is usually determined by the order of thediscovery, MMP-1 being the first. However, MMP-4, -5 and –6 have been eliminated as aconsequence of duplication. MMP nomenclature often includes a characteristic name, for examplegelatinase A/MMP-2. MMPs participate in many physiological processes, such as embryonicdevelopment, organ morphogenesis, blastocyst implantation, ovulation, nerve growth, cervicaldilatation, post-partum uterine involution, mammary development, endometrial cycling, hair folliclecycling, angiogenesis, inflammatory cell function, apoptosis, tooth eruption, bone remodelling andwound healing. To ensure physiologically beneficial MMP activity, regulation takes place at thelevel of transcription, translation, trafficking of membrane-bound forms (secretion and endocytosis),extracellular binding proteins, shedding, oligomerization, internalization and autolysis, activationand in addition, MMP activity is regulated by their physiological inhibitors, i.e.TIMPs. Regulatoryaction or induction is mediated by growth factors, hormones, cytokines, cell-matrix, cell-cellinteractions and cellular transformation as well as by extrinsic chemical/physical factors. AberrantMMP regulation has been associated with numerous tissue destructive diseases, including cancer,arthritides, cardiovascular disease, nephritis, neurological disease, breakdown of the blood brainbarrier, infectious diseases, periodontal disease, lung diseases, gastric ulceration, eye diseases,corneal ulceration, skin diseases, liver fibrosis, atherogenesis, emphysema and chronic woundhealing (Nagase and Woessner 1999).

18

TA

BL

E 1

. Mat

rix

met

allo

pro

tein

ases

N

ame

(sp

ecifi

c n

ame)

Siz

e la

ten

t/act

ive

(kD

a)S

ub

stra

tes

MM

P-1

(Col

lage

nase

-1)

52/4

1T

ypes

I,

II, I

II, V

II, X

, X

I co

llage

ns,

gela

tin,

enta

ctin

, ag

grec

an,

tena

scin

, pe

rleca

n, v

itron

ectin

, IG

FB

P-2,

3,α1

-α1

PI,

α2M

, pro

TNF-

α.M

MP-

2 (G

elat

inas

e A

)72

/62

Typ

es I

, III

, IV

, V

, V

II, X

, X

I co

llage

ns,

gela

tin,

elas

tin,

fibro

nect

in,

lam

inin

, ag

grec

an,

vitro

nect

in,

tena

scin

dec

opr

oIL

-1β,

MC

P, I

GFB

P-3/

5, p

roT

NF-

α, F

GF-

RI,

α1-a

ntic

hym

otry

psin

,α1P

I.M

MP-

3 (S

trom

elys

in-1

)54

/43,

28

Typ

es I

II, I

V,

V,

IX,

X,

XI

colla

gens

, el

astin

, pr

oteo

glyc

ans,

lam

inin

, fib

rone

ctin

, ge

latin

, fib

rin/fi

brin

ogen

, en

aggr

ecan

, vitr

onec

tin, p

erle

can,

dec

orin

, pro

-HB

-EG

F, p

roIL

-1β,

pla

smin

ogen

, E-

cadh

erin

,IG

FB

P-3,

α1-a

ntic

hyα2

M, p

roT

NF-

α.M

MP-

7 (M

atril

ysin

)28

/19

Ela

stin

, pr

oteo

glyc

ans,

lam

inin

, fib

rone

ctin

, ge

latin

, ty

pes

I, III

, IV

, V

, IX

, X

, X

I co

llage

ns,

fibrin

/fibr

inog

en,

agte

nasc

in, v

itron

ectin

, pro

α-d

efen

sin,

dec

orin

, E-c

adhe

rin, p

lasm

inog

en, p

roT

NF

-α,α

1PI.

MM

P-8

(Col

lage

nase

-2)

75/5

8; 5

4/42

Typ

es I,

II, I

II, V

II, X

col

lage

n, g

elat

in, a

ggre

can,

ent

actin

, ten

asci

n, ti

ssue

fact

or p

athw

ay in

hibi

tor,

α1P

I,α2

M.

MM

P-9

(Gel

atin

ase

B)

92/8

2Ty

pes

I, IV

, V, V

II, X

, XI,

XIV

, X

VII,

gel

atin

, el

astin

, fib

rone

ctin

, la

min

in,

aggr

ecan

, vi

trone

ctin

, de

corin

, p

roT

GF

2RA

, pla

smin

ogen

, α1P

I, pr

oTN

F-α,

α1-a

ntic

hym

otry

psin

,α2M

,α1P

I.M

MP-

10 (

Str

omel

ysin

-2)

54/4

3, 2

4T

ypes

III,

IV

, V

, IX

, X

, X

I, pr

oteo

glyc

ans,

lam

inin

, fib

rone

ctin

, ge

latin

, ag

grec

an,

elas

tin,

fibrin

/fibr

inog

en,

envi

tron

ectin

.M

MP-

11 (

Str

omel

ysin

-3)

55/4

4α1

PI,

IGF

BP-

1,α2

MM

MP-

12 (

Met

allo

elas

tase

)54

/43,

22

Typ

es I

, IV

col

lage

n, a

ggre

can,

dec

orin

, ge

latin

, el

astin

, fib

rone

ctin

, fib

rin/fi

brin

ogen

, lam

inin

, pro

teog

lyca

n, v

itrpl

asm

inog

en,α

2M,α

1PI.

MM

P-13

(C

olla

gena

se-3

)60

/48

Typ

e I,

II, II

I, IV

, VI,

VII,

X, I

X, X

IV c

olla

gen,

gel

atin

, fib

rone

ctin

, ent

actin

, agg

reca

n, te

nasc

in, α

1-an

tichy

mot

ryps

MM

P-14

(Mem

bran

e ty

pe 1

m

etal

lopr

otei

nase

)66

/60

Nat

ive

type

s I,

II, I

II co

llage

ns,

gela

tin,

fibro

nect

in,

tena

scin

, pe

rleca

n, n

idog

en,

vitro

nect

in,

fact

or X

II, f

ibrin

, ag

fibrin

ogen

, pro

TN

F-α

, lam

inin

, car

tilag

e pr

oteo

glyc

an c

ore

prot

ein,

CD

44, α

2M,α

1PI.

MM

P-15

(M

embr

ane

type

2

met

allo

prot

eina

se)

76/6

1La

min

in, f

ibro

nect

in, t

enas

cin,

nid

ogen

, ent

actin

, gel

atin

, agg

reca

n, v

itron

ectin

, pro

TN

F-α

, tra

nsgl

utam

inas

e.

MM

P-16

(M

embr

ane

type

3

met

allo

prot

eina

se)

70/5

6G

elat

in, t

ype

III c

olla

gen,

per

leca

n, fi

bron

ectin

, vitr

onec

tin, a

ggre

can,

tran

sglu

tam

inas

e.

MM

P-17

(M

embr

ane

type

4

met

allo

prot

eina

se)

71/6

7G

elat

in, f

ibrin

/fibr

inog

en, α

2M, p

roTN

F-α.

MM

P-18

xCol

4T

ype

I col

lage

n.M

MP-

19 (R

AS

I)57

Typ

e I a

nd IV

col

lage

ns, f

ibro

nect

in, g

elat

in, t

enas

cin,

cas

ein,

lam

inin

, ent

actin

, agg

reca

n, C

OM

P.

MM

P-20

(E

nam

elys

in)

54/4

3A

mel

ogen

in, c

asei

n, g

elat

in, f

ibro

nect

in, t

ype

IV, X

VIII

col

lage

ns, l

amin

in, t

enas

cin

C, a

ggre

can,

CO

MP

.M

MP-

2162

/49

(hum

an)

MM

P-22

42/2

8G

elat

in, c

asei

nM

MP-

23 (C

A-M

MP)

66G

elat

inM

MP-

24 (

Mem

bran

e ty

pe 5

m

etal

lopr

otei

nase

)73

/64

Pro

teog

lyca

n, ty

pe I

colla

gen,

fibr

onec

tin, l

amin

in.

MM

P-25

(M

embr

ane

type

6

met

allo

prot

eina

se)

63/5

8G

elat

in, t

ype

IV c

olla

gen,

fibr

onec

tin.

MM

P-26

(M

atril

ysin

2, E

ndom

etas

e)29

/19

Fib

rone

ctin

, fib

rinog

en, g

elat

in, t

ype

IV c

olla

gen,

α1P

I, la

min

in-1

.M

MP-

2775

Typ

e II

colla

gen,

gel

atin

, fib

rone

ctin

MM

P-28

(E

pily

sin)

59/4

5C

asei

nM

odifi

ed fr

om a

nd re

fere

nces

from

:(A

imes

and

Qui

gley

199

5, B

artle

tt et

al.

1998

, d'O

rtho

et a

l. 19

97, K

ähär

i and

Saa

rialh

o-K

ere

1999

, Knä

uper

et a

l. 19

96, L

iu e

t al.

2000

, Lla

no e

al. 2

001,

Lyn

ch a

nd M

atris

ian

2002

, McC

awle

y an

d M

atris

ian

2000

, Nab

eshi

ma

et a

l. 20

02,P

ark

et a

l. 20

00, U

ria a

nd L

opez

-Otin

200

0, V

äänä

nen

et a

l. 20

01, V

an d

en S

teen

et a

l.

19

2.2.1. General characteristics of MMPs

MMPs consist of a single polypeptide, varying between 20-100 kiloDalton (kDa) in size (Figure 1).Upon translation, the full-length MMP is in a prepro-form and the amino (N)-terminal predomain ofMMPs contains a hydrophobic leader sequence which targets most of these enzymes to thesecretory pathway. Concomitant with secretion, the signal sequence is cleaved off, resulting in, atleast usually, an inactive proMMP. The prodomain has a conserved PRCGVPD motif involved inmaintaining the latency of the MMPs. The cysteine within this sequence (the so called “cysteineswitch”) ligates the catalytic Zn2+ to maintain proMMPs in an inactive state. Upon activation, theprodomain is cleaved off or the enzyme undergoes conformational changes to become an activeproteinase. The catalytic domain contains a HEXGHXXXXXHS motif with three His-residues andGlu-residues considered to be the critical catalytic Zn2+-binding sites and a conserved methionine,which forms a unique “Met-turn” structure (Nagase and Woessner 1999). The catalytic domain maycontain two Zn2+ -ions of which one has been suggested to be structural (Willenbrock et al. 1995).The catalytic domain dictates cleavage-site specificity through its active site cleft and is linked tothe Carboxy (C)-terminal hemopexin domain by a proline-rich linker peptide. The hemopexin-domain has an ellipsoidal disk shape consisting of four antiparallell β-strands and a α-helicalcentral cavity occupied by a Ca2+-ion. Membrane type (MT)-MMPs, which are destined to the cellsurface, contain an additional single-pass transmembrane domain located at the C-terminus(Nagase and Woessner 1999, Sternlicht and Werb 2001).

2.2.2. Collagenases

The three mammalian collagenases are MMP-1 (collagenase-1), MMP-8 (collagenase-2) andMMP-13 (collagenase-3). The collagenases cleave the α-chains of fibrillar collagens I, II and III atthe (P1)Gly775-(P1´)Ile/Leu776 site resulting in the generation of N-terminal ¾ (αA) and C-terminal ¼fragments (αB), which then rapidly denature at body temperature and are further degraded bygelatinolytic MMPs (Kähari and Saarialho-Kere 1999).MMP-8 cleaves triple helical type I collagen more efficiently than type III collagen (Hasty et al.1987), MMP-1 prefers type III collagen over type I collagen (Hasty et al. 1987, Mallya et al. 1990),and MMP-13 type II collagen (Knauper et al. 1996c, Krane et al. 1996). In comparison, MMP-13 isa much stronger gelatinase than MMP-1 or –8 (Knäuper et al. 1996a). Collagenases also degradevarious other extracellular molecules (Table 1). In all collagenases, a Tyr-Asp-Gly triplet isproposed as essential and specific for collagenase activity and the specific action of collagenaseson triple helical collagen is determined by the presence of a 16-amino acid sequence in their C-terminal domain (Hirose et al. 1993). MMP-8 is distinct from the two other collagenases in that itcan be stored in the secondary granules of polymorphonuclear (PMN) neutrophils while MMP-1and –13 requires transcriptional activity and de novo protein synthesis (Hasty et al. 1986, Kähäriand Saarialho-Kere 1999).MMP-1 is expressed by fibroblasts, endothelial cells, macrophages, hepatocytes, chondrocytes,osteoblasts, tumor cells and migrating epidermal keratinocytes (Pilcher et al. 1998) and itsexpression is induced in various inflammatory diseases and cancers (Johansson et al. 1997,2000). Until recently, it was generally accepted that rodents lack a homologue for human MMP-1.However, two MMP-1 homologues have been cloned and characterized in the mouse and areexpressed by extra-embryonic tissue trophoblast giant cells (Balbin et al. 2001).MMP-8 was first cloned from messenger RNA (mRNA) extracted from the peripheral leukocytes ofa patient with chronic granulocytic leukemia (Devarajan et al. 1991, Hasty et al. 1990) and issynthesized during the myelocyte stage of neutrophil development (Hasty et al. 1986). PMNneutrophil-stored MMP-8 is released to the ECM upon chemotactic stimulation in vitro or duringinflammatory conditions in vivo (Tschesche et al. 1991). MMP-8 has been found in two differentforms, of which the neutrophil-derived 75 kDa MMP-8 contains complex N-linked carbohydratesand the ~50 kDa less glycosylated form of MMP-8 is usually detected within all other types of cells(Hasty et al. 1986, Mallya et al. 1990). In cell culture, MMP-8 expression has been detected inmucosal fibroblasts, squamous cell carcinoma (SCC) cells of the tongue (Moilanen et al. 2002),

20

Figure 1. Structure of MMPs. MMPs typically consist of a predomain, a prodomain including thecysteine switch (C), a hinge region, the catalytic domain including the Zn2+ binding active site and ahemopexin domain. The membrane type MMPs also contain a transmembrane domain and MMP-2and –9 fibronectin domains. Modified from (Vu and Werb 2000).

21

chondrocytes (Cole et al. 1996), odontoblasts (Palosaari et al. 2000), melanoma cells(Giambernardi et al. 1998), leukemia cells (Kim et al. 2001) and human endothelial cells(Hanemaaijer et al. 1997). MMP-8 has been found to be expressed in vivo by bronchial epithelialcells and macrophages involved in bronchiectasis (Prikk et al. 2001), oral SCCs (Moilanen et al.2002), chondrocytes in rheumatoid arthritic and osteoarthritic lesions (Chubinskaya et al. 1999),rheumatoid synovial fibroblasts (Hanemaaijer et al. 1997), in human gingival sulcular epithelialcells (Tervahartiala et al. 2000), in cells of human atheroma (Herman et al. 2001) and by plasmacells associated with oral keratocysts (Wahlgren et al. 2001).MMP-13 was first cloned and identified from human breast carcinoma (Freije et al. 1994) andsubsequently cloned from interleukin-1 (IL-1) stimulated chondrocytes (Mitchell et al. 1996). MMP-13 is expressed in culture by skin fibroblasts and keratinocytes (Johansson et al. 2000), leukemiacells (Kim et al. 2001) and in several other cultured carcinoma cells (Giambernardi et al. 1998) aswell as by plasma cells in vitro and in vivo (Wahlgren et al. 2001). MMP-13 expression in tissues isgenerally detected at sites of active remodelling, i.e. fetal bone development and postnatal boneremodelling (Johansson et al. 1997) and in severe inflammations such as osteoarthritic cartilage(Mitchell et al. 1996), rheumatoid synovium (Lindy et al. 1997), macrophages of chronic cutaneousulcers (Vaalamo et al. 1997), intestinal ulcerations, cancers such as malignant tumours, SCCs,cutaneous basal cell carcinoma and chondrosarcomas (Kähari and Saarialho-Kere 1999).

2.2.3. Gelatinases

The two main gelatinases (also called type IV collagenases) MMP-2 (gelatinase A) and MMP–9(gelatinase B) are characterized structurally by three repeats of fibronectin-type II domains insertedin the hemopexin domain interacting with collagens and gelatins (Roeb et al. 2002). MMP-2 and -9are highly efficient in cleaving gelatin along with type IV collagen and several other substrates(Table 1). MMP-2 can also cleave native type I and II collagen to the characteristic αA and αB

identical to those generated by collagenases (Aimes and Quigley 1995, Konttinen et al. 1991).MMP-2 was the first identified type IV collagenase and first purified from a malignant murine PMTsarcoma cell line (Salo et al. 1983) and later cloned by (Huhtala et al. 1990). MMP-2 is typicallyexpressed constitutively by cells of mesenchymal origin (Van den Steen et al. 2002) and itsexpression is associated with many different cell types and cancers, such as melanoma andfibrosarcoma (Huhtala et al. 1991) and numerous cultured carcinoma cells (Giambernardi et al.1998).The full-length MMP-9 gene was cloned from the HT1080 fibrosarcoma cell line by (Huhtala et al.1991). MMP-9 is synthesized during late stages of PMN neutrophil development, stored within thetertiary granules and released upon stimulus. MMP-9 may exist both as a monomer or homodimerand as a covalent complex with neutrophil gelatinase B-associated lipocalin in neutrophils(Kjeldsen et al. 1992, Van den Steen et al. 2002). In contrast to the constitutively expressed MMP-2, MMP-9 expression in other cell types than PMN neutrophils requires transcriptional activity.MMP-9 is expressed by skin and gingival epithelial cells in vitro and vivo (Salo et al. 1991, 1994),alveolar macrophages, plasma cells and lymphocytes (Di Girolamo et al. 1998, Van den Steen etal. 2002) as well as in several cultured carcinoma cell lines (Giambernardi et al. 1998).MMP-9 plays a physiological role in reproduction, growth and development and has beenimplicated to play a role in tumor invasion and inflammatory diseases such as arthritis, cornealulcers, Alzheimers disease, skin blistering diseases (Vu et al. 1998), periodontitis (Westerlund etal. 1996), premature ruptures of amniotic membranes, lung diseases, neuroinflammatory diseases,HIV, vascular diseases and various cancers (Van den Steen et al. 2002).

2.2.4. Stromelysins

This MMP subclass includes MMP-3 (stromelysin-1) and -10 (stromelysin-2), which can cleave theglobular domain, but not the helical type IV collagen in addition to a wide variety of other matrixmolecules (Table 1). Stromelysins are structurally characterized by an insertion of an XPPVPTXXVmotif in the C-terminal part of their catalytic domain (Hirose et al. 1993, Nagase and Woessner

22

1999). MMP-11 (stromelysin-3) and -12 (metalloelastase) are often included in this subgroupalthough they are structurally different from MMP-3 and -10 (Johansson et al. 2000, Kähari andSaarialho-Kere 1999).MMP-3 and –10 are expressed by fibroblastic cells and by normal and transformed epithelial cellsin culture and in vivo (Giambernardi et al. 1998, Johansson et al. 2000, Kähari and Saarialho-Kere1999). MMP-11 has not been found to degrade any ECM components; however, MMP-11 cleavesserine proteinase inhibitors and insulin-like growth factor binding protein-1 in vitro. MMP-11 isexpressed during wound healing, in fibroblasts, in breast cancer, uterus, placenta, cyclingendometrium and involuting mammary gland (Luo et al. 2002).MMP-12 degrades elastin very efficiently and is expressed by macrophages and stromal cells atsites of rapid matrix turnover during murine fetal development, in granulomatous diseases of theintestine and skin (Johansson et al. 2000, Kähari and Saarialho-Kere 1999) and in human alveolarmacrophages associated with pulmonary emphysema (Belaaouaj et al. 1995). Inspite ofmacrophage accumulation in MMP-12 deficent mice there was no enlargement of air space due toexposure to cigarette smoke indicating that MMP-12 plays a major role in the destruction of lungtissues (Hautamäki et al. 1997).

2.2.5. Matrilysins

MMP-7 (matrilysin-1) and MMP-26 (endometase/matrilysin-2) are similar in that they both lack thehinge- and hemopexin-domain. MMP-7 was first isolated from a mixed tumor library and has a highaffinity for elastin (Table 1). MMP-7 is primarily expressed by cells of epithelial origin andspecifically in a lumenal direction. MMP-7 can be stored in the secretory epithelial cells of exocrineglands of the skin, endometrium, gastrointestinal tract and airways as well as in early involutinguterus of rat. MMP-7 is also expressed by malignant epithelial cells in tumors of the gastrointestinaltract, prostate, and breast (Wilson and Matrisian 1996).MMP-26 was cloned from a human endometrial tumor cDNA library and is expressed in humanplacenta and uterus as well as in various tumor cells (Uria and Lopez-Otin 2000). MMP-26 has aunique cysteine switch sequence with a His-residue replacing the common Arg-residue. MMP-26cleaves gelatin and human plasma α1-proteinase inhibitor (α1-PI; Table 1) (Park et al. 2000).

2.2.6. MT-MMPs

The membrane-type MMPs form a distinct class of MMPs in that they have a membrane-spanningdomain near their C-terminus followed by a nine-amino-acid insertion between the pro and catalyticdomain which ends in a RXKR furin activation consensus that may be cleaved by furin or furin-likeenzymes leading to intracellular activation (Apte et al. 1997). MMP-14 (MT1-MMP), MMP-15 (MT2-MMP), MMP-16 (MT3-MMP) and MMP-24 (MT5-MMP) are type I transmembrane MMPs with atraditional 24 amino acid-residue transmembrane domain at their C-termini, while MMP-17 (MT4-MMP) and MMP-25 (MT6-MMP) are glycosylphosphatidylinositol-anchored MMPs. The type Itransmembrane MMPs can activate MMP-2 (Sternlicht and Werb 2001).MMP-23 can be considered a specialized member of the MT-MMPs. The human MMP-23 lacks arecognizable signal sequence at its N-terminus and the Cys-switch. MMP-23 is prominentlyexpressed in reproductive tissue as well as in heart, intestine, colon, placenta, lung and pancreas(Velasco et al. 1999). The mouse orthologue for MMP-23, the CA-MMP, is a type IItransmembrane and contains unique cysteine-rich, proline-rich and IL-1 type II receptor-likedomains (Pei et al. 2000). Collectively, MT-MMPs cleave a wide variety of ECM molecules in vitro(Table 1).MMP-14 was first identified on the surface of invasive tumor cells and was found to increaseinvasiveness of cultured carcinoma cells (Sato et al. 1994). MMP-14 has been identified in both ashed soluble and a membrane-associated form in cultured breast cancer cells and fibroblasts (Li etal. 1998). MMP-14 is expressed by skin fibroblasts (Madlener 1998, Okada et al. 1997), in tumorcells of epithelial origin, bronchial epithelial cells (Maisi et al. 2002) and endothelial cells (Silletti etal. 2001).

23

2.2.7. The X-files of the MMPs

The latest “rookies” to the MMP family have distinct structural and/or functional properties from theother MMP subcategories. These MMPs do not show specific characteristic to any other subclassand thus it can be considered legitimate to assign them to the X-files.MMP-18 (Table 1) was identified as a novel interstitial collagenase (xCol4) from the Xenopuslaevis tadpole. This collagenase deserves a special attribute, since collagenase activity was firstidentified in the Xenopus tadpole tail some 40 years ago, and MMP-18 found in 1996 was the firstidentifiable amphibian collagenase (Stolow et al. 1996).MMP-19 (Table 1) was first cloned from a human mammary gland cDNA library and is most similarto the stromelysin class of MMPs in terms of activity. Expression of MMP-19 has been found inplacenta, lung, pancreas, ovary, small intestine, spleen, thymus and prostate (Cossins et al. 1996;Pendas et al. 1997).Human MMP-20 (enamelysin) was first found in odontoblastic cells of the dental papilla and MMP-20 mRNA is typically expressed during tooth development by ameloblasts of the enamel organ(Bartlett et al. 1998), by human tongue SCC cells, in the tooth pulp and in placenta (Väänänen etal. 2001). MMP-20 cleaves amelogenin very efficiently (Table 1).The human MMP-21 (Table 1) is expressed in adult brain, lung, testis, ovary, colon and leukocytes,kidney along with several carcinoma cell lines (Ahokas et al. 2002).MMP-28 or epilysin was first cloned from a human testis and keratinocyte cDNA library and is mostrelated to MMP-19. Epilysin is highly expressed in testis and also in lungs, heart, colon, intestineand brain and in addition, in cultured keratinocytes and migrating wound keratinocytes. MMP-28 isunusual in being abundantly expressed in normal adult tissue and cleaves casein in vitro (Lohi etal. 2001).

2.2.8. Activation mechanisms of MMPs

The latency of MMPs is regulated through a cysteine-switch comprised of the unpaired Cys73

sulfhydryl group within the conserved P71RCGVPD-motif keeping MMPs inactive by binding to thecatalytic Zn2+. A key step in MMP activation is the dissociation of Cys from the Zn2+ allowing theinteraction of Zn2+ with the H2O required for catalysis and concomitant exposure of the active site.Thus, the Zn2+ is viewed as the “switch” that leads to activation. This is attained by proteolyticremoval of the prodomain, modification of the sulfhydryl-group or by molecular perturbation. Thedisengagement of the Cys-Zn2+ bond may also lead to autoactivation of the MMP (Nagase 1997,Van Wart and Birkedal-Hansen 1990).The activation process of MMPs encompasses three different mechanisms: stepwise activation,activation on the cell-surface and intracellular activation (Figure 2).Non-proteolytic activation without concomitant loss in molecular weight of most MMP zymogenscan be accomplished by SH-reactive agents (iodoacetate, p-aminophenylmercuric acetate(APMA), HOCl, oxidized glutathione, mercurial compounds, anti-rheumatic gold (I) compounds,denaturants (urea, SDS, NaSCN), oxidation and freeze/thaw (Birkedal-Hansen et al. 1993, Murphyet al. 1999, Nagase 1997, Sorsa et al. 1987, Stetler-Stevenson et al. 1989, Van Wart and Birkedal-Hansen 1990). The nonproteolytic activation proceeds in a stepwise manner in which severalintermediates are initially generated by an intramolecular reaction. The final cleavage to an activeMMP follows a bimolecular reaction (Nagase 1997).The proteolytic activation of MMPs also proceeds in a stepwise manner. Activator proteinases firstattack the proteinase susceptible “bait” region located in the middle of the prodomain. Thiscleavage induces conformational changes in the prodomain and renders the final activation sitereadily cleavable by a second proteolysis. The latter reaction is usually catalyzed by an MMP butnot by the activator proteinase (Nagase 1997). Many proteinases activate MMPs and many MMPsactivate each other. The activation cascade of MMPs is fairly complex (Figure 3). However, most

24

Figure 2. Activation of MMPs. proMMP activation can proceed through different intermediate formsby chemical and proteinase activation, both eventually rendering the active MMP enzyme. Adaptedfrom (Nabeshima et al. 2002, Nagase 1997).

data are gained from in vitro experiments and in vivo activation mechanisms for MMPs are not wellcharacterized. Most likely many MMPs are activated in vivo by tissue and plasma proteinases or bybacterial proteinases (Moilanen et al. 2003, Nagase, 1997, Sorsa et al. 1992, 1997). MMP-1 (Sanget al. 1995), -2 (Fridman et al. 1995), -3 (Cowell et al. 1998, Ogata et al. 1992), -10 (Nakamura etal. 1998), -13 (Knäuper et al. 1997) and -26 (Uria and Lopez-Otin 2000) can activate proMMP-9 invitro through a two-step activation mechanism (Van den Steen et al. 2002). TAT-2 can activateproMMP-1, -3, -8, -9 and -20 and to a lesser extent proMMP-2 (Moilanen et al. 2003, Sorsa et al.1997, Väänänen et al. 2001). MMP-3 (Wilson and Matrisian 1996) and –10 (Nakamura et al. 1998)can activate proMMP-7, which in turn can activate proMMP-1 (Sang et al. 1996), -2 and -8 (Balbinet al. 1998) and also partially proMMP-9 (Wilson and Matrisian 1996). MMP-10 activates proMMP-8 in vitro (Knäuper et al. 1996b). ProMMP-13 is activated by MMP-2, -3, -10, -14, -15, TAT-2 andplasmin (Johansson et al. 2000, Knäuper et al. 1996b, Moilanen et al. 2003). At least 9 MMPscontain RXK/RR motifs and can be activated by furin or its related proprotein convertases (Kang etal. 2002, Pei and Weiss 1995, Santavicca et al. 1996, Yana and Weiss 2000). Thus, these MMPsmay be activated intracellularly.The fully active MMPs have Phe or Tyr at their N-termini, generated by MMP activation. Activationof MMPs leading to forms with other N-termini usually results in reduced enzymatic activity of theMMP (Nagase 1997). For MMP-9, it has been demonstrated that mere substrate binding toproMMP-9 induces its enzymatic activity without removal of the prodomain. MMP-9 was shown tobe enzymatically active despite the fact that only the proforms of the enzyme could be detected inplacental tissue extracts (Bannikov et al. 2002). A similar phenomenon for MMP-9 has beendetected in human inflammatory gingival crevicular fluid (GCF) (Westerlund et al. 1996).

25

Figure 3. The activation network of MMPs. The picture illustrates the highly complex activation ofMMPs by other MMPs and serine proteinases. MMP activation data is mostly based on in vitrostudies, little is known about MMP activation in vivo. Adapted from (Balbin et al. 1998, Cowell et al.1998, Knäuper et al. 1996b, 1997, Murphy et al. 1999, Sorsa et al. 1997, Van den Steen et al. 2002).

2.2.8.1. Cell-surface activation by MMP-14

MMP-2, unlike the other MMPs, has a propeptide generally not susceptible to proteolytic initiationof activation by proteinases (Nagase 1997, Sorsa et al. 1997). However, MMP-2 can be activatedby the cell surface-anchored MMP-14, -15, -16 and -24 (Kang et al. 2000, Llano et al. 1999, Satoet al. 1996, 1994, Strongin et al. 1995) and also by the soluble, transmembrane-truncated form ofMMP-14 (Will et al. 1996). MMP-14 is a pericellular activator of MMP-2, -8, -9 and -13 (Knäuper etal. 1996c).MMP-14 activation of MMP-2 occurs through a trimolecular-complex of MMP-2/TIMP-2/MMP-14which concentrates the components to the cell surface (Atkinson et al. 1995, Cowell et al. 1998,Sato et al. 1996, Strongin et al. 1995). MMP-14, through its catalytic domain, acts as a cellmembrane receptor for TIMP-2. Binding of the TIMP-2 N-terminal domain to the MMP-14 catalyticdomain mediates the activation of MMP-2 through the TIMP-2 C-terminal domain. Thus, a secondMMP-14 molecule is needed for activation of MMP-2 (Zucker et al. 1998). The initial cleavage ofMMP-2 destabilizes the structure of the MMP-2 prodomain and autoproteolysis at a second sitethen releases the rest of the prodomain generating fully active MMP-2 (Cowell et al. 1998, Stronginet al. 1995). In the activation process, the C-terminal hemopexin domain of proMMP-2 binds to theC-terminal end of TIMP-2 (Murphy et al. 1992).MT-MMPs (and MMP-11) themselves are not activated by typical chemical activators which induceautocatalytic activation of other secreted MMPs, i.e. APMA or SDS (Will et al. 1996). Both thegolgi-associated endopeptidase furin as well as plasmin have been suggested to be physiologicalactivators of MMP-14 and other MT-MMPs (Okumura et al. 1997, Yana and Weiss 2000) althoughfurin-mediated activation of MMP-14 is not a prerequisite for MMP-2 activation (Cao et al. 1996).Some results demonstrate that the MMP-14 prodomain would be required for the binding of TIMP-

26

2 to the catalytic domain and for the catalytic activity of MMP-14 (Pavlaki et al. 2002). In fact,membrane association alone may expose MMP-14 to furin-independent processing (Yana andWeiss 2000).

2.2.9. Transcriptional regulation of MMPs

Many functions of MMPs overlap and therefore the temporal, spatial and tissue-specific expressionof MMPs is necessary. The basal expression of several MMPs (such as MMP-1,-3,-7,-8,-9,-10,-12and -13) in cultured cells is low, and their transcription is induced by a variety of extracellularstimuli. These include growth factors, cytokines, chemical agents, physical stress and oncogeniccellular transformation acting through intracellular cascades involving secondary messengers.MMP expression can subsequently be down-regulated by suppressive factors. MMP expression isalso regulated by bacterial endotoxin and other virulence factors, phorbol esters, ECM proteins,cell-membrane associated proteins, cell stress and changes in cell shape (Kähäri and Saarialho-Kere 1999).In the genome, chromosome 11 contains a cluster of MMP genes, some of the latest MMP familymembers map to chromosome 1 while MT-MMP genes are spread on several chromosomes. The5’-flanking regulatory regions of most inducible MMP genes exhibit common features, including aTATA box and an activator protein 1 (AP-1) regulatory elements in the proximal promoter region.Extracellular stimuli activate the nuclear AP-1 transcription factor complex composed of membersof the Jun and Fos family, which bind to the AP-1 cis-element and activate transcription of thecorresponding MMP gene. The expression of the AP-1 dimer, c-Jun and c-Fos, are induced as aresult of the activation of mitogen-activated protein kinases (Kähäri and Saarialho-Kere 1999).MMP-2 is often constitutively expressed and controlled through activation and to some degree bypost-transcriptional mRNA stabilization. The MMP-2 gene is relatively unresponsive to stimulationin cultured cells and lacks both the AP-1 element and the classical TATA box. In addition, itcontains several potential SP-1 binding sites not found in collagenase gene promoters, an activatorprotein-2 binding sequence and an transcriptionally important tumor suppressor p53 binding site aswell as heterogenous initiation sites for transcription. Other nuclear factors controlling MMPtranscription are the ETS family of oncoproteins binding PEA3 sites, nuclear factor of κB, signaltransducers and activators of transcription, p53 and negative regulatory elements such astransforming growth factor-β (TGF-β) inhibitory element (Johansson et al. 2000, Kähari andSaarialho-Kere 1999, Mauviel 1993, Van den Steen et al. 2002). Common bi-allelic single-nucleotide polymorphisms that influence the rate of transcription have also been identified withinseveral MMP gene promoters and are associated with increased susceptibility to various types ofcancers (Van den Steen et al. 2002, Ye 2000).Tumor necrosis factor-α (TNF-α) is an inflammatory response-mediating cytokine which issynthesized as a transmembrane molecule that can bind to its receptor by cell-cell contact. TNF-αis proteolytically processed to a soluble homotrimer (Gearing et al. 1995) typically upregulatingmany MMPs in different cell types. TNF-α and IL-1β stimulate MMP-9, but not MMP-2 production inhuman gingival mucosal keratinocytes (Mäkela et al. 1998, Salo et al. 1994). The bacteriallipopolysaccharide (LPS) is a very potent inducer of both MMP-2 and –9 in vitro which may acteither directly or indirectly via cytokine stimulation on gene expression (Van den Steen et al. 2002).TGF-β is part of a super-family of growth factors and there are three structurally very similarmammalian isoforms. TGF-β requires activation to a mature form for receptor binding andintracellular signalling. TGF-β1 typically reduces the expression of collagenase genes and activityin cultured cells while simultaneously elevating the expression of TIMPs (Verrecchia and Mauviel2002). In contrast, TGF-β1 upregulates both MMP-2 and MMP-9 activities in cultured fibroblastsand transiently also MMP-2 mRNA expression (Salo et al. 1991, Van den Steen et al. 2002,Verrecchia and Mauviel 2002). TGF-β induces MMP-2 and –9 production in human gingivalmucosal keratinocytes (Salo et al. 1991) and increases MMP-2 and –9 expression in peripheralblood monocytes and in various tumorigenic and non-tumorigenic cell lines (Van den Steen et al.2002).

27

MMPs can themselves modulate growth factor and cytokine activity. MMP-3 and –13 cleavage ofthe heparan sulphate proteoglycan perlecan results in the release of basic fibroblast growth factor(Whitelock et al. 1996) and the cleavage of decorin by MMP-2, -3 and -7 releases TGF-β bound todecorin (Vu and Werb 2000). MMP-2 and –9 directly process TGF-β into an active ligand (Van denSteen et al. 2002, Yu and Stamenkovic 2000). At least MMP-1, -3 and -7 efficiently and MMP-2and –9 less efficiently, can process proTNF-α into the biologically active form (Gearing et al. 1995).Proteolysis may also downregulate components of mitogenic signaling pathways. Thus, MMPs canregulate the bioavailability and/or activity of growth factors and cytokines by cleavage of bothmatrix and non-matrix substrates or by mediating receptor turnover (Table 1).Post-transcriptional mechanisms also modulate MMP expression. Examples are stabilization ofMMP-1 and –13 mRNA transcripts by phorbol esters and epidermal growth factor (EGF) as well asstabilization of MMP-2 and –9 mRNA transcripts by TGF-β (Van den Steen et al. 2002). Theturnover of MMP-1 mRNA is apparently regulated by AU-rich sequences in the 3´-untranslatedregion and similar sequences may also regulate the stability of other MMP transcripts (Sternlichtand Werb 2001). MMP expression could also be regulated by alternative splicing. Multipletranscripts have been reported for MMP-8, -11, -13, -16, -17, -20, -25 and -26 (Hu et al. 1999, Luoet al. 2002).

2.2.10. In vivo functions of MMPs

None of the individual MMP-null mice generated to date have had an embryonic lethal phenotype.The most severe phenotype is associated with MMP-14 knock-out mice, which developed a severeskeletal phenotype including dwarfism, delayed membraneous ossification of calvarial bones,osteopenia, severe generalized arthritis, reduced activity of osteogenic cells and fibrosis of softtissues. MMP-14 deficiency also imparts a severe defect in collagenolytic activity (Holmbeck et al.1999). Furthermore, MMPs and in particular MMP-14 have been implicated in endothelial cellneovascularization processes through pericellular degradation of the fibrin-rich scaffold both in vitroand in vivo (Hiraoka et al. 1998). In a study with MMP-9 null mice, this gelatinase wasdemonstrated to play a role in vascularization of hypertrophic cartilage, reproduction, apoptosis ofhypertrophic chondorcytes and endochondral ossification during bone formation (Vu et al. 1998).Although MMPs can cleave virtually all ECM molecules in vitro, much less is known about theiractual in vivo substrates. Based on table 1, one would think that some of the ECM moleculespresented there as substrates for MMPs would actually have been proven to be substrates forMMPs in vivo. The literature comes up with very few, one being amelogenin which truly appears tobe the in vivo substrate for MMP-20. The other two evidently in vivo MMP substrates are not ECMmolecules. In addition, while MMP-9 indirectly mediates tissue destruction in vivo (which has beenproposed to be the main function of MMPs from the birth of MMPs) MMP-7 actually acts to protectthe host from microbial attack.The serpin α1-PI has been established as a substrate for MMP-9 in vivo. In studies with miceinduced with the subepidermal blistering disease bullous pemphigoid, it was shown that whileneutrophil elastase is primarily responsible for the destruction of the BM protein BP180 (type XVIIcollagen), MMP-9 is required for the inactivation of the α1-PI which act as an plasma inhibitor ofneutrophil elastase. Both neutrophil elastase and MMP-9 cleave BP180 in vitro, however, in vivoMMP-9 acts upstream of neutrophil elastase indirectly facilitating blister formation (Liu et al. 2000).The antimicrobial peptide α-defensin has been shown to be a substrate for MMP-7 in vivo.Intestinal Paneth cells secrete procryptidin (the mouse homologue for α-defensin) which isactivated into its mature form through cleavage by MMP-7 both in vivo and in vitro. In MMP-7 nullmice, only precursor forms of cryptidins were detected. MMP-7 null mice were also moresusceptible to exogenous bacterial infection in the intestine giving further evidence that MMP-7 isimportant in restricting bacterial colonization and access to the intestinal mucosa (Wilson et al.1999).

28

2.2.11. In vivo localization of MMPs

Many investigators are interested in detecting the net MMP activity in tissues. However, MMPantibodies used for immunohistochemistry poorly discriminate between the pro- and active forms ofMMPs. Excess TIMP activities also interfer with detection of immunoreactive active MMP species.Gel zymography by itself is a very sensitive method for detection of MMP activity, however,homogenization of tissue for this assay excludes localization of the cell type exhibiting activity. Inaddition, the extraction procedure and sample treatment can activate enzymes or permit contact ofenzymes and inhibitors localized in distinct compartments in the intact cells or tissues. In situzymography allows preservation of tissue architecture while detecting actual enzyme activity andimportantly, does not require species-specific reagents. This method was first introduced by Galiset al. (Galis et al. 1995).In situ zymography as a method is not quantitative and difficult to standardize. However, valuableinformation regarding localization of net MMP activity at the tissue level can be outlined by thismethod. The methodology is also rather simplistic and cheap. In principle, any purified MMPsubstrate can be used and conjugated to a fluorophore, such as fluorescein or resorufin (Galis etal. 1995). To eliminate false interpretation due to autofluorescence originating from ECM moleculesin the tissues, substrates can be conjugated to quenched fluorophores (Sarment et al. 1999).Autoradiographic gelatin emulsions are also used in addition to different kinds of films, referred toas film in situ zymography. When using gelatin as a substrate, mostly MMP-2 and –9 activity isdetected by both gel and in situ zymography. However, MMP-14 also causes focal degradation ofgelatin films (d'Ortho et al. 1998).Thus, the clear drawback of this procedure has been the rather non-specific detection of proteaseactivity since the substrates generally used are casein or gelatin. The use of broad-spectrumMMPIs and serine proteinase inhibitors only allow for conclusions of which enzyme family is themajor protease detected (Korostoff et al. 2000). Since MMPs are active at neutral pH, regulatingthe pH in the methodology allows for detection of either MMPs or serine proteinases. In situzymography has thus been shown to be specific for enzyme activity by incubating samples timelyat different temperatures and at different pHs (Berndt et al. 2000, Sarment et al. 1999).

2.3. MMP inhibition

2.3.1. Physiological inhibition of MMPs

MMPs can be self-regulated by their own proteolytic inactivation. Some cleavages inactivate MMPsor generate truncated enzyme species resulting in a concomitant change of action. MMPs are alsoinhibited and cleared by endogenous inhibitors like α2-macroglobulin (α2M), the major plasmainhibitor of MMPs. α2M functions by a mechanism involving the presentation of a cleavable “bait”region that, once proteolytically cleaved, causes a conformational change that entraps theproteinase, which becomes covalently anchored by transacylation. α2M/MMP-complexes areirreversibly cleared by scavenger receptor-mediated endocytosis (Hahn-Dantona et al. 2001).TIMPs are specific, physiological tissue inhibitors of MMPs. At present, four TIMPs (TIMP1-4) havebeen identified in vertebrates. Mammalian TIMPs are 20-30 kDa variably glycosylated two-domainmolecules. TIMPs play an important role in tissue remodelling due to their abilities to inhibit MMPsby forming 1:1 molar stoichiometric enzyme-inhibitor complexes (Brew et al. 2000). The criticalresidue involved in MMP inhibition is located around the disulphide bond between Cys1 and Cys70

resulting in bidental coordination of the catalytic Zn2+ (Nagase 1997).TIMP -1, -2 and -4 are secreted in a soluble form, whereas TIMP-3 is sequestred to the ECM.TIMP-1 and -2 inhibit the activity of most MMPs. As an exception, TIMP-1 is a poor inhibitor ofMMP-14, -15, -16 and -24 as well as MMP-19. TIMP-3 inhibits the activity of MMP -1, -2, -3, -9 and-13 being most potent in inhibiting MMP-9. TIMP-4 inhibits the activity of MMP-2 and -7 somewhatmore potently than that of MMP -1, -3 and -9. TIMP-2 and -3, unlike TIMP-1 are effective inhibitorsof the MT-MMPs (Baker et al. 2002, Brew et al. 2000).

29

TIMPs differ in the types of non-inhibitory complexes they form, mediated by their C-terminaldomains. Thus, TIMP-2 binds tightly to the C-terminal domain of proMMP-2, and TIMP-1 toproMMP-9 of which the former is important in the cell-surface activation of proMMP-2 (Baker et al.2002, Brew et al. 2000). The kinetics of TIMP-MMP interactions indicate that it is a non-competitive, two-step complex formation. In this, an inactive reversible complex (E•I) is formedrapidly which rearranges more slowly to an inactive tight complex (E•I*) (Baker et al. 2002, Brew etal. 2000). The fate of the MMP-TIMP complex is likely to be determined through internalization(Hahn-Dantona et al. 2001). Overexpression of TIMPs1-4 can inhibit invasion of malignant cells invitro and in vivo (Baker et al. 2002). However, TIMPs also exhibit MMP-independent biologicalactivity and paradoxically both tumour progressive and suppressive activity (Baker et al. 2002,Brew et al. 2000).

2.3.2. Synthetic MMPIs

Since MMPs have been implicated to play a detrimental function in virtually all pathologicalconditions, great effort has been put into developing inhibitors against the activity of theseenzymes. Clinically, synthetic MMPIs may inhibit tumour growth either as cytostats by maintainingsmall clusters of metastatic cells in a dormant state or by inhibiting tumour-induced angiogenesis(Coussens et al. 2002).Initially, the task was to design molecules that mimicked the substrate-cleavage site forcollagenases in the collagen molecule. The goal was to obtain enzyme inhibition through chelationof the active-site Zn2+-ion by incorporating a Zn2+-binding group into peptide analogues. In terms ofcollagenase inhibition a right-hand side hydroxyamate group was identified as one of the bestchoices for chelating the active site Zn2+. The hydroxyamate acts as a bidentate ligand with eachoxygen at an optimum distance from the active-site Zn2+-ion. The inhibitor binding is reversible.The broad-spectrum hydroxyamate MMPIs, Batimastat (BB-94) and Marimastat (BB-2516) werethe first MMPIs to enter clinical trials in the treatment of malignant tumours. The non-peptidomimetic MMPIs include at least tetracyclines and bisphosphonates (Hidalgo and Eckhardt2001).

2.3.2.1. Batimastat

Batimastat (Figure 4) is well tolerated, but its utility is limited by poor water solubility, whichrequires intraperitoneal or intrapleural administration. Batimastat inhibits at least MMP-1, -2, -3, -7,-8, -9, –13 and –14 with IC50s between 1-20 nM. Batimastat has been extensively studied inpreclincial animal models and found to effectively inhibit tumor growth, metastasis, tumor-associated angiogenesis and prolong survival time in several murine models including melanoma,hemangioma, xenograft models of human ovarian carcinoma, colorectal carcinoma, breast andpancreatic cancer (Coussens et al. 2002, Rasmussen and McCann 1997).

2.3.2.2. Marimastat

Marimastat (Figure 4) is water soluble and can be administered orally. Marimastat is a broad-spectrum MMPI with IC50s against MMP-1, -2, -3, -7, -8, –9, -12 and –14 of 5, 6, 200, 20, 2, 3, 3and 1.8 nM, respectively (Coussens et al. 2002, Rasmussen and McCann 1997, Whittaker et al.1999). Results with Marimastat-treatment in clinical trials are mixed. Marimastat has been tested inmore than 400 patients in Phase I/II studies in a number of different solid tumors in the USA.Marimastat treatment significantly reduced all the measured cancer specific antigen rates of rise ina dose-dependent fashion (Rasmussen and McCann 1997). Marimastat has also been tested incombinatorial treatments together with conventional cytotoxic drugs.

30

2.3.2.3. Tetracyclines

In 1944, the mycologist B.M Duggar discovered the broad antimicrobial effect of Aureomycin(chlortetracycline hydrochloride). Tetracyclines (TCs; Figure 4) were taken into pharmaceutical usein 1948 and are highly effective against aerobic and anaerobic bacteria, mycoplasma, chlamydia,spirochetes, some protozoa and rickettsiae. The chemically improved compounds doxycycline(1967) and minocycline (1972) are second-generation semisynthetic tetracycline derivatives withenhanced antibacterial activity, rapid intestinal absorption, longer half-life, superior tissue-fluidpenetration and reduced toxicity (Rifkin et al 1993).In the 1980s, Golub and collagues showed that TCs inhibit mammalian collagenase activity andproduction independently from their antimicrobial action (Golub et al. 1998). The TCs can affect theactivity and/or expression of MMP-1, -2, -3, -8, -9, -13 and –20 (Golub et al. 1998, Joshi and Miller1997, Lee et al. 2001, Rifkin et al. 1993, Suomalainen et al. 1992, Väänänen et al. 2001). Theability of the TCs and its chemically modified analogues to inhibit MMPs is based on multiplemechanisms: 1) inhibition of the active MMPs by chelation of the active site cations, 2) interferencewith the processing of the proenzyme into its active form, 3) reduction of MMP mRNA expressionand 4) rendering the MMPs more susceptible to degradation and fragmentation (Rifkin et al. 1993,Smith et al. 1999, Sorsa et al. 1998). TCs are known to inhibit MMPs originating from a variety ofcells: neutrophils, keratinocytes, macrophages, osteoblasts, chondrocytes, synovial fibroblasts, T-lymphocytes, HUVECs and from a wide range of tissues: skin, gingiva, cornea, cartilage, andrheumatoid synovium (Hanemaaijer et al. 1997, Rifkin et al. 1993, Smith et al. 1999). Doxycyclineand minocycline are more potent inhibitors than the parent compound tetracycline. Fordoxycycline, MMP-8 has IC50 values between 25-40 µM, MMP-1 over 200 µM while MMP-13 isextremely sensitive with IC50 values around 2 µM (Greenwald et al. 1998, Sadowski andSteinmeyer 2001, Suomalainen et al. 1992).Doxycycline inhibits MMP-8 activation and activity in vitro resulting in fragmentation of the enzymemost likely due to chelation of Ca2+, which shields MMPs from proteolytic attack. Inhibition bydoxycycline can be reversed by high concentrations of Ca2+ (Smith et al. 1996). Inhibition ofcollagenases by doxycycline occurs in a noncompetitive manner through alteration of MMPconformation rather than by interaction with the catalytic Zn2+ (Sorsa et al. 1994). Inhibition involvesbinding between the inhibitor and the enzyme and/or the inhibitor and the enzyme-substratecomplex (Smith et al. 1999). The differences in sensitivity of the three different collagenases to TCinhibition is likely due to differences in structure/function including high collagenolytic/gelatinolyticactivity, S1 pocket size and broad substrate specificity. TCs have been shown to be effective in thetreatment of human periodontal disease, sterile corneal ulceration, rat adjuvant arthritis, dog andguinea pig osteoarthritis, rodent rickets, human rheumatoid and reactive arthritis, diabetic andovariectomized (OVX) rat osteporosis, aortic aneurysms and in vitro osteoclast-mediated boneresorption and cartilage degradation system (Golub et al. 1998, Greenwald et al. 1998, Rifkin et al.1993). TCs can also block nitric oxide synthase activity, impact arachidonic acid metabolism andinhibit protein kinase C mediated transcriptional activity of several MMPs (Golub et al. 1998).The TCs have also been tested for their ability to inhibit tumor invasion. In cultured highly invasivebreast carcinoma cells, doxycycline inhibited the secretion and activity of MMP-2 and –9. Inaddition, doxycycline inhibited the proliferation of prostate and breast cancer cell lines in vitro,induced apoptosis and decreased the invasion and metastatic potential of highly aggressive breastcancer and melanoma cells. In phase I evaluation studies with cancer patients, side-effects suchas fatigue, confusion, nausea and vomiting were encountered (Hidalgo and Eckhardt 2001).

2.3.2.4. The chemically modified tetracyclines

To identify the site of anticollagenase function, Golub et al. synthesized different chemicallymodified analogs of tetracycline, the CMTs (Figure 4). The CMTs comprise a group of at least 10(CMTs 1-10) analogues plus some special modified CMTs that differ in their MMP specificity andpotency (Table 2). CMTs are modified so that the dimethylamino group from carbon-4 position (theside-chain required for antimicrobial activity in TCs) is removed. The removal of antimicrobial

31

Figure 4. Chemical formulas of the MMP inhibitors used in the study. Adapted from (Greenwald et al.1998, Shapiro et al. 1997, Whittaker et al. 1999).

32

activity retained and enhanced anticollagenase activity (Golub et al. 1998). In CMTs, theadvantages over TCs include lack of gastrointestinal toxicity, fatigue and photosensitivity due tolong-term systemic adminstration, greater plasma concentrations and longer elimination half-lifethus requiring less frequent drug administration. While all CMTs lack antimicrobial activity, onlyCMT-5 lacks any anti-collagenase effect. Apparently, TC/CMT inhibition of the catalytically activeMMP involves interaction by the Ca2+ and Zn2+ binding sites at the carbonyl and hydroxyl groups atC11 and C12, respectively on ring B at physiological pH (Figure 4). These sites are nitrogenated inCMT-5 and may thus account for the poor inhibitory effect of CMT-5 on MMPs (Greenwald et al.1998, Rifkin et al. 1993, Sorsa et al. 1994, Sorsa et al. 1998).The potency of the various TCs as MMP inhibitors appears to be CMT-8>CMT-3>CMT-7>Doxycycline>CMT-1. CMT-1 is more effective than doxycycline, but less effective than thefurther developed CMTs. CMT-3 is the only compound effective against MMP-1. In addition, CMT-3 and –8 are the most effective collagenase activity inhibitors and CMT-8 is the most effectiveinhibitor of MMP-13 activity (Greenwald et al. 1998). CMTs-1, -3, -7 and –8 also inhibits oxidativeactivation, collagenolytic and gelatinolytic activity of proMMP-8 and -9 in vitro (Sorsa et al. 1998).In preclinical studies, CMT-3 inhibited tumor growth, reduced lung and bone metastasis andinduced apoptosis. The compound is orally available and well tolerated with side-effects such asphotosensitivity and fatigue. CMT-3 reduces the plasma concentrations of MMP-2 and –9 as wellas the production of MMP-9 by peripheral blood mononuclear cells (Lokeshwar et al. 2002). CMT-3has also been shown to inhibit vasculogenic mimicry by aggressive melanoma cells in vitro. CMT-3decreased both MMP-2 and –9 active and proforms and MMP-1 expression in aggressivemelanoma cells (Seftor et al. 2002).

Table 2. Non-antimicrobial Chemically Modified Tetracyclines

Compound Description PropertiesCMT-1 4-dedimethylamino-tetracycline CMT discovery compoundCMT-2 tetracyclinonitrile Not absorbed by oral routeCMT-3 6-demethyl, 6-deoxy, 4-dedimethylamino-tetracycline Most lipophilic CMTCMT-4 7-Chloro, 4-dedimethylamino-tetracycline Orally bioavailableCMT-5 tetracyclinpyrazole No MMP inhibition but can

scavenge free radicalsCMT-6 5-hydroxy, 4-dedimethylamino-tetracycline Not absorbed by oral routeCMT-7 12α-deoxy, 4-dedimethylamino-tetracycline Orally bioavailableCMT-8 6α-deoxy, 5-hydroxy-4-dedimethylamino-tetracycline CMT derived from

doxycyclineCMT-9 12α,4α-anhydro, 4-dedimethylamino-tetracycline

CMT-10 7-dimethylamino, 4-dedimethylamino-tetracycline CMT derived fromminocycline

Modified from (Greenwald et al. 1998)

2.3.2.5. The CTTHWGFTLC-peptide

A synthetic decapeptide, the CTTHWGFTLC (CTT)-peptide was initially isolated from a phagedisplay library in which the HWGF-motif selectively inhibits MMP-2 and -9 activity efficiently (IC50

10 µM). The CTT-peptide also inhibits migration of HT1080 fibrosarcoma, C8161 melanoma,SKOV-3 ovarian carcinoma and KS1767 Kaposi’s sarcoma cell lines as well as human endothelialcells. In vivo, the CTT-peptide homes to tumors, suppresses tumor formation, targets tumorvasculature and prolongs survival of tumor-bearing mice. The peptide has not been found to becytotoxic. It has been thought that the Trp-residue in the HWGF-motif may bind to the hydrophobicpocket of the substrate cleft in the gelatinases and that the His-residue may act as a ligand for thecatalytic Zn2+-ion. The cyclic conformation is also critical for peptide function (Koivunen et al.1999). In later experiments, CTT-bearing liposomes were found to be effectively taken up andinternalized by gelatinase-expressing carcinoma cell lines. In addition, uptake of CTT-liposomes

33

containing adriamycin, a widely used anti-cancer drug, was significantly enhanced and increasedcancer cell-killing activity (Medina et al. 2001).

2.3.3. MMPIs in clinical trials

In early phase I clinical trials prolonged treatment with hydroxyamate-based broad-spectrumMMPIs caused musculoskeletal pain and inflammation resulting from the inhibition of MMP activityinvolved in normal turnover of connective tissue in tendons and joints, complications not seen inpreclinical models. The conditions were reversible and patients were able to continue treatmentafter a brief pause in drug receivment. Recently, inhibition of MMPs involved in the release ofmembrane-bound proTNF-α from its receptor has been implicated in the musculoskeletal pain sideeffects.Phase II trials have also turned out to be problematic. MMPIs are cytostatic rather than cytotoxic,and thus conventional measures of drug efficacy such as reduction in tumor size can not be usedto monitor drug activity. Instead, reduction in the increasing rate of tumor markers in serum wasused to define a biologically active dose of MMPIs although changes in biomarker levels in serumdo not necessarily reflect tumor regression.Phase III trials were initiated in the mid-1990s. MMPIs were often administered at randomlyselected doses or at the maximum tolerated dose instead of optimal biological doses. Theconducted phase III trials have not been very encouraging (Table 3). In one study patientsreceiving the MMPI Tanomastat for treatment of pancreatic and small-cell lung cancer showedsignificantly poorer survival than patients receiving a placebo. Despite the large number of patientsaccrued in MMPI clinical trials, there remains no clear demonstration that any dosing schedule orcompound tested has reached levels sufficient to inhibit target MMP activity within the tumor tissue(Coussens et al. 2002). However, CMT-3 (Metastat) was in a multicenter phase I clinical trial.Disease stabilization in patients with nonepithelial type of cancer for up to 26 months wasdemonstrated as well as a significant reduction in plasma MMP-2 levels (Rudek et al. 2001). Dueto positive outcomes and responses of treating patients with HIV-related Kaposi’s sarcoma inphase I clinical trials with CMT-3, phase II clinical trials are now ongoing.

Table 3. Fate of MMPIs in clinical trials for various cancers.

Compound Company Phase of clinical trial

BB-94, Batimastat British Biotechnology Ltd. Terminated since 1998TA-2516, Marimastat Schering Plough/Tanabe Seiyaku Terminated for nowBMS-275291 Bristol-Myers Squibb Phase II/III recruitingAG3340, Prinomastat Agouron/Hoffman-La Roche HaltedBAY12-9566, Tanomastat Bayer HaltedAE-941, Neovastat Aeterna Laboratories Phase III ongoingMMI270 Novartis PhaseI/II conductedCMT-3, Metastat CollaGenex Pharmaceuticals, Inc. Phase IIAdapted and modified from (Coussens et al. 2002, Kähäri and Saarialho-Kere 1999).

2.4. Laminin-5

Ln-5 is a non-collagen heterotrimeric glycoprotein constituted by the association of threegenetically distinct polypeptides, the α3, β3 and γ2 chains. Ln-5 is a major component of theanchoring filaments in the hemidesmosomal complex within the BM (Figure 7,10) mediatingimportant functions in cell adhesion, migration, proliferation, wound healing, homeostasis of skinand development (Colognato and Yurchenco 2000). Ln-5 was discovered in the beginning of the90’s and is also known as kalinin (Rousselle et al. 1991), epiligrin (Carter et al. 1991), nicein(Marinkovich et al. 1993, Vailly et al. 1994) and ladsin (Miyazaki et al. 1993). Under non-reducingconditions, the intracellular heterotrimeric human Ln-5 molecule is a 460 kDa high-molecularweight precursor form. This form contains an unprocessed 200 kDa α3-chain, a 140 kDa β3-chain

34

and an unprocessed 155 kDa γ2-chain. Upon secretion, human Ln-5 heterotrimers exist in 440 kDaand 400 kDa forms, resulting from extracellular processing of the α3 and γ2-chains, while the β3-chain remains unprocessed. In the 440 kDa form, the α3-chain is processed to a 165 kDa form andin the 400 kDa form, also the γ2-chain is processed into a 105 kDa form. In cell culture, the α3-chain is infrequently found in a 145 kDa form, while it is abundant in tissues (Marinkovich et al.1992, Rousselle et al. 1991). The γ2 and α3-chain also exist in different alternative spliced variants(Airenne et al. 2000).

2.4.1. Structure of the heterotrimeric Ln-5 molecule

Ln-5 folds into a rod with two asymmetric globular domains at either end of a long central helicaldomains (Marinkovich et al. 1993). Ln-5 is currently the only laminin with truncations in all threeshort arms being a rod bounded by the G domain at one end and by one or two globular domainsat the other. The α3-chain contains one small globular domain and three EGF-like modules. Theβ3-chain contains one globular domain and six EGF-like repeats (Aumailley and Gayraud 1998).In the human Ln-5 γ2-chain, the first N-terminal amino acids form a signal peptide. The N-terminaldomain V (residues 28-196) is formed of three and one half cysteine-rich EGF-modules. Domain IV(residues 197-381) contains a single cysteine and forms a globular structure. Domain III (residues382-608) is formed by the second half of an EGF module and four additional modules. Domain I/II(residues 608-1193) starts with two closely spaced cysteines and contains one cysteine close tothe C-terminus. The γ2-chain is a α-helical domain with a heptad repeat structure typical for coiledcoil proteins. The γ2-chain lacks the N-terminal globular domain VI present in the other γ−chains,which forms the globular structure at the end of the short arms. There are six putative Asn-linkedglycosylation sites (Kallunki et al. 1992). The N-terminal region of mouse Ln-5 γ2-chain comprises606 residues including a signal peptide of 21 peptides (Sasaki et al. 2001). The mouse and humanLn-5 γ2-chain are highly conserved (Pyke et al. 1994).

2.4.2. Ligands for Ln-5

Ln-5 binds to integrins α6β1, α6β4 and α3β1 (Carter et al. 1991), syndecan (Utani et al. 2001) andpossibly α-dystroglycan presumably through the C-terminal G domain of the α3-chain (Aumailley etal. 2003, Shang et al. 2001). Keratinocyte adhesion to Ln-5 via α3β1 promotes assembly of gapjunctions and gap junction intercellular communication integrating individual cells into synchronizedcolonies (Nguyen et al. 2000b). Ln-5 and α6β4-integrin on the other hand forms a complexinteracting with the intermediate filaments forming stable hemidesmosomal adhesion complex thatretain cells stationary at the BM (Carter et al. 1991).Studies on functional blocking of α3β1 and α6β4 integrins on migratory cells depositing Ln-5 arevery contradictory. Blocking α3- or α6-integrin subunits either increases or decreases normalhuman keratinocytes (NHKs) motility and adhesion (Goldfinger et al. 1999, O'Toole et al. 1997,Rousselle and Aumailley 1994, Zhang and Kramer 1996). Blocking both α3β1 and α6β4 integrinsusually abolishes both adhesion and migration (Goldfinger et al. 1999, Zhang and Kramer 1996).This dispersity may be due to differences in intracellular signalling. α3β1-focal adhesions (FAs)interact with actin-containing stress fibers and mediates HNKs adhesion and spreading on Ln-5 viatwo pathways, one is phosphoinositide 3-OH-kinase (PI3K) dependent and the other pathway isthrough RhoGTPase activity, a regulator of actin stress fibers. Migration of HNKs on collagen isdependent on α2β1-integrin mediated signalling and is only seen after TGF-β1 treatment. Thediverse signalling is thought to serve to regulate communication between cell-substrate and cell-cell adhesions (Decline and Rousselle 2001, Nguyen et al. 2000a, Shang et al. 2001, Zhang andKramer 1996). Both haptotactic and chemotactic migration of HaCaT and A431 carcinoma cells onLn-5 was carried out through α3β1 integrin. α6β4 can inhibit α3β1 controlled migration only when itis haptotactic but not chemotactic and this is mediated through the PI3K pathway. If migration ischemotactic, PI3K plays a stimulatory role, if migration is haptotactic, an inhibitory role. This may

35

partly explain the differential effects of Ln-5 for NHK and tumor cell migration (Hintermann et al.2001).

2.4.3. Processing of the Ln-5 chains

The heterotrimeric Ln-5 molecule is assembled intracellularly by formation of a β3γ2-chain dimerthrough disulphide bonds. The association of the α3-chain to the dimer is a rate-limiting step in theheterotrimer formation, assembling into the 460 kDa unprocessed form intracellularly. Assembly isnecessary for secretion of the complete Ln-5 molecule (Matsui et al. 1995).The α3 and γ2-chain of the 460 kDa intracellular precursor form of Ln-5 synthesized by culturedhuman keratinocytes undergoes proteolytic processing extracellularly (Marinkovich et al. 1992,Rousselle et al. 1991) (Table 4). The murine counterparts are 190/150 kDa α3-chain, 140 kDa β3-chain and 135/100 kDa γ2-chain (Baker et al. 1996b, Hormia et al. 1998, Sasaki et al 2001).The 200 kDa α3-chain shortening to a 165 kDa and 145 kDa protein at the C-terminus is predictedto occur at Gln1337-Asp1338 in the spacer region between the LG3 and LG4 domain (Amano et al.2000, Nguyen et al. 2000a, Tsubota et al. 2000) and is the first cleavage of heterotrimeric Ln-5processing. It has been demonstrated that MMP-2 and -14 cleave the human α3-chain (Veitch etal. 2002) but not the rat Ln-5 α3-chain (Giannelli et al. 1997, Koshikawa et al. 2000). The secondcleavage of the α3-chain to the 145 kDa form occurs within domain IIIA. Both these cleavage sitesare highly conserved between mouse and human (Amano et al. 2000). Although a recent studydemonstrated MMP-14 cleavage of human Ln-5 β3-chain in vitro, this processing has not beendemonstrated in vivo (Udayakumar et al. 2003).The human γ2-chain is processed into the 105 kDa form after incorporation of Ln-5 to the ECM,resulting in a lag before processing of the 440 kDa Ln-5 into the 400 kDa form (Amano et al. 2000,Gagnoux-Palacios et al. 2001, Marinkovich et al. 1992). The extracellular processing of the Ln-5γ2-chain removes the globular domain IV and the EGF-rich domain V of the short arm. The γ2-chain globular domain IV is required for incorporation into the ECM, but not for synthesis orsecretion of the γ2-chain. The γ2-short arm 50/35 kDa fragment, resulting from the processing isdeposited to the ECM by NHKs and mouse epidermal Pam212 cells (Gagnoux-Palacios et al.2001, Sasaki et al. 2001). The YSGD tetrapeptide is the unique physiological cleavage site of theextracellular processing of the γ2-chain (Table 4). The cleavage site is within domain III at thebeginning of the second EGF-like repeat at Gly434-Asp435 and is conserved between humans andmice (Amano et al. 2000, Gagnoux-Palacios et al. 2001, Sasaki et al. 2001). The isoenzymes bonemorphogenetic protein-1 (BMP-1), mammalian Tolloid, mammalian Tolloid-like 1 and mammalianTolloid-like 2 can all process the NHK synthesized Ln-5 α3-and γ2-chain at the physiologicalcleavage site (Amano et al. 2000, Veitch et al. 2002). In one study, MMP-14 readily cleaved thehuman Ln-5 γ2-chain into 105 kDa, ~90 kDa and ~70 kDa forms in vitro (Gilles et al. 2001) whileothers report that MMP-2 and-14 do not process the human Ln-5 γ2-chain (Veitch et al. 2002).BMP-1 and MMP-14 converts both rat and mouse 135 kDa Ln-5 γ2-chain to the 100 kDa form atthe physiological cleavage site, while MMP-2 and -14 can either directly convert the 135 kDa formto an 80 kDa form or through the 100 kDa form (Koshikawa et al. 2000, Sasaki et al. 2001, Schenket al. 2003, Veitch et al. 2002). MMP-2 cleaves the rat Ln-5 γ2-chain in domain III at A586-L587

(Table 4) resulting in an 80 kDa γ2x-chain. The cleavage site precedes two closely spacedcysteines in domain III (Cys-Pro-Ala-Cys), which are thought to be involved in joining the three Ln-5 subunits at the centre of the “cross”. This location indicates that the γ2x-chain remains attachedto the heterotrimer (Giannelli et al. 1997). Mouse epidermal Pam212 cells also secrete an ~70 kDaLn-5 γ2x-chain which is most likely produced through direct processing of the 135 kDa form(Sasaki et al. 2001). Human Ln-5 γ2-chain is cleaved by plasmin (Veitch et al. 2002) while the ratLn-5 γ2-chain is not (Giannelli et al. 1997).

36

Table 4. Cleavage sites in the Laminin-5 γγγγ2-chain.

Laminin-5 γ2-chain155 kDa-> 105 kDa cleavage site 105 kDa-> 80 kDa cleavage site

Human YSG434 - D435ENPDIEC1-5 Human not determinedRat YSG413 – D414ENPDIEC7 Rat AAA586-L587TSCPACYN6

Mouse YSG434 - D435ENPDIEC8, 9 Mouse not determined(1Amano et al. 2000, 2Gagnoux-Palacios et al. 2001, 6Giannelli et al. 1997, 3Kallunki et al. 1992, 4Miyazaki etal. 1993, 8Sasaki et al. 2001, 7Schenk et al. 2003, 9Sugiyama et al. 1995, 5Vailly et al. 1994)

2.4.4. In vivo functions of Ln-5

During mouse development, Ln-5 is prominently expressed in organs of endodermal andectodermal origin where it is deposited to the BM area. Ln-5 is not found in muscle, neural orendothelial BMs (Carter et al. 1991). In adult human tissues, expression of Ln-5 mRNA is confidedto skin, lung and also kidney, thymus, choroid plexus, cerebellum and the brain intermediate zone(Kallunki et al. 1992).Ln-5 is an absolute necessity for hemidesomosome formation. Mutations in the three Ln-5 genesresults in mechanobullous skin blistering disorders, such as junctional epidermolysis bullosa (EB)of which the Herlitz type is lethal and inherited as an autosomal recessive trait. EBs arecharacterized by detachment of the epidermis from the dermis (Pulkkinen and Uitto 1999). Ln α3-chain null mice exhibited very similar blistering defects in skin and oral mucosa as human Herlitztype of EB patients with junctional epidermal-dermal separation and lethal phenotype. Ameloblastsof the α3-null mice incisors were reduced in size relative to the teeth of wild type mice and enameldeposition was abnormal (Ryan et al. 1999a). In both the developing human and murine teeth, Ln-5 γ2-chain mRNA and protein were detected in dental epithelial cells and along the BMs of theenamel organ. Ln-5 γ2-chain mRNA was expressed in differentiating ameloblasts and latercontinued in the secretory ameloblasts (Kainulainen et al. 1998, Sahlberg et al. 1998).

2.5. Cell migration

Active cell motility is essential in physiological tissue development and homeostasis, includingembryological morphogenesis, wound healing, immune surveillance and inflammation. In theprocess of migration, cells must make their way through the macromolecules of the ECM throughadhesion, change in shape as well as ECM proteolysis. Cells may move in a directed (taxis) orrandom (kinesis) manner along soluble (chemo) or solid (hapto) attractants. Thus, haptotaxisoccurs along gradients of ECM solid substrata. Factors stimulating epithelial cell motility includegrowth factors, cytokines, ECM fragments and cryptic sites. With respect to motility, themicroenvironment regulates epithelial morphogenetic movements critical for organogenesis, tissueformation, remodelling, healing and renewal (Quaranta 2002).Cell migration is also a means for tumor cell invasion, the hallmark of malignant tumors. Tumor cellmetastasis proceeds through neovascularisation and penetration of the BM resulting inintravasation of tumor cells to the blood flow or lymphatic vessels followed by extravasation oftumor cells to new sites and growth of tumors. Migrating tumor cells must cross multiple barrierscovering distances up to several hundreds of micrometers (Friedl and Brocker 2000).The pattern of cell motility is the same for both cancer cells and in normal tissue remodelling. Alsonormal cells invade tissues locally as can be seen in any morphogenetic migratory movementsinvolving epithelial cells (Quaranta 2002). Sheet migration is a common strategy for epithelia tocover or envelop tissue which occurs in development as well as during wound repair. EpithelialSCC cells express the same MMPs as epithelial keratinocytes during wound healing. In thisaspect, tumors have been compared to nonhealing wounds (Quaranta 2002)

37

2.5.1. Mechanisms of cell migration

Cell adhesion is essential for cell migration and prior to forward movement, cells require arearrengement of adhesion complexes. When a cell prepares to move, a protrusion orfilopod/lamellipod with ruffling spikes is formed for attachment to the substrate. Upon attachment tothe substrata, integrins cluster in the cell membrane to ligand-binding sites which inducesintracellular signalling and subsequent assembly of cytoskeleton-binding FA complexes (Friedl andBrocker 2000), typically composed of integrins and different laminin-isoforms. FAs provideattachment-dependent outside-in signalling leading to the recruitment and polymerization of actinmonomers and the formation of actin cables or bundles (stress fibers) (Ray and Gately 1996).Following substrate binding at the leading edge, a gradient of binding and traction forces isestablished in the cells resulting in forward movement of the cell in relation to the underlyingsubstrate. In locomotive cells, FAs, then termed focal contacts (FCs), are dynamic structures ofsmaller size, incompletely assembled, less stable and may also contain molecules not present inFAs (Friedl and Brocker, 2000).

2.5.2. Proteolysis and ECM remodelling during cell migration

Cells may move in the three-dimensional ECM without focalized proteolysis simply by changingshape. Focalized pericellular proteolysis mediated by membrane-associated MMPs can help cellsto “make the way” for migration by reducing the mechanical force in the matrix. Focalizedproteolysis can also be used to release matrix-bound growth factors, chemotactic signals or todecrease adhesion of the cells on the surrounding substrata thus enhancing migration. Pericellularproteolysis gives the action of proteinases a spatially restricted and regulated environment throughwhich extracellular signal transduction can be mediated, concentrates MMPs within a near vicinityof their substrates, limits access of MMP inhibitors and enhances MMP activation (Sternlicht andWerb 2001).

2.5.3. MMPs and tumor cell migration

MMPs have been implicated not only in promoting cancer development through removal of theECM barrier but also in the modulation of signals affecting cellular transformation, tumor growth,angiogenesis and apoptosis. The first indication for a role of MMPs in cancer came from Liotta andhis collagues (Liotta et al. 1979) in the 1980s when he and his collagues discovered that a type IVcollagenase was involved in melanoma invasion and metastasis through breakdown of BMs.Several studies show that blocking tumor cell migration with both broad-spectrum MMPIs andserine proteinase inhibitors do not fully abolish the ability of tumor cells from migrating (Friedl andBrocker 2000). Proteases may mediate important migratory functions not directly related to theshear ability of a tumor cell to move and invade. In a tumor cell aspect, such functions would beadhesion to any substrata coming in its way, clearance, finding the path of least resistance anddirectionality. Some MMPs are expressed by the tumor cells, i.e. MMP-7 is expressed by thecarcinoma cells in gastrointestinal, breast, lung and in cutaneous carcinomas (Wilson andMatrisian 1996). MMP-8 is expressed by oral SCC tumor cells (Moilanen et al. 2002). MMPs-2, -9and –14 originate from epithelial cells of various carcinomas (Coussens et al. 2002). MMP-14 isproduced by stromal fibroblasts in breast, colon, and head and neck cancers, and in bothcarcinoma cells and fibroblasts in pancreatic, ovarian, gastric and thyroid cancers (Nabeshima etal. 2002).Recent research has led to the concept that MMPs in tumor tissues are largely produced byreactive stromal cells recruited to the neoplastic environment (Coussens et al. 2002). MMPs fromadjacent stromal cells are often induced and commandeered by the malignant epithelial cells. Inaddition, inflammatory cells are a prominent feature of many malignant tumours and can be asource of certain MMPs in the peritumoural environment. MMP-1, -2, -3, -9, -11, -12, -13 and –14are all localized to the stromal compartment of breast cancer tumors and in colonadenocarcinomas (Basset et al. 1997, Lynch and Matrisian 2002). It has also been noted that

38

increased MMP expression correlates with tumor aggressiveness. MMPs –1, –2, -7, -8, -9, -10 and-11 are expressed and increasingly secreted with tumorigenicity by cultured epithelial and SCCcells (Bachmeier et al. 2000a, Bachmeier et al. 2000b). Increased expression of MMP-2 andcollagenases correlates with the invasiveness of malignant melanomas (Kähäri and Saarialho-Kere1999, Väisänen et al. 1996).Actual evidence for tissue degradative activity of distinct MMPs in tumour tissues in vivo is limited.A study with MDCK-cells overexpressing various MMPs indicated that MMP-14 provided cells withan invasive phenotype and collagen-degrading ability, while other MMPs did not (Hotary et al.2000). In general, MMP-overexpressing transgenic mice develop spontaneous hyperproliferativelesions. MMP-7 knock-out mice exhibit reduction in intestinal tumourigenesis and MMP-2 deficientmice show reduced angiogenesis and tumour progression (Kähäri and Saarialho-Kere 1999). Withsome exceptions, however, there are few indications that the elimination of a single MMP canreduce tumor invasion and metastasis.Since MMPs participate in angiogenesis, MMPs can also be considered critical for metastasisthrough the formation of new blood vessels. This is particularly true for the activity of MMP-2, -9and -14 mediated fibrinolysis, but also for other MMPs that can release or activate growth factorsinvolved in neovascularisation (Hiraoka et al. 1998, McCawley and Matrisian 2000, Vu et al. 1998).MMPs can also regulate angiogenesis by generating anti-angiogenic peptides. MMPs –2, -3, -7, -9and –12 generate angiostatin in vitro by cleaving plasminogen (Pg) and MMP-7 and –9 in vivo.Angiostatin inhibits basic fibroblast growth factor-induced endothelial cell proliferation and tubeformation in vitro (Cornelius et al. 1998). Endostatin is a C-terminal fragment of type XVIII collagenwith antiangiogenic activity. MMP-3, -7, -9, -13 and –20 can generate endostatin-like fragmentsfrom type XVIII collagen in vitro. Recombinant endostatin efficiently blocks angiogenesis andsuppresses primary tumor and metastasis growth in animal models (Nyberg et al. 2003, Sternlichtand Werb 2001). The hemopexin domain of MMP-2, released after autolysis in human tumors canalso inhibit angiogenesis, cell proliferation and migration (Bello et al. 2001).MMP-8 expression was associated with non-metastatic breast carcinoma cell lines both in vitro andin vivo. Knock-down of MMP-8 expression by antisense perturbation resulted in an increase in theinvasive potential of the non-invasive breast carcinoma cell line (Agarwal et al. 2003). MMPs mayalso cleave apoptotic signalling molecules and affect immunoevasion (Lynch and Matrisian 2002,Sheu et al. 2001). Proteolytic exposure of a cryptic site in the type IV collagen molecule by at leastMMP-2 can regulate angiogenesis and tumor growth in vivo (Xu et al. 2001). MMP-9 associateswith the hyaluronan receptor CD44 on invasive carcinoma cells and promotes tumor invasion andangiongenesis (Yu and Stamenkovic 2000) while MMP-14 can cleave CD44 and induce carcinomacell migration (Kajita et al. 2001). MMP-14 and MMP-2 are typically localized at the leading front ofinvading cultured carcinoma cells for efficient pericellular proteolysis and MMP-2 activation(Nabeshima et al. 2002, Nakahara et al. 1997).

2.5.4. Ln-5 in physiological cell migration

Ln-5 mediates adhesion of many different cell lines and blocking the Ln α3-chain effectively inhibitsadhesion of NHKs (Rousselle and Aumailley 1994, Rousselle et al. 1991). NHKs deposit Ln-5 tothe ECM, however, NHKs are usually stationary but can be induced to migrate by TGF-β1, whichdecreases the adhesion of NHKs on human Ln-5. TGF-β1-treated migrating NHKs and MCF-10Acells deposit the 200 kDa unprocessed α3-chain (Decline and Rousselle 2001, Nguyen et al.2000b, Zhang and Kramer 1996). Plasmin-processing of the α3-chain has been found to inducenucleation of HDs, at least in carcinoma cells (Goldfinger et al. 1999, Goldfinger et al. 1998, Veitchet al. 2002). The excised 40 kDa fragment resulting from α3-chain processing does not exhibit anysignificant cell scattering-activity (Tsubota et al. 2000). Domain III of the γ2-chain is involved inmediating cell migration. In transgenic mice, potential involvement of the γ2-chain in cell migrationwas shown to be related with a 613 bp upstream region flanking the LAMC2-gene and containing aGATAA box, as well as AP-1 and Sp1 binding sites (Salo et al. 1999).

39

2.5.5. Ln-5 and tumor cell migration

Ln-5 was initially found to have cell-scattering effects when purified from human gastric carcinomacells and rat NBT carcinoma cells and when applied to both normal and carcinoma cell lines itstimulated cell migration and disruption of intercellular connections. The Ln-5-like moleculeinduced actin cytoskeleton reorganization and FC formation in the tumor cells (Miyazaki et al.1993). It was also demonstrated that addition of soluble Ln-5 to culture media induces rapidadhesion and spreading of HaCat and pancreatic carcinoma cell lines (Baker et al. 1996a). Cellmigration on rat Ln-5 is induced by MMP-2 and -14-mediated cleavage of the γ2-chain whileMCF10 cell migration on Ln-5 can be induced by exogenous addition of MMP-2 (Giannelli et al.1997, Koshikawa et al. 2000). Constitutive HT1080 and BRL cell migration on Ln-5 is mediated byLn-5 cleavage through endogenously produced MMP-2 and –14 (Koshikawa et al. 2000). Veryaggressive melanoma cells secrete both the 100 and 85 kDa Ln-5 γ2-chains along with MMP-2, -9and –14 (Seftor et al. 2002). Recently, a study with human prostate cancer cell lines showed thatthe human Ln-5 β3-chain was cleaved by MMP-14 and induced carcinoma cell migration(Udayakumar et al. 2003). Also hepatoma, breast and colon carcinoma cell migrations are inducedon Ln-5 via the MMP cleavage mechanism (Koshikawa et al. 2000). In vivo, the 80 kDa Ln-5 γ2x-chain is present in rodent tissues undergoing remodelling and in mouse carcinomas but not inquiescent tissues (Giannelli et al. 1997, Giannelli et al. 1999). MMP-2, -14 and Ln-5 are co-localized in breast and colon carcinoma tissues in vivo (Koshikawa et al. 2000). EGF inducesmigration and expression of both MMP-14 and Ln-5 in the cell periphery of the outgrowth (Gilles etal. 2001) and cleavage of the Ln-5 γ2-chain liberates a fragment with EGF-family of ligandhomology (Schenk et al. 2003). Endogenously produced MMP-2 activated TGF-β1, subsequentlystimulating noninvasive hepatocellular carcinoma cells (HCC) cells to migrate on Ln-5 and with aconcomitant upregulation of α3β1-integrin (Giannelli et al. 2001, Giannelli et al. 2002).In vivo, the loss of BM continuity has been associated with increasing malignancy of carcinomas,based mostly on morphologic data. BMs are not only degraded, but also synthesized incarcinomas (Lohi 2001). The Ln-5 γ2-chain protein and mRNA is expressed in colonadenocarcinomas, ductal mammary carcinomas, melanomas, ameloblastomas, pancreaticcarcinomas, prostate carcinomas, SCCs and most types of lung carcinomas by the cancer cellsand in most cases in the budding tumor cells confined at the invasive front (Fukushima et al. 1998,Hao et al. 1996, Kainulainen et al., 1998, Määttä et al. 1999, Pyke et al. 1994, 1995, Tani et al.1997). Ln-5 is neoexpressed in thyroid tumors although Ln-5 is totally lacking in the normal adultas well as fetal thyroid tissue (Lohi et al. 1998). A peculiar observation is that gastric carcinomacells secrete monomeric Ln-5 γ2-chain (Koshikawa et al. 1999). Ln-5 has also been suggested tobe a marker for malignant transformation of tumors in cervical carcinomas (Skyldberg et al. 1999).The significance of the Ln-5 γ2-chain expression for the malignant behaviour of tumors is atpresent unknown. Ln-5 γ2-chain expression does not seem to have a major influence on tumourclinical behaviour (Määttä et al. 1999).

2.6. Periodontal tissue

The healthy periodontium, providing support to maintain teeth in adequate function, is composed offour principal components: the gingiva, periodontal ligament, alveolar bone and cementum (Figure5). The gingiva is histologically composed of the stratified squamous keratinizing gingivalepithelium which covers the underlying connective tissue. The gingival epithelium can be dividedinto three different types based on their location and composition: the outer oral epithelium, thesulcular epithelium (SE) lining the gingival sulcus extending from the tip of the gingival crest to thecoronal most portion of the junctional epithelium. The SE and oral gingival epithelium are verysimilar in cellular structure and composition. The junctional epithelium extends from the base of thegingival sulcus to an arbitrary point close to the alveolar bone crest and forms the tissueattachment of the gingiva to tooth structures. The junctional epithelium is associated with two BM-like structures, the internal basal lamina involved in the dento-epithelial complex and the externalbasal lamina facing the underlying connective tissue (Bartold et al. 2000). The internal basal

40

lamina is unique in that it lacks typical structural molecules for BMs, the major in human tissuesthus being Ln-5. Since Ln-5 is unable to self-polymerize, it has been suggested that Ln-5 may binddirectly to the tooth enamel surface (Hormia et al. 1998, Oksanen et al. 2001). The mainmacromolecule in the gingival connective tissue is type I collagen. The gingival connective tissueserves primarily to protect the root surface and alveolar bone form the external oral environmentand additionally, provides support of epithelial tissues and the teeth within their alveolar housing(Bartold et al. 2000).

Figure 5. Structure of the tooth and the surrounding periodontal tissue. Picture reprinted withpermission from Taina Tervahartiala from her doctoral thesis, printed in Helsinki 2003.

2.7. Skin

Skin is the largest organ of the body comprising some 16% of the body weight (Figure 6). Skincovers the body surface being the primary defence against microbial invasion, dehydration andmechanical, chemical, osmotic, thermal and photic damage. Skin consists of two main layers, thekeratinizing stratified squamous surface epithelium or epidermis and the subjacent collagen-richconnective tissue layer providing support and nourishment, the corium or dermis (Figure 6). TheBM structure separates the epidermis from the underlying dermis. Beneath the dermis is a looserconnective tissue layer, the superficial fascia, or hypodermis, which in many places is largelytransformed into subcutaneous adipose tissue. The hypodermis is loosely connected to anunderlying deep fascia, aponeurosis, or periosteum (Bloom and Fawcett 1968).

41

Figure 6. Cutaneous wound healing. Upon injury of intact skin, a clot forms to prevent further bloodflow and is further covered by a scab. Neutrophils are the first inflammatory cells to arrive to thewounded site, which occurs within hours. Within 3-7 days, keratinocytes migrate resealing thedenuded surface, fibroblasts migrate to the wound site forming the granulation tissue andneutrophils are replaced by macrophages. Wound contraction with extensive connective tissueremodelling occurs for several weeks after injury.

42

2.7.1. The basement membrane

The BM is a specialized extracellular structure, whose main function is to act as a selective andphysical barrier between cells and the underlying connective tissue (Figure 7). The BM alsoprovides a base for tissue repair and acts as a dynamic link between the attached epithelial cellsand the underlying connective tissue, mediating signals for cellular movement, polarization,proliferation and differentiation. The integrity of BMs is indeed important as a barrier to the spreadof tumor cells. The 80 nm thick BM or basal lamina consists of two distinct layers: a 40 nmelectron-lucent layer (lamina lucida or rara) adjacent to the basal plasma membrane of the cellsthat rest on the lamina and a 40 nm electron-dense layer (lamina densa) just below (Figure 7). Thebasal lamina underlying the epidermis of the skin is made up of the HDs and FAs (Ray and Gately1996).

2.7.1.1. The hemidesmosomal complex

The HDs are highly specialized junctional complexes attaching epithelial cells to the underlying BMin stratified and other complex epithelia such as the skin, the cornea, parts of the gastrointestinaland respiratory tract, oral mucosa, mammary gland and the amnion (Ray and Gately 1996).Ultrastructurally, HDs are small electron dense domains of the plasma membrane on the ventralsurface of basal keratinocytes in human skin. Their most conspicuous component is a tripartitecytoplasmic plaque to which bundles of intermediate filaments are attached. HDs are associatedwith a sub-basal dense plate in the lamina lucida and are connected via fine thread-like anchoringfilaments to the lamina densa. The cross-banded anchoring fibrils project into the upper regions ofthe papillary dermis where they loop back and reinsert into the lamina densa. These structuresconstitute the functional unit called the hemidesmosomal adhesion complex (Borradori andSonnenberg 1999).The cytoplasmic constituents include at least bullous pemphigoid antigen 230 and plectinbelonging to the plakin family. The transmembrane proteins of HDs are α6β4 integrin and bullouspemphigoid antigen 180 (type XVII collagen). The ectodomain of bullous pemphigoid antigen 180spans the lamina lucida and a flexible tail intertwines within the lamina densa. α6β4 integrin islocalized at the basal pole of the basal epithelial cells where it coordinates a structural linkagebetween intermediate filaments and the ECM portion of the BM (Borradori and Sonnenberg 1999).The association of Ln-5 with α6β4 integrin is crucial for the nucleation of HD assembly and for themaintenance of stable adhesion (Baker et al. 1996a). The lamina densa separates the connectivetissue from the BM structure. The structures in the lamina lucida include the anchoring filamentswhich are thin, threadlike structures connecting HDs to the lamina densa. These layers arecomposed of type IV and VII collagen, Ln-5, Ln-6 (α3β1γ1), Ln-7 (α3β2γ1), Ln-8 (α4β1γ1), Ln-10(α5β1γ1), nidogen, perlecan, fibulin –1 and –2. The anchoring fibrils extend from the lower portionof the lamina densa to the underlying mesenchyme and are almost exclusively composed of typeVII collagen (Burgeson and Christiano 1997). The truncated Ln-5 γ2-chain lacks the domain VIcontaining the high-affinity nidogen binding site found in the Ln γ1-chain. Thus, Ln-5 is not directlyconnected to the type IV collagen network and perlecan. The Ln-5 through its β3-chain forms acovalently linked complex with the α3-chain of Ln-6, containing nidogen-binding sites (Amano et al.2000, Borradori and Sonnenberg 1999). However, Sasaki et al. demonstrated that the N-terminalregion of the Ln-5 γ2-chain contains two binding sites each for fibulin-2 and heparin and additionalbinding sites for fibulin-1 and nidogen-1 (Sasaki et al. 2001). A model predicts that whilemonomeric Ln-5 is concentrated below the hemidesmosomal plaque linking the α6β4-integrin (andprobably bullous pemhigoid antigen 180) to type VII collagen, the Ln-5-6-7 complex (but not theLn-5 monomer) is participating in the stabilization of the BM in the interhemidesmosomal space(Aumailley and Gayraud 1998, Borradori and Sonnenberg 1999, Rousselle et al. 1997)

43

Figure 7. The basement membrane. The functional role of Laminin-5 in the hemidesmosomalcomplex is illustrated. Not all molecules are shown.

2.7.2. Estrogen and the skin

Estrogen is a reproductive hormone mediating essential biological processes in the body. In thedeveloped world, due to longer life span, women spend more than one-third of their life-timepostmenopausally in a state of profound estrogen deprivation. Hill has estimated that by 2030 theworld population of postmenopausal women will be of the order of 1.2 billion (Hill 1996).Menopause is marked by a sharp change in hormonal balance and a decline in estrogen proceedsto reduced progesterone levels and permanent amenorrhea. The menopause is known to havedeleterious effects on bone mass, increase the risk of cardiovascular disease as well as breast andendometrial cancers (Hill 1996). However, the effects of estrogen-deprivation on skin have beenless well documented. The skin is a target for several hormones, including estrogen, and mostcells within the skin contain estrogen receptors (ERs). A higher ER content has been found inwomen than in men. Estrogen increases the mitotic rate of epidermal cells and enhancesvascularization. Estrogens have been extensively used in hormone replacement therapy (HRT)and birth control (Ashcroft et al. 2003).

2.7.2.1. Estrogen and skin collagen content

Type I collagen comprises about 80% of dermal collagen and plays a major role in providing tensilestrength to skin. Skin thickness is proportional to the collagen content (Shah and Maibach 2001).Premenopausal women were found to deposit more collagen in acute test wounds than age-

44

matched men (Jorgensen et al. 2002). In postmenopausal women, skin collagen content andthickness declines exponentially after 2-4 years into the menopause (Affinito et al. 1999). Inaddition, BM type IV collagen decreases with age while BM thickness increases (Vazquez et al.1996). Studies addressing the relationship between skin collagen content, skin thickness andchronological age are controversial. Apparently, the reduction in collagen content correlates withyears from the menopause rather than chronological age (Affinito et al. 1999, Brincat et al. 1985,1987a, 1987b). However, in postmenopausal women treated with HRT the age-related decrease isreversed. Skin thickness increases within 6 months of HRT and then seems to plateau (Brincat etal. 1985, 1987a, 1987b). However, there are also contradictory results, since one year treatmentwith systemic estrogen had no effect on skin collagen in postmenopausal women (Haapasaari etal. 1997). In addition, topical estrogen treatment increases type I and III collagen synthesis andcontent in human skin (Shah and Maibach 2001).

2.8. General characteristics of inflammation

Inflammations share several similar characteristics although distinct features exist in respect to thedisease etiology and the involved tissue. Wound healing can be considered a specialized form ofinflammation. The difference between acute and chronic inflammation is not clearly distinguishable(Figure 8). In general, an acute inflammation becomes chronic when a prolonged inflammatoryresponse resides in the affected tissue. Recovery from acute inflammation involves elimination ofthe factor (bacteria, foreign bodies etc.) causing the inflammatory response, formation of anintermediate tissue, removal of necrotic debris, recruitment of cells to the site of injury (fibroblastsand endothelial cells), neoangiogenesis, epithelial cell migration (if present), as well asregeneration and remodelling of the tissue. Inflammation per se is necessary to protect the hostfrom infection, but persistent inflammation can also cause disease and tissue destruction. Acuteinflammation may proceed to chronic inflammation if the cause of the initial injury is not completelyeliminated but may also develop in the absence of acute inflammation. Causes of chronicinflammation can be infections, remnants of dead organisms, foreign bodies, metabolic products,immune reactions, chemical and physical factors/irritants or other disease-related factors(Trowbridge & Emling 1989).In most inflammations, neutrophils are the first line of defence and within a few days, neutrophilsare replaced by macrophages and lymphocytes. Neutrophils and macrophages release proteases,pro-inflammatory cytokines and engage in essential functions including phagocytosis of debris,microbial agents and degraded matrix components (Trowbridge & Emling 1989). MMP expressionby inflammatory cells is spatially and temporarily restricted, however, the battery of MMPs whichcan be expressed by inflammatory cells contributing to the inflammatory response in tissues isimpressive: monocytes express MMP-1, -7, -8 and -9; macrophages MMP-1, -2, -3, -8, -9, -10, -12and -13; PMN neutrophils store MMP-8 and MMP-25 in their secondary granules as well as MMP-9in the tertiary granules and also MMP-19 and MMP-25 are localized to neutrophils; plasma cellsexpress at least MMP-2, -8, -9 and –13. Thus, the inflammatory cells have an impressive capacityfor MMP secretion at the site of inflammation (Barrick et al. 1999, Prikk et al. 2001, Wahlgren et al.2001) . The MMPs within neutrophils can be secreted to ECM after stimulus with various cytokines,of which IL-8 is one of the most potent. Since neutrophils do not synthesize TIMPs, neutrophildegranulation is a powerful ECM degradative pool (Van den Steen et al. 2002).

45

Figure 8. Typical features and differences in acute and chronic inflammation. Adapted from (Nwomehet al. 1998b).

2.8.1.Chronic inflammatory periodontal disease

Adult periodontitis is characterized by chronic periodontal inflammation and irreversible attachmentloss associated with alveolar bone destruction leading to tooth loss. The host response isresponsible for the matrix degradation (Trowbridge and Emling 1989, Van Dyke and Serhan 2003).The initial phase of chronic periodontal disease, gingivitis, is the most common chronicinflammatory disease in humans (Williams et al. 1992). Gingivitis and periodontitis are the result ofan induction of host inflammatory responses to the accumulation of periodontopathogens and theirproducts (i.e. LPS, leukotoxin, proteolytic enzymes) on tooth surfaces adjacent gingival tissues.Initially, gingivitis represents a generalized acute inflammatory response but if the inflammatoryresponse contained within the gingivitis lesion spreads to the deeper periodontal tissues andalveolar bone is lost, the resultant lesion is termed periodontitis. The progression from gingivitis toperiodontitis is unclear and people can have gingivitis which never progress into periodontitis(Bartold et al. 2000, Van Dyke and Serhan 2003). Typically, the gingival pocket epitheliumproliferates into the underlying connective tissue in the form of a network of finger-like projectionscoinciding with ECM degradation and loss of tooth attachement. Subsequent to the initialinflammatory response, connective tissue destruction occurs within 3-4 days after plaqueaccumulation. As the inflammatory process develops, alveolar bone loss results in tooth mobilityand, ultimately, tooth loss if the disease is left untreated and continues to progress (Bartold et al.2000, Van Dyke and Serhan 2003).

46

2.8.1.1. MMPs in chronic periodontal disease

In periodontitis, the degradation of collagen fibers of the periodontal ligament and alveolar boneresorption are the most profound manifestations of the uncontrolled soft and hard tissuedegradation. High levels of proteinase activity in periodontal tissue and GCF correlates withdisease severity (Van Dyke and Serhan 2003).In inflamed periodontal tissue, gingival fibroblasts express MMP-1, -2, -3, -8, -11 and -14 whilegingival keratinocytes express MMP-1, -2, -3, -7, -8, -9, -10 and -13. Macrophages and endothelialcells express MMP-1, -2, -3 and -9. GCF typically contains IL-1α, IL-1β, TNF-α, IL-6, IL-8 and IFN-γ (Birkedal-Hansen et al. 1993, Van Dyke and Serhan 2003). Active MMP-8 and -13 aresignificantly elevated in GCF from patients with AP, in comparison to GCF from healthy subjects.MMP-8 levels also correlate well with clinical status of the patients (Ingman et al. 1996, Mäntyla etal. 2003, Sorsa et al. 1988, Tervahartiala et al. 2000). MMP-2 and -14 mRNA are expressed byfibroblast-type cells in both healthy and diseased gingival tissue, although the cells expressingMMP-2 and -14 tend to localize closer to the site of inflammation in AP (Dahan et al. 2001,Tervahartiala et al. 2000). MMP-2 mRNA expression is increased in SE of AP patients whencompared to clinically healthy gingiva (Tervahartiala et al. 2000). While proMMP-2 and -9 arepresent in both healthy and inflamed gingival tissue extracts, active MMP-2 is seen only in inflamedsamples and total MMP-9 protein and mRNA are elevated in patients with periodontal disease(Korostoff et al. 2000, Westerlund et al. 1996). In AP patients, TIMP-1 mRNA and protein levelsare usually unchanged in both GCF and at the tissue level (Ingman et al. 1996).

2.8.1.2. MMP inhibition in periodontal disease

TCs have been successfully used in the treatment of human inflammatory periodontal disease. Thenon-antimicrobial effect of the TCs promotes connective tissue attachment, inhibits MMPs andosteoclast function. Due to side-effects of TCs, the low-dose doxycycline (LDD) therapy wasproposed. By administering specially-formulated capsules each containing as little as 20% (20 mg)as opposed to commercially available (100 mg) doxycycline capsules, the side-effects could beovercome. More importantly, the treatment significantly inhibited the pathologically-excessivecollagenase activity in the gingival tissue and GCF of AP patients, significantly decreasedperiodontitis-associated bone loss as well as MMP-8 and -13 levels in GCF (Golub et al. 1997,Rifkin et al. 1993). Currently, the only approved MMP-inhibition based pharmaceutical on themarket is LDD (trade name Periostat) for treatment of human inflammatory periodontal disease inthe USA. Also CMT-1, -3, and –8 effectively inhibited the elevated gelatinase activity, reducedbone loss and reduced the proinflammatory cytokine levels associated with LPS-inducedperiodontitis in rats (Ramamurthy et al. 2002) and CMT-1 inhibited also collagenase activity in thegingiva of diabetic rats (Yu et al. 1993).

2.8.2. Cutaneous wound healing

Wound healing is an acute inflammation which is very commonly referred to as a “pathologicalevent” (Figure 6). Wound healing can be regarded to be pathological if one considers any traumapathological, if the wound is of substantial depth or size, or if there is chronicity associated with thewound. Nutritional deficiencies, chronic diseases, adrenal steroids, diabetes and estrogen-deficiency affect wound healing adversely. Wound healing is, however, a physiologicalinflammation in that it is necessary for the organism to reseal any openings in the surfaces of thebody. It is highly unlikely that any human will live her entire life without gaining any minor skininjury. In this regard, it is more accurate to say that living organisms have a well developed“physiological” means to deal with torn surfaces. Wounds are divided into acute and chronic basedon healing speed and wound quality Chronic wounds are etiologically different but they share oneor more characteristics: repeated trauma, bacterial infection and local tissue ischemia (Chen et al.1999).

47

2.8.2.1. Clotting

When the skin is injured, a blood clot consisting of thrombocytes, neutrophils and monocytesembedded in fibrin fibers is formed. The function of the fibrin clot is to provide a temporary shieldand cover the wound and also to act as a provisional matrix for cell attachment, migration, woundcontraction and re-epithelialization during the tissue repair process. The clot also serves as areservoir of cytokines and growth factors that are released as activated platelets degranulate(Martin 1997).

2.8.2.2. The inflammatory response

Neutrophils normally begin arriving at the wound site within minutes of injury (Martin 1997,Yamaguchi and Yoshikawa 2001). Immunohistochemical staining for monocytes/macrophages andB lymphocytes have shown that there is a delayed inflammatory response in aged mice whencompared to young animals (Ashcroft et al. 1997b). There is also an age-related increase in pro-inflammatory cytokines such as TNF-α by neutrophils (Ashcroft et al. 1999). In this regard,elevated TNF-α, IL-1, -6 and -8 levels have been associated with chronic wounds (Chen et al.1999, Trengove et al. 1999). Estrogen has been shown to enhance the oxidative metabolism ofactivated neutrophils during phagocytosis and also to down-regulate the macrophage migrationinhibitory factor, a pro-inflammatory cytokine produced by monocytes, T-lymphocytes, endothelialcells and keratinocytes. This leads to reduced inflammation, enhanced matrix deposition andaccelerated wound healing. These effects depend upon a predominant ER-α response.Neutrophils and ER-positive cells are more numerous in untreated wounds from elderly humansthan in subjects with estrogen-treated wounds. Estrogen also modulates the inflammatoryresponse not only by inhibiting neutrophil chemotaxis to the wound but also by altering theexpression of neutrophil adhesion molecules such as L-selectin, thereby dampening the migrationof cells from the vasculature into the injured tissue (Ashcroft et al. 1999).

2.8.2.3. Re-epithelialization

Re-epithelialization involves the formation of a provisional wound bed matrix, the migration ofepidermal keratinocytes from the cut edge of the wound, the proliferation of keratinocytes feedingthe advancing and migrating epithelial tongue, the stratification and differentiation of newepithelium and the restoration of BM integrity. The keratinocytes forming the cut edge of the woundbegin to migrate within 24-48 hours. The keratinocytes migrate from the edge of the wound under aprovisional matrix of fibrin and fibronectin and over or through the viable dermis (Martin 1997,Yamaguchi and Yoshikawa 2001). Initially, keratinocytes form FCs with the fibrillar collagen,wherafter cell protrusion is favoured by gaps and pathways of least resistance. The gaps may beprovided by collagenolytic activity. Several studies with postmenopausal women and OVX animalsshow that estrogen-deprivation leads to delayed re-epithelialization and reduced type I and type IVcollagen deposition (Ashcroft et al. 1997a, 1997b). However, women receiving systemic HRT foronly 3 months show a marked increase in the rate of reepithelialization and collagen depositionalmost comparable to that of premenopausal women (Ashcroft et al. 1997a, 1999). Studies withtopical estrogen treatment of both elderly human subjects as well as OVX animals showaccelerated wound healing (Ashcroft et al. 1997a).

2.8.2.4. Wound contraction

Wound repair involves phenotypic change of fibroblasts from quiescent to proliferating cells, andsubsequently to migratory mediated through different chemotactic signals, and then to stationarymatrix producing and contractile cells. Wound contraction is accomplished by myofibroblasts thatcontain α-smooth muscle actin and mediate contractile forces produced by granulation tissue inwounds. Granulation tissue is composed of a fibronectin-rich scaffold, fibroblasts, new bloodvessels and increasing amounts of collagen types I/III. During granulation tissue formation,

48

fibroblasts are stimulated by TGF-β to deposit new ECM proteins. The apoptosis of myofibroblastsand vascular cells finally mediates the transition from a granulation tissue into a scar. The finalphase is that of remodelling in which collagen is synthesized, degraded and reorganized(Yamaguchi and Yoshikawa 2001). Studies on the effects of estrogen on fibroblast infiltration,granulation tissue formation, collagen deposition and angiogenesis are very controversial. It isknown though, that estrogen modulates the expression of cytokines such as IL-1, expressed bymacrophages and fibroblasts, known to stimulate formation of granulation tissue. Aging is alsoassociated with reduced capillary growth and delayed angiogenesis (Schnaper et al. 1996). In a ratmodel of ovariectomy, it was shown that wound contraction was delayed 4 months post-OVX butnot 2 weeks post-OVX compared to Sham-operates (Calvin et al. 1998).

2.8.2.5. Growth factors regulating wound healing

Growth factors enhancing keratinocyte migration are EGF, TGF-α, keratinocyte growth factor(KGF), TGF-β1, hepatocyte growth factor, insulin like growth factor-1 and insulin. The levels andfunction of many of these growth factors are regulated by pro-inflammatory cytokines such as IL-1and TNF-α during cutaneous injury. These growth factors act as both mitogenic as well aspromigratory factors on the epidermis to drive wound closure. EGF induces lamellipodial extensionand the assembly of FA complexes as part of the crawling response of tissue culture fibroblastsand epithelial cells (Martin 1997). KGF protein expression was enhanced in cutaneous wounds ofestrogen-treated OVX rats when compared to untreated controls (Ramamurthy et al. 2003). Woundfluids from chronic wounds inhibited the proliferation of cultured cells while that from acute woundsstimulated growth. MMPs within the fluid from chronic wounds degrade exogenously added EGFsignificantly more than fluid from acute wounds in vitro (Trengove et al. 1999).

2.8.2.5.1. TGF-β and scarring

The two extremities of life have one thing in common, reduced or neglible scarring following woundhealing. Until late fetal stages, there is generally no sign of a connective tissue scar where awound has healed: the repair is perfect. In addition, oral mucosal wound healing is characterizedby less scarring and accelerated wound closure.Scar tissue is composed of excessive accumulation of disorderly arranged collagens, mostly type Iand III, proteoglycans and persistent myofibroblasts. During the course of wound healing, type Icollagen mRNA expression starts within 24 hours after wounding, being prominent in fibroblasts upto 21. Collagen α1(I) chain expressing fibroblasts are initially localized to the deeper layers ofgranulation tissue but at later stages (day 8 and forward) they are mainly detected beneath thenewly reconstituted epidermis (Scharffetter et al. 1989). ECM accumulation in fibrotic conditions isaccompanied by elevated mRNA steady-state levels of fibrillar collagens (Verrecchia and Mauviel2002). In the embryo, TGF-β1 is expressed transiently and at low levels after injury, but at the adultwound site it is present at high levels for the duration of healing and beyond. Adult gingivalfibroblasts located in the papillary connective tissue share many properties with fetal fibroblasts(Martin 1997). TGF-β1 induces the synthesis and inhibits the degradation of ECM, stimulates theformation of granulation tissue and collagen deposition. TGF-β also acts as a negative regulator ofepithelial cell proliferation, but in contrast, stimulates the proliferation of mesenchymal cells. TNF-αinhibits collagen transcription and suppresses TGF-β stimulation of collagen mRNA expression(Jacinto et al. 2001). Wound TGF-β1 levels are decreased at day 7 of wound healing in elderlywomen when compared to young women or women receiving HRT. Cell culture studies indicatethat estrogen is the major reproductive hormone involved in dermal fibroblast production and/orsecretion of TGF-β1. Thus, TGF-β1 is hormonally modulated during wound healing (Ashcroft et al.1997a). Interestingly, lack of estrogen and TGF-β1 improves scarring, which is superior in oldersubjects in terms of color, texture and contour when compared to hypertrophic scarring in youngwomen (Ashcroft et al. 1997a, 1997b).

49

2.8.2.6. MMPs in wound healing

MMPs are crucial for wound healing (Ågren et al. 2001, Lund et al. 1999). However, most studiesdemonstrate that MMP expression and activity are elevated in chronic wounds when compared toacute wounds (Saarialho-Kere 1998, Trengove et al. 1999). Usually MMP activity decreasesconsistently in patients with chronic wounds when progress from nonhealing to healing occurs.MMP-1 is prominently expressed in leading edge migrating keratinocytes in both acute and chronichuman wounds (Ashcroft et al. 1997c, Inoue et al. 1995, Pilcher et al. 1999, 1998; Saarialho-Kere1998). In human cutaneous wound healing, extensive research has been conducted upon thefunction of MMP-1 in the process of leading edge keratinocytes. When migrating keratinocytescomes into contact with fibrillar collagen, α2β1 integrin rapidly triggers MMP-1 expression, which issustained by subsequent crosstalk with activated EGF receptor in an autocrine manner. MMP-1then cleaves fibrillar collagen I leaving behind denatured collagen, i.e. gelatin. It has beenhypothesized, that in partially degraded collagen, an RGD site is exposed, recognized by integrinαvβ3 leading to upregulation of MMP-2, which then cleaves the gelatin (Pilcher et al. 1997, 1998,1999).Both in acute human open dermal wounds and in chronic nonhealing ulcers, the elevated level ofMMP-8 in wound exudates and tissue extracts is a hundred fold higher than MMP-1 levels. MMP-8was primarily present in the proform in acute wound fluids and in its active form in chronic woundfluids (Nwomeh et al. 1998a, 1999). At the initial phases of wound healing, large amounts ofdamaged type I collagen is present and MMP-8 has a high affinity for type I collagen (Armstrongand Jude 2002).MMP-13 is abundantly expressed by activated fibroblasts as well as macrophage-like cells inchronic wounds distinct from areas of stromal cells expressing MMP-1. In contrast, MMP-13 mRNAexpression is usually low or undetectable in acute human wounds (Saarialho-Kere 1998, Vaalamoet al. 1997). Since the mouse has only just recently been found to have two MMP-1 genes, futurestudies remain to show the expression and function of murine MMP-1 in wound healing (Balbin etal. 2001). Most murine studies have analysed the expression of MMP-13 during cutaneous woundhealing and it appears that this MMP does mimick the expression of human MMP-1 at least tosome extent. In acute murine wounds, MMP-13 is expressed by the leading edge keratinocytesand has also been detected in macrophage-like cells in the granulation tissue directly beneath thescab (Lund et al. 1999, Madlener 1998, Okada et al. 1997).MMP-9 expression is usually confined to leading edge keratinocytes of human burning wounds,oral mucosal wounds and in the epidermis of various blistering diseases (Ashcroft et al. 1997c,Oikarinen et al. 1993, Salo et al. 1994) while it has not been detected in the keratinocytes ofhuman cutaneous chronic wounds (Saarialho-Kere et al. 1993). In addition, MMP-9 is usuallyexpressed by neutrophils and macrophage-like cells in the wound granulation tissue in humanacute, chronic and blistering wounds (Ashcroft et al. 1997c, Oikarinen et al. 1993, Salo et al. 1994,Saarialho-Kere et al. 1993). In acute murine dermal wounds, MMP-9 is prominently expressed inmigrating keratinocytes (Lund et al. 1999, Madlener 1998, Okada et al. 1997) and MMP-9 has alsobeen detected in distinct cells of the mouse granulation tissue (Madlener 1998). MMP-2 ispredominantly expressed by fibroblasts within the granulation tissue and basal expression isreadily elevated upon tissue injury in the dermis of acute, burn and chronic wounds as well as invarious blistering diseases (Oikarinen et al. 1993, Salo et al. 1994). MMP-2 protein has also beendetected in the leading epidermal wound edge (Ashcroft et al. 1997c, Oikarinen et al. 1993, Salo etal. 1994). Gelatinolytic activity in chronic wound fluids is significantly elevated when compared toacute wounds. In addition, the ratio of MMP-9 to TIMP-1 is aberrant in chronic wound fluids. Thereis also a shift from more latent to active forms of gelatinases in the wound fluid from chronicwounds (Bullen et al. 1995, Trengove et al. 1999). In healing murine wounds, MMP-2 and -14mRNA expression co-localizes and is detected mostly in fibroblasts of the wound granulation tissuejuxtaposed to the proliferative epithelial cell layer (Lund et al. 1999, Madlener 1998, Okada et al.,1997).TIMP-1 is usually expressed in epidermis and in the stroma of acute cutaneous wounds, however,TIMP-1 is usually undetectable in chronic wound tissue and wound fluids (Bullen et al. 1995,

50

Nwomeh et al. 1999, Vaalamo et al. 1997). In mice, TIMP-1 was not detected in epithelia, incontrast to human acute and chronic wounds (Madlener 1998). Also diminished levels of α1-PI andα2M in chronic wounds have been reported (Yager et al. 1996). MMP-2, MMP-9 and neutrophilelastase levels were found to be increased at late stages of wound healing in elderly persons(Ashcroft et al. 1997c).

2.8.2.7. MMP inhibition in wound healing

Most studies on acute wound healing combined with the experimental use of MMPIs have resultedin delayed re-epithelialization. GM6001-treatment, however, was found to increase incisionalwound strength without affecting collagen content. By using an exprimental model for spongegranulomas, GM6001 reduced the inflammatory response and decreased type I collagen content.GM6001 upregulated MMP activity, reduced wound fluid TNF-α levels and increased total TGF-β1levels in the sponge granulomas (Witte et al. 1998). In contrast, GM6001-treatment resulted indelayed re-epithelialization of human suction blisters and in murine acute cutaneous wounds(Ågren et al. 2001, Lund et al. 1999). When Pg-deficient mice were additionally treated withGM6001, the acute wounds did not heal at any time point due to a reduction in the ability of theleading edge keratinocytes to degrade fibrin and subsequently make their way through the fibrinclot (Lund et al. 1999). In human suction blisters, GM6001 treatment resulted in an upregulation ofstromal MMP-1 and -2 mRNA expression and in epithelial MMP-9 expression, while in the Pg-deficient mice, MMP-9 expression was also detected in the macrophage-like cells. MMP-2 wasdetected in the dermo-epidermal junction and GM6001-treatment prevented activation of MMP-2.In the GM6001-treated Pg-deficient mice, MMP-2 expression changed so that MMP-2 mRNA wasalso detected in the leading edge keratinocytes. Overall MMP expression was upregulated inGM6001-treated Pg-deficient mice, a pattern not seen in Pg-deficient animals without GM6001treatment (Lund et al. 1999).

2.8.2.8. Ln-5 in wound healing

Within 8 hours of injury to the epidermis or the oral mucosa in both human and mouse and prior tomigration, a precursor form of Ln-5 is synthesized and deposited into the BM by keratinocytes atthe wound edge (Kainulainen et al. 1998, Larjava et al. 1993, Nguyen et al. 2000b, Pyke et al.1994). Early deposition contrasts with the delayed expression of other BM components, such astype VII collagen and type IV collagen, which are missing under the migrating keratinocytes.Keratinocyte attachment and spreading on collagen via the α2β1 integrin is associated withdeposition of Ln-5 concurrent with adhesion. The γ2-chain short arm (domain III) is the ligand forthe α2β1 integrin making it likely that the α2β1 binds the 140 kDa γ2-chain precursor form (Declineand Rousselle 2001). Ln-5 expression is upregulated by TGF-β in cultured oral mucosalkeratinocytes while TNF-α had no effect on Ln-5 expression in epidermal keratinocytes(Kainulainen et al. 1998).

51

3. Aims of the study

1. Human periodontal disease exhibits all the typical features of chronic inflammation. Theadvantage of periodontitis as a model for chronic tissue destructive inflammation is easy andusually non-invasive sample access. MMPs are known to be elevated in inflammatory diseasesand an MMPI has successfully been used in the treatment of adjunctive AP. Therefore, weassessed the effect of two inflammatory cytokines, TGF-β1 and TNF-α on MMP-2 and MMP-9expression in cultured human oral mucosal keratinocytes. We further studied the in vivolocalization of gelatinolytic activity and the expression of gelatinases MMP-2, MMP-9 and theLn-5 γ2-chain in inflamed human gingival tissue. Concomitantly, we investigated the effect ofdifferent synthetic MMPIs on MMP-2 and MMP-9 expression and activity in cultured oralmucosal keratinocytes as well as on in situ gelatinolytic activity.

2. Reproductive hormones such as estrogen are known to influence wound healing. MMPs areelevated in wound tissue from women who are past menopause but no studies haveinvestigated the effect of estrogen on MMP-8 expression during wound healing. Estrogen isknown to modulate the Ln-5 γ2-chain but no studies have previously investigated the effect ofestrogen on Ln-5 γ2-chain expression and processing during wound healing. The mechanismsby which MMPIs impact wound healing are poorly understood and generally tested on acutewound healing. We therefore investigated the effect of an MMPI and estrogen on cutaneouswound healing in OVX rats by studying re-epithelialization, collagen content, MMP expressionand the Ln-5 γ2-chain processing.

3. MMP-2 and -14 are known to cleave the Ln-5 γ2-chain and induce cell migration. Migratingkeratinocytes deposit Ln-5 while migrating on many substrata, such as collagen during woundhealing. Carcinoma cells synthesize Ln-5 and may use it for spreading and invasion. Thus, weinvestigated the MMP expression, Ln-5 deposition and effect of synthetic MMPIs on migrationof cultured human oral mucosal keratinocytes. We also assessed if other MMPs in addition tothe previously characterized MMP-2 and -14 could cleave the Ln-5 γ2-chain and inducecarcinoma cell migration in addition to testing the effect of MMPIs on cell migration over Ln-5.

52

4. Materials and Methods

4.1. Animals (III & IV)

Rat wound tissue biopsies were obtained from Professor Nungavarm S Ramamurthy, Departmentof Oral Biology and Pathology, School of Dental Medicine, State University of New York at StonyBrook, USA. Animal experiments were conducted with the appropriate IACUC approvals granted toDr.Nungavarm Ramamurthy. Wound tissues were excised from six months old adult femaleSprague-Dawley rats which were either Sham operated or bilaterally ovariectomized. 120 dayslater the rats were anesthetized with a a mixture of Ketamine and Xylazine. Eight full thickness skinwounds were made in the dorsal thorax (4 on each side) using a 6 mm diameter circular biopsypunch and the wounds were allowed to heal by secondary intention. Wound biopsies werestandardized and white petrolatum was applied to the wounds as previously described (Eckles etal. 1990). The Sham rats were divided into two groups: placebo and CMT-8. OVX rats were dividedinto three groups: placebo, CMT-8 and estrogen. There were six rats in each group except for theestrogen group which had 4 rats. The CMT-8 (15 mg/kg body weight) was administered to the ratsby oral gavage in 1 ml of 2% carboxymethyl cellulose. Placebo rats received only 1 ml of 2%carboxymethylcellulose. Estrogen, 100 µg/ml was dissolved in corn oil and 10 µg was administereddaily by subcutaneous injection. On day 7 after creating the wounds, the animals wereanesthetized, blood samples were collected and the skin containing four wounds was excised forhistological analysis. Only the granulation tissue was removed from three other wounds forbiochemical analysis using a 3 mm disposable biopsy punch. The wound closure and amount ofgranulation tissue were measured as previously described (Ramamurthy et al. 1999).

4.2. Tissue samples (I & II)

Inflamed buccal oral mucosa from AP patients (n=5) and clinically healthy gingival tissuespecimens were obtained from patients prior to any therapy, following the patients informedconsent. The samples were snap-frozen in liquid nitrogen. The collection of human tissue samplesfor the studies was approved by the ethical committee of the Institute of Dentistry, University ofHelsinki, Finland.

4.3. Materials (I-V)

Ln-5 was purified from serum-free DME of 804G cells as described (Koshikawa et al. 2000,Koshikawa et al. 1992). Non-functional Ln-5 mouse mAb TR-1 (Plopper et al. 1996) waschemically conjugated to protein A-Sepharose 4B as described (Koshikawa et al. 1992).Concentrated culture media was passed over the TR-1 antibody column and absorbed Ln-5 waseluted with 0.05% trifluoroacetic acid, pH 2.5.APMA, an optimal organomercurical activator of latent proMMPs (Van Wart and Birkedal-Hansen1990) was purchased from Sigma Chemicals, St.Louis, MO. The broad spectrum MMP-inhibitors,BB-2516 (Marimastat) and BB-94 (Batimastat) were kindly provided by British Bio-Technology,Ltd., Oxford, England. LW-1 (3,4-dihydro-1-oxo-1,2,3,-benzotriazine-3-(3-tetrahydrofuranyl)carbonate, LW-2 (1,2-dihydro-3,6-dioxo-2-phenyl-pyridazine-1-methylcarbonate), LW-3 (3,4-dihydro-1-oxo-1,2,3,benzotriazine-3-(2-methoxy) ethylcarbonate), LW-4 (1,2-dihydro-2-ethoxycarbonyl-(1-oxo-isochinolin-5-phenyl) ethylcarbonate and LW-5 (1(2H)-phtalazinone-2-(4-methoxy-phenyl) carbonate) were kindly provided by Dr.R.Klause, Drug Research Department,Luipold Pharma, M nich, Germany. All the CMTs (Figure 4) were kindly provided by CollaGenexPharmaceuticals Inc., Newtown, Pennsylvania. Recombinant TGF-β and TNF-α were obtainedfrom R&D Systems, Abingdon, United Kingdom.

53

4.4. Cell lines (I & V)

Human oral mucosal keratinocytes (HMKs) were obtained from surgical gingival biopsy andcultured as described (Salo et al. 1994). The HMKs were found to be spontaneously immortalizedand its passages 13-18 were used in the experiments. The chromosome number of the cells wasexamined and found to be around the hypertriploid range (70-76). Primary cultures of normalhuman skin keratinocytes were obtained during mastectomy for breast cancer (supplied by Dr.JuhaPeltonen, University of Turku, Turku, Finland). Cells were maintained in specialized keratinocytemedium (KGM, Clonetics Corp., San Diego, CA).A549 is a human lung carcinoma and MCF-7 is a human breast adenocarcinoma cell line. Thesecell lines were maintained in DME (Gibco BRL) with 10% fetal calf serum (v/v) (Irvine Scientific),penicillin, streptomycin at 37°C in 5% CO2/95% air.

4.5. MMP inhibition of cell proliferation (I)

The effect of MMP inhibitors was analyzed as described (Kueng et al. 1989). 20 000 keratinocyteswere seeded into 96-well plates (Nunclon, Denmark) and allowed to attach for 24 hours at 37°C.The cells were then incubated with 200 nM to 50 µM concentrations of CMTs, LWs or Batimastat.Cells were treated with 100 ng/ml recombinant TNF-α to stimulate MMP-9 production. The cellswere fixed with 4% (v/v) formaldehyde in phosphate-buffered saline (PBS) containing 5% (w/v)sucrose and stained with 0.1% crystal violet in boric acid (pH 6.0). After destaining with 10% aceticacid, the absorbance was measured with a Multiscan MS plate reader (Version 4.0, Labsystems,Helsinki, Finland) at 595 nm. Cell numbers obtained were used to equalize the quantity of mediumMMP levels analyzed by zymography. The effect of TGF-β on keratinocyte MMPs was analyzed byzymography of both medium and cell-associated fractions. HMKs were cultured into 80%confluence and then 10 ng/ml TGF-β1 was added for 48 hours. Culture medium was collected andthe cells were extracted at 4°C for 10 min in PBS containing 2% Triton X-100 (Sigma).

4.6. Cell migration assays

4.6.1. Radial migration (I)

Radial migration was measured by seeding 50 000 cells into cylinders placed in uncoated 24-wellculture plates and allowed to attach overnight. The cylinders were removed and non-adherent cellswere washed away and adherent cells were allowed to migrate out from the cell disk for 4 days inthe presence or absence of inhibitors at 37°C. The culture medium was collected for zymographyand the cells were fixed with 4% (v/v) formaldehyde in PBS containing 5% (w/v) sucrose andstained with 0.1% crystal violet in boric acid (pH 6.0). The amount of migration was measured bydensitometric scanning of the area covered by the cells.

4.6.2. Scratch assay (I)

Confluent keratinocyte cultures were seeded onto uncoated coverslips and a scratch “wound” wasmade with a pipette tip. Cells were then allowed to migrate in the presence or absence of 10 ng/mlTGF-β1 and 1 µM batimastat for 20 hours. The cells were then fixed as described above and cellmigration was measured under the microscope by counting the cells moved into the “wounded”space.

4.6.3. Transwell assay (V)

Cell migration was performed in Transwell chambers (Corning Star) as previously described(Koshikawa et al. 2000). Filters were coated with 750 ng/ml Ln-5 or 1 µg/ml type IV collagenovernight at 4°C and thereafter blocked with 5% milk powder. DME including 1% Hepes, penicillin,

54

streptomycin, glutamine (w/v) and 0.1% BSA (w/v) was added to the lower chamber. MMP-3, -8, -13, -14, or –20 (Table 5) was added to the lower chamber at a final concentration of 10 nM. Cellswere resuspended in DME (see above) and 100 µl was seeded to the upper chamber. Cellmigration assays were carried out in the presence or absence of 500 nM, 1 µM Batimastat, 100 µMCTT-peptide or 100 µM unrelated control peptide CERGGLETSC. After 16 h of incubation, thecells that had migrated onto the lower surface of the filters were stained with 0.5% crystalviolet/20% methanol and counted.

Table 5. MMPs used in the study.

Enzyme Type and Source Ratio(enz:sub)

Reference

MMP-2 human recombinant 1:10 (Koivunen et al. 1999, Sorsa et al. 1997,Westerlund et al. 1996)

MMP-2 human gingival fibroblastderived

1:10 (Westerlund et al. 1996)

MMP-3 human recombinant 1:10 Sigma Chemicals, St.Louis, MOMMP-3 synovial fibroblast derived 1:5 (Sorsa et al. 1994)MMP-7 human recombinant 1:10 Chemicon International Inc., Temecula, CAMMP-8 human recombinant 1:20 Chemicon International Inc., Temecula, CAMMP-8 human neutrophil derived 1:10 (Konttinen et al. 1991)MMP-9 human neutrophil derived 1:10 (Kjeldsen et al. 1992, Sorsa et al. 1997,

Westerlund et al. 1996)MMP-12 human recombinant 1:30 Elastin Products Company Inc., Owensville,

MOMMP-13 human recombinant 1:5 Invitek GmbH, GermanyMMP-13 synovial fibroblast (Lindy et al. 1997)MMP-14 human recombinant 1:10 Invitek GmbH, GermanyMMP-20 human recombinant 1:20 Dr.John Bartlett, Harvard Forsyth Dental

Center, Boston, MA

4.7. Collagenase and gelatinase activity assays (III & IV)

For hydroxyproline (a specific iminoacid marker of collagen) analysis, the samples were hydrolysedin 6 M hydrochloride acid (HCl) at 106°C for 24 hours and then assayed by Stegmenn’scolorimetric assay. For collagenase activity assay, wound tissues from each group were pooledand extracted in collagenase buffer (pH 7.4) containing 5 M urea and precipitated with 65%ammonium sulphate. Collagenase and gelatinase activities in the pooled extracts were measuredas described (Golub et al. 1994) using radiolabeled collagen or gelatin as substrates.

4.8. Zymography (I,IV,V)

Zymography was performed as previously described (Mäkelä et al. 1994, Ryan et al. 1996). A549serum-free culture media was collected and spun down at 800 rpm to remove cells. Thesupernatant was concentrated using ammonium sulphate at 80% saturation. The protein pellet wasreconstituted in 300 µl of 20 mM Tris-HCl, pH 7.5; 0.1 M NaCl-buffer. Media was dialyzed againstthe same buffer and prepared for zymography. Wound tissue extracts and culture media wereanalyzed by separating proteins according to size by 10% sodium dodecyl sulphate-polyacrylamidegel electrophoresis (SDS-PAGE) containing 1 mg/ml 2-methoxy-2,4-diphenyl-3(2H)-furanone-labeled gelatin (O'Grady et al. 1984) or unlabeled gelatin. The lysis of gelatin was monitored bylong-wave UV light or by staining the gel with with 0.1% Serva Blue R (Serva Feinbiochemica,GmbH, Heidelberg, Germany) and subsequent destaining with 20% methanol, 10% acetic acid.

55

4.9. Western immunoblotting (I-V)

Proteins were characterized by separation with SDS-PAGE under non-reducing or reducingconditions. Conditioned media from the HMK culture was shaked overnight at 4°C with gelatin-sepharose-4B (Pharmacia, Sweden) to bind MMP-9. MMP-9 was released from the pellet byheating the samples at 90°C for 3 min with Laemmli buffer (Laemmli 1970). Wound tissue extracts,culture media or protein samples were equally treated with Laemmli buffer (with or withoutreducing agents) and the proteins were separated according to size by 8% or 11% SDS-PAGE.The proteins were then electrotransferred to nitrocellulose membranes (Schleicher & Schuell,Dassel, Germany) in Blotting-buffer (25 mM Tris-HCl, pH 8.3; 192 mM Glycine, 20% methanol).Non-specific binding was blocked with 5% milk powder (Valio, Finland) or 3% gelatin. Themembranes were incubated with primary antibodies (Table 6) overnight and subsequently withsecondary antibodies conjugated to alkaline phosphatase or horseradish peroxidase. Proteinswere detected by incubating the membranes with nitroblue tetrazolium (Sigma, St.Louis, MO) and5-bromo-4-chloro-3-indolyl phosphate (Sigma, St. Louis, MO), diaminobenzidine tetrahydrochloridesolution or by the enhanced chemiluminesence technique (Amersham Pharmacia Biotech Inc, NJ,USA). When the enhanced chemiluminesence technique was used, the membranes were exposedto Hyperfilm-ECL autoradiography films. The intensities of immunoreacted proteins weresemiquantitated with a Bio-Rad Model GS-700 Imaging Densitometer using the MolecularAnalyst - program.

Table 6. Antibodies used in the study.

Antibody Type Dilution ReferenceMMP-2 mouse anti-human 1:1000 Chemicon International, Temecula, CAMMP-2 rabbit anti-human 1:200 Henning Birkedal-Hansen, University of

Alabama1MMP-2 rabbit anti-human 1:500-2000 (Westerlund et al. 1996)MMP-3 rabbit anti-human 1:1000 Chemicon International, Temecula, CA

1MMP-8 rabbit anti-human 1:250-500 (Hanemaaijer et al. 1997, Sorsa et al. 1994)MMP-9 rabbit anti-human 2 µg/ml Oncogene Science Inc., Cambridge, MA

1MMP-9 rabbit anti-human 1:300-1500 (Kjeldsen et al. 1992, Westerlund et al. 1996)MMP-13 mouse anti-human 1:500 Oncogene Research Products, Cambridge,

MA1MMP-13 rabbit anti-human 1:500 (Freije et al. 1994)

2Type ICollagen

mouse anti-human 1:1500 Sigma, St Louis, MO

2MMP-14 rabbit anti-human 1:100-500 Biogenesis Ltd., Poole, UKMMP-20 mouse anti-human 1:1000 (Väänänen et al. 2001)CD-45 mouse anti-human 1:50 Santa Cruz Biotechnology Inc., USA

Laminin-5γ2-chain

rabbit anti-human 1:50-1000 (Pyke et al. 1995, Salo et al. 1999)

Laminin-5 rabbit anti-human GB3 commercialAntibodies have been verified to crossreact with murine tissue antigens as previouslydescribed 1(Ramamurthy et al. 1999, Ryan et al. 1999b) or by the 2manufacturer.

4.10. In situ zymography (II)

Untreated frozen tissue sections were thawed and warmed to room temperature. Sections werecovered with either in situ zymography (ISZ)-buffer (50 mM Tris-HCl, pH 7.4, 1 mM CalCl2) aloneor with ISZ-buffer containing 0.5 mM APMA. All samples were incubated for 30 min at 37°C afterwhich the samples were covered with ISZ-buffer alone or ISZ-buffer containing one of the followingproteinase activity modifiers: 0.5 mM APMA, 0.5 mM Pefabloc (Boehringer Mannheim, GmbH,

56

Germany), 100 µM CTT-peptide, 100 µM CERGGLETSC-peptide, 100 µM GACFSIAHECGA-peptide or 10 mM Ethylenediaminotetraacetic-acid (EDTA) . Incubation was carried out for 30 minat 37°C after which the buffers were discarded. All samples were covered with a 1:1 mixture of 1%(w/v) low melting temperature agarose (Sigma, St.Louis, MO) in ISZ-buffer and 1 mg/ml OregonGreen-conjugated gelatin (Molecular Probes, Inc., Eugene, OR). Samples were covered with themixture alone or with the mixture containing one of the following: 0.5 mM PefablocSC Plus, 100µM CTT-peptide, 100 µM CERGGLETSC-peptide, 100 µM GACFSIAHECGA-peptide or 10 mMEDTA. The mixture was covered with a coverslip and allowed to gel at room temperature and thenthe samples were incubated in a humid chamber in the dark at 37°C for up to 96 hours.Gelatinolytic activity was evidenced as dark areas in the otherwise uniform fluorescence substratelayer.

4.11. Immunohistochemistry (I-IV)

Tissue sections were thawed or deparaffinized and rehydrated. The Vectastain ABC Elite Kits(Vector Labs, Burlingame, CA) were used for immunostaining as described. HMKs were seeded inKGM medium on uncoated coverslips for 3 days. Cells were fixed in 4% formaldehyde and 5%sucrose in Ca2+ and Mg2+-containing PBS (PBS+) for 30 min. Sections were pretreated with 0.4%pepsin (paraffin sections only) and endogenous peroxidase activity was blocked by incubation in0.6% H2O2 in methanol. Cultured cells were permeabilized with 0.5% Triton X-100 in PBS for 4min. Samples were blocked with goat/horse normal serum in 2% bovine serum albumin andincubated with antibodies overnight at +4°C. Following incubation with antibodies, samples wereincubated with biotinylated secondary antibody and subsequently with Avidin-Biotin enzyme-complex or a rhodamine-conjugated secondary antibody (Boehringer Mannheim, Indianapolis,USA). Sections were stained using 3-amino-9-ethylcarbazole as chromogen and counterstainedwith Mayer’s hematoxylin (Merck KGaA, Darmstadt, Germany). Samples processed for dualimmunostaining were incubated with a second primary antibody and stained with SG-color (VectorLabs). Controls for immunostaining included omitting antibodies from the staining sequence orincubation with nonrelevant immunoglobulins. The intensity of immunoreactivity was detected byconnecting a Kappa CF 15/4-camera to a Leica DMRB microscope and a Kappa MCUII-multicontrol unit from the camera to the computer. The software used for analysis was Leica Qwin.

4.12. In situ hybridization (I-IV)

cDNAs presented in Table 7 were used as templates for in vitro transcription with a riboprobetranscription kit (Boehringer Mannheim GmbH, Germany) according to the manufacturersinstructions. The sense and antisense cRNA transcripts were labelled with digoxigenin-11-UTP. Insitu hybridization was performed principally as previously described (Tervahartiala et al. 2000). Forin situ hybridization, solutions were treated with 0.1% diethylpyrocarbonate. Samples were fixed in4% paraformaldehyde, proteolyzed in 0.2 M HCl at room temperature and in 0.1-0-5 µg/mlProteinase K (Finnzymes, Helsinki, Finland) at 37°C. The reaction was stopped by 100 mM glycine(in PBS) and acetylation was done with 0.25%-0.5% acetic anhydride (in 100 mM triethanolamine).Samples were dehydrated in a graded alcohol series prior to prehybridization. Hybridization buffercontaining 400-800 ng/ml digoxigenin-labeled antisense or sense probe was applied to thesections and incubated under stringent conditions at 55-58°C overnight. The hybridization buffercontained 10 mM dithiotreitol, 250 µg/ml yeast total-RNA (Boehringer Mannheim GmbH), 250µg/ml salmon sperm DNA (Sigma), 50% deionized formamide, 4 x SSC, 10 (w/v) dextran sulphate,2 x Denhardt´s solution (Sigma). Washing was done under stringent conditions as previouslydescribed (Tervahartiala et al. 2000). The digoxigenin-label was detected by incubating sampleswith alkaline phosphatase-conjugated anti-digoxigenin Fab-fragments (diluted 1:100) and by usingFast Red tablets as chromogen. The samples were counterstained with Mayer’s hematoxylin(Merck KGaA, Darmstadt, Germany).

57

Table 7. cDNA used for cRNA probe synthesis for in situ hybridization.

cDNA Accessionnumber

Fragment Length(base pairs)

Reference

Laminin-5γ2-chain

NM_005562 PstI-EcoRI 845 (Kallunki et al. 1992)

MMP-2 NM_004530 ScaI-SacI 635 (Huhtala et al. 1990)MMP-8 NM_002424 SphI 95 (Hasty et al. 1990)MMP-9 NM_004994 HindIII-EcoRI 574 (Huhtala et al. 1991)

MMP-13 NM_002427 BamHI-HindIII 729 (Freije et al. 1994)Footnote: The probes have been tested by Northern hybridization for specificity to human sequencesand by sequencing. The cDNA has been analyzed for homology between human and murine cDNAs.

4.13. MMP cleavage in vitro (II & V)

MMPs (Table 5) were incubated with 52 µM β-casein or 0.1-0.4 µg Ln-5 γ2-chain at indicatedenzyme:substrate ratios (Table 5) for different time intervals at 37°C. Some enzymes wereactivated with 0.5 mM APMA prior to incubation with the substrate. Enzyme activity was inhibitedwith 50-200 µM CTT-peptide, CERGGLETSC-peptide, GACFSIAHECGA-peptide, 10 mM EDTA or10-50 µM Marimastat. The reactions were stopped by boiling in Laemmli buffer (with or withoutreducing agent) and the cleavage of β-casein was evidenced by separating fragments according tosize with 11% SDS-PAGE, visualized with 0.1% Serva Blue R and destained with 20% methanol,10% acetic acid. Fragmentation of the 21 kDa β-casein band was regarded as enzyme activity.Cleavage of the Ln-5 γ2-chain molecule was detected by Western immunoblotting using a Ln-5 γ2-chain specific antibody (Table 6).

4.14. N-terminal sequencing (V)

3-4 µg Ln-5 γ2-chain was incubated with MMP-3, -8, -12, -13, -14 and –20 at 37°C in the presenceor absence of 1 mM APMA. After incubations, fragments were separated by 8% SDS-PAGE andproteins transferred onto ProBlott PVDF membrane (Applied Biosystems, Foster City, CA). Themembranes were treated for sequencing as described (Mozdzanowski et al. 1992). The N-terminalprotein sequences were determined by a 492 Procise protein sequencer (Applied Biosystems).

58

5. Results

5.1. MMP inhibition and cell growth (I)

None of the LWs nor Batimastat affected the cell number during the 2-day exposure as comparedwith control cultures. CMT-3 and -8 at a concentration of 20 µM reduced the cell number by about50% (P<0.001; Student’s paired t-test) as compared with the control cells. At 20 µM, CMT-1reduced the cell numbers by about 20% (P < 0.001). CMT-5 had no effect on keratinocyte growth.The results were similar in cultures in the absence or presence of TNF-α.

5.2. MMP inhibition and gelatinase production by cells (I)

The main gelatinolytic proteinase secreted by the HMKs was MMP-2 and only minimal amounts ofMMP-9 could be detected in the medium of 48 – and 96-hours cultures. MMP-2 production was notmarkedly influenced by any of the LWs. TNFα-induced MMP-9 synthesis was inhibited dose-dependently by the LWs. At 50 µM, inhibition was 61% for LW-1 and 79% for LW-2. The CMTsinhibited MMP-2 production dose-dependently. The inhibitory effect of CMT-1 was evident only at20 µM (26%) and 50 µM (64%), CMT-3 and CMT-8 inhibited MMP-2 production already at 10 µM(60-70%). CMT-5 did not have any effects on MMP-2 or MMP-9 secretion. TNFα-induced MMP-9was inhibited only at the highest concentration (50 µM) of CMT-1, -3 and -8. CMT-3 at 10 and 20µM increased MMP-9 levels in medium of TNFα-treated keratinocytes. Batimastat had minimaleffects on production of the two keratinocyte gelatinases at the concentrations studied (3-300 µM).

5.3. MMP inhibition and cell migration

5.3.1. Radial migration (I)

Addition of TNF-α to the culture medium induced MMP-9 expression by 15-fold while it had noeffect on constitutive MMP-2 expression. The migration of TNFα -treated cells was not differentfrom that of the control cells. Similar results were obtained using both oral mucosal and skinkeratinocytes. Using the radial migration method, CMT-3 and -8 at concentrations that inhibitedMMP-2 secretion but either increased or had no effect on MMP-9 had marked inhibitory effect (upto 90%) on HMK cell migration. CMT-1, inhibiting MMP-2 less efficiently, reduced cell migrationcorrespondingly to a lesser extent, even though the compound had about the same effect on cellgrowth as CMT-3 and -8. CMT-5 had no effect on MMP-2 expression and also no effect on cellmigration. Batimastat inhibited migration by about 60% at 1 µM. LW-1 and -2 had little or no effecton MMP-2 production and also had negligble effect on cell migration. The migration experimentswere repeated using NHKs and the results were very similar to those with the HMKs.

5.3.2. Scratch assay (I)

Inhibitors were also tested on cell migration using the “scratch” wound healing in vitro assay withconfluent keratinocytes. These cultures were exposed to TGF-β for 20 hours in the presence orabsence of 1 µM Batimastat. TGF-β increased secreted MMP-2 levels by twofold. In 2% Triton-X100 extracts of the cells that represent both the cell membrane-associated and cytoplasmicenzyme pools, MMP-2 levels were also increased by about twofold. In addition to proforms of theenzyme, active forms of MMP-2 representing about 30% of the total secreted enzyme could bedetected in the culture medium of TGFβ-treated keratinocytes. TGF-β increased medium MMP-9levels by fivefold while its activity could not be detected in the cell extracts. Eight times more cellsmigrated in the TGFβ-treated cultures, while in control cultures only a few cells migrated into thewounded space. Batimastat fully inhibited the TGFβ-induced epithelial cell migration.

59

5.3.3. Transwell assay (V)

Batimastat effectively inhibited the migration of MCF-7 cells over Ln-5 substrate. 100 µM CTT-peptide effectively reduced A549 cell migration over Ln-5 and type IV collagen, without fullyabolishing it. Migration of A549 cells incubated with 100 µM non-relevant CERGGLETSC-peptidewas comparable to migration of PBS-treated A549 cells. A549 cells plated on filters lacking Ln-5did not show any migration. The A549 lung carcinoma cells were found to endogenously produceMMP-2, -3, -8, -13, -14 and –20 as detected by Western immunblot.

5.4. Immunolocalization of MMP-2, MMP-9, Ln-5 γγγγ2-chain and CD45-positive cells ininflamed human gingival tissue (II)

Intense MMP-9 immunoreactivity was detected in clearly defined areas of the inflamed humangingival connective tissue, directly beneath the SE. Extracellular MMP-9 immunoreactivity couldoccasionally be detected within the SE of inflamed human gingiva. In the deeper layers of inflamedgingival connective tissue associated with less infiltration of inflammatory cells, only sporadicallydistributed, small patches of MMP-9 immunoreactivity could be detected. Gingival connectivetissue areas with strong MMP-9 immunoreactivity were otherwise predominantly associated withheavy infiltration by inflammatory cells. Immunostaining with CD45 antibody showed that MMP-9immunoreactivity colocalized with areas infiltrated by CD45-positive cells. Similar to MMP-9immunoreactivity, CD45-positive cells were confined to the gingival connective tissue area directlybeneath the SE and only sporadically distributed CD45 immunoreactive cells could be detected inthe deeper layers of the connective tissue. In inflamed human gingival specimens, MMP-2immunoreactivity was confined to the BM region just beneath the SE in areas where inflammatorycells were abundant. In unaffected areas lacking inflammatory cells, MMP-2 immunoreactivity washardly detectable. MMP-2 immunoreactivity was also detected in gingival fibroblasts. Ln-5 γ2-chainimmunoreactivity was detected in the BM region and occasionally in basal SE cells of the inflamedgingival tissue. MMP-2 and Ln-5 γ2-chain were found to colocalize in the BM area. In unaffectedareas of the gingival connective tissue, no CD45, MMP-2 or -9 immunoreactivity could be detected.

5.5. MMP-2 and MMP-9 mRNA expression in inflamed human gingival tissue (I & II)

MMP-2 mRNA expression was clearly induced in severly inflamed human gingival tissue in all fourpatients examined within the area of inflamed pocket epithelium showing local invasion into theconnective tissue stroma. The expression was observed in both basal and suprabasal cells. MMP-2 was less intense in SE of gingival specimens that showed a weaker inflammatory reaction. Inepithelium of weakly inflamed gingival epithelium that showed no growth into the subepithelialconnective tissue, MMP-2 was weakly induced, predominantly in the suprabasal cells. In additionto being expressed in the SE, MMP-2 mRNA was noted in some fibroblast-like cells. MMP-9mRNA was confined to inflammatory cells located beneath the SE in inflamed gingival tissue. Afew epithelial cells were also found to express MMP-9 mRNA in the inflamed area.

5.6. In situ gelatin zymography (II)

By in situ zymography, gelatinolytic activity was predominantly detected in the connective tissuedirectly beneath SE in both APMA-activated and native tissue samples of inflamed human gingiva.Neglible gelatinolytic activity was detected in the SE. Gelatinolytic activity increased temporallybeing readily detectable on day 2 and clearly increasing up to day 3. In samples incubated with100 µM CTT-peptide the intensity of gelatinolytic activity was clearly reduced or totally abolishedin both native and APMA-treated inflamed human gingival tissue. Weak gelatinolytic activity couldbe detected in clinically healthy gingival tissue and this was completely inhibited by 100 µM CTT-peptide. In the deeper layers of inflamed human gingival tissue, associated with less inflammatorycell infiltration, only sporadically distributed patches of gelatinolytic activity could be detected evenafter several days of incubation and 100 µM CTT-peptide effectively inhibited all gelatinolytic

60

activity. Also EDTA abolished all gelatinolytic activity in both native and APMA-treated samples.Samples incubated with 100 µM of the unrelated GACFSIAHECGA- and CERGGLETSC-peptidesshowed no inhibitory effect on gelatinolytic activity. Only minimal reduction of gelatinolytic activitypresent in inflamed human gingival tissue specimens incubated with 0.5 mM of the serineproteinase inhibitor Pefabloc was observed.

5.7. Immunolocalization of MMP-2 and Ln-5 in epithelial in vitro wounds (I)

Immunostaining demonstrated that MMP-2 was localized in specific streaks resembling ECMcontactlike structures at the ventral surface of the keratinocytes. These structures were presentonly in the migrating cells which were more numerous in TGFβ-treated cultures than in controlcultures. MMP-9 staining was diffuse, with no specific localization in the migrating or stationarycells. Ln-5 was present under the migrating as well as stationary keratinocytes throughout thecultures.

5.7.1. Controls for in situ hybridization and immunohistochemistry (I-IV)

Incubating samples for immunohistochemistry with nonrelevant IgG or omitting the primaryantibody resulted in no specific staining. Incubating samples for in situ hybridization withcorresponding sense probes resulted in no specific hybridization signal.

5.8. Effect of estrogen and CMT-8 in rat cutaneous wound healing (III & IV)

Histologic examination of the wounds showed re-epithelialization over the healing dermis in allexperimental groups 7 days postwounding. Only a minimal delay in re-epithelialization could bedetected (<10%) in some of the untreated OVX animals with a concomitant decrease in theamount of granulation tissue. Newly re-epithelialized wounds from OVX animals showeddetachment of epidermis from the underlying dermis in the epidermal-dermal junction. In OVXanimals treated with CMT-8, epidermal attachment was almost comparable to Sham-operated rats.Skin samples and sections from all experimental groups were prepared in the same manner,suggesting that the ruptures in the epidermal-dermal junctions were due to physical trauma ratherthan to mechanical stress during sample preparation.

5.8.1. Wound collagen content (III)

The wound collagen content in the 120 days post OVX rats was significantly reduced by 50% (p <0.05) compared to Sham and CMT-8 treated Sham as detected by hydroxyproline assay. Treatingthe rats with CMT-8 or estrogen significantly increased the type I collagen deposition (p < 0.05) inthe wound tissue comparable to that of Sham operated animals. Type I collagen immunoreactivitywas prominent in the wound bed between the two leading epithelial fronts of the regenerating day7 wounds in all samples and groups examined. Intense type I collagen immunoreactivity was alsoseen in endothelial layers. In all groups, type I collagen immunoreactivity was also found in thegranulation tissue beneath the newly organized epidermis juxtaposed to the intact dermis. Withinthe wound area, the intensity of immunoreactivity for type I collagen appeared the same in bothSham and OVX rats. CMT-8 treatment increased type I collagen in OVX rats, however, estrogenhad the most profound effect on type I collagen by increasing it within the wound area. Depositionof type I collagen in the re-epithelialized area was clearly less in the OVX group, compared to theCMT-8 treated OVX group.

5.8.2. Collagenase and gelatinase activity (III & IV)

Only APMA-activated samples demonstrated collagenolytic activity from day 7 cutaneous woundextracts. CMT-8 treatment of Sham rats decreased collagenolytic activity in comparison to Sham.

61

Collagenolytic activity in OVX wound samples was decreased compared to Sham while CMT-8 andestrogen treatment decreased collagenolytic activity further in the day 7 wound extracts.Using radiolabeled gelatinase assay, both total (+APMA) and active gelatinase activity werehighest in the Sham group when compared with the other groups in the day 7 wound tissueextracts. OVX reduced gelatinase activity and both Sham-operates treated with CMT-8 and OVXrats demonstrated even lower gelatinase activity. Estrogen-treatment reduced active gelatinase incomparison to OVX rats. Gelatin zymography from wound tissue extracts demonstrated thepresence of both pro-and active forms of MMP-9 and proforms of MMP-2. Gelatin zymographyshowed similar results for gelatinolytic activity as detected by the functional assay although CMT-8and estrogen treatment of OVX rats more markedly decreased gelatinolytic enzymes in thezymograms. Western immunoblotting for MMP-2 and -9 showed similar results.

5.8.3. Expression of collagenases in rat day 7 cutaneous wounds (III)

5.8.3.1. MMP-8 and MMP-13 protein and mRNA expression

MMP-8 immunoreactivity was localized to the supbrabasal layers of the proliferating woundepithelium in all groups examined. In Sham and CMT-8 treated OVX day 7 wounds, MMP-8immunoreactivity was also detected in inflammatory cells, including PMN neutrophils within thegranulation tissue beneath the proliferating wound epithelium and near the intact dermis.MMP-8 mRNA was expressed by cells in the proliferating and migratory wound epithelia, in dermalfibroblasts and inflammatory cells in both Sham and OVX rats. Less intense expression wasdetected in the wound epithelium of CMT-8 treated rats.MMP-13 mRNA was confined to basal cells in the migratory and proliferative epithelium anddecreased progressively away from the wound edge, being absent in the intact epidermis.Expression intensity was similar in all groups examined. However, MMP-13 mRNA expressingcells appeared more numerous in estrogen-treated OVX and in these wounds, MMP-13 mRNAwas also detected in monocyte/macrophage-like cells. MMP-13 protein within the wound area wasconfined to dermal fibroblasts, inflammatory and pericyte-like cells in the granulation tissue in frontor directly beneath the leading edge epithelial front. The immunohistochemical staining was verysimilar within all groups, except in the OVX wounds, where MMP-13 was almost undetectable. InSham and CMT-8 treated OVX-rats, MMP-13-positive inflammatory cells had an appearanceresembling either plasma cells or PMN-type cells, while other cells were more pericyte-like. In there-epithelialized area of day 7 wounds, MMP-13 was detected in fibroblast-and pericyte-like cellsas well as in inflammatory cells within the granulation tissue in both OVX and CMT-8 treated OVXrats. MMP-13 immunoreactive cells appeared more numerous in the CMT-8 treated OVX groupcompared to OVX.

5.8.3.2. Western immunoblotting of MMP-8 and MMP-13 (III)

Pro-and active forms of MMP-8 in skin wound extracts were reduced in both CMT-8 treated Shamand OVX rats when compared to Sham. MMP-8 pro-and active forms were further reduced inestrogen-and CMT-8 treated OVX wounds when compared to both OVX and Sham ones.Active and complexed forms but neglible levels of proMMP-13 were detected in skin woundextracts from all groups examined. Active forms of MMP-13 were reduced in both CMT-8 treatedSham and OVX rats when compared to Sham ones.

5.8.4. Expression of gelatinases and MMP-14 in rat day 7 cutaneous wounds (IV)

5.8.4.1. MMP-2

Immunoreactive MMP-2 protein expression was prominent in both wound keratinocytes of theregenerating epidermal edge and in the intact epidermis within all experimental groups. MMP-2immunoreactivity was also present in fibroblasts, endothelial cells as well as in inflammatory cells

62

in the deeper layers of the granulation tissue. By double immunostaining MMP-2 was found tocolocalize with immunoreactive MMP-8 protein in keratinocytes of the leading epidermal edge.MMP-2 mRNA was confined to basal cells of the regenerating epidermal edge and decreasedprogressively away from the wound edge, being absent in the intact epidermis in all groupsexamined.

5.8.4.2. MMP-9

MMP-9 protein expression was mainly localized to inflammatory cells and fibroblasts within thewound stroma. Weak MMP-9 protein expression was detected in basal keratinocytes within theleading edge of the regenerating epidermis. Most prominent expression was detected in Shamrats, where immunoreactive MMP-9 protein was confined to fibroblasts and inflammatory cellslocated in the wound bed in front of the leading epidermal edge and in the subepithelial layerjuxtaposed to the intact dermis. No clear differences in intensity or number of cells expressingMMP-2 or MMP-9 expression could be detected by histological techniques between the differenttreatment groups.

5.8.4.3. MMP-14

MMP-14 protein expression was detected in a few keratinocytes within the leading edge of theregenerating epidermis in all experimental groups. Additionally, immunoreactivity for MMP-14 wassporadically located to the BM region extending to a length of approximately three to five basalwound keratinocytes. Immunoreactivity for MMP-14 was very prominent and abundant infibroblasts and occasional inflammatory cells located in the granulation tissue. Most prominentimmunoreactivity for MMP-14 was detected in the subepithelial layer directly beneath the BM.When the wound area was studied, no differences in MMP-14 immunoreactivity between thedifferent experimental groups could be detected which was in line with Western immunoblot resultsfor MMP-14. In the re-epithelialized area, MMP-14 immunoreactivity was most prominent in thearea of epidermal-dermal splits present in untreated OVX wounds. In contrast, both MMP-14protein and epidermal-dermal splits were clearly reduced in the CMT-8 treated rats.

5.8.5. Basement membrane

5.8.5.1. Ln-5 γ2-chain expression (IV)

By in situ hybridization, Ln-5 γ2-chain mRNA expression was localized to keratinocytes in theleading edge of regenerating wound epidermis in day 7 wounds with less intense expression inOVX rats as compared to Sham. Ln-5 γ2-chain immunoreactivity was localized to the cytoplasm ofbasal wound keratinocytes and to the BM area in all experimental groups. The distribution of Ln-5γ2-chain immunoreactive protein in the BM area of OVX rats was clearly more diffuse anddiscontinuous when compared to Sham. In wounds from CMT-8 or estrogen-treated animals, Ln-5γ2-chain immunoreactivity appeared more prominent compared with both Sham and OVX rats.When the newly re-epithelialized area in day 7 wounds was studied, Ln-5 γ2-chainimmunoreactivity was clearly more prominent in apperance within the BM area of CMT-8 treatedrats when compared to OVX rats. Detachment of the epidermis colocalized with diffuse anddiscontinous expression of Ln-5 γ2-chain immunoreactivity in OVX rat wounds. Ln-5 γ2-chainimmunoreactivity appeared to be present on both the epidermal and the dermal side of the split,although most abundantly on the epidermal side.Semi-quantitative data on Ln-5 γ2-chain immunoreactive protein intensity in the BM area and thebasal keratinocytes showed that OVX rats had reduced Ln-5 γ2-chain total immunoreactivity whencompared to Sham rats. CMT-8 or estrogen-treated OVX rats exhibited higher levels of the Ln-5γ2-chain immunoreactivity in comparison to both Sham and OVX rats.

63

5.8.5.2. Ln-5 γ2-chain processing (IV)

In tissue extracts from day 7 wounds, weak immunoreactivity for the unprocessed 135 kDa Ln-5γ2-chain form was detected. Immunoreactivity for the processed, mature 100 kDa form of the Ln-5γ2-chain was detected in both Sham and CMT-8 treated Sham, but was reduced in OVX rats andtreating OVX rats with CMT-8 or estrogen further decreased the immunoreactivity. The 80 kDa Ln-5 γ2x-chain was readily detected in Sham as well as CMT-8 or estrogen-treated OVX rats, but wasalmost absent in CMT-8 treated Sham and totally absent in OVX.

5.9. Ln-5 γγγγ2-chain and ββββ-casein processing by MMPs in vitro (II & V)

Human recombinant autoactivated MMP-2 proteolytically processed the Ln-5 γ2-chain in vitro fromthe 100 kDa form to the 80 kDa Ln-5 γ2x-chain. 100 µM of CTT-peptide fully inhibited thisconversion. The efficacy of CTT-peptide inhibition was also confirmed by autoactivated MMP-2 andAPMA-activated MMP-9 cleavage of β-casein in vitro. The CTT-peptide dose-dependently inhibitedthe cleavage of the 21 kDa β-casein molecule by both gelatinases with an apparent IC50 of 25 µM.The unrelated GACFSIAHECGA or CERGGLETSC-peptides did not have any effect on MMP-2and –9 activity.The Ln-5 γ2-chain was cleaved by MMP-3, -8, -12, -13, -14 and –20 at indicated enzyme tosubstrate ratios (Table 5) as detected by Western immunoblotting with specific anti-Ln-5 γ2-chainantibody (Table 6). MMP-7 did not cleave the Ln-5 γ2-chain. Without MMPs, Ln-5 γ2-chain wasdetected as the 135 kDa and 100 kDa forms. MMPs increased the Ln-5 γ2-chain processing dose-dependently. Within 1 hour, MMP-20 processed the 140 kDa and 100 kDa forms of the Ln-5 γ2-chain into 80 kDa and 66 kDa forms. After 48 h of incubation with MMP-20, Ln-5 γ2-chain was fullyconverted into a 66 kDa form as well as to low molecular weight 30-40 kDa forms. Bothcollagenases MMP-8 and MMP-13 generated an 80 kDa Ln-5 γ2x-fragment within 1 h of incubationand after 3 h the 66 kDa form was clearly detectable. After 24 h of incubation, the 140 kDa and 100kDa forms of the Ln-5 γ2-chain were almost fully processed into the 80 and 66 kDa forms and the80 kDa Ln-5 γ2x-chain was continuously further processed to the 66 kDa form. MMP-13 furthergenerated 20-40 kDa low molecular weight Ln-5 γ2-fragments. MMP-2, -3, -12 and –14 processedthe Ln-5 γ2-chain solely into an 80 kDa form.Ln-5 γ2-chain was incubated with MMP-8, -13 and -20 in the presence or absence of 50 µMMarimastat which fully inhibited MMP processing of the Ln-5 γ2-chain.

5.9.1. N-terminal sequences of MMP cleaved Ln-5 γγγγ2-chain fragments (V)

The N-termini of MMP-3, -12, -13 and –20 processed 80 kDa Ln-5 γ2x-fragment wereLTSCPACYNQ and the cleavage occurred between Ala586 and Leu587 (Table 8). The 80 kDa Ln-5γ2x-fragment produced by MMP-8 was one amino acid shorter than the 80 kDa form produced byMMP-3, -12, -13 or –20 being TSCPACYNQV and the cleavage occurred between Leu587 andThr588. The 100 kDa forms of Ln-5 γ2-chain were sequenced after MMP-12 or -14 treatment andthe N-terminal sequences of both these polypeptides were almost identical, DENPDIE(A/E)(S/A)D,the cleavage site in both being Gly413-Asp414. A schematic diagram of MMP-mediated cleavages ofthe Ln-5 γ2-chain is presented in Figure 9.

64

TABLE 8. N-terminal sequences of MMP-cleaved laminin-5 γγγγ2-chain fragments.

MMP Size (kDa) N-terminus Cleavage site in the Ln-5 γγγγ2-chainMMP-3 80 LTSCPACYNQ Ala586 – Leu587

MMP-8 80 TSCPACYNQV Leu587 – Thr588

MMP-12 80 LTSCPACYNQ Ala586 – Leu587

100 DENPDIEASD Gly413 – Asp414

MMP-13 80 LTSCPACYNQ Ala586 – Leu587

MMP-14 100 DENPDIEEAD Gly413 – Asp414

MMP-20 80 LTSCPACYNQ Ala586 – Leu587

The amino acid cysteine (C) was not recognized by automatic microsequencing and wasdeduced from cDNA.

5.9.2. Cell migration over MMP cleaved Ln-5

Added MMP-3, -13, -14 or –20 at final concentrations of 10 nM significantly induced migration ofMCF-7 breast carcinoma cells over Ln-5 substrate. Of the five MMPs tested, MMP-14 was mosteffective in inducing migration of MCF cells over Ln-5 substrate and also MMP-3, -13 and -20clearly induced migration of MCF-7 cells in relation to PBS-treated controls, while MMP-8 did not.MCF-7 cells do not synthesize any MMPs in detectable amounts (Giambernardi et al. 1998). MCF-7 cells plated on filters lacking Ln-5 did not show any migration.

Figure 9. The structure of the heterotrimeric Ln-5 molecule. The three chains (αααα3, ββββ3 and γγγγ2) areconnected by disulphide bonding. MMP cleavage sites in the Ln-5 γγγγ2-chain are indicated with arrows.Arabic letters represent the different domains within the laminin chains.

65

6. Discussion

6.1. MMP inhibition and HMK cell proliferation

MMP inhibitors can have multiple effects on cells, affecting MMP expression, secretion, activationand catalytic competence as well as cell proliferation and migration. HMK cell growth was reducedmost by CMT-3 and -8 and less by CMT-1 while CMT-5 did not have any effect. These results arein line with a study with human and rat prostate cancer cell lines, where all CMTs (except CMT-5)and doxycycline were found cytotoxic (Lokeshwar et al. 2002). Batimastat or the LWs on the otherhand, did not show any cell proliferation inhibition effects. Batimastat is designed to chelate theZn2+-binding groups in MMPs and thus may not interfer with other molecules crucial for cellviability. CMTs bind Ca2+ and are thus likely to affect intracellular Ca2+-levels resulting incytotoxicity also independent of MMP inhibition. In this aspect, CMTs have been shown to interferewith Ca2+-receptors in bone and interfere with the osteoclast function (Greenwald et al. 1998). Cellproliferative inhibition was independent of TNF-α treatment indicating that the CMTs would actmore direct on cellular mitogenic mechanisms.

6.2. MMP and Ln-5 γγγγ2-chain expression in HMKs and inflamed gingival tissue

MMP-2 was the primary gelatinase secreted by the HMKs while MMP-9, but not MMP-2 synthesiswas induced by TNF-α in line with previous results (Mäkelä et al. 1998, Salo et al. 1994). MMP-2was instead induced by TGF-β treatment and more importantly, MMP-2 was activated. MMP-9production was increased upon TGF-β treatment but activation did not increase. Both MMP-2 andMMP-9 expression have been previously found to increase upon TGF-β treatment of oral mucosalkeratinocytes (Salo et al. 1994). MMP-2 expression by the cultured HMKs could be confirmed invivo in human gingival tissue epithelial cells by both immunohistochemistry and in situhybridization. MMP-2 expression was most dramatic in inflamed human gingival samples. Thisindicates that while oral mucosal keratinocytes may express baseline levels of MMP-2, expressionis induced upon stimulus such as inflammatory cytokines and in response to LPS, which is aproduct of periopathogenic bacteria in gingival inflammation (Van den Steen et al. 2002).Interestingly, it has been shown that periodontal pathogens such as Porphyromonas Gingivalisinhibit migration of HMKs on Ln-5 through proteolysis of FC proteins. This may impair keratinocytemotility and subsequently the wound healing process associated with periodontitis (Hintermann etal. 2002). In addition, when cultured human gingival fibroblasts were incubated withperiopathogenic bacteria this resulted in an increase in MMP-2 expression (Dahan et al. 2001) andMMP-2 has been found to dampen inflammation by cleaving monocyte chemoattractant protein-3(McQuibban et al. 2000). Thus, MMP-2 could be involved in host defense during inflammation.Protrusions of SE into the gingival connective tissue is typically associated with periodontalinflammation which could be associated with MMP-2 action and possible cleavage of the Ln-5 γ2-chain in the BM area subsequently inducing the benign invasive phenotype of the SE cells.We detected MMP-9 mRNA and protein expression mainly in inflammatory cells (CD45-positivecells) of inflamed human gingival tissue with minimal expression in basal SE cells in line withpreviously published work (Westerlund et al. 1996). Since MMP-9 expression is induced by TNF-αin both oral mucosal and skin keratinocyte cultures, the stromal cells may orchestrate theexpression of MMP-9 in different tissues in vivo. In this context, TGF-β has been found to induceMMP-9 expression in human skin explants and in the presence of TNF-α, proMMP-9 wasproteolytically activated. No similar effect was found with isolated fibroblasts or keratinocytes (Hanet al. 2001). In comparison, MMP-9 is prominently expressed in migrating skin woundkeratinocytes, while MMP-2 is associated with more constitutive expression (Okada et al. 1997).Thus, both gelatinases are most likely subject to tissue specific regulation.By in situ gelatin zymography, gelatinolytic activity was detected in the stroma beneath the SE ininflamed human peridontal tissue in line with previous reports (Korostoff et al. 2000). By using theMMP-2 and -9 specific inhibitor, the CTT-peptide, we could demonstrate that most gelatinolyticactivity in inflamed human periodontal tissue was accounted for by MMP-2 and -9. The CTT-

66

peptide may thus be very useful in the qualitative analysis of in situ gelatin zymography. In situzymography also demonstrated that MMP-2 and -9 mediated gelatin degradation in inflamedhuman periodontal tissue was clearly increased when compared to periodontal tissue of healthyhumans. This is in line with previous reports (Korostoff et al. 2000, Westerlund et al. 1996).Ln-5 γ2-chain was confined to the BM region of inflamed gingival SE, in line with previous reports(Hormia et al. 1998, Uitto et al. 1998). Intracellular Ln-5 and MMP-13 expression in inflamedhuman gingiva has been found to correlate and be co-expressed in cells by immunohistochemistry(Uitto et al. 1998). In our study, we found that MMP-2 and the Ln-5 γ2-chain co-localized in the BMregion of inflamed human gingiva. MMP-2 has previously been shown to cleave the Ln-5 γ2-chain(Giannelli et al. 1997) and we show that MMP-13 cleaves the Ln-5 γ2-chain within the same site.While immunohistochemistry showed that MMP-9 appears to be the main gelatinase in the stromalcompartment associated with CD-45 positive inflammatory cells, MMP-2 may instead beassociated with actions of the gingival keratinocytes.

6.3. MMP inhibition in HMKs and inflamed periodontal tissue

Batimastat and CMTs are effective inhibitors of MMP-2 and -9 activity in vitro (Coussens et al.2002, Hidalgo and Eckhardt 2001, Rasmussen and McCann 1997, Rifkin et al. 1993). Batimastatdid not affect synthesis of the two gelatinases further confirming that Batimastat mostly acts onMMP activity, not transcription. MMP-2 production was reduced by the CMTs, CMT-3 and –8 beingmost effective, while TNFα-induced MMP-9 synthesis increased after addition of 10-20 µM CMT-3.Previous studies with both melanoma and prostate cancer cell lines demonstrated effectivedownregulation of both MMP-2 and -9 expression and secretion in culture media (Lokeshwar et al.2002, Seftor et al. 2002). However, MMP inhibitors have also been found to increase MMPsynthesis during acute wound healing (Ågren et al. 2001, Lund et al. 1999). Thus, non-tumorigeniccells most likely respond differentially to MMPIs than aggressive tumor cells or cells associatedwith chronic inflammation. In addition, intracellular Ca2+-levels affect MMP expression differentiallythrough secondary messengers, which is probably cell type specific. TC can prevent the proteolyticcleavage of membrane-bound proTNF-α into its active, soluble form in human LPS-treatedmonocytes without affecting TNF-α synthesis (Ramamurthy et al. 2002). This indicates that TC-based MMP inhibition affect not only MMPs but also cytokines involved in MMP induction andrepression cascades.Our results are the first to show that the CTT-peptide effectively inhibits in vivo gelatinolytic activityassociated with chronic inflammation. The CTT-peptide has previously been shown to inhibit tumorcell migration, MMP-2 and -9 activity in vitro and angiogenesis (Koivunen et al. 1999). LDD(Periostat) has been successfully used in the treatment of periodontal disease in the USA andinhibits both collagenases and gelatinases effectively (Golub et al. 1998). Also the CMTs arepotent inhibitors of tissue destruction and collagenase activity in experimental periodontitis(Ramamurthy et al. 2002, Yu et al. 1993). Taking that the CTT-peptide inhibits MMP-2 and -9 withIC50 of 10 µM, it may well be of potential use in the future treatment of chronic inflammation. Inaddition, both doxycycline and CMTs are cytotoxic, whereas the CTT-peptide is not. Batimastat isnot cytotoxic, however, this broad-spectrum MMPI causes unwanted side-effects. Although manychronic inflammatory diseases have been associated with elevated MMP activity, there are stillproblems analyzing the efficacy of MMP inhibition in vivo. These problems include samplingrelevant tissues and biologic fluids (e.g. rheumatoid synovium and the adjacent synovial fluid) todemonstrate MMP inhibition and subsequent ECM protection at the lesion site. In this regard,periodontal disease offers the advantages of accessible tissue and adjacent easily-sampled non-invasive biologic fluids; i.e. GCF, mouthrinse or saliva and well documented clinical parameters ofdisease severity (Golub et al. 1997, Mäntylä et al. 2003). This may, at least in part, explain whyLDD is currently the only MMP-inhibition based drug on the market, being targeted for adjunctivetreatment for periodontal disease, acne and rosecea.

67

6.4. Cutaneous wound healing

Wound healing and estrogen-deficiency results in many of the features associated with chronicinflammation. In our study, we did not find any differences in re-epithelialization rates in contrast tosome previous studies (Ashcroft et al. 1997a, 1997b, 1999). However, in these studies, animalsonly 18 days post-OVX were used. In the animal model for estrogen-deficiency used herein, ratswere kept ovariectomized 120 days before wounding. It has been shown that in women,chronological age is not the major factor influencing the impact of estrogen-deficiency in skin,rather time after menopause (Brincat et al.1985, 1987a, 1987b). Women are not trulyhypoestrogenic until 2 years post menopause. Many of the animal models for estrogen-deprivationpresent contradictory results. This is due to the use of different species and gender of animals,various types of administered hormones, routes of administration, dosage and frequencies of thehormone, duration of estrogen-deprivation, diversity of methods assessing wound healing and awide variety of tissues and cell culture in which the healing process is studied (Calvin 2000). Touse an animal model for investigating estrogen-deficiency and therapies for clinical use on human,eventually this field needs a more standardized animal model.In our study, wound collagen content was significantly decreased in OVX animals and treatmentwith CMT-8 or estrogen significantly increased wound tissue collagen content. These results are inline with previous human studies (Ashcroft et al. 1997a, 1997b, 1999). A decrease in woundcollagen content in OVX animals could imply aberrant collagen turnover as seen in diabetic ratsand concomitantly reduced wound strength and quality. Estrogen has been shown to dampen theinflammatory response and neutrophil chemotaxis thus leading to decreased collagen degradation(Ashcroft et al. 1999). Our study is the first to demonstrate the effect of CMTs on cutaneous woundhealing associated with estrogen-deprivation. CMT-1 has previously been shown to inhibit bodyweight loss associated with a reduction in skin collagen in diabetic rats without affectinghyperglycemia. CMT-1 stimulated collagen protein synthesis in diabetic animals through increasedsteady-state levels of pro α1(I) mRNA collagen, however, CMT-1 did not increase mRNA levels innon-diabetic rats (Craig et al. 1998, Yu et al. 1993).The wounds from the OVX animals showed one thing in common, abnormal attachment of theepidermis to the dermis. This was clearly seen in the areas where re-epithelialization wascomplete. Junctional detachment of the epidermis from the dermis is associated with severalblistering diseases, where BM molecules such as Ln-5 are targeted or depleted. Wounds fromestrogen or CMT-8 treated animals showed a similar epidermal-dermal junction as seen in Shamoperates. In this context, anti-inflammatory properties of TCs have been useful in the treatment ofspecific cutaneous diseases, including EB and bullous pemphigoid (Loo et al. 2001).Immunostaining for Ln-5 γ2-chain in the wounds demonstrated a more discontinuous staining andless Ln-5 γ2-chain mRNA in the BM area in OVX rats when compared to the other treatmentgroups. In this regard, skin BM type IV collagen decreases with age (Vazquez et al. 1996). Thus,while estrogen-deficiency eventually leads to wound closure, delayed wound healing may resultfrom either reduced attachment of the newly formed epidermis to the underlying dermis decreasingwound strength and/or aberrant collagen turnover resulting in delayed granulation tissue turnoverand concomitant wound contraction. In line with this, wound contraction in rats 4 months post OVXwas delayed as compared to Sham operates (Calvin et al. 1998). Overall, macroscopic woundhealing was enhanced in treated OVX animals in comparison to untreated OVX animals.One interesting observation in our study was the MMP-8 mRNA expression in the woundkeratinocytes. MMP-8 has previously only been found in dermal fibroblasts and infiltratingneutrophils (Fisher et al. 2001, Hanemaaijer et al. 1997). MMP-8 expression has been detected inkeratinocytes by RT-PCR (Bachmeier et al. 2000b), however, no study thus far has presented withany in situ hybridization data from wound tissues. In other studies, MMP-8 mRNA expression hasbeen found in plasma cells of oral cysts (Wahlgren et al. 2001), bronchial epithelial cells (Prikk etal. 2001), oral SCC cells (Moilanen et al. 2002) and gingival SE cells (Tervahartiala et al. 2000).However, we could not detect MMP-8 protein other than in a few scattered keratinocytes of thesuprabasal epithelial layer in the wound. It thus appears that studies investigating thetranscriptional control and especially the stability of the MMP-8 mRNA transcript would be

68

necessary. Indeed, also other investigators have noted that MMP RNA and protein profilescorrelate poorly during murine wound healing which has been attributed to differences in stabilityand degradation rates between mRNA and protein (Wall et al. 2002).In our study, MMP-8 was also immunohistochemically detected within inflammatory cells in thegranulation tissue. Overall, these results point to a possbility that MMP-8 could participate incollagen turnover and clearance during murine wound healing. We found active forms of MMP-8 inwound tissue extracts from all experimental groups. In humans, MMP-8 levels have been found tobe severalfold higher than MMP-1 in wound fluids from both acute and chronic wounds. MMP-8 isalso significantly elevated in chronic wound fluids when compared to acute wound fluids (Nwomehet al. 1999, Nwomeh et al. 1998a). MMP-8 has also been found to cleave tissue factor pathwayinhibitor involved in regulating coagulation cascades upon vascular injury (Cunningham et al.2002). Increased MMP-8 activity may thus also lead to coagulation disorders in chronic wounds. Inthis regard, we found that CMT-8 decreased MMP-8 mRNA expression in the wound keratinocyteswhen compared to OVX. Also Western immunoblotting demonstrated that CMT-8 and estrogen-treatment reduced both pro-and active forms of MMP-8. Recently, MMP-8 expression was found tobe under hormonal control. Male MMP-8 knock-out mice was more suspectible to cancer thanfemale mice (Balbin et al. 2003).If ignoring the fact that rodents possess two MMP-1 genes and old-fashionably referring to MMP-13 as the murine orthologue for human MMP-1, our results are in line with previous studies (Lundet al. 1999, Madlener 1998, Okada et al. 1997). MMP-13 has been suggested to be the maincollagenase responsible for collagen degradation in murine keratinocytes similar to MMP-1 inhumans (Netzel-Arnett et al. 2002). We also found active forms of MMP-13 in wound tissueextracts evidencing that MMP-13 participate in matrix turnover during murine wound healing. MMP-9 protein was only detected at low levels and mostly in inflammatory cells in day 7 wounds. Thismay be due to the fact that others have reported MMP-9 expression during acute murine woundhealing to be maximal at day 5 and thereafter it ceases rapidly (Lund et al. 1999, Madlener 1998,Okada et al. 1997).Interestingly, MMP-2 mRNA and protein was detected in the leading edge keratinocytes within allgroups in addition to fibroblasts. MMP-2 mRNA expression has been mostly reported to beconfined to dermal fibroblasts (Lund et al. 1999, Madlener 1998, Oikarinen et al. 1993, Okada et al.1997, Salo et al. 1994). Some reports exist on MMP-2 protein expression in wound epidermal cells(Oikarinen et al. 1993, Salo et al. 1994), normal skin epidermal cells (Dumas et al. 1999), andcultured skin keratinocytes secreting MMP-2 (Bachmeier et al. 2000a, Kobayashi et al. 1998).However, MMP-2 expression was confined to the basal keratinocytes of intact skin in elderlypersons (Ashcroft et al. 1997c). We also detected MMP-2 expression in SE cells of inflamedhuman periodontal tissue (Giambernardi et al. 1998) and bronchial epithelial cells (Maisi et al.2002). In addition, in the wound scratch assay presented herein, HMKs expressed mainly MMP-2and expression was induced after TGF-β treatment.In the rat wounds, MMP-2 mRNA was expressed with a similar intensitry in all groups examined. Ina study with diabetic mice and in line with our data on gelatinolytic activity for OVX rats, diabeticmice actually showed lower levels of proMMP-2 in wound tissue extracts during early woundhealing stages (Wall et al. 2002). Since our and other studies (Giannelli et al. 1997) demonstratethat MMP-2 is directly involved in inducing cell migration reduced MMP-2 production could beassociated with delayed re-epithelialization in certain situations. Estrogen was found to increaseMMP-2 production in suction blister fluids from patients receiving systemic estrogen therapy(Haapasaari et al. 1997).Abundant MMP-14 expression in fibroblasts are in line with previous studies (Lund et al. 1999,Madlener 1998, Okada et al. 1997). We did, however, also find MMP-14 immunoreactivityassociated with a few basal keratinocytes and in the BM region. A recent study demonstrates thatprimary murine keratinocytes indeed synthesize MMP-14 in line with our findings (Netzel-Arnett etal. 2002). Most intense MMP-14 immunoreactivity was found in wounds from the OVX animalsconfined to the BM area with epidermal-dermal detachment. MMP-14 is capable of cleaving theBM component Ln-5 γ2-chain and was co-localized with the Ln-5 γ2-chain in the BM area. Thismay indicate that MMP-14 is associated with degradation of BM molecules in the OVX animals. In

69

line with this, Ln-5 γ2-chain staining in the BM area correlated well with the junctional epidermal-dermal split, being weakest in the OVX rats. Western immunoblot data presented interestingresults; the 80 kDa Ln-5 γ2x-chain was clearly present in wound tissue extracts from Sham-operates, but was lacking in OVX wound tissue extracts. Interestingly, both CMT-8 and estrogen-treatment resulted in the neoexistence of the Ln-5 γ2x-chain. However, treating Sham-operateswith CMT-8 also nearly fully abolished the Ln-5 γ2x-chain. It therefore appears that while treatmentof estrogen-deprived wounds was beneficial in terms of Ln-5 γ2-chain processing, treating Shamoperates with CMT-8 may not be. Our study is novel in presenting the effect of CMTs and estrogenon Ln-5 γ2-chain expression and processing in cutaneous wound healing. Ln-5 deposition is criticalfor wound keratinocytes and Ln-5 neosynthesis starts within hours of wounding (Larjava et al.1993, Nguyen et al. 2000b). Upon re-epithelialization, hemidesmosomal Ln-5 is critical for theintegrity of the epidermal-dermal junction and stability of the regenerated skin. Clearly, the impactof estrogen and MMPIs on Ln-5 γ2-chain with focus on wound healing and inflammation deservesfurther studies in the future.

6.5. Therapeutic aspects of estrogen and MMPIs in cuteneous wound healing

Chronic wound therapy lacks specific treatment, in large part due to the lack of knowledge of themolecular abnormalities within the wound that prevent it from healing (Chen et al. 1999). Acutewound healing is a precisely defined orchestra of spatially and temporally regulated proteaseactivity. The use of an MMPI for the treatment of acute wound healing is not necessary, since itmay interfere with the tightly controlled proteolytic events and Ca2+-levels essential for woundhealing. Upon damage, the leading edge keratinocytes become transiently leaky to Ca2+ leading toAP-1 activation in cells of the wound margin (Jacinto et al. 2001). AP-1 is involved in the inductionof several MMPs and Ln-5 expression and possibly cell migration on Ln-5 (Kähäri and Saarialho-Kere 1999, Olsen et al. 2000, Salo et al. 1999). Interference with the intracellular Ca2+-balance withan MMPI may thus be detrimental for acute wound healing. However, chronic wound healing hasbeen associated with increased Ca2+-levels (Sank et al. 1989) which may further add to thebenefits of using MMPIs for chronic wound healing. In this regard, CMT-1 was found to normalizecollagen-synthesis in diabetic rats without affecting collagen-synthesis in non-diabetic rats (Craig etal. 1998, Yu et al. 1993).Estrogen has been shown to improve acute wound healing in terms of re-epithelialization andscarring (Ashcroft et al. 1997a). HRT for women past menopause can reverse changes associatedwith skin fragility and delayed wound healing, however, HRT is also associated with an increasedrisk for breast cancer. Thus, the use of locally applied estrogen for acute wound treatment andmost importantly for treatment of chronic wounds could be beneficial. In a chronic wound, not onlyMMPs are upregulated, TIMP activity is either unchanged or reduced which leads to uncontrolledMMP activity. At present there are no studies addressing the effect of estrogen on TIMPexpression in cutaneous wounds or skin cells. Estrogen do modulate TIMP production in other celltypes and CMT-3 decreases TIMP-1 and –2 production in prostate cancer cell lines (Lokeshwar etal. 2002).The complex pattern of MMP expression and activity in chronic inflammations and wound healingmay be the result of an altered balance in matrix deposition and degradation. Impaired ECMturnover or misdirected ECM deposition may result in excessive MMP activity leading touncontrolled matrix degradation (Stetler-Stevenson 1996). Thus, MMPs may be either upregulatedor downregulated in chronic inflammations depending on signals mediated by the ECM. We haveshown that the use of an MMPI for the treatment of wounds associated with chronicity can improvewound healing and wound integrity.

6.6. Cell migration and Ln-5 γγγγ2-chain processing

The Ln-5 γ2-chain was processed by MMP-3, -8, -12, -13 and -20 in addition to the previouslydescribed MMP-2 and -14 (Giannelli et al. 1997, Koshikawa et al. 2000). Cleavage of the Ln-5 γ2-chain resulted in the formation of the 80 kDa Ln-5 γ2x-chain. N-terminal sequencing of the 80 kDa

70

Ln-5 γ2x-chain revealed that the cleavage site for MMP-3, -12, -13, and –20 was identical to thepreviously described cleavage site for MMP-2 (Giannelli et al. 1997). However, the cleavage sitefor MMP-8 in the Ln-5 γ2-chain differed by one amino acid. The cleavage by MMP-12 and -14resulting in processing of the 135 kDa Ln-5 γ2-chain to the 100 kDa form was identical to thepreviously described physiological cleavage site (Amano et al. 2000, Gagnoux-Palacios et al.2001, Schenk et al. 2003, Veitch et al. 2002) and this cleavage site is conserved within humansand rodents (Sasaki et al. 2001). MMP-14 cleaves the human Ln-5 γ2-chain in vitro generatingfragments of roughly 90 kDa and 70 kDa (Gilles et al. 2001). These results contrast with anotherstudy where MMP-2 and -14 did not cleave the human Ln-5 γ2-chain (Veitch et al. 2002). However,in the latter study, immunoprecipitated Ln-5 was used as a substrate in contrast to detecting thenative Ln-5 γ2-chain (Gilles et al. 2001). Although the cleavage site for the 80 kDa Ln-5 γ2-chain inhumans and rodents is not conserved, the cleavage sites contain amino acids of the same polarityin both the human and rodent Ln-5 γ2-chain (personal communication with Dr.Vito Quaranta).Human Ln-5 γ2-chain is also more resistant to MMP proteolysis, another reason why cleaved Ln-5γ2-chain is more seldom detected in media from cultured human cells or tissues (personalcommunication with Dr.Naohiko Koshikawa). In addition, MMPs show divergence in the cleavagesites within various substrates (Netzel-Arnett et al. 1993). Thus, it appears that conformationalrestrictions within biological molecules determine the access of MMPs to cleave a certainmolecule.Interestingly, while MMP-3, -13, -14 and -20 induced MCF-7 cell migration, MMP-8 did not. In onestudy, the expression of MMP-8 was associated with a non-invasive phenotype of breastcarcinoma cells while perturbing MMP-8 expression led to a more invasive phenotype (Agarwal etal. 2003). Further studies will confirm how critical the one amino acid difference in reality is. MCF-7cells express neglible levels of MMPs (Giambernardi et al. 1998) and so MCF-7 cell migration isonly induced after exogenous addition of MMPs. However, it has been shown that constitutiveHT1080 and BRL cell migration on Ln-5 is mediated by endogenously produced MMP-2 and –14cleavage of the Ln-5 γ2-chain (Koshikawa et al. 2000). In addition, A549 lung carcinoma cells usedin this study migrated on Ln-5 and type IV collagen and were found to secrete MMP-2, -3, -13, -14and -20 all of which also efficiently induced MCF-7 cell migration on Ln-5.The domainIII EGF-like fragment released by MMP-2 cleavage of the Ln-5 γ2-chain from 100 kDato 80 kDa exhibits promigratory properties, binds to the cell surface EGF receptor and increasesMMP-2 expression (Schenk et al. 2003). In line with this, also selective EGF-like repeats in theTenascin C molecule can bind and activate the EGF receptor (Swindle et al. 2001). The Ln-5 γ2-chain domainIII/EGF-like fragment could thus potentially act in a paracrine manner binding to manyneighbouring cells and induce migration of a larger cell cluster which can be seen in certain typesof cancer. Highly invasive melanoma cells for example are typically migrating as larger clustersand produce MMP-2 and the Ln-5 γ2x-chain is present in the culture media (Seftor et al. 2002).Why are so many MMPs processing the Ln-5 γ2-chain? It makes sense in away. MMP expressionis often spatially and temporally restricted. For any mechanism to be used in the whole organism,several pathways must exist. One excellent example is the co-localization of MMP-20 and Ln-5 inameloblasts of the developing tooth (Bartlett et al. 1998). Ln-5 is also the major component of theinternal BM facing the tooth enamel surface in adult tooth (Hormia et al. 1998). Patients withjunctional EB have hypoplastic defects in their enamel, which may be caused by detachment of theameloblast layer from the enamel surface (Kainulainen et al. 1998). We and others have alsodemonstrated co-localization of MMP-2 and MMP-13 with Ln-5 γ2-chain in inflamed periodontaltissues (Uitto et al. 1998) and MMP-14 and Ln-5 γ2-chain in the BM area within cutaneous wounds.Our study with HMKs demonstrated that TGF-β-induced HMK cell migration was dependent onMMP-2 expression, secretion and activity along with Ln-5 deposition, in line with previous studies(Kainulainen et al. 1998). In summary, all the MMPs that have been demonstrated here to cleavethe Ln-5 γ2-chain are present during remodelling processes and wound healing as well as incancers (see Review of the literature).Careful interpretations should always be made when comparing results between cell lines, whichdiffer in tumorigenicity, integrin expression, MMP expression and the capability of forming HDs.

71

The Ln-5 γ2x-chain is rarely found in culture media of non-migrating NHKs (Gagnoux-Palacios etal. 2001, Gilles et al. 2001). In a study with nonmigratory HCC cells on Ln-5, MMP-2 activated thelatent TGF-β1 secreted by the HCC cells which stimulated α3-integrin expression subsequentlymodulating cell migration over Ln-5. This emphasizes that cells that do not secrete MMP-2 couldpotentially become invasive or activated because of MMP-expression by surrounding stromal cells(Giannelli et al. 2001, Giannelli et al. 2002). In contrast, the cleaved 80 kDa Ln-5 γ2x-chain isclearly detected in culture media of highly aggressive melanoma cells correlating with their abilityto invade (Seftor et al. 2002). While not many studies have addressed the Ln-5 processing in vivoyet, it is known that in quiescent human tissues, the Ln-5 γ2-chain is mostly present in the 105 kDaprocessed form and the Ln α3-chain is always present in its processed form (Marinkovich et al.1992). MMP-2 was found to modulate the processing of the Ln-5 γ2-chain to the Ln-5 γ2x-chain inmammary tissue from sexually mature, pregnant or involuting rats but not in sexually immarturerats. When sexually immature rats were injected with estrogen, MMP-2 expression was inducedand the Ln-5 γ2-chain processed (Giannelli et al. 1999). In our study, the Ln-5 γ2x-chain wasclearly detected in the wound tissue extracts, except for the OVX group. Thus, the furtherprocessing of the Ln-5 γ2-chain is apparently associated with remodelling, migration orpathological situations in vivo and should be studied under such conditions.Normal cell migration differs from tumor cell invasion only in terms of regulation. Any epithelial cellmaking its way through the connective tissue during normal remodelling uses the samemechanisms as an invading tumor cell. The tumor cell just don´t have a brake anymore. Inaddition, it makes sure that the brakes can´t be fixed by producing molecules that accelerate itsway through the matrix. In normal cells, Ln-5 is assembled intracellulary and secreted as aheterotrimer without continuously ongoing synthesis. In contrast, carcinoma cells synthesize andsecrete monomeric Ln-5 γ2-chain, which is detected in the leading edge of tumors (Koshikawa etal. 1999, Olsen et al. 2000). Hepatocyte growth factor and TGF-β1 synergistically upregulate theLn-5 γ2-chain expression but not Ln α3 or β3 chain expression in cultured carcinoma cells (Olsenet al. 2000). The Ln-5 γ2-chain expressed by tumor cells may not be fully identical in form and/orfunction to the form synthesized by normal cells. CD44, fibronectin and tenascin are found in tumortissues in alternatively spliced forms most likely promoting tumor growth and invasion (Stetler-Stevenson 1996). In addition, MT1-MMP and Ln-5 γ2-chain expression in the invasive front ofcolon carcinomas is upregulated by the nuclear β−catenin/T-cell factor-4 transcription complex(Hlubek et al. 2001, Takahashi et al. 2002). Thus, the Ln-5 γ2-chain appears to be one of thosemolecules which carcinoma cells may use for accelerating their way through matrix.However, if carcinoma cells can secrete the monomeric γ2-chain, when and how is this monomerprocessed? One possibility is that Ln-5 and its processing is associated with FCs during migration.It would make sense in that the migratory phenotype in cells is associated with the formation anddissociation of FCs which would include Ln-5 γ2-chain deposition and processing. If so, Ln-5 wouldalso be perfectly located for focalized proteolytic processing by the MMPs generating apromigratory Ln-5 γ2-fragment and inducing migration. In our study with HMKs, MMP-2 and Ln-5were detected in specific streaks resembling ECM contactlike structures. It also makes sense dueto the observation that Ln-5 is neosynthesized by regenerating epidermal cells residing on apartially intact BM, such as in suction blisters (Kainulainen et al. 1998), where Ln-5 is available andstill the epidermal keratinocytes produce and deposit novel Ln-5 for some purpose. Furthermore,tumor cells may have changed their ability to take advantage of the promigratory role of the Ln-5γ2-fragment and exclusively synthesize the Ln-5 γ2-chain in the FCs of invading cells, excludingthe stop signal which may reside in the ability of the cells coming from behind to form stableadhesion complexes, i.e. HDs. Indeed, highly invasive tumor cells do not generally form HDs andLn-5 γ2-chain synthesis is mostly associated with the invasive front cells in tumors (Hao et al.1996).Strikingly, blocking the Ln-5 α3-chain does not block HCC cell adhesion to either uncleaved orMMP-2-cleaved Ln-5 but instead efficiently blocks cell migration on MMP-2 cleaved Ln-5 γ2-chain(Giannelli et al. 2001, Giannelli et al. 1997). In this regard, it makes even more sense that tumorcells would synthesize monomeric γ2-chain, getting rid of one possible control mechanism, i.e. the

72

α3-chain. In this context, the Ln α3-chain is known to be the rate-limiting factor in Ln-5 assemblyand secretion. Thus, the spatially and timely directed processing of the Ln-5 α3-chain and γ2-chainby different proteinases could be one means by which cells control migration. Based on resultspresented herein and in other studies (Schenk et al. 2003), MMP cleavage of the Ln-5 γ2-chainresults in a promigratory capable of inducing migration (A schematic summary is presented inFigure 10).

6.7. MMP inhibition-last aspects

MMPI development for clinical use in treatment of human cancers and chronic inflammation shouldby no means be buried just because of an initial setback. As shown in this study and by manyothers (Hidalgo and Eckhardt 2001), MMPIs have potential in both cancer and chronicinflammatory diseases treatment. Many of the MMP-knockouts have demonstrated resistance toseveral diseases, for example to endotoxin-induced septic chock and to experimental bullouspemphigoid in MMP-9 deficient mice (Van den Steen et al. 2002), which should be enough toattract the pharmacological development of MMPIs. However, to develop effective, target-specificMMPIs, the MMPs being expressed and activated in a certain type of cancer at certain stages or ina particular chronic inflammatory environment must be carefully determined. There are many new,exciting techniques emerging for this purpose (Lopez-Otin and Overall 2002). In addition, we mustalso determine the true function of MMPs in pathological conditions. MT1-MMP activity was foundto reduce tumor cell migration due to the fact that excess MT1-MMP activity degraded thesubstrate critical for tumor cell migration (Deryugina et al. 2003). However, MT1-MMP activity alsoenhances tumor cell proliferation in 3D type I collagen gels in comparison to MT1-MMP deficienttumor cells and this can be inhibited by MMPIs (Hotary et al. 2003). Thus, the true meaning ofMMP activity in different disease states in vivo must be resolved.There are several issues that need to be resolved before MMPIs can be considered for moreextensive use in humans. A major problem is the time of drug administration; based on both animalmodels and human clinical trials, an MMPI is most likely to have full effect when administeredduring the early phases of the disease. In addition, dose-evaluation and modes to evaluate dose-efficacy must be developed. The effect of an MMPI can indeed vary tremendeously depending onthe time of administration. As known for CMTs, application of the drug during activation of MMPs invitro drops IC50s dramatically in comparison to adding the drug after activation (Golub et al. 1998).The goal must be the definition of a dose that is suitable for continuous oral administration andresults in sustained plasma concentrations exceeding the inhibitory concentrations for MMPs invitro with tolerable toxicity. Critically, the development of MMPIs lacks endpoint assessments forthe efficacy of the compound in modulating the activity of its target MMP in vivo and therelationship between target modulation and clinical response. Many investigators point out thepressing need to develop novel clinical trials which include the use of parameters of tumorprogression instead of response rate as the principal endpoint, randomized instead of single-armphase II trials and randomized discontinuation trials (Hidalgo and Eckhardt 2001). Interestingly,CMT-3 can be monitored by analysis of changes in MMP-9 plasma levels, a measurement thatdoes appear to correlate with efficacy (Hidalgo and Eckhardt 2001). MMPs could also possibly beanalyzed in other tissues, such as circulating blood cells, oral, lung and tear fluids (Holopainen etal. 2003, Mäntylä et al. 2003, Prikk et al. 2001, Sorsa et al. 1999). There is also a pressing need todevelop and validate markers of tumor progression, a task that might best be approached byradiologists and imaging scientists (Bremer et al. 2001).

73

Figure 10. Schematic illustration of a possible role for Ln-5 in inducing cell migration. In a quiescentcell, the Ln-5 molecule (with a processed αααα3- and γγγγ2-chain) is part of the anchoring filaments in theBM hemidesmosomal complex. A signal triggering the cell to move results in neoexpression of Ln-5(containing an unprocessed αααα3-chain) and MMP activation. MMPs cleave the Ln-5 γγγγ2-chaingenerating a cryptic domainIII fragment which binds to the EGF-receptor of the cell (and possibly toneighbouring cells) inducing intracellular signalling. Subsequently, the cell is triggered to formfilopodia or invadiopodia (in the case of a cancer cell) and migrate. Concomitantly, MMP-2 (and mostlikely other MMPs) synthesis is induced. The figure is drawn based on results in the present studyand from (Koshikawa et al. 2000, Schenk et al. 2003).

74

Animal models must be more standardized and developed to mimic the human disease moreprecisely (Coussens et al. 2002). A “Wild West” mentality in the use of animal models will onlyresult in contradictory and confusing results with information not very valuable for the assessmentof a human disease. Malignant cells are genetically highly unstable so therapies targeting thesecells can result in modifications resulting in resistance to cancer therapy. MMP activity in cancersusually originates from both tumor and stroma and importantly, stromal cells are much less likely todevelop drug resistance (McCawley and Matrisian 2000). MMPs and Ln-5 could exhibit potentialuse in diagnostics as prognostic markers of cancers and chronic inflammatory diseases (Skyldberget al. 1999, Väisänen et al. 1996). Indeed, one such diagnostic point-of-care test for monitoringMMP-8 in the GCF of chronic periodontal patients has been developed (Mäntylä et al. 2003, Sorsaet al. 1999). Also new, exciting aspects of MMP inhibition have been presented that involve anoncatalytic targeting of MMPs. These include malignant growth arrest by the MMP-2 hemopexindomain generated through autolysis, development of substrate-targeted inhibitors and MMP-activable cytotoxics (Bello et al. 2001, Overall and Lopez-Otin 2002). In the development ofMMPIs, one must also emphasize that proteolytic activity may result in anti-angiogenic peptides,activation of anti-tumor molecules and that, in fact, upregulated MMP activity may result in theexcess degradation of ECM molecules and subsequent reduction in migration of tumor cells. Thesame goes for any chronic inflammation of course, and the same problems reside there.Of the MMPIs used in the studies presented herein, two are being developed for the use in clinicaltreatment of humans. We used CMT-8, which exhibits very similar MMP-inhibitory efficacy as theCMT-3. Thus, the more lipophilic CMT-3 (Metastat) is currently in clinical trials for treatment forvarious cancers, one being human immunodeficiency virus-related Kaposi’s sarcoma. This isparticularly interesting since human immunodeficiency virus-1 protease inhibitors were found toregress Kaposi’s sarcoma tumor formation and inhibit MMP-2 activation (Sgadari et al. 2002).While most MMPIs are more or less cytostatic, not cytotoxic, CMT-3 exhibits both features, makingit an interesting candidate for cancer treatment. In this aspect, an interesting target for cytotoxicMMPIs could be diseases associated with fibrosis. A study with exprimental fibrosis in ratsdemonstrate that doxycycline attenuated neovascularisation, inhibited early fibrous-tissueformation and regression (Lamparter et al. 2002). In addition, since TGF-β1, which is involved inexcess scar tissue formation, is activated by MMPs (at least MMP-2), then there certainly shouldbe a pharmacological interest in the development of certain types of MMPIs for fibrotic diseases.The CTT-peptide is also being developed for future clinical trials for the treatment of cancer.However, as expected with any peptide, it can be easily degraded in vivo, restricting its oralavailability. Attempts are made to “shield” or envelope the peptide for delivery to its target where itcan be liberated. The CTT-peptide has also been used for the delivery of anti-cancer drugs totumors, based on its ability to home to tumors expressing MMP-2 and MMP-9 (Medina et al. 2001).Thus, the CTT-peptide is not only a potential anti-cancer compound per se due to its gelatinaseinhibitory activity, but can also be used as a tool for more targeted delivery of other anti-cancerdrugs as well as gene delivery. In this way, the development of more specific MMP-inhibitors mayalso indirectly have benefits in treatment of cancer patients.Clearly, the function of MMPs is only now emerging. While little is known about the true in vivofunction of MMPs, it is becoming evident that MMPs are not mere matrix degrading enzymes.MMPs participate in virtually all extracellular events, from cleavage and remodelling of largemacromolecules, activation and inhibition of growth factors and cytokines, unmasking of crypticsites within macromolecules involved in cell migration and proliferation, generation ofantiangiogenic peptides, tumor suppressing and supporting activities, inflammatory responses anddevelopment. With this in mind, the development of future MMPIs poses a true but beneficialchallenge.

75

7. Conclusions

1. Both HMKs and murine skin keratinocytes express MMP-2. TGF-β induced HMK cell migrationwas dependent on MMP-2 production, activity and Ln-5 deposition while TNF-α induced MMP-9 played a minor role. This indicates a role for MMP-2 in keratinocyte migration during bothperiodontal and cutaneous wound healing as well as in the host defense in inflammations.

2. This study is novel in presenting in vivo expression of MMP-2 in HMKs. In inflamed gingivaltissue, MMP-2 and Ln-5 localizes to the BM region of SE while MMP-9 expression isassociated with the inflammatory cells. In vivo gelatinolytic activity was localized to the stromadirectly beneath the SE. These results indicate a role for MMP-2 in keratinocyte migration andproliferation during inflammation. Inhibition with the CTT-peptide demonstrates that mostgelatinolytic activity in inflamed periodontal tissue in vivo is mediated by MMP-2 and MMP-9.The use of the CTT-peptide improves the specificity and qualitative analysis of in situ gelatinzymography. LDD (Periostat) is currently used for treatment of periodontal disease in the USA.Thus, the CTT-peptide, being a more specific and non-cytotoxic MMPI, could be of potentialuse in the therapy for chronic inflammatory diseases.

3. This study is novel in investigating the effects of CMTs on wound healing associated withestrogen-deprivation and on Ln-5 γ2-chain processing. Estrogen-deprivation affected woundhealing in OVX rats adversely in terms of collagen content, wound strength and Ln-5 γ2-chainprocessing and expression. CMT-8 or estrogen treatment increased collagen deposition,wound strength and normalized Ln-5 γ2-chain processing and expression. MMP activity andexpression was down-regulated by both OVX and further by treating animals with estrogen orCMT-8. This study is also novel in demonstrating MMP-8 expression in murine skin woundkeratinocytes. The role for MMP-8 could be type I collagen degradation, similar to MMP-1activity in human cutaneous wound healing. Contradictory results from different studies on theeffects of estrogen on skin clearly show that a more valid, standardized animal model shouldbe established for future studies. While HRT most likely improve skin quality and woundhealing in postmenopausal women, the present study and other related work also indicate thatlocal estrogen application could be a means to treat impaired wound healing. CMT-8 treatmentshowed similar effect as estrogen and thus, the related compound CMT-3 which is beingdeveloped for cancer treatment, could be of potential use in the treatment of chronic wounds.The CMTs tested on HMKs in culture were found to inhibit cell proliferation and thus, the use ofCMTs should be considered especially in the treatment of fibrotic lesions.

4. This study is novel in presenting evidence that several other MMPs in addition to the previouslydescribed MMP-2 and –14 can cleave the Ln-5 γ2-chain and induce carcinoma cell migration.This is presumably mediated through liberation of a promigratory EGF-like fragment. Ln-5 γ2-chain processing may be important for many biological processes including development,wound healing and remodelling. This also further demonstrates that multiple MMPs canpotentially induce tumor cell invasion through the processing of the Ln-5 γ2-chain, which isusually neoexpressed at the invasive front of tumors and deposited by migrating cells. Of theMMPs tested, N-terminal sequences showed that the Ln-5 γ2-chain contain multiple cleavagesites for different MMPs. Thus, MMPs may regulate other cellular processes in addition tomigration through the Ln-5 γ2-chain.

5. The present study shows that MMPs are expressed and activated at sites and conditionstypically involved in remodelling and cell migration. MMPs play multiple roles in bothphysiological and pathological conditions. In this study, we have presented novel evidence forthe use of different MMPIs in the treatment of both inflammatory processes as well ascarcinoma cell migration. In summary, while the development of MMPIs should take newdirections toward more precise methods to evaluate efficacy, the role of MMPs in pathologicalconditions demonstrate that the MMPIs are still attractive for drug development.

76

References

Affinito P, Palomba S, Sorrentino C, Di Carlo C, Bifulco G, Arienzo MP and Nappi C (1999) Effects ofpostmenopausal hypoestrogenism on skin collagen. Maturitas 33: 239-47.

Agarwal D, Goodison S, Nicholson B, Tarin D and Urquidi V (2003) Expression of matrix metalloproteinase 8(MMP-8) and tyrosinase-related protein-1 (TYRP-1) correlates with the absence of metastasis in an isogenichuman breast cancer model. Differentiation 71: 114-25.

Ahokas K, Lohi J, Lohi H, Elomaa O, Karjalainen-Lindsberg ML, Kere J and Saarialho-Kere U (2002) Matrixmetalloproteinase-21, the human orthologue for XMMP, is expressed during fetal development and incancer. Gene 301: 31-41.

Aimes RT and Quigley JP (1995) Matrix metalloproteinase-2 is an interstitial collagenase. Inhibitor- free enzymecatalyzes the cleavage of collagen fibrils and soluble native type I collagen generating the specific 3/4- and1/4-length fragments. J Biol Chem 270: 5872-6.

Airenne T, Lin Y, Olsson M, Ekblom P, Vainio S and Tryggvason K (2000) Differential expression of mouselaminin gamma2 and gamma2* chain transcripts. Cell Tissue Res 300: 129-37.

Alberts B, Bray D, Lewis J, Raff M, Roberts K and Watson JD (1994) Molecular biology of the cell. New York:Garland Publishing, Inc.

Amano S, Scott IC, Takahara K, Koch M, Champliaud MF, Gerecke DR, Keene DR, Hudson DL, Nishiyama T,Lee S, Greenspan DS and Burgeson RE (2000) Bone morphogenetic protein 1 is an extracellular processingenzyme of the laminin 5 gamma 2 chain. J Biol Chem 275: 22728-35.

Apte SS, Fukai N, Beier DR and Olsen BR (1997) The matrix metalloproteinase-14 (MMP-14) gene isstructurally distinct from other MMP genes and is co-expressed with the TIMP-2 gene during mouseembryogenesis. J Biol Chem 272: 25511-7.

Armstrong DG and Jude EB (2002) The role of matrix metalloproteinases in wound healing. J Am Podiatr MedAssoc 92: 12-8.

Ashcroft GS, Dodsworth J, van Boxtel E, Tarnuzzer RW, Horan MA, Schultz GS and Ferguson MW (1997a)Estrogen accelerates cutaneous wound healing associated with an increase in TGF-beta1 levels. Nat Med 3:1209-15.

Ashcroft GS, Greenwell-Wild T, Horan MA, Wahl SM and Ferguson MW (1999) Topical estrogen acceleratescutaneous wound healing in aged humans associated with an altered inflammatory response. Am J Pathol155: 1137-46.

Ashcroft GS, Horan MA and Ferguson MW (1997b) Aging is associated with reduced deposition of specificextracellular matrix components, an upregulation of angiogenesis, and an altered inflammatory response in amurine incisional wound healing model. J Invest Dermatol 108: 430-7.

Ashcroft GS, Horan MA, Herrick SE, Tarnuzzer RW, Schultz GS and Ferguson MW (1997c) Age-relateddifferences in the temporal and spatial regulation of matrix metalloproteinases (MMPs) in normal skin andacute cutaneous wounds of healthy humans. Cell Tissue Res 290: 581-91.

Ashcroft GS, Mills SJ, Lei K, Gibbons L, Jeong MJ, Taniguchi M, Burow M, Horan MA, Wahl SM and NakayamaT (2003) Estrogen modulates cutaneous wound healing by downregulating macrophage migration inhibitoryfactor. J Clin Invest 111: 1309-18.

Atkinson SJ, Crabbe T, Cowell S, Ward RV, Butler MJ, Sato H, Seiki M, Reynolds JJ and Murphy G (1995)Intermolecular autolytic cleavage can contribute to the activation of progelatinase A by cell membranes. JBiol Chem 270: 30479-85.

Aumailley M, El Khal A, Knoss N and Tunggal L (2003) Laminin 5 processing and its integration into the ECM.Matrix Biol 22: 49-54.

Aumailley M and Gayraud B (1998) Structure and biological activity of the extracellular matrix. J Mol Med 76:253-65.

Bachmeier BE, Boukamp P, Lichtinghagen R, Fusenig NE and Fink E (2000a) Matrix metalloproteinases-2,-3,-7,-9 and-10, but not MMP-11, are differentially expressed in normal, benign tumorigenic and malignanthuman keratinocyte cell lines. Biol Chem 381: 497-507.

Bachmeier BE, Nerlich AG, Boukamp P, Lichtinghagen R, Tschesche H, Fritz H and Fink E (2000b) Humankeratinocyte cell lines differ in the expression of the collagenolytic matrix metalloproteinases-1,-8, and -13and of TIMP-1. Biol Chem 381: 509-16.

Baker AH, Edwards DR and Murphy G (2002) Metalloproteinase inhibitors: biological actions and therapeuticopportunities. J Cell Sci 115: 3719-27.

Baker SE, DiPasquale AP, Stock EL, Quaranta V, Fitchmun M and Jones JC (1996a) Morphogenetic effects ofsoluble laminin-5 on cultured epithelial cells and tissue explants. Exp Cell Res 228: 262-70.

77

Baker SE, Hopkinson SB, Fitchmun M, Andreason GL, Frasier F, Plopper G, Quaranta V and Jones JC (1996b)Laminin-5 and hemidesmosomes: role of the alpha 3 chain subunit in hemidesmosome stability andassembly. J Cell Sci 109: 2509-20.

Balbin M, Fueyo A, Knauper V, Lopez JM, Alvarez J, Sanchez LM, Quesada V, Bordallo J, Murphy G andLopez-Otin C (2001) Identification and enzymatic characterization of two diverging murine counterparts ofhuman interstitial collagenase (MMP-1) expressed at sites of embryo implantation. J Biol Chem 276: 10253-62.

Balbin M, Fueyo A, Knauper V, Pendas AM, Lopez JM, Jimenez MG, Murphy G and Lopez-Otin C (1998)Collagenase 2 (MMP-8) expression in murine tissue-remodeling processes. Analysis of its potential role inpostpartum involution of the uterus. J Biol Chem 273: 23959-68.

Balbin M, Fueyo A, Tester AM, Pendas AM, Pitiot AS, Astudillo A, Overall CM, Shapiro SD and Lopez-Otin C(2003) Loss of collagenase-2 confers increased skin tumor susceptibility to male mice. Nat Genet 35: 252-7.

Bannikov GA, Karelina TV, Collier IE, Marmer BL and Goldberg GI (2002) Substrate binding of gelatinase Binduces its enzymatic activity in the presence of intact propeptide. J Biol Chem 277: 16022-7.

Barrick B, Campbell EJ and Owen CA (1999) Leukocyte proteinases in wound healing: roles in physiologic andpathologic processes. Wound Repair Regen 7: 410-22.

Bartlett JD, Ryu OH, Xue J, Simmer JP and Margolis HC (1998) Enamelysin mRNA displays a developmentallydefined pattern of expression and encodes a protein which degrades amelogenin. Connect Tissue Res 39:101-9.

Bartold PM, Walsh LJ and Narayanan AS (2000) Molecular and cell biology of the gingiva. Periodontol 24: 28-55.

Basbaum CB and Werb Z (1996) Focalized proteolysis: spatial and temporal regulation of extracellular matrixdegradation at the cell surface. Curr Opin Cell Biol 8: 731-8.

Basset P, Okada A, Chenard MP, Kannan R, Stoll I, Anglard P, Bellocq JP and Rio MC (1997) Matrixmetalloproteinases as stromal effectors of human carcinoma progression: therapeutic implications. MatrixBiol 15: 535-41.

Belaaouaj A, Shipley JM, Kobayashi DK, Zimonjic DB, Popescu N, Silverman GA and Shapiro SD (1995)Human macrophage metalloelastase. Genomic organization, chromosomal location, gene linkage, andtissue-specific expression. J Biol Chem 270: 14568-75.

Bello L, Lucini V, Carrabba G, Giussani C, Machluf M, Pluderi M, Nikas D, Zhang J, Tomei G, Villani RM, CarrollRS, Bikfalvi A and Black PM (2001) Simultaneous inhibition of glioma angiogenesis, cell proliferation, andinvasion by a naturally occurring fragment of human metalloproteinase-2. Cancer Res 61: 8730-6.

Berndt A, Johannesson J, Haas M, Arkona C, Katenkamp D, Wiederanders B, Hyckel P and Kosmehl H (2000)Simultaneous enzyme overlay membrane (EOM)-based in situ zymography and immunofluorescencetechnique reveals cathepsin B-like activity in a subset of tumour vessels. Histochem Cell Biol 114: 63-8.

Birkedal-Hansen H, Moore WG, Bodden MK, Windsor LJ, Birkedal-Hansen B, DeCarlo A and Engler JA (1993)Matrix metalloproteinases: a review. Crit Rev Oral Biol Med 4: 197-250.

Bloom W and Fawcett DW (1968) A textbook of histology. Philadelphia: W.B. Saunders Company.Borradori L and Sonnenberg A (1999) Structure and function of hemidesmosomes: more than simple adhesion

complexes. J Invest Dermatol 112: 411-8.Bremer C, Tung CH and Weissleder R (2001) In vivo molecular target assessment of matrix metalloproteinase

inhibition. Nat Med 7: 743-8.Brew K, Dinakarpandian D and Nagase H (2000) Tissue inhibitors of metalloproteinases: evolution, structure

and function. Biochim Biophys Acta 1477: 267-83.Brincat M, Kabalan S, Studd JW, Moniz CF, de Trafford J and Montgomery J (1987a) A study of the decrease of

skin collagen content, skin thickness, and bone mass in the postmenopausal woman. Obstet Gynecol 70:840-5.

Brincat M, Moniz CJ, Studd JW, Darby A, Magos A, Emburey G and Versi E (1985) Long-term effects of themenopause and sex hormones on skin thickness. Br J Obstet Gynaecol 92: 256-9.

Brincat M, Yuen AW, Studd JW, Montgomery J, Magos AL and Savvas M (1987b) Response of skin thicknessand metacarpal index to estradiol therapy in postmenopausal women. Obstet Gynecol 70: 538-41.

Bullen EC, Longaker MT, Updike DL, Benton R, Ladin D, Hou Z and Howard EW (1995) Tissue inhibitor ofmetalloproteinases-1 is decreased and activated gelatinases are increased in chronic wounds. J InvestDermatol 104: 236-40.

Burgeson RE and Christiano AM (1997) The dermal-epidermal junction. Curr Opin Cell Biol 9: 651-8.Calvin M (2000) Oestrogens and wound healing. Maturitas 34: 195-210.Calvin M, Dyson M, Rymer J and Young SR (1998) The effects of ovarian hormone deficiency on wound

contraction in a rat model. Br J Obstet Gynaecol 105: 223-7.Cao J, Rehemtulla A, Bahou W and Zucker S (1996) Membrane type matrix metalloproteinase 1 activates pro-

gelatinase A without furin cleavage of the N-terminal domain. J Biol Chem 271: 30174-80.

78

Carter WG, Ryan MC and Gahr PJ (1991) Epiligrin, a new cell adhesion ligand for integrin alpha 3 beta 1 inepithelial basement membranes. Cell 65: 599-610.

Chen C, Schultz GS, Bloch M, Edwards PD, Tebes S and Mast BA (1999) Molecular and mechanistic validationof delayed healing rat wounds as a model for human chronic wounds. Wound Repair Regen 7: 486-94.

Chubinskaya S, Kuettner KE and Cole AA (1999) Expression of matrix metalloproteinases in normal anddamaged articular cartilage from human knee and ankle joints. Lab Invest 79: 1669-77.

Cole AA, Chubinskaya S, Schumacher B, Huch K, Szabo G, Yao J, Mikecz K, Hasty KA and Kuettner KE (1996)Chondrocyte matrix metalloproteinase-8. Human articular chondrocytes express neutrophil collagenase. JBiol Chem 271: 11023-6.

Colognato H and Yurchenco PD (2000) Form and function: the laminin family of heterotrimers. Dev Dyn 218:213-34.

Cornelius LA, Nehring LC, Harding E, Bolanowski M, Welgus HG, Kobayashi DK, Pierce RA and Shapiro SD(1998) Matrix metalloproteinases generate angiostatin: effects on neovascularization. J Immunol 161: 6845-52.

Cossins J, Dudgeon TJ, Catlin G, Gearing AJ and Clements JM (1996) Identification of MMP-18, a putativenovel human matrix metalloproteinase. Biochem Biophys Res Commun 228: 494-8.

Coussens LM, Fingleton B and Matrisian LM (2002) Matrix metalloproteinase inhibitors and cancer: trials andtribulations. Science 295: 2387-92.

Cowell S, Knäuper V, Stewart ML, D'Ortho MP, Stanton H, Hembry RM, Lopez-Otin C, Reynolds JJ and MurphyG (1998) Induction of matrix metalloproteinase activation cascades based on membrane-type 1 matrixmetalloproteinase: associated activation of gelatinase A, gelatinase B and collagenase 3. Biochem J 331:453-8.

Craig RG, Yu Z, Xu L, Barr R, Ramamurthy N, Boland J, Schneir M and Golub LM (1998) A chemically modifiedtetracycline inhibits streptozotocin-induced diabetic depression of skin collagen synthesis and steady-statetype I procollagen mRNA. Biochim Biophys Acta 1402: 250-60.

Cunningham AC, Hasty KA, Enghild JJ and Mast AE (2002) Structural and functional characterization of tissuefactor pathway inhibitor following degradation by matrix metalloproteinase-8. Biochem J 367: 451-8.

Dahan M, Nawrocki B, Elkaim R, Soell M, Bolcato-Bellemin AL, Birembaut P and Tenenbaum H (2001)Expression of matrix metalloproteinases in healthy and diseased human gingiva. J Clin Periodontol 28: 128-36.

Decline F and Rousselle P (2001) Keratinocyte migration requires alpha2beta1 integrin-mediated interactionwith the laminin 5 gamma2 chain. J Cell Sci 114: 811-23.

Deryugina EI, Ratnikov BI and Strongin AY (2003) Prinomastat, a hydroxamate inhibitor of matrixmetalloproteinases, has a complex effect on migration of breast carcinoma cells. Int J Cancer 104: 533-41.

Devarajan P, Mookhtiar K, Van Wart H and Berliner N (1991) Structure and expression of the cDNA encodinghuman neutrophil collagenase. Blood 77: 2731-8.

Di Girolamo N, Tedla N, Lloyd A and Wakefield D (1998) Expression of matrix metalloproteinases by humanplasma cells and B lymphocytes. Eur J Immunol 28: 1773-84.

d'Ortho MP, Stanton H, Butler M, Atkinson SJ, Murphy G and Hembry RM (1998) MT1-MMP on the cell surfacecauses focal degradation of gelatin films. FEBS Lett 421: 159-64.

Dumas V, Kanitakis J, Charvat S, Euvrard S, Faure M and Claudy A (1999) Expression of basement membraneantigens and matrix metalloproteinases 2 and 9 in cutaneous basal and squamous cell carcinomas.Anticancer Res 19: 2929-38.

Eckles TA, Reinhardt RA, Dyer JK, Tussing GJ, Szydlowski WM and DuBous LM (1990) Intracrevicularapplication of tetracycline in white petrolatum for the treatment of periodontal disease. J Clin Periodontol 17:454-62.

Fisher GJ, Choi HC, Bata-Csorgo Z, Shao Y, Datta S, Wang ZQ, Kang S and Voorhees JJ (2001) Ultravioletirradiation increases matrix metalloproteinase-8 protein in human skin in vivo. J Invest Dermatol 117: 219-26.

Freije JM, Diez-Itza I, Balbin M, Sanchez LM, Blasco R, Tolivia J and Lopez-Otin C (1994) Molecular cloningand expression of collagenase-3, a novel human matrix metalloproteinase produced by breast carcinomas. JBiol Chem 269: 16766-73.

Fridman R, Toth M, Pena D and Mobashery S (1995) Activation of progelatinase B (MMP-9) by gelatinase A(MMP-2). Cancer Res 55: 2548-55.

Friedl P and Brocker EB (2000) The biology of cell locomotion within three-dimensional extracellular matrix. CellMol Life Sci 57: 41-64.

Fukushima Y, Ohnishi T, Arita N, Hayakawa T and Sekiguchi K (1998) Integrin alpha3beta1-mediatedinteraction with laminin-5 stimulates adhesion, migration and invasion of malignant glioma cells. Int J Cancer76: 63-72.

79

Gagnoux-Palacios L, Allegra M, Spirito F, Pommeret O, Romero C, Ortonne JP and Meneguzzi G (2001) Theshort arm of the laminin gamma2 chain plays a pivotal role in the incorporation of laminin 5 into theextracellular matrix and in cell adhesion. J Cell Biol 153: 835-50.

Galis ZS, Sukhova GK and Libby P (1995) Microscopic localization of active proteases by in situ zymography:detection of matrix metalloproteinase activity in vascular tissue. Faseb J 9: 974-80.

Gearing AJ, Beckett P, Christodoulou M, Churchill M, Clements JM, Crimmin M, Davidson AH, Drummond AH,Galloway WA, Gilbert R and et al. (1995) Matrix metalloproteinases and processing of pro-TNF-alpha. JLeukoc Biol 57: 774-7.

Giambernardi TA, Grant GM, Taylor GP, Hay RJ, Maher VM, McCormick JJ and Klebe RJ (1998) Overview ofmatrix metalloproteinase expression in cultured human cells. Matrix Biol 16: 483-96.

Giannelli G, Bergamini C, Fransvea E, Marinosci F, Quaranta V and Antonaci S (2001) Human hepatocellularcarcinoma (HCC) cells require both alpha3beta1 integrin and matrix metalloproteinases activity for migrationand invasion. Lab Invest 81: 613-27.

Giannelli G, Falk-Marzillier J, Schiraldi O, Stetler-Stevenson WG and Quaranta V (1997) Induction of cellmigration by matrix metalloprotease-2 cleavage of laminin-5. Science 277: 225-8.

Giannelli G, Fransvea E, Marinosci F, Bergamini C, Colucci S, Schiraldi O and Antonaci S (2002) Transforminggrowth factor-beta1 triggers hepatocellular carcinoma invasiveness via alpha3beta1 integrin. Am J Pathol161: 183-93.

Giannelli G, Pozzi A, Stetler-Stevenson WG, Gardner HA and Quaranta V (1999) Expression of matrixmetalloprotease-2-cleaved laminin-5 in breast remodeling stimulated by sex steroids. Am J Pathol 154:1193-201.

Gilles C, Polette M, Coraux C, Tournier JM, Meneguzzi G, Munaut C, Volders L, Rousselle P, Birembaut P andFoidart JM (2001) Contribution of MT1-MMP and of human laminin-5 gamma2 chain degradation tomammary epithelial cell migration. J Cell Sci 114: 2967-76.

Goldfinger LE, Hopkinson SB, deHart GW, Collawn S, Couchman JR and Jones JC (1999) The alpha3 lamininsubunit, alpha6beta4 and alpha3beta1 integrin coordinately regulate wound healing in cultured epithelialcells and in the skin. J Cell Sci 112: 2615-29.

Goldfinger LE, Stack MS and Jones JC (1998) Processing of laminin-5 and its functional consequences: role ofplasmin and tissue-type plasminogen activator. J Cell Biol 141: 255-65.

Golub LM, Evans RT, McNamara TF, Lee HM and Ramamurthy NS (1994) A non-antimicrobial tetracyclineinhibits gingival matrix metalloproteinases and bone loss in Porphyromonas gingivalis-induced periodontitisin rats. Ann N Y Acad Sci 732: 96-111.

Golub LM, Lee HM, Greenwald RA, Ryan ME, Sorsa T, Salo T and Giannobile WV (1997) A matrixmetalloproteinase inhibitor reduces bone-type collagen degradation fragments and specific collagenases ingingival crevicular fluid during adult periodontitis. Inflamm Res 46: 310-9.

Golub LM, Lee HM, Ryan ME, Giannobile WV, Payne J and Sorsa T (1998) Tetracyclines inhibit connectivetissue breakdown by multiple non- antimicrobial mechanisms. Adv Dent Res 12: 12-26.

Greenwald RA, Golub LM, Ramamurthy NS, Chowdhury M, Moak SA and Sorsa T (1998) In vitro sensitivity ofthe three mammalian collagenases to tetracycline inhibition: relationship to bone and cartilage degradation.Bone 22: 33-8.

Gross J and Lapier CM (1962) Collagenolytic activity in amphibian tissue: a tissue culture assay. Proc Natl AcadSci U S A 48: 1014-1022.

Haapasaari KM, Raudaskoski T, Kallioinen M, Suvanto-Luukkonen E, Kauppila A, Läärä E, Risteli J andOikarinen A (1997) Systemic therapy with estrogen or estrogen with progestin has no effect on skin collagenin postmenopausal women. Maturitas 27: 153-62.

Hahn-Dantona E, Ruiz JF, Bornstein P and Strickland DK (2001) The low density lipoprotein receptor-relatedprotein modulates levels of matrix metalloproteinase 9 (MMP-9) by mediating its cellular catabolism. J BiolChem 276: 15498-503.

Han YP, Tuan TL, Hughes M, Wu H and Garner WL (2001) Transforming growth factor-beta and tumor necrosisfactor-alpha-mediated induction and proteolytic activation of MMP-9 in human skin. J Biol Chem 276: 22341-350.

Hanemaaijer R, Sorsa T, Konttinen YT, Ding Y, Sutinen M, Visser H, van Hinsbergh VW, Helaakoski T,Kainulainen T, Rönkä H, Tschesche H and Salo T (1997) Matrix metalloproteinase-8 is expressed inrheumatoid synovial fibroblasts and endothelial cells. Regulation by tumor necrosis factor- alpha anddoxycycline. J Biol Chem 272: 31504-9.

Hao J, Yang Y, McDaniel KM, Dalkin BL, Cress AE and Nagle RB (1996) Differential expression of laminin 5(alpha 3 beta 3 gamma 2) by human malignant and normal prostate. Am J Pathol 149: 1341-9.

Hasty KA, Hibbs MS, Kang AH and Mainardi CL (1986) Secreted forms of human neutrophil collagenase. J BiolChem 261: 5645-50.

80

Hasty KA, Jeffrey JJ, Hibbs MS and Welgus HG (1987) The collagen substrate specificity of human neutrophilcollagenase. J Biol Chem 262: 10048-52.

Hasty KA, Pourmotabbed TF, Goldberg GI, Thompson JP, Spinella DG, Stevens RM and Mainardi CL (1990)Human neutrophil collagenase. A distinct gene product with homology to other matrix metalloproteinases. JBiol Chem 265: 11421-4.

Hautamäki RD, Kobayashi DK, Senior RM and Shapiro SD (1997) Requirement for macrophage elastase forcigarette smoke-induced emphysema in mice. Science 277: 2002-4.

Herman MP, Sukhova GK, Libby P, Gerdes N, Tang N, Horton DB, Kilbride M, Breitbart RE, Chun M andSchonbeck U (2001) Expression of neutrophil collagenase (matrix metalloproteinase-8) in human atheroma:a novel collagenolytic pathway suggested by transcriptional profiling. Circulation 104: 1899-904.

Hidalgo M and Eckhardt SG (2001) Development of matrix metalloproteinase inhibitors in cancer therapy. J NatlCancer Inst 93: 178-93.

Hill K (1996) The demography of menopause. Maturitas 23: 113-27.Hintermann E, Bilban M, Sharabi A and Quaranta V (2001) Inhibitory role of alpha 6 beta 4-associated erbB-2

and phosphoinositide 3-kinase in keratinocyte haptotactic migration dependent on alpha 3 beta 1 integrin. JCell Biol 153: 465-78.

Hintermann E, Haake SK, Christen U, Sharabi A and Quaranta V (2002) Discrete proteolysis of focal contactand adherens junction components in Porphyromonas gingivalis-infected oral keratinocytes: a strategy forcell adhesion and migration disabling. Infect Immun 70: 5846-56.

Hiraoka N, Allen E, Apel IJ, Gyetko MR and Weiss SJ (1998) Matrix metalloproteinases regulateneovascularization by acting as pericellular fibrinolysins. Cell 95: 365-77.

Hirose T, Patterson C, Pourmotabbed T, Mainardi CL and Hasty KA (1993) Structure-function relationship ofhuman neutrophil collagenase: identification of regions responsible for substrate specificity and generalproteinase activity. Proc Natl Acad Sci U S A 90: 2569-73.

Hlubek F, Jung A, Kotzor N, Kirchner T and Brabletz T (2001) Expression of the invasion factor laminin gamma2in colorectal carcinomas is regulated by beta-catenin. Cancer Res 61: 8089-93.

Holmbeck K, Bianco P, Caterina J, Yamada S, Kromer M, Kuznetsov SA, Mankani M, Robey PG, Poole AR,Pidoux I, Ward JM and Birkedal-Hansen H (1999) MT1-MMP-deficient mice develop dwarfism, osteopenia,arthritis, and connective tissue disease due to inadequate collagen turnover. Cell 99: 81-92.

Holopainen JM, Moilanen JA, Sorsa T, Kivelä-Rajamäki M, Tervahartiala T, Vesaluoma MH and Tervo TM(2003) Activation of matrix metalloproteinase-8 by membrane type 1-MMP and their expression in humantears after photorefractive keratectomy. Invest Ophthalmol Vis Sci 44: 2550-6.

Hormia M, Sahlberg C, Thesleff I and Airenne T (1998) The epithelium-tooth interface--a basal lamina rich inlaminin-5 and lacking other known laminin isoforms. J Dent Res 77: 1479-85.

Hotary K, Allen E, Punturieri A, Yana I and Weiss SJ (2000) Regulation of cell invasion and morphogenesis in athree-dimensional type I collagen matrix by membrane-type matrix metalloproteinases 1, 2, and 3. J Cell Biol149: 1309-23.

Hotary KB, Allen ED, Brooks PC, Datta NS, Long MW and Weiss SJ (2003) Membrane type I matrixmetalloproteinase usurps tumor growth control imposed by the three-dimensional extracellular matrix. Cell114: 33-45.

Hu SI, Klein M, Carozza M, Rediske J, Peppard J and Qi JS (1999) Identification of a splice variant of neutrophilcollagenase (MMP-8). FEBS Lett 443: 8-10.

Huhtala P, Chow LT and Tryggvason K (1990) Structure of the human type IV collagenase gene. J Biol Chem265: 11077-82.

Huhtala P, Tuuttila A, Chow LT, Lohi J, Keski-Oja J and Tryggvason K (1991) Complete structure of the humangene for 92-kDa type IV collagenase. Divergent regulation of expression for the 92- and 72-kilodaltonenzyme genes in HT-1080 cells. J Biol Chem 266: 16485-90.

Ingman T, Tervahartiala T, Ding Y, Tschesche H, Haerian A, Kinane DF, Konttinen YT and Sorsa T (1996)Matrix metalloproteinases and their inhibitors in gingival crevicular fluid and saliva of periodontitis patients. JClin Periodontol 23: 1127-32.

Inoue M, Kratz G, Haegerstrand A and Stahle-Backdahl M (1995) Collagenase expression is rapidly induced inwound-edge keratinocytes after acute injury in human skin, persists during healing, and stops at re-epithelialization. J Invest Dermatol 104: 479-83.

Jacinto A, Martinez-Arias A and Martin P (2001) Mechanisms of epithelial fusion and repair. Nat Cell Biol 3:E117-23.

Johansson N, Ahonen M and Kähäri VM (2000) Matrix metalloproteinases in tumor invasion. Cell Mol Life Sci57: 5-15.

Johansson N, Saarialho-Kere U, Airola K, Herva R, Nissinen L, Westermarck J, Vuorio E, Heino J and KähäriVM (1997) Collagenase-3 (MMP-13) is expressed by hypertrophic chondrocytes, periosteal cells, andosteoblasts during human fetal bone development. Dev Dyn 208: 387-97.

81

Jorgensen LN, Sorensen LT, Kallehave F, Vange J and Gottrup F (2002) Premenopausal women deposit morecollagen than men during healing of an experimental wound. Surgery 131: 338-43.

Joshi N and Miller DQ (1997) Doxycycline revisited. Arch Intern Med 157: 1421-8.Kainulainen T, Häkkinen L, Hamidi S, Larjava K, Kallioinen M, Peltonen J, Salo T, Larjava H and Oikarinen A

(1998) Laminin-5 expression is independent of the injury and the microenvironment during reepithelializationof wounds. J Histochem Cytochem 46: 353-60.

Kajita M, Itoh Y, Chiba T, Mori H, Okada A, Kinoh H and Seiki M (2001) Membrane-type 1 matrixmetalloproteinase cleaves CD44 and promotes cell migration. J Cell Biol 153: 893-904.

Kallunki P, Sainio K, Eddy R, Byers M, Kallunki T, Sariola H, Beck K, Hirvonen H, Shows TB and Tryggvason K(1992) A truncated laminin chain homologous to the B2 chain: structure, spatial expression, andchromosomal assignment. J Cell Biol 119: 679-93.

Kang T, Nagase H and Pei D (2002) Activation of membrane-type matrix metalloproteinase 3 zymogen by theproprotein convertase furin in the trans-Golgi network. Cancer Res 62: 675-81.

Kang T, Yi J, Yang W, Wang X, Jiang A and Pei D (2000) Functional characterization of MT3-MMP intransfected MDCK cells: progelatinase A activation and tubulogenesis in 3-D collagen lattice. Faseb J 14:2559-68.

Kim MH, Albertsson P, Xue Y, Nannmark U, Kitson RP and Goldfarb RH (2001) Expression of neutrophilcollagenase (MMP-8) in Jurkat T leukemia cells and its role in invasion. Anticancer Res 21: 45-50.

Kjeldsen L, Bjerrum OW, Hovgaard D, Johnsen AH, Sehested M and Borregaard N (1992) Human neutrophilgelatinase: a marker for circulating blood neutrophils. Purification and quantitation by enzyme linkedimmunosorbent assay. Eur J Haematol 49: 180-91.

Knäuper V, Lopez-Otin C, Smith B, Knight G and Murphy G (1996a) Biochemical characterization of humancollagenase-3. J Biol Chem 271: 1544-50.

Knäuper V, Murphy G and Tschesche H (1996b) Activation of human neutrophil procollagenase by stromelysin2. Eur J Biochem 235: 187-91.

Knäuper V, Smith B, Lopez-Otin C and Murphy G (1997) Activation of progelatinase B (proMMP-9) by activecollagenase-3 (MMP- 13). Eur J Biochem 248: 369-73.

Knäuper V, Will H, Lopez-Otin C, Smith B, Atkinson SJ, Stanton H, Hembry RM and Murphy G (1996c) Cellularmechanisms for human procollagenase-3 (MMP-13) activation. Evidence that MT1-MMP (MMP-14) andgelatinase a (MMP-2) are able to generate active enzyme. J Biol Chem 271: 17124-31.

Kobayashi T, Hattori S, Nagai Y, Tajima S and Nishikawa T (1998) Differential regulation of MMP-2 and MMP-9gelatinases in cultured human keratinocytes. Dermatology 197: 1-5.

Koivunen E, Arap W, Valtanen H, Rainisalo A, Medina OP, Heikkilä P, Kantor C, Gahmberg CG, Salo T,Konttinen YT, Sorsa T, Ruoslahti E and Pasqualini R (1999) Tumor targeting with a selective gelatinaseinhibitor. Nat Biotechnol 17: 768-74.

Konttinen YT, Lindy O, Kemppinen P, Saari H, Suomalainen K, Vauhkonen M, Lindy S and Sorsa T (1991)Collagenase reserves in polymorphonuclear neutrophil leukocytes from synovial fluid and peripheral blood ofpatients with rheumatoid arthritis. Matrix 11: 296-301.

Korostoff JM, Wang JF, Sarment DP, Stewart JC, Feldman RS and Billings PC (2000) Analysis of in situprotease activity in chronic adult periodontitis patients: expression of activated MMP-2 and a 40 kDa serineprotease. J Periodontol 71: 353-60.

Koshikawa N, Giannelli G, Cirulli V, Miyazaki K and Quaranta V (2000) Role of cell surface metalloproteaseMT1-MMP in epithelial cell migration over laminin-5. J Cell Biol 148: 615-24.

Koshikawa N, Moriyama K, Takamura H, Mizushima H, Nagashima Y, Yanoma S and Miyazaki K (1999)Overexpression of laminin gamma2 chain monomer in invading gastric carcinoma cells. Cancer Res 59:5596-601.

Koshikawa N, Yasumitsu H, Umeda M and Miyazaki K (1992) Multiple secretion of matrix serine proteinases byhuman gastric carcinoma cell lines. Cancer Res 52: 5046-53.

Krane SM, Byrne MH, Lemaitre V, Henriet P, Jeffrey JJ, Witter JP, Liu X, Wu H, Jaenisch R and Eeckhout Y(1996) Different collagenase gene products have different roles in degradation of type I collagen. J BiolChem 271: 28509-15.

Kueng W, Silber E and Eppenberger U (1989) Quantification of cells cultured on 96-well plates. Anal Biochem182: 16-9.

Kähäri VM and Saarialho-Kere U (1999) Matrix metalloproteinases and their inhibitors in tumour growth andinvasion. Ann Med 31: 34-45.

Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4.Nature 227: 680-5.

Lamparter S, Slight SH and Weber KT (2002) Doxycycline and tissue repair in rats. J Lab Clin Med 139: 295-302.

82

Larjava H, Salo T, Haapasalmi K, Kramer RH and Heino J (1993) Expression of integrins and basementmembrane components by wound keratinocytes. J Clin Invest 92: 1425-35.

Lee HM, Golub LM, Cao J, Teronen O, Laitinen M, Salo T, Zucker S and Sorsa T (2001) CMT-3, a non-antimicrobial tetracycline (TC), inhibits MT1-MMP activity: relevance to cancer. Curr Med Chem 8: 257-60.

Li H, Bauzon DE, Xu X, Tschesche H, Cao J and Sang QA (1998) Immunological characterization of cell-surface and soluble forms of membrane type 1 matrix metalloproteinase in human breast cancer cells and infibroblasts. Mol Carcinog 22: 84-94.

Lindy O, Konttinen YT, Sorsa T, Ding Y, Santavirta S, Ceponis A and Lopez-Otin C (1997) Matrixmetalloproteinase 13 (collagenase 3) in human rheumatoid synovium. Arthritis Rheum 40: 1391-9.

Liotta LA, Abe S, Robey PG and Martin GR (1979) Preferential digestion of basement membrane collagen by anenzyme derived from a metastatic murine tumor. Proc Natl Acad Sci U S A 76: 2268-72.

Liu Z, Zhou X, Shapiro SD, Shipley JM, Twining SS, Diaz LA, Senior RM and Werb Z (2000) The serpin alpha1-proteinase inhibitor is a critical substrate for gelatinase B/MMP-9 in vivo. Cell 102: 647-55.

Llano E, Pendas AM, Freije JP, Nakano A, Knäuper V, Murphy G and Lopez-Otin C (1999) Identification andcharacterization of human MT5-MMP, a new membrane- bound activator of progelatinase a overexpressedin brain tumors. Cancer Res 59: 2570-6.

Lohi J (2001) Laminin-5 in the progression of carcinomas. Int J Cancer 94: 763-7.Lohi J, Leivo I, Owaribe K, Burgeson RE, Franssila K and Virtanen I (1998) Neoexpression of the epithelial

adhesion complex antigens in thyroid tumours is associated with proliferation and squamous differentiationmarkers. J Pathol 184: 191-6.

Lohi J, Wilson CL, Roby JD and Parks WC (2001) Epilysin, a novel human matrix metalloproteinase (MMP-28)expressed in testis and keratinocytes and in response to injury. J Biol Chem 276: 10134-44.

Lokeshwar BL, Selzer MG, Zhu BQ, Block NL and Golub LM (2002) Inhibition of cell proliferation, invasion,tumor growth and metastasis by an oral non-antimicrobial tetracycline analog (COL-3) in a metastaticprostate cancer model. Int J Cancer 98: 297-309.

Loo WJ, Kirtschig G and Wojnarowska F (2001) Minocycline as a therapeutic option in bullous pemphigoid. ClinExp Dermatol 26: 376-9.

Lopez-Otin C and Overall CM (2002) Protease degradomics: a new challenge for proteomics. Nat Rev Mol CellBiol 3: 509-19.

Lund LR, Romer J, Bugge TH, Nielsen BS, Frandsen TL, Degen JL, Stephens RW and Dano K (1999)Functional overlap between two classes of matrix-degrading proteases in wound healing. Embo J 18: 4645-56.

Luo D, Mari B, Stoll I and Anglard P (2002) Alternative splicing and promoter usage generates an intracellularstromelysin 3 isoform directly translated as an active matrix metalloproteinase. J Biol Chem 277: 25527-36.

Lynch CC and Matrisian LM (2002) Matrix metalloproteinases in tumor-host cell communication. Differentiation70: 561-73.

Madlener M (1998) Differential expression of matrix metalloproteinases and their physiological inhibitors inacute murine skin wounds. Arch Dermatol Res 290 Suppl: S24-9.

Maisi P, Prikk K, Sepper R, Pirilä E, Salo T, Hietanen J and Sorsa T (2002) Soluble membrane-type 1 matrixmetalloproteinase (MT1-MMP) and gelatinase A (MMP-2) in induced sputum and bronchoalveolar lavagefluid of human bronchial asthma and bronchiectasis. Apmis 110: 771-82.

Mallya SK, Mookhtiar KA, Gao Y, Brew K, Dioszegi M, Birkedal-Hansen H and Van Wart HE (1990)Characterization of 58-kilodalton human neutrophil collagenase: comparison with human fibroblastcollagenase. Biochemistry 29: 10628-34.

Marinkovich MP, Lunstrum GP and Burgeson RE (1992) The anchoring filament protein kalinin is synthesizedand secreted as a high molecular weight precursor. J Biol Chem 267: 17900-6.

Marinkovich MP, Verrando P, Keene DR, Meneguzzi G, Lunstrum GP, Ortonne JP and Burgeson RE (1993)Basement membrane proteins kalinin and nicein are structurally and immunologically identical. Lab Invest69: 295-9.

Martin P (1997) Wound healing--aiming for perfect skin regeneration. Science 276: 75-81.Massova I, Kotra LP, Fridman R and Mobashery S (1998) Matrix metalloproteinases: structures, evolution, and

diversification. Faseb J 12: 1075-95.Matsui C, Wang CK, Nelson CF, Bauer EA and Hoeffler WK (1995) The assembly of laminin-5 subunits. J Biol

Chem 270: 23496-503.Mauviel A (1993) Cytokine regulation of metalloproteinase gene expression. J Cell Biochem 53: 288-95.McCawley LJ and Matrisian LM (2000) Matrix metalloproteinases: multifunctional contributors to tumor

progression. Mol Med Today 6: 149-56.McQuibban GA, Gong JH, Tam EM, McCulloch CA, Clark-Lewis I and Overall CM (2000) Inflammation

dampened by gelatinase A cleavage of monocyte chemoattractant protein-3. Science 289: 1202-6.

83

Medina OP, Söderlund T, Laakkonen LJ, Tuominen EK, Koivunen E and Kinnunen PK (2001) Binding of novelpeptide inhibitors of type IV collagenases to phospholipid membranes and use in liposome targeting to tumorcells in vitro. Cancer Res 61: 3978-85.

Mitchell PG, Magna HA, Reeves LM, Lopresti-Morrow LL, Yocum SA, Rosner PJ, Geoghegan KF and HamborJE (1996) Cloning, expression, and type II collagenolytic activity of matrix metalloproteinase-13 from humanosteoarthritic cartilage. J Clin Invest 97: 761-8.

Miyazaki K, Kikkawa Y, Nakamura A, Yasumitsu H and Umeda M (1993) A large cell-adhesive scatter factorsecreted by human gastric carcinoma cells. Proc Natl Acad Sci U S A 90: 11767-71.

Moilanen M, Pirilä E, Grenman R, Sorsa T and Salo T (2002) Expression and regulation of collagenase-2(MMP-8) in head and neck squamous cell carcinomas. J Pathol 197: 72-81.

Moilanen M, Sorsa T, Stenman M, Nyberg P, Lindy O, Vesterinen J, Paju A, Konttinen YT, Stenman UH andSalo T (2003) Tumor-associated trypsinogen-2 (trypsinogen-2) activates procollagenases (MMP-1, -8, -13)and stromelysin-1 (MMP-3) and degrades type I collagen. Biochemistry 42: 5414-20.

Mozdzanowski J, Hembach P and Speicher DW (1992) High yield electroblotting onto polyvinylidene difluoridemembranes from polyacrylamide gels. Electrophoresis 13: 59-64.

Murphy G, Stanton H, Cowell S, Butler G, Knäuper V, Atkinson S and Gavrilovic J (1999) Mechanisms for promatrix metalloproteinase activation. Apmis 107: 38-44.

Murphy G, Willenbrock F, Ward RV, Cockett MI, Eaton D and Docherty AJ (1992) The C-terminal domain of 72kDa gelatinase A is not required for catalysis, but is essential for membrane activation and modulatesinteractions with tissue inhibitors of metalloproteinases. Biochem J 283: 637-41.

Myllyharju J and Kivirikko KI (2001) Collagens and collagen-related diseases. Ann Med 33: 7-21.Mäkelä M, Salo T, Uitto VJ and Larjava H (1994) Matrix metalloproteinases (MMP-2 and MMP-9) of the oral

cavity: cellular origin and relationship to periodontal status. J Dent Res 73: 1397-406.Mäkelä M, Sorsa T, Uitto VJ, Salo T, Teronen O and Larjava H (1998) The effects of chemically modified

tetracyclines (CMTs) on human keratinocyte proliferation and migration. Adv Dent Res 12: 131-5.Mäntyla P, Stenman M, Kinane DF, Tikanoja S, Luoto H, Salo T and Sorsa T (2003) Gingival crevicular fluid

collagenase-2 (MMP-8) test stick for chair-side monitoring of periodontitis. J Periodontal Res 38: 436-9.Määttä M, Soini Y, Pääkko P, Salo S, Tryggvason K and Autio-Harmainen H (1999) Expression of the laminin

gamma2 chain in different histological types of lung carcinoma. A study by immunohistochemistry and in situhybridization. J Pathol 188: 361-8.

Nabeshima K, Inoue T, Shimao Y and Sameshima T (2002) Matrix metalloproteinases in tumor invasion: Rolefor cell migration. Pathol Int 52: 255-64.

Nagase H (1997) Activation mechanisms of matrix metalloproteinases. Biol Chem 378: 151-60.Nagase H and Woessner JF, Jr. (1999) Matrix metalloproteinases. J Biol Chem 274: 21491-4.Nakahara H, Howard L, Thompson EW, Sato H, Seiki M, Yeh Y and Chen WT (1997)

Transmembrane/cytoplasmic domain-mediated membrane type 1-matrix metalloprotease docking toinvadopodia is required for cell invasion. Proc Natl Acad Sci U S A 94: 7959-64.

Nakamura H, Fujii Y, Ohuchi E, Yamamoto E and Okada Y (1998) Activation of the precursor of humanstromelysin 2 and its interactions with other matrix metalloproteinases. Eur J Biochem 253: 67-75.

Netzel-Arnett S, Mitola DJ, Yamada SS, Chrysovergis K, Holmbeck K, Birkedal-Hansen H and Bugge TH (2002)Collagen dissolution by keratinocytes requires cell surface plasminogen activation and matrixmetalloproteinase activity. J Biol Chem 277: 45154-61.

Netzel-Arnett S, Sang QX, Moore WG, Navre M, Birkedal-Hansen H and Van Wart HE (1993) Comparativesequence specificities of human 72- and 92-kDa gelatinases (type IV collagenases) and PUMP (matrilysin).Biochemistry 32: 6427-32.

Nguyen BP, Gil SG and Carter WG (2000a) Deposition of laminin 5 by keratinocytes regulates integrin adhesionand signaling. J Biol Chem 275: 31896-907.

Nguyen BP, Ryan MC, Gil SG and Carter WG (2000b) Deposition of laminin 5 in epidermal wounds regulatesintegrin signaling and adhesion. Curr Opin Cell Biol 12: 554-62.

Nwomeh BC, Liang HX, Cohen IK and Yager DR (1999) MMP-8 is the predominant collagenase in healingwounds and nonhealing ulcers. J Surg Res 81: 189-95.

Nwomeh BC, Liang HX, Diegelmann RF, Cohen IK and Yager DR (1998a) Dynamics of the matrixmetalloproteinases MMP-1 and MMP-8 in acute open human dermal wounds. Wound Repair Regen 6: 127-34.

Nwomeh BC, Yager DR and Cohen IK (1998b) Physiology of the chronic wound. Clin Plast Surg 25: 341-56.Nyberg P, Heikkilä P, Sorsa T, Luostarinen J, Heljasvaara R, Stenman UH, Pihlajaniemi T and Salo T (2003)

Endostatin inhibits human tongue carcinoma cell invasion and intravasation and blocks the activation ofmatrix metalloprotease-2, -9, and -13. J Biol Chem 278: 22404-11.

Ogata Y, Enghild JJ and Nagase H (1992) Matrix metalloproteinase 3 (stromelysin) activates the precursor forthe human matrix metalloproteinase 9. J Biol Chem 267: 3581-4.

84

O'Grady RL, Nethery A and Hunter N (1984) A fluorescent screening assay for collagenase using collagenlabeled with 2-methoxy-2,4-diphenyl-3(2H)-furanone. Anal Biochem 140: 490-4.

Oikarinen A, Kylmäniemi M, Autio-Harmainen H, Autio P and Salo T (1993) Demonstration of 72-kDa and 92-kDa forms of type IV collagenase in human skin: variable expression in various blistering diseases, inductionduring re-epithelialization, and decrease by topical glucocorticoids. J Invest Dermatol 101: 205-10.

Okada A, Tomasetto C, Lutz Y, Bellocq JP, Rio MC and Basset P (1997) Expression of matrixmetalloproteinases during rat skin wound healing: evidence that membrane type-1 matrix metalloproteinaseis a stromal activator of pro-gelatinase A. J Cell Biol 137: 67-77.

Oksanen J, Sorokin LM, Virtanen and Hormia M (2001) The junctional epithelium around murine teeth differsfrom gingival epithelium in its basement membrane composition. J Dent Res 80: 2093-7.

Okumura Y, Sato H, Seiki M and Kido H (1997) Proteolytic activation of the precursor of membrane type 1matrix metalloproteinase by human plasmin. A possible cell surface activator. FEBS Lett 402: 181-4.

Olsen J, Lefebvre O, Fritsch C, Troelsen JT, Orian-Rousseau V, Kedinger M and Simon-Assmann P (2000)Involvement of activator protein 1 complexes in the epithelium-specific activation of the laminin gamma2-chain gene promoter by hepatocyte growth factor (scatter factor). Biochem J 347: 407-17.

O'Toole EA, Marinkovich MP, Hoeffler WK, Furthmayr H and Woodley DT (1997) Laminin-5 inhibits humankeratinocyte migration. Exp Cell Res 233: 330-9.

Overall CM and Lopez-Otin C (2002) Strategies for MMP inhibition in cancer: innovations for the post-trial era.Nat Rev Cancer 2: 657-72.

Palosaari H, Wahlgren J, Larmas M, Rönkä H, Sorsa T, Salo T and Tjäderhane L (2000) The expression ofMMP-8 in human odontoblasts and dental pulp cells is down-regulated by TGF-beta1. J Dent Res 79: 77-84.

Park HI, Ni J, Gerkema FE, Liu D, Belozerov VE and Sang QX (2000) Identification and characterization ofhuman endometase (Matrix metalloproteinase-26) from endometrial tumor. J Biol Chem 275: 20540-4.

Pavlaki M, Cao J, Hymowitz M, Chen WT, Bahou W and Zucker S (2002) A conserved sequence within thepropeptide domain of membrane type 1 matrix metalloproteinase is critical for function as an intramolecularchaperone. J Biol Chem 277: 2740-9.

Pei D, Kang T and Qi H (2000) Cysteine array matrix metalloproteinase (CA-MMP)/MMP-23 is a type IItransmembrane matrix metalloproteinase regulated by a single cleavage for both secretion and activation. JBiol Chem 275: 33988-97.

Pei D and Weiss SJ (1995) Furin-dependent intracellular activation of the human stromelysin-3 zymogen.Nature 375: 244-7.

Pendas AM, Knäuper V, Puente XS, Llano E, Mattei MG, Apte S, Murphy G and Lopez-Otin C (1997)Identification and characterization of a novel human matrix metalloproteinase with unique structuralcharacteristics, chromosomal location, and tissue distribution. J Biol Chem 272: 4281-6.

Pilcher BK, Dumin J, Schwartz MJ, Mast BA, Schultz GS, Parks WC and Welgus HG (1999) Keratinocytecollagenase-1 expression requires an epidermal growth factor receptor autocrine mechanism. J Biol Chem274: 10372-81.

Pilcher BK, Dumin JA, Sudbeck BD, Krane SM, Welgus HG and Parks WC (1997) The activity of collagenase-1is required for keratinocyte migration on a type I collagen matrix. J Cell Biol 137: 1445-57.

Pilcher BK, Sudbeck BD, Dumin JA, Welgus HG and Parks WC (1998) Collagenase-1 and collagen inepidermal repair. Arch Dermatol Res 290 Suppl: S37-46.

Plopper G, Falk-Marzillier J, Glaser S, Fitchmun M, Giannelli G, Romano T, Jones JC and Quaranta V (1996)Changes in expression of monoclonal antibody epitopes on laminin-5r induced by cell contact. J Cell Sci109: 1965-73.

Prikk K, Maisi P, Pirilä E, Sepper R, Salo T, Wahlgren J and Sorsa T (2001) In vivo collagenase-2 (MMP-8)expression by human bronchial epithelial cells and monocytes/macrophages in bronchiectasis. J Pathol 194:232-8.

Pulkkinen L and Uitto J (1999) Mutation analysis and molecular genetics of epidermolysis bullosa. Matrix Biol18: 29-42.

Pyke C, Romer J, Kallunki P, Lund LR, Ralfkiaer E, Dano K and Tryggvason K (1994) The gamma 2 chain ofkalinin/laminin 5 is preferentially expressed in invading malignant cells in human cancers. Am J Pathol 145:782-91.

Pyke C, Salo S, Ralfkiaer E, Romer J, Dano K and Tryggvason K (1995) Laminin-5 is a marker of invadingcancer cells in some human carcinomas and is coexpressed with the receptor for urokinase plasminogenactivator in budding cancer cells in colon adenocarcinomas. Cancer Res 55: 4132-9.

Quaranta V (2002) Motility cues in the tumor microenvironment. Differentiation 70: 590-8.Ramamurthy NS, McClain SA, Pirilä E, Maisi P, Salo T, Kucine A, Sorsa T, Vishram F and Golub LM (1999)

Wound healing in aged normal and ovariectomized rats: effects of chemically modified doxycycline (CMT-8)on MMP expression and collagen synthesis. Ann N Y Acad Sci 878: 720-3.

85

Ramamurthy NS, Pirilä E, Lee P, Maisi P, McClain SA, Kucine A, Wahlgren J, Salo T, Wolff M, Sorsa T andGolub LM (2003) Delayed wound healing, altered collagen level and keratinocyte growth factor expression inskin of ovariectomized rat is normalized by estrogen or chemically modified doxycycline. In Menopause: thestate of the art-in research and management, (ed. H. P. G. Schneider), pp. 236-243. London: ParthenonPublishing group Ltd.

Ramamurthy NS, Rifkin BR, Greenwald RA, Xu JW, Liu Y, Turner G, Golub LM and Vernillo AT (2002) Inhibitionof matrix metalloproteinase-mediated periodontal bone loss in rats: a comparison of 6 chemically modifiedtetracyclines. J Periodontol 73: 726-34.

Rasmussen HS and McCann PP (1997) Matrix metalloproteinase inhibition as a novel anticancer strategy: areview with special focus on batimastat and marimastat. Pharmacol Ther 75: 69-75.

Ray MC and Gately LE, 3rd (1996) Basement membrane zone. Clin Dermatol 14: 321-30.Rifkin BR, Vernillo AT and Golub LM (1993) Blocking periodontal disease progression by inhibiting tissue-

destructive enzymes: a potential therapeutic role for tetracyclines and their chemically-modified analogs. JPeriodontol 64: 819-27.

Roeb E, Schleinkofer K, Kernebeck T, Potsch S, Jansen B, Behrmann I, Matern S and Grotzinger J (2002) TheMMP-9 hemopexin domain is a novel gelatin binding domain and acts as an antagonist. J Biol Chem 15: 15.

Rousselle P and Aumailley M (1994) Kalinin is more efficient than laminin in promoting adhesion of primarykeratinocytes and some other epithelial cells and has a different requirement for integrin receptors. J CellBiol 125: 205-14.

Rousselle P, Keene DR, Ruggiero F, Champliaud MF, Rest M and Burgeson RE (1997) Laminin 5 binds theNC-1 domain of type VII collagen. J Cell Biol 138: 719-28.

Rousselle P, Lunstrum GP, Keene DR and Burgeson RE (1991) Kalinin: an epithelium-specific basementmembrane adhesion molecule that is a component of anchoring filaments. J Cell Biol 114: 567-76.

Rudek MA, Figg WD, Dyer V, Dahut W, Turner ML, Steinberg SM, Liewehr DJ, Kohler DR, Pluda JM and ReedE (2001) Phase I clinical trial of oral COL-3, a matrix metalloproteinase inhibitor, in patients with refractorymetastatic cancer. J Clin Oncol 19: 584-92.

Ryan MC, Lee K, Miyashita Y and Carter WG (1999a) Targeted disruption of the LAMA3 gene in mice revealsabnormalities in survival and late stage differentiation of epithelial cells. J Cell Biol 145: 1309-23.

Ryan ME, Ramamurthy NS, Sorsa T and Golub LM (1999b) MMP-mediated events in diabetes. Ann N Y AcadSci 878: 311-34.

Ryan ME, Ramamurthy S and Golub LM (1996) Matrix metalloproteinases and their inhibition in periodontaltreatment. Curr Opin Periodontol 3: 85-96.

Saarialho-Kere UK (1998) Patterns of matrix metalloproteinase and TIMP expression in chronic ulcers. ArchDermatol Res 290 Suppl: S47-54.

Saarialho-Kere UK, Kovacs SO, Pentland AP, Olerud JE, Welgus HG and Parks WC (1993) Cell-matrixinteractions modulate interstitial collagenase expression by human keratinocytes actively involved in woundhealing. J Clin Invest 92: 2858-66.

Sadowski T and Steinmeyer J (2001) Effects of tetracyclines on the production of matrix metalloproteinases andplasminogen activators as well as of their natural inhibitors, tissue inhibitor of metalloproteinases-1 andplasminogen activator inhibitor-1. Inflamm Res 50: 175-82.

Sahlberg C, Hormia M, Airenne T and Thesleff I (1998) Laminin gamma2 expression is developmentallyregulated during murine tooth morphogenesis and is intense in ameloblasts. J Dent Res 77: 1589-96.

Salo S, Haakana H, Kontusaari S, Hujanen E, Kallunki T and Tryggvason K (1999) Laminin-5 promotesadhesion and migration of epithelial cells: identification of a migration-related element in the gamma2 chaingene (LAMC2) with activity in transgenic mice. Matrix Biol 18: 197-210.

Salo T, Liotta LA and Tryggvason K (1983) Purification and characterization of a murine basement membranecollagen-degrading enzyme secreted by metastatic tumor cells. J Biol Chem 258: 3058-63.

Salo T, Lyons JG, Rahemtulla F, Birkedal-Hansen H and Larjava H (1991) Transforming growth factor-beta 1up-regulates type IV collagenase expression in cultured human keratinocytes. J Biol Chem 266: 11436-41.

Salo T, Makela M, Kylmaniemi M, Autio-Harmainen H and Larjava H (1994) Expression of matrixmetalloproteinase-2 and -9 during early human wound healing. Lab Invest 70: 176-82.

Sang QA, Bodden MK and Windsor LJ (1996) Activation of human progelatinase A by collagenase andmatrilysin: activation of procollagenase by matrilysin. J Protein Chem 15: 243-53.

Sang QX, Birkedal-Hansen H and Van Wart HE (1995) Proteolytic and non-proteolytic activation of humanneutrophil progelatinase B. Biochim Biophys Acta 1251: 99-108.

Sank A, Chi M, Shima T, Reich R and Martin GR (1989) Increased calcium levels alter cellular and molecularevents in wound healing. Surgery 106: 1141-7; discussion 1147-8.

Santavicca M, Noel A, Angliker H, Stoll I, Segain JP, Anglard P, Chretien M, Seidah N and Basset P (1996)Characterization of structural determinants and molecular mechanisms involved in pro-stromelysin-3activation by 4-aminophenylmercuric acetate and furin-type convertases. Biochem J 315: 953-8.

86

Sarment DP, Korostoff J, D'Angelo M, Polson AM, Feldman RS and Billings PC (1999) In situ localization andcharacterization of active proteases in chronically inflamed and healthy human gingival tissues. J Periodontol70: 1303-12.

Sasaki T, Gohring W, Mann K, Brakebusch C, Yamada Y, Fassler R and Timpl R (2001) Short arm region oflaminin-5 gamma2 chain: structure, mechanism of processing and binding to heparin and proteins. J Mol Biol314: 751-63.

Sato H, Takino T, Kinoshita T, Imai K, Okada Y, Stetler Stevenson WG and Seiki M (1996) Cell surface bindingand activation of gelatinase A induced by expression of membrane-type-1-matrix metalloproteinase (MT1-MMP). FEBS Lett 385: 238-40.

Sato H, Takino T, Okada Y, Cao J, Shinagawa A, Yamamoto E and Seiki M (1994) A matrix metalloproteinaseexpressed on the surface of invasive tumour cells. Nature 370: 61-5.

Scharffetter K, Kulozik M, Stolz W, Lankat-Buttgereit B, Hatamochi A, Sohnchen R and Krieg T (1989)Localization of collagen alpha 1(I) gene expression during wound healing by in situ hybridization. J InvestDermatol 93: 405-12.

Schenk S, Hintermann E, Bilban M, Koshikawa N, Hojilla C, Khokha R and Quaranta V (2003) Binding to EGFreceptor of a laminin-5 EGF-like fragment liberated during MMP-dependent mammary gland involution. J CellBiol 161: 197-209.

Schnaper HW, McGowan KA, Kim-Schulze S and Cid MC (1996) Oestrogen and endothelial cell angiogenicactivity. Clin Exp Pharmacol Physiol 23: 247-50.

Seftor RE, Seftor EA, Kirschmann DA and Hendrix MJ (2002) Targeting the tumor microenvironment withchemically modified tetracyclines: inhibition of laminin 5 gamma2 chain promigratory fragments andvasculogenic mimicry. Mol Cancer Ther 1: 1173-9.

Sgadari C, Barillari G, Toschi E, Carlei D, Bacigalupo I, Baccarini S, Palladino C, Leone P, Bugarini R, MalavasiL et al. (2002) HIV protease inhibitors are potent anti-angiogenic molecules and promote regression ofKaposi sarcoma. Nat Med 8: 225-32.

Shah MG and Maibach HI (2001) Estrogen and skin. An overview. Am J Clin Dermatol 2: 143-50.Shang M, Koshikawa N, Schenk S and Quaranta V (2001) The LG3 module of laminin-5 harbors a binding site

for integrin alpha3beta1 that promotes cell adhesion, spreading, and migration. J Biol Chem 276: 33045-53.Shapiro LE, Knowles SR and Shear NH (1997) Comparative safety of tetracycline, minocycline, and

doxycycline. Arch Dermatol 133: 1224-30.Sheu BC, Hsu SM, Ho HN, Lien HC, Huang SC and Lin RH (2001) A novel role of metalloproteinase in cancer-

mediated immunosuppression. Cancer Res 61: 237-42.Silletti S, Kessler T, Goldberg J, Boger DL and Cheresh DA (2001) Disruption of matrix metalloproteinase 2

binding to integrin alpha vbeta 3 by an organic molecule inhibits angiogenesis and tumor growth in vivo. ProcNatl Acad Sci U S A 98: 119-24.

Skyldberg B, Salo S, Eriksson E, Aspenblad U, Moberger B, Tryggvason K and Auer G (1999) Laminin-5 as amarker of invasiveness in cervical lesions. J Natl Cancer Inst 91: 1882-7.

Smith GN, Jr., Brandt KD and Hasty KA (1996) Activation of recombinant human neutrophil procollagenase inthe presence of doxycycline results in fragmentation of the enzyme and loss of enzyme activity. ArthritisRheum 39: 235-44.

Smith GN, Jr., Mickler EA, Hasty KA and Brandt KD (1999) Specificity of inhibition of matrix metalloproteinaseactivity by doxycycline: relationship to structure of the enzyme. Arthritis Rheum 42: 1140-6.

Sorsa T, Ding Y, Salo T, Lauhio A, Teronen O, Ingman T, Ohtani H, Andoh N, Takeha S and Konttinen YT(1994) Effects of tetracyclines on neutrophil, gingival, and salivary collagenases. A functional and western-blot assessment with special reference to their cellular sources in periodontal diseases. Ann N Y Acad Sci732: 112-31.

Sorsa T, Ingman T, Suomalainen K, Haapasalo M, Konttinen YT, Lindy O, Saari H and Uitto VJ (1992)Identification of proteases from periodontopathogenic bacteria as activators of latent human neutrophil andfibroblast-type interstitial collagenases. Infect Immun 60: 4491-5.

Sorsa T, Mäntyla P, Rönkä H, Kallio P, Kallis GB, Lundqvist C, Kinane DF, Salo T, Golub LM, Teronen O andTikanoja S (1999) Scientific basis of a matrix metalloproteinase-8 specific chair-side test for monitoringperiodontal and peri-implant health and disease. Ann N Y Acad Sci 878: 130-40.

Sorsa T, Ramamurthy NS, Vernillo AT, Zhang X, Konttinen YT, Rifkin BR and Golub LM (1998) Functional sitesof chemically modified tetracyclines: inhibition of the oxidative activation of human neutrophil and chickenosteoclast pro- matrix metalloproteinases. J Rheumatol 25: 975-82.

Sorsa T, Salo T, Koivunen E, Tyynelä J, Konttinen YT, Bergmann U, Tuuttila A, Niemi E, Teronen O, Heikkilä P,Tschesche H, Leinonen J, Osman S and Stenman UH (1997) Activation of type IV procollagenases byhuman tumor-associated trypsin- 2. J Biol Chem 272: 21067-74.

Sorsa T, Suomalainen K, Turto H and Lindy S (1987) Latent human leukocyte collagenase can be activated bygold thioglucose and gold sodium thiomalate, but not by auranofin. Biosci Rep 7: 965-8.

87

Sorsa T, Uitto VJ, Suomalainen K, Vauhkonen M and Lindy S (1988) Comparison of interstitial collagenasesfrom human gingiva, sulcular fluid and polymorphonuclear leukocytes. J Periodontal Res 23: 386-93.

Sternlicht MD and Werb Z (2001) How matrix metalloproteinases regulate cell behavior. Annu Rev Cell Dev Biol17: 463-516.

Stetler-Stevenson WG (1996) Dynamics of matrix turnover during pathologic remodeling of the extracellularmatrix. Am J Pathol 148: 1345-50.

Stetler-Stevenson WG, Krutzsch HC, Wacher MP, Margulies IM and Liotta LA (1989) The activation of humantype IV collagenase proenzyme. Sequence identification of the major conversion product followingorganomercurial activation. J Biol Chem 264: 1353-6.

Stolow MA, Bauzon DD, Li J, Sedgwick T, Liang VC, Sang QA and Shi YB (1996) Identification andcharacterization of a novel collagenase in Xenopus laevis: possible roles during frog development. Mol BiolCell 7: 1471-83.

Strongin AY, Collier I, Bannikov G, Marmer BL, Grant GA and Goldberg GI (1995) Mechanism of cell surfaceactivation of 72-kDa type IV collagenase. Isolation of the activated form of the membrane metalloprotease. JBiol Chem 270: 5331-8.

Sugiyama S, Utani A, Yamada S, Kozak CA and Yamada Y (1995) Cloning and expression of the mouselaminin gamma 2 (B2t) chain, a subunit of epithelial cell laminin. Eur J Biochem 228: 120-8.

Suomalainen K, Sorsa T, Golub LM, Ramamurthy N, Lee HM, Uitto VJ, Saari H and Konttinen YT (1992)Specificity of the anticollagenase action of tetracyclines: relevance to their anti-inflammatory potential.Antimicrob Agents Chemother 36: 227-9.

Swindle CS, Tran KT, Johnson TD, Banerjee P, Mayes AM, Griffith L and Wells A (2001) Epidermal growthfactor (EGF)-like repeats of human tenascin-C as ligands for EGF receptor. J Cell Biol 154: 459-68.

Takahashi M, Tsunoda T, Seiki M, Nakamura Y and Furukawa Y (2002) Identification of membrane-type matrixmetalloproteinase-1 as a target of the beta-catenin/Tcf4 complex in human colorectal cancers. Oncogene 21:5861-7.

Tani T, Lumme A, Linnala A, Kivilaakso E, Kiviluoto T, Burgeson RE, Kangas L, Leivo I and Virtanen I (1997)Pancreatic carcinomas deposit laminin-5, preferably adhere to laminin- 5, and migrate on the newlydeposited basement membrane. Am J Pathol 151: 1289-302.

Tervahartiala T, Pirilä E, Ceponis A, Maisi P, Salo T, Tuter G, Kallio P, Törnwall J, Srinivas R, Konttinen YT andSorsa T (2000) The in vivo expression of the collagenolytic matrix metalloproteinases (MMP-2, -8, -13, and -14) and matrilysin (MMP-7) in adult and localized juvenile periodontitis. J Dent Res 79: 1969-77.

Trengove NJ, Stacey MC, MacAuley S, Bennett N, Gibson J, Burslem F, Murphy G and Schultz G (1999)Analysis of the acute and chronic wound environments: the role of proteases and their inhibitors. WoundRepair Regen 7: 442-52.

Trowbridge HO and Emling RC (1989) Inflammation: a review of the process. Chicago: Quintessence PublishingCo., Inc.

Tschesche H, Bakowski B, Schettler A, Knäuper V and Reinke H (1991) Leukodiapedesis, release of PMNleukocyte proteinases and activation of PMNL procollagenase. Biomed Biochim Acta 50: 755-61.

Tsubota Y, Mizushima H, Hirosaki T, Higashi S, Yasumitsu H and Miyazaki K (2000) Isolation and activity ofproteolytic fragment of laminin-5 alpha3 chain. Biochem Biophys Res Commun 278: 614-20.

Udayakumar TS, Chen ML, Bair EL, Von Bredow DC, Cress AE, Nagle RB and Bowden GT (2003) Membranetype-1-matrix metalloproteinase expressed by prostate carcinoma cells cleaves human laminin-5 beta3 chainand induces cell migration. Cancer Res 63: 2292-9.

Uitto VJ, Airola K, Vaalamo M, Johansson N, Putnins EE, Firth JD, Salonen J, Lopez-Otin C, Saarialho-Kere Uand Kähäri VM (1998) Collagenase-3 (matrix metalloproteinase-13) expression is induced in oral mucosalepithelium during chronic inflammation. Am J Pathol 152: 1489-99.

Uria JA and Lopez-Otin C (2000) Matrilysin-2, a new matrix metalloproteinase expressed in human tumors andshowing the minimal domain organization required for secretion, latency, and activity. Cancer Res 60: 4745-51.

Utani A, Nomizu M, Matsuura H, Kato K, Kobayashi T, Takeda U, Aota S, Nielsen PK and Shinkai H (2001) Aunique sequence of the laminin alpha 3 G domain binds to heparin and promotes cell adhesion throughsyndecan-2 and -4. J Biol Chem 276: 28779-88.

Vaalamo M, Mattila L, Johansson N, Kariniemi AL, Karjalainen-Lindsberg ML, Kähäri VM and Saarialho-Kere U(1997) Distinct populations of stromal cells express collagenase-3 (MMP-13) and collagenase-1 (MMP-1) inchronic ulcers but not in normally healing wounds. J Invest Dermatol 109: 96-101.

Wahlgren J, Maisi P, Sorsa T, Sutinen M, Tervahartiala T, Pirilä E, Teronen O, Hietanen J, Tjäderhane L andSalo T (2001) Expression and induction of collagenases (MMP-8 and -13) in plasma cells associated withbone-destructive lesions. J Pathol 194: 217-24.

88

Vailly J, Verrando P, Champliaud MF, Gerecke D, Wagman DW, Baudoin C, Aberdam D, Burgeson R, Bauer Eand Ortonne JP (1994) The 100-kDa chain of nicein/kalinin is a laminin B2 chain variant. Eur J Biochem 219:209-18.

Wall SJ, Bevan D, Thomas DW, Harding KG, Edwards DR and Murphy G (2002) Differential expression ofmatrix metalloproteinases during impaired wound healing of the diabetes mouse. J Invest Dermatol 119: 91-8.

Van den Steen PE, Dubois B, Nelissen I, Rudd PM, Dwek RA and Opdenakker G (2002) Biochemistry andmolecular biology of gelatinase B or matrix metalloproteinase-9 (MMP-9). Crit Rev Biochem Mol Biol 37: 375-536.

Van Dyke TE and Serhan CN (2003) Resolution of inflammation: a new paradigm for the pathogenesis ofperiodontal diseases. J Dent Res 82: 82-90.

Van Wart HE and Birkedal-Hansen H (1990) The cysteine switch: a principle of regulation of metalloproteinaseactivity with potential applicability to the entire matrix metalloproteinase gene family. Proc Natl Acad Sci U SA 87: 5578-82.

Vazquez F, Palacios S, Aleman N and Guerrero F (1996) Changes of the basement membrane and type IVcollagen in human skin during aging. Maturitas 25: 209-15.

Veitch DP, Nokelainen P, McGowan KA, Nguyen TT, Nguyen NE, Stephenson R, Pappano WN, Keene DR,Spong SM, Greenspan DS, Findell PR and Marinkovich MP (2002) Mammalian Tolloid metalloproteinase,and not matrix metalloprotease 2 nor membrane type 1 metalloprotease, processes laminin-5 inkeratinocytes and skin. J Biol Chem 7: 7.

Velasco G, Pendas AM, Fueyo A, Knäuper V, Murphy G and Lopez-Otin C (1999) Cloning and characterizationof human MMP-23, a new matrix metalloproteinase predominantly expressed in reproductive tissues andlacking conserved domains in other family members. J Biol Chem 274: 4570-6.

Verrecchia F and Mauviel A (2002) Transforming growth factor-beta signaling through the Smad pathway: rolein extracellular matrix gene expression and regulation. J Invest Dermatol 118: 211-5.

Westerlund U, Ingman T, Lukinmaa PL, Salo T, Kjeldsen L, Borregaard N, Tjäderhane L, Konttinen YT andSorsa T (1996) Human neutrophil gelatinase and associated lipocalin in adult and localized juvenileperiodontitis. J Dent Res 75: 1553-63.

Whitelock JM, Murdoch AD, Iozzo RV and Underwood PA (1996) The degradation of human endothelial cell-derived perlecan and release of bound basic fibroblast growth factor by stromelysin, collagenase, plasmin,and heparanases. J Biol Chem 271: 10079-86.

Whittaker M, Floyd CD, Brown P and Gearing AJ (1999) Design and therapeutic application of matrixmetalloproteinase inhibitors. Chem Rev 99: 2735-76.

Will H, Atkinson SJ, Butler GS, Smith B and Murphy G (1996) The soluble catalytic domain of membrane type 1matrix metalloproteinase cleaves the propeptide of progelatinase A and initiates autoproteolytic activation.Regulation by TIMP-2 and TIMP-3. J Biol Chem 271: 17119-23.

Willenbrock F, Murphy G, Phillips IR and Brocklehurst K (1995) The second zinc atom in the matrixmetalloproteinase catalytic domain is absent in the full-length enzymes: a possible role for the C-terminaldomain. FEBS Lett 358: 189-92.

Williams D, Hughes F, Odell E and Farthing P (1992) Pathology of periodontal disease, pp. 40-96. Oxford:Oxford University Press.

Wilson CL and Matrisian LM (1996) Matrilysin: an epithelial matrix metalloproteinase with potentially novelfunctions. Int J Biochem Cell Biol 28: 123-36.

Wilson CL, Ouellette AJ, Satchell DP, Ayabe T, Lopez-Boado YS, Stratman JL, Hultgren SJ, Matrisian LM andParks WC (1999) Regulation of intestinal alpha-defensin activation by the metalloproteinase matrilysin ininnate host defense. Science 286: 113-7.

Witte MB, Thornton FJ, Kiyama T, Efron DT, Schulz GS, Moldawer LL and Barbul A (1998) Metalloproteinaseinhibitors and wound healing: a novel enhancer of wound strength. Surgery 124: 464-70.

Vu TH, Shipley JM, Bergers G, Berger JE, Helms JA, Hanahan D, Shapiro SD, Senior RM and Werb Z (1998)MMP-9/gelatinase B is a key regulator of growth plate angiogenesis and apoptosis of hypertrophicchondrocytes. Cell 93: 411-22.

Vu TH and Werb Z (2000) Matrix metalloproteinases: effectors of development and normal physiology. GenesDev 14: 2123-33.

Väisänen A, Tuominen H, Kallioinen M and Turpeenniemi-Hujanen T (1996) Matrix metalloproteinase-2 (72 kDtype IV collagenase) expression occurs in the early stage of human melanocytic tumour progression andmay have prognostic value. J Pathol 180: 283-9.

Väänänen A, Srinivas R, Parikka M, Palosaari H, Bartlett JD, Iwata K, Grenman R, Stenman UH, Sorsa T andSalo T (2001) Expression and regulation of MMP-20 in human tongue carcinoma cells. J Dent Res 80: 1884-9.

89

Xu J, Rodriguez D, Petitclerc E, Kim JJ, Hangai M, Moon YS, Davis GE, Brooks PC and Yuen SM (2001)Proteolytic exposure of a cryptic site within collagen type IV is required for angiogenesis and tumor growth invivo. J Cell Biol 154: 1069-79.

Yager DR, Zhang LY, Liang HX, Diegelmann RF and Cohen IK (1996) Wound fluids from human pressureulcers contain elevated matrix metalloproteinase levels and activity compared to surgical wound fluids. JInvest Dermatol 107: 743-8.

Yamaguchi Y and Yoshikawa K (2001) Cutaneous wound healing: an update. J Dermatol 28: 521-34.Yana I and Weiss SJ (2000) Regulation of membrane type-1 matrix metalloproteinase activation by proprotein

convertases. Mol Biol Cell 11: 2387-401.Ye S (2000) Polymorphism in matrix metalloproteinase gene promoters: implication in regulation of gene

expression and susceptibility of various diseases. Matrix Biol 19: 623-9.Yu Q and Stamenkovic I (2000) Cell surface-localized matrix metalloproteinase-9 proteolytically activates TGF-

beta and promotes tumor invasion and angiogenesis. Genes Dev 14: 163-76.Yu Z, Ramamurthy NS, Leung M, Chang KM, McNamara TF and Golub LM (1993) Chemically-modified

tetracycline normalizes collagen metabolism in diabetic rats: a dose-response study. J Periodontal Res 28:420-8.

Zhang K and Kramer RH (1996) Laminin 5 deposition promotes keratinocyte motility. Exp Cell Res 227: 309-22.Zucker S, Drews M, Conner C, Foda HD, DeClerck YA, Langley KE, Bahou WF, Docherty AJ and Cao J (1998)

Tissue inhibitor of metalloproteinase-2 (TIMP-2) binds to the catalytic domain of the cell surface receptor,membrane type 1-matrix metalloproteinase 1 (MT1-MMP). J Biol Chem 273: 1216-22.

Ågren MS, Mirastschijski U, Karlsmark T and Saarialho-Kere UK (2001) Topical synthetic inhibitor of matrixmetalloproteinases delays epidermal regeneration of human wounds. Exp Dermatol 10: 337-48.

90

91

Original publications

Reprinted with the kind permission from Bentham Science Publishers Ltd (Current MedicinalChemistry), Blackwell Publishing (Wound Repair and Regeneration) and Elsevier Ltd (Biochemicaland Biophysical Research Communications, Experimental Cell Research).