Skin wound healing in diabetic b6 integrin-deficient mice

JASPER N. JACOBSEN,1,2 BJØRN STEFFENSEN,3 LARI HAKKINEN,1 KAREN A. KROGFELT2

and HANNU S. LARJAVA1

1Laboratory of Periodontal Biology, Department of Oral Biological and Medical Sciences, University of BritishColumbia, Vancouver, BC, Canada; 2Statens Serum Institut, Department of Microbiological Surveillance andResearch, Copenhagen, Denmark; and 3Department of Periodontics and Biochemistry, University of Texas

Health Science Center, San Antonio, TX, USA

Jacobsen JN, Steffensen B, Hakkinen L, Krogfelt KA, Larjava HS. Skin wound healing in diabetic b6integrin-deficient mice. APMIS 2010; 118: 753–64.

Integrin avb6 is a heterodimeric cell surface receptor, which is absent from the normal epithelium, butis expressed in wound-edge keratinocytes during re-epithelialization. However, the function of theavb6 integrin in wound repair remains unclear. Impaired wound healing in patients with diabetes con-stitutes a major clinical problem worldwide and has been associated with the accumulation ofadvanced glycated endproducts (AGEs) in the tissues. AGEs may account for aberrant interactionsbetween integrin receptors and their extracellular matrix ligands such as fibronectin (FN). In thisstudy, we compared healing of experimental excisional skin wounds in wild-type (WT) and b6-knock-out (b6) ⁄ )) mice with streptozotocin-induced diabetes. Results showed that diabetic b6) ⁄ ) mice had asignificant delay in early wound closure rate compared with diabetic WT mice, suggesting that avb6integrin may serve as a protective role in re-epithelialization of diabetic wounds. To mimic the glycosy-lated wound matrix, we generated a methylglyoxal (MG)-glycated variant of FN. Keratinocytesutilized avb6 and b1 integrins for spreading on both non-glycated and FN-MG, but their spreadingwas reduced on FN-MG. These findings indicated that glycation of FN and possibly other integrinligands could hamper keratinocyte interactions with the provisional matrix proteins during re-epitheli-alization of diabetic wounds.

Key words: Wound healing; integrins; fibronectin; diabetes mellitus; advanced glycated endproducts.

Hannu S. Larjava, Laboratory of Periodontal Biology, Department of Oral Biological and MedicalSciences, Faculty of Dentistry, University of British Columbia, 2199 Wesbrook Mall, Vancouver, BC,Canada V6T 1Z3. e-mail: larjava@interchange.ubc.ca

Diabetes mellitus affects approximately 170million people worldwide, and this number isprojected to double by the year 2030 (1). Dia-betes has been associatedwith a number of physi-ologic factors that predispose to aberrant woundhealing, such as impaired growth factor produc-tion (2–4), angiogenic response (4, 5), macro-phage function (6), collagen accumulation,epidermal barrier function, quantity of granula-tion tissue (4), keratinocyte and fibroblast migra-tion and proliferation (7), and synthesis of ECM

components and their remodeling by matrixmetalloproteinases (MMPs) (8). In addition, aprolonged state of systemic hyperglycemia is animportant factor for the molecular pathophysi-ology of diabetic complications because of thegradual build-up of advanced glycated endprod-ucts (AGEs) in the tissues. AGEs are formed byan irreversibly, non-enzymatic reaction betweenthe amino groups of tissue proteins, such ascollagen and fibronectin (FN) in the extracellularmatrix (ECM), and reducing sugars, which areabundant in hyperglycemia (9). These modifica-tions of the ECM proteins alter cell–ECMReceived 8 February 2010. Accepted 1 June 2010

APMIS 118: 753–764 � 2010 The Authors

Journal Compilation � 2010 APMIS

DOI 10.1111/j.1600-0463.2010.02654.x

753

interactions, including cell adhesion and migra-tion that are critical for wound healing (9). Itis likely that the changes are mediated by aber-rant interactions of the glycated ECM proteinswith the cell surface receptors of the integrinfamily.Integrins are heterodimeric, transmembrane

receptors that consist of an a and a b subunit(10–12). Through specific interactions withECM proteins, integrins mediate cell adhesion,migration, proliferation and survival of manycell types, including keratinocytes (13, 14). Dur-ing wound healing, keratinocyte integrins regu-late wound re-epithelialization (14, 15). Manyintegrins recognize a specific conserved tri-pep-tide motif, arginine–glycine–asparagine (RGD),which is found in a variety of ECM proteinssuch as FN, vitronectin, and tenascin-C (16,17). In cell cultures, the expression of keratino-cyte-specific avb6 integrin facilitates keratino-cyte adhesion and migration on the early woundprovisional matrix (18, 19). Furthermore, theavb6 integrin activates transforming growthfactor (TGF)-b1, an important molecule thatpromotes wound re-epithelialization by bindingto the latent TGF-b complex (20, 21). In fact,avb6 integrin may be critically important forTGF-b1 bioactivity in vivo (22). Studies of acutewound healing in young b6 integrin-deficientmice (b6) ⁄ )) and b6 integrin over-expressingmice did not detect any abnormalities in thehealing of experimental skin wounds (23, 24),suggesting that this integrin may not be neces-sary for normal wound healing. In contrast, inmice where wound healing was compromised bycorticosteroid-induced immunosuppression orby old age, b6 integrin deficiency resulted indelayed wound healing (25). In diabetic ulcers,wound closure is delayed or lacking as a resultof reduced epithelial migration. In addition, thelevel of TGF-b in diabetic wounds is reducedcompared with non-diabetic wounds (26).Therefore, we hypothesized that avb6 integrinplays a role in compromised wound healingassociated with the diabetic state. To this end,we examined the role of avb6 integrin in woundhealing in adult mice in which diabetes wasinduced by intraperitoneal injections of strepto-zotocin (STZ). Furthermore, we investigatedthe effects of glycation of the provisional woundmatrix protein FN on integrin-mediated adhe-sion of human skin keratinocytes.

MATERIALS AND METHODS

Animals

The animal studies were conducted in compliance withthe Canadian Council on Animal Care (CACC) andapproved by the University of British Columbia Ani-mal Care Committee. Forty-eight 12- to 14-month-oldmale wild-type (WT) FVB and age- and sex-matchedb6) ⁄ ) mice with the same genetic background wereused in this study (generous gift fromDr. Dean Shepp-ard, University of California, San Francisco). Eachexperimental group was subdivided into STZ-treated(WT: n = 14; b6) ⁄ ): n = 15) and vehicle-treated con-trol groups (WT: n = 9; b6) ⁄ ): n = 10).

Streptozotocin induction of diabetes in mice

Mice were maintained under fasting conditions for4 h and subsequently anesthetized with a mixture ofisoflurane and oxygen. After induction of anesthesia,mice in the experimental group received peritonealinjections of STZ (55 mg ⁄kg of body weight; SigmaAldrich Inc., St. Louis, MO, USA) in citrate buffer(P-4809; Sigma Aldrich Inc.). Animals in the controlgroup were injected with an equivalent volume ofcitrate buffer only. This treatment was carried outfor six consecutive days. Tail blood samples werecollected on a weekly basis from fasting animalsand analyzed for serum glucose level using the Accu-chek� system (Roche Diagnostics, F. Hoffmann-LaRoche Ltd, Basel, Switzerland) to verify successfuldevelopment of experimental diabetes in the STZ-treated animals. A serum glucose concentrationexceeding 17 mmol ⁄L was considered to reflect theonset of diabetes (27, 28).

Experimental wounds

Animals were anesthetized as described above, andthe dorsal skin was shaved and depilated (Veet�;Reckitt-Benckiser, NA Inc., Parsippany, NJ, USA).Four full-thickness, 4-mm excisional wounds werecreated on the dorsal skin of each mouse using a ster-ile, disposable biopsy punch (Miltex, Inc., York, PA,USA). After wounding, mice were maintained inseparate cages and had access to food and waterad libitum.To assess the rate of wound healing, images of all

wounds were recorded at the time of wounding (day0) and on a daily basis post-wounding using a digitalcamera (Nikon Coolpix 995; Nikon, Tokyo, Japan).All wound pictures were standardized according to ameasurement ruler included in the images. Thewound surface areas were quantified using the NIHImageJ program (National Institute of Health; http://rsb.info.nih.gov/ij/), and the healing was expressedas a percentage of the initial wound area on day 0.

JACOBSEN et al.

754 � 2010 The Authors Journal Compilation � 2010 APMIS

A total of n = 36–56 wounds per group were ana-lyzed from each time point.

Histologic and immunohistochemical examination of

wounds

For histologic and immunohistochemical analyses,tissue biopsies were obtained from the wound sites ondays 5 and 10 after wounding. To this end, mice wereeuthanized by CO2 inhalation and wound biopsiescontaining the complete wounded area as well asadjacent wound margin and normal skin were har-vested using a 6-mm sterile biopsy punch. The tissuesamples were mounted in Tissue-TEK (Sakura, Tor-rance, CA, USA) and snap-frozen on dry ice immedi-ately after collection. Frozen sections (6 lm) wereprepared and stored at )80 �C until use. For orienta-tion and to localize mid-wound sections, every tenthslide was stained with Harris’ hematoxylin and eosin(H&E). Only the mid-wound sections were used forcomparative analyses. Epithelial gap distance wasmeasured as the distance between epithelial tonguesin the wound cross-sections using NIH ImageJ, andhistology was scored from 1 (lowest) to 4 (highest)based on previously used criteria (29). A subset of5-day-old wounds was fixed with 10% buffered neu-tral formalin and stained using a modified Movat’spentachrome staining procedure. Briefly, the fixedsections were placed in Alcian blue dye overnight andrinsed in running water for 5 min. The sections werethen soaked in warm (60 �C) alkaline alcohol for60 min, followed by the standard staining procedureas described previously (30). Histologic assessment ofgranulation tissue formation used an ordinate scalefrom 1 to 4 reflecting the degree of density and orga-nization of neo-synthesized collagen fiber bundles aslow (1), medium (2), or high (3). The presence ofmature collagen among well-granulated woundsyielded a score of 4. All histologic scoring wasperformed in a blinded manner, evaluating n = 5–6wounds (two sections per wound) from five to sixmice per group.

For the detection of avb6 integrins in experimentaland control wounds, frozen sections were fixed for5 min with acetone pre-cooled to )20 �C, rinsed withphosphate-buffered saline (PBS; 137 mM NaCl,2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4 indistilled water, pH 7.4) containing 1 mg ⁄mL bovineserum albumin (BSA; Sigma Aldrich Inc.), and incu-bated with normal blocking serum (Vectastain;Vector Laboratories, Burlingame, CA, USA) at roomtemperature for 30 min. For detection of avb6 inte-grin, sections were incubated with rabbit anti-humanmonoclonal antibody b6B1 (a generous gift fromDr. Dean Sheppard, University of California, SanFrancisco, CA) diluted 1:10 dilution in PBS contain-ing 1 mg ⁄mL BSA in a humidified chamber at 4 �Covernight. After washing with PBS ⁄BSA, sections

were first incubated with an anti-rabbit biotinylatedsecondary antibody for 60 min at room temperature,and then with Vectastain� ABC reagent (Vectastain�

Elite Kit) according to the manufacturer’s guidelines.The reaction was visualized using the VIP (purplestain) substrate kit. The primary antibody was omit-ted in the negative controls, none of which showedpositive staining.

Glycation of fibronectin with methylglyoxal

Fibronectin was purified from human plasma by gela-tin-Sepharose affinity chromatography according toestablished procedures (31). Briefly, plasma wasdiluted in chromatography buffer (50 mM Tris;0.15 M NaCl, pH 7.4) and particulate matter wassedimented by centrifugation at 2500 g for 20 min.The supernatant was loaded on a gelatin-Sepharoseaffinity column (GE Healthcare Bio-sciences AB,Uppsala, Sweden). After extensive washes with chro-matography buffer alone, non-specifically boundproteins were removed with buffer containing 1 MNaCl. Gelatin-binding MMP-2 and MMP-9 wereeluted with 2% DMSO and bound FN was elutedwith 3 M urea in chromatography buffer (32). Elimi-nation of MMP-2 and -9 was verified by gelatin enzy-mography using 10% SDS-PAGE gels copolymerizedwith heat-denatured type I collagen (33, 34). PurifiedFN was quantified using the BCA assay (Pierce,Rockford, IL, USA), frozen slowly, and stored at)80 �C until analysis.For glycation, purified human FN (866 lg ⁄mL)

was incubated with 67 mM methylglyoxal (MG;Sigma Chemical Co., St. Louis, MO, USA) in PBS.The native FN control was diluted to the same con-centration in PBS without MG. Both MG-treatedand control FN were dialyzed against PBS for thesame periods of time (9). To confirm the conjugationof MG to FN molecules, an affinity-purified rabbitanti-MG-AGE polyclonal antibody (kindly providedby Dr. Shamsi, University of Cleveland, OH) (35)was used as a probe. Glycated FN and untreatedcontrol samples were separated with 7.5% SDS-PAGE gels under reducing conditions (65 mMdithiothreitol) and then transferred to polyvinylidenedifluoride membranes (Millipore Corp., Bedford,MA, USA). The membranes were blocked with10 mg ⁄mL heat-denatured BSA in 50 mM Tris, pH7.4, at 37 �C for 60 min, and then incubated withrabbit anti-MG-AGE at a 1:2000 dilution in 20 mMTris ⁄150 mM NaCl ⁄0.1% Tween 20 buffer (pH 7.4)containing 1% (w ⁄ v) BSA at 4 �C overnight. TheMG-treated protein was detected with a horseradishperoxidase-conjugated goat anti-rabbit antibody(IgG H+L) at a 1:5000 dilution in PBS ⁄Tweencontaining 5 mg ⁄mL BSA, followed by chemilumi-nescence reagent (PerkinElmer LAS, Inc., Boston,MA, USA) for 1 min. The image was captured using

DIABETIC WOUNDHEALING AND TISSUE BIOLOGY

� 2010 The Authors Journal Compilation � 2010 APMIS 755

the ChemiDoc XRS Imaging system (Bio-Rad Labo-ratories, Hercules, CA, USA).

Plate-binding assay for coating efficiency

To determine whether FN and FN-MG were coatedwith the same efficiency, the proteins were incubated in96-well plates at identical, serially diluted concentra-tions from 2 to 0.002 lg ⁄well in 0.1 M NaHCO3 ⁄Na2CO3 (pH 9.6) at 4 �C overnight. Heat-denaturedBSA (10 mg ⁄mL) in 50 mM Tris buffer (pH 7.4) wasused to block non-specific binding sites by incubatingfor 1 h at 22 �C. After thorough washing with 50 mMTris buffer containing 0.05% (v ⁄ v) Tween 20, plateswere incubated with a polyclonal rabbit anti-FN anti-body (36) for 60 min at 22 �C. Bound proteins weredetected by incubating with alkaline phosphatase-conjugated goat anti-rabbit IgG (1:1000 dilution;Bio-Rad Laboratories) for 1 h at 22 �C. After severalrinses, 1 mg ⁄mL phosphatase substrate (SigmaAldrich Inc.) was added to the wells. The reactionswere quantified at 405 nm using an Opsys MR platereader (Dynes, Chantilly, VA, USA).

Cell spreading assay

A 48-well culture plate (Falcon Multiwell�; BD Bio-sciences, Franklin Lakes, NJ, USA) was coated with5, 20, or 50 lg ⁄mL FN or FN-MG diluted in PBScontaining Ca2+ and Mg2+ and incubated for 1 h atroom temperature. Wells were then blocked with 1%heat-denatured BSA in PBS for 30 min at roomtemperature and subsequently washed twice in PBS.Sub-confluent human skin HaCaT keratinocytes (agenerous gift from Dr. Fusenig, German Cancer Cen-ter, Heidelberg, Germany) were seeded at 2.0 · 104

cells per well in serum-free Dulbecco’s modifiedEagle’s medium (Flow Laboratories, Irvine, UK)supplemented with 23 mM Na2CO3, 20 mM Hepes(Gibco�; Invitrogen Corporation, Carlsbad, CA,USA), antibiotics (50 lg ⁄mL streptomycin sulfateand 100 U ⁄mL penicillin; Sigma Aldrich, Inc.), 10%heat-inactivated fetal bovine serum (Gibco Life Tech-nologies), and 50 lM cycloheximide (Sigma Aldrich,Inc.) to prevent de novo protein synthesis. Cells wereincubated for 45 min to 3 h at 37 �C in the presenceof 5% CO2 to allow cell spreading. At indicated timepoints, the experiment was terminated by the additionof appropriate volume of 2· cell fixative (8% formal-dehyde, 10% sucrose in PBS). The wells were thenfilled with distilled water to the top and covered witha glass plate for analysis. Cells with a ring of lamellarcytoplasm were considered spread. Cell spreadingwas quantified using a phase-contrast microscope(Nikon TMS; Nikon) equipped with a grid-eyepieceas detailed previously (19). Briefly, the number of cellsshowing spreading in four randomly selected fieldsfrom each of three replicate wells was counted, and

cells that were surrounded by a ring of lamellarcytoplasm were considered spread.

Blocking of integrin function

Cells were seeded at 1.0 · 104 cells per well in 96-wellplates (Nunc, Roskilde, Denmark) coated with FN orFN-MG as in the presence or absence of monoclonalantibodies with the capacity to specifically block thefunctions of the b1 [mAb13; 20 lg ⁄mL (37)], av(L230; 20 lg ⁄ml), or b6 [10D5; 50 lg ⁄ml (38)] inte-grin subunits. Antibody L230 was purified fromculture supernatants of hybridoma cells (ATCCHB8448) in our laboratory. All antibody concentra-tions used were previously optimized to produce max-imal inhibition of cell spreading (data not shown).The cells were incubated with the anti-integrin anti-bodies for 15 min on ice prior to seeding and transferto the cell culture incubator (5% CO2; 37 �C) for celladhesion and spreading. When about 50% of cellswithout inhibitory antibodies showed spreading, allcells were fixed and cell spreading was quantified.

Statistical analysis

A Student’s t-test (two-tailed, Bonferroni-adjusted)was used to assess pairwise differences in wound heal-ing kinetics between each day of healing, whereas epi-thelial gap distances from mid-wound H&E sectionswere analyzed using a Mann–Whitney U-test (Graph-Pad Prism 5.0; GraphPad Software, Inc., La Jolla,CA, USA). For all cell spreading experiments, weused a two-way ANOVA with Bonferroni post-test todescribe differences between cell behavior on nativevs glycated FN for various concentrations of FN,time, and anti-integrin antibody blockings. A p-valueless than 0.05 was considered statistically significant.

RESULTS

STZ induction of experimental diabetes

For six consecutive days beginning on day 50prior to generating experimental wounds, adultFVB WT and b6 integrin knockout mice weretreated intraperitoneally with STZ to inducepancreatic b-cell autoimmunity and a chronicstate of systemic hyperglycemia. During these50 days, we observed an insignificant loss ofanimals in the control group, but considerablelosses among the diabetic animals, especially inthe b6) ⁄ ) group (Fig. 1A). Blood glucose mea-surements (Fig. 1B,C), each time taken after 4 hof fasting, showed that a glucose concentrationabove 17 mmol ⁄L reflecting a diabetic state (27)

JACOBSEN et al.

756 � 2010 The Authors Journal Compilation � 2010 APMIS

was achieved within the first 2 weeks of STZtreatment for both FVB and b6) ⁄ ) mice. Asteady state of severe hyperglycemia (25–30 mmol ⁄L) developed approximately 10 daysprior to wounding in STZ-treated animals,whereas all control mice maintained normalblood glucose levels of 10–11 mmol ⁄L

throughout the study period. Body weights(Fig. 1D,E) decreased only in the diabeticpopulations, a clinical sign of impaired glucoseuptake. Mice in the b6) ⁄ ) group had lower bodyweights compared with their FVB WT litter-mates. Tail neuropathy was consistent with astate of chronic hyperglycemia (Fig. 1F,G).

Wound healing kinetics

To evaluate the contribution of b6 integrin towound repair in normal and diabetic mice, wefirst quantified clinical wound healing from digi-tized images. A representative panel is shownfrom day 0, 5, and 10 from all four experimentalgroups (Fig. 2A). The final sample consisted of36–56 wounds per experimental group. Changesin size of wound areas are presented in Fig. 2B–F. Wounds in the control groups were closed byday 7 with no difference in healing rates betweenthe WT and b6) ⁄ ) mice (Fig. 2B). In compari-son, among the diabetic animals, b6) ⁄ ) miceshowed a small but significant delay in woundhealing from day 1 to 3 compared with WT, butno other differences were observed and com-plete wound closure was reached on day 9(Fig. 2C). Induction of diabetes in the WT ani-mals caused a small but significant delay inwound healing on days 3 and 4 (p < 0.05), andthe wounds closed at the same time as WT ani-mals without diabetes (Fig. 2D). Finally, thediabetic b6) ⁄ ) mice exhibited approximately awhole day delay in wound healing comparedwith b6) ⁄ ) mice without diabetes. The differencewas statistically significant from day 1 to day 6(Fig. 2E). The combined effects of the b6 knock-out and the diabetic state resulted in an overalldelay in wound closure of nearly 2 days com-pared with the WT mice (Fig. 2F).

Histology and immunohistochemistry

Histologic staining of 5-day-old wound speci-mens (Fig. 3A) confirmed the clinical wound-healing data with a significant difference inepithelial gap distance between control anddiabetic animals (p < 0.01). Deficiency of b6integrin did not lead to increased epithelial gapdistance in normal wound repair (p = 0.578),but was more influential in diabetic woundhealing, yet statistically insignificant at day 5(p = 0.112) (Fig. 3B). In concordance with

A

B C

D E

F G

Fig. 1. (A) Survival data for 50 days pre-woundingand 10 days post-wounding period of wild-type (WT;n = 9, control; n = 14, diabetic) and b6) ⁄ ) (n = 10,control; n = 15, diabetic) mice. (B) Blood glucoselevels for WT control vs diabetic animals and (C) forb6) ⁄ ) control vs b6) ⁄ ) diabetic animals. (D) Meanbody weight for control vs diabetic animals and (E)for b6) ⁄ ) control vs b6) ⁄ ) diabetic animals. (F) Rep-resentative images of mouse tails from the untreatedcontrol group compared with the streptozotocin-induced diabetes group (G) displaying clinical signsof neuropathy.

DIABETIC WOUNDHEALING AND TISSUE BIOLOGY

� 2010 The Authors Journal Compilation � 2010 APMIS 757

these observations, histologic scoring at day 5showed no difference among the controlanimals, but a reduced score for the diabeticanimals, and in particular, the b6 integrin-defi-cient group (Fig. 3C). The Movat’s modifiedpentachrome staining was carried out on a setof day 5 wounds to assess the formation of gran-ulation tissue and collagen deposition (Fig. 4A).The wounds in the control animals revealedmore mature collagen (yellow) at the connectivetissue sites than in the diabetic wounds, whereneo-synthesized collagen (light bluish-green)

seemed to dominate, indicating delayed woundhealing. Histologic scoring revealed a tendencytoward reduced formation of granulation tissueamong the diabetic animals, especially in theb6) ⁄ ) mice, but statistical significance could notbe achieved (Fig. 4B). As a control experiment,we localized the presence ⁄absence of the avb6integrin at the wound edge sites by immunohis-tochemistry (Fig. 5). The keratinocytes at thewound edges of WT animals stained positive forthe avb6 integrin in both untreated and STZ-treated mice. The b6) ⁄ ) mice all stained negativeapart from some background staining of hairfollicles in the unwounded normal skin.

Glycation of FN impairs HaCaT cell spreading in vitro

We hypothesized that the delayed woundre-epithelialization could be associated withaberrant cross-linking to important integrin-binding ligands of the wound bed matrixbecause of the formation of AGEs. Therefore,we performed an in vitro spreading assay usinghuman HaCaT keratinocytes seeded on the inte-grin-binding ligand FN. For these experiments,we used FN that had been treated with MG togenerate AGEs to mimic the diabetes-inducedhyperglycemic effect on ECM tissue proteins.Gel electrophoresis and Western blotting were

used to verify the glycation of FN by MG as ithas been reported in other studies (9, 35).MG-treated (FN-MG) and untreated FNshowed similar mobility in SDS-PAGE gels(Fig. 6A), but Western blotting using antibodieswith specificity for the MG-conjugated part ofthe FN-MG protein-sugar complex confirmedthe reaction (Fig. 6B). Importantly, glycatedand native FN showed similar coating efficiencyto ELISA plates used in the subsequent cellexperiments (Fig. 6C).The spreading frequency of HaCaT keratino-

cytes was determined upon seeding the cells onFN vs FN-MG-coated substrates. The spread-ing of HaCaT cells on FN-MG substrate wassignificantly reduced compared with the FNsubstrate (Fig. 6D). On substrates coated with10 lg ⁄mL of control FN, approximately 50%of HaCaT cells showed spreading, whereastwice the coating concentration for FN-MGwas necessary to obtain the same spreading fre-quency (Fig. 6D). In addition, a significantlyslower HaCaT cell spreading was observed on

A

B C

D

F

E

Fig. 2. (A) Representative clinical photographs ofexcisional wounds from wild-type (WT) and b6) ⁄ )

mice treated with citrate buffer only (control) or55mg ⁄kg streptozotocin in citrate buffer (diabetic) atdays 0, 5, and 10 after wounding. Wound closure ratein the control mice was similar compared withWT andb6) ⁄ ) mice, whereas a delay in wound closure couldbe observed for the diabetic WT and especially thediabetic b6) ⁄ ) mice. Scale bar: 10 mm. (B–F) Relativewound area change over time. Data show mean ±SEM per cent of wound area relative to wound size atday 0 within each group (n = 36–56 wounds pergroup). Statistical significance between the two groupsat each time point was assessed by Student’s t-test.*p < 0.05, **p < 0.01, ***p < 0.001.

JACOBSEN et al.

758 � 2010 The Authors Journal Compilation � 2010 APMIS

FN-MG compared with FN in time courseexperiments (Fig. 6E). We then investigatedwhether HaCaT keratinocytes utilized the sameintegrin receptors for binding FN-MG and FN.Incubation of cells with function-blocking anti-bodies against av and b6 integrin subunitsresulted in about 50% inhibition of cell spread-ing on both FN-MG and FN substrates(Fig. 6F). Likewise, incubation with an anti-body against the b1 integrin subunit resulted inabout 60% inhibition of cell spreading on bothsubstrates, and applying a combination of anti-bodies to av and b1 integrins resulted in morethan 90% inhibition (Fig. 6F). These resultsindicate that both av and b1 integrins can sup-port HaCaT keratinocyte spreading on FNregardless of its glycation state.

DISCUSSION

Diabetes mellitus constitutes a major clinicalproblem and is associated with many compli-cations, including impaired wound healing. Asre-epithelialization is a crucial step of tissuerepair characterized by migrating wound edge

keratinocytes over a provisional matrix torestore barrier function, we investigated the roleof the putative cell adhesion receptor avb6 inte-grin in epidermal wound healing in a diabeticmouse model. The induction of hyperglycemiain rodents using repeated injections of 55 mg ⁄kgof STZ mimics diabetes mellitus type I that iscaused by a T-cell-mediated auto-immune reac-tion against the pancreatic islet b-cells (39, 40).This approach results eventually in a chronicstage of hyperglycemia. It is different from thediabetic state caused by a single higher dose(160–250 mg ⁄kg) of STZ, which induces anacute and transient state of diabetes character-ized by direct cytotoxicity, DNA damage, andapoptosis among the pancreatic b-cells (41–43).Our observation of an increased frequencyof premature deaths among the diabetic b6-knockout mice post-wounding may be due tothe combination of a chronic hyperglycemiaand a mildly exaggerated inflammation in thelungs and skin, which is associated with theb6-deficient phenotype (17).In other diabetic rodent models, wound

healing studies are often performed using theleptin receptor-deficient db ⁄db mouse, which

A

B C

Fig. 3. (A) Hematoxylin and eosin (H&E) staining of day 5 mid-wound sections from wild-type (WT) and b6) ⁄ )

mice treated with citrate buffer (control) and 55mg ⁄kg streptozotocin in citrate buffer (diabetic). E, epithelium;GT, granulation tissue; CT, connective tissue; WC, wound crust. Dashed lines indicate epithelial gap distance.Scale bar: 1 mm. (B) Epithelial gap distances were measured from day 5 mid-wound H&E sections as the distancebetween epithelial tongues from all the four groups. Mann–Whitney U-test revealed overall significant differencesbetween control and diabetic animals (p < 0.01), and also between diabetic WT and b6) ⁄ ) mice, respectively(p < 0.05). (C) Histology was scored from 1 (lowest) to 4 (highest) based on the degree of cellular invasion, GTformation, vascularity, and re-epithelialization. n = 10–12 sections (two sections per wound) from five to sixmice per group. Scale bars in B and C show mean ± SD. **p < 0.01.

DIABETIC WOUNDHEALING AND TISSUE BIOLOGY

� 2010 The Authors Journal Compilation � 2010 APMIS 759

represents a model for type II diabetes mellituscharacterized by hyperglycemia, obesity, hyper-insulinemia, and impaired wound healing (44,45). The db ⁄db mouse is genetically diabetic andthereby negates the need for STZ administration(46). However, as our goal was to specificallyassess the role of avb6 integrin in diabeticwound healing, we decided to use the well-established model of chemically induced type Idiabetes mellitus by intraperitoneal adminis-tration of STZ (41) in b6 integrin-deficient mice.By this experimental approach, we successfullygenerated chronic hyperglycemia, which in theseanimals predisposes to multiple vascular com-plications and cutaneous dysfunctions such asimpaired wound repair due to AGE formationsimilar to humans (47).Our quantitative analysis of clinical wound

pictures revealed that the lack of b6 integrin didnot cause any significant delay in wound healingamong vehicle-treated (citrate buffer; control)mice, which is in accordance with previous stud-ies performed on young and adult, female andmale mice of the same genetic backgrounds (23,

A

B

Fig. 4. (A) Movat’s modified pentachrome stainingof representative day 5 mid-wound sections fromwild-type (WT) and b6) ⁄ ) mice treated with citratebuffer (control) and 55mg ⁄kg citrate-buffered strepto-zotocin (STZ, diabetic). Control animals (WT andb6) ⁄ )) show similar dense and well-organized neo-synthesized collagen fiber bundles (blue-green) in thegranulation tissue (GT). Small amounts of maturecollagen (yellow) appear near the epidermis (E) andat the wound edge (WE) proximal to the unwoundedconnective tissue (arrows) and clarified by magnifiedpicture insets. STZ-treated diabetic mice show lessformation of GT, especially in the case of the b6) ⁄ )

group, and no mature collagen is observed in eitherWT or b6) ⁄ ) mice. Scale bars: 100 lm. (B) The degreeof GT formation from five to six mice per group wasscored by an ordinate scale (1–4) according to thedegree of density and organization of neo-synthesizedcollagen fiber bundles as low (1), medium (2), or high(3). Presence of mature collagen among well-granu-lated wounds yielded a score of 4. No statistical differ-ence could be established, but a tendency towardreduced GT formation among the diabetic animals,especially the b6) ⁄ ) mice, was observed. Bars showmean ± SD, n = 10–12 sections (two sections perwound) from five to six wounds.

Fig. 5. Immunolocalization of b6 integrin in day 5mid-wound sections from wild-type (WT) and b6) ⁄ )

mice treated with citrate buffer (control) and55mg ⁄kg citrate-buffered streptozotocin (diabetic).b6) ⁄ ) mice showed a complete lack of b6 integrinstaining in all specimens examined, with the exceptionof limited background staining of hair follicles. In theWT mice, the immunoreactivity of b6 integrin wasconfined to the wound edge of the epithelium contain-ing populations of proliferating and migrating kerati-nocytes. WE, wound edge; GT, granulation tissue;CT, connective tissue; HF, hair follicles. Scale bar:200 lm.

JACOBSEN et al.

760 � 2010 The Authors Journal Compilation � 2010 APMIS

24). However, in the diabetic mice, we observeda statistically significant delay in wound healingin the b6) ⁄ ) group from day 1 to day 3 post-wounding (p < 0.05), suggesting that avb6integrin might play a protective role in there-epithelialization phase of diabetic woundrepair. No histologic differences at day 5 post-wounding were found between the non-diabeticWT and b6) ⁄ ) animals when comparing epithe-lial wound gaps, granulation tissue formation,and collagen deposition. However, among thediabetic animals, we observed a reduction ingranulation tissue formation, complete absenceof mature collagen deposition and a minortendency toward reduced keratinocyte migra-tion in the b6) ⁄ ) mice. Surprisingly, a recentstudy reported accelerated wound repair in b6integrin-deficient mice in a dexamethasone(DEX)-impaired wound healing model (48).DEX is a glucocorticoid steroid hormone usedas an immunosuppressive drug with unfortunateside effects such as skin atrophy and impairedwound healing caused in part by dysregulationof TGF-b1 (49). Xie et al. observed increasedproliferation of epidermal and hair follicle-asso-ciated keratinocytes in the DEX-treated b6) ⁄ )

mice compared with DEX-treated WT animals,which was linked to a reduction in pro-inflam-matory cytokines and TGF-b1 (48). In the

A

C

D

E

F

B

Fig. 6. (A) Binding affinity of human fibronectin(FN) and the methylglyoxal-treated FN (FN-MG).Immunolabeling with specific antibodies (Ab) showedsimilar binding affinity to the microtiter plate betweenthe two fibronectin preparations. (B) SDS-PAGE and(C) Western blot using an antibody that recognizesspecifically FN-MG. (D) HaCaT keratinocytes wereallowed to spread for 1 h on various concentrations ofnative fibronectin or FN-MG or (E) on 20 lg ⁄mLfibronectin or FN-MG over time. In both cases, cellsshowed reduced spreading on FN-MG (s) comparedwith FN (d) in a concentration- and time-dependentmanner. (F) HaCaT spreading was also quantified inthe presence of function-blocking anti-integrin anti-bodies targeted against the av, b1 and b6 subunits.Data show mean ± SEM of three independent exper-iments, n = 4 (D-E) and n = 8 (F). The effect of inte-grin blocking accounted for 88% of the total variation(p < 0.001), whereas no statistical difference wasobserved between keratinocyte spreading on FN andFN-MG within the same anti-integrin antibodyblocking. All keratinocyte spreading data (D–F) wereevaluated by a two-way ANOVA with Bonferronipost-tests. **p < 0.01, ***p < 0.001.

DIABETIC WOUNDHEALING AND TISSUE BIOLOGY

� 2010 The Authors Journal Compilation � 2010 APMIS 761

STZ-induced diabetes model, the delay inwound healing was mainly associated with ini-tial epithelial migration to the wounds, ratherthan later stages of re-epithelialization that aremore dependent on keratinocyte proliferation.Therefore, the re-epithelialization defectobserved in STZ-treated b6) ⁄ ) mice comparedwith WT diabetic mice is most likely a result ofkeratinocyte migration impairment rather thanproliferation as supported by reduced spreadingon glycated FN.As described previously, a prolonged state of

hyperglycemia can lead to tissue damagethrough aberrant cross-linking of proteins in theECM and their receptors because of the accu-mulation of AGEs (50), which can potentiallylead to impaired keratinocyte function duringre-epithelialization. In this study, we generatedglycated FN through reaction with MG, a reac-tive a-oxaloaldehyde derived from the glucosemetabolism and one of the most potent AGE-formers found in vitro (51) and in vivo (35).Indeed, intraperitoneal administration of MGresults in induced diabetes-like microvascularalterations and impaired wound healing in rats(52), and MG glycation of provisional matrixconstituent type I collagen inhibits keratinocytemigration and spreading, possibly throughmodification of the putative binding site of inte-grin a2b1 (53, 54). We here demonstrated thatkeratinocyte spreading was significantly reducedin a time- and dose-dependent manner on MG-glycated FN.Functional inhibition of integrin receptors

using specific blocking antibodies demonstratedthat the utilization of integrin receptors thatbind native FN was not changed by glycation ofFN, suggesting that the reduced spreading onMG-glycated FN was caused by masking of theintegrin-binding sites. Intriguingly, AGE modi-fication of FN alters the putative tripeptide-binding site (arginine–glycine–aspartic acid,RGD) which is a ligand for a variety of inte-grins, including avb6 (18, 55). It is possible,therefore, that AGE-induced alteration of theRGD sequence in FN may cause aberrant cross-linking to many potential integrin receptors,and thereby hamper keratinocyte functionduring the re-epithelialization phase of diabeticwound healing. Taken together, our data pointto a protective role of avb6 integrin in earlydiabetic wound healing during the phase of

re-epithelialization. Moreover, we have demon-strated reduced spreading frequency of humankeratinocytes on glycated FN, which could playa role in diabetic wound repair.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

The authors thank Chrisitan Sperantia for excellenttechnical assistance. Yanshuang Xie, Leeni Koivisto,and Gethin Owen are acknowledged for their kindhelp on mouse surgery, in vitro experiments andimmunolabeling, respectively. We also thank ZhihuaChen for support with purification and biochemicalmodification of FN. This work was supported by agrant from the Canadian Institutes of HealthResearch (CIHR) to H.L., the Tech Transfer Unit atthe University of Copenhagen to K.A.K. (274-05-0435), and National Institutes of Health (NIH) grantsDE017139 and DE016312 to B.S. J.N.J. acknow-ledges Memorial scholarship of E. and C. Elmquist,The Oticon fund, A.V. Lykfeldt’s fund, the Beckettfund, T. and A. Frimodt’s fund, C. and O. Brorsonstravel grant, and C.M. ⁄M.C. Sørensen and Mrs. S.F.Lerchel fund for generous financial support.

REFERENCES

1. Wild S, Roglic G, Green A, Sicree R, King H.Global prevalence of diabetes: estimates for theyear 2000 and projections for 2030. Diabetes Care2004;27:1047–53.

2. Goren I, Muller E, Pfeilschifter J, Frank S.Severely impaired insulin signaling in chronicwounds of diabetic ob ⁄ob mice: a potential role oftumor necrosis factor-alpha. Am J Pathol2006;168:765–77.

3. Galkowska H, Wojewodzka U, Olszewski WL.Chemokines, cytokines, and growth factors inkeratinocytes and dermal endothelial cells in themargin of chronic diabetic foot ulcers. WoundRepair Regen 2006;14:558–65.

4. Falanga V. Wound healing and its impairment inthe diabetic foot. Lancet 2005;366:1736–43.

5. Galiano RD, Tepper OM, Pelo CR, BhattKA, Callaghan M, Bastidas N, et al. Topicalvascular endothelial growth factor acceleratesdiabetic wound healing through increasedangiogenesis and by mobilizing and recruitingbone marrow-derived cells. Am J Pathol 2004;164:1935–47.

6. Maruyama K, Asai J, Ii M, Thorne T, LosordoDW, D’Amore PA. Decreased macrophage num-ber and activation lead to reduced lymphatic

JACOBSEN et al.

762 � 2010 The Authors Journal Compilation � 2010 APMIS

vessel formation and contribute to impaired dia-betic wound healing. Am J Pathol 2007;170:1178–91.

7. Gibran NS, Jang YC, Isik FF, Greenhalgh DG,Muffley LA, Underwood RA, et al. Diminishedneuropeptide levels contribute to the impairedcutaneous healing response associated with diabe-tes mellitus. J Surg Res 2002;108:122–8.

8. Lobmann R, Ambrosch A, Schultz G, WaldmannK, Schiweck S, Lehnert H. Expression of matrix-metalloproteinases and their inhibitors in thewounds of diabetic and non-diabetic patients.Diabetologia 2002;45:1011–6.

9. Murillo J, Wang Y, Xu X, Klebe RJ, Chen Z,Zardeneta G, et al. Advanced glycation of type Icollagen and fibronectin modifies periodontal cellbehavior. J Periodontol 2008;79:2190–9.

10. Stupack DG. Integrins as a distinct subtype ofdependence receptors. Cell Death Differ 2005;12:1021–30.

11. De Arcangelis A, Georges-Labouesse E. Integrinand ECM functions: roles in vertebrate develop-ment. Trends Genet 2000;16:389–95.

12. Hynes RO. Integrins: bidirectional, allosteric sig-naling machines. Cell 2002;110:673–87.

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

14. Santoro MM, Gaudino G. Cellular and molecularfacets of keratinocyte reepithelization duringwound healing. Exp Cell Res 2005;304:274–86.

15. Clark EA, Brugge JS. Integrins and signal trans-duction pathways: the road taken. Science 1995;268:233–9.

16. Lu M, Munger JS, Steadele M, Busald C, TellierM, Schnapp LM. Integrin alpha8beta1 mediatesadhesion to LAP-TGFbeta1. J Cell Sci 2002;115:4641–8.

17. Munger JS, Huang X, Kawakatsu H, GriffithsMJ, Dalton SL, Wu J, et al. The integrin alpha vbeta 6 binds and activates latent TGF beta 1: amechanism for regulating pulmonary inflamma-tion and fibrosis. Cell 1999;96:319–28.

18. Busk M, Pytela R, Sheppard D. Character-ization of the integrin alpha v beta 6 as a fibro-nectin-binding protein. J Biol Chem 1992;267:5790–6.

19. Koivisto L, Larjava K, Hakkinen L, Uitto VJ,Heino J, Larjava H. Different integrins mediatecell spreading, haptotaxis and lateral migration ofHaCaT keratinocytes on fibronectin. Cell AdhesCommun 1999;7:245–57.

20. Annes JP, Munger JS, Rifkin DB. Making senseof latent TGFbeta activation. J Cell Sci 2003;116:217–24.

21. Annes JP, Chen Y, Munger JS, Rifkin DB. Inte-grin alphaVbeta6-mediated activation of latent

TGF-beta requires the latent TGF-beta bindingprotein-1. J Cell Biol 2004;165:723–34.

22. Aluwihare P, Mu Z, Zhao Z, Yu D, Weinreb PH,Horan GS, et al. Mice that lack activity of alp-havbeta6- and alphavbeta8-integrins reproducethe abnormalities of Tgfb1- and Tgfb3-null mice.J Cell Sci 2009;122:227–32.

23. Hakkinen L, Koivisto L, Gardner H, Saarialho-Kere U, Carroll JM, Lakso M, et al. Increasedexpression of beta6-integrin in skin leads to spon-taneous development of chronic wounds. Am JPathol 2004;164:229–42.

24. Huang XZ, Wu JF, Cass D, Erle DJ, Corry D,Young SG, et al. Inactivation of the integrin beta6 subunit gene reveals a role of epithelial integrinsin regulating inflammation in the lung and skin.J Cell Biol 1996;133:921–8.

25. AlDahlawi S, Eslami A, Hakkinen L, LarjavaHS. The alphavbeta6 integrin plays a role in com-promised epidermal wound healing. WoundRepair Regen 2006;14:289–97.

26. MiQ, Riviere B, ClermontG, SteedDL, VodovotzY.Agent-basedmodel of inflammation andwoundhealing: insights into diabetic foot ulcer pathologyand the role of transforming growth factor-beta1.WoundRepair Regen 2007;15:671–82.

27. Michaels J, Churgin SS, Blechman KM, GreivesMR, Aarabi S, Galiano RD, et al. db ⁄db miceexhibit severe wound-healing impairments com-pared with other murine diabetic strains in a sili-cone-splinted excisional wound model. WoundRepair Regen 2007;15:665–70.

28. Qiu Z, Kwon AH, Kamiyama Y. Effects ofplasma fibronectin on the healing of full-thicknessskin wounds in streptozotocin-induced diabeticrats. J Surg Res 2007;138:64–70.

29. Greenhalgh DG, Sprugel KH, Murray MJ, RossR. PDGF and FGF stimulate wound healing inthe genetically diabetic mouse. Am J Pathol 1990;136:1235–46.

30. Schmidt R., Wirtala J. Modification of Movatpentachrome stain with improved reliability ofelastin staining. J Histotechnol 1996;19:325–7.

31. Engvall E, Ruoslahti E. Binding of soluble formof fibroblast surface protein, fibronectin, to colla-gen. Int J Cancer 1977;20:1–5.

32. Pal S, Chen Z, Xu X, Mikhailova M, SteffensenB. Co-purified gelatinases alter the stability andbiological activities of human plasma fibronectin.J Periodont Res 2010;45:292–5.

33. Overall CM, Limeback H. Identification andcharacterization of enamel proteinases isolatedfrom developing enamel. Amelogeninolytic serineproteinases are associated with enamel matura-tion in pig. Biochem J 1988;256:965–72.

34. Steffensen B, Wallon UM, Overall CM. Extracel-lular matrix binding properties of recombinantfibronectin type II-like modules of human 72-kDa

DIABETIC WOUNDHEALING AND TISSUE BIOLOGY

� 2010 The Authors Journal Compilation � 2010 APMIS 763

gelatinase ⁄ type IV collagenase. High affinitybinding to native type I collagen but not nativetype IV collagen. J Biol Chem 1995;270:11555–66.

35. Shamsi FA, Partal A, Sady C, Glomb MA, Naga-raj RH. Immunological evidence for methylgly-oxal-derived modifications in vivo. Determinationof antigenic epitopes. J Biol Chem 1998;273:6928–36.

36. Stanley CM, Wang Y, Pal S, Klebe RJ, HarklessLB, Xu X, et al. Fibronectin fragmentation is afeature of periodontal disease sites and diabeticfoot and leg wounds and modifies cell behavior.J Periodontol 2008;79:861–75.

37. Akiyama SK, Yamada SS, Chen WT, YamadaKM. Analysis of fibronectin receptor functionwith monoclonal antibodies: roles in cell adhe-sion, migration, matrix assembly, and cytoskeletalorganization. J Cell Biol 1989;109:863–75.

38. Huang X, Wu J, Spong S, Sheppard D. The inte-grin alphavbeta6 is critical for keratinocyte migra-tion on both its known ligand, fibronectin, and onvitronectin. J Cell Sci 1998;111(Pt 15):2189–95.

39. Rossini AA, Like AA, Chick WL, Appel MC,Cahill GF Jr. Studies of streptozotocin-inducedinsulitis and diabetes. Proc Natl Acad Sci USA1977;74:2485–9.

40. Xie X, Li S, Liu S, Lu Y, Shen P, Ji J. Proteomicanalysis of mouse islets after multiple low-dosestreptozotocin injection. Biochim Biophys Acta2008;1784:276–84.

41. Like AA, Rossini AA. Streptozotocin-inducedpancreatic insulitis: new model of diabetes mell-itus. Science 1976;193:415–7.

42. Riley WJ, McConnell TJ, Maclaren NK,McLaughlin JV, Taylor G. The diabetogeniceffects of streptozotocin in mice are prolongedand inversely related to age. Diabetes 1981;30:718–23.

43. Rayat GR, Rajotte RV, Lyon JG, Dufour JM,Hacquoil BV, Korbutt GS. Immunization withstreptozotocin-treated NOD mouse islets inhibitsthe onset of autoimmune diabetes in NOD mice.J Autoimmun 2003;21:11–5.

44. Coleman DL. Obese and diabetes: two mutantgenes causing diabetes-obesity syndromes in mice.Diabetologia 1978;14:141–8.

45. Coleman DL. Diabetes-obesity syndromes inmice. Diabetes 1982;31:1–6.

46. Galiano RD, Michaels J, Dobryansky M, LevineJP, Gurtner GC. Quantitative and reproduciblemurine model of excisional wound healing.Wound Repair Regen 2004;12:485–92.

47. Brownlee M. Biochemistry and molecular cellbiology of diabetic complications. Nature 2001;414:813–20.

48. Xie Y, Gao K, Hakkinen L, Larjava HS. Micelacking beta6 integrin in skin show acceleratedwound repair in dexamethasone impaired woundhealing model. Wound Repair Regen 2009;17:326–39.

49. Beer HD, Fassler R, Werner S. Glucocorticoid-regulated gene expression during cutaneouswound repair. Vitam Horm 2000;59:217–39.

50. Peppa M, Stavroulakis P, Raptis SA. Advancedglycoxidation products and impaired diabeticwound healing. Wound Repair Regen 2009;17:461–72.

51. Ahmed N, Argirov OK, Minhas HS, CordeiroCA, Thornalley PJ. Assay of advanced glycationendproducts (AGEs): surveying AGEs by chro-matographic assay with derivatization by 6-am-inoquinolyl-N-hydroxysuccinimidyl-carbamateand application to Nepsilon-carboxymethyl-lysine- and Nepsilon-(1-carboxyethyl)lysine-mod-ified albumin. Biochem J 2002;364:1–14.

52. Berlanga J, Cibrian D, Guillen I, Freyre F, AlbaJS, Lopez-Saura P, et al. Methylglyoxal adminis-tration induces diabetes-like microvascularchanges and perturbs the healing process ofcutaneous wounds. Clin Sci (Lond) 2005;109:83–95.

53. Morita K, Urabe K, Moroi Y, Koga T, Nagai R,Horiuchi S, et al. Migration of keratinocytes isimpaired on glycated collagen I. Wound RepairRegen 2005;13:93–101.

54. Chong SA, Lee W, Arora PD, Laschinger C,Young EW, Simmons CA, et al. Methylglyoxalinhibits the binding step of collagen phagocytosis.J Biol Chem 2007;282:8510–20.

55. Sakata N, Sasatomi Y, Meng J, Ando S, UesugiN, Takebayashi S, et al. Possible involvement ofaltered RGD sequence in reduced adhesive andspreading activities of advanced glycation endproduct-modified fibronectin to vascularsmooth muscle cells. Connect Tissue Res 2000;41:213–28.

JACOBSEN et al.

764 � 2010 The Authors Journal Compilation � 2010 APMIS