SPARC-thrombospondin-2-double-null Mice Exhibit Enhanced Cutaneous Wound Healing and Increased Fibrovascular Invasion of Subcutaneous Polyvinyl Alcohol Sponges

http://jhc.sagepub.com/Journal of Histochemistry & Cytochemistry

http://jhc.sagepub.com/content/53/5/571The online version of this article can be found at:

 DOI: 10.1369/jhc.4A6425.2005

2005 53: 571J Histochem CytochemGooden, Robert B. Vernon, Thomas N. Wight, Paul Bornstein and E. Helene Sage

Pauli A. Puolakkainen, Amy D. Bradshaw, Rolf A. Brekken, May J. Reed, Themis Kyriakides, Sarah E. Funk, Michel D.Increased Fibrovascular Invasion of Subcutaneous Polyvinyl Alcohol Sponges

SPARC-thrombospondin-2-double-null Mice Exhibit Enhanced Cutaneous Wound Healing and  

Published by:

http://www.sagepublications.com

On behalf of: 

  Official Journal of The Histochemical Society

can be found at:Journal of Histochemistry & CytochemistryAdditional services and information for    

  http://jhc.sagepub.com/cgi/alertsEmail Alerts:

http://jhc.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

What is This? 

- May 1, 2005Version of Record >>

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

© The Histochemical Society, Inc.

0022-1554/05/$3.30

571

ARTICLE

Volume 53(5): 571–581, 2005Journal of Histochemistry & Cytochemistry

http://www.jhc.org

SPARC-thrombospondin-2-double-null Mice Exhibit Enhanced Cutaneous Wound Healing and Increased Fibrovascular Invasion of Subcutaneous Polyvinyl Alcohol Sponges

Pauli A. Puolakkainen, Amy D. Bradshaw, Rolf A. Brekken, May J. Reed, Themis Kyriakides, Sarah E. Funk, Michel D. Gooden, Robert B. Vernon, Thomas N. Wight, Paul Bornstein,and E. Helene Sage

Hope Heart Program, Benaroya Research Institute at Virginia Mason, Seattle, Washington (PAP,ADB,RAB,SEF,MDG,RBV, TNW,EHS); Departments of Medicine (PAP,MJR,PB) and Biochemistry (TK,PB), University of Washington, Seattle, Washington; and Department of Surgery, Helsinki University Central Hospital, Helsinki, Finland (PAP)

SUMMARY

Secreted protein acidic and rich in cysteine (SPARC) and thrombospondin-2(TSP-2) are structurally unrelated matricellular proteins that have important roles in cell–extracellular matrix (ECM) interactions and tissue repair. SPARC-null mice exhibit acceler-ated wound closure, and TSP-2-null mice show an overall enhancement in wound healing.To assess potential compensation of one protein for the other, we examined cutaneouswound healing and fibrovascular invasion of subcutaneous sponges in SPARC-TSP-2 (ST)double-null and wild-type (WT) mice. Epidermal closure of cutaneous wounds was found tooccur significantly faster in ST-double-null mice, compared with WT animals: histologicalanalysis of dermal wound repair revealed significantly more mature phases of healing at 1,4, 7, 10, and 14 days after wounding, and electron microscopy showed disrupted ECM at 14days in these mice. ST-double-null dermal fibroblasts displayed accelerated migration, rela-tive to WT fibroblasts, in a wounding assay in vitro, as well as enhanced contraction of na-tive collagen gels. Zymography indicated that fibroblasts from ST-double-null mice alsoproduced higher levels of matrix metalloproteinase (MMP)-2. These data are consistentwith the increased fibrovascular invasion of subcutaneous sponge implants seen in thedouble-null mice. The generally accelerated wound healing of ST-double-null mice reflectsthat described for the single-null animals. Importantly, the absence of both proteins resultsin elevated MMP-2 levels. SPARC and TSP-2 therefore perform similar functions in the regu-lation of cutaneous wound healing, but fine-tuning with respect to ECM production andremodeling could account for the enhanced response seen in ST-double-null mice.

(J Histochem Cytochem 53:571–581, 2005)

S

ecreted protein acidic

and rich in cysteine (SPARC)

,

thrombospondin (TSP)-1 and -2, tenascin C and X,osteopontin, and SC1/hevin are matricellular proteinsthat do not contribute structurally to extracellularmatrix (ECM). Rather, they bind to a variety of struc-tural proteins (e.g., collagens) and modulate the inter-action of cells with the ECM (Bornstein and Sage 2002).

SPARC (also known as BM-40 and osteonectin) isa 32-kDa calcium-binding glycoprotein secreted byvarious cells, e.g., fibroblasts, endothelial cells, andplatelets (Stenner et al. 1986; Sage et al. 1989a; Ka-sugai et al. 1991; Vuorio et al. 1991). SPARC modu-lates the interaction of cells with the ECM, is coun-teradhesive for cells from diverse sources, inhibits cellspreading, and regulates the production of severalECM proteins (Bradshaw and Sage 2001). Two prin-cipal functions of SPARC are modification of cellshape and inhibition of cell-cycle progression (Brad-shaw and Sage 2001). SPARC is expressed during de-velopment and in remodeling tissues in adults (Reed et

KEY WORDS

SPARC

thrombospondin

wound healing

matricellular

extracellular matrix

epidermis

dermis

angiogenesis

collagen

Correspondence to: E. Helene Sage, PhD, Hope Heart Program,Benaroya Research Institute at Virginia Mason, 1201 9th Ave.,Seattle, WA 98101. E-mail: [email protected] or [email protected]

Received for publication May 26, 2004; accepted September 29,2004 [DOI: 10.1369/jhc.4A6425.2005].

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

572

Puolakkainen, Bradshaw, Brekken, Reed, Kyriakides, Funk, Gooden, Vernon, Wight, Bornstein, Sage

al. 1993; Puolakkainen et al. 1999; Bornstein andSage 2002). Targeted disruption of the SPARC gene inmice results in a complex phenotype characterized byearly cataractogenesis (Gilmour et al. 1998; Norose etal. 1998; Yan et al. 2002), increased amounts ofsubcutaneous adipose tissue (Bradshaw et al. 2003a),decreased amounts of collagen in the skin (Bradshawet al. 2003b), and progressively severe osteopenia(Delany et al. 2000). The curled tails of these mice arealso suggestive of altered collagen fibrillogenesis. Fur-thermore, enhanced fibrovascular invasion of subcuta-neous polyvinyl alcohol (PVA) sponge implants (Brad-shaw et al. 2001), enhanced growth of malignanttumors (Brekken et al. 2003), and reduced foreignbody response (Puolakkainen et al. 2003) have beenreported in SPARC-null mice. We have previouslydescribed the spatial and temporal distribution ofSPARC during cutaneous wound healing (Reed et al.1993) and during the healing of intestinal anasto-moses (Puolakkainen et al. 1999) and have recentlydescribed accelerated cutaneous wound closure inSPARC-null mice (Bradshaw et al. 2002). Dermal fi-broblasts from these animals displayed higher rates ofmigration, relative to cells from wild-type (WT) mice,in a wounding model in vitro (Bradshaw et al. 2002).

TSP-2 is a matricellular glycoprotein that influencesthe formation of collagen fibrils and the interactionsof cells with ECM. It is produced in many connectivetissues during development and in response to injuryin adult animals (Kyriakides et al. 1999b; Bornstein etal. 2000). The phenotype of TSP-2-null mice includesenhanced vascular density and angiogenesis, a bleed-ing disorder, increased bone growth, and abnormalcollagen fibrils associated with laxity of tendons andligaments as well as an increased fragility of skin(Kyriakides et al. 1998). Dermal fibroblasts fromTSP-2-null mice exhibit defects in adhesion that aredue to augmented levels of matrix metalloproteinase(MMP)-2 (Yang et al. 2000). TSP-2 is expressed duringcutaneous wound healing (Kyriakides et al. 1999b),and TSP-2-null mice displayed accelerated excisionalwound healing with irregularly organized and highlyvascularized granulation tissue (Kyriakides et al. 1999b).No difference was found in the rate of re-epithelializa-tion, but there was less scarring in TSP-2 mice, relativeto WT animals. The fibrovascular invasion of subcu-taneous PVA sponges was also increased in TSP-2-nullmice (Kyriakides et al. 1999a).

Clearly, SPARC and TSP-2 are distinct matricellu-lar proteins, with apparently similar functions (e.g.,ECM production and/or assembly) as well as dispar-ate ones (e.g., inhibition of angiogenesis by TSP-2). Tofurther evaluate the function of SPARC and TSP-2 invivo, to discriminate between their contributions, andto assess potential compensation of one protein for theother, we measured cutaneous wound healing, inva-

sion of PVA sponges, and changes in ECM in SPARC-TSP-2 (ST) double-null mice, compared with WT ani-mals. We report enhanced epidermal closure and der-mal wound healing, as well as increased fibrovascularinvasion of PVA sponges, which are associated with al-tered ECM in ST-double-null mice.

Materials and Methods

Analysis of Genomic DNA by Polymerase Chain Reaction (PCR)

Tail-derived DNA was digested with proteinase K (100 ng/

�

l)at 55C overnight, followed by precipitation in isopropanol.PCR was performed as described by Graves and Yablonka-Reuveni (2000). The following primers were used. SPARC(WT allele): 5

�

-GATGAGGGTGGTCTGGCCCAGCCCTA-GATGCCCCTCAC-3

�

(forward) and 5

�

-CACCCACACAG-CTGGGGGTGATCCAGATAAGCCAAG-3

�

(reverse); SPARC(null allele): forward primer as above; 5

�

-GTTGTGCCC-AGTCATAGCCGAATAGCCTCTCCACCCAAG-3

�

(reverse,located in Neo insert) (Figure 1). TSP-2 (WT allele): 5

�

-GGT-GACCACGTCAAGGACAC-3

�

(forward) and 5

�

-TGG-CCACGTACATCCTGCT 3

�

(reverse); TSP-2 (null allele):5

�

-GATCAGCAGCCTCTGTTCACATAC-3

�

(forward, lo-cated in Neo insert) and 5

�

-GGAGAAGAATTAGGGAGGC-TTAGGG-3

�

(reverse, located in TSP-2 intron 3) (Figure 1).

Animal Model

C57Bl/6

�

129SvJ SPARC-TSP-2 (ST) double-null mice andcorresponding WT animals (produced from heterozygousmatings) were generated from the corresponding single

-

nullanimals (Kyriakides et al. 1998; Bradshaw et al. 2002). Thegenotypes of the mouse colony were monitored by PCR us-ing genome-specific primers on purified tail DNA. All micewere handled and studies were carried out according to theguidelines of the American Association for Accreditation ofLaboratory Care and The Hope Heart Institute Animal Careand Use Committee. All mice were housed in a modifiedpathogen-free facility before and throughout the study.Twenty-eight ST-double-null and 25 WT mice (3–7 monthsold) were used for the experiments. For surgery, the micewere anesthetized by isoflurane inhalation (Abbott Labora-tories; North Chicago, IL), and their backs were shaved andcleaned. Each animal received two excisional punch biopsies(Acu-Punch; Acuderm, Inc., Ft. Lauderdale, FL) of 5-mm di-ameter and two incisional (1 cm) full-thickness cutaneouswounds. Incisions were made with a scalpel and were closedwith staples. All instruments were rinsed in endotoxin-freewater and were sterilized prior to use. Animals were housedindividually and were monitored for 4–6 hr after woundingto ensure that premature closure of the biopsy wound edgesdid not occur. There were no significant differences in thewound sizes at time 0. The animals were sacrificed at 1, 4, 7,10, and 14 days after wounding.

Assessment of Wound Healing

At the time of sacrifice, the wound images were captureddigitally, and the wound area and the extent of epithelial

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

Enhanced Wound Repair in SPARC-TSP-2-null Mice

573

closure were calculated with computerized planimetry usingAdobe Photoshop and public domain NIH-Image software.Dermal healing was assessed in hematoxylin- and eosin-stained sections of wounds with a maturity scoring systemintroduced by Greenhalgh et al. (1990) as modified bySpenny et al. (2002). The amount and type of cell accumula-tion, the type of granulation tissue, and scarring are assessedin this model, with a score ranging from 1 to 15.

Immunohistochemical Staining

Immunolocalization of SPARC was performed as previouslydescribed (Sage et al. 1989b). Wound samples were im-mersed in Methacarn fixative (60% methanol, 30% chloro-form, 10% glacial acetic acid) for 24 hr, embedded in paraf-fin, and cut into 5-

�

m sections. Sections were heated ona hotplate for 1 hr (60C), deparaffinized, and rehydratedthrough an ethanol gradient ending with water. Sectionswere incubated in 1% hydrogen peroxide for 30 min toblock endogenous peroxidases, followed by blocking in Au-

tozyme (Biomeda; Foster City, CA) at 37C for 10 min and in25% Sea-Block (East Coast Biologics; North Berwick,ME) in 0.2% PBS-Tween for 20 min. Incubation with a goatpolyclonal anti-SPARC primary antibody (10

�

g/ml) (Brekkenet al. 2003) was carried out for 1 hr at room temperature.After a 1-hr incubation with a donkey anti-goat secondaryantibody (Jackson Immunoresearch Laboratories; West Grove,PA) (2.5

�

g/ml) at room temperature, immune complexeswere detected with 3,3

�

-diaminobenzidine (Sigma ChemicalCo.; St Louis, MO). The sections were counterstained withhematoxylin, dehydrated with a graded series of ethanol so-lutions, cleared with xylene, and mounted with Permount(Fisher Scientific; Fair Lawn, NJ). Stained sections were cap-tured digitally with a cooled CCD camera (SPOT RT Diag-nostic Instruments Inc.; Sterling Heights, MI). Control sec-tions were treated with secondary antibody only. Tissuesections of testis from ST-double-null and WT mice servedas negative and positive controls, respectively.

For assessment of cell proliferation, six mice of each geno-type were injected intraperitoneally (2

�

g/g) on the day ofsampling with 300

�

g/ml of 5-bromo-2

�

-deoxyuridine (BrdU)(Sigma) and were sacrificed after 8 hr. Samples were pre-pared for immunostaining as described (Reed et al. 1996).The number of BrdU-positive cells/wound in the wound areawas calculated from these slides.

Reverse Transcriptase PCR (RT-PCR)

RNA was isolated from wound tissue with Tri Reagent (Mo-lecular Research Center, Inc.; Cincinnati, OH) according tothe manufacturer’s instructions. PCR was performed asdescribed (Graves and Yablonka-Reuveni 2000) with theprimer sequences (sense/anti-sense) for SPARC, rpS6, trans-forming growth factor (TGF)-

�

, vascular endothelial growthfactor (VEGF), MMP-2, collagen I, and collagen VI, as de-scribed by Brekken et al. (2003).

Electron Microscopy

Samples were immersed in Karnovsky’s fixative and wereprocessed for transmission electron microscopy according toestablished methods (Wight et al. 1997).

Isolation and Culture of Dermal Fibroblasts

Dermal cells were isolated from the skin as described (Brad-shaw et al. 2002). In brief, pieces of shaved skin were incu-bated in 0.25% trypsin (Sigma) in Solution A (10 mM glu-cose, 3 mM KCl, 130 mM NaCl, 1 mM Na

2

HPO

4

/7H

2

O,30 mM Hepes, pH 7.4) overnight at 4C to separate the epi-dermis from the dermis. The dermal pieces were incubatedin 0.25%

Clostridium histolyticum

collagenase (Worthing-ton Enzymes; Freehold, NJ) at 37C for 4–6 hr. The digesteddermis was triturated and plated in Dulbecco’s modified Ea-gle’s medium (DMEM) supplemented with 10% fetal calfserum (Gibco-BRL; Gaithersburg, MD), 500 U/ml penicillinG, 500 U/ml streptomycin sulfate, and 2.5

�

g/ml fungizone(growth medium). The majority of cells in these primary cul-tures were dermal fibroblasts (Bradshaw et al. 2002). All ex-periments were performed with cells at early passage (P1–P4), before cellular senescence or transformation.

Figure 1 The SPARC and TSP-2 genes are disrupted in ST-double-null mice. PCR products of genomic DNA isolated from WT (�/�)and double-null (�/�) mouse-tail biopsies are shown. The WT al-lele of the TSP-2 gene was detected with primers (TS2G-A andTS2G-B) to sequences within the gene that yielded a 539-bp se-quence; the disrupted allele was detected with primers to se-quences within the gene (T2IN3) and the neomycin insert (447neo) to yield a 900-bp sequence. The WT and disrupted alleles ofthe SPARC gene were similarly detected with primer pairs MG-SPARC.FOR and MGSP.REV, and MGSPARC.FOR and NEO.REV thatyield 296- and 450-bp sequences, respectively. MW markers appearon the right of each panel.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

574

Puolakkainen, Bradshaw, Brekken, Reed, Kyriakides, Funk, Gooden, Vernon, Wight, Bornstein, Sage

In Vitro Wounding Model for Cell Migration

For migration studies, equal numbers of ST-double-null andWT dermal fibroblasts were plated in six-well plates ingrowth medium and were grown to confluence. A rubberspatula was used to remove a defined area of cells, i.e., to cre-ate a wound in the monolayer. An area of each well was des-ignated by marks on the dish such that the same field could befollowed over time. The degree of cell migration was moni-tored by photography of the same field at 24-hr intervals (24,48, and 72 hr after wounding). Scanned images were im-ported into NIH-Image and were quantified as the percent ofarea occupied by cells vs the total area of the original wound.

Collagen Gel Contraction Assays

Assays of cell-mediated contraction of native, fibrillar type-Icollagen gels were performed according to Vernon andGooden (2002). The disk-shaped, 12.7-mm diameter gels,which were polymerized from 0.5 mg/ml rat tail collagen(BD Biosciences; Bedford, MA), contained ST-double-null orWT dermal fibroblasts at three different concentrations. (6,12, and 24

�

10

3

cells/400

�

l gel). The gels were culturedfor 20 hr at 37C in DMEM/10% fetal calf serum to permitcontraction and subsequently were fixed with neutral-buff-ered formalin (Sigma), immersed in deionized water, andviewed under darkfield illumination with a stereomicroscopeequipped with a CCD (SPOT) camera. Areas of gels weremeasured from digital images with NIH-Image.

Zymography

Equivalent numbers of ST-double-null and WT dermal fibro-blasts were cultured in DMEM/1% fetal calf serum. The culturemedia were conditioned for 3 days, after which zymographywas performed as described (Koike et al. 2002). Control me-dium was defined as that which had no contact with the cells.

Fibrovascular Invasion of PVA sponges

Prior to implantation, circular PVA sponges of uniform size(10-mm diameter, clinical PVA sponges grade 3; M-PACTWorldwide Management Co., Eudora, KS) were treated as pre-viously described (Bradshaw et al. 2002). The sponges were im-planted into five ST-double-null and four WT mice (threesponges per animal), sampled at 14 days after implantation,processed for histology, and stained with hematoxylin andeosin. For quantification of invasion, microscopic images werecaptured with a CCD (SPOT) camera, imported into NIH-Image, and the area of fibrovascular invasion calculated as a per-centage of the total area of the sponge (Bradshaw et al. 2002).

Results

ST-double-null Mice

ST-double-null mice appeared grossly normal. How-ever, they did have kinked tails and early cataract for-mation similar to SPARC-null mice. Previously, it hasbeen shown that SPARC-null mice do not produceSPARC protein (Bradshaw et al. 2003b) and that TSP-2-null mice do not produce TSP-2 (Kyriakides et al.1998). Confirmation of the disruption of both genes inthe ST-double-null mice is shown in Figure 1.

Wound Healing Is Acceleratedin ST-double-null Mice

Cell migration, proliferation, and ECM remodeling arecritical components of cutaneous wound healing. Epi-dermal closure, as reflected by a decreased woundarea, occurred faster in ST-double-null mice, relativeto WT animals. This difference between ST-double-nulland WT mice was statistically significant (

p

�

0.01) at1-, 4-, 7-, and 10-day time points after wounding (Fig-ures 2A and 2B). Some of the wounds in the WT miceremained open 14 days after wounding, whereas allthe wounds in the ST-double-null mice were epithe-lialized in less than 10 days. In comparison, by day 11,five of six SPARC-null wounds showed no visible scabor opening, whereas only three of six WT wounds ex-hibited complete closure (Bradshaw et al. 2002). Incontrast, no differences in wound re-epithelializationwere found between TSP-2-null and WT mice (Kyri-akides et al. 1999b).

Wound healing was accelerated in both incisional(Figure 3A) and excisional (Figure 3B) wounds in ST-double-null mice at all time points, in comparison towounds in WT mice. Because cell accumulation, de-velopment and maturation of granulation tissue, andscarring occurred earlier, the maturity score was higherin the ST-double-null animals. The difference was sig-nificant (

p

�

0.01) at 4, 7, and 10 days (Figure 3C).Wound maturity scoring produced comparable resultsfrom incisional (data not shown) and excisional wounds(Figure 3C).

SPARC Is Present in Healing Wounds

Because critical components of cutaneous wound heal-ing are influenced by the expression of SPARC, weconfirmed its appearance in healing wounds of WTanimals. Immunostaining for SPARC revealed increasedexpression in the wound area at 4 and 7 days afterwounding, in comparison to uninjured dermis. SPARCwas located intracellularly in fibroblasts and endothe-lial cells of the wound area (data not shown) (Reed etal. 1993). No staining for SPARC was seen in thewounds or skin from ST-double-null mice. RT-PCRrevealed SPARC mRNA in the wounds of WT animalsat all time points, with highest levels at days 4 and 7after wounding (data not shown). The expression pat-tern of TSP-2 in healing wounds has previously beenreported (Kyriakides et al. 1999b). At day 3, TSP-2was detected in only a small number of fibroblast-likecells in the wound area, and no ECM-associated ex-pression was observed. In 7-, 10-, and 14-day wounds,TSP-2 was prominent in both the cells and ECM withinthe wound area (Kyriakides et al. 1999b). Thus, bothSPARC and TSP-2 are expressed in a temporal fashionthat is relevant to healing cutaneous wounds of WTanimals.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

Enhanced Wound Repair in SPARC-TSP-2-null Mice

575

Cell Proliferation in ST-double-null vs WT Wounds

Both SPARC and TSP-2 affect the cell cycle (Brekkenand Sage 2000; Armstrong et al. 2002). We thereforeasked whether an increase in cell proliferation in vivo

contributed to the accelerated wound repair in ST-double-null mice. We found no significant differencein the amount of cell proliferation assessed by BrdUincorporation between wounds from ST-double-nulland WT mice. The mean number (

�

SD) of BrdU-posi-tive cells in wounds from ST-double-null mice at 4 dayswas 105

�

40 and at 7 days was 131

�

7. Corre-sponding values for WT mice were 97

�

9 and 104

�

37, respectively. These results indicate that alteredrates of cell proliferation are not responsible for theaccelerated wound repair at 4 days.

ECM Is Altered in ST-double-null Mice

Because a lack of SPARC or TSP-2 results in changesin the organization and composition of ECM, weasked whether such changes could contribute to theaccelerated wound healing in ST-double-null mice.Electron microscopy revealed that tissue from the 14-day wounds of WT animals contained collagen fibrilsof variable size. Moreover, the fibrils were arranged inparallel and formed the anticipated, well-organized ar-chitecture of the ECM (Figures 4A and 4B). In con-trast, collagen fibrils in wound tissue from ST-double-null mice were small and rather uniform in size, andthe organization of the ECM was loose and disrupted(Figures 4C and 4D). Previously, small collagen fibrilsof uniform size were described in the skin and in for-eign body capsules of SPARC-null mice (Bradshawet al. 2003b, Puolakkainen et al. 2003), and irregularcollagen fibers arranged in a basket-weave fashion werereported in wounds from TSP-2-null mice (Kyriakideset al. 1999b). Thus, changes in ECM composition areseemingly correlated with alterations in wound repairin the double-null mice.

Dermal Cell Migration Is Acceleratedin ST-double-null Mice

We also asked whether there were differences in ratesof migration between primary WT and ST-double-nulldermal cells. Confluent dermal cell monolayers werewounded, after which cell migration was analyzed as afunction of time. The denuded area of the monolayerwas covered faster by cells from ST-double-null mice,a result indicating that the fibroblasts from the skin ofST-double-null mice migrated significantly (

p

�

0.01)more rapidly than those from WT animals (Figure 5).Because wound healing is complex and relies in parton cell migration, further studies were performed invivo on PVA sponge implants to determine the effectof combined SPARC and TSP-2 deficiency. Fibrovas-cular invasion of the sponges, determined as the per-centage of the total area of the sponge devoted tocells, ECM, and vasculature, was increased signifi-cantly (

p

�

0.01) in ST-double-null mice, relative to

Figure 2 Epidermal closure is enhanced in ST-double-null mice.(A) Stereomicroscopic image of excisional wounds at 1, 4, and 10days after wounding. Part of the wounds in WT mice remainednon-epithelialized at 10 days. Bar 2 mm. (B) The areas (mm2 �SD) of the wounds were measured by NIH-Image software. Thewound area was significantly smaller in ST-double-null mice (graybars), relative to WT animals (white bars), an indication of en-hanced epidermal closure. *p�0.01.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

576

Puolakkainen, Bradshaw, Brekken, Reed, Kyriakides, Funk, Gooden, Vernon, Wight, Bornstein, Sage

WT animals (Figure 6). No significant differences be-tween genotypes were found in the morphology of thetissue invading the sponge implants. The finding ofenhanced fibrovascular invasion was consistent with

that seen in SPARC-null and TSP-2-null mice (Brad-shaw et al. 2001; Kyriakides et al. 2001).

One possibility for the enhanced migration of ST-double-null over WT cells could be an increased secre-

Figure 3 Dermal wound healing is accelerated in ST-double-null mice. Hematoxylin- and eosin-stained sections of representative incisional(A) and excisional (B) wounds at 1, 4, 7, 10, and 14 days after wounding indicate more mature and advanced wound healing in ST-double-null mice, as reflected by cell accumulation (short arrows), development of granulation tissue (arrowheads), and scarring (long arrows). As-terisks indicate edges of wounds. Magnification: �2.5–10. (C) Dermal wound healing was analyzed according to a modified Greenhalghclassification of wound maturity (score from 1 to 15) (mean � SD). A significantly enhanced dermal wound maturity score was found in ST-double-null mice (gray bars), relative to WT animals (white bars), at all time points but 14 days. *p�0.01.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

Enhanced Wound Repair in SPARC-TSP-2-null Mice

577

tion of MMPs. We examined levels of MMP2 becauseaugmented levels of this proteinase have been reportedin cells from TSP-2-null mice (Yang et al. 2000). Equiv-alent numbers of ST-double-null and WT dermalfibroblasts were cultured in DMEM/1% fetal calf se-rum. The supernate was conditioned for 3 days, afterwhich zymography was performed. Cells from ST-double-null mice produced more MMP-2 than cellsfrom their WT counterparts (Figure 7). No differencewas found in the expression of transforming growthfactor beta (TGFbeta), vascular endothelial cell growthfactor (VEGF), collagen I, or collagen VI mRNA be-tween wounds from ST-double-null and WT mice(data not shown). Levels of TGFbeta mRNA were max-imal at 4 and 7 days, whereas those of VEGF wereconstant at different time points.

Fibroblasts from ST-double-null Animals Exhibit Enhanced Contraction of Native Collagen Gels

The capacity of dermal fibroblasts to remodel ECMwas tested in a collagen gel contraction assay. Over a

fourfold range of cell concentration, fibroblasts fromST-double-null mice contracted collagen gels to agreater extent than did fibroblasts from WT mice (Fig-ure 8). This result was distinct from observations ofcollagen gel contraction by single-null SPARC or TSP-2dermal fibroblasts. Relative to WT, contraction of col-lagen gels by SPARC-null fibroblasts was similar (Brad-shaw et al. 2002), whereas gel contraction by TSP-2-null fibroblasts was diminished (Kyriakides et al. 1999b).

A summary of the data relating to wound healing inST-double-null, SPARC-null, and TSP-2-null mice isshown in Table 1.

Discussion

A number of matricellular proteins including SPARC,TSP-1 and -2, osteopontin, and tenascin C (Mackie etal. 1988; Reed et al. 1993; Liaw et al. 1998) show in-creased expression in response to injury, such as cuta-neous wound healing. Previously, the role of SPARCand TSP-2 in wound healing has been studied in mice

Figure 4 ECM is altered in ST-double-null mice. (A) Transmission electron microscopy of wound tissue from day 14 wounds of WT animalsshowed collagen fibrils of variable size (arrows). (B) The collagen fibrils (e.g., arrow) in WT wounds were parallel and formed a denselypacked, well-organized dermal ECM. (C) In contrast, collagen fibrils (arrows) in the wound tissue from ST-double-null mice were small anduniform in size. (D) The organization of the collagen fibrils (e.g., arrow) in ST-double-null ECM was loose and disrupted. Original magnifica-tions: �50,000 (A,C) and �30,000 (B,D).

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

578

Puolakkainen, Bradshaw, Brekken, Reed, Kyriakides, Funk, Gooden, Vernon, Wight, Bornstein, Sage

with a targeted disruption of each gene: SPARC-nullmice exhibited accelerated epidermal closure, andTSP-2-null mice showed an enhanced dermal woundhealing (Table 1).

Here we report significantly enhanced epithelialclosure in mice lacking gene function for both SPARCand TSP-2. That no difference in the re-epithelializa-

tion rate was detected in TSP-2-null mice (Kyriakideset al. 1999b) indicates that SPARC functions in vivomore specifically in epidermal migration and closurethan TSP-2. This result is consistent with our findingthat dermal fibroblasts cultured from ST-double-nullmice showed enhanced migration, relative to fibro-blasts from WT mice in vitro. Accelerated migrationrelative to WT controls was also seen in dermal fibro-blasts from SPARC-null mice (Bradshaw et al. 2002)(Table 1).

Greenhalgh et al. (1990) introduced a scoring sys-tem to assess wound repair and to quantify the matu-rity of the wound as a function of time. This scoreallows comparison of wounds, e.g., after treatment orbetween genotypes. We have used a simplified scoringmethod that excludes the epithelial component (Spennyet al. 2002) and found significantly more maturephases of both excisional and incisional wound heal-ing at 1, 4, 7, 10, and 14 days after wounding in ST-double-null mice. This result is in accord with a previ-ous study that showed an irregularly organized buthighly vascularized granulation tissue in TSP-2-nullmice (Kyriakides et al. 1999b). In addition, TSP-2-nullwounds manifested a higher cellularity and healed withless scarring than control wounds (Kyriakides et al.1999b) (Table 1). We conclude that both SPARC andTSP-2 affect the dermal component of cutaneous wound

Figure 5 Enhanced migration of dermal fibroblasts from ST-dou-ble-null mice. Dermal fibroblasts were isolated and cultured untilconfluence. Wounds were created in the cell monolayers, and cellmigration into the denuded area was analyzed as the percentageof area covered by the cells as a function of time. Fibroblasts fromthe skin of ST-double-null mice (gray bars) migrated significantlyfaster than those from WT animals (white bars) at 24 and 48 hr.*p�0.01.

Figure 6 Fibrovascular invasion of PVA sponges was increased inST-double-null mice. PVA sponges were implanted subcutaneouslyinto ST-double-null and WT mice and were evaluated histologi-cally at 14 days after implantation. Fibrovascular invasion of thesponges was determined as the percentage of the total area of thesponge devoted to cells, vasculature, and ECM. The invasion wasincreased significantly (*p�0.01) in ST-double-null relative to WTmice.

Figure 7 Production of MMP-2 by dermal fibroblasts. Equivalentnumbers of ST-double-null and WT dermal fibroblasts were cul-tured in DMEM/1% fetal calf serum. The culture media were condi-tioned for 3 days, after which zymography was performed. Controllane contained an equivalent amount of medium not conditionedby cells. The representative zymogram indicates that cells from ST-double-null mice produced higher amounts of MMP-2 (arrow) thancells from their WT counterparts. The band migrating at �90 kDarepresents MMP-9 (arrowhead). Molecular mass standards � 10�3

in kDa, on right.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

Enhanced Wound Repair in SPARC-TSP-2-null Mice 579

healing by their modulation of angiogenesis and bythe production and assembly of ECM.

Given the capacity of SPARC to retard cell-cycleprogression in vitro, we asked whether differences incell proliferation might account for the increased heal-ing of ST-double-null mice. There was no significantdifference in cell proliferation assessed by BrdU incor-poration between wounds from ST-double-null andWT mice. This finding is consistent with data fromSPARC-null mice, in which cell proliferation was ex-

cluded as a mechanism that could account for the ac-celerated wound healing in these animals (Bradshawet al. 2002) (Table 1).

The presence of SPARC in the wound area, withmaximal levels at days 4 and 7 after wounding,supports our previous conclusions (Reed et al. 1993;Puolakkainen et al. 1999) and speaks to a potentiallysignificant function for this protein during the phaseof granulation tissue formation of cutaneous woundhealing. TSP-2 has previously been shown to be pres-ent in both early and late wounds, either associatedwith cells or more widely distributed in cells and ECM(Kyriakides et al. 1999b). This temporal expression isimportant because the protein that is produced firstmay determine the course of the wound-healing pro-cess (see Table 1), as was observed in a study on TSP-1-and TSP-2-null mice (Agah et al. 2002).

Altered ECM is characteristic of mice with targeteddisruptions of genes for several matricellular proteins(Bornstein and Sage 2002). By hydroxyproline analy-sis, the concentration of collagen in SPARC-null skinwas found to be half that of WT skin (Bradshaw et al.2002). Electron microsopy studies have revealed thatthe collagen fibrils are smaller and more uniform insize in SPARC-null mice, relative to their WT counter-parts (Bradshaw et al. 2003b; Puolakkainen et al.2003). The wound beds of TSP-2-null mice also con-tained disorganized collagen fibers (Kyriakides et al.1999b). Given these findings, we asked whether al-tered ECM might account for the enhanced woundhealing observed in ST-double-null mice. By electronmicroscopy, we found that the collagen fibers in day-14 wounds of ST-double-null mice were small anduniform in diameter, whereas those from WT micewere larger and more variable in size. In WT animalsthe collagen fibrils were parallel and constituted awell-organized architecture. In contrast, in ST-double-null mice, a severely disrupted ECM was found.

Figure 8 ST-double-null dermal fibroblasts exhibit enhanced con-traction of native collagen gels. Over a fourfold range of cell concen-tration, fibroblasts from ST-double-null mice (open circles) con-tracted the gels to a greater extent than fibroblasts from WT mice(closed circles). Values are mean � SD (n4 replicates).

Table 1 Characteristics of SPARC-null, TSP-2-null, and ST-double-null mice in healing dermal excisional wounds

SPARC-null TSP-2-null ST-double-null

Epithelial closure ↑ � ↑(Rete pegs, thickened epidermis)

Dermal wound healing Granulation tissue � Inflammation � ↑Collagen deposition � Vascularized granulation tissue ↑

Total cell content ↑Scarring ↓

Cell proliferation in vivo � NS �

Cell migration in vitro ↑ NS ↑Fibrovascular invasion in vivo ↑ ↑ ↑MMP-2 protein in vitro �a ↑ ↑Collagen fibrils/ECM in vivo Small, uniform in size Lack parallel orientation Small, uniform in size

Collagen content ↓ Loose, disrupted ECMCollagen gel contraction in vitro � ↓ ↑

a�, Bradshaw, AD and Sage, EH, unpublished data.Findings in wounds from SPARC-null, TSP-2-null, and ST-double-null mice are listed, in comparison to their own WT counterparts. ↑, accelerated or enhanced; �,no difference between null and WT; ↓, decreased; NS, not studied.

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

580 Puolakkainen, Bradshaw, Brekken, Reed, Kyriakides, Funk, Gooden, Vernon, Wight, Bornstein, Sage

Collagen gel contraction has been proposed as amodel in vitro for wound contraction in vivo that re-sults in part from the contractile action of myo-fibroblasts within the wound bed (Grinnell 1994). Itis noteworthy, therefore, that the contraction of gelsin vitro by ST-double-null dermal fibroblasts was in-creased relative to WT fibroblasts, a result in accor-dance with the increased rate of wound closure inST-double-null mice. The mechanism by which con-comitant ablation of SPARC and TSP-2 enhances thecapacity of fibroblasts to remodel collagen in vitro isunclear, particularly because single-null SPARC andTSP-2 fibroblasts exhibited, respectively, no differencein collagen gel contraction (Bradshaw et al. 2002) anddiminished gel contraction (Kyriakides et al. 1999b),relative to WT fibroblasts. It is possible that the en-hanced contractile action of fibroblasts characteristicof the ST-double-null phenotype is, in part, a conse-quence of synthesis of an abnormal pericellular ECMin association with elevated levels of MMP-2.

Fibrovascular invasion of PVA sponges was increasedin SPARC-null mice (Bradshaw et al. 2002), and Kyria-kides et al. (2001) reported increased angiogenesis andfibrotic encapsulation of sponges in TSP-2-null mice.Therefore, we were not surprised to find a significantlyincreased fibrovascular invasion of PVA sponges in ST-double-null animals, relative to WT mice. The fibrovas-cular invasion of PVA sponges has been shown to re-flect angiogenic responses. Thus, part of the enhancedwound repair in ST-double-null mice might be due toincreased angiogenesis during the healing process.

In summary, our findings indicate that the wound-healing phenotype of ST-double-null mice representscollectively the characteristics of the single-null ani-mals. Lack of SPARC seems to dominate epithelialclosure, whereas lack of TSP-2 dominates dermalwound healing. Both lead to altered ECM in thewound tissue. We propose that the enhanced cutane-ous wound healing seen in ST-double-null mice resultsfrom accelerated epidermal closure and alterations inthe ECM. Collectively, and consistent with earlier re-ports, these findings indicate that SPARC and TSP-2have important roles in epidermal wound healing andin the repair of the dermis in response to injury.

AcknowledgmentsThis work was supported in part by grants from the Na-

tional Institutes of Health (F32 HL-10352 to RAB, K01 AR-0022200 to ADB, R01 AR-45418 to PB, and R01 GM-40711and R01 HL-59574 to EHS), from the National ScienceFoundation to the University of Washington Engineered Bio-materials (EEC9529161), from The Gilbertson Foundationto The Hope Heart Program, and from the Helsinki Univer-sity Central Hospital Research Funds (EV0, Finland) (to PAP).

The authors thank Eileen Neligan for assistance with themanuscript.

Literature CitedAgah A, Kyriakides TR, Lawler J, Bornstein P (2002) The lack of

thrombospondin-1 (TSP 1) dictates the course of wound healingin double-TSP1/TSP2-null mice. Am J Pathol 161:831–839

Armstrong LC, Bjorkblom B, Hankenson KD, Siadak AW, StilesCE, Bornstein P (2002) Thrombospondin 2 inhibits microvascu-lar endothelial cell proliferation by a caspase-independent mech-anism. Mol Biol Cell 13:1893–1905

Bornstein P, Armstrong LC, Hankenson KD, Kyriakides TR, YangZ (2000) Thrombospondin 2, a matricellular protein with di-verse functions. Matrix Biol 19:557–568

Bornstein P, Sage EH (2002) Matricellular proteins: extracellularmodulators of cell function. Curr Opin Cell Biol 14:608–616

Bradshaw AD, Graves DC, Motamed K, Sage EH (2003a) SPARC-null mice exhibit increased adiposity without significant differ-ences in overall body weight. Proc Natl Acad Sci USA 100:6045–6050

Bradshaw AD, Puolakkainen P, Dasgupta J, Davidson JM, WightTN, Sage EH (2003b) SPARC-null mice display abnormalities inthe dermis characterized by decreased collagen fibril diameterand reduced tensile strength. J Invest Dermatol 120:949–955

Bradshaw AD, Reed MJ, Carbon JG, Pinney E, Brekken RA, SageEH (2001) Increased fibrovascular invasion of subcutaneouspolyvinyl alcohol sponges in SPARC-null mice. Wound Rep Reg9:522–530

Bradshaw AD, Reed MJ, Sage EH (2002) SPARC-null mice exhibitaccelerated cutaneous wound closure. J Histochem Cytochem50:1–10

Bradshaw AD, Sage EH (2001) SPARC, a matricellular protein thatfunctions in cellular differentiation and tissue response to injury.J Clin Invest 107:1049–1054

Brekken RA, Puolakkainen P, Graves DC, Workman G, Lubkin SR,Sage EH (2003) Enhanced growth of tumors in SPARC null miceis associated with changes in the ECM. J Clin Invest 111:487–495

Brekken RA, Sage EH (2000) SPARC, a matricellular protein: at thecrossroads of cell-matrix communication. Matrix Biol 19:569–580

Delany AM, Amling M, Priemel M, Howe C, Baron R, Canalis E(2000) Osteopenia and decreased bone formation in osteonectin-deficient mice. J Clin Invest 105:915–923

Gilmour DT, Lyon GJ, Carlton MBL, Sanes JR, Cunningham JM,Anderson JR, Hogan BLM, et al. (1998) Mice deficient for thesecreted glycoprotein SPARC/osteonectin/BM-40 develop nor-mally but show severe age-onset cataract formation and disrup-tion of the lens. EMBO J 17:1860–1870

Graves DC, Yablonka-Reuveni Z (2000) Vascular smooth musclecells spontaneously adopt a skeletal muscle phenotype: a uniqueMyf5(�)/MyoD(�) myogenic program. J Histochem Cytochem48:1173–1193

Greenhalgh DG, Sprugel KH, Murray MJ, Ross R (1990) PDGFand FGF stimulate wound healing in the genetically diabeticmouse. Am J Pathol 136:1235–1246

Grinnell F (1994) Fibroblasts, myofibroblasts and wound contrac-tion. J Cell Biol 124:401–404

Kasugai S, Todescan R, Nagata T, Yao K-L, Butler WT, Sodek J(1991) Expression of bone matrix proteins associated with min-eralized tissue formation by adult rat bone marrow cells in vitro:inductive effects of dexamethasone on the osteoblastic pheno-type. J Cell Physiol 147:111–117

Koike T, Vernon RB, Hamner MA, Sadoun E, Reed MJ (2002)MT1-MMP, but not secreted MMPs, influences the migration ofhuman microvascular endothelial cells in 3-dimensional collagengels. J Cell Biochem 86:748–758

Kyriakides TR, Leach KJ, Hoffman AS, Ratner BD, Bornstein P(1999a) Mice that lack the angiogenesis inhibitor, thrombospon-din 2, mount an altered foreign body reaction characterized byincreased vascularity. Proc Natl Acad Sci USA 96:4449–4454

Kyriakides TR, Tam JWY, Bornstein P (1999b) Accelerated woundhealing in mice with disruption of the thrombospondin 2 gene. JInvest Dermatol 113:782–787

by guest on June 15, 2013jhc.sagepub.comDownloaded from

The

Jour

nal o

f His

toch

emis

try

& C

ytoc

hem

istr

y

Enhanced Wound Repair in SPARC-TSP-2-null Mice 581

Kyriakides TR, Zhu YH, Smith LT, Bain SD, Yang Z, Lin MT,Danielson KG, et al. (1998) Mice that lack thrombospondin 2display connective tissue abnormalities that are associated withdisordered collagen fibrillogenesis, an increased vascular density,and a bleeding diathesis. J Cell Biol 140:419–430

Kyriakides TR, Zhu YH, Yang Z, Huynh G, Bornstein P (2001) Al-tered extracellular matrix remodeling in sponge granulomas ofthrombospondin 2-null mice. Am J Pathol 159:1255–1262

Liaw L, Birk D, Ballas CB, Whitsitt JS, Davidson JM, Hogan BLM(1998) Altered wound healing in mice lacking a functional os-teopontin gene (ssp1). J Clin Invest 101:1468–1478

Mackie EJ, Halfter W, Liverani D (1988) Induction of tenascin inhealing wounds. J Cell Biol 107:2757–2767

Norose K, Clark JI, Syed NA, Basu A, Heberkatz ES, Sage EH,Howe CC (1998) SPARC deficiency leads to early-onset catarac-togenesis. Invest Ophthalmol Vis Sci 39:2674–2680

Puolakkainen P, Bradshaw A, Kyriakides TR, Reed M, Brekken R,Wight T, Bornstein P, et al. (2003) Compromised production ofextracellular matrix in mice lacking Secreted Protein, Acidic andRich in Cysteine (SPARC) leads to a reduced foreign body reac-tion to implanted biomaterials. Am J Pathol 162:627–635

Puolakkainen P, Reed M, Vento P, Sage EH, Kiviluoto T, Kivi-laakso E (1999) Expression of SPARC in healing intestinal anas-tomosis and short bowel syndrome in rats. Dig Dis Sci 44:1554–1564

Reed MJ, Penn PE, Li Y, Birnbaum R, Vernon RB, Johnson TS,Pendergrass WR, et al. (1996) Enhanced cell proliferation andbiosynthesis mediate improved wound repair in refed, caloric-restricted mice. Mech Ageing Dev 89:21–43

Reed MJ, Puolakkainen P, Lane TF, Dickerson D, Bornstein P, SageEH (1993) Differential expression of SPARC and thrombospon-

din 1 in wound repair: Immunolocalization and in situ hybridiza-tion. J Histochem Cytochem 41:1467–1477

Sage EH, Decker J, Funk S, Chow M (1989a) SPARC: a Ca2�-bind-ing extracellular protein associated with endothelial cell injuryand proliferation. J Mol Cell Cardiol 21:13–22

Sage EH, Vernon RB, Decker J, Funk S, Iruela-Arispe ML (1989b)Distribution of the calcium-binding protein SPARC in tissues ofembryonic and adult mice. J Histochem Cytochem 37:819–824

Spenny ML, Muangman P, Sullivan SR, Bunnett NW, Ansel JC, Ol-erud JE, Gibran NS (2002) Neutral endopeptidase inhibition indiabetic wound repair. Wound Repair Regen 10:295–301

Stenner DD, Tracy RP, Riggs BL, Mann KG (1986) Human plate-lets contain and secrete osteonectin, a major protein of mineral-ized bone. Proc Natl Acad Sci USA 83:6892–6896

Vernon RB, Gooden MD (2002) An improved method for the col-lagen gel contraction assay. In Vitro Cell Dev Biol 38:97–101

Vuorio T, Kahari V-M, Black C, Vuorio E (1991) Expression of os-teonectin, decorin, and transforming growth factor beta 1 genesin fibroblast cultures from patients with systemic sclerosis andmorphea. J Rheumatol 18:247–253

Wight TN, Lara S, Riessen R, Le Baron R, Isner J (1997) Selectivedeposits of versican in the extracellular matrix of restenotic le-sions from human peripheral arteries. Am J Pathol 151:963–973

Yan Q, Clark JI, Wight T, Sage EH (2002) Alterations in the lenscapsule contribute to cataractogenesis in SPARC-null mice. J CellSci 115:2747–2756

Yang Z, Kyriakides TR, Bornstein P (2000) Matricellular proteins asmodulators of cell-matrix interactions: adhesive defect in throm-bospondin-2-null fibroblasts is a consequence of increased levelsof matrix metalloproteinase-2. Mol Biol Cell 11:3353–3364

by guest on June 15, 2013jhc.sagepub.comDownloaded from