In Vivo Evidence for a Bridging Role of a Collagen V Subtype at the Epidermis–Dermis Interface

ORIGINAL ARTICLE

In Vivo Evidence for a Bridging Role of a Collagen V Subtype at the Epidermis–Dermis Interface Christelle Bonod-Bidaud1,2,7, Muriel Roulet1,2,7, Uwe Hansen3,8, Ahmed Elsheikh4,8, Marilyne Malbouyres1,2, Sylvie Ricard-Blum2,5, Cle´ment Faye2,5, Elisabeth Vaganay1,2, Patricia Rousselle2,6 and Florence Ruggiero1,2 Collagen V is the defective product in most cases of classical Ehlers–Danlos syndrome (EDS), a connective tissue disorder typically characterized by skin fragility and abnormal wound healing. Collagen V assembles into diverse molecular forms. The predominant a1(V)2a2(V) heterotrimer controls fibrillogenesis in skin and other tissues. The a1(V)3 minor form is thought to occur in skin, but its function is unknown. To elucidate its role, we generated transgenic mice that overexpress the human a1(V)3 homotrimer in the epidermis. The transgenederived product is deposited as thin unstriated fibrillar material in the basement membrane zone of embryonic and perinatal epidermis and hair follicles. Accumulation of a1(V)3-containing fibrils leads to ultrastructural modifications at the epidermis–dermis interface and provokes changes in biomechanical properties, although not statistically significant. Using superparamagnetic immunobeads to isolate authentic suprastructures and protein-binding assays, we demonstrate that the homotrimer is part of a protein network containing collagen IV, laminin-111, and the dermal collagen VI. Our data show that the homotrimer serves as a bridging molecule that contributes to the stabilization of the epidermal–dermal interface. This finding strongly suggests that collagen V may be expressed in skin as different subtypes with important but distinct roles in matrix organization and stability. Journal of Investigative Dermatology (2012) 132, 1841–1849; doi:10.1038/jid.2012.56; published online 22 March 2012

INTRODUCTION Collagen V is a quantitatively minor component of collagen I–rich tissues such as the dermis, tendon/ligament, bone, blood vessel, and cornea. In contrast to cornea where the concentration of this collagen type is relatively high (20–25%), collagen V represents only 1–3% of the total collagen content in skin (Fichard et al., 1995). The low representation of collagen V in skin contrasts with the severity of the skin phenotype in patients with classic Ehlers–Danlos syndrome (EDS), a heterogeneous heritable connective tissue 1

Institut de Ge´nomique Fonctionnelle de Lyon, ENS de Lyon, UMR CNRS 5242, SFR BioSciences Gerland, Lyon, France; 2Universite´ Lyon 1, Villeurbanne, France; 3Institute for Physiological Chemistry and Pathobiochemistry, University Hospital of Muenster, Muenster, Germany; 4 School of Engineering, University of Liverpool, Liverpool, UK; 5UMR CNRS 5086, BMSSI, Institut de Biologie et Chimie des Prote´ines, Lyon, France and 6 SFR BioSciences Gerland-Lyon Sud, Institut de Biologie et Chimie des Prote´ines, FRE 3310, CNRS, Lyon, France 7

These authors contributed equally to this work.

8

These authors contributed equally to this work.

Correspondence: Florence Ruggiero, Institut de Ge´nomique Fonctionnelle de Lyon, ENS de Lyon, UMR CNRS 5242, 46 Alle´e d 0 Italie, 69364 Lyon, France. E-mail: [email protected] Abbreviations: d.p.c., days post coitum; EDS, Ehlers–Danlos syndrome; K14, keratin 14; PBS, phosphate-buffered saline; RT, reverse-transcriptase; SPR, surface plasmon resonance; TGF-b, transforming growth factor-b Received 16 September 2011; revised 16 January 2012; accepted 30 January 2012; published online 22 March 2012

& 2012 The Society for Investigative Dermatology

disorder essentially caused by mutations in collagen V genes, COL5A1 and COL5A2. Mutations in the human third chain a3(V) (COL5A3) gene have not been reported. This disorder is primarily characterized by soft and hyperextensible skin, impaired wound healing with atrophic scars, and reduced tensile strength (Malfait et al., 2005). Ultrastructural analyses of skin biopsies of EDS patients essentially revealed defects in collagen fibrillogenesis with a decrease in fiber density and the presence of large irregular collagen fibers (Giunta and Steinmann, 2000; Bouma et al., 2001). Collagen V assembles in different isoforms that differ in chain composition and whose in vivo significance is unclear (Ricard-Blum and Ruggiero, 2005). The most abundant and widely distributed isoform is the a1(V)2a2(V) heterotrimer, but different minor forms are also expressed in tissues as the a1(V)3 homotrimer first observed in cell cultures and embryonic tissues (Haralson et al., 1984; Moradi-Ameli et al., 1994). In skin, the major role of a1(V)2a2(V) heterotrimer in the nucleation and control of collagen I/V heterofibrils has been well documented using mouse models for EDS (ChanutDelalande et al., 2004; Wenstrup et al., 2004). Interestingly, skin analysis of the pN (exon 6 deletion in Col5a2 gene) mouse model for EDS also revealed that collagen V may have isoform-specific roles and underscored the role of a1(V)3 homotrimer in skin development (Chanut-Delalande et al., 2004). Along this line, in vitro fibrillogenesis experiments showed that, unlike the heterotrimer, recombinant a1(V)3 www.jidonline.org 1841

C Bonod-Bidaud et al.

RESULTS Generation of K14-COL5A1 mice and characterization of the transgene-derived product

Transgenic mouse lines (K14-COL5A1) overexpressing the human proa1(V) chain in the epidermis under the control of the K14 promoter (Vassar et al., 1989) were created and characterized (Supplementary Figure S1 online). Correct expression of the transgene was attested by immunofluorescence staining of HaCat cells transiently transfected with the K14-COL5A1 construct, with the 18G5 antibody that recognizes the N-terminal end of the proa1(V) chain (BonodBidaud et al., 2007; Supplementary Figure S1A and B online). The construct was then injected into fertilized eggs to generate the K14-COL5A1 transgenic mouse lines. Three independent transgenic founders (Tg1-3) were identified by PCR using specific primers; the strongest expression in skin was observed in founder 1 (Supplementary Figure S1C online). The tissue specificity of the K14 promoter was 1842 Journal of Investigative Dermatology (2012), Volume 132

Ep

17

.5

d. p. c. 7. 5 W

T–

–1 Tg

homotrimer did not form heterotypic fibrils when mixed with collagen I, but polymerize into distinctly thin filaments (Chanut-Delalande et al., 2001). The a1(V)3 homotrimer was instead localized as thin filamentous structures at the surface of large collagen I fibrils and did not regulate fibril assembly. The localization of these thin fibrils indicates that the homotrimer can act as a molecular linker between collagen fibrils and/or macromolecules in the extracellular matrix (ChanutDelalande et al., 2001). Consistent with this finding, very thin (5–10 nm in diameter) unstriated fibrils made of collagen V were previously reported immediately near epithelial basement membranes and extending into the adjacent interstitial matrix (Fitch et al., 1984; Modesti et al., 1984; Birk et al., 1986; Adachi and Hayashi, 1994). However, the paucity of collagen V homotrimer in tissues did not allow further biochemical and functional analyses. To further detail the in vivo assembly and function of collagen V homotrimer in skin, we generated a transgenic mouse line (K14-COL5A1) overexpressing the human proa1(V) chain under the control of the keratin 14 (K14) epidermis-specific promoter (Vassar et al., 1989). This strategy was designed to produce the homotrimer in cells that do not express the proa2(V), with the aim to provoke its accumulation at the epidermis–dermis interface during skin development. Proa2(V) mRNA were not detected in the epidermis during mouse development (Andrikopoulos et al., 1992), whereas proa1(V) mRNA was detected in the ectoderm at 10.5 d.p.c. (days post coitum), but expression ceased at 12.5 d.p.c. when epidermal cells started to differentiate (Roulet et al., 2007). We showed that overexpression of a1(V)3 homotrimer provokes a large increase in the thin fibril density present underneath the epidermal basement membrane and results in changes in the ultrastructural and biomechanical properties of the skin. The higher content in collagen V thin fibrils at the epidermal– dermal interface allowed the detailed analysis of the in vivo role of a1(V)3 homotrimer. Our findings unravel an important role of the minor a1(V)3 molecular form in epidermis–dermis organization and stability during mouse skin development.

d. p. c.

Bridging Role of Collagen V Homotrimer in Skin

D Ep

pb

D

200

hCOL5A1 mCol5a2 mGapdh

300 160 Tg–5 d WT–5 d Ep D Ep D

WT

Tg

250

mCol1a1

600

mCol6a1

600

mTn-X

150

mTgf1

200

hCOL5A1

160

mTgf2

300

mCol5a2

100

mTgf3

160

mGapdh

200

mGapdh

pb

kDa 200

WT Tg

WT

Tg

WT Tg

1 hpNα1(V) mα2(V)

2

3

4

kDa

116

250 150 100

97

50 37 25

hproα1(V) hpNα1(V)

IgG heavy chains IgG light chains bp

ce

bp

ce

Figure 1. Transgene expression. (a, b) Skin was dissected from 17.5 d.p.c. (days post coitum) and 5-day-old mice, both wild-type (WT) and transgenic (Tg). (a) The epidermis (Ep) was separated from the dermis (D). Reversetranscriptase (RT)-PCR was performed with specific primers on (a) the dermis and epidermis or (b) total skin. (c) Immunoblot analysis of human proa1(V) and mouse proa2(V) expression in 17.5 d.p.c. skin embryos. Membranes were probed with (left) 13F6 antibody or with (right) polyclonal antibodies that recognize the murine proa2(V) chain. (d) Crude extracts (lanes 2 and 4) and isolated suprastructural fragments (lanes 1 and 3) from mouse skin. bp, bead pellet; ce, crude extract.

checked by performing reverse-transcriptase (RT)-PCR on RNA isolated from different tissues using primers specific for the transgene product (Supplementary Figure S1D online). We also checked that transgenic epidermis does not express endogenous proa2(V), because it may form a hybrid heterotrimeric molecule with the human proa1(V) transgene product. We thus chemically separated the epidermis from the dermis (Trost et al., 2007) and performed RT-PCR on isolated epidermis and dermis RNA extracts from embryonic (17.5 d.p.c.) and perinatal (5 days old) skin using mCol5a2 and hCOL5A1 specific primers. We showed that mCol5a2 expression is restricted to the dermis in skin, excluding a possible production of the heterotrimer in transgenic epidermis (Figure 1a). Substantial expression of hCOL5A1 was detected in the dermis (Figure 1a), suggesting that treatment with sodium thiocyanate to separate the epidermis from the dermis (Trost et al., 2007) did not prevent the presence of contaminant epidermal cells, probably from hair follicle sheath that also expresses K14, which remain attached to the dermis. Overexpression of human proa1(V) in the epidermis did not change significantly the expression of major skin matrix components and signaling factor mRNAs (Col1a1, Col6a1, Tenascin-X, Tgfb1-3) as analyzed by semiquantitative RT-PCR (Figure 1b) and densitometry (not shown). The expression of the proa1(V) homotrimer in transgenic mice was confirmed by immunoblot analysis of 17.5 d.p.c.

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

embryo skin extracts using specific antibody to human proa1(V) chain. After secretion, procollagen V is proteolytically processed into mature protein. Thus, different procollagen V molecular forms can be identified from tissue extracts: the proa1(V) that contains the N- and C-propeptides, intermediate processed forms, and the proNa1(V) chain, the latter corresponding to the mature chain. An intense band corresponding to human proNa1(V) chain was observed in transgenic skin extracts (Figure 1c, left panel) but not in wildtype extracts, indicating that the transgene-derived product was correctly produced and secreted by the epidermis. By re-probing the membrane with antibody to mouse a2(V) chain, a band corresponding to the a2(V) chain position was equally detected in wild-type and transgenic skin extracts (Figure 1c, right panel). Thereafter, crude extracts from perinatal mouse skin containing suprastructural fragments obtained by mechanical disruption of skin were used for the isolation of a1(V)3 homotrimer aggregates by superparamagnetic immunobeads. The human proa1(V) chain and the processed pNa1(V) form were detected in crude transgenic skin extracts (Figure 1d, lane 2), as well as in the isolated suprastructures (Figure 1d, lane 1), whereas no signal could be detected in wild-type mice (Figure 1d, lanes 3 and 4). The a1(V) homotrimer is deposited in the basement membrane zone as thin unstriated fibrils associated with ultrastructural changes in skin

Immunohistochemical localization of the a1(V) chain was performed in cross-sections of 5-day-old mouse skin using the 18G5 antibody directed against the N-terminal end of the proa1(V) chain (Bonod-Bidaud et al., 2007) and detecting both endogenous and transgene-derived products. In wildtype and transgenic skin, collagen V was detected in the dermis as a weak and diffuse signal (Figure 2a). However, the staining was substantially more intense at the dermis– epidermis junction zone in transgenic skin (Figure 2a, top right panel). In addition, a strong signal, undetectable in wild-type skin, was observed around hair follicles (Figure 2a, bottom right panel). Interestingly, hair follicles appeared more numerous in transgenic mouse skin (Figure 2a). This was further confirmed in cross-sections stained with Masson’s trichrome (Figure 2b) and by scanning electron microscopy showing hair density of 5-day-old wild-type and transgenic mouse pelage (Supplementary Figure S2 online). Hair count performed on depilated skin revealed that the number of hair strands was in average 18.5% higher in K14-COL5A1 transgenic mice than in wild-type mice (Figure 2b, bright field and histogram). The high number of hair strands in transgenic skin can explain the substantial expression of the transgene detected in the dermis after sodium thiocyanate treatment (Figure 1a). Under the transmission electron microscope (TEM), thin unstriated fibrils (about 5 nm in diameter) are observed in the dermal–epidermal junction zone of 17.5 d.p.c. wild-type embryos (Figure 2c, left panel arrows) and 5-day-old mice (Figure 2c, right panel). A particularly strong accumulation of this thin fibrillar material was observed underneath the dermal–epidermal junction of K14-COL5A1 transgenic skin

(Figure 2c, right panel, arrows). Interestingly, with the increase in collagen V homotrimer thin fibrils at the basement membrane zone, we showed that the number of hemidesmosomes per micrometer of apical membrane of basal epithelial cells was in average 1.7 times higher in modified mice (0.8 mm1 vs. 0.47 mm1 in wild-type) (Figure 2d). Immunogold labeling of the dermal–epidermal junction zone of the K14-COL5A1 transgenic skin with mAb specific to the human proa1(V) chain (13F6) confirmed that the transgene-derived product localizes underneath the dermal–epidermal basement membrane and is deposited as thin unstriated filamentous suprastructures and as a component of networklike suprastructures (Figure 3b–d compared with wild type Figure 3a). Interestingly, these suprastructures were decorated with gold particles without any pretreatment with acetic acid, indicating that collagen V homotrimer epitopes were unmasked and readily accessible to antibody. Furthermore, in agreement with the immunohistochemical staining, mainly network-like suprastructures were labeled in the extracellular matrix surrounding the hair follicles of transgenic mice (Figure 3e and f). As expected, the dermal collagen fibrils were not labeled at all. Overexpression of human a1(V)3 homotrimer led to other minor changes in the morphology of basal epidermal cells. Whereas the basal layer cells of wildtype epidermis are lined up and contiguous, lacunae were often observed between basal cells of 17.5 d.p.c. transgenic epidermis (Figure 2c, left panel, asterisks). Because the presence of lacunae between cells can reflect high rate of proliferation, epidermal basal cell proliferation was estimated by the number of Ki-67-positive cells in transgenic skin compared with wild type. No significant difference in proliferation rate was observed in transgenic and wild-type skin (data not shown). Overexpression of a1(V)3 homotrimer modifies skin biomechanical properties

To analyze possible alteration of the skin extensibility, 1-month-old mice were killed and held by the skin of their back. No change in skin extensibility was observed in K14COL5A1 mice (in average 44 mm, n ¼ 23). However, when the mice were held by the skin of their back for few minutes and put back down, the fold formed by skin pinching relaxed much slower in transgenic mice than in wild type, indicating that overexpression of a1(V) homotrimer may alter skin deformability. In parallel experiments, the skin of mice was stretched to breaking point while simultaneously measuring the applied force and resulting extension (Supplementary Figure S3 online). The data are plotted as stress versus strain (Figure 4a). On average, the transgenic skin showed a flatter slope (lower stiffness) and a lower rupture stress than the wild-type skin. Figure 4b shows the load extension behavior of the wild-type and transgenic skin specimens subjected to splitting tests (Supplementary Figure S3 online). The behavior went through three stages starting with low stiffness (low resistance to splitting) up to an extension of approximately 1.2 to 1.4 mm, followed by a higher stiffness stage up to an extension of about 2.5 mm. At the end of this stage, some of www.jidonline.org 1843

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

WT–5 d

WT–5 d

K14-COL5A1–5 d

50 µm

K14-COL5A1–5 d

50 µm

100 µm

100 µm

Hair follicle number

80

100 µm

100 µm

WT–17.5 d.p.c.

100 µm

100 µm

**

60 40 20 0 WT

K14COL5A1

WT–5 d

K14-COL5A1–17.5 d.p.c.

Ke

*

Fb Ke

1 µm

K14-COL5A1–5 d Fb

*

2 µm

2 µm

1 µm

K14-COL5A1–17.5 d.p.c.

1.2 HD number µm–1

WT–17.5 d.p.c. Ke

Ke

***

1.0 0.8 0.6 0.4 0.2

1 µm

1 µm

0.0 WT

K14COL5A1

Figure 2. Analysis of skin structure and ultrastructure. (a) Immunolocalization of the transgene using the monoclonal 18G5 antibody in (left panel) 5-day-old wild-type (WT) and (right panel) transgenic mice (K14-COL5A1). (b, upper panel) Masson’s trichrome staining of skin cross-sections of 5-day-old WT and transgenic mice. (b, lower panel) Light microscope micrographs of depilated 5-day-old epidermis and histogram of hair follicle counting. **Po0.01. (c, d) Ultrastructural analysis of 17.5 d.p.c. (days post coitum) (left) and 5-day-old (right) transgenic and WT mouse skin. (c) Asterisks indicate lacunae in the basal layer of the transgenic epidermis; arrows show the accumulation of dense unstriated fibrillar material underneath the basement membrane in transgenic skin compared with WT. (d) Arrows indicate hemidesmosomes. Histogram of hemidesmosome (HD) counting, ***Po0.001. Fb, fibroblast; Ke, keratinocyte.

1844 Journal of Investigative Dermatology (2012), Volume 132

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

Keratinocyte Keratinocyte

BM

BM

fib

hs

Keratinocyte

Keratinocyte

BM BM fib

Figure 3. Analysis of skin by immunogold electron microscopy. Ultrastructural analysis of the skin from (a) wild-type and (b–d) transgenic mice and hair follicles of (e, f) transgenic mice by using an antibody against human proa1(V) chain. Arrowheads indicate representative gold particles. BM, basement membrane; fib, dermal collagen fibrils; hs, hair shaft. Bar ¼ 200 nm.

0.12

0.3 (Wild-type mice) (Transgenic mice)

Wild-type mice Transgenic mice

0.1 Splitting load (N)

Stress (MPa)

0.25 0.2 0.15 0.1

0.08 0.06 0.04 0.02

0.05

0

0 0

0.2

0.4

0.6 –1

Strain (mm mm )

0.8

0

1

2

3

4

5

Extension (mm)

Figure 4. Analysis of biomechanical properties of 1-month-old mouse skin. (a) Stress/strain behavior of mouse skin. (b) Splitting tests results.

the tissue specimens ruptured, whereas others continued to resist additional splitting load while experiencing low stiffness. However, no notable difference could be observed between the two specimen types. Collagen V homotrimer serves as a bridging molecule at the epidermal–dermal interface

The unique location and structure of a1(V)3 homotrimer suggest that it may serve as a molecular bridge between the epidermis basement membrane and dermal collagen fibers. To test this possibility, collagen V–containing suprastructures were isolated from mechanical disrupted skin extracts using magnetic immunobeads (Figure 5a). The isolated suprastructures had an amorphous and network-like morphology. We showed that the human proa1(V) chain, as well as the ECM components, collagen VI (Figure 5a (A and B) and Supplementary Figure S4 online), laminin g1 (a chain of the laminin111) (Figure 5a (C) and Supplementary Figures S4, S5 online), collagen IV (Figure 5a (D and E)), and tenascin-X (Figure 5a (F)) are structural components of these networks. We thus reasoned that a1(V)3 homotrimer is a part of a protein network containing basement membrane components and dermal collagen VI. To confirm this hypothesis, we next used surface plasmon resonance (SPR) technology to identify direct interaction with the a1(V)3 homotrimer.

Because the homotrimer is present in low amount in tissues, we used the previously characterized recombinant a1(V)3 homotrimer that contains the entire N-propeptide (Fichard et al., 1997) as the immobilized ligand, as the proa1(V) chain was shown to retain a part of the N-propeptide in skin (Colige et al., 2005). We showed that soluble pepsinized collagen VI and IV, but not laminin, bound to recombinant proa1(V)3 homotrimer spotted onto Gold affinity chips, and the complexes formed were stable as shown by the slow dissociation phase (Figure 5b). However, using solid-phase assay and purified laminin-111 (Rousselle et al., 1997), we were able to show direct binding between immobilized proa1(V) chain and soluble laminin-111, thus confirming the immunogold labeling (Figure 5c). To detail the a1(V) domain(s) binding sites, recombinant fragments of the Npropeptide (the full-length N-propeptide Na1(V), and the NC3 domain containing the TSPN domain and the variable region) were further analyzed for their capacity to bind collagens IV and VI and laminin-111. Both collagens bind to the Na1(V) and to the NC3 domain using SPR, suggesting that the binding sites are both located in the globular domain of the N-propeptide, whereas no binding was obtained between the Na1(V) and laminin-111 using solid-phase assay, indicating that the laminin-111 binding site is located in the C-terminal non-globular end of the N-propeptide (Figure 5c). Combined, our data fit with a head-to-head polymerization of the a1(V)3 monomers to bridge basement membrane components to dermal proteins (Figure 5d). DISCUSSION Earlier studies have shown that the collagen V heterotrimer is present in most tissues as a buried component in collagen I fibrils, and that its immunodetection in tissues required the use of epitope unmasking (Birk, 2001). However, unmasked collagen V was also immunolocalized as thin filaments in the immediate vicinity of basement membranes (Gay et al., 1981; Modesti et al., 1984; Birk et al., 1986). From these immunostainings, an anchoring function between basement and stromal matrix was proposed for unmasked collagen V. www.jidonline.org 1845

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

Resonance units (RU)

1,000 800 600 Collagen IV Collagen VI

400 200 0 –10

10 20 30 40 Time (minutes)

0

bp

50

60

Injected proteins Immobilized recombinant fragments Collagen IV Collagen VI

bp

Nα1(V)

TSPN VR COL

+

+

NC3

TSPN VR

+

+

bp 6,000 OD (578 nm)

5,000 4,000 3,000

pN[α1(V)]3

2,000

Nα1(V)

1,000 0 0

5 10 15 –1 Laminin 111 (μg ml )

20

Keratinocyte

Basement membrane Collagen [α1(V)]3 Collagen VI

Col I/Col IV heterofibrils Fibroblast Collagen I Collagen [α1(V)]2α2(V)

Figure 5. Binding partners of collagen V. (a) Transmission electron microscope (TEM) analysis of suprastructures isolated with superparamagnetic immunobeads. (A, B) Double labeling with antibodies against collagen VI (large gold particles, black arrowheads) and the human proa1(V) chain (small gold particles, white arrowheads). (C) Double labeling with antibodies against laminin-g-1 chain (black arrowheads) and the human proa1(V) chain (white arrowheads). (D, E) Double labeling with antibodies against collagen IV (large gold particles, black arrowheads) and the human proa1(V) chain (small gold particles, white arrowheads). (F) Double labeling with antibodies against tenascin-X (large gold particles, black arrowheads) and the human proa1(V) chain (small gold particles, white arrowheads). Bp, superparamagnetic immunobeads. Bar ¼ 100 nm. (b) Surface plasmon resonance (SPR) assays. Top: Injection of soluble pepsinized collagens IV and VI over spotted recombinant collagen V homotrimer. Bottom: Interactions between immobilized recombinant domains of collagen V and collagens IV and VI. (c) Solid-phase binding assays. Soluble lanimin-111 binding to immobilized recombinant collagen V homotrimer (pN[a1(V)]3 and N-propeptide (Na1(V)). OD, optical density. (d) Diagram showing the organization and location of collagen V subtypes (homotrimer vs. heterotrimer) expressed in skin.

Our previous in vitro fibrillogenesis experiments strongly suggested that the exposed collagen V in tissues corresponds to the homotrimer collagen V subtype (Chanut-Delalande et al., 2001). Here, by generating transgenic mice overexpressing the collagen V homotrimer in the epidermis, we clearly identify the a1(V)3 homotrimer as a candidate molecule to serve bridging function in skin. This is in 1846 Journal of Investigative Dermatology (2012), Volume 132

agreement with previous TEM observations of the basement membrane–stromal architecture (Adachi et al., 1997). Moreover, consistent with this bridging function, the detailed nature of direct physical interactions between collagen V homotrimer and components of basement membrane/dermal zone has been determined. Using superparamagnetic immunobead technology, which allows the characterization of

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

native protein networks in tissues (Villone et al., 2008), we demonstrated that collagen V homotrimer directly binds to collagen IV–containing networks and collagen VI–beaded filaments, and thus is part of the structural network bridging epidermal basal membrane to dermal extracellular matrix. Furthermore, we evidenced that direct interactions are at least mediated by the a1(V) N-terminal recombinant domain. In agreement with these results, we recently identified, from a yeast two-hybrid screen, physical interactions between the a1(V) N-propeptide and collagen VI chains (Symoens et al., 2010). Collagens are large multimeric proteins that contain several binding sites for different molecules. The formation of a protein network that bridges collagen V homotrimer fibrils to collagen IV and laminin-111 on the basement membrane side, and collagen VI on the dermal side, is thus conceivable. Combined, the data allow us to propose a model for the organization and location of collagen V subtypes in skin (Figure 5d). The striking feature of K14-COL5A1 transgenic mice was the accumulation of thin fibrillar material composed of collagen V homotrimer underneath the epidermal basement membrane. Stiffness and rupture stress of the transgenic skin were lower compared with wild type, both parameters being influenced by the density of collagen fibers in tissues. Collagen fiber density and organization of transgenic dermis as judged by histology and TEM appeared normal (not shown). Thus, the results were interpreted as a higher stiffness of the epidermis–dermis cohesion in transgenic skin resulting in rupture at site of lower stiffness, i.e., the dermis. This might be correlated with the higher number of hemidesmosomes observed in transgenic skin compared with wild type. Our data provide the in vivo evidence that distinct molecular forms of a same collagen type can differentially influence extracellular matrix architecture and mechanical properties. To our knowledge, this is previously unreported. Regulation of assembly and disassembly of hemidesmosomes is of crucial importance in migration of keratinocytes during biological and pathological processes (Tsuruta et al., 2011). The increase in the number of hemidesmosmes observed in the K14-COL5A1 transgenic mice may also have an impact on their ability to migrate. The fact that collagens V and VI and tenascin-X are part of the same protein complex may be correlated with their implication in identical or overlapping connective tissue disorders. Mutations in tenascin-X and collagen V genes are both responsible for the classic form of EDS (Malfait and De Paepe, 2005). The Ullrich congenital muscular dystrophy caused by mutations in collagen VI genes exhibits clinical and ultrastructural characteristics overlapping with classic EDS (Kirschner et al., 2005). Lack of one of the elements of the protein network we have established here may destabilize the whole structure and cause defects in skin cohesion and function. Overexpression of the transgene in mouse skin also resulted in the detection of a strong signal of the transgene product around hair follicle and an unexpected increase in hair number. Interestingly, a large number of hair follicles was initially reported in pN/ mouse skin (Andrikopoulos et al., 1995), and we showed that the targeted deletion in Col5a2 gene in pN/ mice results in abnormal predominant

deposition of homotrimers into the dermal matrix (ChanutDelalande et al., 2004). Thus, the two mouse models provide corroborative evidence for an unexpected role of collagen V homotrimer in hair cycling. Control of hair follicle cycling involves growth factors such as members of the transforming growth factor-b (TGF-b) (Stenn and Paus, 2001). Although we did not detect significant change in TGF-b1-3 expression in transgenic skin (Figure 1b), the possible role of collagen V in hair cycling deserves further investigation. We recently reported that TGF-b1 binds to the a1(V) N-propeptide (Symoens et al., 2010). The bioavailability of TGF-b in skin can be controlled by the partial N-propeptide cleavage occurring during pro-chain maturation (Bonod-Bidaud et al., 2007). Overexpression of the a1(V)3 in the perifollicular zone may disturb TGF-b delivery, which results in an increase of hair number. In conclusion, the generation and analysis of the K14COL5A1 mice has shed light on the role of a minor collagen V subtype in skin development and function. Our data demonstrated a specific role and distribution for the different collagen V subtypes in skin as illustrated in Figure 5d and has sorted out key elements of the native protein network occurring at the basement membrane zone. The K14-COL5A1 transgenic mouse line we have fully characterized in this study represents a good model to study the influence of collagen V in skin wound healing, a reported skin defect in EDS patients. MATERIALS AND METHODS Generation of transgenic mouse lines Transgenic mouse lines (K14-COL5A1) overexpressing the human proa1(V) chain in the epidermis under the control of the K14 promoter (Vassar et al., 1989) were created and characterized as described in Supplementary Materials online. Mice were kept according to French national requirements and all experimental procedures were approved by the Direction of the Veterinary Service of Rhone Department (DDSV, Lyon, France).

RT-PCR Total RNAs were extracted from 17.5 d.p.c. or 5-day-old mouse skin with the RNeasy kit according to the manufacturer’s instructions (Qiagen, Hilden, Germany). RT reactions were performed using 1 mg of RNA, with 1 mg of 500 mM hexameric oligonucleotides and 200 U of Moloney murine leukemia virus enzyme. Col1a1, Col6a1, Tenascin-X, and Tgfb1-3 were amplified using the primers (to be continued in Supplementary Materials online).

Immunodetection Whole protein extracts of 17.5 d.p.c. transgenic and wild-type skin embryos were analyzed by western blot to detect the transgenederived product using specific monoclonal antibodies (see Supplementary Materials online). For histology and immunochemistry, tissue samples of 5-day-old mice were fixed in 4% paraformaldehyde, embedded in paraffin, and processed as detailed in Supplementary data online.

Measurement of hair number Depilatory cream was applied on the skin of the back of the 5-day-old mice that were killed. After treatment for 20 minutes, hair www.jidonline.org 1847

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

was removed with a scraper. Thereafter, hair follicles were counted with a stereomicroscope at low magnification. Seven transgenic mice and four wild-type mice were analyzed, and four different fields were counted for each mouse.

Transmission and scanning electron microscopy and immunogold electron microscopy Routine TEM and scanning electron microscopy methods are detailed in Supplementary Materials online. For immunogold electron microscopy, skin from newborn and 7-day-old mice were sliced into small pieces and fixed in 4% paraformaldehyde and 0.25% glutaraldehyde in 100 mM sodium cacodylate buffer, pH 7.4, at 41C. The samples were rinsed in phosphate-buffered saline (PBS), dehydrated in an ascending series of ethanol from 30% to 70%, and embedded in the acrylic resin LR White (London Resin Company, London, UK). Ultrathin sections were cut with an ultramicrotome and collected on Formvar carbon-coated nickel grids for immunogold electron microscopy. Furthermore, ultrathin sections were first incubated with 100 mM glycine in PBS for 2 minutes, washed twice with PBS, and incubated with 2% BSA in PBS for 30 minutes. Thereafter, mAbs against the human proa1(V) chain (13F6) diluted 1:50 in 0.2% BSA were incubated for 2 hours. After washing with PBS, the sections were incubated with gold-conjugated antibodies against mouse immunoglobulins (12-nm gold particles). Finally, all sections were rinsed with water and stained with uranyl acetate for 8 minutes. Electron micrographs were taken at 80 kV with an EM 410 electron microscope (Philips, Amsterdam, The Netherlands).

Isolation of authentic suprastructures by superparamagnetic immunobeads Crude extracts from mouse skin containing suprastructural fragments obtained by mechanical disruption of tissue were used for isolation by superparamagnetic immunobeads (Invitrogen, Oslo, Norway; Kassner et al., 2003). The superparamagnetic polysterene beads covered with affinity-purified secondary antibodies (anti-rabbit immunoglobulins) were covalently coupled to primary antibodies (CO20511, BIOLOGO) against human collagen V as target to enhance the separation efficiency as previously described (Villone et al., 2008) (detailed in Supplementary Materials online).

(SFR Biosciences-Gerland, Lyon, France) for providing help and advice, Nadine Aguilera for maintaining transgenic strains, and Gerburg Hoelscher for excellent technical help. SPR arrays were performed at the core facility PAP (SFR Biosciences Gerland, Lyon, France), and we are grateful to C Marquette and L Blum for providing access to the spotter. This work was supported by funding to FR from the ‘‘GIS-maladies rares’’, the ‘‘Association pour la Recherche sur le Cancer’’, and in part by the Sonderforschungsbereich (SFB) project A2. SUPPLEMENTARY MATERIAL Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid

REFERENCES Adachi E, Hayashi T (1994) Anchoring of epithelia to underlying connective tissue: evidence of frayed ends of collagen fibrils directly merging with meshwork of lamina densa. J Electron Microsc (Tokyo) 43:264–71 Adachi E, Hopkinson I, Hayashi T (1997) Basement-membrane stromal relationships: interactions between collagen fibrils and the lamina densa. Int Rev Cytol 173:73–156 Andrikopoulos K, Suzuki HR, Solursh M et al. (1992) Localization of proalpha2(V) collagen transcripts in the tissues of the developing mouse embryo. Dev Dyn 195:113–20 Andrikopoulos K, Liu X, Keene DR et al. (1995) Targeted mutation in the Col5a2 gene reveals a regulatory role for type V collagen during matrix assembly. Nat Genet 9:31–6 Birk DE (2001) Type V collagen: heterotypic type I/V collagen interactions in the regulation of fibril assembly. Micron 32:223–37 Birk DE, Fitch JM, Linsenmayer TF (1986) Organization of collagen types I and V in the embryonic chicken cornea. Invest Ophthalmol Vis Sci 27:1470–7 Bonod-Bidaud C, Beraud M, Vaganay E et al. (2007) Enzymatic cleavage specificity of the proalpha1(V) chain processing analysed by site-directed mutagenesis. Biochem J 405:299–306 Bouma P, Cabral WA, Cole WG et al. (2001) COL5A1 exon14 splice acceptor mutation causes a functional null allele, haploinsufficiency of alpha1(V) and abnormal heterotypic interstitial fibrils in Ehlers–Danlos syndrome II. J Biol Chem 276:13356–64 Chanut-Delalande H, Fichard A, Bernocco S et al. (2001) Control of heterotypic fibril formation by collagen V is determined by chain stoichiometry. J Biol Chem 276:24352–9 Chanut-Delalande H, Bonod-Bidaud C, Cogne S et al. (2004) Development of a functional skin matrix requires deposition of collagen V heterotrimers. Mol Cell Biol 24:6049–57

Stress/strain tests and splitting tests were conducted as described in Supplementary Materials online.

Colige A, Ruggiero F, Vandenberghe I et al. (2005) Domains and maturation processes that regulates the activity of ADAMTS-2, a metalloproteinase cleaving the aminopropeptide of fibrillar procollagen type I, II, III and V. J Biol Chem 280:34397–408

Protein-binding assays

Faye C, Chautard E, Olsen BR et al. (2009) The first draft of the endostatin interaction network. J Biol Chem 284:22041–7

Biomechanical analysis

Protein-binding assays were performed using surface plasmon resonance in a Biacore Flexchip system (GE Healthcare, Uppsala, Sweden) using SPR arrays (Faye et al., 2009) or with solid-phase binding assays as previously described (Rousselle et al., 1997). Details are provided in Supplementary Materials online. CONFLICT OF INTEREST

Fichard A, Kleman JP, Ruggiero F (1995) Another look at collagen V and XI molecules. Matrix Biol 14:515–31 Fichard A, Tillet E, Delacoux F et al. (1997) Human recombinant alpha1(V) collagen chain. Homotrimeric assembly and subsequent processing. J Biol Chem 272:30083–7 Fitch JM, Gross J, Mayne R et al. (1984) Organization of collagen types I and V in the embryonic chicken cornea: monoclonal antibody studies. Proc Natl Acad Sci USA 81:2791–5

The authors state no conflict of interest.

Gay S, Martinez-Hernandez A, Rhodes RK et al. (1981) The collagenous exocytoskeleton of smooth muscle cells. Collagen Relat Res 1:377–84

ACKNOWLEDGMENTS

Giunta C, Steinmann B (2000) Compound heterozygosity for a disease causing G1489E [correction of G1489D] and disease-modifying G530S substitution in COL5A1 of a patient with the classical type of Ehlers–Danlos syndrome: an explanation of intrafamilial variability? Am J Med Genet 90:72–9

We are grateful to Elaine Fuchs for generously providing the K14 expression cassette, to David Birk for the gift of antibody against the proa2(V) chain, and to Lydia Sorokin for the gift of antibody against the laminin g-1 chain. We thank Denise Aubert and the transgenic mouse core facility PBES

1848 Journal of Investigative Dermatology (2012), Volume 132

C Bonod-Bidaud et al. Bridging Role of Collagen V Homotrimer in Skin

Haralson MA, Mitchell WM, Rhodes RK et al. (1984) Evidence that the collagen in the culture medium of Chinese hamster lung cells contains components related at the primary structural level to the alpha1(V) collagen chain. Arch Biochem Biophys 229:509–18

Roulet M, Ruggiero F, Karsenty G et al. (2007) A comprehensive study of the spatial and the temporal expression of the col5a1 gene in mouse embryos: a clue for understanding collagen V function in developing connective tissues. Cell Tissue Res 327:323–32

Kassner A, Hansen U, Miosge N et al. (2003) Discrete integration of collagen XVI into tissue-specific collagen fibrils or beaded microfibrils. Matrix Biol 22:131–43

Rousselle P, Keene DR, Ruggiero F et al. (1997) Laminin 5 binds the NC-1 domain of type VII collagen. J Cell Biol 138:719–28

Kirschner J, Hausser I, Zou Y et al. (2005) Ullrich congenital muscular dystrophy: connective tissue abnormalities in the skin support overlap with Ehlers–Danlos syndromes. Am J Med Genet 132:296–301 Malfait F, De Paepe A (2005) The molecular basis of classic Ehlers–Danlos syndrome: a comprehensive study of biochemical and molecular findings in 48 unrelated patients. Am J Med Genet C 139:17–23 Malfait F, Coucke P, Symoens S et al. (2005) The molecular basis of classic Ehlers–Danlos syndrome: a comprehensive study of biochemical and molecular findings in 48 unrelated patients. Hum Mutat 25: 28–37

Stenn KS, Paus R (2001) Controls of hair follicle cycling. Physiol Rev 81:449–94 Symoens S, Renard M, Bonod-Bidaud C et al. (2010) Identification of binding partners interacting with the a1-N-propeptide of type V collagen. Biochem J 433:371–81 Trost A, Bauer JW, Lanschu¨tzer C et al. (2007) Rapid, high-quality and epidermal-specific isolation of RNA from human skin. Exp Dermatol 16:185–90 Tsuruta D, Hashimoto T, Hamill KJ et al. (2011) Hemidesmosomes and focal contact proteins: functions and cross-talk in keratinocytes, bullous diseases and wound healing. J Dermatol Sci 62:1–7

Modesti A, Kalebic T, Scarpa S et al. (1984) Type V collagen in human amnion is a 12 nm fibrillar component of the pericellular interstitium. Eur J Cell Biol 35:246–55

Vassar R, Rosenberg M, Ross S et al. (1989) Tissue-specific and differentiationspecific expression of a human K14 keratin gene in transgenic mice. Proc Natl Acad Sci 86:1563–7

Moradi-Ameli M, Rousseau JC, Kleman JP et al. (1994) Diversity in the processing events at the N-terminus of type-V collagen. Eur J Biochem 221:987–95

Villone D, Fritsch A, Koch M et al. (2008) Supramolecular interactions in the dermo-epidermal junction zone: anchoring fibril–collagen VII tightly binds to banded collagen fibrils. J Biol Chem 283:24506–13

Ricard-Blum S, Ruggiero F (2005) The collagen superfamily: from the extracellular matrix to the cell membrane. Pathol Biol 53:430–42

Wenstrup RJ, Florer JB, Brunskill EW et al. (2004) Type V collagen controls the initiation of collagen fibril assembly. J Biol Chem 279:53331–7

www.jidonline.org 1849