Cysteine-rich Protein 61 (CCN1) Domain-specific Stimulation of Matrix Metalloproteinase-1 Expression through αVβ3 Integrin in Human Skin Fibroblasts*

Background: Elevated MMP-1 is a prominent feature of human skin aging. Results: The CCN1 VWC domain interacts with αVβ3 integrin and cooperates with adjacent IGFBP and TSP1 domains to stimulate MMP-1 expression. Conclusion: CCN1 structural domains contribute to human skin aging. Significance: Targeting the CCN1/αVβ3 integrin pathway may provide a therapeutic approach to improve the health of aged human skin. Human skin largely comprises collagenous extracellular matrix. The hallmark of skin aging is fragmentation of collagen fibrils. Matrix metalloproteinases (MMPs) are largely responsible for collagen degradation. MMP-1, principally derived from dermal fibroblasts, is the major protease capable of initiating degradation of native fibrillar collagens. Presently, we report that CCN1, a secreted and extracellular matrix-associated protein, is elevated in aged human skin dermal fibroblasts in vivo and stimulates MMP-1 expression through functional interaction with αVβ3 integrin in human dermal fibroblasts. CCN1 contains four conserved structural domains. Our results indicate that the three N-terminal domains (IGFBP, VWC, and TSP1), but not the C-terminal CT domain, are required for CCN1 to stimulate MMP-1 expression. This stimulation is dependent on interaction between the active structural domains and αVβ3 integrin. The interaction of VWC domain with integrin αVβ3 is necessary and requires functional cooperation with adjacent IGFBP and TSP1 domains to stimulate MMP-1 expression. Finally, induction of MMP-1 expression in dermal fibroblasts by CCN1 N-terminal domains resulted in fragmentation of type I collagen fibrils in a three-dimensional collagen lattice model. These data suggest that domain-specific interactions of CCN1 with αVβ3 integrin contribute to human skin aging by stimulating MMP-1-mediated collagen fibril fragmentation.

Human skin, like all human organs, undergoes progressive alterations as a consequence of natural aging. Additional damage is superimposed on natural aging as a result of chronic exposure to UV irradiation from sunlight. The bulk of skin comprises collagenous extracellular matrix, and the molecular hallmark of aged human skin is altered collagen fibril in dermal connective tissue (1,2). Age-related, progressive alterations of dermal connective tissue collagen correlate with a decline of skin functions including increased fragility and reduced wound healing and with increased cancer incidence (2)(3)(4)(5)(6).
In human skin, collagen turnover is carried out by matrix metalloproteinases (MMPs), 2 a family of zinc-containing proteinases that specifically degrade extracellular matrix including type I collagen fibrils. To date, 23 members of the MMP family are known in humans. 18 members are expressed in human skin. Of these, MMP-14 is most highly expressed (7). Although MMP-14 has been shown to degrade type I collagen (8 -10), its expression is not altered by stimuli, such as UV irradiation and reactive oxygen species, or wounding, which greatly increase collagen fibril fragmentation (7,11). Also, MMP-14 is not increased in aged or photoaged human skin, which have elevated levels of collagen defragmentation (12). 3 In contrast, MMP-1 (also known as collagenase 1) is produced principally in fibroblasts and appears to be the major protease capable of initiating degradation of native fibrillar collagens (13)(14)(15)(16)(17)(18). MMP-1 cleaves native type I collagen fibrils at a single site, resulting in fragments that are one fourth and three fourths of the total length of the native molecule. MMP-1 expression is very low in normal human skin but is elevated in aged human skin and is associated with fragmented and disorganized collagen fibrils in the dermis (2). However, the precise molecular mechanisms underlying elevated MMP-1 expression in aged human skin are not well understood.
CCN1 is a secreted, extracellular matrix-associated protein that belongs to the CCN (cysteine-rich protein 61, connective tissue growth factor, nephroblastoma overexpressed) family (19,20). CCN proteins play fundamental roles in growth, differentiation, angiogenesis, migration, and extracellular matrix regulation (19,21). We previously reported that CCN1 is expressed in human skin dermis and is substantially elevated in the dermis of naturally aged, photoaged, and acutely UV-irradiated human skin (22,23). In cell culture models, CCN1 recombinant protein regulates cell adhesion, cell migration, cell-matrix interactions, and the synthesis of extracellular matrix (24 -26). Furthermore, elevated CCN1 up-regulates MMP-1 protein expression in cultured primary human skin dermal fibroblasts (22,23). CCN1 (and other CCN family proteins) is composed of an N-terminal signal peptide followed by four conserved structural/functional domains (19,21). These domains share a high degree of sequence homology with (i) insulin-like growth factor-binding proteins (IGFBP), (ii) Von Willebrand factor type C repeat (VWC), (iii) thrombospondin type I repeat (TSP1), and (iv)) C-terminal cysteine knots (CT). Interestingly, these four distinct structural domains are separated by protease-sensitive residues, and each of these domains is encoded by a separate exon. The multimodular structural organization of CCN1 protein suggests that its diverse biological functions are programmed by the combinatorial actions of individual domains, either acting independently or interdependently (27)(28)(29). CCN1 exerts a range of functions by interacting with numerous integrins, in a cell type-and function-specific manner (27). In this study, we determined the role of CCN1 structural domains and their binding to specific integrins in the regulation of MMP-1 expression, in human skin fibroblasts.

Procurement of Human Skin Samples and Laser Capture
Microdissection (LCM)-Punch biopsies (4 mm) of sun-protected buttock human skin were obtained from young (25-30 years) and aged (Ն80 years) healthy volunteers as described previously (30,31). Human skin dermal fibroblasts were subsequently captured by LCM as described previously (7,30). Briefly, OCT-embedded human skin cryosections (15 m) were stained with hematoxylin and eosin, and ϳ200 dermal fibroblasts were captured by LCM (Leica ASLMD system; Leica Microsystems, Wetzlar, Germany). Total RNA was prepared from LCM-captured dermal fibroblasts using a commercial kit (RNeasy Micro kit, Qiagen). A TaqMan PreAmp Master Mix kit (Applied Biosystems) was used to preamplify cDNA for quantitative real-time PCRs. The quality and quantity of amplified cDNA were determined by Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA). Quantitation of transcript levels of several different genes in samples of total RNA and amplified cDNA yielded essentially identical results (data not shown). All procedures involving human subjects were approved by the University of Michigan Institutional Review Board, and all participating subjects provided written informed consent prior entering the study.
Immunohistology-Immunohistology was performed as described previously (18). Briefly, slides were fixed in 2% paraformaldehyde and permeabilized with 0.5% Triton X-100 in phosphate-buffered saline (PBS), blocked with rabbit serum (5% in PBS), and incubated for 1 h at room temperature with CCN1 primary antibody (Santa Cruz Biotechnology) followed by incubation of secondary antibody for 1 h at room tempera-ture. After staining, the slides were examined using a digital imaging microscope (Zeiss). Isotype-control immunoglobulin substituted for the CCN1 primary antibody yielded no detectable staining (data not shown).
RNA Isolation and Quantitative Real-time RT-PCR-Total RNA was extracted using TRIzol reagent (Invitrogen). cDNA for PCR templates was prepared by reverse transcription of total RNA (100 ng) using a Taqman Reverse Transcription kit (Applied Biosystems). Real-time PCR was performed on a 7300 Sequence Detector (Applied Biosystems) using Taqman Universal PCR Master Mix Reagents (Applied Biosystems). MMP-1, CCN1, and 36B4 primers and probes were purchased from Applied Biosystems. Target gene mRNA expression levels were normalized to the housekeeping gene 36B4 as an internal control for quantification.
Cell Culture-Primary human skin dermal fibroblasts were established from healthy adult human skin biopsy and cultured as described previously (32).
CCN1 Constructs and Transfection-CCN1 domain-specific or domain-specific deletion expression vectors were generated by PCR using domain-specific primers. PCR products were inserted into a pCDNA3.1 expression vector, and orientation and sequences were confirmed by DNA sequencing. These constructs (2 g) were transiently transfected into human skin dermal fibroblasts by electroporation (Amaxa Nucleofector, Koeln, Germany).
Western Blot Analysis, Immunoprecipitation, and Integrin Neutralization-Forty-eight h after transfection with CCN1 expression constructs, conditioned medium, cellular protein, and ECM were extracted for analysis. To prepare ECM, cells were removed by 5 mM EDTA for 1 h at 37°C. ECM was collected by scraping the tissue culture plates with a Cell Lifter Polyethyene (Costar; Fisher Scientific) in whole cell extraction buffer (25 mM HEPES (pH 7.7), 0.3 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.1% Triton X-100, 0.5 mM DTT, 20 mM ␤-glycerolphosphate, 0.1 mM Na 3 VO 4 , 2 g/ml leupeptin, and 100 g/ml PMSF). Protein concentrations of whole cell extract, ECM proteins, and conditioned medium were determined by Bradford assay (Bio-Rad Laboratories). Western blotting was performed as described previously (33). From each sample, total cellular protein (ϳ50 g/lane), ECM protein (ϳ50 g/lane), and 50 l of conditioned medium were subjected electrophoresis in a 4 -20% gradient SDS-polyacrylamide gel and transferred to a PVDF membrane (Millipore). MMP-1 and CCN1 antibodies were purchased from Santa Cruz Biotechnology. Protein bands were visualized and quantified with enhanced chemifluorescence (Vistra ECF Western blotting System; GE Healthcare) following the manufacturer's protocol. The intensities of each band were quantified by a STORM Molecular Imager (Molecular Dynamics) and normalized using ␤-actin as a loading control. The intensity of protein bands was quantified by ImageQuant software (an integral application of the STORM Molecular Imager system). For immunoprecipitation experiments, cellular and ECM extracts were incubated overnight at 4°C with HA antibody (Pierce) to capture HAtagged ␣V␤3 integrin. After immunoprecipitation, proteins were separated in a 4 -20% gradient SDS-polyacrylamide gel followed by Western blotting with V5 tag antibody (Invitrogen) to detect tagged CCN1 proteins. Integrin neutralization assays were performed using Integrin Antibody Investigation Kits (␣ Integrin Blocking Kit (ECM430) and ␤ Integrin Blocking Kit (ECM440); Chemicon) or ␣V␤3 integrin neutralization antibody (MAB1976B). Briefly, cells were transfected with control vector (pCDNA3.1) or relevant CCN1 expression vectors. After 24 h, cells were harvested and then incubated with neutralization antibodies (10 g/1 ϫ 10 6 cells) for 1 h at 37°C. The cells were then cultured for 2 days prior to analysis.
Three-dimensional Collagen Lattice Cell Culture-Primary human skin dermal fibroblasts were embedded in three-dimensional collagen lattices as described previously (17). Briefly, collagen lattices were prepared by mixing an appropriate volume of rat tail type I collagen (BD Biosciences) and a medium mixture (DMEM, NaHCO 3 (44 mmol/liter), l-glutamine (4 mmol/ liter), folic acid (9 mmol/liter), neutralized with 1 N NaOH to pH 7.2) at a concentration of 2 mg/ml. Two ml of the collagen/ cell (1 ϫ 10 5 /35-mm plate) suspension was added to each well. To assess collagen fragmentation, lattices were washed extensively with PBS (at least three times for 30 min), and then trypsin-EDTA (100 ng/ml; Invitrogen), in serum-free media, was added to activate MMP-1 secreted by the cells. After 2 days, conditioned media were collected, concentrated, and analyzed by 10% SDS-PAGE. Collagen bands were visualized by staining with SimplyBlue (Invitrogen). For the integrin neutralization assays, cells were first transfected with relevant CCN1 expression vectors. After 24-h transfection, cells were harvested and then incubated with ␣V␤3 integrin neutralization antibody for 1 h at 37°C. The cells were then cultured in three-dimensional collagen lattices for 2 days with ␣V␤3 integrin neutralization antibody prior to analysis.
Atomic Force Microscopy (AFM) Imaging-Collagen lattices were washed with PBS, mounted on a microscope glass slide (25 ϫ 75 ϫ 1.0 mm, Fisher Scientific), and allowed to air dry for at least 24 h before AFM image analysis. AFM images were obtained using a Dimension Icon AFM system (Bruker-AXS, Santa Barbara, CA) as described previously (34). AFM images were obtained with ScanAsyst mode in air using a siliconetched cantilever (NSC15/AIBS; MikroMasch, San Jose, CA) with a full tip cone angle ϳ40°and the tip radius of curvature ϳ10 nm. AFM images were acquired at a scan rate of 0.977 Hz, 512 ϫ 512-pixel resolution. AFM images were obtained from the Electron Microbeam Analysis Laboratory (EMAL) at the University of Michigan College of Engineering and were analyzed using Nanoscope Analysis software (Nanoscope Analysis v120R1sr3; Bruker-AXS).
Statistical Analysis-Data are expressed as mean Ϯ S.E. Student's t test was used to evaluate the statistical differences between the groups. All p values are two-tailed, and values Ͻ0.05 were considered statistically significant.

CCN1, a Positive Regulator of MMP-1 Expression, Is Elevated in Human Skin Dermal Fibroblasts in Vivo-We previously
reported that CCN1 is elevated in full thickness aged human skin which includes epidermis and dermis (22). However, expression levels of CCN1 in dermal fibroblasts, the major cells responsible for collagen homeostasis, in young and aged human skin are unknown. To address this question, we isolated dermal fibroblasts in young and aged human skin by LCM (Fig. 1A). Total RNA was prepared from LCM-captured dermal fibroblasts, and CCN1 mRNA levels were determined by quantitative real-time PCR. As shown in Fig. 1B, CCN1 mRNA in dermal fibroblasts in aged skin was elevated 3-fold compared with dermal fibroblasts in young skin. Consistent with this observation, CCN1 protein was markedly elevated in aged human skin dermis compared with young skin dermis (Fig. 1C). Importantly, MMP-1, which is induced by CCN1 in human dermal fibroblasts (22), was significantly increased in LCM-captured aged dermal fibroblasts, compared with dermal fibroblasts young skin (Fig. 1D). These data indicate that elevated CCN1 correlates with increased expression of MMP-1 in dermal fibroblasts in aged human skin.

Characterization of CCN1 Structural Domain(s) Involved in Stimulation of MMP-1 Expression in Human Dermal
Fibroblasts-Although CCN1 is elevated in aged human skin and CCN1 can up-regulate MMP-1 in human dermal fibroblasts (22), it is unknown which CCN1 domain(s) regulate MMP-1 expression. To dissect domain-specific function, we created expression vectors for individual CCN1 domains and domain-specific deletions, as shown in Fig. 2A. These constructs were transiently transfected into human dermal fibroblasts, and Western blotting was performed to determine expression levels in fibroblast cell extracts, culture media, and ECM extracts (Fig. 2B). CCN1 lacking signal peptide (NS-CCN1) was detected only in the cell extract. All other constructs were detected in both cell extract and culture medium, indicating that CCN1 domain-specific constructs were expressed and secreted into the media. Single domain constructs were undetectable in the ECM fraction, whereas all other constructs were detectable in the ECM extracts.
Importantly, only full-length CCN1 (FL-CCN1) and CCN1 in which the CT domain was deleted (CT-DEL) were able to up-regulate MMP-1 mRNA and protein expression (Fig. 2, C  and D). Deletion of the IGFBP, VWC, or TSP1 domain, either alone or in combination, abrogated the ability of CCN1 to induce MMP-1 expression. Interestingly, whereas the IGFBP, VWC, or TSP1 domain alone was unable to regulate MMP-1 expression (Fig. 2, C and D), combined expression of these three individual domains, which mimics expression of the CT domain deletion protein, resulted in stimulation of MMP-1 expression (Fig. 2E). This effect required expression of all three structural domains. Combined expression of any two domains had no effect on MMP-1 expression (Fig. 2E). Collectively, these data demonstrate that stimulation of MMP-1 expression by CCN1 requires IGFBP, VWC, and TSP1 structural domains, but not the CT domain.
We next performed immunoprecipitation assays to explore the mechanism by which ␣V␤3 integrin mediates CCN1 regulation of MMP-1 expression, HA-tagged ␣V and ␤3 integrin proteins were co-expressed with V5-tagged full-length, CT domain deletion or single domain proteins, in human dermal fibroblasts. As shown in Fig. 3C, both full-length (FL-CCN1) and the CT domain deletion proteins (CT-DEL) co-immunoprecipitated with ␣V␤3 integrin. The VWC, but not other single domain proteins, also interacted with ␣V␤3 (Fig. 3D). Interestingly, whereas the VWC single domain is able to interact with ␣V␤3 integrin (Fig. 3D), the VWC domain alone was unable to stimulate MMP-1 expression (Fig. 2, C-E). CCN1 lacking secretion signal peptide (NS-CCN1) did not interact with ␣V␤3 integrin (Fig. 3C).
Functional analysis indicated that stimulation of MMP-1 expression by full-length CCN1 (FL-CCN1) and CT domain deletion proteins (CT-DEL) was significantly reduced by ␣V␤3 integrin blocking antibody (Fig. 3E). These data indicate that CCN1 proteins containing IGFBP, VWC, and TSP1 domains interact with ␣V␤3 integrin and that this interaction mediates stimulation of MMP-1 expression in human dermal fibroblasts.
CCN1 Domain-specific Fragmentation and Lattice Contraction of Three-dimensional Collagen Lattices-Finally, we examined the functional consequences of expression of full-length CCN1 (FL-CCN1, CT domain deletion CCN1 (CT-DEL), positive control) and nonsecreted CCN1 lacking secretion signal peptide (NS-CCN1, negative control) by human dermal fibroblasts in three-dimensional collagen lattices. As shown in Fig. 4, expression of FL-CCN1 and CT-DEL but not NS-CCN1 or control vector, resulted in collagen gel contraction (Fig. 4A). This contraction was blocked by ␣V␤3 integrin neutralizing antibody (Fig. 4B). Collagen gel contraction in response to expression of FL-CCN1 or CT-DEL, was accompanied by increased levels of MMP-1-generated collagen fragments (Fig. 4C). This collagen fragmentation was blocked by neutralizing ␣V␤3 integrin antibody (Fig. 4D). AFM images show that collagen fibrils in lattices with control vector (Fig. 4E) or expression of NS-CCN1 (Fig. 4F) were intact and well organized. In contrast, collagen fibrils were fragmented and disorganized in lattices with expression of FL-CCN1 (Fig. 4G) and CT-DEL (Fig. 4H). This collagen fibril disruption was blocked by ␣V␤3 neutralizing antibody (Fig. 4, I and J). These data suggest that the CT domain deletion CCN1 is functionally equivalent to full-length CCN1 and that collagen lattice contraction, and collagen fibril fragmentation, and disorganization are associated with CCN1 stimulation of MMP-1. Finally, we show that CCN1 activates focal adhesion kinase (FAK) and paxillin, two key effectors of integrin signaling, and the downstream target MAPK/ERK (Fig.  4K), which is a major driving force for stimulation of MMP-1 expression (35).

DISCUSSION
CCN1 proteins are composed of an N-terminal secretion signal peptide followed by four conserved structural domains (19,21,27). Our data indicate that the secretion signal and the first three domains (IGFBP, VWC, and TSP1) are required for CCN1 stimulation of MMP-1 expression, in human dermal fibroblasts. The C-terminal CT domain is dispensable for this function. Interestingly, CCN family member CCN5 contains IGFBP, VWC, and TSP1 domains, but naturally lacks the CT domain (36,37).
Absence of the CCN1 CT domain may have distinct consequences for different functional roles. It has been reported that the CT domain is required for cell adhesion, but its deletion has no effect on stimulation of growth factor-induced proliferation (28). Presently, we find that CCN1 lacking the CT domain is equivalent to full-length wild type CCN1 in its ability to bind to

. Characterization of CCN1 structural domains involved in MMP-1 expression in human dermal fibroblasts.
A, diagram of CCN1 domain expression constructs that were used to determine domain-specific regulation of MMP-1 expression is shown. SP, signal peptide. B, CCN1 domain constructs, as described in A, were transiently transfected into human dermal fibroblasts, and protein expressed by each construct in cells, culture media, and ECM was determined by Western blotting. Black arrows indicate CCN1-specific bands. White arrows indicate undetected corresponding CCN1-specific bands in culture media and ECM fraction. ␤-Actin was used as loading control, and data are representative of three independent experiments. C and D, CCN1 domain constructs, as described in A, were transiently transfected into human dermal fibroblasts, and MMP-1 mRNA (C) and protein (D) levels were quantified by real-time PCR and Western blotting, respectively. MMP-1 mRNA levels were normalized to mRNA levels of the housekeeping gene, 36B4 (internal control). ␤-Actin was used as Western blot loading control. Inset shows representative Western blot. n ϭ 4; *, p Ͻ 0.05. E, combined expression of IGFBP, VWC, and TSP1 CCN1 domains, but not any combination of two domains, up-regulates MMP-1 expression. Indicated CCN1 domain constructs, as described in A, were transiently transfected into human dermal fibroblasts, and MMP-1 protein levels were determined by Western blotting and quantified by STORM Molecular Imager. ␤-Actin was used as loading control. Inset shows representative Western blot. n ϭ 3; *, p Ͻ 0.05.
␣V␤3 integrin and stimulate MMP-1 expression in human dermal fibroblast; indicating that the CT domain is nonessential for these activities. Our data demonstrate that secretion is vital for CCN1 function because signal peptide-deleted, nonsecreted CCN1 had no effect on MMP-1 expression.
Our data further suggest that upon secretion, CCN1 interacts with ECM and this ECM-CCN1 interaction may be important in CCN1 stimulation of MMP-1. The modular structural organization of CCN1 protein suggests that its diverse biological functions may result from the combined actions of individual domains, either acting independently or interdependently (38). A remarkable finding of our study is that each CCN1 domain is independently stable; however, the individual domains are unable to regulate MMP-1 expression. We found that co-expression of IGFBP, VWC, and TSP1 domains, but not any combination of two individual domains, resulted in stimulation of MMP-1 expression. These data indicate that these three domains cooperate with each other to stimulate MMP-1 in human dermal fibroblasts.
CCN1 protein exerts a range of diverse functions by interacting with multiple integrins (27,39). CCN1 interacts with distinct integrins in a cell type-and function-specific manner to mediate disparate biologic activities. Because CCN1 can stimulate MMP-1 in human dermal fibroblasts, an important question is the identity of a cell surface CCN1-binding integrin that mediates CCN1 stimulation of MMP-1.
Screening of integrins by antibody neutralization revealed that ␣V␤3 integrin acts as a cell surface receptor for CCN1; CCN1 physically interacts with ␣V␤3 integrin to mediate MMP-1 induction. ␣V␤3 integrin is implicated in the patho- integrin. V5-tagged FL-CCN1, CT-DEL CCN1, or CCN1 lacking secretion signal peptide (NS-CCN1) was co-expressed with HA-tagged ␣V␤3 integrin in human dermal fibroblasts. Cell extracts were prepared, and ␣V␤3 integrin chains were immunoprecipitated (IP) with HA antibody, and the immunoprecipitated proteins were analyzed by Western blotting with V5 antibody. Data are representative of three independent experiments. D, VWC domain, but not other individual domains, interacts with ␣V␤3 integrin. V5-tagged single domains were co-expressed with HA-tagged ␣V␤3 integrin in human dermal fibroblasts. Cell extracts were prepared, and ␣V␤3 integrin chains were immunoprecipitated with HA antibody, and the immunoprecipitated proteins were analyzed by Western blotting with V5 antibody. Data are representative of three independent experiments. E, CCN1 stimulation of MMP-1 expression is reduced by function blocking antibody to ␣V␤3 integrin. ␤-Actin was used as loading control. MMP-1 expression was determined by Western blotting and quantified by STORM Molecular Imager. Inset shows representative Western blot. n ϭ 3; *, p Ͻ 0.05. physiology of wound healing, angiogenesis, and tumor metastasis (40,41). Our data provide evidence that ␣V␤3 integrin contributes to human skin connective tissue aging by stimulating MMP-1 expression through functional interaction with CCN1. It has been reported that CCN1 interacts with ␣V␤3 integrin through VWC domain (27,28). We also observed that the VWC domain is able to interact with ␣V␤3 integrin (Fig.  3D). However, we found that the VWC domain alone is unable to stimulate MMP-1 expression (Fig. 2, C and D). Interestingly, combined expression of the VWC domain with domains adjacent to VWC can stimulate MMP-1 expression (Fig. 2E). These data indicate that the interaction of the VWC domain with ␣V␤3 integrin requires functional cooperation with adjacent IGFBP and TSP1 domains. It appears that although the interaction of VWC domain with ␣V␤3 integrin is necessary, this interaction alone is not sufficient to stimulate MMP-1 expres-sion. It is possible that the adjacent IGFBP and TSP1domains may interact with other integrins, which may act together with ␣V␤3 integrin to mediate CCN1 stimulation of MMP-1. The nature of the possible complex interactions between CCN1 domains and integrins remains to be determined.
CCN1 binding to the ECM may alter fibroblasts interactions with the ECM and with associated integrins. ECM-bound CCN1 may function as a "docking" protein that coordinates interactions between ECM proteins and cell surface integrins, ultimately orchestrating the cell response to the ECM microenvironment. Consistent with this notion, we found that CCN1 activates integrin outside-in signaling resulting in activation of FAK, a key effector of the integrin signaling pathway, and its downstream target paxillin, in human dermal fibroblasts (Fig.  4K). This finding suggests that signaling downstream of CCN1/ ␣V␤3 integrin may drive MMP-1 expression through a FAK- . Expression of full-length CCN1 or CCN1 lacking the CT domain induces contraction, fragmentation, and disorganization of three-dimensional collagen lattices. Human dermal fibroblasts were transfected with the indicated constructs and cultured in three-dimensional collagen lattices. For the neutralization assays, CCN1-transfected cells were treated with ␣V␤3 integrin antibody and cultured in three-dimensional collagen lattices, as indicated under "Experimental Procedures." A, full-length CCN1 (FL-CCN1) or CCN1 lacking the CT domain (CT-DEL) induces collagen gel contraction, compared with control vector (Ctrl) or CCN1 lacking the secretion signal peptide (NS-CCN1). B, CCN1-induced collagen gel contraction is inhibited by function blocking antibody to ␣V␤3 integrin. C, FL-CCN1 or CT-DEL induces collagen fragmentation, compared with control vector (Ctrl) and NS-CCN1. D, CCN1-induced collagen fragmentation is inhibited by function-blocking antibody to ␣V␤3 integrin. Conditioned media were collected, concentrated, and resolved in a 10% SDS-polyacrylamide gel. Collagen bands were visualized by staining with SimplyBlue. E-J, representative images from AFM of three-dimensional collagen lattices are shown. E, control; F, NS-CCN1; G, FL-CCN1; H, CT-DEL; I, FL-CCN1 with function-blocking antibody to ␣V␤3 integrin; J, CT-DEL with function-blocking antibody to ␣V␤3 integrin. The white and red arrows indicate intact and fragmented collagen fibrils, respectively. All data (A-J) are representative of at least three independent experiments. Scale bars, 200 m. K, CCN1 activates integrin downstream targets FAK, paxillin, and MAPK/ERK. Human skin dermal fibroblasts were transfected with CCN1 expression vector. Cell extracts were prepared, and activation of FAK, paxillin, and MAPK/ERK, and ␤-actin protein levels were determined 2 days after transfection by Western blotting. Numbers indicate -fold change compared with control (CTRL). n ϭ 3. dependent pathway. In addition, we found that CCN1 also activates ERK1/ERK2 (Fig. 4K) and AP-1 transcription factor (22), the major regulators of MMP-1 expression (35). It is well documented that integrin signaling activates the MAPK/ERK/ AP-1 pathway (42,43). These data suggest that CCN1/␣V␤3 integrin signaling may stimulate MMP-1 through FAK and MAPK pathways.
Dermal fibroblasts reside in collagen-rich ECM microenvironment in human skin. We provide evidence that CCN1 structural domains (IGFBP, VWC, and TSP1) can alter collagen ECM environment. CCN1 can stimulate degradation and disorganizations of type I collagen fibrils, the major structural protein in skin dermis, in three-dimensional collagen lattices. Furthermore, CCN1 can alter collagen gel contraction which is a visible readout of changes in ECM integrity and cell-matrix interactions. These data further suggest that elevated CCN1 in aged dermal fibroblasts promotes type I collagen fragmentation and altered dermal structural integrity, the most prominent feature of aged human skin.
In summary, we propose a working model in which CCN1 contributes to skin aging by stimulating MMP-1 expression through domain-specific interactions with ␣V␤3 integrin (Fig.  5). CCN1 is elevated in aged human dermal fibroblasts in vivo, the major cells responsible for collagen homeostasis in skin. Upon secretion, CCN1 binds to ECM and through its IGFBP, VWC, and TSP1 domains interacts with ␣V␤3 integrin. Interaction of CCN1 domains with ␣V␤3 integrin triggers FAK and MAPK/AP-1 pathways and thus stimulates MMP1 expression. CCN1-associated up-regulation of MMP-1 causes dermal col-lagen fragmentation and thus contributes to skin aging. Our data suggest that targeting the CCN1/␣V␤3 integrin pathway may be an effective therapeutic strategy to reduce collagen loss and thereby lessen the deleterious impact of aging on the health of human skin.