Abstract
Expression of SPARC (
T
SPARC (
Increases in the expression of SPARC are often observed in tissues undergoing remodeling. Induction of SPARC is associated with tumor metastasis, angiogenesis, fibrosis, and bone metabolism (Termine et al. 1981; Iruela-Arispe et al. 1995; Kuhn and Mason 1995; Ledda et al. 1997). Increased amounts of SPARC protein and mRNA are also observed during dermal wound healing. Although SPARC is expressed in uninjured skin and increases in SPARC expression are observed at early time points after injury, the highest levels occur at later times during remodeling of the ECM (Reed et al. 1993; Hunzelman et al. 1998). Given the capacity of SPARC to affect a number of processes involved in wound repair, we sought to determine whether healing was altered in the absence of SPARC.
For these experiments we used mice with a targeted deletion of the SPARC gene. The phenotype of SPARC-null mice is ostensibly normal except that the mice display early-onset cataractogenesis and progressively severe osteopenia (Gilmore et al. 1998; Norose et al. 1998; Delany et al. 2000). To this interesting phenotype we now add our recent observations that a significant increase in the rate of wound closure occurs in the absence of SPARC. Identification of proteins, and their mechanism of action, that participate in response to injury provides opportunities for solving the complex pathways that govern dermal wound repair and for novel treatment of wound pathologies.
Materials and Methods
Animals
C57BL/6 × 129SVJ SPARC-null mice were generated as described in Norose et al. (1998). SPARC-null and wild-type mice were housed in a modified pathogen-free facility before and throughout the course of the wound-healing studies. The genotypes of the mouse colony were monitored by polymerase chain reaction (PCR) using genome-specific primers on purified tail DNA. Two wounds per animal were generated with a 5-mm Acu-punch biopsy tool (Acuderm; Ft Lauderdale, FL) in the dorsa of anesthetized mice, as described in Reed et al. (1996). Animals were housed individually and were monitored for 4–6 hr after wounding to ensure that premature closure of the wound edges did not occur. We did not observe significant differences in the size of the wounds at time 0. Wounds were left undressed. Three age-matched animals per time point from each genotype (SPARC-null and wild-type) were used in each experiment. Eight separate wound-healing experiments were carried out with animals of different ages and wound sizes, with similar results. A trend towards accelerated healing in the SPARC-null vs the wild-type mice was consistently observed, and earlier experiments were used as guides for selection of optimal time points in later trials. Wounds were measured by a micrometer at designated times after injury. Two measurements were taken for each wound, one representing the horizontal axis of the ellipse and the other representing the vertical axis. The area of the ellipse was calculated for each wound. At each time point, the wounds from three animals of each genotype were excised and cut into two equal pieces at the mid-section of the wound, fixed in formalin, and embedded in paraffin for sectioning. Tissue sections were subsequently stained with hematoxylin and eosin (H and E) or picrosirius red to visualize cell bodies and ECM.
In Situ Hybridization
In situ hybridization (ISH) analysis was carried out as described in Reed et al. (1996). Briefly, riboprobes were synthesized with an RNA transcription kit (Promega, Madison, WI) in the presence of α-[35S]dUTP (Amersham; Arlington Heights, IL). The template was a 312-bp HindIII/EcoR1 fragment from the 3 UTR of mouse α1(I) collagen in pBSK (provided by Dr. Paul Bornstein, University of Washington) (Slack et al. 1991). The specific activity of the probes was ~3–4 × 107 cpm/μg. A riboprobe transcribed in the sense orientation was used as a control. The method of ISH was a modified protocol described by Holland et al. (1987). The slides were photographed with a Leitz DMRB photomicroscope equipped for darkfield transillumination.
Immunohistochemistry
Sections of wounds from wild-type and SPARC-null mice were deparaffinized and either subjected to antigen retrieval with a 20-min incubation in steaming citrate buffer (0.1 M), as recommended for the Ki-67 antibody (Dako; Carpinteria, CA), or not subjected to antigen retrieval in the case of the anti-proliferating cell nuclear antigen (PCNA) antibody (biotin-conjugated IgG; Zymed, South San Francisco, CA). Sections were blocked in 2% normal goat serum and were incubated in primary antibody for 1 hr. The Ki-67 antigen was detected with a secondary antibody conjugated to biotin, and the signals for both Ki-67 and PCNA were amplified by incubation with the ABC Elite kit (Vector Labs; Burlingame, CA). Horseradish peroxidase activity associated with the primary antibodies was visualized with 3,3′-diaminobenzidine (DAB) substrate with or without nickel enhancement. Slides were counterstained with methylene green (Ted Pella; Redding, CA). Stained sections were viewed on a Leica DMR microscope and images were captured either by Kodachrome 400 film or by digital capture with an RT spot camera (Diagnostics Instruments; Sterling Heights, MI) linked to a Macintosh G4 computer (Apple; Cupertino, CA).
Primary Cell Isolation, Migration Assays, and Collagen Gel Contraction Assays
Dermal cells were isolated from the skin of wild-type and SPARC-null animals, as described in Bradshaw et al. (1999). Briefly, pieces of shaved skin were incubated in 0.25% trypsin (Sigma; St Louis, MO) in Solution A (10 mM glucose, 3 mM KCl, 130 mM NaCl, 1 mM Na2HPO4/·7H2O, 30 mM Hepes, pH 7.4) overnight at 4C before separation of the epidermis from the dermis. The dermal pieces were incubated in 0.25% collagenase (Worthington Enzymes; Freehold, NJ) at 37C for 4–6 hr. The digested dermis was triturated and plated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (Gibco-BRL; Gaithersburg, MD), 500 U/ml penicillin G, 500 U/ml streptomycin sulfate, and 2.5 g/ml fungizone (growth medium). The majority of cells in these primary cultures were dermal fibroblasts. However, other cell types were also represented. Endothelial cells, smooth muscle cells, and keratinocytes were detected with lineage-specific primers by reverse transcriptase PCR (RT-PCR; DC Graves, unpublished experiments). Significant differences in the proportion of particular cell types in either wild-type or SPARC-null cultures were not detected. The levels of contaminating cell types were estimated to be less than 10% in both wild-type and SPARC-null cultures. All experiments were performed with cells at early passage (P1-P4), before cellular senescence.
For migration experiments, equal numbers of wild-type and SPARC-null dermal cells (as quantified by a hemacytometer) were plated in 6-well plates in growth medium and were grown to confluence. A rubber spatula was used to remove a defined area of cells and thereby create a wound in the monolayer. Uniform wounds were made in wild-type and SPARC-null culture wells and were photographed at time 0. An area of the well was designated by marks on the dish such that the same field could be followed over time. The degree of migration was monitored by photography of the same field at 24-hr intervals. Each experiment was composed of triplicate wells for both wild-type and SPARC-null cultures. Scanned images were imported into an NIH image software program and were quantified as the percent of area occupied by cells vs total area included within the image. Duplicate wells were plated in parallel to monitor proliferation of the cultured cells. Cells were removed by trypsin digestion and were quantified by a hemacytometer at 24, 48, and 72 hr, coinciding with the time of photography of the migration assays. Higher-magnification images of the cell migration front were taken to monitor the number of mitotic cells in the wound area of the dish.
Collagen gel contraction assays were carried out as described in Vernon and Sage (1996). Briefly, 50,000 primary dermal cells isolated from wild-type animals were resuspended in growth medium containing increasing concentrations of rat tail collagen (BD Biosuppliers; Bedford, MA). The cells in collagen solution were placed in 24-well plates pre-coated with 1% agarose to prevent adhesion of the collagen matrix to the sides of the well. The plates were placed at 37C to allow polymerization of the collagen gel and subsequent cell contraction. Gel contraction was analyzed 24 hr after initial cell seeding by measurement of the diameter of the collagen gel.
Hydroxyproline Analysis
Measurement of hydroxyproline content of SPARC-null and wild-type skin was carried out as described by Woessner (1961). Briefly, dorsal skin samples from age-matched wild-type and SPARC-null mice were taken with an 8-mm punch biopsy tool. The samples were placed in 6 N HCl in sealed tubes and were heated at 120C for 48 hr. The samples were removed to a 90–110C oven with loosened caps and were heated until dry. At room temperature (RT), 1 ml of 0.25 M sodium phosphate buffer (pH 6.5) was added to each sample and the solution was vortexed. Fifty μl of chloramine T (Sigma) was added to each sample, which was incubated at RT for 20 min, followed by the addition of 50 μl p-dimethyl-aminobenzaldehyde (Sigma) and a further incubation at 60C for 30 min. The absorbance of the samples was read at 540 nm. The concentration of hydroxyproline was determined with the Molecular Devices Softmax program. A standard curve was established with cis-4-hydroxy-L-proline (Sigma).
Results
Wound Measurements and Histological Analysis
Cutaneous wound healing relies on cell migration, proliferation, and ECM remodeling to repair the injury, three processes that are influenced by the expression of SPARC (Lane and Sage 1994; Singer and Clark 1999). We sought to determine whether excisional wound healing was affected by the absence of SPARC. Uniform 5-mm wounds were made with a sterile punch biopsy tool in the dorsum of age-matched wild-type and SPARC-null mice. A significant increase in the rate of closure was observed in SPARC-null vs wild-type animals. Shown in Figure 1 are representative sections of wounds from one experiment with wild-type (Figures 1A, 1C, and 1E) and SPARC-null mice (Figures 1B, 1D, and 1F) stained with H and E. The hyperproliferative epithelium (arrows) at the leading edge of the wound were in closer apposition in SPARC-null mice by both Day 4 (compare Figures 1A and 1B) and Day 7 (Figures 1C and 1D) after wounding. By Day 11 (shown at higher magnification in Figures 1E and 1F), five of six SPARC-null wounds showed no visible scab or opening, whereas only three of six wild-type wounds exhibited complete closure.
The rates of wound closure were monitored by measurement of the wounds at Days 4, 7, and 11 with a micrometer (Figure 2). On Day 4 the average size of the wounds in SPARC-null mice was 4.46 mm2, whereas the average of the wild-type wounds was 6.32 mm2. The differences observed on Day 4 were more pronounced on Day 7. SPARC-null wounds averaged 2.23 mm2 compared to wild-type wounds, which averaged 4.96 mm2. Measurements on Day 11 did not yield significant numbers because only one wound of six was unclosed in SPARC-null animals (three of six were unclosed and therefore measurable in wild-type animals). A significant decrease in the size of the wounds in the SPARC-null mice was observed. These data indicate an increased capacity of the null mice to close excisional wounds. In addition, the rate of closure observed at early time points implicated enhanced
Wild-type (
wound contraction in SPARC-null vs wild-type mice. Wound contraction is a significant component of dermal excisional wound healing in mice and in humans (Singer and Clark 1999). Because hair follicles within the biopsy were removed upon wounding, the closest follicle on either edge of the wound provided a marker by which to monitor contraction. At Day 6–7, wounds from SPARC-null mice exhibited an average follicle distance of 1.7 mm (±0.5), whereas follicles flanking wild-type wounds averaged a distance of 3.0 mm (±0.8; p<0.05). The difference in average follicle spacing was also observed in Day 11–12 wounds (wild-type 0.96, ±0.015; SPARC-null 0.52, ±0.2; p<0.005). Therefore, accelerated closure of SPARC-null vs wild-type wounds appeared to be based, at least in part, on increased contraction.
The experiments shown are representative of eight separate wound-healing experiments performed in our laboratory. Although the parameters of these studies differed (the age of the mice and the size of the wound), the results were consistent. In every case, an increase in the rate of closure was observed in SPARC-null mice over that of appropriate wild-type control mice at a minimum of one time point. The trend of accelerated wound repair in the absence of SPARC has been reproducible in every experiment. Figures 1–4 represent one experiment performed with 5-mm wounds generated in 6-month-old mice.
To determine the degree to which the absence of SPARC might affect collagen I expression in response to injury, we performed ISH analysis of wounds from wild-type and SPARC-null animals. A representative wild-type (Figure 3A) and SPARC-null wound (Figure 3B) at Day 5 is shown; arrows indicate the margins of the wound bed. We did not observe significant differences in the level or distribution of α1(I) collagen mRNA expression between wild-type and SPARC-null wounds. In addition, picrosirius red staining of Day
Five-mm excisional wounds were made in the dorsum of SPARC-null and wild-type mice. Wounds were measured by a micrometer at designated times after injury and the area of the wound was determined according to the procedure described in Materials and Methods. A total of 18 wounds (9 animals) contributed to the Day 4 time point, whereas 12 wounds were measured on Day 7. Error bars represent the SEM. KO, SPARC knockout (= null) mice.
11 wounds showed no substantial differences in collagen deposition at the wound site of SPARC-null compared to wild-type mice (Figures 3C and 3D). Collagen I was also assessed by immunohistochemistry with antibodies against the N-terminal propeptide of Type I procollagen to differentiate newly synthesized from previously-deposited collagen, as well as by immunoblotting analysis (data not shown). Although some variability among wounds was observed, the overall extent of fibrillar collagen deposition in the SPARC-null wounds did not appear to be significantly different from that in wild-type wounds. By immunohistochemistry there were also no gross differences in Type IV collagen and fibronectin between wild-type and SPARC-null wounds (data not shown). However, the collagen content of unwounded skin did appear to be decreased in SPARC-null vs wild-type skin (see below).
Proliferation in SPARC-null vs Wild-type Wounds
SPARC interferes with the cell-cycle progression induced by a variety of growth factors in vitro (Motamed 1999). We therefore asked whether an increase in cell proliferation in vivo contributed to accelerated wound repair in SPARC-null mice. Sections of wild-type and SPARC-null wounds were stained with an anti-PCNA biotinylated primary antibody, because expression of PCNA has been shown to mark mitotically active cells in tissue sections (Wasseem and Lane 1990). The percentage of PCNA-positive cells did not appear to be significantly different in wild-type vs SPARC-null mice at the time points examined. For example, on Day 4 the percentage of PCNA-positive cells in SPARC-null wounds was 9.83%, whereas 10.54% of the cells were PCNA-positive in wild-type wounds (Table 1). Similar results were obtained with Ki-67 as a marker of cell proliferation (Endl and Gerdes 2000). We conclude that an increase in the percentage of proliferating cells in vivo probably does not account for the accelerated rate of wound closure observed in SPARC-null animals on Days 4 and 7 (Figures 1 and 2). However, we cannot rule out the possibility that there are early increases in cell proliferation before Day 4 that might contribute to the overall increase in the rate of wound closure consistently observed in SPARC-null mice (see Discussion).
Migration of Dermal Fibroblasts In Vitro
Because cell migration is another critical component of cutaneous wound repair, we asked whether there were differences in migration rates between primary wild-type and SPARC-null dermal cells. Wild-type and SPARC-null cells were plated at equal densities and were allowed to reach confluence. A wound of uniform size was made across the monolayer with a soft rubber spatula to avoid excessive injury to the remaining cells, and a designated field was photographed at time 0. The same field was photographed at 24 and 48 hr after wounding. The images were scanned for quantification of cell invasion, as shown in Table 2. The experiment was repeated four separate times with primary isolates from three different preparations, with similar results. The SPARC-null dermal cells consistently showed an enhanced capacity to invade the abraded area on the dish, compared to wild-type cells. To assess whether differences in the rates of proliferation affected invasion, we plated parallel cultures to monitor cell division in two separate experiments. At each migration time point, the cells were detached with trypsin and were counted by a hemacytometer. It was anticipated that SPARC-null cells would divide more rapidly in vitro relative to wild-type cells (Bradshaw et al. 1999). However, the increased proliferation of the SPARC-null vs wild-type cells was always significantly less than the increased rate of repopulation of the wound area exhibited by the SPARC-null cells (e.g., in Experiment 2 an 18% increase in proliferation of SPARC-null cells over that of wild-type was observed, whereas a 37% increase in migration of SPARC-null cells vs wild-type cells was seen over the 24-hr time period). In addition, the migration assays were monitored for the presence of mitotic cells at and around the migration front (wound edge). Although both wild-type and SPARC-null cultures contained proliferating cells, a significant difference in the percentage
Representative wounds from Day 4 wild-type (
of proliferating cells at the wound edge was not observed between the two cultures.
Collagen Composition of SPARC-null Skin
Although differences in the migration of SPARC-null vs wild-type cells were apparent, the marked rate of wound closure observed at early time points in the SPARC-null mice was suggestive of differences in wound contraction. In addition, preliminary observations of SPARC-null skin implied differences in the collagenous ECM of the dermis compared to that of wild-type. For example, SPARC-null skin appeared to be more prone to tearing than wild-type skin. In addition, collagenase treatments performed during primary dermal fibroblast isolation (see Materials and Methods) revealed an increased sensitivity of the SPARC-null dermis to digestion. Whereas wild-type dermis typically required 4–5 hr for complete dissolution of the tissue by collagenase, parallel samples of SPARC-null dermis required 2–3 hr under identical conditions (unpublished experiments).
Percentage of proliferating cells (± SD) a
aAt least four fields from no less than three wounds per time point stained with anti-PCNA antibodies were quantified to determine the percentage of proliferating cells in wounds from wild-type or SPARC-null mice (~1500–3000 total cells per age and genotype were counted). p values were not significant for either time point.
To quantify potential differences in collagen composition, we performed hydroxyproline analyses on SPARC-null and wild-type skins. As shown in Figure 4, SPARC-null skin showed an average hydroxyproline content of 8.7 μg/mg tissue, whereas wild-type skin yielded a value of 17.6 μg/mg tissue. The level of hydroxyprolyl post-translational modification of collagen chains did not appear to be significantly different between SPARC-null and wild-type mice, as assessed by SDS-PAGE analysis of collagen extracted from skin and tails (data not shown). We conclude that the skin of SPARC-null mice has a substantially reduced collagen concentration in comparison to that of wild-type mice. Because wound healing in mice results in part from enhanced contraction of the dermal matrix, the accelerated rate of closure observed in SPARC-null animals could be a consequence of the decreased content of dermal collagen relative to that of wild-type mice, or due to an increased contractile capacity of dermal fibroblasts in the absence of SPARC.
Quantification of migration for wounding assays performed In Vitro a
aPercent of cell occupancy was determined with NIH image on scanned fields from each time point. Quantification of two representative experiments using two separate preparations of primary isolates is shown. Two separate wells were quantified for Experiment 1 for each genotype (wild-type #1 and #2; SPARC-null #1 and #2), whereas three separate wells were quantified in Experiment 2 (wild-type #1–3 and SPARC-null #1–3). Four separate migration assays with three different primary isolates were performed with similar results. Some variability between primary isolates was encountered. However, an increase in the rate of migration of SPARC-null cultures over that of wild-type was consistently observed. Migration at 0 hr is defined as 0%. p<0.05 for both time points in Experiment 1 and for the 24-hr time point in Experiment 2.
The hydroxyproline content of wild-type (white bar) and SPARC-null skin (black bar) was expressed as μg hydroxyproline/mg dry tissue (see Materials and Methods). Skin removed from 12 separate age-matched animals of each genotype contributed to the determination. Error bars represent SEM. p<0.002.
Collagen Gel Contraction Assays
To determine the capacity of SPARC-null dermal cells to contract collagen gels, SPARC-null and wild-type fibroblasts were plated in gels of increasing collagen concentration. As shown in Figure 5, SPARC-null fibroblasts did not exhibit an increased capacity for collagen gel contraction in comparison to wild-type cells. An equal number of SPARC-null (Figure 5, diamonds) or wild-type dermal fibroblasts (Figure 5, circles) contracted a collagen gel of 0.5 mg/ml to an area of 19.6 mm2, whereas an area of ~75 mm2 was observed with a collagen gel of 1.25 mg/ml. Therefore, there was an inverse correlation between the degree of contraction and the concentration of the collagen gel for both cell genotypes. We have observed in some preparations of fibroblasts a decrease in the capacity of SPARC-null cells to contract collagen gels in comparison to wild-type cells. This phenomenon has been variable. More importantly, SPARC-null cells consistently showed an inverse correlation between the capacity of the cells to contract collagen gels and the concentration of collagen within the gel itself. Because SPARC-null fibroblasts
Primary fibroblasts contract collagen gels of lower concentration to a greater degree than observed with collagen gels of higher concentration. Equal numbers of wild-type (open circles) and SPARC-null dermal fibroblasts (open diamonds) were seeded in triplicate into collagen gels of increasing concentration in 24-well plates for 24 hr. The area of the resulting contracted gel was calculated from measurement of the diameter of each gel. The results shown are representative of three separate experiments. Error bars represent the SEM, most of which lie within the symbol.
did not exhibit an increased capacity for collagen gel contraction in comparison to wild-type cells, the increased wound closure observed in the absence of SPARC most likely results from the altered collagenous ECM of the dermis. Given that SPARC-null skin contained approximately half the amount of collagen (as measured by hydroxyproline analysis) as that of wild-type skin, we propose that the skin of SPARC-null mice is more susceptible to contraction than that of wild-type. This property could account, at least in part, for the enhanced rate of wound closure observed in SPARC-null vs wild-type mice.
Discussion
The function(s) of SPARC, a matricellular protein with several apparent effects in vitro, has been elusive in vivo. Recent evidence from SPARC-null mice indicates that this protein is required for lens transparency and maintenance of bone mass. We now report that the absence of SPARC in mice leads to accelerated dermal wound repair. SPARC-null mice were able to close excisional wounds more rapidly than wild-type counterparts. Given the capacity of SPARC to retard cell cycle progression in culture, we sought to determine whether differences in cell proliferation might account for the increased healing observed in SPARC-null vs wild-type mice. The percentage of proliferating cells at Days 4 and 7 did not appear to differ significantly in SPARC-null vs wild-type wounds, as determined by quantification of PCNA- and Ki-67-immunoreactive cells. However, we can not exclude the possibility that differences in the rate of proliferation of individual cells present in the wound bed might be accelerated in SPARC-null mice, i.e., the absence of SPARC might render cells more sensitive to mitogenic stimuli and thus shorten the time required for cell-cycle traverse at early times after injury. Nevertheless, a substantial increase in the profile of dividing cells was not apparent in SPARC-null wounds and therefore is most likely not the principal mechanism that leads to accelerated wound closure in these mice.
Consistent differences in the deposition of certain ECM components (fibrillar collagen, Type IV collagen, and fibronectin) at the wound site were not detected between SPARC-null and wild-type wounds, although some variability among wounds was observed. However, the collagen composition of the uninjured dermal ECM in SPARC-null mice is significantly reduced in comparison to wild-type mice. Hydroxyproline analysis confirmed that SPARC-null skin contains 50% less collagen than age-matched wild-type skin. However, abundant production of newly synthesized collagen in the wound bed was not substantially reduced in the absence of SPARC. Therefore, we propose that collagen production in response to wounding is more robust than during routine maintenance of the dermal collagen layer, most likely due to the presence of multiple collagen-stimulating factors in the wound. Alternatively, SPARC might be involved in the incorporation and/or stabilization of collagen into the more mature fibers of remodeled, structured ECM rather than the expression and secretion of collagen I per se. In this case, alterations in collagen expression and accumulation would not be substantial in the days and weeks after injury but would become more apparent at later time points during scar formation. Experiments are under way to address these possibilities.
The reduced collagen content of the SPARC-null dermis appears to render the skin intrinsically more susceptible to cell contraction than that of wild-type mice. Collagen gel contraction assays demonstrated a linear relationship between the concentration of collagen gels and their propensity for contraction by dermal fibroblasts. In general, collagen gel contraction by SPARC-null vs wild-type dermal fibroblasts was not increased. Therefore, the increased wound contraction was most likely due not to an increased cellular capacity for contraction in the absence of SPARC but instead to an altered ECM. Interestingly, the tight-skin mouse (Tsk) exhibits delayed wound contraction in response to excisional wounding (Ehrlich and Needle 1983). This mouse has a substantially increased dermal collagen component in comparison to wild-type mice (Osborn et al. 1983). Moreover, similar to the SPARC-null case, primary fibroblasts isolated from Tsk animals are not deficient in their capacity to contract collagen gels (Ehrlich and Needle 1983). Alterations in the interactions of fibroblasts with their pericellular ECM in the Tsk mouse have been proposed as a mechanism responsible for the decreased contraction of their dermal wounds (Hembry et al. 1986). Perhaps the increased collagen content of the Tsk mouse presents an ECM less amenable to contraction, whereas the decreased collagen content in the SPARC-null mouse provides an ECM that is more readily contracted.
The capacity of SPARC to influence the amount of collagen in the uninjured dermis was somewhat unexpected. Although SPARC binds to a number of different collagens, including Types I, III, IV, and V, all of which are represented in the dermal or epidermal ECM, a prior function of SPARC in collagen accumulation has not been described (Lane and Sage 1994; Singer and Clark 1999). Whether SPARC acts directly to influence collagen fiber assembly in the extracellular milieu or through other mechanisms remains to be determined. Unlike SPARC-null mesangial cells, SPARC-null fibroblasts did not display significant differences in Type I collagen production under standard growth conditions, in comparison to wild-type cells (Francki et al. 1999). Preliminary results from ultrastructural analysis of the collagenous matrix of the dermis indicate that the collagen fibrils formed in the absence of SPARC are substantially altered (our unpublished experiments). Further experiments are in progress to characterize the nature of the aberrant collagen fibril organization.
The activity of MMPs is critical for both collagen gel contraction and wound contraction (Parks 1999). For example, mice deficient in stromelysin-1 (MMP-3) exhibited delayed closure of excisional wounds, whereas incisional wound repair in these animals was not affected (Bullard et al. 1999a). Because contraction is believed to be of greater consequence in the healing of excisional wounds compared to incisional wounds (Mast 1992), MMP-3 was implicated as a factor important for wound contraction. In addition, primary fibroblasts isolated from MMP-3-null mice did not exhibit efficient collagen gel contraction in vitro (Bullard et al. 1999b). In preliminary experiments we have not observed significant differences in the expression of MMPs in SPARC-null vs wild-type fibroblasts in vitro. Furthermore, because SPARC-null fibroblasts did not contract collagen gels more efficiently than wild-type fibroblasts, we do not believe that the mechanism of increased wound closure in the absence of SPARC derives from altered MMP levels or activation in dermal fibroblasts. It is possible, however, that the extracellular milieu of SPARC-null dermis manifests altered MMP activity. Experiments are under way to address MMP activity in SPARC-null animals in vivo.
Although the composition of the ECM in the absence of SPARC undoubtedly influences the process of accelerated closure, the effect of SPARC on growth factor activity might also contribute to the increase in the rate of wound repair in SPARC-null mice. The capacity of SPARC to bind to and diminish the activity of growth factors, such as PDGF and VEGF, and to associate with the ECM make it a prime candidate for a modulator of cell behavior in response to injury. The absence of SPARC might allow increased migratory and/or contractile activity mediated by growth factors and would thereby accelerate wound repair (Clark et al. 1989; Pierce et al. 1991; Raines et al. 1992; Kupprion et al. 1998).
The concept that matrix-associated proteins might modulate cell behavior in response to injury is gaining support with the publication of recent studies involving transgenic animals. A number of matricellular proteins show increased expression in response to injury: thrombospondins 1 and 2, osteopontin, and tenascin C (Reed et al. 1993; Liaw et al. 1998; Mackie et al. 1988). Thrombospondin 2-null mice display an increase in vascular structures in non-injured skin and an accelerated excisional wound healing response that is probably related to their increased angiogenic potential (Kyriakides et al. 1998, 1999). Mice that do not produce osteopontin exhibit a decreased level of wound debridement and a greater disorganization of the newly formed matrix after injury, but they do not exhibit a substantial difference in rates of healing from those of wild-type mice (Liaw et al. 1998). Tenascin C-null mice do not appear to have gross differences in their capacity to repair excisional wounds, although a decrease in fibronectin deposition is observed in tenascin C-null vs wild-type mice (Forsberg et al. 1996). Hence, the expression of matricellular proteins at sites of injury might govern in part cellular interaction with growth factors and ECM to fine-tune the complex array of processes that contribute to healing.
In conclusion, cutaneous wound healing provides an excellent milieu for the study of factors that influence cell interaction with ECM. For repair of damage to the skin, the matrix must be degraded, rearranged, and eventually reconstructed to restore a functional barrier for the organism. We have found evidence that the absence of SPARC expression enhances wound closure in mice. Future experiments will provide valuable insights into the molecular mechanisms governing SPARC activity in the skin and in wound repair.
Footnotes
Acknowledgements
Supported by National Institutes of Health grants GM 40711 and HL 59574 (EHS), AG 15837 (MJR), and DK 07467 (ADB), by National Science Foundation EE 04–150 (EHS), and by a Beeson Scholar Award (MJR).
We are indebted to Emmett Pinney for the hydroxyproline analysis. In addition, we would like to thank Juliet G. Carbon for valuable assistance with the mouse colony, and Dr Pauli Puolakkainen and the members of the Sage Laboratory for helpful discussions and suggestions.