Accelerated Wound Healing in Mice With a Disruption of the Thrombospondin 2 Gene

Accelerated Wound Healing in Mice With a Disruption of the Thrombospondin 2 Gene Themis R. Kyriakides, Jessica W.Y. Tam, and Paul Bornstein Department of Biochemistry, University of Washington, Seattle, Washington, U.S.A.

Mice that lack the extracellular matrix protein thrombospondin 2 have, among several abnormalities, an increase in vascular density, abnormal collagen fibrils, and dermal fibroblasts that are defective in adhesion. These findings suggested that responses involving these processes, such as wound healing, might be altered. To investigate the healing process, excisional wounds were made with the aid of a biopsy punch. Such wounds, observed over a 14 d period, appeared to heal at an accelerated rate and with less scarring in thrombospondin 2-null mice. Histologic analysis of thrombospondin 2-null wound sites revealed the presence of an irregularly organized and highly vascularized granulation tissue. In addition, thrombospondin 2-null wounds retained a higher total cellular content, than control wounds. No differences in wound re-epithelization rates were

observed, but thrombospondin 2-null epithelia formed rete pegs and were thicker than control epithelia. By immunohistochemistry, we detected elevated levels and an irregular deposition pattern for fibronectin in thrombospondin 2-null wounds, observations that correlated with the abnormal collagen organization in the granulation tissue. Immunostaining for thrombospondin 2 in control wounds showed that the protein is present in both early and late wounds, in a scattered cell-associated pattern or widely distributed cell- and matrix-associated pattern, respectively. Our results suggest that thrombospondin 2 plays a crucial part in the organization and vascularization of the granulation tissue during healing, possibly by modulating fibroblast–matrix interactions in early wounds and regulating the extent of angiogenesis in late wounds. Key words: angiogenesis/fibrillogenesis/matricellular/remodeling. J Invest Dermatol 113:782–787, 1999

T

when cultured in vitro demonstrated an adhesion defect. These observations suggested that TSP2 has important roles in the formation of collagen fibrils, in the interaction of cells with the ECM, and in the vascularization of tissues. Recently, we observed that following the subcutaneous implantation of silicone disks in TSP2-null mice, the resulting foreign body capsules were highly vascularized, and contained irregularly shaped collagen fibers (Kyriakides et al, 1999). These observations corroborated several aspects of the phenotype and suggested that the tendency for these abnormalities to develop persisted in adult TSP2-null mice. As the formation of foreign body capsules involves many of the processes that make up the wound healing response (Woodward, 1982; Williams, 1989; Tang and Eaton, 1995) we hypothesized that the latter may also be altered in TSP2-null mice. To investigate the healing response, wounds were made on the backs of TSP2-null and control mice with the aid of a biopsy punch, and the gross and histologic appearance of the wounded areas was observed for a period of 14 d. Wound tissue was also harvested and processed for histologic analysis at selected time points post-wounding. The total cellularity, vascularity, and appearance of the granulation tissue and regenerating epithelium were examined. In addition, we performed immunohistochemical analysis of the deposition of fibronectin, a component of the ECM that participates in wound healing. Control wound sections were also analyzed by immunohistochemistry to determine the spatiotemporal distribution of TSP2. This study provides evidence for expression of TSP2 during the wounding response. Our results suggest that TSP2 performs a number of important functions during wound healing but that, surprisingly, its absence may lead to improved healing.

hrombospondin 2 (TSP2) is a member of a protein family that is comprised of five secreted modular glycoproteins (TSP 1–5; Frazier, 1991; Bornstein, 1992; Adams and Lawler, 1993; Bornstein and Sage, 1994). It is thought that these proteins are multifunctional components of the extracellular matrix (ECM) that can modulate cell–matrix interactions (Bornstein, 1995). Their biologic roles, however, are poorly understood. TSP2, like its structural homolog TSP1 (Good et al, 1990; Tolsma et al, 1993) has been shown to modulate the activity of basic fibroblast factor on endothelial cells and to inhibit angiogenesis (Volpert et al, 1995). TSP2 mRNA and protein have been detected in mouse embryos predominantly in connective tissues such as dermis, tendon, and ligaments (Iruela-Arispe et al, 1993; Kyriakides et al, 1998a; Tooney et al, 1998). A structural role for TSP2 in these tissues has not been demonstrated. In adult mice, the distribution of TSP2 is limited and is associated predominantly with cells such as the skin fibroblasts (Kyriakides et al, 1998a). Consistent with its distribution pattern and anti-angiogenic activity, disruption of the Thbs2 gene in mice resulted in the development of a pleiotropic phenotype that included aberrations in collagen fibrillogenesis and angiogenesis, findings which were more prominent in tissues such as skin (Kyriakides et al, 1998b). In addition, TSP2-null skin fibroblasts Manuscript received April 1, 1999; revised July 13, 1999; accepted for publication August 5, 1999. Reprint requests to: Dr. Paul Bornstein, Department of Biochemistry, Box 357350, University of Washington, Seattle, WA 98195, U.S.A. Abbreviation: TSP, thrombospondin.

·

0022-202X/99/$14.00 Copyright © 1999 by The Society for Investigative Dermatology, Inc.

782

VOL. 113, NO. 5 NOVEMBER 1999

THE ROLE OF TSP2 IN WOUND HEALING

783

MATERIALS AND METHODS Mice Three month old 129SV/J age-matched wild-type and TSP2-null mice, generated by gene targeting (Kyriakides et al, 1998b), were used in this study. All wounded mice were housed individually. A total of three mice/genotype/time point, and two wounds/mouse were utilized. Wounds Excisional wounds were made with the aid of 6 mm punches (Acuderm, Ft Lauderdale, FL) essentially as described by Subramaniam et al (1997), with the only exception that Avertin (Aldrich, Milwaukee, WI) was used for anesthesia. At selected time points, mice were killed and wounds were harvested with a rim of approximately 2 mm of unwounded tissue. A mark was made on the top left corner of each excised wound for orientation. Processing and staining Wound tissues were sandwiched in porous holders to prevent them from curling, fixed in 10% formalin (Z-fix, Anatech, Battle Creek, MI), processed, cut in half, and embedded in paraffin. To reduce variability during cutting and embedding, wounds were laid flat and cut with a single edge razor blade placed vertically over the middle of the wound. The two resulting pieces were placed in paraffin wound-face down and were kept vertical until the paraffin hardened. Five micrometer sections were stained with hematoxylin and eosin (H&E) as described previously (Kyriakides et al, 1998a). To visualize nuclei, sections were stained with 5 µg per ml DAPI (Sigma, St Louis, MO) in 1% bovine serum albumin, 0.01% Triton X-100. To visualize red blood cells, which autofluoresce, sections were deparaffinized and rehydrated and examined with the aid of fluorescence optics. A Nikon Eclipse 800 microscope equipped with fluorescence optics was used for all examinations. Immunohistochemistry Immunolocalization of TSP2 was performed with two rabbit anti-TSP2 antibodies as described previously (Kyriakides et al, 1998a). All of the data shown were obtained with a polyclonal antibody made against the full-length TSP2 molecule; similar results were obtained with a polyclonal antibody against the amino-terminal domain of TSP2. Identical procedures were followed for the detection of TSP1 (with a rabbit anti-mouse TSP1 antibody, a kind gift of Dr. D. Mosher, University of Wisconsin, Madison, WI) and fibronectin (with a rabbit anti-fibronectin antibody, Sigma). Immunolocalization for PECAM-1 (CD31) was performed as described above except that tissues were treated with 0.025% trypsin/pronase for 30 min at 37°C prior to blocking, according to the supplier’s (Pharmingen, San Diego, CA) instructions. The antibody (clone MEC 13.3) was applied to tissues at a concentration of 10 µg per ml. Quantitation A digital camera was used to capture a series of images from sections stained with DAPI and a grid was placed over the images to enable the estimation of the number of nuclei per unit area. Estimation of epithelial thickness was performed on H&E-stained sections with the aid of an ocular micrometer. For each section, measurements were taken at the two edges of the wound and, depending on length, one or more measurements were taken from the middle of the wound. Areas of epidermal invagination (rete pegs) were not included in the measurements. Estimation of wound neovascularization was performed by counting PECAM-1-positive figures within granulation tissue. Slides were labeled with a code and a series of digital images from the center of each wound was generated. Two investigators counted the number of blood vessels in each image and the files were then renamed and counted again. The collected data revealed insignificant variation between the repeat measurements and the measurements of the two investigators.

RESULTS Gross and histologic appearance of healing wounds Wounds of TSP2-null mice were indistinguishable from control wounds up to day 5. On this day some null animals had lost the wound scab. By day 7, most TSP2-null mice had wounds without a scab, whereas control animals retained their scabs for 2 more days (Fig 1a). Knockout animals examined on day 10 showed a significant improvement in the overall appearance of the wound, with reduced inflammation as indicated by reduced redness, and minimal scarring (Fig 1c). In contrast, control animals either retained a portion of the scab, or had shed it but exhibited a scarred, inflamed wound (Fig 1b). By day 14, wounds of the TSP2-null animals were not easily visible as they exhibited no redness and minimal scarring. Control animals showed a significant improvement in healing, but their wounds still demonstrated

Figure 1. Excisional wounds heal at an accelerated rate in TSP2null mice. One week after the creation of a full thickness wound, wildtype mice retained a substantial scab (left mouse in a) whereas the scab had detached in the TSP2-null mice (right mouse in a). Close up views of day 10 wounds in a wild-type (b) and TSP2-null (c) animal. The latter has a wound without a scab, and with reduced inflammation and little scarring.

superficial signs of inflammation and retained a visible scar (data not shown). Examination of H&E-stained sections of wounds revealed dramatic differences in the behavior of the regenerating epithelium, with the formation of characteristic rete pegs (epidermal invaginations into the dermis) in day 7 wounds of TSP2-null mice (Fig 2b). The rete pegs, which were most obvious between days 5 and 8, were absent from control animals at all time points, and disappeared from TSP2-null animals by day 10. The regenerating epithelium of null animals was also thicker during the same interval (Fig 3). The difference in thickness diminished by day 14 (data not shown). Wounds of both TSP2-null and control animals retained similar levels of inflammatory cells throughout healing (see Fig 2 for day 10 wounds). The organization of the granulation tissue in the remodeling phase, day 14, however, was markedly different between control (Fig 4a) and TSP2-null animals (Fig 4b) with the latter having collagen fibers that lacked the extensive parallel orientation seen in controls. Rather, collagen fibers in mutant wounds were organized in a basket-weave fashion that resembled, somewhat, the appearance of uninjured dermis. This organization may account for the more diffuse deposition pattern of fibronectin in these animals at this time point (see Fig 7b). Differences in the vascularity of granulation tissue were also evident. Sections examined with the aid of fluorescence optics revealed the presence of a significantly higher number of red blood cells in TSP2-null mice in the later phases (days 10–14) of the healing process (data not shown). This observation was also made in H&E-stained sections but was not as easily demonstrable. Immunolocalization of endothelial cells with anti-PECAM-1 antibodies permitted a more direct evaluation of wound neovascularization. Representative sections are shown in Fig 5(a, b). Quantitation of blood vessels, shown in Fig 5(c), demonstrated the prolonged neovascularization of TSP2-null wounds. Unlike control wounds in which a decrease in vascularization occurs with time, TSP2-null wounds remain highly vascularized over the observed period. Attempts to visualize the vascular endothelium with antivon Willebrand factor antibodies were unsuccessful. As blood vessels in uninjured dermis adjacent to the wound site stained positive for von Willebrand factor we conclude that endothelial cells in healing wounds do not express this factor. Consistent with

784

KYRIAKIDES ET AL

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 2. Abnormal re-epithelialization of day 7 TSP2-null wounds. Sections of day 7 wounds were stained with H&E. The epithelium covering control wounds (a) remained flat over the wound bed. On the contrary, in TSP2-null wounds (b) the newly formed epithelium invaded the granulation tissue to form rete pegs (arrows). Arrowhead in (b) denotes an area with a high local concentration of inflammatory and red blood cells. Scale bar: 100 µm.

Figure 3. Wound epithelium thickness is increased in day 7 TSP2null mice. H&E-stained day 7 wound sections were analyzed for the thickness of the newly formed epithelial layer with the aid of an ocular micrometer. A total of five sections/wound from six wounds, three mice/ genotype were measured. *p , 0.05.

an increase in vascularity, we also observed an increase in the total cellularity of TSP2-null wounds, which was quantitated following visualization of nuclei with DAPI (Fig 6). The difference in overall cellularity was more pronounced in day 10 wounds, but remained significant through day 14. Immunohistologic analysis of healing wounds Immunohistochemical analysis with an anti-fibronectin antibody showed that this protein is prominently present in day 14 wounds of both control (Fig 7a) and TSP2-null (Fig 7b) animals. The immunostaining pattern in control wounds appeared punctate and resembled the organized alignment of the collagen fibers. On the contrary, the pattern of fibronectin staining in TSP2-null wounds appeared patchy, a distribution which may reflect the abnormal arrangement of collagen fibers (see Fig 4d). Examination of sections from several animals gave us the impression that there was more immunoreaction product in TSP2-null wounds. A more sophisticated quantitative method, however, will have to be used to determine the relative fibronectin levels. Differences in the immunostaining patterns of fibronectin at earlier time points (days 7–10) were also observed, but were more subtle (data not shown). In control wounds the immunostaining pattern for TSP2 varied during the course of healing. In day 3 wounds (Fig 8a), the presence of TSP2 protein was detected only in a small number of fibroblast-like cells. No matrix-associated staining was observed. In day 7, 10, and 14 wounds (Fig 8b–d), the presence of TSP2 protein was prominent, with a greater number of positive fibroblasts and some matrix-associated staining. In day 7 and 10 wounds, the pattern appeared to be widespread in the wound bed, including areas that were highly vascularized (arrows in Fig 8b, c). By day

Figure 4. Abnormal organization of granulation tissue in day 14 wounds. H&E-stained sections of day 14 wounds from control (a) and TSP2-null (b) animals. In control wounds collagen fibers were organized in an arrangement parallel to the epidermis. The collagen fibers in TSP2null wounds were arranged in a basket-weave fashion and lacked an orientation to the epidermis. Scale bar: 50 µm.

14, TSP2 protein appeared to decrease, especially in the matrix, but remained high in cells (arrows in Fig 8d). Low levels of TSP2 protein, however, could still be detected in some areas of the matrix. Examination of day 5 wounds showed inconsistent results with some wounds having patterns that resembled day 3 wounds and some that resembled day 7 wounds (data not shown). Immuno-

VOL. 113, NO. 5 NOVEMBER 1999

Figure 5. TSP2-null wounds remain vascularized during remodeling of granulation tissue. Sections of wounds were stained with antiPECAM1 antibodies and visualized by the peroxidase reaction. Representative sections displaying blood vessels within granulation tissue from day 14 wounds are shown for a wild-type (a) and a TSP2-null wound (b). Scale bar: 50 µm. (c) Results of vessel quantitation for day 7, 10, and 14 wounds. A total of 24 sections (area per high power field, 0.04 mm2) from eight wounds from four animals per time point were quantitated for each genotype. *p ø 0.05.

THE ROLE OF TSP2 IN WOUND HEALING

785

Figure 7. Abnormal pattern of fibronectin deposition in TSP2-null granulation tissue. Sections of day 14 wounds were stained with an antifibronectin antibody and visualized by the peroxidase reaction. The deposition of reaction product in wild-type wounds (a) appeared in a punctate pattern (arrows) and was unevenly distributed. In TSP2-null wounds (b) the deposition was patchy (arrows), and was prominent throughout the granulation tissue. Scale bar: 100 µm.

staining of TSP2-null wounds, as a negative control, revealed no specific immunoreactivity (data not shown). Preliminary immunohistochemical analysis of wounds for TSP1 revealed that this protein is present in the early inflammatory phase (day 3), but not in the later tissue formation phases (days 7–14). Our observations are generally in agreement with previous studies that investigated the deposition of TSP1 during healing (Raugi et al, 1987; DiPietro et al, 1996). Our inability to detect TSP1 in the late phases of healing, during which it has been observed in human wounds (Raugi et al, 1987), may reflect species differences or differences in the sensitivity of the antibodies. Importantly, we did not observe changes in the TSP1 immunostaining patterns of control and TSP2-null wounds (data not shown). DISCUSSION

Figure 6. Wounds in TSP2-null mice retain a high total cellular content during healing. Sections of wounds were stained with DAPI and the number of nucleated cells within the granulation tissue was estimated from 15 sections (four mice per genotype, five sections per wound).*p , 0.05.

The importance of the matricellular protein TSP2 in the processes of collagen fibrillogenesis and angiogenesis was suggested by the abnormalities we observed in TSP2-null mice (Kyriakides et al, 1998b); however, the mechanisms by which TSP2 participates in these processes are not well understood. A number of clues to possible mechanisms have been gathered from in vivo analysis of TSP2-null mice. For example, we have observed that following the implantation of solid PDMS (silastic) disks or cube-shaped polyvinyl alcohol sponges, TSP2-null mice mount exaggerated fibroplastic and angiogenic responses (Kyriakides et al, 1999; unpublished observations). In the present investigation we examined

786

KYRIAKIDES ET AL

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 8. Immunolocalization of TSP2 in wild-type granulation tissue. Sections from day 3 (a), day 7 (b), day 10 (c), and day 14 (d) were stained with an anti-TSP2 antibody and visualized by the peroxidase method. TSP2 was localized to a few selected cells (arrows in a) in day 3 wounds. At days 7, 10, and 14, TSP2 appeared in both cells and the surrounding matrix (b–d). In day 7 and 10 wounds some of the immunoreactivity was found in areas that were highly vascularized (arrows in b and c). In day 14 wounds TSP2 was more prominent in individual cells (arrows in d) than in the matrix. No immunoreactivity was detected in wounds from TSP2-null mice (data not shown). Scale bars: 50 µm.

whether these abnormalities extended to the remodeling and neovascularization of healing excisional wounds. As the presence of TSP2 in the wound healing response had not been demonstrated, we also performed an immunohistochemical analysis of its distribution. Wound healing consists of a series of distinct but overlapping processes including inflammation, granulation tissue formation, and remodeling (Clark, 1993; Mutsaers et al, 1997). Key early events during healing are the invasion of collagen-secreting fibroblasts and new vasculature into the wound bed. At this early time (day 3), deposition of TSP2 was sparse and localized to a small number of cells, believed to be fibroblasts. Later, the response consisted of processes whose function is to form and then remodel granulation tissue, also known as provisional matrix, to a state resembling that of uninjured dermis. During this phase a decrease in the degree of vascularization, possibly due to regression, was also observed. The deposition of TSP2 at this time (days 7–14) was pronounced and widespread. Thus, an association can be made between low levels of TSP2 and increased angiogenesis in early wounds, and high levels of TSP2 and regressing angiogenesis in late wounds. In addition, high levels of TSP2 coincide with granulation tissue formation and matrix remodeling. These associations are also supported by observations made during the healing of wounds in TSP2-null mice. Consistent with the presence of very few TSP2-positive cells in early wounds (days 0–4) in control mice, we were unable to observe any early abnormalities in the wounds of TSP2-null mice. The most prominent alterations in the appearance of wounds in the latter animals occurred between days 7 and 10, a time corresponding to maximal TSP2 deposition in control mice. These associations provide strong circumstantial evidence for the participation of TSP2 in the wound-healing response. Wounds in TSP2-null mice appeared to heal at an accelerated rate as judged by premature loss of the scab and an apparent reduction in inflammation and scarring. The mechanism leading to this accelerated response is unknown, but the ability of the epithelium to invade the granulation tissue to form rete pegs and increase in thickness may be associated with scab loss. The thickened epithelium may prevent an adequate visualization of the underlying inflammatory response, which was nevertheless apparent in histologic sections (Fig 2b), and may also account for the improved appearance of the wounds. The lack of organized parallel collagen fibers, that are the hallmark of a scarred wound (De Vries et al, 1995; Nodder and Martin, 1997), from the wound bed of TSP2null mice may account for the reduced scarring seen in these animals. As TSP2 is absent from murine epithelial cells (Kyriakides et al,

1998a; Tooney et al, 1998), and we were unable to detect it in either the normal or wound epidermis, we found the formation of rete pegs in TSP2-null wounds surprising. Possibly, the abnormal structure and composition of the granulation tissue, or increased vascular density, can produce signals that alter the migration and proliferation patterns of re-epithelializing cells. The finding that rete peg formation is absent from late wounds, despite their prolonged neovascularization, suggests that changes within the granulation tissue can affect the behavior of the newly formed epithelium. A similar appearance of the epithelium with formation of rete pegs is seen in human psoriatic skin lesions. Interestingly, this condition is believed to be associated with an increase in the dilatation and tortuosity of the microvasculature, as well as an increase in total microvasculature (Barton et al, 1992). It should be noted that the re-epithelization rate and reduction of the wound area were indistinguishable between TSP2-null and control animals. The latter observation was unexpected in view of the fact that TSP2-null dermal fibroblasts do not contract collagen gels efficiently.1 Collagen gel contraction has been proposed as an in vitro model for predicting the rate of wound contraction, which is believed to result from the contractile action of myofibroblasts within the wound bed (Grinnell, 1994; Moulin et al, 1996). In contrast, Gross et al (1995) have shown that the systematic removal of granulation tissue does not compromise wound closure; this finding suggests that results obtained by the collagen gel contraction method may not be reliable in predicting the rate of wound contraction in vivo. Our observations suggest that, at least for 4 and 6 mm diameter wounds, the inability of TSP2-null dermal fibroblasts to contract collagen gels efficiently does not compromise wound closure and support the findings of Gross et al (1995). It should be noted, however, that we have not examined granulation tissue fibroblasts, which may have normal contractile properties. We are currently isolating wound fibroblasts that we will evaluate in the collagen gel contraction assay. As expected, we observed abnormally shaped collagen fibers and a prolonged period of neovascularization in TSP2-null wounds. The role of TSP2 in collagen fibrillogenesis has yet to be elucidated. To date, we have been unable to detect TSP2 as a component of normal dermis or tendon collagen fibers, and thus we postulate that TSP2 may function at the cellular level during fibrillogenesis. We have found that TSP2-null fibroblasts have reduced levels of 1Kyriakides TR, Zhu YH, Tam JWY, Zang Z, Bornstein P: Accelerated wound healing in mice with a disruption of the thrombospondin 2 gene. Mol Biol Cell 9 (Suppl.59a), 1998 (abstr.)

VOL. 113, NO. 5 NOVEMBER 1999

(α1 integrin and increased levels of matrix metalloproteinase 2.2 If and how these alterations affect fibrillogenesis remains to be determined. Preliminary electron microscopic observations also revealed that tendon fibroblasts from TSP2-null mice do not properly organize the pericellular environment in which fibril assembly occurs (D. Birk, personal communication). It is possible that fibroblasts are unable to establish and maintain effective interactions with growing collagen fibrils due to cell surface changes that compromise adhesion. The mechanism by which TSP2, and possibly TSP1, inhibit angiogenesis also remains elusive. The observations we have made in the TSP2-null mouse suggest that TSP2 is a potent inhibitor of angiogenesis, especially during tissue repair. Furthermore, the lack of the anti-angiogenic activity of TSP2 cannot be compensated by TSP1, which is also present in healing wounds (Raugi et al, 1987; DiPietro et al, 1996). It is of interest to note that preliminary analysis of wounds in TSP1-null mice also showed prolonged neovascularization.3 The fact that both TSP2-null and TSP1-null mice display prolonged neovascularization, despite the presence of TSP1 and TSP2, respectively, suggests that these two proteins act either at different times or by different mechanisms. The latter possibility is supported by the finding that blocking TSP1 function by the administration of anti-sense oligonucleotides, thus creating conditions similar to those in TSP1 null animals, led to a decreased re-epithelization rate and a delay in wound healing (DiPietro et al, 1996). This observation is in contrast to the finding that wound healing is accelerated in TSP2-null mice. Both TSP2 and TSP1 have been termed matricellular proteins (Bornstein, 1995), a category which includes other extracellular matrix proteins such as osteopontin, SPARC/osteonectin, tenascinC, and vitronectin. These proteins are believed to have dynamic functions in the ECM, including the modulation of cell–matrix interactions. An analysis of the wound healing response in mice that are null for some of these proteins has revealed alterations in the organization of the remodeling matrix. For example, in wounds of mice that lack osteopontin collagen fibrils in the deep wound were shown to be of a smaller than normal diameter (Liaw et al, 1998). Mice that lack tenascin-C, appeared to deposit reduced levels of fibronectin within the wound (Forsberg et al, 1996). Despite the presence of these aberrations, wound healing in these mice was not compromised. This finding was especially surprising as changes in the deposition of fibronectin, a key ECM component during healing (ffrench-Constant et al, 1989), would be expected to have negative consequences for the process. In addition, tenascin-C was previously shown to participate in the wound healing response, where it was believed to influence the migration and proliferation of keratinocytes at the wounded area (Mackie et al, 1988; Latijnhouwers et al, 1996). On the contrary, TSP2-null animals show significant changes in the healing response. The mechanisms involved in the development of these abnormalities remain to be determined. Whether the accelerated wound healing response observed in TSP2-null mice is desirable will be judged in part by the outcome of ongoing evaluations of the tensile strength of incisional wounds. If the tensile strength of healed wounds in TSP2-null mice is not compromised, the inhibition of synthesis of TSP2 might be used to improve healing in human wounds.

This work was supported by National Institutes of Health grants HL18645 and DE08229 and by the University of Washington Engineered Biomaterials Engineering Research Center (NSF grant EEC9529161). We thank Patricia Jun

2Yang Z, Kyriakides TR, Bornstein P: Altered expression of integrins and matrix metalloproteinases in thrombospondin 2-null mouse fibroblasts. Mol Biol Cell 9 (Suppl.429a), 1998 (abstr.) 3Polverini PJ, DiPietro LA, Dixit VM, Hynes RO, Lawler J: Thrombospondin 1 knockout mice show delayed organization and prolonged neurovascularization of skin wound. FASEB J 9:A272 1995 (abstr.)

THE ROLE OF TSP2 IN WOUND HEALING

787

for technical assistance and Drs May Reed and John Olerud for advice and a critical reading of the manuscript.

REFERENCES Adams JC, Lawler J: Diverse mechanisms for cell attachment to platelet thrombospondin. J Cell Sci 104:1061–1071, 1993 Barton SP, Abdullah MS, Marks R: Quantification of microvascular changes in the skin in patients with psoriasis. Br J Dermatol 126:569–574, 1992 Bornstein P: The thrombospondins: Structure and regulation of expression. FASEB J 6:3290–3299, 1992 Bornstein P: Diversity of function is inherent in matricellular proteins: an appraisal of thrombospondin 1. J Cell Biol 130:503–506, 1995 Bornstein P, Sage EH: Thrombospondins. Methods Enzymol 245:62–85, 1994 Clark RAF: Biology of dermal wound repair. Dermatol Clin 11:647–666, 1993 De Vries HJC, Zeegelaar JE, Middelkoop E, Van Gijsbers G, Marle J, Wildevuur CHR, Westerhof W: Reduced wound contraction and scar formation in punch biopsy wounds. Native collagen dermal substitutes. A clinical study. Br J Dermatol 132:690–697, 1995 DiPietro LA, Nissen NN, Gamelli RL, Koch AE, Pyle JM, Polverini PJ: Thrombospondin 1 synthesis and function in wound repair. Am J Pathol 148:1851–1860, 1996 Forsberg E, Hirsch E, Frohlich L, et al: Skin wounds and severed nerves heal normally in mice lacking tenascin-C. Proc Natl Acad Sci USA 93:6594– 6599, 1996 Frazier WA: Thrombospondins. Curr Opin Cell Biol 3:792–799, 1991 ffrench-Constant C, Van De Water L, Dvorak HF, Hynes RO: Reappearance of an embryonic pattern of fibronectin splicing during wound healing in the adult rat. J Cell Biol 109:903–914, 1989 Good DJ, Polverini PJ, Rastinejad F, Le Beau MM, Lemons RS, Frazier WA, Bouck NP: A tumor suppressor-dependent inhibitor of angiogenesis is immunologically and functionally indistinguishable from a fragment of thrombospondin. Proc Natl Acad Sci USA 87:6624–6628, 1990 Grinnell F: Fibroblasts, myofibroblasts, and wound contraction. J Cell Biol 124: 401–404, 1994 Gross J, Farinelli W, Sadow P, Anderson R, Bruns R: On the mechanism of skin wound ‘contraction’: a granulation tissue ‘knockout’ with a normal phenotype. Proc Natl Acad Sci USA 92:5982–5986, 1995 Iruela-Arispe ML, Liska DJ, Sage EH, Bornstein P: Differential expression of thrombospondin 1, 2 and 3 during murine development. Dev Dyn 197:40– 56, 1993 Kyriakides TR, Zhu Y, Yang Z, Bornstein P: The distribution of the matricellular protein thrombospondin 2 in tissues of embryonic and adult mice. J Histochem Cytochem 46:1007–1015, 1998a Kyriakides TR, Zhu Y, Smith LT, et al: Mice that lack thrombospondin 2 display connective tissue abnormalities that are associated with disordered collagen fibrillogenesis, an increased vascular density, and a bleeding diathesis. J Cell Biol 140:419–430, 1998b Kyriakides TR, Leach KJ, Hoffman AS, Ratner BD, Bornstein P: Mice that lack the angiogenesis inhibitor, thrombospondin 2, mount an altered foreign body reaction characterized by increased vascularity. Proc Natl Acad Sci USA 13:4449–4454, 1999 Latijnhouwers MA, Van Bergers M, Bergen BH, Spruijt KI, Andriessen MP, Schalkwijk J: Tenascin expression during wound healing in human skin. J Pathol 178:30–35, 1996 Liaw L, Birk DE, Ballas CB, Whitsitt JS, Davidson JM, Hogan BLM: Altered wound healing in mice lacking a functional osteopontin gene (ssp1). J Clin Invest 101:1468–1478, 1998 Mackie EJ, Halfter W, Liverani D: Induction of tenascin in healing wounds. J Cell Biol 107:2757–2767, 1988 Moulin V, Castilloux G, Jean A, Garrel DR, Auger FA, Germain L: In vitro models to study wound healing fibroblasts. Burns 22:359–362, 1996 Mutsaers SE, Bishop JE, McGrouther G, Laurent GJ: Mechanisms of tissue repair: from wound healing to fibrosis. Int J Biochem Cell Biol 29:5–17, 1997 Nodder S, Martin P: Wound healing in embryos: a review. Anat Embryol 195: 215–228, 1997 Raugi GJ, Olerud JE, Gown AM: Thrombospondin in early human wound tissue. J Invest Dermatol 89:551–554, 1987 Subramaniam M, Saffaripour S, Van De Water L, Frenette PS, Mayadas TN, Hynes RO, Wagner DD: Role of endothelial selectins in wound repair. Am J Pathol 150:1701–1709, 1997 Tang L, Eaton JW: Inflammatory responses to biomaterials. Am J Clin Pathol 103:466–471, 1995 Tolsma SS, Volpert OV, Good DJ, Frazier WA, Polverini PJ, Bouck N: Peptides derived from two separate domains of the matrix protein thrombospondin-1 have anti-angiogenic activity. J Cell Biol 122:497–511, 1993 Tooney P, Sakai T, Sakai K, Aeschlimann D, Mosher D: Restricted localization of thrombospondin-2 protein during mouse embryogenesis: a comparison to thrombospondin 1. Matrix Biol 17:131–143, 1998 Volpert OV, Tolsma SS, Pellerin S, Feige J-J, Chen H, Mosher DF, Bouck N: Inhibition of angiogenesis by thrombospondin 2. Biochem Biophys Res Commun 217:326–332, 1995 Williams DF: A model for biocompatibility and its evaluation. J Biomed Eng 11: 185–191, 1989 Woodward SC: How fibroblasts and giant cells encapsulate implants: considerations on design of glucose sensors. Diabetes Care 5:278–281, 1982