Repair of wounds located on the limbs, but not the body, of horses is often accompanied by the formation of exuberant granulation tissue (EGT) which behaves clinically like a benign tumor and resembles the human keloid 12 in that the evolving scar is trapped in the proliferative phase of repair, leading to fibrosis. This condition ultimately generates extensive scarring, detrimental to function. Indeed, a recent study reported that approximately 7% of injuries leading to the retirement of racehorses are the result of a wound 3. Efforts to enhance wound repair and to prevent unwanted scar formation in both humans and horses have been disappointing, possibly owing to insufficient knowledge of the underlying mechanisms.

We recently identified genes expressed in wound margin biopsies of horses, using suppression subtractive hybridization (SSH) 4. Two hundred and twenty six cDNAs were found to be differentially expressed during the proliferative phase of repair of normal body wounds, 129 of which matched against GenBank databases. In the study described herein, we targeted a previously identified SSH-derived cDNA fragment corresponding to β-catenin (CTNNB1) and a second corresponding to equine platelet endothelial adhesion molecule-1 (PECAM1) for their potential contribution to the proliferative phase of repair, which is aberrant following wounding on the limbs of horses. CTNNB1 had already been demonstrated to play a role in fibroproliferative disorders 56 and PECAM1 was chosen because of its known interaction with β-catenin 7.

Beta-catenin is an 88-kDa highly conserved protein that is a structural component of the adherens junction (AJ), a cadherin-dependent adhesive structure intricately linked to the actin microfilament network and located between adjacent epithelial cells. Beta-catenin is a central mediator in the canonical wingless (Wnt) signaling pathway, which exerts remarkable control over cellular phenotype and behavior 8. When the Wnt pathway is quiescent, β-catenin participates in AJs which must disassemble to allow for cell migration, a process critical to epithelialization following wounding. When freed from intercellular contacts, β-catenin is phosphorylated and targeted for degradation 6. Activation of the Wnt pathway inhibits phosphorylation, leading to cytosolic stabilization of β-catenin. This stabilized form subsequently translocates to the nucleus where it functions as a transcriptional co-activator of T cell factor/lymphoid enhancer factor (Tcf/Lef) target genes 9.

Beta-catenin plays a cell-specific role during wound repair. Protein levels are transiently elevated in mesenchymal cells during the proliferative phase of repair 5, which is thought to regulate the growth of dermal fibroblasts 1011. Moreover, β-catenin stimulates the growth of microvascular endothelial cells 12. In contrast, β-catenin inhibits migration of human epithelial cells in vitro, which may impair wound closure in vivo 13. Hyperplastic wounds in man exhibit a prolonged phase of elevated β-catenin protein levels and concurrent tcf-dependent transcriptional activation of target genes involved in fibroproliferative processes, such as alpha smooth muscle actin (α-SMA), fibronectin (FN) and collagen type III (COLIII) 5. A transgenic mouse model, in which stabilized β-catenin is expressed in mesenchymal cells, develops aggressive fibromatosis (desmoid tumors) and hyperplastic cutaneous wounds 11.

PECAM1 (CD31) is a 130-kDa transmembrane glycoprotein member of the immunoglobulin superfamily that is constitutively expressed on platelets, specific cells of the immune system, as well as on endothelial cells, particularly at intercellular junctions 14. Its ectodomain mediates adhesion 15 while its cytoplasmic portion serves as a scaffold for cytoskeletal proteins 16 and for signaling 17. PECAM1 has been implicated in a number of important biological processes including modulation of adhesion 18, cell migration 19, inflammatory cell infiltration 2021, phagocytosis by macrophages 22, endothelial permeability 23 and angiogenesis 2425.

PECAM1 binds tyrosine-phosphorylated β-catenin having left AJs and become localized to the cytosol 16. Precisely, PECAM1 functions as a reservoir for tyrosine-phosphorylated β-catenin, maintaining localization at the plasma membrane. This dynamic interaction is involved in the proliferation phase of angiogenesis 16. Biswas et al. 7 identified PECAM1 as a modulator of microvascular endothelial cell proliferation via its participation in the Wnt signaling pathway, thereby stabilizing and promoting the accumulation of transcriptionally active β-catenin.

In the present study we hypothesized that the temporal expression pattern of these two genes with potential roles in angiogenesis and cell proliferation would differ between body wounds which heal normally and limb wounds which are predisposed to the formation of EGT in horses. The specific objectives of this study were to characterize the full-length equine CTNNB1 and PECAM1 cDNAs, and to study the spatio-temporal expression profile of their mRNAs and proteins during the repair of body and limb wounds. The ultimate objective of our research program is to contribute to a better understanding of the dermal repair process, and permit the development of novel diagnostic and therapeutic strategies to resolve wound healing complications in the horse and which could ultimately be extrapolated to man.

The purpose of this study was to determine the temporal expression of CTNNB1/β-catenin and of PECAM1 during the repair of body and limb wounds in horses in an effort to define some of the molecular mechanisms leading to undue scarring. At the time of the study, the cDNAs were not characterized for the horse therefore both were cloned and sequenced. Hybridization of the size-selected libraries revealed the clones containing the full-length sequence of PECAM1 but only a truncated fragment of CTNNB1. Fortunately, the latter allowed the design of a homologous oligonucleotide that was used, along with a heterologous one, to amplify the full-length equine CTNNB1 cDNA sequence.

Results show equine CTNNB1 to be a highly conserved gene and to contain all 16 exons reported in humans. The excellent identity of the sequences for CTNNB1 in mammals (close to 100%) validates the quality of the equine sequence we obtained despite the required amplification by PCR. A high rate of degradation may have contributed to our inefficiency in obtaining the correct hybridization size for the full-length CTNNB1 gene in horse; indeed, in man there exists in the 3'-UTR one copy of AU-rich element (ATTTA), a motif known to contribute to short lived mRNAs.

CTNNB1, first cloned by McCrea et al. 27 in Xenopus laevis, shows no similarity in sequence to the genes for the α-catenins reported in other species while the β-catenin protein shares 70% amino acid identity with both plakoglobin, found in desmosomes, and the product of the Drosophila segment polarity gene "armadillo". Analysis of equine predicted β-catenin, against a protein conserved domains database (PROSITE), showed the presence of 9 Armadillo/plakoglobin repeat profiles, an approximately 40 amino acid long repeated sequence motif. Proteins that contain this type of domain combine structural roles and signaling functions, by generating and transducing signals affecting gene expression 28. The calculated expected molecular weight (85,5 kDa) concords with that previously reported for other species.

A preliminary mRNA gene expression profiling study identified a cDNA fragment that corresponded to PECAM1, suggesting its up-regulation in response to dermal wounding 4. The equine sequence determined herein contained the 16 exons present in the full-length human PECAM1, the predominant isoform detected in human tissue and endothelial cells 26. This contrasts with the main isoform found in murine endothelium which lacks exons 14 and 15 29. While it is possible that other isoforms also exist in horses, our data suggest that, at least in skin, the full-length form predominates.

Processes occurring during the proliferative phase of wound repair ensure restoration of the structural and functional characteristics of skin which depend on both the epithelial and the mesenchymal components of the wound. The severity of scarring of horse limb wounds is related to an excessive proliferative phase where angiogenesis and fibroplasia are exacerbated 30313233. This may result from dysregulation of the molecular components governing the growth of mesenchymal cells. In the study reported herein we have focused on two molecules, CTNNB1 and PECAM1, for their proposed role in endothelial cell and fibroblast proliferation.

The temporal patterns we report for mRNA expression of CTNNB1 and PECAM1 resemble one another. Both genes were constitutively expressed in normal intact skin, but while dermal trauma induced the expression of both mRNAs, the response was more acute for PECAM1 whose levels were significantly increased over baseline one week following wounding compared to two weeks for CTNNB1. This significant up-regulation of mRNA was maintained for the duration of the study for PECAM1 at both anatomic locations and for CTNNB1 in thoracic wounds; conversely, expression of CTNNB1 in limb wounds was not significantly elevated following wounding. Finally, mRNA expression in thoracic wounds significantly surpassed that of limb wounds for PECAM1, one week post-wounding, and for CTNNB1, three, four and six weeks post-wounding.

While possible that the temporal pattern of mRNA expression might reflect the different levels at the various sampling sites, the wounds were very close together such that we think it unlikely that anatomic or metabolic differences (blood supply, muscle coverage, etc) are important. Previous studies using a similar model 30313233 have not reported an influence of the exact location on the body or the lower limb. Likewise a current study has shown no difference in the rate of closure between wounds located distally or more proximally on the lower limb (Monteiro et al, unpublished data).

The correspondence in the patterns of mRNA expression following wounding is consistent with the PECAM1/β-catenin protein partnership proposed by Biswas et al. 73435. Indeed, the cytoplasmic portion of PECAM1 binds cytosolic β-catenin and this dynamic interaction is involved in angiogenesis 16. More specifically, by enhancing the accumulation of stabilized β-catenin, PECAM1 modulates microvascular endothelial cell proliferation via transcriptional activation of growth promoting genes. Furthermore, a significant increase in PECAM1 mRNA levels preceded that of CTNNB1 following wounding in our study, consistent with the fact that upon cytokine stimulation endothelial cells initially induce PECAM1, followed by an increased expression of β-catenin 7.

While one might have anticipated higher levels of equine CTNNB1 mRNA in limb wounds predisposed to the development of EGT, a form of fibroproliferative disease, our opposite findings may reflect the mechanism of post-transcriptional regulation of β-catenin during wound repair, which controls how mRNA is translated into protein. Indeed, Cheon et al. documented that in the absence of any change in mRNA level, β-catenin protein, expressed primarily in dermal mesenchymal cells, was nonetheless transiently increased during the proliferative phase of normal wound healing in humans while exhibiting a sustained elevation in hyperplastic wounds 5. Thus, a lack of correlation between mRNA and protein levels might underlie our unexpected findings; specifically, why CTNNB1 mRNA levels remained elevated when immunostaining for β-catenin waned, and why CTNNB1 mRNA levels in thoracic wounds surpassed those of limb wounds at weeks 3, 4 and 6 of healing.

In the horse, epithelial cells migrate slowly on limb wounds which are concurrently afflicted by over-abundant fibroblast proliferation; both are influences attributed to β-catenin protein 1013. Although immunohistochemical studies did not suggest quantitative differences in β-catenin expression between body and limb wounds, our investigation did reveal that while the protein remains transcriptionally active (nuclear stain) in dermal mesenchymal cells for the duration of the study, the intensity of the stain diminishes between the 4^th and 6^th weeks of healing. This pattern suggests that the signal would have disappeared entirely from normally healed wounds had the study been extended a few weeks. Conversely, strong immunostaining for β-catenin was detected in biopsies of 3 and 6 month old EGT, a condition characterized by exaggerated angiogenesis and fibroplasia and which resembles hyperplastic scarring in man in which a prolonged duration of β-catenin protein elevation has been reported 5. Our findings may indicate that mesenchymal cells in EGT are behaving as though in a prolonged active proliferative phase of wound healing. In other words, the high levels of β-catenin would encourage proliferation and motility of the dermal cells, generating a larger dermal component to the wounds and hampering remodeling which requires reduced β-catenin levels 5.

While the design of our study cannot establish a cause-and-effect relationship between the prolonged elevation of β-catenin and the development of equine EGT, transgenic mouse models expressing a stabilized form of β-catenin in mesenchymal cells develop hyperplastic cutaneous wounds 11. Fibroblasts derived from these mice display increased proliferation, motility and invasiveness when grafted into nude mice. Moreover, primary cell cultures demonstrate Tcf-dependent transcriptional activation 11 consistent with the hypothesis that nuclear β-catenin transactivation of target genes is a primary component of fibrosis.

In contrast to the membrane staining of epithelial cells reported by Cheon et al 5, we identified nuclear β-catenin within epithelial cells of intact skin. Our finding is similar to that of Tsuji et al. 36 who report nuclear immunostaining for β-catenin from the upper spinous to the granular cells in normal, intact human epidermis. The presence of stabilized, transcriptionally active β-catenin no doubt reflects keratinocyte proliferation and differentiation in response to continued renewal of the epithelial component of equine skin 3738.

While PECAM1 is best known as a promoter of angiogenesis, its ability to facilitate neutrophil transendothelial migration 2021 and thus boost the acute inflammatory response 39 is particularly intriguing in view of the data obtained herein. Indeed, it is tempting to draw a parallel between the inefficient and prolonged inflammatory response to wounding in the horse limb 33 and the significantly inferior expression of PECAM1 mRNA in wounds at this site compared to those of the thorax, one week following injury. Conversely, while differences in expression between the two sites disappear from the second week on, one might have expected an increase in limb wounds where new blood vessels are particularly abundant within the granulation tissue 33, in relation to PECAM1's role in angiogenesis. Interestingly, the expression pattern of PECAM1 isoforms has been shown to change during tube formation in vitro, indicating specialized roles for specific isoforms of PECAM1 during angiogenesis 26. While our data suggest that we have cloned the predominant isoform of equine PECAM1, it is possible that another isoform is responsible for modulating endothelial cell adhesive properties during angiogenesis in the horse. Finally, the PECAM1 mRNA expression pattern found in our study corroborates previous assertions concerning the sluggishness hindering the repair of equine wounds, particularly those located on the extremities 13033. To this effect, a recent study in mice showed that both mRNA and protein expression of PECAM1 were induced three days following dermal wounding but mRNA expression returned to baseline by day 12 40. In contrast, expression of equine PECAM1 mRNA was significantly up-regulated in both thoracic and limb wounds for the 6-week duration of the study.

Immunohistochemistry revealed that, as expected of an endothelial cell marker, PECAM1 is principally present in blood vessels. As such, our data reflect that previously reported by Lepault et al. 33 who documented, histologically, more pronounced angiogenesis in healing limb than body wounds of horses. The concomitant presence of PECAM1 and β-catenin on endothelial cells of wound granulation tissue further supports the dynamic interaction between these two molecules, which may modulate the proliferation phase of angiogenesis 716.

This study is the first to characterize equine cDNA for CTNNB1 and PECAM1 and to document that the genes are expressed over the different phases of wound repair in horses. Our findings suggest that β-catenin and PECAM1 might interact to modulate endothelial cell and fibroblast proliferation during wound repair in the horse.

Most previous studies investigating the roles of β-catenin and PECAM1 in wound healing have used in vitro techniques. Since wound repair is a complex process involving the interplay of several cell types, signaling pathways, extracellular matrix components and soluble factors, the role of various factors and their interactions may best be evaluated using an in vivo approach. Thus, while data interpretation can be challenging, the value of our study lies in the fact that the findings are more representative of what truly occurs in the patient.

In an effort to develop targeted therapies to prevent the formation of EGT leading to excessive scarring of equine limb wounds, future studies should quantitatively verify the temporal protein expression of β-catenin and PECAM1 and attempt to elucidate how the two molecules interact. The clues gained from studying the equine model may eventually lead to new insights into how to prevent unwanted scar formation in humans.

VM, ZI and JLL were responsible for generating the majority of the data, including analysis of the wound healing experiments: ZI was responsible for the cloning; VM and JLL contributed to the cloning and carried out the immunoblotting and the immunohistochemistry; VM helped to draft the manuscript.

JGL and CLT were the senior investigators who conceived the study and participated in its design and coordination; CLT drafted the manuscript and JGL helped in the drafting.

All authors read and approved the final manuscript.