The Matricellular Protein CCN1 Mediates Neutrophil Efferocytosis in Cutaneous Wound Healing

Neutrophil infiltration constitutes the first step in wound healing, although their timely clearance by macrophage engulfment, or efferocytosis, is critical for efficient tissue repair. However, the specific mechanism for neutrophil clearance in wound healing remains undefined. Here we uncover a key role for CCN1 in neutrophil efferocytosis by acting as a bridging molecule that binds phosphatidylserine, the “eat-me” signal on apoptotic cells, and integrins αvβ3/αvβ5 in macrophages to trigger efferocytosis. Both knockin mice expressing a mutant CCN1 that is unable to bind αvβ3/αvβ5 and mice with Ccn1 knockdown are defective in neutrophil efferocytosis, resulting in exuberant neutrophil accumulation and delayed healing. Treatment of wounds with CCN1 accelerates neutrophil clearance in both Ccn1 knockin mice and diabetic Leprdb/db mice, which suffer from neutrophil persistence and impaired healing. These findings establish CCN1 as a critical opsonin in skin injury and suggest a therapeutic potential for CCN1 in certain types of non-healing wounds.


INTRODUCTION
Wound healing encompasses a highly coordinated series of events requiring the interaction of various cell types to achieve tissue integrity and homeostasis after injury. To this end, wound repair proceeds in three overlapping but functionally distinct phases: initiating with an inflammatory phase marked by infiltration of neutrophils and macrophages, followed by a proliferative phase that includes tissue formation and extracellular matrix deposition, and concluding with a maturation phase that brings about matrix remodeling and resolution of Users may view, print, copy, and download text and data-mine the content in such documents, for the purposes of academic research, subject always to the full Conditions of use:http://www.nature.com/authors/editorial_policies/license.html#terms 2 with distinct integrins in a cell type-specific manner 26,27 . CCN1 is organized into four conserved domains with sequence similarities to insulin-like growth factor binding proteins (IGFBP), von Willebrand factor type C repeat (vWC), thrombospondin type 1 repeat (TSR), and a cysteine-knot in the carboxyl-terminus (CT). Specific integrin binding sites have been identified in the vWC, TSR, and CT domains 26,27 . We have previously shown that CCN1 functions to dampen and resolve fibrosis in wound healing by triggering cellular senescence in myofibroblasts through engagement of integrin α 6 β 1 via its CT domain during the tissue maturation phase 28,29 . Here we show that surprisingly, CCN1 is also indispensable for the clearance of neutrophils, thus serving a distinct function in the early inflammatory phase of wound healing. Mechanistically, it acts as a bridging molecule by binding PS on apoptotic neutrophils through its TSR domain and to integrins α v β 3 /α v β 5 on macrophages through its vWC domain, thereby activating Rac1 in macrophages to trigger efferocytosis. Application of CCN1 protein on slow-healing wounds with persistent neutrophil accumulation, including wounds of diabetic mice, accelerates neutrophil clearance. These findings reveal CCN1 as the key opsonin for neutrophil efferocytosis in cutaneous wound healing, and suggest a potential therapeutic role for CCN1 in certain types of slow healing wounds.

Ccn1 knockdown impedes cutaneous wound healing
Our investigation on the role of Ccn1 in cutaneous wound healing uncovered a biphasic pattern of expression in mRNA and protein, with an early peak in the inflammatory phase (day 3) and a late peak in the maturation phase (days 7-9), suggesting distinct functions in these phases (Fig. 1a,b). To evaluate this possibility, we applied antisense oligonucleotides (AS-ODN) to excisional cutaneous wounds to knockdown Ccn1 expression (Fig. 1c). When wounds were treated with daily administration of Ccn1 AS-ODN from days 3-7 to target the inflammation and tissue formation phases of wound healing, wound closure was delayed with associated robust inflammatory cell accumulation and poor granulation tissue formation (Fig. 1d,e). Accumulation of inflammatory cells and impaired granulation tissue formation can be seen when AS-ODN was applied from days 3-5 to target early Ccn1 expression specifically, but not when applied from days 7-8 to target late expression in the maturation phase ( Supplementary Fig. S1), during which CCN1 was previously shown to trigger myofibroblast senescence to restrict fibrosis 28 . These findings indicate that CCN1 may play distinct roles in the early and late phases of wound healing, corresponding to its biphasic expression pattern.
In sense-ODN-treated control wounds, resolution of inflammation, formation of granulation tissue, and re-epithelialization were observed within a 9-day period ( Fig. 1d-f, 1i). By contrast, immunohistochemical and double immunofluorescence staining of AS-ODNtreated wounds for neutrophils and macrophages with anti-Ly6G and CD68 antibodies, respectively, revealed an aberrant and exuberant accumulation of neutrophils ( Fig. 1f and Supplementary Fig. S2). Myeloperoxidase (MPO) activity was >4-fold higher in AS-ODNtreated wounds than sense-ODN-treated controls, confirming a large accumulation of neutrophils (Fig. 1g). Consistently, the pro-inflammatory cytokines Il-1β and Tnfα were expressed at 7 and 2.3-fold higher levels in AS-ODN-treated wounds, respectively (Fig. 1h). Neutrophils and macrophages are major cell types expressing CCN1 in the wounds during the inflammatory phase as judged by colocalization of CCN1 with these immune cells, and AS-ODN effectively eliminated CCN1 protein expression in these cells ( Supplementary Fig.  S3). Consistent with accumulation of neutrophils and the short life span of these cells, enhanced accumulation of apoptotic cells was observed in AS-ODN-treated wounds ( Supplementary Fig. S4). Moreover, AS-ODN-treated wounds exhibited impaired reepithelialization as shown by immunostaining of the stratified epithelial cell marker cytokeratin 14 (CK14) (Fig. 1i). However, CCN1 does not have a direct effect on keratinocyte migration ( Supplementary Fig. S5), suggesting that impaired reepithelialization is secondary to other defects in these wounds. Together, these results indicate that Ccn1 knockdown leads to neutrophil persistence and delayed wound healing, suggesting that wounds deficient in CCN1 function are stalled at the inflammatory phase.
Impaired skin wound healing in Ccn1 D125A/D125A knockin mice CCN1 exerts diverse effects on cellular functions primarily through direct binding to distinct integrins in a cell type-specific manner 26 . To dissect CCN1 functions through distinct integrin receptor pathways, we examined knockin mice expressing mutant CCN1 unable to bind specific integrins. For example, Ccn1 DM/DM mice carry mutations in the overlapping binding sites for integrin α 6 β 1 , the major CCN1 receptor in fibroblastic cells 30 , and integrin α M β 2 , the adhesion receptor for monocytes 31 . However, Ccn1 DM/DM knockin mice did not exhibit any observable inflammatory phenotype, but showed exacerbated fibrosis due to the inability to activate α 6 β 1 in fibroblasts 28 . Although lacking an RGD sequence, CCN1 binds integrins α v β 3 and α v β 5 through a non-canonical site in its vWC domain 32 . To assess the role of CCN1 mediated through α v integrins, we constructed knockin mice in which the Ccn1 genomic locus was replaced by Ccn1 D125A (Fig. 2a), a mutant allele encoding a single amino acid substitution (Asp-125 to Ala) that abolished CCN1 binding to α v integrins 32 . Knockin of the mutant allele was confirmed by standard molecular methods including direct sequencing, and illustrated by the presence of the predicted PCR products (Fig. 2b). Whereas Ccn1-null mice are embryonic lethal due to cardiovascular defects 33,34 , Ccn1 D125A/D125A mice are viable, fertile, and exhibit no apparent abnormalities, indicating the CCN1-D125A mutant protein is biologically active and does not impair essential functions during development.
Expression of mutant Ccn1 is elevated in cutaneous wounds in Ccn1 D125A/D125A mice, similar to wild type (WT) mice (Fig. 2c). Ccn1 D125A/D125A mice experienced significantly slower cutaneous wound closure than WT mice, with poor granulation tissue formation and aberrant accumulation of inflammatory cells similar to the effects of Ccn1 knockdown (Fig.  2d,e). Immunostaining for Ly6G and CD68 and double immunofluorescence showed that neutrophils were accumulated in wounds of Ccn1 D125A/D125A mice (Fig. 2f), and this result was confirmed by MPO assay (Fig. 2g). Moreover, immunostaining for CK14 and vimentin indicated that re-epithelialization and granulation tissue formation, respectively, were impaired ( Fig. 2f,g). positive endothelial cells and blood vessels, fewer vessel-associated mural cells or pericytes as judged by staining for NG2 and desmin, as well as diminished expression of angiogenic growth factors ( Supplementary Fig. S6). Since the α 6 β 1 binding site required for CCN1 induction of myofibroblast senescence was not affected by the D125A mutation 28 , senescence occurred successfully in wounds of Ccn1 D125A/D125A mice ( Supplementary Fig.  S7a). However, collagen deposition and tissue maturation was delayed, peaking at day 11 instead of day 9 in WT, consistent with delayed healing in these mice ( Supplementary Fig.  S7b). The phenotypes of Ccn1 D125A/D125A mice closely resemble the effects of Ccn1 knockdown, suggesting that critical CCN1 functions in the inflammatory phase of wound healing are mediated through α v integrins.

Neutrophil depletion accelerates repair in Ccn1 knockin mice
Consistent with increased neutrophil accumulation in Ccn1 D125A/D125A wounds, elevated expression of the pro-inflammatory cytokines Tnfα and Il-1β was observed (Fig. 3a,b). Expression of the neutrophil chemoattractants Cxcl1 (KC, GROα) and Cxcl5 (LIX), as well as their receptor Cxcr2, was elevated and more sustained in Ccn1 D125A/D125A wounds ( Fig.  3c-e), suggesting that there may be enhanced recruitment of neutrophils to the wound site. Since macrophages in wounds are mostly derived from monocytes recruited from circulation, the elevated and sustained expression of Ccl2 encoding the monocyte chemotactic protein-1 (MCP-1) was also consistent with increases in macrophages (Fig. 3f). These cytokine expression patterns may be due to aberrant recruitment of, or gene expression by, inflammatory cells, or alternatively, impaired neutrophil apoptosis or clearance 6 . However, there was no difference in the number of neutrophils in the peripheral blood of WT and Ccn1 D125A/D125A mice (Fig. 4a), suggesting no disparity in the systemic level of neutrophil production and mobilization. The expression of granulocyte colony stimulating factor (Gcsf) in the wounds was also similar ( Supplementary Fig. S8). Moreover, CCN1 protein had no effect on apoptosis of bone marrow-derived neutrophils ( Supplementary Fig. S9), and there was no defect in Ccn1 D125A/D125A macrophage gene expression when exposed to M1 or M2-polarizing agents ( Supplementary Fig. S10). Together, these results indicate that neutrophil accumulation in Ccn1 D125A/D125A mice is not likely due to aberrant production, mobilization, or apoptosis of neutrophils, nor deranged gene expression in macrophages.
It was observed that deletion of macrophages in the inflammatory phase of wound healing led to vascularization defect, reduced granulation tissue formation and impaired reepithelialization, similar to aspects of the Ccn1 D125A/D125A phenotype 37 . Since deletion of macrophages may lead to impaired neutrophil clearance, we tested the hypothesis that neutrophil persistence may be the cause of wound healing defects in Ccn1 D125A/D125A mice by depleting neutrophils using a well-characterized monoclonal antibody (mAb) against Ly6G (RB6-8C5) 38 . Previous studies showed that neutrophil depletion during wound healing either had no effect or accelerated re-epithelialization 39,40 . A single intraperitoneal (i.p.) injection of anti-Ly6G mAb 1 day before wounding decreased neutrophils from ~15-20% of blood cells to <4% by day 2 post-wounding in both WT and Ccn1 D125A/D125A mice, thereafter neutrophils recovered to the normal level by day 8 (Fig. 4a) 40 . In WT mice, wound closure was unaffected by anti-Ly6G mAb treatment up to day 7 but was slightly retarded by day 9, although the difference did not reach statistical significance (p>0.05) (Fig. 4b). By contrast, neutrophil depletion accelerated wound closure in Ccn1 D125A/D125A mice to essentially the same rate as similarly treated WT mice (*p<0.001; **p<0.0001) (Fig. 4b).
Anti-Ly6G mAb-treated Ccn1 D125A/D125A wounds showed much reduced neutrophil accumulation and improved granulation tissue formation as judged by staining with H&E and immunostaining for Ly6G and CD68 (Fig. 4c,d). MPO assay also confirmed reduction of neutrophils in day 7 wounds (Fig. 4e). Moreover, neutrophil depletion in Ccn1 D125A/D125A mice resulted in greatly improved re-epithelialization, as evidenced by CK14 staining (Fig. 4f). Concomitantly, substantial reduction in expression of the monocyte chemokine Ccl2 as well as the neutrophil chemokines Cxcl1 and Cxcl5 and their receptor Cxcr2 was also observed (Fig. 4g). Thus, neutrophil depletion accelerated wound closure and rectified many of the wound healing-associated defects in Ccn1 D125A/D125A mice, indicating that neutrophil persistence is the underlying cause of elevated chemokine expression, impaired granulation tissue formation, and delayed wound healing in Ccn1 D125A/D125A mice.

CCN1 mediates neutrophil efferocytosis through α v integrins
Since integrin α v β 3 is known to function as an efferocytosis receptor on macrophages 21 , we hypothesized that CCN1 might regulate phagocytosis of apoptotic neutrophils by macrophages through α v integrins. To evaluate this possibility, we harvested neutrophils and macrophages from bone marrow of WT and Ccn1 D125A/D125A mice (Supplementary Figs. S9,S10). When WT macrophages (labeled with CellTracker™ Green CMFDA) were incubated with aged WT neutrophils (labeled blue with DAPI), efferocytosis was evident with multiple neutrophils detected inside macrophages (Fig. 5a). However, WT neutrophils were inefficiently engulfed by macrophages from Ccn1 D125A/D125A mice, indicating a role for CCN1 in efferocytosis (Fig. 5a). Efferocytosis was substantially diminished when either neutrophils or macrophages were from Ccn1 D125A/D125A mice, and further reduced when both neutrophils and macrophages were from these mice. These results indicate that CCN1 functions to regulate both cell types in efferocytosis in a non-cell-autonomous manner (Fig.  5b).
Next, we examined the effect of exogenously added CCN1 protein on efferocytosis (Fig.  5c). WT CCN1 protein was able to restore efferocytosis in Ccn1 D125A/D125A macrophages to WT levels, showing that CCN1 can directly promote efferocytosis. However, the CCN1-D125A mutant protein inhibited efferocytosis in WT macrophages (Fig. 5c), suggesting that it might be able to act in a dominant negative manner. Indeed, addition of an increasing amount of CCN1-D125A protein progressively inhibited the stimulatory effect of WT CCN1 on efferocytosis, showing dominant negative action (Fig. 5d). This observation may help to explain the reduced efferocytsis observed when either neutrophils or macrophages were isolated from Ccn1 D125A/D125A mice (Fig. 5b), since the mutant CCN1 protein inhibits WT CCN1 function.
Blocking α v integrins with known antagonists such as Echistatin or RGD peptide inhibited efferocytosis, whereas the RGE control peptide had no effect (Fig. 5e). Inhibition by RGD peptide was observed even in the presence of exogenously added CCN1 protein, consistent with α v integrins acting downstream of CCN1 (Fig. 5f). Furthermore, macrophages pretreated with function-blocking mAbs against integrins α v , β 3 , β 5 , and α v β 3 abrogated CCN1stimulated efferocytosis, whereas IgG control or anti-β 1 mAb had no effect (Fig. 5g). Together with the loss of efferocytosis activity in the D125A mutant, these results establish that CCN1 acts through integrins α v β 3 and α v β 5 .
Phagocytosis requires cytoskeletal reorganization and movement to form phagosomes, a process dependent on activation of the small GTPase Rac1 24 . Activation of α v β 5 is thought to trigger the recruitment of the p130 Cas /CrkII/Dock180 complex, which activates Rac1 and promotes efferocytosis 41 . We found that CCN1 stimulated the formation of the p130 Cas / CrkII complex in macrophages, as shown by reciprocal co-immunoprecipitation ( Fig. 6a,b). Tyrosine phosphorylation of the focal adhesion kinase (FAK) was also dependent upon the formation this p130 Cas /CrkII signaling complex 42 , which occurred in a CCN1-dependent manner (Fig. 6c). Furthermore, we demonstrate that CCN1 induced the activation of Rac1 in bone marrow-derived macrophages (Fig. 6d). CCN1-induced formation of the p130 Cas /CrkII signaling complex, FAK phosphorylation, and activation of Rac1 are each dependent on CCN1 interaction with α v β 3 /α v β 5 , as they were abolished in the D125A mutant ( Fig. 6).

CCN1 functions as a bridging molecule
As phagocytic receptors on macrophages, integrins α v β 3 and α v β 5 do not directly recognize the so called "eat-me" signals, such as PS, on apoptotic cells. Instead, a molecule that can bind both PS and the integrins is required to bridge the apoptotic cells to α v β 3 /α v β 5 on macrophages 10 . To test whether CCN1 may act as a bridging molecule, we first examined its ability to bind PS directly using solid-phase ELISA. We found that PS efficiently bound to CCN1-WT protein as well as to the CCN1-D125A 32 (unable to bind α v β 3 /α v β 5 ) and CCN1-DM 30 (unable to bind α 6 β 1 /α M β 2 ) with half maximal binding occurring at ~20-25 nM CCN1, but not to laminin (Fig. 7a). Although laminin was previously shown to bind to glycolipids and phospholipids, it required much higher concentration (>10 µg) of lipids for binding 43 . To confirm that CCN1 binding to PS mediates its function in efferocytosis, PScontaining liposomes were used as competitive inhibitors (Fig. 7b). Although exogenously added CCN1 protein enhanced efferocytosis, its effect was strongly inhibited when CCN1 was pre-loaded with PS-liposomes, indicating CCN1 binding to PS is critical for its bridging function.
CCN1 contains a thrombospondin type I repeat (TSR) in its third domain (Fig. 7c). As the TSR motif in BAI-1 can bind PS 44 , we tested whether CCN1 binding to PS is mediated through its TSR. For this purpose, several deletion mutants were constructed (Fig. 7c) and the binding of purified deletion mutant proteins to PS was tested using membranes on which various phospholipids were spotted (Fig. 7d). Bound CCN1 was detected using anti-CCN1 vWC domain antibodies, which recognized all deletion mutants equally (Fig. 7c). Consistent with the solid-phase binding assay above, CCN1 showed strong binding to PS, but not to other negatively charged lipids. It also bound to the mitochondrial inner membrane lipid cardiolipin and sulfatide, as observed for the TSR of BAI-1 44 . Both CCN1-ΔCT and D125A-ΔCT proteins were also capable of binding to PS, cardiolipin, and sulfatide, indicating that the C-terminal domain (CT) of CCN1 is dispensable for PS binding. These deletion mutants also bound phosphatidic acid (PA), which was not recognized by fulllength CCN1, suggesting that removal of the CT domain may reveal a cryptic binding site for PA in TSR. By contrast, CCN1-I.II, which is lacking both the TSR and CT domains, did not bind to any lipid on the strip, indicating that CCN1 binding to PS requires the TSR domain (Fig. 7d). Since the mutant protein CCN1-I.II does not bind PS and the D125A-ΔCT does not bind integrins α v β 3 /α v β 5 , these mutant proteins did not stimulate neutrophil efferocytosis (Fig. 7e). By contrast, CCN1-WT and CCN1-ΔCT, being able to bind both PS and integrins α v β 3 /α v β 5 , enhanced neutrophils efferocytosis by macrophages (Fig. 7e). Together, these results show that CCN1 is a novel opsonin molecule for efferocytosis by bridging apoptotic neutrophils to macrophages through direct interaction with PS and integrins α v β 3 /α v β 5 , leading to the activation of Rac1 and phagocytosis.

CCN1 accelerates neutrophil clearance in slow-healing wounds
In view of the activity of CCN1 in efferocytosis, we tested its ability to accelerate neutrophil clearance in slow-healing wounds. A single dose of recombinant WT CCN1 protein or vehicle (PBS) was applied to wounds of Ccn1 D125A/D125A mice at day 6 post-wounding and wounds were analyzed one day later (Fig. 8a). H&E staining showed high neutrophil accumulation in PBS-treated Ccn1 D125A/D125A mice, whereas the number of neutrophils significantly decreased in CCN1-treated mice with proportionately more associated macrophages (Fig. 8a). MPO assays confirmed that treatment of wounds with WT CCN1 significantly reduced neutrophils in Ccn1 D125A/D125A mice, whereas treatment with CCN1-D125A protein had no effect (Fig. 8b). Moreover, expression of the pro-inflammatory cytokine Il-1β was reduced upon treatment with CCN1 protein in Ccn1 D125A/D125A mice, consistent with curtailed inflammation (Fig. 8c).
Chronic non-healing or slow-healing wounds, including diabetic wounds, are often associated with prolonged or exacerbated inflammation, and excessive neutrophils have been reported in chronic pressure and venous ulcers 7,9,45 . Delayed wound healing with prolonged inflammation and neutrophil persistence is also reported in wounds of Lepr db/db mice, which carry an inactivating mutation in the leptin receptor gene and develop obesity and diabetes 8 . To investigate the activity of CCN1 in a diabetic context, excisional wounds of Lepr db/db mice were treated with daily administration of CCN1 protein from days 5 to 8 and analyzed on day 9 (Fig. 8d,e). H&E staining showed inflammatory cell accumulation in PBS-treated control wounds, but not in CCN1-treated wounds (Fig. 8d). Immunostaining for Ly6G confirmed that the robust neutrophil accumulation was eliminated by CCN1 treatment in Lepr db/db wounds (Fig. 8d). Correspondingly, CCN1 treatment results in a dramatic reduction (>5 fold) in MPO activity in Lepr db/db wounds to the level similar to that of unwounded skin (Fig. 8e). Next, we assessed whether defects in neutrophil clearance in Lepr db/db wounds might be related to impaired Ccn1 expression. In contrast to significant increases in Ccn1 mRNA and protein in WT wounds (Fig. 1a,b), Ccn1 mRNA in Lepr db/db wounds showed only marginal increases upon wounding that did not reach statistical significance (Fig. 8f). Moreover, the total CCN1 protein actually decreased upon wounding in Lepr db/db mice, possibly due to a protease-rich environment in which neutrophils are aberrantly accumulated (Fig. 8g). Thus, there is a deficiency in Ccn1 expression in Lepr db/db wounds, leading to accumulation of neutrophils. These results underscore the importance of CCN1 in promoting neutrophil efferocytosis in wound healing, suggesting a therapeutic potential for CCN1 in the treatment certain types of slow-healing wounds including diabetic wounds.

DISCUSSION
Restoration of tissue integrity and homeostasis following injury to the skin is of vital importance, as the integument provides the first barrier against invading microbes and pathogens. Acute inflammation occurring immediately after injury plays a critical role for host defense and debridement of necrotic tissues, although deregulated inflammation or its persistence can cause further tissue damage and lead to chronic non-healing wounds. Impaired wound healing causes considerable morbidity in afflicted patients, and is a leading cause of diabetes-related amputations 46 . Thus, effective management of inflammation resolution is critical to wound care, and targeting excessive neutrophil activity has been proposed for treating chronic wounds such as venous ulcers 45 . Here we have uncovered a critical role for CCN1 in triggering efferocytosis of neutrophils by macrophages to promote resolution of inflammation in skin wounds. CCN1 appears to be the first bridging molecule identified to play this role in cutaneous wound healing. Moreover, application of CCN1 accelerates neutrophil clearance in diabetic mouse wounds, suggesting a therapeutic potential for CCN1 in the treatment of certain types of impaired wound healing.
To dissect the functions of CCN1 through distinct integrins, we have constructed an allelic series of knockin mice in which Ccn1 is replaced by alleles expressing CCN1 mutants unable to bind specific integrins. This approach circumvents the embryonic lethality of Ccn1 null mice due to cardiovascular defects 34 , and avoids potential limitations in cell-type specific deletions as CCN1 may be secreted by diverse cell types in the tissue microenvironment. In skin wound healing, analysis of Ccn1 DM/DM knockin mice expressing CCN1 unable to bind α 6 β 1 /α M β 2 led to the discovery that CCN1 controls fibrosis through triggering myofibroblast senescence by engagement of α 6 β 1 28 , and studies on Ccn1 D125A/D125A mice reported herein revealed a critical function in promoting the resolution of inflammation through efferocytosis. Interestingly, fibroblast-specific deletion of Ccn1, even in combination with deletion of the homologous family member Ccn2, did not result in any change in wound closure rate 47 , suggesting that CCN1 produced from various cell types in the wound microenvironment adequately compensates for the loss of fibroblastderived CCN1. Indeed, CCN1 is expressed in macrophages 48 , neutrophils ( Supplementary  Fig. S2), endothelial cells 33 , and fibroblasts 28 . No apparent defect in macrophage function was noted in Ccn1 DM/DM knockin mice, even though CCN1 can induce gene expression in macrophages through α M β 2 in culture 49 , suggesting that inflammatory cytokines or regulators other than CCN1 may prevail in regulating macrophage gene expression in the cutaneous wound.
Multiple efferocytosis receptors and bridging molecules have been described, and they may play differential roles in various developmental and pathological contexts depending on their cell type specificity and time course of expression. For example, both Gas6 and Protein S contribute to the phagocytosis of photoreceptor outer segments by retinal pigment epithelial cells, but no other efferocytosis defect was observed among a myriad of phenotypes in Gas6, Protein S, and TAM receptor mutant mice 25,50 . Mice lacking TSP-1, a bridging molecule, exhibit impaired angiogenesis in cutaneous wound healing but efferocytosis defect has not been reported 51 . Likewise, MFG-E8-null mice show reduced angiogenesis in skin wound healing without notable efferocytosis defect 52 . Indeed, whereas MFG-E8 is critical for the elimination of apoptotic B cells in the germinal centers 53 and mammary epithelial cells during mammary gland involution 54 , MFG-E8 is downregulated in virtually all inflammatory conditions and inflammation-related pathologies 55 , suggesting that it may not participate in neutrophil elimination.
CCN1 binds to α v β 3 /α v β 5 on macrophages, PS on apoptotic neutrophils, and α 6 β 1 on myofibroblasts through its vWC, TSR, and CT domains, respectively, to exert distinct functions during wound healing (Fig. 9). Since CCN family proteins emerged through evolution by exon shuffling 26,27 , it appears likely that this process recombined various structural domains from extracellular matrix proteins to confer unique functions governing multiple aspects of wound healing. CCN1 is known to dampen and resolve fibrosis in the maturation phase of wound healing by inducing myofibroblast senescence 28 . Our current findings suggest the possibility that CCN1 might further contribute to the restriction of fibrosis by promoting efferocytosis of apoptotic myofibroblasts. In support of this hypothesis, we found that CCN1 can promote macrophage engulfment of apoptotic fibroblastic cells ( Supplementary Fig. S10c), and apoptotic myofibroblasts have been associated with restriction of fibrosis 56 .
Since CCN1 can induce angiogenesis through interaction with integrin α v β 3 in endothelial cells 35,36 , it is not surprising that Ccn1 D125A/D125A mice also exhibited impaired angiogenesis during wound healing. These knockin mice show reduced CD31-positive endothelial cells, fewer vessel-associated pericytes, and diminished expression of angiogenesis-related genes ( Supplementary Fig. S6). However, neutrophil depletion rectified many of the wound healing defects by accelerating wound closure, improving angiogenesis, and promoting granulation tissue formation (Fig. 4), suggesting that neutrophil persistence, rather than impaired angiogenesis, underlies the wound healing defects in Ccn1 D125A/D125A mice. As there are multiple angiogenic factors in the wound microenvironment, it is possible that CCN1 plays a redundant or relatively minor role in regulating angiogenesis during wound healing, and the angiogenic defect in Ccn1 D125A/D125A mice may have been exacerbated by neutrophil persistence. However, since the angiogenesis and efferocytosis functions of CCN1 are mediated through the same integrin and cannot be separated by mutation, understanding the angiogenic role of CCN1 in wound healing may require further scrutiny.
Diabetic wounds often suffer from enhanced inflammation and delayed healing, and Lepr db/db mice exhibit these phenotypes 8,40 . In contrast to WT mice in which Ccn1 expression is enhanced upon wounding (Fig. 1a,b), Ccn1 mRNA is not significantly elevated in Lepr db/db wounds and CCN1 protein actually decreased upon wounding (Fig. 8f,g), potentially due to robust protease activity produced by accumulated neutrophils 45 . Thus, Lepr db/db wounds appear to suffer from impaired Ccn1 expression, leading to a deficit in CCN1 function. Consistently, CCN1 therapy of cutaneous wounds in Lepr db/db mice efficiently eliminated excess accumulation of neutrophils (Fig. 8). Since these results are obtained in Lepr db/db mice, the specific role of CCN1 in normal and impaired human wounds remains to be determined. It is interesting to note that the CCN1-D125A mutant protein can inhibit efferocytosis in a dominant negative manner (Fig. 5d), most likely by sequestering neutrophils through PS binding but being unable to bridge them to macrophages. Thus, cleavage of CCN1 through its protease-sensitive central hinge region between domains II and III 57 in certain pathological contexts may potentially generate CCN1 fragments that inhibit efferocytosis in a dominant negative manner (Fig. 7e). For example, CCN1 is cleaved in lung epithelial cells from patients with chronic obstructive pulmonary disease (COPD) 58 , a disease in which impaired efferocytosis has been observed 15 .
Whether CCN1 plays a critical role in efferocytosis in other inflammatory or pathological contexts beyond cutaneous wound repair is currently unknown. In this regard, it is noteworthy that aberrant Ccn1 expression has been reported in myriad inflammatory conditions, including sepsis 59 , inflammatory bowel disease 60, 61 , rheumatoid arthritis 62 , COPD 63 , cholestasis 64 and cardiac injury repair 65 . It will be important to assess whether the opsonin function of CCN1 is important in these contexts. Moreover, CCN1 controls fibrosis by inducing senescence not only in skin wounds, but also in liver injuries induced by either hepatotoxin or cholestasis 29 , indicating that the role of CCN1 in tissue injury repair is potentially wide-ranging and not limited to cutaneous wounds.
Prolonged inflammation is the hallmark of non-healing wounds, including diabetic wounds, pressure ulcers, and venous ulcers 1,66 . Excessive neutrophils contributes to impaired healing, and thus wound closure in diabetic mice was accelerated by 50% upon neutrophil depletion 40 . Here we show that CCN1 treatment leads to effective neutrophils removal in slow-healing wounds, including wounds of Ccn1 D125A/D125A and in Lepr db/db diabetic mice (Fig. 8). These findings suggest a potential therapeutic value in the application of CCN1 in certain types of non-healing wounds 45 . Since CCN1 plays multiple and disparate roles in wound healing (Fig. 9), the possibility that application of CCN1 in wound care may both accelerate inflammation resolution and dampen fibrosis and reduce scarring clearly warrants further investigation.

Animals and analysis of cutaneous wound healing
Animal protocols were approved by the Institutional Animal Care and Use Committee of the University of Illinois at Chicago. Two independent lines of Ccn1 D125A/D125A mice (A5 and F5) were generated in a svJ129-C57BL/6 mixed background (Fig. 2a) and backcrossed >6 times into the C57BL/6 background; results from both lines were similar. Genetically diabetic mice (BKS.Cg-m+/+ Lepr db /J) were from Jackson Laboratories, and intercrossed to yield homozygous db/db mice. Typically, 10 to 12-week-old male mice with similar body weight (approximately 25 to 28 g for C57BL/6 mice) were used for cutaneous wound healing. Mice were anesthetized with intraperitoneal (IP) injection of ketamine/xylazine (100 mg per kg i.p.) and the dorsum of each mouse was clipped free of hair. Full-thickness excisional wounds through panniculus carnosus were created with a 6 mm biopsy punch (Miltex) and left open during the healing process 28 . For healing kinetic studies, digital images of wounds were taken and wound diameters were measured using Photoshop CS2 (Adobe). For histological analyses, wound tissues were snap-frozen in optimum cutting temperature (OCT) compound (Tissue-Tek) and sectioned serially (7 µm thickness) using a Leica CM1950 UV cryostat. Alternatively, wound tissues were embedded in paraffin and sections were stained with Hematoxylin/Eosin (H&E) or probed with anti-Ly6G rat mAb (BD Pharmingen; 1A8, 1:100 dilution) or anti-CD68 rabbit polyclonal antibody (Abcam, 1:100 dilution), followed by biotin-conjugated goat anti-rat IgG secondary antibody coupled with streptavidin-HRP with 3,3'-Diaminobenzidine (DAB; Sigma) as substrate, and counterstained with hematoxylin. Apoptosis in wounds was detected using anti-activated caspase-3 polyclonal antibody (Cell signaling) and visualized under fluorescence microscopy (Leica DM4000B). For Western blot analysis, wound tissues were homogenized in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-Cl pH 7.2, 150 mM NaCl, 0.1% SDS, 0.5% Na-deoxycholate, 1% Triton X-100, and protease inhibitors cocktail) using polytron homogenizer, followed by sonication. Cellular senescence was evaluated by senescence-associated β-galactosidase (SA-β-gal) staining of wound tissues 28 .

Myeloperoxidase assay
The samples from freshly isolated wound tissues were prepared as described 68 , and MPO assay was performed in assay buffer (50 mM hexadecyl trimethyl ammonium bromide, 0.167 mg per ml of O-dianisidine dihydrochloride and 0.0005 % hydrogen peroxide). MPO activity (absorbance) was measured using plate reader (Lab system, Multiscan MS) at 450 nm. Purified human myeloperoxidase (Sigma) was used to generate a standard curve. Data were expressed as absorbance per wound weight.

RNA isolation and qRT-PCR
Skin wounds or cultured bone marrow-derived macrophages were homogenized using TRIzol reagent (Invitrogen) and total RNA was isolated using RNeasy® Mini Kit (Quiagen). Total RNA (2 µg) were reverse transcribed to cDNA using MMLV-Reverse Transcriptase (Promega), and qRT-PCR was performed with the iCycler Thermal Cycler (Bio-Rad) using iQ SYBR Green Supermix (Bio-Rad). The specificity of qRT-PCR was confirmed by agarose gel electrophoresis and melting-curve analysis. A housekeeping gene (cyclophilin E) was used as an internal standard. The primers used are listed in Supplementary Table I.

Isolation of bone marrow-derived macrophages and neutrophils
Murine neutrophils and macrophages were isolated from bone marrow of either WT or Ccn1 D125A/D125A mice as described with modifications 69,70 . For neutrophils isolation, the resuspended bone marrow in Hank's balanced salt solution (HBSS, 5ml) was carefully layered over 62% Percoll (5 ml, Amersham) and centrifuged at 1000 xg for 30 min. After the gradient centrifugation, a cloudy layer containing neutrophils was removed from on top of the Percoll layer and resuspended in RPMI media for culture. An aliquot of cell suspension (~20 µl) was spread on to slide glass to assess the purity of isolated neutrophils by Giemsa staining (Sigma). For macrophages, the bone marrow cells were re-suspended in macrophage complete media (DMEM/F12 with 20% (v/v) L-929 cells conditioned media), counted, and 4×10 5 cells were plated in sterile plastic petri dish (100 mm) and incubated in 37°C in 5% CO 2 for 7 days with medium being refreshed on day 3 for full differentiation into macrophages. The macrophages differentiation was checked using F4/80 immunostaining.

Efferocytosis assay
Bone marrow-derived macrophages (10 5 cells per well) from either WT or Ccn1 D125A/D125A mice were plated on 12-well culture dish 1 day before the assay and labeled with CellTracker™ Green 5-chloromethylfluorescein diacetate (CMFDA; 5 µM, Invitrogen). Neutrophils were aged for 16 hrs and then incubated at 55°C for 2 hrs prior to use as substrate for efferocytosis. Aged neutrophils (5×10 5 cells per well) were labeled with DAPI (10 mg per ml) and incubated with macrophages for 60 min or 90 min. Where indicated, CCN1 was pre-incubated with neutrophils for 1h before incubation with macrophages. After incubation, cells were washed with ice-cold PBS 4 times fixed with 4% paraformaldehyde for 5 min. Efferocytosis was measured using the fluorescence microscopy by counting macrophages containing >2 neutrophils in 12 randomly-selected high-powered fields and efferocytosis index was expressed as a percentage positive macrophages over total counted (~300 cells). For inhibition of efferocytosis, PS-containing liposomes were prepared as described 22 and pre-loaded to CCN1 for 30min. For fibroblasts efferocytosis, human normal skin fibroblasts (1077SK; ATCC CRL-2094) were cultured, labeled with CellTracker™ Blue 4-chloromethyl-6,8-difluoro-7-hydroxycoumarin (CMF2HC; 5 µM, Invitrogen), and induced to undergo apoptosis by exposure to staurosporine (1 µM; Invitrogen) for 16 h. Efferocytosis assay was performed as described above.

Depletion of neutrophils in mice
Neutrophils depletion was achieved as described 38 . Briefly, anti-Ly6G mAb (hybridoma clone# RB6-8C5; originally developed by Dr. Robert I. Coffman, DNAX Research Institute, Palo Alto, CA and obtained from Dr. David Sibley, Washington University, St. Louis, MO) was produced under serum free condition using a CELLine bioreactor (CL350; Argos technologies, IL). Anti-Ly6G mAb (0.5mg per mouse) was delivered in a single i.p. injection 1 day before wounding. Neutrophils depletion was confirmed by whole blood count on day 2 post-wounding.

Cell migration assay
Human immortalized HaCaT keratinocytes were plated on to non-TC treated dishes that were pre-coated with rat tail collagen I (4.3 mg per ml; BD Bioscience) and serum-starved for 48 h. Cells were pretreated with mitomycin C (10 µg per ml; Sigma) for 30 min, followed by treatment of hEGF (10 ng per ml) or recombinant CCN1 protein (either WT or D125A mutants; 2 µg per ml each). Cell migration was assessed after 48 h and quantified using Image J software.

PS binding
Standard binding assays usually coat plastic wells with phospholipids, followed by addition of binding protein and immunodetection of the binding protein. However, CCN1 is a cell adhesion molecule and characteristically coats plastic surfaces efficiently even in a short incubation period, making it impossible to detect specific binding using this protocol. Thus, we pre-coated 96-well plates with serial dilution of CCN1 proteins (WT, DM, and D125A) and laminin at 4°C overnight. After washing with PBST (2×), uncoated area was blocked with BSA (10 mg per ml) for 1 h. A total 300 ng of PS was then added to each well for 1 h, followed by washing with PBST (3×). Anti-PS mAb (Abcam; 1µg per well) was added, and standard ELISA protocol was employed using 3,3',5,5' Tetramethyl Benzine as a substrate and measured at OD450 nm 20 . Alternatively, lipid membranes (Echelon biosciences) were incubated overnight with CCN1 proteins (WT, CCN1-N, CCN1-ΔCT, and D125A-ΔCT; 10 µg each), and anti-CCN1 vWC antibodies were used for detection.

GTP-Rac1 pull-down assay
Activated, GTP-bound Rac1 was detected using Rac1 activation assay kit (Cell Biolabs, Inc) according to manufacturer's instruction. Briefly, bone marrow-derived macrophages were treated with CCN1-WT or CCN1-D125A (4µg each) for 60 min. Cells were then washed with cold PBS twice and lysed. The cell lysates were incubated with 20 μg of PAK1-PBD (p21-binding domain of p21-activated protein kinase) agarose for 1h at 4°C. The precipitation of GTP-Rac1 was detected by immunoblotting.

Statistics
Results are expressed as the mean ± standard deviation (S.D.), and statistical analysis was performed by one-sided two-sample t-tests and one-way analysis of variance (ANOVA; for multiple group comparison) with Bonferroni post-hoc tests. A p<0.05 was considered significant. All experiments were performed in triplicate, unless otherwise indicated.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. *p<0.005, **p<0.002. Statistical analyses were conducted using one-sided two-sample ttests (wound diameter). (e) Wound tissue sections (day 9) were stained with H&E; wound edge was marked with dash lines and area in square box was shown in higher magnification on the right (bar=100 µm). (f) Adjacent wound sections (day 9) were stained with either anti-Ly6G mAb or anti-CD68 polyclonal antibody to visualize neutrophils or macrophages, respectively, and counterstained with hematoxylin. Double immunofluorescence with anti-Ly6G mAb (Green) or anti-CD68 polyclonal antibody (Red) provides contrast for neutrophils and macrophages. Bars=100 µm (g) MPO activity in wounds (day 9) treated with either sense-ODN or AS-ODN (n=4 per treatment). (h) Expression of Il-1β and Tnfα was measured by qRT-PCR at day 9, normalized to cyclophilin E (n=4 per time point per treatment). (i) Wound sections (day 9) were stained for CK-14 (red) to visualize stratified epithelial cells and counterstained with DAPI. Bar=500 µm. qRT-PCR and MPO assay were conducted in triplicates. All data are expressed as mean ± standard deviation in triplicate determinations.    Wound tissues (day 7) from mice treated with anti-Ly6G mAb or PBS were stained with H&E; area in box was enlarged on the right. Bar=100 µm. (d) Adjacent sections were stained with either anti-Ly6G or anti-CD68 and counterstained with hematoxylin (Bar=100 µm). Ly6G-positive neutrophils are greatly reduced after neutrophil depletion. (e) MPO activity was measured in wounds of mice treated with anti-Ly6G mAb or PBS (n=4 per data point). (f) Immunofluorescence staining of CK14 (red) in day 7 wounds of Ccn1 D125A/D125A mice treated with either PBS or anti-Ly6G and counterstained with DAPI (blue). Bar=500 µm. (g) Expression of chemokines (Cxcl1, Cxcl5, and Ccl2) and chemokine receptor (Cxcr2) in wounds of Ccn1 D125A/D125A mice treated with either anti-Ly6G mAb or PBS was measured by qRT-PCR and normalized to cyclophilin E (n=4 per data point). Data are expressed as mean ± standard deviation in triplicate determinations. Neutrophils and macrophages from WT or Ccn1 D125A/D125A mice were incubated in various combinations as indicated for 90 min. Reduced efferocytosis was observed when either marrow-derived macrophages (Ccn1 WT/WT ) was evaluated after treatment with BSA, CCN1-WT, or CCN1-D125A mutant proteins (4 µg per ml) for 60 min using PAK-PBD agarose (20 µg for each sample) pull-down assay. Total lysates were used to verify total Rac1 expression by immunoblotting (lower panel).

Figure 7. CCN1 functions as a bridging molecule by direct binding to PS through its TSR domain
(a) CCN1 binding to PS was assessed by solid-phase ELISA. Serial dilution of recombinant CCN1 proteins (WT, D125A, or DM mutants) or Laminin were pre-coated onto 96-well plate and blocked with BSA (10 mg per ml). After PS (300 ng per well) was added and incubated for 1 hr, bound-PS was detected by anti-PS mAb (1 µg per well). Coated proteins were plotted with logarithmic scale (n=3 per each concentration) (b) Efferocytosis assay was performed using WT neutrophils and macrophages. Where indicated, CCN1 protein alone or pre-incubated with liposomes containing phospho-L-Serine was added to the assay (n=4 per treatment). *p<0.001 in one-sided two-sample t-tests. were pre-incubated with aged neutrophils and efferocytosis was measured (n=4 per treatment). Data expressed as mean ± standard deviation in triplicate determinations.  In the inflammatory phase, CCN1 acts as a bridging molecule to stimulate neutrophil clearance by binding integrins α v β 3 /α v β 5 in macrophages through its vWC domain and PS in apoptotic cells through its TSR domain. In the maturation phase, CCN1 functions to dampen and restrict fibrosis by triggering myofibroblast senescence by binding integrin α 6 β 1 through its CT domain 28 .