A Modified Collagen Dressing Induces Transition of Inflammatory to Reparative Phenotype of Wound Macrophages

Collagen containing wound-care dressings are extensively used. However, the mechanism of action of these dressings remain unclear. Earlier studies utilizing a modified collagen gel (MCG) dressing demonstrated improved vascularization of ischemic wounds and better healing outcomes. Wound macrophages are pivotal in facilitating wound angiogenesis and timely healing. The current study was designed to investigate the effect of MCG on wound macrophage phenotype and function. MCG augmented recruitment of macrophage at the wound-site, attenuated pro-inflammatory and promoted anti-inflammatory macrophage polarization. Additionally, MCG increased anti-inflammatory IL-10, IL-4 and pro-angiogenic VEGF production, indicating a direct role of MCG in resolving wound inflammation and improving angiogenesis. At the wound-site, impairment in clearance of apoptotic cell bioburden enables chronic inflammation. Engulfment of apoptotic cells by macrophages (efferocytosis) resolves inflammation via a miR-21-PDCD4-IL-10 pathway. MCG-treated wound macrophages exhibited a significantly bolstered efferocytosis index. Such favorable outcome significantly induced miR-21 expression. MCG-mediated IL-10 production was dampened under conditions of miR-21 knockdown pointing towards miR-21 as a causative factor. Pharmacological inhibition of JNK attenuated IL-10 production by MCG, implicating miR-21-JNK pathway in MCG-mediated IL-10 production by macrophages. This work provides direct evidence demonstrating that a collagen-based wound-care dressing may influence wound macrophage function and therefore modify wound inflammation outcomes.

upon micro-environmental conditions, macrophages possess a wide range of functions which are modulated through the release of several factors 21 . Successful and timely resolution of inflammation involves switching of macrophages from an inflammatory (mφ inf ) to a reparative (mφ heal ) phenotype. In the current work, we sought to evaluate the effect of MCG on macrophage function and polarization in the context of wound inflammation.

MCG augmented recruitment of macrophage to the wound site and attenuated pro-inflammatory macrophage polarization.
To determine whether MCG treatment affect the macrophage abundance at the wound-site during inflammatory phases, poly vinyl alcohol (PVA) sponges soaked in MCG stock solution were implanted subcutaneously in mice. Wound inflammatory cells were harvested from PVA sponges on day 3 and day 7 post-wounding (PW), stained with FITC conjugated F4/80, a murine macrophage marker and analyzed using flow cytometry. MCG treated wound cells displayed significantly higher abundance of F4/80 + macrophages as compared to those from untreated wounds at both time points (Fig. 1). To determine whether MCG played a role in macrophage polarization, wound macrophages isolated from PVA sponges (CD11b + ) were stained with F4/80-FITC and PE-conjugated mφ inf /mφ heal surface markers. The double positive cells were analyzed for mφ inf surface markers CD40, CD11c, CD16/32 CD284, and mφ heal surface markers CD206 and CD23 ( Fig. 2A). Significantly lower expression of all mφ inf surface markers and higher expression of mφ heal surface markers were observed in mφ exposed to MCG on day 3 and day 7 post-implantation respectively, indicating a shift in wound macrophage polarization to an anti-inflammatory, reparative mφ heal phenotype in response to MCG ( Fig. 2 and Supplementary Fig. S1). Upregulation of anti-inflammatory IL-10 and VEGF in MCG-treated wound cells and cultured macrophages. Anti-inflammatory mφ heal macrophages produce copious amount of Interleukin-10 (IL-10), Interleukin-4 (IL-4) and Vascular Endothelial Growth Factor (VEGF) which helps in resolution of inflammation [22][23][24] and promotes angiogenesis [25][26][27] . To determine whether MCG promoted an anti-inflammatory milieu at the wound-site, these anti-inflammatory cytokines were quantified from conditioned media of wound inflammatory cells derived from MCG-treated wounds. IL-10 protein was strongly upregulated in MCG-treated wound inflammatory cells (Fig. 3A). When studied in individual wound cell population, MCG was observed to induce IL-10 ( Fig. 3B,C) and pro-angiogenic VEGF (Fig. 3D) in wound macrophages. To test a direct effect of MCG on macrophage IL-10, IL-4 and VEGF production, wound macrophages and differentiated THP-1 derived macrophages were utilized. Measurement of protein by ELISA demonstrated significant induction of IL-10, IL-4 and VEGF protein following treatment with MCG in both wound macrophages and differentiated THP-1 cells MCG induced the efferocytosis-JNK-miR-21 pathway. We have reported that engulfment of apoptotic cells by macrophages (aka, efferocytosis) is a signaling cue that drives polarization to anti-inflammatory mφ heal phenotype via miR-21-programmed cell death 4 (PDCD4)-IL-10 pathway 18 . Thus, the effect of MCG treatment on macrophage efferocytosis activity was determined. A significantly elevated efferocytosis index was noted in macrophages treated with MCG as compared to matched untreated controls (Fig. 5A,B). Successful efferocytosis is known to induce miR-21 expression, which via phosphatase and tensin homolog (PTEN) and PDCD4 silencing, switches macrophage to an anti-inflammatory mφ heal phenotype 18 . In this work, MCG-induced efferocytosis was associated with elevated miR-21 expression (Fig. 5C). Interestingly, MCG-induced IL-10 expression was blunted under conditions of miR-21 knockdown (Fig. 5D, Supplementary www.nature.com/scientificreports www.nature.com/scientificreports/ Fig. S3A). This line of evidence recognizes miR-21 as a mechanism implicated in MCG-induced IL-10 production by macrophages. We have previously reported that pharmacological inhibition of c-Jun N-terminal kinase (JNK) or knockdown of cellular c-Jun resulted in significant downregulation of inducible IL-10 protein expression, demonstrating a direct role of c-Jun and JNK in LPS-induced IL-10 expression in human monocyte-derived macrophages 18 . The JNK inhibitor (420119 JNK Inhibitor II) significantly inhibited MCG-induced IL-10 production (Fig. 5E). To further determine if MCG → miR-21 → IL-10 induction is via JNK pathway, THP-1 cells were transfected with miRIDIAN hsa-miR-21 mimic to increase cellular miR-21 abundance ( Supplementary Fig. S3B) followed by knockdown of c-Jun using siRNA( Supplementary Fig. S3C) and treatment with MCG. Knocking down c-Jun under these conditions resulted in abrogation of MCG-induced IL-10 even in high miR-21 conditions suggesting a central role of cJun-JNK pathway in MCG → miR21 induced IL-10 production (Fig. 5F). Finally, a summary of the proposed pathway implicated in anti-inflammatory effect of MCG via IL-10 production has been presented (Fig. 6).

Discussion
The microenvironment of tissue injury is characterized by the infiltration of visiting immune cells in a situation where collagen and its fragmented products are in abundance. Collagen and its degradation products have been recognized to induce signal transduction which in turn modulates several physiological functions like cell adhesion and migration, hemostasis and immune function 28,29 . Interestingly, a clostridial collagenase ointment Wound inflammatory cells on d3 were harvested from MCG treated PVA sponges subcutaneously implanted in C57BL/6 mice. The wound inflammatory cells were harvested from the sponges, and subjected to ELISA for (A) IL-10 protein expression analysis. (B) d3 wound macrophages (CD11b + ) were harvested from MCG treated PVA sponges subcutaneously implanted in C57BL/6 mice and subjected to ELISA for IL-10 protein expression. (C,D) d7 wound macrophages (CD11b + ) were harvested from MCG treated PVA sponges subcutaneously implanted in C57BL/6 mice and subjected to ELISA for (C) IL-10 and (D) VEGF protein expression. Data are mean ± SEM (n = 5-6); *p < 0.05 compared to wound cells/macrophages harvested from untreated PVA sponges.  30 . Collagen degraded to its peptide components is readily phagocytosed by macrophages at the wound-site 31,32 . Whether such engulfment of collagen peptides induces any cellular signaling in macrophages remains unknown. This report provides maiden evidence demonstrating that a modified collagen based wound dressing (MCG), generated by tryptic digestion and composed of short and long chain peptides of collagen, induces mφ heal -like polarization in wound macrophages, including production of copious amounts of anti-inflammatory and pro-angiogenic response by these cells.
Circulating monocytes recruited to the wound tissue differentiate to macrophages that are critical in orchestrating the inflammatory and subsequent repair process at the wound-site [33][34][35] . Consistent with findings of the current study, an increased macrophage infiltration in excisional wounds treated with modified collagen gel in a porcine model was also noted suggesting that MCG possess a macrophage chemoattractant property 16 . LC-MS/ MS studies from our laboratory have demonstrated that MCG is composed of long and short chain peptides derived from collagen 16 . Synthetic polypeptides such as pentameric (Pro-Pro-Gly) 5 form as well as peptides of bovine collagen digested by collagenase are known to display potent chemoattractant activity for rat alveolar macrophages 36 and human monocytes 37 . The exact mechanism of collagen peptide mediated macrophage chemoattractant function remains unclear. It is plausible that collagen peptides promote production of Monocyte Chemoattractant Protein-1 (MCP-1), a potent macrophage chemo-attractant, thereby increasing the macrophage infiltration.
Once extravasated, macrophage phenotype and function is guided by environmental cues 18,38 . At the wound-site, macrophages are known to serve functionally distinct roles including the classical (proinflammatory, mφ inf ) and alternative (anti-inflammatory, prohealing, mφ heal ) activation states 35,[38][39][40] . While the pro-inflammatory mφ inf macrophages are responsible for the clearing of infectious agents, the mφ heal macrophages are more reparative in nature and enable timely resolution of inflammation and promote angiogenesis 15,34,41 . Chronic diabetic ulcers with unresolved inflammation display aberrant mφ inf :mφ heal macrophage ratio and an imbalance between pro-and anti-inflammatory environment 42,43 . CD40, CD16-32, CD11c and CD284 (TLR4) are well-established markers of mφ inf macrophage polarization 34,39,40,44,45 . Functional wound macrophages treated with MCG in vivo displayed a decrease in mφ inf macrophage polarization at the inflammatory phases indicative of a shift in the wound macrophage polarization from mφ inf to mφ heal . IL-10, also known as human cytokine synthesis inhibitory factor (CSIF) 46 , is a cytokine with anti-inflammatory properties 47-49 while VEGF is a potent angiogenic factor 50 . MCG-induced shift in the phenotype of the wound macrophages was coupled with induction of IL-10 and VEGF. These findings are consistent with increased IL-10, Mannose Receptor C-Type 1 (Mrc-1) and C-C Motif Chemokine Receptor 2 (CCR2) expression in MCG-treated wounds as previously reported in a porcine model 15 . Following MCG treatment, increased mφ heal macrophage polarization was associated with an increased wound angiogenesis 15,16 . Given that anti-inflammatory tissue mφ have been directly implicated in angiogenesis 27,51 , it is plausible that the MCG-induced mφ heal polarization of macrophages promoted wound angiogenesis 15 .
Mechanism of macrophage polarization includes complex interplay of multiple signaling pathways and transcription factors 52 . This work identified miR-21 as a major driver of MCG-induced macrophage polarization. Although sample size was modest, the findings were robust primarily because of consistent effect. An overload of apoptotic cells at the wound-site causes inflammation to persist 53 . We have recently underscored a major role of efferocytosis and microRNA-21 (miR-21) in macrophage transition from mφ inf to an anti-inflammatory mφ heal phenotype featuring increased IL-10 18 . Efferocytosis or successful engulfment of apoptotic cells is known to promote an anti-inflammatory response in macrophages 54,55 including induction of miR-21 expression 18 . An impairment of efferocytosis in diabetic wounds led to unresolved inflammation 19,53 . miR-21 promoted anti-inflammatory mφ heal like response in human macrophages by directly targeting phosphatase and tensin homolog (PTEN) and programmed cell death protein 4 (PDCD4) that subsequently inhibited NF-κB → TNF-α or promoted JNK → AP-1 → IL-10 production 18 . Blocking of JNK resulted in an attenuation of MCG-induced IL-10 production suggesting that the anti-inflammatory effects of MCG involves miR-21 targeting PDCD4 followed by activation of JNK → AP-1 → IL-10 pathway 18 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Collagen based wound dressings have been widely used in effective treatment of chronic wounds 11,56 . Recent studies from our laboratory have reported improved wound macrophage function and epithelialization using another stabilized collagen matrix dressing 12 . The current understanding of the mechanisms of action of these dressings include (i) serving as a substrate for high matrix metalloproteinase (MMP) in chronic wound environment; (ii) the chemotactic property of the collagen breakdown products for cells critical in formation of granulation tissue and (iii) exudate management as a result of the high absorptive property. This work identified and characterized a novel mechanism of action of collagen based wound dressings in modifying wound macrophage inflammatory response. MCG promoted an anti-inflammatory proangiogenic mφ heal -like macrophage phenotype via miR-21-cJun/JNK mediated signaling pathway. Results of this study were obtained from the study of macrophages that were activated in vivo employing a standardized PVA sponge implantation model 19,30,53,57 . The reported observations provide a valuable paradigm that is ready to be tested on macrophages directly isolated from the chronic wound patients. The findings of this work add a new dimension to our understanding of macrophage-ECM interactions. It provides firm mechanistic explanation addressing the significance of collagen based wound-care dressings.

Materials and Methods
Polyvinyl alcohol (PVA) sponge implantation model. The animal studies were approved by, and all methods were performed in accordance with the relevant guidelines and regulations set by The Ohio State University's Institutional Animal Care and Use Committee and Indiana University's Institutional Animal Care and Use Committee. Circular sterile PVA sponges (8 mm diameter) were subcutaneously implanted on the back of 8-12-week old C57BL/6 mice under anesthesia induced by isoflurane inhalation as previously described 19,30,53,57 . PVA sponges were soaked in MCG stock solution (2.5 g/ml) for overnight (16 h). Soaking PVA sponges (each sponge of volume ~100 mm 3 ) resulted in each sponge getting coated with ~125 mg of MCG solution. The weight of sponges post-soaking was measured to ensure equal loading of the MCG solution to each sponge. PVA sponges containing either MCG or saline (control) were then subcutaneously inserted into each animal. MCG was obtained as Stimulen TM gel from Southwest Technologies Inc. (North Kansas City, MO) 15,16 . Harvesting of the PVA sponges were done on day 3/day 7 post-implantation following euthanasia. All sponges were removed and placed in sterile saline. Repeated compression of the sponges in saline resulted in a wound cell suspension which was then filtered with a 70 µm nylon cell strainer to eliminate all debris, followed by hypotonic lysis with ice cold deionized water to remove the red blood cells 19,53,57 . Wound macrophages (CD11b + ) were obtained from day 3/ day 7 wound cell infiltrate by magnetic bead based sorting as previously described 19,53 . Immunostaining and flow cytometry. Markers used to determine monocyte and/or macrophage subsets comprised: FITC-F4/80 (Serotec) and PE-CD16/32, PE-CD11c, PE-CD40, PE-CD284, PE-CD206, PE-CD23(eBioscience). The cells were stained with surface markers and recorded using BD Accuri flow cytometry (BD Biosciences) and analyzed as previously described 17,30,58,59 . Cell culture, differentiation, and treatment. The human THP-1 monocytic cell-line was cultured and differentiated to macrophages using PMA treatment as previously described 18,57 . The macrophages were treated with MCG as described previously 16 . www.nature.com/scientificreports www.nature.com/scientificreports/ RNA extraction, reverse transcription and quantitative real time-PCR (qRT-PCR). Total RNA was isolated from the cells using the mirVana RNA Isolation Kit (Ambion, Austin, TX), according the manufacturer's directions as previously described 15,16,18,19,53,57 . miRNA expression was determined using specific Taqman assays and miRNA RT Kit (Applied Biosystems, Foster City, CA) 15,16,18,19,53,57,60 . Enzyme-linked immunosorbent assay (ELISA). For the assay, wound cells or wound mφ (isolated from subcutaneously implanted PVA sponges) or THP-1 cells were seeded in 24-well plates cells (equally seeded for control and treatment). Cytokine levels were measured from commercially available ELISA kits as previously des cribed 12,18,19,30,53,57 and are expressed as per million cells.
Apoptotic cell clearance (efferocytosis) assay. Mouse macrophages that infiltrated PVA sponges were isolated as previously described 19,53 and seeded into 8-chambered slides. Apoptosis was induced in mouse thymocytes and the apoptotic cells were used to perform the efferocytosis assay as previously described 19,53 . THP-1 cells with stable knockdown of miR-21. Stable knockdown of miR-21 was achieved in THP-1 cells using lenti-miR-000-zip (control) or lenti-miR-21-zip vectors and puromycin selection as previously described 18 followed by treatment with PMA to differentiate into macrophages 18,57 . miRIDIAN miR mimic and small interfering RNA delivery. Delivery of miRNA mimic and small interfering (si) RNA were be performed as previously described 18,61 . Immunocytochemistry and imaging. Cytospin smears from THP-1 cells suspension were fixed in cell fixation buffer (BD Cytofix, BD Biosciences, CA) for 10 min. Following fixation, cells were washed, blocked in 10% normal goat serum (NGS) for 30 min and were incubated in primary antibodies for cJun (1:100; Abcam). Fluorescence tagged secondary antibody detection was performed with Alexa Fluor 488 secondary antibody (1:200, Life Technologies) as described previously 30 . Fluorescent images were collected using collected using confocal microscopy (LSM880). Image analysis was performed using Zen (Zeiss) software to quantitate fluorescence intensity.
Statistical analysis. Data are reported as mean ± SEM of 3-6 experiments as shown in the figure legends.
Student's t-test (two-tailed) and ANOVA were applied to determine the significance. p < 0.05 was considered to be statistically significant.