Adenosine diphosphate contributes to wound healing in diabetic mice through P2Y1 and P2Y12 receptors activation

Several studies have shown the importance of purinergic signaling in various inflammatory diseases. In diabetes mellitus, there is an increase in the activity of some nucleotidases suggesting that this signaling may be affected in the diabetic skin. Thus, the aim of our study was to investigate the effect of ADP on wound healing in diabetic skin. Swis and C57BL/6 mice were pharmacologic induced to type 1 diabetes and submitted to a full-thickness excisional wound model to evaluate the effect of ADP as a topic treatment. Adenosine diphosphate accelerated cutaneous wound healing, improved the new tissue formation, and increased collagen deposit by positively modulating P2Y1 and P2Y12 and TGF-β production. In parallel, ADP reduced reactive oxygen species production and TNF-α levels, while increased IFNγ, IL-10 and IL-13 levels in the skin. Also, ADP induced the migration of neutrophils, eosinophils, mast cells, TCRγ4+, and TCRγ5+ cells while reduced Treg cells towards the lesion at day 7. In accordance, ADP increased the proliferation and migration of fibroblast, induced myofibroblast differentiation and keratinocyte proliferation in a P2Y12-dependent manner. We provide the first evidence of ADP acting as a potent mediator on skin wound resolution and a possible therapeutic approach for diabetic patients worldwide.


Introduction
Wound healing is a complex, dynamic and multi-mediated process characterized by a highly regulated cascade of events requiring the interaction of many cell types, including inflammatory and immune cells. Normal healing process occurs over a range of overlapping events: inflammation, granulation tissue formation, and remodeling. Impaired wounds are often associated with pathologic inflammation due to a persistent, incomplete, or uncoordinated healing process [1,2].
Patients suffer from abnormalities of wound healing worldwide; in particular, in senescence, diabetes, ischemia, peripheral vascular disease, or cancer [3,4]. Chronic wounds are reported to affect around 6.5 million patients just in USA; the estimated annually cost is more than US$ 25 billion for wound-related complications and healthcare system. In Brazil, the most populous country in Latin America, about 40 to 60% of non-traumatic lower limb amputations occur in diabetic patients; whereas, about 85% are related to foot ulcers [5][6][7].
ADP plays a pivotal role in the physiologic process of hemostasis and platelet aggregation. ADP activates P2Y1, P2Y12, and P2Y13 receptors, which are expressed by monocytes/macrophages, lymphocytes, mast cells, fibroblasts, keratinocytes, endothelial cells, eosinophils, platelets, neutrophils, and dendritic cells [8][9][10]. Neuroprotective function for ADP was demonstrated in zebrafish retina since it mitigated the excessive cell death and tissue damage; additionally, it stimulated cellular proliferation after injury [11]. This study aims to explore the role of ADP in accelerating wound healing in diabetic mice, considering that chronic wounds are a relevant health problem evidenced by the lack of an effective treatment, especially in diabetic patients, and also the pro-inflammatory, cell proliferative and pro-resolution effects of purinergic antagonists. The full-thickness excisional wound murine model was used to address our aim since it is the most reproducible and feasible model for tissue repair.

Mice
Male Swiss and C57BL/6 mice, obtained from the Institute of Science and Technology in Biomodels at Oswaldo Cruz Foundation, were used for full-thickness excisional wound models.
For the cutaneous leishmaniasis lesion model, we used male BALB/c mice, obtained from the Microbiology and Parasitology Department animal facility at Biomedical Institute in Fluminense Federal University. All procedures described were approved by the Ethics Committee for the Use of Animals of the Federal University of Rio de Janeiro (CEUA/UFRJ: 093/15 and IMPPG 128/15).

Induction of diabetes mellitus
Diabetes was induced by alloxan (65 mg/kg, i.v.) in mice fasted for 12 h [12,13]. Non-diabetic mice were injected with saline. Diabetes was confirmed 7 days later when blood glucose concentration was at least 350 mg/dl. The glucose levels were still elevated (over 350 mg/dl) at day 30 after alloxan injection.

Full-thickness excisional wound model
At day 7 after alloxan or saline administration, mice were i.p. anesthetized (ketamine 112 mg/kg and xylazine 7.5 mg/kg) and a full-thickness excisional wound (10 mm in diameter) was executed on the dorsum using biopsy punch. Wounds were treated once a day for 5 or 14 days with topical application of 5'-AMP, ADP, ATP, adenosine, or pyrophosphate (Sigma-Aldrich, St Louis, MO) at 30 μM (30 μL -15.4 µg/kg), or vehicle (30 μL of saline/mouse) or until the determined day for sample collection.

Wound area quantification
To determine the wound closure rate, the wound area was evaluated at days 0, 3, 7, 10, and 14 after wounding. Photos were taken at a standard distance using a tripod and were analyzed using ImageJ software. Data were expressed as a percentage of the initial wound area.

Hematoxylin and eosin and total collagen quantification
Wounds tissues were paraffin-embedded and cut in 5-μm thick sections. Hematoxylin and eosin staining were performed, as previously described. Skin sample sections (7-μm) were stained with Picro-Sirius Red for total amount of collagen, as previously reported (Barros et al., 2019).
The quantification was determined by morphometric analysis using a quantitative imaging software (ImagePro Plus, version 4.5.1). The percentage of collagen per field was obtained by dividing the total area by the fibrosis area.

Ecto-nucleotidase activity
Ecto-nucleotidase activity was determined in the wound homogenates by the rate of inorganic phosphate (Pi) release using the malachite green reactions, as described elsewhere [14]. The concentration of Pi released in the reaction was determined by a Pi standard curve and expressed as nucleotidase activity (nmol Pi x h -1 x mg ptn -1 ).

Imunohistochemistry
Wound samples collected at day 7 were paraffin-embedded, sections (7-μm) cut and immunostained for several markers, as described previously [15]. The specific markers are detailed in the supplemental material. The data were expressed as number of positive cells per field. For collagen type markers, we developed a score method for the semi-quantification of collagen deposits performed by two different observers.

Cell proliferation
Primary neonate dermal fibroblasts (2 x 10 4 cells) were used for BrdU staining proliferation assay, as previously described [17]. The images were captured using a fluorescence microscope and analyzed using ImageJ software. Results were expressed as the percentage of BrdU + cells by total number of cells labeled with DAPI.

Wound scratch assay
Primary dermal fibroblasts were seeded and grown until 90% confluence to evaluate migrationinduced effect of ADP (10, 30, or 100 µM), as previously reported [18]. Pictures of the scratched areas were taken at 0, 6, 12, 18, and 24 h using an inverted microscope equipped with a digital camera (BEL Engineering -Monza, Italy). The areas were measured using the ImageJ software, and fibroblast migration expressed as % of area still open related to the initial area (0 h).

Flow Cytometry
Flow cytometry of the wound tissues was performed as previously described [19]. Briefly, wound tissues were digested by an enzyme cocktail (supplemental material) and the cells (10 6 cells/mL) were subjected to FACs procedure, stained, and analyzed. Lymphocyte populations recovered from skin and draining inguinal, axillary, and brachial lymph nodes were also analyzed. For skin Treg cell analysis, samples were enriched by Percoll gradient for mononuclear cells. Samples were acquired with BD FACS Canto II (BD Biosciences, San Jose, CA) and then analyzed with FlowJo software. Gating strategy and the list of antibodies are described in the supplemental material.

Eosinophils and mast cells infiltrate
Skin sections (5-μm) were stained with modified Sirius Red or Alcian Blue for eosinophils and mast cells, respectively, as described elsewhere [20,21]. Images were taken using a digital camera coupled to the microscope (Olympus BX53) at 40x magnification. Twenty fields were analyzed per wound/animal (n=3) and the data were expressed as number of eosinophils or mast cells/mm 2 .

Myeloperoxidase activity
The number of neutrophils was indirectly determined by myeloperoxidase enzyme activity in the wounds removed 7 days after wounding, as previously described [22]. Neutrophils numbers were estimated by a standard curve, using neutrophils obtained 6 h after i.p. administration of 3% thioglycollate (>90% of neutrophils). Proteins were measured by the Bradford method. Results were expressed as number of neutrophils/mg of protein.

ELISA
Cytokine quantification was performed in wounds obtained at day 7 using PeproTech kits following manufacturer's instructions. The results were expressed as pg or ng of cytokine/mg of protein.

Superoxide assay
The superoxide production assay was performed by the nitroblue tetrazolium (NBT) reaction with reactive oxygen species resulting in formazan as final product [23]. Briefly, the wounds were removed at day 3 and homogenized in PBS containing protease inhibitors for the assay.
The formazan formed was measured by ELISA reader (620 nm, Spectra Max-250, Molecular Devices). Results were expressed as µg of formazan/mg of protein.

Cytometric Bead Array (CBA)
Cytokine concentration in the wounds was determined by flow cytometry using the kit CBA Mouse Inflammation (BD Biosciences, San Diego, CA), following manufacturer's instructions.
This CBA kit allows measurements of IL-6, IL-10, C-C motif chemokine ligand 2, IFN-γ, TNF-, and IL-12p70. Sample processing and data analysis was acquired by FACS Calibur flow cytometer (BD Bioscences) and FCAP Array software, respectively. Results were expressed as pg or ng of cytokine/mg of protein.

Statistical analysis
Statistical differences in the wound closure experiments were determined using two-way ANOVA with the Bonferroni post-test using the GraphPad Prism software. The significance of other experiments was determined by unpaired Student's t test.

ADP improves wound healing in diabetic mice
Swiss male diabetic and non-diabetic mice were topically treated with saline or ADP 30 μM (30 μL -15.4 µg/kg), every day for 5 days after wounding. ADP was effective in accelerating the wound closure in diabetic mice, but without changing the wound healing in non-diabetic mice (Figure 1a and 1b). ADP-treated diabetic mice presented 60 % wound closure versus 2 % in saline-treated mice at day 7. More importantly, the wound closure profile of the diabetic animals treated with ADP was similar to saline-treated non-diabetic mice. Moreover, the only effective dose able to accelerate wound closure was 30 μM. Indeed, higher ADP doses seemed to delay wound healing observed until day 14 ( Figure 1c).

Clopidogrel (Clop) impairs ADP-induced wound closure
To assess the role of P2Y12 and ADP in our model, a P2Y12 irreversible antagonist, Clop (5 mg/kg), impaired the ADP-treated wound closure in diabetic mice. It was characterized by an increase of the lesion size and a worsening of the wound general aspects, at all the time-points evaluated ( Figure 1a). Furthermore, Clop administration also worsened the saline-treated wound of diabetic mice. Still, an endogenous critical role of ADP/P2Y12 for tissue repair was suggested since Clop treatment also impaired saline-or ADP-treated wounds of non-diabetic mice ( Figure 1b and Supplemental Figure 1).
Assuming that ADP is the major antagonist of P2Y12 [25,26], our data demonstrates an unequivocal proof of ADP's role in accelerating wound closure of diabetic mice.

P2Y1 is also involved in ADP effects
ADP receptors involvement in wound healing was verified by another P2Y12 antagonist (MRS2395) and by a P2Y1 antagonist (MRS2179), both at 30 μM/mouse. Both antagonists impaired the wound closure induced by ADP until day 7; however, at day 10 and 14 the wound healing profile was identical to that of ADP-treated group (Figure 2a-b). P2Y1 and P2Y12 antagonists alone did not alter the wound closure in diabetic mice. We observed in the Figure 1 that Clop treatment not only impaired the beneficial ADP effect, but also worsened the healing process, and we expected to observe the same response with MSR2395. However, the administration routes were different between these drugs, and Clop is an irreversible antagonist, unlike the MRS2395 and MRS2179 that are competitive antagonists.

Apyrase worsens wound healing
Apyrase removes the γ-phosphate from ATP and the β-phosphate from ADP, yielding AMP [27]. Apyrase treatment worsened wound healing of diabetic mice, compared to saline-or ADPtreated diabetic wound (Figure 2c), confirming the crucial role of this nucleotide in tissue repair.

Different nucleotides do not accelerate the wound healing
In order to certificate that ADP is the nucleotide responsible for the observed effect in our model, adenosine, 5'AMP, pyrophosphate or ATP were topically applied on the wounds of diabetic mice at 30 μM/mouse, the same ADP concentration used. Among the nucleotides tested only ATP showed a slight improvement of wound closure at day 7 (Figure 2d-g).

ADP reduces ecto-nucleotidase activity in the wounds of diabetic mice
We tested a possible enzyme deregulation related to ADP degradation during diabetes. The ecto-nucleotidase activity observed in the ADP-treated wounds obtained from diabetic mice was reduced compared to saline-treated wounds from diabetic mice, and also when compared to both non-diabetic mice groups (Figure 2h). The same profile was observed in blood samples obtained from ADP-treated diabetic mice (data not shown). It seems that ADP treatment downregulates ecto-nucleotidase activity only in diabetic mice, which seems to favor wound healing.

ADP increases P2Y1 + and P2Y12 + cells in
ADP-treated diabetic wounds presented higher expression of P2Y1 and P2Y12 at day 7, compared to saline-treated wounds. Clop treatment impaired ADP-induced P2Y1 and P2Y12 expression, whereas it did not change the receptors expression in saline-treated diabetic mice ( Figure   2i-j).

ADP improves tissue formation
Saline-treated wounds of diabetic mice presented edematous dermis, leukocyte infiltration (predominantly by mononuclear cells), and null (or partial) formation of epidermis at day 7. In the reticular dermis, exuberant granulation tissue formation and congested neovessels were observed ( Figure 3a). Interestingly, ADP-treated wounds presented a chronological change of the regenerative process. The epidermis was regenerated and integrated to the underlying dermis, with hyperplasic suprabasal layers, and hyperkeratosis. In the dermis, there was an exuberant granulation tissue with inflammatory cell infiltrate comprising eosinophils, mast cells, myeloid progenitors, neutrophils, and mononuclear cells. Clop administration impaired tissue regeneration in saline-treated wounds, where denuded epidermis areas, necrotic dermis with an inflammatory infiltrate composed predominantly of polymorphonuclear cells, striking bleeding, and the absence of granulation tissue were observed. Clop administration also impaired wound healing in ADP-treated mice; however, with milder effects. In this case, wounds displayed a more organized reticular dermis, with collagen bundles parallel to the skin surface and interspersed with fibroblasts; a few vessels and inflammatory infiltrate were also noticed ( Figure 3a).
Picro Sirius Red-stained photomicrographs (red staining) showed higher collagen fibers deposition in ADP-treated wounds of diabetic animals compared to the saline-treated wounds.
Clop administration impaired collagen deposit in both ADP and saline-treated wounds. Collagen fibers quantification confirmed that ADP treatment enhanced collagen deposition while Clop administration impaired its accumulation (Figure 3b). ADP seemed to accelerate the switch of type III to type I collagen, a more mature fiber. Nevertheless, Clop administration reduced type I collagen deposit, without changing type III collagen production in both salineand ADP-treated wounds (Fig. 3c-d). The results depicted in the bar graphs represent the photomicrographs.

ADP induces keratinocyte proliferation
We also evaluated if ADP enhances re-epithelization. At day 7 after wounding, ADP-treated wounds presented higher number of Ki67 + cells, a cell proliferation marker, in the layer adjacent to the basal membrane when compared to saline-treated wounds. At day 14, the Ki67 + cells in ADP-treated wounds were reduced but still higher than in saline-treated wounds. Corroborating this result, epidermis area is also larger in ADP-treated wounds at day 7 compared to saline-

ADP modulates the inflammatory response
At day 3 after wounding, ADP promoted increased INF-γ and reduced TNF-α levels without affecting IL-10 levels, while increased IL-10 and IL-13 levels were observed at day 7 ( Figure   4b). No differences were detected in IL-6, IL-12p70 and C-C motif chemokine ligand 2 (CCL2/MCP-1) levels between groups (data not shown). We also observed reduction of reactive oxygen species production at day 7 post wounding after ADP treatment, while Clop administration restored reactive oxygen species production (Figure 4c), suggesting again the participation of P2Y12. These results suggest that ADP treatment controls inflammatory response associated with pro-resolution effects.

ADP increases myofibroblasts population and TGF-β production
Myofibroblasts present high ability of promoting extracellular matrix protein production and wound contraction [1]. We observed that ADP treatment increased α-smooth muscle actin (α-SMA) expression in the dermis, which was reduced by Clop, while in Clop+ADP group a less dramatic reduction of myofibroblasts was observed (Figure 5a). ADP effect on α-SMA increased expression was also confirmed by WB assay (Figure 5b TGF-β is a pivotal cytokine in wound healing [1], and ADP treatment seemed to increase TGF-β production by keratinocytes (Figure 5f -epidermis) and to increase the number of TGF-β + cells in the dermis (Figure 5g).

ADP treated wounds present a different leukocyte profile
Unbalance of numbers and/or activation of local leukocytes are common under diabetes condition, which compromises tissue repair [28]. Interestingly, we observed an increase of neutrophil recruitment in ADP-treated wounds (Figure 6a upper line and 6b). In parallel, a decrease in the inducible nitric oxide synthase + cells and an increase in the arginase + cells were detected in ADP-treated wounds (Figure c-d). This suggests that macrophage population switched towards an alternative-activated phenotype instead of an alteration in frequency, since F4/80 + macrophage numbers were similar between groups (Figure 6a  T cells are resident in normal human skin and participate in cutaneous immunosurveillance, contributing to skin homeostasis [29]. Thus, we evaluated T cell profile in the skin and wound-draining lymph nodes of diabetic mice after ADP treatment. Interestingly, the percentage of Treg cells (Foxp3 + /CD4 + /CD3 + ) was selectively reduced in the ADP-treated wounds, but not in the draining lymph nodes (Figure 7a).
In parallel, ADP did not alter CD4 + and CD8 + T cells frequencies in the skin and in the lymph nodes; however, ADP-treated wounds showed increased proportions of Vγ4 + γ T cells and Vγ5 + γ T cells. Again, no changes in T cell population were seen in the draining lymph nodes after wounding (Figure 7b).

ADP accelerates wound closure only in diabetes condition
We also evaluated ADP effect on cutaneous ulcer induced by Leishmania amazonensis infection, and no improvement was observed (Figure 8a-b). These results indicate that ADP may be effective only in wounds due to metabolic diseases such as diabetes.

Discussion
Herein we demonstrate that ADP acts in accelerating skin wound healing in diabetic mice. Due to the large number of patients suffering from diabetes worldwide, presenting a poor life quality and high risk of complications as chronic wounds, we emphasize the importance to understand the pathophysiology of wound healing and to search for novel substances for wound treatment. ADP is an endogenous nucleotide involved in platelet aggregation, inflammation, and repair, without apparent side effects, being quickly metabolized. We have unequivocal evidence that ADP is a pivotal mediator for tissue repair and possibly a promising therapeutic agent for wound healing.
Controlled inflammation is essential for wound healing. The absence of inflammatory response or its exaggerated activation impairs wound healing towards the proliferative and remodeling/healing phases [30]. Our findings point ADP as a key modulator of cell activation, inflammation, and restoration of tissue integrity in diabetic mice wounds.
Adenosine diphosphate was able to improve wound healing only of diabetic mice. To note, ADP-treated wounds of diabetic mice heal at the same rate as wounds of non-diabetic healthy mice. Deficient wound healing in diabetic mice may be due to an insufficient ADP production, upregulation of ADP degradation, or ineffective expression/activation of purinergic receptors. Due to ADP liability its quantification was unfeasible. Thus, we moved to investigate the activity of enzymes that degrade extracellular nucleotides [31,32]. Data from literature demonstrated that nucleotidases activity is increased in diabetic patients and in associated pathologies. Moreover, hydrolysis of adenine nucleotides is increased in platelets from diabetic patients [33,34]. Thus, we demonstrated a reduction in nucleotidase activity by ADP treatment in diabetic wounds. Furthermore, we also observed an increase on P2Y1 and P2Y12 receptors expression after ADP treatment. Taken together, both data may contribute to the positive effect of ADP in diabetic wounds.
Several strategies were used to certificate the ADP effect in wound healing of diabetic mice. Initially, we demonstrated that ATP, 5'AMP, adenosine, and pyrophosphate were not as effective as ADP. Additionally, an important role of endogenous ADP and P2Y12 in tissue repair was confirmed via Clop, since it impaired the beneficial effect of exogenous ADP on diabetic wounds, and also on non-diabetic wounds [26, 27, 34].
ADP's receptors are expressed in important cells for tissue repairment [9,26,29], and, probably, this fact allowed ADP to improves tissue formation with less edema, collagen deposit, accelerated re-epithelization, and increased cell infiltration such as leukocytes and fibroblasts from the edge towards the center of the lesion. These cells characterize granulation tissue formation, crucial for healing process. The mentioned results place ADP as a pro-inflammatory and pro-resolution molecule, providing superior quality and organization in tissue formation.
Fibroblasts/myofibroblasts are contractile cells that approach the edges of the wounds and produce extracellular matrix, primarily collagen, the major mature scar component [35].
The increase of myofibroblasts population induced by ADP helps to explain the increase in collagen deposit and in dermal TGF-β + cells. The shift from type III to type I collagen, triggered by ADP treatment, provides a more mature connective tissue and scar [35].
The balance of pro-and anti-inflammatory cytokines is essential for successful healing [36]. In our analyzes, the increment of IFN-γ levels, at day 3, as well as IL-10 and IL-13, at day 7 after wounding, suggests an anticipation in the shift of inflammation to resolution phase induced by ADP treatment. Also, the increased amount of TGF-β in the wound after ADP treatment corroborates an earlier resolution phase [37].
An intense inflammatory infiltrate in the wound was observed after ADP treatment. It is noteworthy that the inflammatory process, during normal wound healing, is characterized by spatial and temporal changes in leukocytes patterns. The well-defined chronology of these events is essential for ideal repair [38]. Tissue macrophages exposed to IL-4 and IL-13 are converted to a wound healing programmed cell [39]. The high concentration of IL-13 at day 7, together with a shift of macrophage phenotype from M1 to M2 after ADP treatment, corroborates to our hypothesis that ADP is a pro-resolution molecule.
Neutrophils can also manage the innate immune response during wound healing by reg- In conclusion, we demonstrated, for the first time, that ADP promotes skin homeostasis by inducing a brief and balanced inflammatory and immune response, followed by an adequate proliferative and remodeling phase, results summarized in the cartoon (Figure 9). Thus, exogenous ADP could be a new insight for therapeutic agents in diabetic wounds. 36. Kondo T, Ohshima T (1996) The dynamics of inflammatory cytokines in healing process of mouse skin wound: a preliminary study for possible wound age determination. J Leg Med 108 (5)         and (c) TCRγ4 + and TCRγ5 + cells (relative to total γδ + T lymphocytes). Data were expressed as mean standard error of the mean. *P<0.05 by Student's t test compared to saline-treated group;

Carson LF, Hladik C (2009) Histotechnology: A Self-Instructional
n=4-5 per group, data are representative of two independent experiments, except for γδ + T lymphocytes data that represent one experiment.