Signaling of Prostaglandin E Receptors, EP3 and EP4 Facilitates Wound Healing and Lymphangiogenesis with Enhanced Recruitment of M2 Macrophages in Mice

Lymphangiogenesis plays an important role in homeostasis, metabolism, and immunity, and also occurs during wound-healing. Here, we examined the roles of prostaglandin E2 (PGE2) receptor (EP) signaling in enhancement of lymphangiogenesis in wound healing processes. The hole-punch was made in the ears of male C57BL/6 mice using a metal ear punch. Healing process and lymphangiogenesis together with macrophage recruitment were analyzed in EP knockout mice. Lymphangiogenesis was up-regulated in the granulation tissues at the margins of punched-hole wounds in mouse ears, and this increase was accompanied by increased expression levels of COX-2 and microsomal prostaglandin E synthase-1. Administration of celecoxib, a COX-2 inhibitor, suppressed lymphangiogenesis in the granulation tissues and reduced the induction of the pro-lymphangiogenic factors, vascular endothelial growth factor (VEGF) -C and VEGF-D. Topical applications of selective EP receptor agonists enhanced the expressions of lymphatic vessel endothelial hyaluronan receptor-1 and VEGF receptor-3. The wound-healing processes and recruitment of CD11b-positive macrophages, which produced VEGF-C and VEGF-D, were suppressed under COX-2 inhibition. Mice lacking either EP3 or EP4 exhibited reduced wound-healing, lymphangiogenesis and recruitment of M2 macrophages, compared with wild type mice. Proliferation of cultured human lymphatic endothelial cells was not detected under PGE2 stimulation. Lymphangiogenesis and recruitment of M2 macrophages that produced VEGF-C/D were suppressed in mice treated with a COX-2 inhibitor or lacking either EP3 or EP4 during wound healing. COX-2 and EP3/EP4 signaling may be novel targets to control lymphangiogenesis in vivo.


Introduction
Lymphangiogenesis, the formation of new lymphatic vessels from pre-existing ones, plays important physiological roles in the homeostasis of interstitial fluids, metabolism, and immunity. Recent evidence indicates that the structure and function of lymphatic vessels in adult mammals may be modulated by inflammation induced by harmful external or internal stimuli, including pathogens, damaged cells, and irritants [1]. Similar to angiogenesis, lymphangiogenesis occurs in adult tissues during inflammatory diseases, including wound healing [2,3], chronic airway inflammation [4,5], inflammatory bowel diseases [6][7][8], and tumor metastasis [9][10][11]. Newly formed lymphatic networks form a drainage system for extravasated interstitial fluids, and inhibition of lymphangiogenesis and lymphatic drainage increases the severity of inflammation in a mouse model of chronic inflammation [12].
Inflammation is an inherently beneficial event that leads to the removal of causal factors and restoration of tissue structure and physiological function. Prostaglandins (PGs), which are generated from arachidonic acid via cyclooxygenase (COX) and specific PG synthases, play a key role in the generation of the inflammatory response, and their biosynthesis is increased significantly in inflamed tissues. We reported previously that microsomal prostaglandin E synthase-1 (mPGES-1) and COX-2-derived PGs, including PGE 2 , enhance angiogenesis during the development of chronic inflammation and tumors through the induction of VEGF-A, a potent pro-angiogenic factor in granulation tissues [21][22][23][24][25]. In addition, we demonstrated that COX-2 plays crucial roles in lymphangiogenesis during secondary lymphedema and tumor development [26,27]. These findings suggested that COX-2-derived PGs are endogenous regulators of lymphangiogenesis in some pathological conditions. However, the PGE receptors responsible for lymphangiogenesis, and whether endogenous PGE 2 enhances lymphangiogenesis during wound healing, remain unknown.
Here, we assessed lymphangiogenesis and the expression levels of COX-2 and mPGES-1 in the proliferative granulation tissues formed in the margin of punched-hole injuries in mice. We also examined the function of EP signaling in wound-healing and lymphangiogenesis using selective/stable PGE 2 analogues and EP receptor knockout mice. The results indicate that COX-2 and EP3/EP4 receptor signaling can promote wound-healing and lymphangiogenesis with enhanced recruitment of M2 macrophages that produce pro-lymphangiogenic VEGF-C/ D. Thus, EP3/EP4-modulating factors are promising molecular targets for controlling lymphangiogenesis during wound-healing processes.

Materials and Methods Animals
Male C57/BL6 mice (8-10 weeks old) weighing 20-25 g were purchased from CLEA Japan Inc. (Tokyo, Japan). Generation and maintenance of the EP3 -/and EP4 -/mice have been reported previously [28,29]. EP3 -/mice are backcrossed more than nine generations to C57BL/6 mice, and have no visible phenotype difference from C57BL/6 WT counter parts. Most EP4 -/mice die postnatally as a result of patent ductus arteriosus and do not survive at all in the C57BL/6 background. Therefore, the F2 progenies of surviving EP4 -/mice and their WT littermates were maintained independently in the mixed genetic background of 129/Ola and C57BL/6. For the experiments using EP4 -/mice, F2-WT mice having this genetic background were used as a control. All animal experimental procedures were approved by the Animal Experimentation and Ethics Committee of the Kitasato University School of Medicine (2013-072), and were performed in accordance with the guidelines for animal experiments set down by Kitasato University School of Medicine, which are in accordance with the 'Guidelines for proper conduct of animal experiments' from the Science Council of Japan. Mice were maintained at a constant humidity (50 ± 5%) and temperature (25 ± 1°C) and were kept continuously on a 12-hour light/dark cycle under specific pathogen free conditions. All animals were provided food and water ad libitum. All surgery was performed under isoflurane anesthesia, and all efforts were made to minimize suffering. Tissue collection procedures were initiated after animals had been euthanized by Isoflurane overdose and unresponsive to all stimuli. Once euthanasia was performed, collection of tissues was initiated immediately and was performed following the same procedure and timing. This study was not a survival study and no unexpected animal death was observed.

Hole-punch assay
The hole-punch assay was performed as described previously [30]. Briefly, a 1.5 mm hole was made in the center of both ears of mice using a metal ear punch under isoflurane anesthesia. Oral administration of a COX-2 inhibitor (celecoxib; 100 mg/kg) was performed daily. At each time point, the mice were euthanized by Isoflurane overdose and the ears were either whole mounted for immunocytochemistry or analyzed by real-time quantitative RT-PCR after harvesting. In another experiment, EP1-4 receptor agonists (50 nmol/site) were applied to both ears once a day, from day 0 to day 3. On day 3, the mice were euthanized and the ears were analyzed via real-time quantitative RT-PCR after harvesting. The EP1 receptor agonist (ONO-DI-004), EP2 receptor agonist (ONO-AE1-259-01), EP3 receptor agonist (ONO-AE-248), and EP4 receptor agonist (ONO-AE1-329) were kindly donated by ONO Pharmaceutical (Osaka, Japan).

Real-time quantitative RT-PCR
Total RNA was extracted from mouse tissues using TRIzol 1 Reagent (Life Technologies, Carlsbad, CA, USA), and single-stranded cDNA was generated from 1 μg of total RNA via reverse transcription using the ReverTra Ace 1 qPCR RT Kit (TOYOBO Co., Ltd., Osaka, Japan), according to the manufacturer's instructions. Quantitative PCR amplification was performed using SYBR 1 Premix Ex Taq™ II (Tli RNaseH Plus) (Takara Bio, Inc. Shiga, Japan). The genespecific primers used in the experiments are listed in Table 1. PCR was performed in 20 μL reactions, which were subjected to 40 cycles of amplification in a thermal cycler. Data were normalized to the expression level of GAPDH in each sample, and were expressed as (/GAPDH mRNA)(x10 -n ).

Statistical analysis
Data are expressed as the mean ± SEM. Comparisons between two groups were performed using Student's t-tests. Comparisons of multiple groups were performed using factorial analysis of variance, followed by Dunnett's test. P < 0.05 was considered statistically significant.

Lymphangiogenesis and upregulation of COX-2 and mPGES-1 at margins of healing wounds in the ear
To estimate lymphangiogenesis in the inflammatory granulation tissues formed during wound healing, the ears of mice were subjected to hole-punch injuries. When all layers of the mouse ear skin were pierced, granulation tissues were formed at the edges of the hole-punch injuries, and consequently the holes were closed. When we measured the diameters of the holes, they became an indicator of wound healing processes ( Fig 1A). As shown in the time course in Fig  1A, the closure of wound significantly occurred from day 3 post-injury compared the size of wounds at day 0. To visualize blood and lymphatic vascular tissue in whole-mounted ear skin samples on day 7, immunofluorescent staining was performed using antibodies against CD31 and LYVE-1. At this stage, new lymphatic vessels that grew thinner lengthened than normal lymphatic vessels were observed in the margins of the punch-hole injuries (Fig 1B), although we could not detect fine lymphatic vessels in the early phage (S1 Fig). A time course analysis revealed that LYVE-1 mRNA expression in the granulation tissues of the wound margins was biphasically increased on day1-2 and 5-7 ( Fig 1C). In addition, VEGFR-3 mRNA expression was increased significantly on day 5-7 ( Fig 1D). LYVE-1-positive macrophages were seen in the early stage after wounding ( Fig 1E) (Fig 2A). The level of COX-2 mRNA in the granulation tissues was increased significantly on day 1-4 compared with day 0 (Fig 2B), and the mPGES-1 mRNA level was also increased from day 2 onwards (Fig 2C). A large number of CD11b-and S100A4-positive cells were found to express COX-2 and mPGES-1 (Fig 2D), and the percentages of COX-2 and mPGES-1 expression on S100A4-positive cells or CD11b-positive cells were approximately 90% and 80%, respectively (S3 Fig).

Suppression of lymphangiogenesis in wound granulation tissues by a COX-2 inhibitor
To determine whether COX-2 enhances lymphangiogenesis, we evaluated the effects of celecoxib, a selective COX-2 inhibitor, on this process. When celecoxib was administered daily (p.o., 100 mg/kg), LYVE-1 and VEGFR-3 mRNA levels were reduced significantly on day 5, and there were concomitant reductions in the levels of the mRNAs encoding the pro-lymphangiogenic factors VEGF-C and VEGF-D (Fig 3A-3D). Administration of celecoxib did not affect the levels of COX-2 and mPGES-1 mRNA (S4 Fig). Immunofluorescent staining of LYVE-1 and CD31 in whole-mounted day 7 ear skin samples revealed that celecoxib-treated mice had a lower number of micro-lymphatic vessels than control mice (Fig 3E). Quantification of the number of sprouts revealed a significant reduction from 19.3 in control mice to 12.6 in celecoxib-treated mice ( Fig  3F). These results suggest that COX-2 up-regulates lymphangiogenesis in wound-induced granulation tissues. In addition, we tested the in vitro direct effects of PGE 2 on proliferation of human lymphatic microvascular endothelial cells (HMVECs). Evaluation of HMVECs revealed no marked increase in the proliferation rate under PGE 2 and EP agonist stimulation (Fig 3G). These results suggest that of PGE 2 does not play a direct role in forming new lymphatic vessels.
Reduction in the recruitment of CD11b-positive macrophages producing VEGF-C/D by a COX-2 inhibitor The formation of granulation tissues proximal to the punched-holes increased over time in the WT mice. However, when celecoxib (100 mg/kg) was administered, this closure of wounds was suppressed significantly from day 3-6 ( Fig 4A). S100A4-positive fibroblasts and CD11b-positive macrophages were recruited into the granulation tissues. Although the density of S100A4-positive fibroblasts was comparable in vehicle-treated and celecoxib-treated mice ( Fig  4B), the density of CD11b-positive macrophages in celecoxib-treated mice was significantly lower than that in the control mice (Fig 4C).
Immunostainings detected VEGF-C and VEGF-D in S100A4-positive fibroblasts in the day 3 granulation tissues (Fig 4D), and the percentages of VEGF-C and VEGF-D expression on S100A4-positive cells were 83.8 ± 2.8% and 88.9 ± 2.2%, respectively (S5 Fig). CD11b-positive macrophages were also stained by the antibodies against VEGF-C and VEGF-D (Fig 4D), and the percentages of VEGF-C and VEGF-D expression on CD11b-positive cells were 89.4 ± 3.4% and 88.8 ± 2.9%, respectively (S5 Fig). Further we determined the expressions of the markers of M1/M2 macrophages in the granulation tissues (Fig 5). mRNA levels of IL-1β and iNOS (both markers of M1 macrophages) in the granulation tissues were not reduced under COX-2 inhibition (Fig 5A and 5B), whereas those of MR (CD206) and Arginase1(Ang-1) (both markers of M2 macrophages) were reduced in mice treated with a COX-2 inhibitor (Fig 5C and 5D). Double immunostaining of CD11b and IL-1β     revealed no marked reduction in the cell population positive to both CD11b and IL-1β in the granulation tissues under COX-2 inhibition (Fig 5E). By contrast, the cell population of double positive to CD11b and MR was reduced in mice treated with a COX -2 inhibitor (Fig 5F). These suggested that recruitment of CD11b/MR-double positive macrophages regulated by COX-2.

Suppression of lymphangiogenesis in EP3 and EP4 receptor knockout mice
To identify the receptors of COX-2-derived PGs relevant to the enhancement of lymphangiogenesis, selective EP1-4 receptor agonists were applied topically to both ears of WT mice once a day from day 0-3 (50 nmol/site). The expression levels of the LYVE-1 and VEGFR-3 mRNAs were enhanced significantly in mice treated with a selective EP3 agonist (ONO-AE-248) or EP4 agonist (ONO-AEI-329) (Fig 7). These results suggest that lymphangiogenesis in granulation tissues of the wound margins depends on EP3/EP4 receptor signaling. To assess the roles of EP3/4 receptor signaling stimulated by endogenous PGE 2 , we determined whether lymphangiogenesis in granulation tissues was suppressed in mice lacking EP3 or EP4. On day 5, the expression levels of the LYVE-1 and VEGFR-3 mRNAs were significantly lower in EP3 -/and EP4 -/mice than WT mice (Figs 8A, 8B, 9A and 9B). These changes were accompanied by reductions in the expression levels of the VEGF-C/D mRNAs (Figs 8C, 8D, 9C and 9D). Immunofluorescent staining of whole-mounted day 7 punched ear skin samples using antibodies against LYVE-1 and CD31 revealed that EP3 -/and EP4 -/mice had fewer micro-lymphatic vessels than WT mice (Figs 8E and 9E). When quantified, the number of sprouts of lymphatic vessels in EP3 -/mice (11.6) was significantly lower than the number in WT mice (19.3) ( Fig  8F). This change was accompanied by reduced recruitment of CD11b-positive macrophages in the EP3 -/mice ( Fig 8G). The same results were observed when EP4 -/mice were compared with their WT counterparts (Fig 9F and 9G). Closures of wounds were suppressed in both EP3 -/and EP4 -/mice in comparison with WT counter parts (Figs 8H and 9H), suggesting wound-healing process was enhanced with EP3/4 signaling. Taken together, these results suggest that wound-healing, lymphangiogenesis and recruitment of CD11b-positive macrophages producing VEGF-C/D in the granulation tissues is dependent on COX-2 and EP3/EP4 receptor signaling.
Finally we determined the cell population of M1/M2 macrophages in the granulation tissues in EP3 -/mice and EP4 -/mice (Fig 9). Double immunostaining of CD11b and IL-1β revealed no marked reduction in the cell population positive to both CD11b and IL-1β in the granulation tissues in EP3 -/mice in comparison with WT counter parts (Fig 10A). By contrast, the cell population of double positive to CD11b and MR was reduced in EP3 -/mice (Fig 10A). The same was true in EP4 -/mice (Fig 10B). These suggested that recruitment of CD11b/MRdouble positive macrophages regulated by EP3/EP4 signaling.

Discussion
Similar to the way in which new blood capillaries sprout from existing capillaries or post-capillary veins during angiogenesis [31], in the wound healing process, lymphatic capillaries grow by    sprouting from existing lymphatic vessels. As shown in Fig 1B, a larger number of delicate new lymphatic vessels than normal lymphatic vessels were generated in the granulation tissues at the margins of punched-hole injuries in mice. During the healing of these injuries, COX-2 and mPGES-1 were up-regulated, and COX-2-derived endogenous PGs, possibly PGE 2 , facilitated lymphangiogenesis in the granulation tissues of the wound margins. The closure of wounds was dependent on COX-2 ( Fig 4A). The results obtained using EP3 -/and EP4 -/mice suggested that wound-healing, lymphangiogenesis, recruitment of CD11b-positive macrophages, and expressions of VEGF-C/D were up-regulated by EP3/EP4 signaling. These findings indicate that EP3/EP4 signaling plays a significant role in lymphangiogenesis.
PGs are produced ubiquitously and act locally to elicit a diverse set of pharmacological effects that modulate multiple physiological systems. PGs are generated by the sequential metabolism of arachidonic acid by COX and prostaglandin synthase enzymes. COX-1 is expressed constitutively in a variety of cells and tissues, whereas COX-2 is induced particularly by inflammatory stimulants such as cytokines and growth factors. Here, the expression levels of the LYVE-1 and VEGFR-3 mRNAs were increased significantly in the granulation tissues of the wound margins (Fig 1), and these changes were accompanied by simultaneous increases in the mPGES-1 and COX-2 mRNA levels (Fig 2). The progressive formation of granulation tissues and lymphangiogenesis in the margins of punched-hole injuries were suppressed by a COX-2 inhibitor (Figs 3 and 4A). COX-2 and mPGES-1, both of which are inducible enzymes, may contribute to wound-induced lymphangiogenesis and the generation of PGE 2 at sites of wound healing, suggesting that these enzymes are possible targets to control lymphangiogenesis. LYVE-1 expression was transiently increased in early stage (day1, 2) of wound healing (Fig 1C). This may be due to the expression of LYVE-1 in macrophages [32][33][34][35].
COX-2 inhibitors, such as celecoxib used in the present study were widely used as the antiinflammatory drugs that inhibit the extravasation of plasma to induce the inflammatory edema [36]. By contrast, it is accepted that there is no effective drug to treat primary lymphedema and secondary lymphedema at present. When infections are involved, antibiotics are widely used in lymphedema patients, whereas the use of diuretics is not recommended due to the lack of efficacy to reduce the edema. The disorder of electrolytes with use of diuretics is a serious adverse effect of them. Recently, we developed a novel approach to dissect the etiology of the secondary lymphedema using a mouse tail [26]. In this model, we had clarified COX-2 derived PGs had a prolymphangiogenic activities. Blockade of COX-2 resulted in the edema formation with increased interstitial pressure. Based on these reports together with present data suggested that supplementation of PGs may be effective to prevent the lymphedema.
The diverse physiological effects of PGs are mediated by G-protein-coupled prostanoid receptors, which are classified according to the prostanoid ligand that binds with greatest affinity [37]. Especially, the actions of PGE 2 , one of the most abundant PGs, are mediated by four subtypes of PGE receptors, namely, EP1, EP2, EP3, and EP4, and there is heterogeneity in the coupling of these receptors to intracellular signal transduction pathways. Here, we applied selective EP receptor agonists to the healing wound for several days after injury. The expression of levels of the LYVE-1 and VEGFR-3 mRNAs in the granulation tissues of the wound margins were enhanced markedly following application of a selective EP3 agonist (ONO-AE-248) or EP4 agonist (ONO-AEI-329)-treated mice (Fig 7). Furthermore, lymphangiogenesis and the expression levels of the VEGF-C/D mRNAs were lower in EP3 -/and EP4 -/mice than WT mice (Figs 8 and 9). These findings suggest that EP3 and EP4 receptor signaling enhances lymphangiogenesis during wound healing via the induction of VEGF-C and VEGF-D. Of the four EP receptors, EP3 and EP4 bind PGE 2 with highest affinity, whereas EP1 and EP2 bind with lower affinity [38]. The EP2 and EP4 receptors activate adenylate cyclase via Gs, thereby increasing intracellular cyclic adenosine monophosphate (cAMP) [39,40], whereas the EP1 receptor couples to Gq and promotes increases in intracellular calcium levels [41,42]. The EP3 receptor couples to a Gi-type G-protein, leading to the inhibition of intracellular cAMP [43]; however, its individual splice variants can also couple to Gs, leading to stimulation of cAMP and IP3 generation [44][45][46]. As shown here, EP3 and EP4 receptors appear to play similar roles in lymphangiogenesis, including the induction of VEGF-C/D and the recruitment of CD11bpositive macrophages (Figs 8 and 9), suggesting that EP4 and a splicing variant of EP3 activate adenylate cyclase and elevate cAMP levels. Thus, the cAMP-dependent signaling pathway may enhance EP3/EP4-dependent lymphangiogenesis and recruitment of CD11b-positive macrophages that produce VEGF-C/D.
To clarify the functional relevance of COX-2-and EP3/4-induced lymphangiogenesis to wound healing process, inhibition of lymphangiogenesis via vascular endothelial growth factor receptor 3 blockade [12] would be very helpful. If wound healing delay with this antibody treatment was observed in vehicle-treated mice or in WT mice, but not in mice treated with a COX-2 inhibitor, or in mice lacking EP3/4receptors, we can provide the conclusive evidence that COX-2/EP3/4-dependent lymphagiogenesis facilitates wound healing processes.
Macrophages play a crucial role in the maintenance of tissue homeostasis through remodeling and repair, as well as the secretion of a wide range of growth factors and cytokines [47]. In several disease models, tumor-and inflammation-induced lymphangiogenesis is influenced by stromal cells, and is mainly dependent on macrophage recruitment and activation [4,17,48]. Some studies have indicated that co-expression of macrophages and lymphatic markers, and incorporation of CD11b-positive macrophages into newly formed lymphatics, occurs during the early stage of lymphatic formation [19,49,50]. Further, it had been reported that an increase in dietary salt resulted in the increased storage of sodium in the skin, and lymphatic growth through the activity of macrophages [51]. A high-sodium diet induced COX-2 in macrophages, resulting in enhanced EP4 receptor signaling, and subsequent triggering of parallel pathways in the kidney and in skin that help dispose of excess sodium [52]. High salt levels interfere with alternative activation of macrophages (M2), which function in attenuating tissue inflammation and promoting wound healing [53]. In the present wound healing model, the density of CD11b-positive macrophages with M2 macrophage markers was suppressed significantly in celecoxib-treated, EP3 -/-, and EP4 -/mice (Figs 4C, 8G and 9G), suggesting that recruitment of CD11b-positive M2 macrophages was regulated by COX-2 derived PG. On the other hand, although COX-2 inhibition did not affect the density of S100A4-positive fibroblasts, it did reduce the progressive formation of granulation tissues in the margin of punchedhole injuries significantly (Fig 4A and 4B). VEGF-C/D were detected in both S100A4-positive fibroblasts and CD11b-positive macrophages in wound granulation tissues (Fig 4D), suggesting that the progressive formation of granulation tissues and recruitment of CD11b-positive macrophages may play a major role in lymphangiogenesis. We are producing flox mice of EP3/4 now. In the near future, we can test cell specific knockout in macrophages in the present model.
In the present study, we had determined the expressions of markers of M1 macrophages and M2 macrophages in the wound granulation tissues. Interestingly, the population of alternative M2 macrophages, but not M1 was recruited to the granulation tissues in a COX-2-dependent manner and an EP3/4-dependent manner. These suggested that endogenous PGE 2 enhanced recruitment of M2 macrophages to the granulation tissues.
Furthermore, the expression levels of chemokines that can elicit macrophage recruitment, including MCP-1, MCP-2, MCP-3, and M-CSF, as well as those of growth factors/chemokines relevant to wound healing process, were increased in the wound granulation tissues. Among the chemokines and growth factors examined, the expression levels of the mRNAs encoding MCP-1, MCP-2, MCP-3, FGF-2, and SDF-1 were reduced after COX-2 inhibition (Fig 6A-6C, 6E and 6H). As discussed above, we also observed a decrease in the progressive formation of granulation tissues and recruitment of CD11b-positive macrophages following COX-2 inhibition. These results suggest that attenuated expression of these chemokines may result in reduced recruitment of macrophages and the progressive formation of granulation tissues.