Pro-angiogenic effect of RANTES-loaded polysaccharide-based microparticles for a mouse ischemia therapy

Peripheral arterial disease results from the chronic obstruction of arteries leading to critical hindlimb ischemia. The aim was to develop a new therapeutic strategy of revascularization by using biodegradable and biocompatible polysaccharides-based microparticles (MP) to treat the mouse hindlimb ischemia. For this purpose, we deliver the pro-angiogenic chemokine Regulated upon Activation, Normal T-cell Expressed and Secreted (RANTES)/CCL5 in the mouse ischemic hindlimb, in solution or incorporated into polysaccharide-based microparticles. We demonstrate that RANTES-loaded microparticles improve the clinical score, induce the revascularization and the muscle regeneration in injured mice limb. To decipher the mechanisms underlying RANTES effects in vivo, we demonstrate that RANTES increases the spreading, the migration of human endothelial progenitor cells (EPC) and the formation of vascular network. The main receptors of RANTES i.e. CCR5, syndecan-4 and CD44 expressed at endothelial progenitor cell surface are involved in RANTES-induced in vitro biological effects on EPC. By using two RANTES mutants, [E66A]-RANTES with impaired ability to oligomerize, and [44AANA47]-RANTES mutated in the main RANTES-glycosaminoglycan binding site, we demonstrate that both chemokine oligomerization and binding site to glycosaminoglycans are essential for RANTES-induced angiogenesis in vitro. Herein we improved the muscle regeneration and revascularization after RANTES-loaded MP local injection in mice hindlimb ischemia.

A randomized controlled study highlighted that transplantation of autologous bone marrow-derived mononuclear cells could reduce the amputation associated with limb ischemia 6 . Chemokines or chemokine receptors such as CXCR3 could be of importance in cell therapy approaches since they are known to participate in arteriogenesis, angiogenesis and muscle regeneration 7 . Monocyte Chemotactic Protein-1 (MCP-1)/CCL2 is a chemokine involved in early inflammation and muscle regeneration following hindlimb ischemia in mice 7 . The effect of Stromal Derived Factor-1 (SDF-1)/CXCL12 on mobilization and migration of EPC from the bone marrow to the ischemic site has been reported 8 . The CC-chemokine Regulated upon Activation, Normal T-cell Expressed and Secreted (RANTES)/CCL5 is both a T cell chemoattractant and an immune-regulatory molecule. We previously described, on various cell types, that RANTES binds to its specific G protein-coupled receptors (GPCR) CCR1, CCR3 and CCR5 and to proteoglycans such as syndecan (SDC)-1, SDC-4 and CD44 [9][10][11][12] . The binding of RANTES to its specific GPCR or to heparan sulfate proteoglycans resulted in the migration of human hepatoma cells or endothelial cells [11][12][13] . RANTES has been detected in plasma samples of patients with cardiovascular diseases 14 and is involved in cardiac inflammatory disorders after organ transplantation or arterial injury 15 .
The expression of RANTES receptors, CCR1 and CCR5, on various cell types implicated in atherosclerosis further illustrates their role in this disease 15 . It has been previously demonstrated that RANTES may induce the production of VEGF, thus leading to angiogenesis in various disease animal models 12,[16][17][18] . Our hypothesis is that RANTES may promote chemokine-induced angiogenesis and accelerate the tissue regeneration after hindlimb ischemia. The beneficial effects of proangiogenic chemokines in the treatment of experimental hindlimb ischemia relate to their capacity to induce the homing of proangiogenic bone marrow derived cells. However, the use of a chemokine is limited by its sensitivity to the intense proteolytic activity found in inflammatory injured tissues and its short half-life, in vivo 19 . A biomaterial approach that could control and sustain the local prolonged release of RANTES may constitute therapeutic tools optimizing its effects on stem cell homing to the injured sites and neovascularization 20 . Our laboratory previously developed polysaccharide-based hydrogel architectures and three-dimensional scaffolds suitable for endothelial cell therapy, growth factor release applications and neo-vascularization in mice [21][22][23][24][25] .
Hence, the aim of this study was to develop a new therapeutic angiogenic approach by delivering RANTES, incorporated into polysaccharide-based microparticles, in the mice ischemic hindlimb. We demonstrated that local administration of this chemokine following the ischemic injury could promote the early reparative events preceding the neovessel formation, muscle regeneration and the restoration of blood flow.

Results
Improvement of clinical score after loaded-microparticle injection. Chemically cross-linked hydrogels based on biodegradable polysaccharides pullulan and dextran developed by our laboratory can be loaded with peptides and have been shown to favor the delivery of proangiogenic factors 25 . As compared to negative control microparticles (MP) alone incubated with avidin-FITC where no green fluorescent was observed, localization of biotinylated RANTES within the MP appeared homogeneously and weak inside the MP, but stronger on their surface, as evidenced by the green fluorescence ring (Fig. 1a). After MP incubation in PBS for 1, 7 or 14 days, RANTES was released at 2.5 ± 0.004 pg/mL (n = 3). Moreover, MP biodegradation induced by dextranase and pullulanase demonstrated that 6 ± 0.03 pg/mL of RANTES could be released from the digested MP (n = 3, Supplementary Fig. 1a). The localization of biotinylated RANTES was assessed after RANTES revelation with streptavidin-FITC labeling and by immunohistochemistry at 5 and 10 day post-injection in muscle tissue sections. We observed expression of RANTES around beads and in tissue periphery at 5 day post-injection. At 10 day post-injection, RANTES was observed as a diffuse signal in the tissue, confirming that an actual release took place ( Supplementary Fig. 1b).
Injection of PBS, VEGF or RANTES solutions or injection of MP loaded with PBS, VEGF or RANTES was performed directly into the muscle immediately after the femoral artery ligature in mice. The MP were easily identified on tissue sections using optical microscopy (Fig. 1b). The clinical scores of muscle regeneration after hindlimb ischemia were calculated at day 0, 5 and 10. The injection of VEGF solution increased significantly the mice clinical score at day 10, but not at day 5, as compared to PBS or RANTES solution injection (Fig. 1c, n = 5, P < 0.05). It is to note that the clinical score calculated at day 5 or 10 after RANTES solution injection was quite similar but not significantly different of PBS solution injection.
The injection of RANTES-or VEGF-loaded MP increased at the same extent the clinical score at day 5 or 10, suggesting that they both improved the animal behaviour (Fig. 1c, n = 5, P < 0.05). It is worth noting that negative control MP had no significant effect. We then sacrificed the animals at day 10 which corresponds to the period required to observe endothelial progenitor cell (EPC) recruitment to injured muscles and their regeneration 26,27 .
Then we wanted to analyze if the increased clinical score associated with healthier animal behavior after RANTES-loaded MP treatment was associated with muscle regeneration.

Muscle regeneration of hindlimb ischemia after loaded-microparticle injection. The location
of nuclei in the muscle myofibers is relative to the evolution of ischemia or muscle regeneration process 28,29 . In a healthy muscle, the nuclei are present at the periphery of mature muscle myofibers ( Fig. 2a left panel, black arrows). Muscle injury leads to muscle damage characterized by a loss of myofiber organization. Muscle regeneration and maturation is then characterized by migration of nuclei of myofibers from the center (associated with immature myofibers) to the periphery (associated with mature myofibers) 30,31 . After PBS solution or negative control MP injection in ischemic muscle, the nuclei appear predominantly at the center of the myofibers. Treatment with VEGF or RANTES solutions or VEGF-loaded MP led to a partial migration of nuclei from the center (  Representative confocal microscopy image of pullulan/dextran-based microparticles (MP) incubated with avidin-FITC alone (Avidin-loaded control MP) or with 10 nM biotinylated-RANTES followed by avidin-FITC to assess RANTES localization. Scale Bar: 50 µm. (b) 10 nM biotinylated RANTES-loaded MP (15 µg/mL) were injected intramuscularly in 6-week old male BALB/c white mice into their ischemic limb. At day 0, RANTES-loaded MP were identified in the injection area (I), as indicated with arrows on hematoxylin-eosin stained sections of the ischemic tissue, observed with optical microscopy. Bar: 1 mm. (c) Histograms represent clinical scores calculated as the sum of the grade obtained for Tarlov score, ischemic score and modified ischemic score at days 0, 5 and 10 after hindlimb ischemia. The effect of two different ways of treatment administration by injection of PBS, VEGF (2 nM) or RANTES (10 nM) solution or by PBS-, VEGF-or RANTES-loaded MP intramuscular injection in ischemic limb was evaluated,*P < 0.05 versus PBS solution at day 10. A quantitative analysis of the number of myofibers with peripheral and central nuclei indicated that the percentage of peripheral nuclei, marker of myofiber maturity, is significantly higher after RANTES injection. It is to note that the delivery with MP increased significantly RANTES effect as compared to the injection of RANTES in solution (Fig. 2b). Myogenin D is considered as a biomarker of muscle functionality 32 . Therefore, we assessed the expression of myogenin D by immunohistochemistry assay (Fig. 2c).The ratio of myogenin D positive cells normalized to the total number of cells, evidenced by DAPI nuclei staining, was significantly increased by 2 fold after the injection of VEGF solution or VEGF-loaded MP. In contrast, this ratio was unchanged after RANTES solution treatment but was significantly increased by 4-fold after injection of RANTES-loaded MP (Fig. 2d, n = 5, P < 0.05). The aim of the next step was to evaluate if muscle regeneration is associated with tissue revascularization.
Revascularization of hindlimb ischemia after loaded-microparticle injection. Muscle ischemia results from a tissue devascularization induced by surgery and ligature procedures. Thus, myogenesis and angiogenesis are the main processes involved in muscle regeneration 33,34 . Quantification of the blood microvessel density was evaluated on hematoxylin-eosin stained muscle cross sections of mice sacrificed at day 10. Only VEGF-or RANTES-loaded MP significantly increased by 2.6-fold or 2.3-fold, respectively, the number of blood vessel per field, whereas the other treatments did not affect it significantly (Fig. 3a). The injection of either RANTES solution led to similar microvessel density, without reaching a statistical significance versus PBS solution (Fig. 3a).The neo-formed microvessels at the periphery of the injection site, obtained after injection of PBS-, VEGF-or RANTES-loaded MP were polymorph, mainly small and thin with regular shape. Moreover, the presence of erythrocytes or polynuclear neutrophils within almost all capillaries on hematoxylin-eosin stained cross sections evidenced the functionality of the neo-formed microvessels (Fig. 3b). These neo-formed microvessels were composed of endothelial cells (EC) as assessed by CD31 immunolabelling and of vascular smooth muscle cells (VSMC) as revealed by alpha smooth muscle actin (SMA) immunolabelling ( Fig. 3c left panel). Expression of these markers suggests the formation of mature blood neovessels with an intima layer containing EC and a media layer containing VSMC. The muscle regeneration induced after RANTES-loaded MP treatment may be related to a direct pro-angiogenic effect on mature endothelial cells 12 , or a recruitment of EPC as it was previously described for the chemokine SDF-1 33 . The presence of cells derived from EPC in the zone of revascularization was evidenced by immunolabelling of their membrane with the specific markers. Undifferentiated progenitor cells can be identified by CD34 expression 34,35 .
Hence, at the periphery of microvessels some cells expressed CD34, suggesting that these cells kept a membrane marker of undifferentiated progenitor cells. These cells also expressed endothelial-specific markers such as CD31 (data not shown) and von Willebrandt factor (vWF, Fig. 3c right panel), suggesting an initiation of the differentiation process into EPC. To evaluate the cellular processes involved in RANTES-induced revascularization, we therefore investigated the in vitro RANTES-induced EPC migration and a two-dimensional (2D) vascular network formation assay.
The identification of CD34 + CD31 + vWF + cells in the tissue of RANTES-loaded MP treated mice led us to investigate the RANTES-induced biological effects on EPC in vitro. Among these effects, we focused on mechanisms involved in angiogenesis i.e EPC spreading, migration and sprouting assays.
Considering the biological effects of RANTES on EPC, and the role of chemokines binding to glycosaminoglycans or oligomerization in their induced biological effects, we evaluated the effects of RANTES mutants on EPC migration and sprouting.

RANTES-mediated effects depend on its oligomerization and its binding to glycosaminoglycans.
The mutation [E66A]-RANTES affects the oligomerization status leading to a dimeric form of RANTES.
[E66A]-RANTES increased the length of the formed vascular sprout by 15 ± 4% compared to untreated cells  (Fig. 5b, n = 3, P < 0.05). To assess the role of RANTES binding to GAG, the experiments were performed either with the [ 44 AANA 47 ]-RANTES, that is deficient in glycosaminoglycan binding sites, or with the chemokine preincubated with low molecular weight heparin (LMWH), or after the cell pretreatment with a beta-D-xyloside, an inhibitor of GAG biosynthesis. The mutant [ 44 AANA 47 ]-RANTES increased EPC chemotaxis by 49 ± 6% and vascular sprout length by 21 ± 4%, relative to untreated cells (UT) ( Fig. 5a and b, n = 3, P < 0.01 and P < 0.001, respectively). Thus, [ 44 AANA 47 ]-RANTES is 11 ± 4% and 2.9 fold less efficient than RANTES to induce EPC migration or vascular sprout formation respectively (Fig. 5b, n = 3, P < 0.01). These results demonstrated that RANTES oligomerization is essential in EPC migration and vascular sprout formation; whereas RANTES binding to GAG is necessary only for EPC vascular sprout formation.
The length of vascular sprout formed in a 2D angiogenesis assay decreased by 95 ± 5% after cell pretreatment with 1 mM beta-D-xyloside (Fig. 5f, n = 3, P < 0.05). According to these results, the cellular effect of RANTES may rely on its binding to glycosaminoglycans, but also to specific G protein-coupled receptors (GPCR). Hence the binding of RANTES to its or GPCR or to glycosaminoglycans mostly carried by membrane proteoglycans such as SDC-1, SDC-4 or CD44, may constitute the first step of RANTES cell signalling leading to these observed proangiogenic cellular effects.
Altogether, our results demonstrate that EPC express membrane receptors and proteoglycans able to bind RANTES. RANTES binding to these cells leads to pro-angiogenic cellular effects such as EPC migration and vascular sprouting.

Discussion
Patients with critical limb ischemia develop a blood monocytosis that may contribute to the tissue repair process, notably by the secretion of growth factors or chemokines 7 . Angiogenesis is important for neovascularization of ischemic tissues and protein-based or gene-based therapy using angiogenic factors such as bFGF or VEGF, is under investigation in clinical practice 1 . However, these strategies appear to be not very convincing in patients with myocardial ischemia 7 . The rapid diffusion of an angiogenic protein and its degradation by inflammation-related proteolytic enzymes may reduce its local concentration and its angiogenic effect. Therefore, the use of a biomaterial to deliver locally an angiogenic factor such as VEGF, with a controlled release, at the injured site, can be considered as an advanced and challenging therapeutic strategy 36,37 . Incorporation of the antibodies revealed with Alexa fluor 488-secondary antibody (green). Nuclei were stained with DAPI (blue).  chemokine SDF-1 into a heparin-based hydrogel was shown to induce chemotaxis in vitro and potentiate EPC enrichment subcutaneously in mice 20 . It has been demonstrated that accumulation of infused endothelial progenitors can be enriched using biomaterial-based delivery of VEGF and SDF-1 38 . Such an added effect of RANTES-loaded MP and VEGF-loaded MP co-delivery in our mice hindlimb ischemia model needs to be investigated. However, as a competition in the ionic and electrostatic association between RANTES or VEGF and the pullulan-dextran scaffold may occur, some preliminary experiments are required to better define the  respective concentrations of RANTES-loaded MP and VEGF-loaded MP inducing angiogenesis without adverse effects. The respective optimal amount of the chemokine and of the growth factor to enable the most efficient neo-revascularization is not well defined, specifically in vivo. Among chemokines released in ischemic lesions, RANTES is secreted by many cell types such as macrophages, activated T cells, platelets, endothelial cells and smooth muscle cells. Through its binding to its receptors CCR1, CCR3 or CCR5, this chemokine is known for its role in the homing and activation of inflammatory cells. The pro-angiogenic role of RANTES in cancer has been suggested by its ability to increase vascularity 39 . RANTES pro-angiogenic effects have been evoked previously but controversy still remains 39 . Indeed, Barcelos et al. reported that exogenous RANTES reduced angiogenesis in mice at day 14 in a disc-induced inflammatory angiogenesis assay 40 . Furthermore, Cochain et al. did not evidence any control of the RANTES-CCR5 axis on post-ischemic vessel growth 41 . In contrast, it was demonstrated that RANTES is required for angiogenesis following peripheral ischemia in a rat hindlimb ischemia model 42 . Moreover, Ambati et al. reported that CCR5-deficient mice experienced a sustained inhibition of corneal neovascularization after chemical and mechanical denudation of corneal epithelium 43 . We previously reported that RANTES may promote angiogenesis in a subcutaneous model of RANTES-base scaffold local implantation in rat 12 . Indeed, the incorporation of RANTES in a commercial nitrocellulose scaffold through non-covalent interactions resulted in a local angiogenesis after its subcutaneous implantation in rat 12 . In the present study, RANTES incorporated in pullulan/dextran microparticles (MP) was mostly localized on their surface, with some RANTES identified in the core of the MP. Since covalent chemical bonds between RANTES molecules and the polymers could lead to potential chemical modification and prevent the biological effect of RANTES, we selected another strategy to incorporate them into the MP. Freeze dried MP were rehydrated with a solution of RANTES, leading to the formation of a gel-like scaffold negatively charged able to retain the molecules within the hydrophilic polymer network. Our results demonstrated that RANTES could be released from MP for at least 14 days and that MP enzyme-induced biodegradation could provide an optimal angiogenic dose of RANTES 12,[39][40][41] . We hypothesize that electrostatic interactions between positively charged residues in RANTES and the negatively charged phosphoester bonds of the scaffold might be responsible for the high affinity of RANTES for the MP and that a slow degradation preserves RANTES from homeostasis degradation. Indeed, pullulan-dextran cross-linking occurs through the formation of phosphoester linkages between two hydroxyl groups within the polymer chains 44 .
Increasing the concentration of charged polymers is an attractive approach to retain growth factors within hydrogel networks through restricted diffusion out of the networks 45 . A bio-inspired approach to increase the magnitude of these electrostatic interactions relies on animal-derived sulfated glycosaminoglycans such as chondroitin sulfate and heparan sulfate to obtain growth factors-binding biomaterials 45 . We recently demonstrated that a scaffold containing fucoidan, a sulfated marine polysaccharide, and VEGF, significantly decreased its release and guided the generation of functional vascular networks 25 . Similarly, we demonstrate that the effect of RANTES on revascularization and on muscular regeneration was notably improved when it was injected with MP, suggesting that the biomaterial may preserve the chemokine from proteolysis. Our study analyzed the RANTES-induced effects on angiogenesis and muscle regeneration at day 10 post-ischemia. A time point at day 5 post-ischemia would be helpful at allowing better delineations of association/causality. Moreover, perfusion imaging using a Laser Doppler imaging system would be helpful to measure the blood flow of the ischemic limb versus the control non-ischemic limb before and after surgery. Finally, our MP could be considered as tunable reservoirs through the number of phosphoesters linkages, since this number depends on the amount of cross-linker that is used during the MP preparation step.
The cooperation between inflammation and vessel formation represents an essential axis of the muscular regeneration process 46 . Recruited monocytes-macrophages secrete growth factors, cytokines and chemokines involved in myogenic cell recruitment 7 , and satellite cell proliferation and migration 47 . The RANTES-induced myogenic cell migration may participate in the tissue regeneration 48 . The expression of CCR5 and SDC-4 by myoblasts may reinforce the role of RANTES in the recruitment and the proliferation of myoblasts 49 .
The presence of functional RANTES is required for angiogenesis following peripheral ischemia in rats 42 . The positive effects of RANTES to induce neo-angiogenesis and muscle regeneration may be of interest to restore blood flow to ischemic tissues after vessel occlusion. In addition to its effect on endothelial cells, RANTES also induces the recruitment of smooth muscle progenitor cells 50 . The mobilization of EPC from bone marrow to the blood flow, their migration to the site of ischemia, and their incorporation into neo-vessels are essential steps of limb tissue repair process 51 . However, it is to note that in our study, very few CD34 + CD31 + vWF + cells could be observed. The difficulty resides in the low number of circulating EPC in the blood as well as that of recruited and incorporated EPC in the targeted vasculature.
Therefore, to assess the in vivo potential of RANTES-loaded MP to enhance the specific recruitment of EPC, future studies should be conducted based on the administration of RANTES-loaded MP combined with an intravenous injection of fluorescent-labeled EPC in Nude mice as described 20 . It was previously demonstrated that SDF-1/CXCL12 could regulate EPC differentiation to endothelial mature cells 33 . Other chemokines may exert such a pro-angiogenic effect: MCP-1/CCL2 is the first CC chemokine reported to play a direct role in tumor angiogenesis 52 , eotaxin also induces in vivo angiogenic responses by using endothelial cells expressing CCR3 53 and numerous studies highlighted the role of CXCL4 in angiogenesis 54 . The reduced pro-angiogenic effects observed after cell treatment by [E66A]-RANTES as compared to RANTES highlighted a slight role of chemokine oligomerization. However, the relationship of RANTES with other chemokines with well-known angiogenic effects should also be taken into account. It has been described a CXCL4-CCL5 heterodimer formation, and a peptide inhibition of CXCL4-CCL5 interactions resulted in a reduced progression of human aortic aneurysm 55 .
In parallel, the injection of EPC may represent a therapeutic alternative to the local injection of pro-angiogenic growth factors. To improve the therapeutic effects of EPC delivered by intra-muscular or intra-arterial injection, the aim of actual studies is to develop adjunctive therapies, along with molecular and bioengineering tools 56 . As only 5-30% of microvessels originated from engrafted EPC 34 , the injection of both EPC and a pro-angiogenic factor, such as RANTES, may represent an innovative therapeutic strategy leading to an increased revascularization after hindlimb ischemia. RANTES-induced homing of EPC may represent a mechanism underlying angiogenesis. It was previously demonstrated that platelet aggregation releases chemokines such as RANTES. Platelets can enhance neovascularization at least partly by RANTES-enhanced CD34 + cell adhesion 57 . In addition to the role of RANTES in the induction and modulation of inflammation, RANTES stimulates the homing and recruitment of EPC in renal vascular regeneration 58 , in atherosclerotic plaque 59 and in wound healing 60 . Our data also demonstrated that the binding of RANTES to glycosaminoglycans, as evidenced by [ 44 AANA 47 ]-RANTES, to GPCR, CCR5, or to proteoglycans, SDC-4 and CD44, present at the membrane of EPC, is essential to the RANTES-induced in vitro biological effects such as spreading, migration and vascular sprout formation. Nevertheless, the moderate decrease of cell migration towards RANTES mutants as compared to cell migration towards RANTES suggest that RANTES binding to glycosaminoglycans or RANTES oligomerization are not the only mechanisms involved in RANTES-induced cell migration. It would be of interest to better characterize the respective effect of CCR5, SDC-4 and CD44 expressed by EPC when these cells migrate towards RANTES mutant. It is conceivable that RANTES may interact with the protein core of SDC-4 or CD44. For this purpose, experiments based on neutralizing antibodies specific of SDC-4 or CD44 will be conducted.
The role of CCR5 will also be further investigated by siRNA interference or using specific neutralizing antibodies. It has previously been described that the [ 44 AANA 47 ]-RANTES mutant showed a 80-fold reduction in affinity for CCR1, despite normal binding to CCR5, and was able to induce monocyte chemotaxis at micromolar concentrations, conversely to nanomolar concentrations for native RANTES 61 . The lower chemotactic effect induced by [E66A]-RANTES compared to [ 44 AANA 47 ]-RANTES may be related to the use of a positively charged and fully exposed motif, KKWVR, of CCL5 oligomer in GAG binding 62 . Moreover, future studies will be carried out to assess the putative effect of RANTES on human EPC survival, proliferation and differentiation potential into mature human endothelial cells. Moreover, the lower biological effects induced in vitro by [ 44 AANA 47 ]-RANTES encourage us to deliver it in vivo to verify its lower capacity compared to RANTES to induce revascularization and muscle regeneration after mice hindlimb ischemia induction.
Altogether, our results highlight the therapeutic role of RANTES-loaded MP after hindlimb ischemia that is involved in the chemoattraction of inflammatory cells, of mature endothelial cells and of endothelial progenitor cells, leading to the revascularization and tissue regeneration. Therefore, future studies based on the injection of biomaterials containing both human EPC and RANTES in order to deliver them locally by intra-muscular injection at the site of ischemia may represent a putative innovative therapeutic approach to counteract hindlimb ischemia and to avoid an invasive surgery.

Methods
Materials. RANTES  Microparticle preparation. Microparticles (MP) were obtained using water-in-oil (w/o) emulsification process 66 . Briefly, 75% pullulan (Mw 200000, Hayashibara Inc., Japan) and 25% dextran (Mw 500000, Sigma-Aldrich, Lyon, France) were dissolved in water then dispersed in canola oil under mechanical stirring 66 . Polysaccharides were cross-linked by sodium trimetaphosphate at 50 °C for 20 min. Resulting MP were washed with PBS then sieved using a vibrating shaker (AS 200, Retsch, Eragny sur Oise, France) to obtain particles of 300-500 µm diameter. MP were freeze-dried and stored at room temperature. For experiments, MP were suspended in PBS (15 mg/mL). The association of RANTES (10 nM) to the MP for 2 hours at 37 °C was verified by the use of biotinylated RANTES revealed with avidin-FITC (DAKO, Les Ulis, France) by confocal microscopy (Leica SP8 tandem, Wetzlar, Germany).
The resorption of RANTES-loaded MP was assessed after MP incubation with RANTES (10 nM) for 2 hours at 37 °C. Then MP were transferred to PBS and incubated for 30 minutes, 24 hours, 7 days and 14 days at 37 °C. At day 14, MP were incubated with a pullulanase/dextranase solution to simulate their bio-resorption (1:1 v/v) (Sigma Aldrich). The release profile of RANTES from MP was assessed by ELISA assay (R&D system, Lille, France).

Mouse model of hindlimb ischemia. The animal protocol was approved by the Bichat Hospital
Institutional Animal Care and Use Committee and all experiments conformed to European Community guidelines for the care and use of laboratory animals. Six week-old male white BALB/c mice, weighting 20-25 g (Janvier, CERJ, Laval, France), were anesthetized with intraperitoneal ketamine (10%, 8 mg/kg) and xylazine (5%, 8 mg/kg) solution (Bayer, Puteaux, France). BALB/c mice represent a good model for hindlimb ischemia development 67 and the sample size (n = 5 mice per group) was calculated for the study according to the objective and design of the study, literature analysis and statistical test based on acceptable level of significance; power of the study, expected effect size, underlying event rate, standard deviation 68 . The right profunda femoris artery was ligated to induce a hindlimb ischemia whereas the left muscle was kept healthy. Ten minutes later, treatments were administered by an intramuscular injection of 20 µL in the central part of ischemic limb at a distance equivalent to the middle of the distance between the two articulations. Thirty mice were divided into six groups (with n = 5 mice per group) according to the treatment injected: a) 10 nM RANTES-loaded MP (300 µg of MP at 15 µg/µL in PBS), b) 2 nM VEGF-loaded MP, c) MP incubated with PBS, d) 10 nM RANTES solution, e) 2 nM VEGF solution, f) PBS solution. Mice were euthanized by a pentobarbital injection at day 10 after the surgery.
The ischemic and non-ischemic (control) gastrocnemius muscles were collected, frozen in isopentane solution cooled in liquid nitrogen and stored at −80 °C for histological analysis.
Clinical score of muscle regeneration. Mice behavior was recorded at day 0, 5 and 10 after the induction of hindlimb ischemia. At day 0, the observation was performed immediately after hindlimb ischemia. At each time point, a clinical score was calculated for each mice group as the sum of the grade obtained for three different scales referring to completely different parameters: the Tarlov scale 69 , a mouse limb ischemia grading scale 70 , and a modified ischemia scale to detect less severe levels of ischemia 71 , to evaluate the functional grading of muscle regeneration (Table 1).
The number per field of neo-formed capillaries was counted on H&E sections. To localize biotinylated RANTES at 5 and 10 days, muscle tissue cryosections were revealed with streptavidin-FITC labeling and with an antibody anti-RANTES (10 µg/mL, polyclonal goat IgG, R&D System) revealed with a secondary antibody coupled to HRP then with DAB and counterstained with hemalun.
Spreading assay. After a 24 h serum deprivation, 20 × 10 3 EPC per well were seeded on a fibronectin (100 μg/ mL, BD Biosciences Pharmingen, Le Pont de Claix, France) pre-coated 8-well Labtek and incubated for 2 h at 37 °C with 3 nM RANTES to assess their spreading. In a parallel experiment, EPC were pre-incubated for 2 h at 37 °C with anti-CCR5, anti-CD44, or anti-SDC-4 antibodies or their isotypes (5 µg/mL). Alternatively, cells were preincubated with beta-D-xyloside (1 mM) for 72 h, or RANTES was pre-incubated with low molecular weight heparin (10 µg/mL, Sigma-Aldrich) for 2 h, as previously described 12 . Cells were permeabilized in 0.05% Triton X-100 (Sigma-Aldrich), fixed with PFA (1%) then stained with AlexaFluor 568 phalloidin (Molecular Probes, Invitrogen) and filamentous actin was observed with a fluorescence microscope (Zeiss, AXIOPHOT, N°/ MicMac, France S.A). Ten fields of stained cells were photographed for each treatment. Cell areas expressed in square inches were evaluated on 30 cells by treatment with the Scion Imager software (Scion Image Software and National Institutes of Health, Release Beta 3b Software).  Ethical approval. All applicable institutional guidelines for the care and use of animals were followed. All procedures performed in studies involving animals were in accordance with the ethical standards of the institution or practice at which the studies were conducted. This article does not contain any studies with human participants performed by any of the authors.