miR‐126‐3p is essential for CXCL12‐induced angiogenesis

Abstract Atherosclerosis, in the ultimate stage of cardiovascular diseases, causes an obstruction of vessels leading to ischemia and finally to necrosis. To restore vascularization and tissue regeneration, stimulation of angiogenesis is necessary. Chemokines and microRNAs (miR) were studied as pro‐angiogenic agents. We analysed the miR‐126/CXCL12 axis and compared impacts of both miR‐126‐3p and miR‐126‐5p strands effects in CXCL12‐induced angiogenesis. Indeed, the two strands of miR‐126 were previously shown to be active but were never compared together in the same experimental conditions regarding their differential functions in angiogenesis. In this study, we analysed the 2D‐angiogenesis and the migration assays in HUVEC in vitro and in rat's aortic rings ex vivo, both transfected with premiR‐126‐3p/‐5p or antimiR‐126‐3p/‐5p strands and stimulated with CXCL12. First, we showed that CXCL12 had pro‐angiogenic effects in vitro and ex vivo associated with overexpression of miR‐126‐3p in HUVEC and rat's aortas. Second, we showed that 2D‐angiogenesis and migration induced by CXCL12 was abolished in vitro and ex vivo after miR‐126‐3p inhibition. Finally, we observed that SPRED‐1 (one of miR‐126‐3p targets) was inhibited after CXCL12 treatment in HUVEC leading to improvement of CXCL12 pro‐angiogenic potential in vitro. Our results proved for the first time: 1‐the role of CXCL12 in modulation of miR‐126 expression; 2‐the involvement of miR‐126 in CXCL12 pro‐angiogenic effects; 3‐the involvement of SPRED‐1 in angiogenesis induced by miR‐126/CXCL12 axis.

are the small soluble proteins belonging to the family of chemoattractant cytokines that are secreted in ischemic areas. 2,3 Interestingly, we and others have already shown that chemokines can also be involved in angiogenesis. 4, presence in the ischemic tissue allows for the recruitment of endothelial progenitor cells leading to local reendothelialization. 6 In addition, CXCL12 stimulates angiogenesis by binding to their specific seven transmembrane domains receptors coupled to G proteins, such as CXCR4 or CXCR7. This binding induces the intracellular signalling pathways involved in angiogenesis such as the MAPK Erk1/2. 7 This phenomenon can also be stimulated by modulation of microRNA (miRs) expression. 8 miRs are small, single-stranded, non-coding RNAs involved in the regulation of gene expression. By binding to their specific targeted mRNAs, they can induce their total degradation or repress their protein translation. 9 Over the last few years, miRs have been extensively studied in the context of angiogenic processes. Indeed, three classes of miRs could be dissociated: pro-angiogenic miRs, anti-angiogenic miRs and miRs with a dual role. 10 Among them, miR-126, strongly expressed in EC, has been found to be implicated in angiogenesis. 11,12 miR strand selection determines which one of the two strands (−5p or −3p) becomes the active strand, and this varies according to cell type and disease state. 13 Interestingly, it was previously shown that both strands of miR can be functional and have different targets. 14 -16 In some miR species, including miR-126, both the passenger strand (miR-126-5p) and guide strand (miR-126-3p) have been shown to improve the biological effects, complicating the interpretation of their action. 16 The pro-angiogenic role of miR-126-3p has been extensively studied. Indeed, it is implicated in the Erk1/2 signalling pathway induced by VEGF-A through a repression of the protein SPRED-1 17 whose role is to inhibit the expression of the small protein G Ras. 18 Although mainly degraded, miR-126-5p has also demonstrated its pro-angiogenic role in vitro, but also in vivo by reduction of intimal hyperplasia and by stimulation of EC proliferation. 19 In EC, the link between miR-126-3p and chemokine CXCL12 has been already demonstrated. 20 Indeed, we have previously shown that overexpression of miR-126 led to an increase of CXCR4 protein expression, a CXCL12's receptor. 21 Finally, overexpression of miR-126 caused an increase of CXCL12 synthesis by EC. 20 The originality and interest of our study was to compare in the same experimental conditions the effect of both miR-126-3p and −5p strands in two angiogenesis models, in vitro and ex vivo. Also, our aim was to determine whether the modulation of their expression (miR-126-3p and −5p) is involved in the vascular tube formation induced by CXCL12. Growth Factor, 0.2 μg.mL -1 hydrocortisone, 0.5 ng.mL -1 VEGF, 10 ng. mL -1 bFGF, 20 ng.mL -1 Insulin like Growth Factor, 1 μg.mL -1 ascorbic acid and 100 Units.mL -1 of penicillin and 100 µgmL -1 of streptomycin.

| Cell culture
The cells were cultured in an incubator at 37°C under a controlled atmosphere of 5% CO 2 .
The Huh7 human hepatoma cell line was grown in Dulbecco's minimal essential medium supplemented with glucose (1 g.L -1 ), 10% of Foetal Bovine Serum, streptomycin (100 UI.mL -1 ) and penicillin (100 UI.mL -1 ) (Invitrogen). Cells were grown at 37°C in disposable plastic flasks, in a humidified atmosphere containing 5% CO 2 . The medium was replaced twice weekly, and cells were trypsinized and diluted every 3 days at a ratio of 1:3.

| Western blot
For the SPRED-1 protein expression analysis, 20 μg of total proteins were loaded on a 7% poly-acrylamide gel and then transferred to a nitrocellulose membrane (ref 10600001, GE Healthcare). The membranes were saturated twice 1-hour with baths containing TBS/T (TBS, 0.1% Tween 20) and 5% milk. An anti-SPRED-1 (E-5) antibody (sc-393198, Santa Cruz) was added diluted to 1/500e in TBS/T and 5% milk overnight. The detection was made by incubation with a secondary goat anti-mouse antibody diluted to 1

| Aortic ring assay
To study the role of miR-126 and CXCL12 in ex vivo angiogenesis, aortas were collected from 5 weeks old Sprague-Dawley rats, frag-

| 2D-angiogenesis
To study the role of miR-126 and CXCL12 in vascular tube formation in

| Migration assay
HUVEC migration was studied using a modified Boyden Chamber. First, the upper chamber was precoated with fibronectin (100 µg.mL -1 ) overnight at 4°C. After removing the excess of fibronectin, chambers were saturated with DMEM containing 0.1% BSA for 30 minutes at 37°C, 5% CO 2 . Then, 5.10 4 HUVEC transfected or not were deposited on the upper chamber containing 500 μL of complete ECBM2 medium supplemented with 12% of FBS. Migration was stimulated by adding 1 mL of complete medium with or without CXCL12 at 6 nmol/L during 24 hours. At the end of the experiment, cells were fixed using 4% of paraformaldehyde, coloured using haematoxylin-Hemalum and quantification of migrated cells was performed under phase contrast microscope.

| Statistical analysis
All the results are presented with mean ± SEM. For statistical analysis, non-parametric tests were performed using GraphPad Prism software. Independent sample t tests (Mann and Whitney) were applied to compare two groups when the data followed a normal distribution and one-way analysis of variance (ANOVA) was used to compare among several groups. p <.05 indicated statistically significant differences.

| CXCL12-induced miR-126 expression in vitro and ex vivo
We and others showed that miR-126-3p regulates CXCL12 expression. Knowing that miR-126 is encoded by the egfl7 gene, and in order to study a potential reverse effect of CXCL12 on miR-126 expression, we first decided to study the egfl7 promoter activity. The results showed ( Figure 1A) that there was a significant increase of 2.35 ± 0.35-fold of promoter activity after stimulation by CXCL12 as compared to untreated cells. To confirm this result, we decided to analyse the miR-126-3p expression in HUVEC and in rat aortas ex vivo after CXCL12 (6 nmol.L -1 ) stimulation for 24 hours. The results showed that there was a significant increase of miR-126-3p level up

F I G U R E 1 CXCL12-induced miR-126-3p endogenous expression in vitro and ex vivo. (A)
To study the effect of CXCL12 on egfl7-miR-126 promotor activity, Huh7 were co-transfected with plasmid pGL3Basic-miR-126-EGFL7-Promoter (Addgene) and control pGL4.73 [hRluc/SV40] (Promega) for 24 hours. Then the cells were stimulated or not stimulated by CXCL12 for 24 h at 6 nmol.L -1 . Detection of luminescence was performed using Dual-Glo ® Luciferase Assay System (Promega). To study the effect of CXCL12 on miR-126 expression level, HUVEC (B) or rat aortas (C) were stimulated or not stimulated by CXCL12 for 24 h at 6 nmol.L -1 . After total RNA extraction, miR-126 level expression was analysed using qRT-PCR with U6 snRNA as endogenous control. The results are expressed as mean ± SEM. Three independent experiments were performed for in vitro experiments and six for ex vivo experiments. **p <.01 vs Untreated cells; *p <.05 vs Untreated aortas. To analyse the up and down regulation of miR-126-3p and miR-126-5p, HUVEC were transfected with 20 nmol.L -1 of premiR-126-3p, premiR-126-5p or inhibitors for 24 h. After total RNA extraction, miR-126-3p (D) and miR-126-5p (E) expression levels were analysed performing qRT-PCR using U6 snRNA as endogenous control. The results are expressed with mean ± SEM. Three independent experiments were performed. **p <.01 vs SCL; *p <.05 vs SCL.

A B C D E
to 88 ± 5% when HUVEC were stimulated with CXCL12 as compared to untreated cells ( Figure 1B). In addition, in our ex vivo model, there was a significant increase up to 48 ± 32-fold after CXCL12 stimulation as compared to untreated aortas ( Figure 1C).
Then, we studied the implications of both strands of miR-126 (miR-126-5p and miR-126-3p) in angiogenesis processes induced by CXCL12. First, we set up the conditions needed to up-regulate (premiR) and down-regulate (antimiR) both miRs strands in HUVEC.
The results showed that there was a significant increase of miR-126-3p level up to 4 ± 1.2-fold after premiR-126-3p transfection as compared to scramble negative control (SCL) ( Figure 1D). Our results demonstrated that CXCL12 enhanced miR-126 expression and we validated the up-and down-regulation of the various miRs species.
The second step of this work was to study the role of these miRs on HUVEC migration and two-dimensional (2D) angiogenesis test.
Then, we studied the effect of the various miRs treatments on 2Dangiogenesis by analysing the number of meshes formed by HUVEC on Matrigel layer. The most striking result was a significant increase of the number of meshes up to 3 ± 0.1 fold after premiR-126-3p transfection as compared to SCL ( Figure 2B). In contrast, there was no significant effect after transfection of premiR-126-5p only. The transfection of both premiR-126-3p/5p gave an intermediary result with a significant increase of the number of meshes up to 65 ± 16%.
However, this result was significantly lower than with transfection of premiR-126-3p only, suggesting an inhibitory effect of miR-126-5p on miR-126-3p pro-angiogenic action. In the case of the inhibitory treatments, there was a significant decrease of 2D-angiogenesis up to 42 ± 10% after transfection of antimiR-126-3p only as compared to SCL. Interestingly, the transfection with both antimiR-126-3p/5p led to a significant decrease of 53 ± 15% as compared to SCL. In contrast, there wasn't any effect on 2D-angiogenesis after transfection of antimiR-126-5p only.
Taken together, our data suggest that miR-126-3p/5p and miR-126-3p act as positive regulators of HUVEC migration and vascular tubes formation, but miR-126-5p did not seem to have any effect on this physiological process.

| miR-126-3p was implicated in CXCL12-induced HUVEC migration
The next step of our study was to analyse the impact of miR-126 deregulation on HUVEC migration, induced by CXCL12. Our results indicated ( Figure 3A) a significant increase of HUVEC migration up to 24 ± 16%, when the cells were stimulated by CXCL12 as compared to SCL. Moreover, there was a significant decrease of HUVEC migration up to 22 ± 8% after antimiR-126-3p transfection and CXCL12 stimulation as compared to stimulation with CXCL12 only.
In addition, the results showed that there was a significant increase of cell migration up to 25 ± 7% after premiR-126-5p transfection and CXCL12 stimulation as compared to the transfection with premiR-126-5p only. In contrast, there was no effect on HUVEC migration Our data suggest that miR-126-3p, but not miR-126-5p, was necessary in CXCL12-induced HUVEC migration.

| miR-126-3p was required for CXCL12-induced 2D-angiogenesis in vitro
The results showed ( Figure 3B)  Then, we studied the role of miR-126-3p and miR-126-5p separately in HUVEC 2D-angiogenesis. Our results showed ( Figure 3B) that there was a significant increase of this effect up  Our result suggests that miR-126-3p/5p and miR-126-3p but not miR-126-5p were necessary for CXCL12-induced angiogenesis ex vivo. In contrast, modulation (up or downregulation) of miR-126-5p seems to alter CXCL12 angiogenic properties ex vivo.

| SPRED-1 inhibition induced 2D-angiogenesis in vitro
Since the −5p species had no significant effects in our previous models, we decided to focus on the −3p species for the rest of the study.
To determine the essential role of miR-126-3p in CXCL12-induced angiogenesis, we focused on SPRED-1 expression, since SPRED-1 is a known target of miR-126-3p. First, we confirmed that there was a significant decrease of SPRED-1 protein level after premiR-126-3p transfection in HUVEC ( Figure 5A). We hypothesized that its downregulation is necessary for CXCL12-induced angiogenesis. To study this, we decided to analyse 2D-angiogenesis in HUVEC in the presence of siRNA-SPRED-1 after CXCL12 stimulation. We validated the siRNA-SPRED-1 efficiency showing the abolition of SPRED-1 mRNA expression compared to SCL ( Figure 5B). Second, we analysed 2Dangiogenesis in HUVEC after siRNA-SPRED-1 transfection. The results showed that there was a significant increase of 2D-angiogenesis up to 50 ± 15% after siRNA-SPRED-1 transfection as compared to HUVEC transfected with SCL ( Figure 5C).
Finally, our results suggest that SPRED-1 is a negative regulator of HUVEC's 2D-angiogenesis.

| CXCL12 decreased SPRED-1 expression leading to increase of 2D-angiogenesis
In the last part of this study, we wanted to know if downregulation of SPRED-1 by miR-126-3p is implicated in CXCL12-dependent angiogenesis.
Since we proved that CXCL12 increased miR-126-3p level in HUVEC, we hypothesized that CXCL12 can modulate the SPRED-1 expression. For the first time, the results showed that CXCL12 triggered a significant decrease of SPRED-1 protein level at 24 hours of treatment ( Figure 6A).
Interestingly, the results showed that siRNA-SPRED-1 transfection and CXCL12 stimulation led to a significant increase of 2D-angiogenesis up to 59 ± 9% as compared to CXCL12 alone ( Figure 6B, white arrows). Overall, these results demonstrated that inhibition of SPRED-1 potentializes CXCL12 pro-angiogenic effect. Moreover miR-126-3p was crucial for CXCL12-induced angiogenesis.
In conclusion, these results suggested that the pro-angiogenic effect of CXCL12 was dependent on miR-126-3p and an alternative signalling pathway parallel to SPRED-1. Despite the potentiating effect of the absence of SPRED-1, the presence of miR-126-3p was essential for the pro-angiogenic effect of CXCL12.

| D ISCUSS I ON
Angiogenesis is a physiological process, necessary for cardiovascular disorders regeneration, particularly after ischemic injuries. In this context, the pro-angiogenic factors such as chemokines and miRs can be implicated to stimulate angiogenesis. 22 In this study, we decided to focus on CXCL12 and its effect on EC through miRs modulation. Among miRs, miR-126, strongly expressed by EC, has been identified as a pro-angiogenic factor, which acts by decreasing SPRED-1 level and stimulating the Erk1/2 signalling pathway. 17 Like most miRs, miR-126 is produced from a double stranded duplex precursor that imbeds miR-126-3p and miR-126-5p complementary strands. 23 Depending on the tissue or cell type, the guide and passenger strands of a miR can act in synergy or as antagonists to regulate various biological processes. 13 For example, although they have different mRNA targets, miR-30-3p and miR-30-5p or miR-145-3p and miR-145-5p act in synergy in the tumour progression of glioma or bladder tumour cells. 24,25 Conversely, depending on their rate and location, miR-155-3p and miR-155-5p may act together or against each other in dendritic and astrocytic cells. 26,27 Therefore, depending on the tissue environment and pathophysiological conditions, the two strands (−3p or −5p) of the same miR may have different roles and this requires studying them simultaneously in order to know the involvement of each strands in a pathophysiological process. This is why we wanted to study the role of both strands (miR-126-3p and miR-126-5p) in angiogenesis. These roles have never been compared HUVEC. 36 In this context, we demonstrated that CXCL12 stimulation was associated with the activation of egfl7-miR-126 promoter.
Since Ets1/2 is known as a specific transcription factor for egfl7, 37 we could hypothesize that CXCL12 induced egfl7 transcriptional activity through an increase of Ets1/2 ( Figure S1). The egfl7 transcription start site contains 2 Ets binding sites that bind Ets1/2 transcription factor. 17 Mutation of the Ets binding element decreases promoter transactivation and decreases miR-126 expression. 37 It has been previously shown that different growth factors and chemokines activate the Ets transcription factor, for example: CXCL12-induced colorectal cancer cells migration via upregulation of Ets1. 38 Based on these studies and in light of our results we hypothesize that CXCL12 Our results and others 17 suggest that miR-126-3p had a strong pro-angiogenic potential. Our model showed that miR-126-5p alone had no effect on 2D-angiogenesis. However, its over-expression (after premiR-126-3p/5p co-transfection) reduced the pro-angiogenic effect of miR-126-3p in HUVEC. In contrast, Zhou et al, 41 found in retinal EC, that silencing the miR-126-3p repressed angiogenesis, while the overexpression of miR-126-5p increased angiogenesis. 41 We believe that this discrepancy was due to the differences between the experimental models and experimental conditions (cells types, presence of growth factors, and time of vascular tubes formation).
We further demonstrated that miR-126-3p is crucial for CXCL12induced migration and 2D-angiogenesis in both in vitro and ex vivo models. Indeed, we showed that in absence of miR-126-3p (after anti-miR-126-3p transfection) there was an abolition of CXCL12 pro-angiogenic properties. These data suggest that the presence of miR-126-3p is essential to stimulate the pro-angiogenic pathways induced by CXCL12. However, we observed conflicting results in our 2D-angiogenesis ex vivo model. Indeed, in the absence of CXCL12 stimulation, we observed that the inhibition of miR-126-3p leads to an increase of 2D-angiogenesis ex vivo. Interestingly, we and others have previously shown that the absence of miR-126-3p leads to CXCL12 synthesis and secretion in HUVEC culture medium. 21,42 In addition, since in our ex vivo experimental condition the miR transfection was done into the whole aorta, not only the EC but also the smooth muscle cells (SMC) and the fibroblasts could be transfected. In this context, it has been demonstrated by Jansen et al, 43 that inhibition of miR-126-3p in SMC leads to an increase of its proliferation. Furthermore, it has been shown that the absence of miR-126-3p can lead to VEGF-A synthesis. 44 Since CXCL12 and antimiR-126-3p have been previously shown to enhance VEGF-A expression [44][45][46] and both of them can stimulate PKC/Erk1/2 pro-angiogenic pathways through the stimulation of Raf protein, [47][48][49] we hypothesize that antimiR-126-3p could have proangiogenic action in our long-term ex vivo tissue culture model. In addition, since in ex vivo experiments the rat aortas were kept in ex vivo tissue culture for 9 days, we hypothesized that after antimiR-126 transfection there was an increase of SMC proliferation associated with VEGF synthesis, which could explain the pro-angiogenic effect in the absence of CXCL12. However, in the presence of CXCL12 we showed that miR-126-3p was crucial for the chemokine pro-angiogenic effects.
Then, we hypothesized that SPRED-1 (a miR-126-3p known target) 17  Our results showed for the first time that CXCL12 enhance miR-126-3p expression and its inhibition leads to a decrease of angiogenesis induced by CXCL12 in vitro. Moreover CXCL12 induced a decrease in SPRED-1 (miR-126-3p known target) and this downregulation improves CXCL12-induced angiogenesis in vitro. In this context, we hypothesized that CXCL12 induced miR-126-3p expression through the expression of Ets1/2 transcription factor complex channel remains inactivated. Therefore, the inhibition of SPRED-1 (after siRNA-SPRED-1 transfection) was not sufficient to restore the CXCL12 pro-angiogenic effects in antimiR-126-transfected HUVEC.
Taken together, under these conditions, it is clear that miR-126-3p plays a key role in CXCL12-induced activation of both Erk1/2 and PI3K/Akt pro-angiogenic pathways (Figure 7).

| CON CLUS ION
In conclusion, in this study we focused on two pro-angiogenic (c) SPRED-1 was implicated in CXCL12-induced angiogenesis; (d) miR-126-3p enhanced cell migration and vascular tubes formation, in HUVEC, however, the miR-126-5p had no effect on both processes.

ACK N OWLED G EM ENTS
This work was supported in collaboration by the Direction de Carole PLANES for giving us access to their animals in order to allow us to carry out our experiments.

CO N FLI C T O F I NTE R E S T
The author declares that there is no conflict of interest. Writing-review & editing (lead).

DATA AVA I L A B I L I T Y S TAT E M E N T
Data supporting the findings of this study could be obtained from the corresponding author upon reasonable request.