Delivery of MicroRNA-126 by Apoptotic Bodies Induces CXCL12-Dependent Vascular Protection

See allHide authors and affiliations

Science Signaling  08 Dec 2009:
Vol. 2, Issue 100, pp. ra81
DOI: 10.1126/scisignal.2000610


Apoptosis is a pivotal process in embryogenesis and postnatal cell homeostasis and involves the shedding of membranous microvesicles termed apoptotic bodies. In response to tissue damage, the CXC chemokine CXCL12 and its receptor CXCR4 counteract apoptosis and recruit progenitor cells. Here, we show that endothelial cell–derived apoptotic bodies are generated during atherosclerosis and convey paracrine alarm signals to recipient vascular cells that trigger the production of CXCL12. CXCL12 production was mediated by microRNA-126 (miR-126), which was enriched in apoptotic bodies and repressed the function of regulator of G protein (heterotrimeric guanosine triphosphate–binding protein) signaling 16, an inhibitor of G protein–coupled receptor (GPCR) signaling. This enabled CXCR4, a GPCR, to trigger an autoregulatory feedback loop that increased the production of CXCL12. Administration of apoptotic bodies or miR-126 limited atherosclerosis, promoted the incorporation of Sca-1+ progenitor cells, and conferred features of plaque stability on different mouse models of atherosclerosis. This study highlights functions of microRNAs in health and disease that may extend to the recruitment of progenitor cells during other forms of tissue repair or homeostasis.


Apoptotic cell death is a process that is important not only for embryonic development but also for cell homeostasis in adults, and it has been implicated in many diseases, including atherosclerosis (16). In the course of apoptotic programs, a series of molecular events culminates in fragmentation of DNA and blebbing of the cell membrane that is accompanied by the release of small membranous particles from the cell surface, which are termed apoptotic bodies (7, 8). Endothelial cell–derived microparticles released in response to proapoptotic stimuli can originate from atherosclerotic plaques, and their abundance in the circulation is associated with endothelial dysfunction and arterial stiffness, indicating that they may serve as a prognostic marker for atherosclerotic vascular disease (8). The release of these microparticles has been viewed as an attempt by the cell to reverse or escape death by removing parts of the caspase machinery and to prevent “eat-me” signals attracting phagocytes for cell elimination (2, 3); however, a function of microparticles in paracrine signaling to neighboring cells to modulate immune cell responses or repair mechanisms has not been fully elucidated.

Whereas apoptotic bodies induce proliferation and differentiation of endothelial progenitor cells in vitro, which implies a role in tissue repair and angiogenesis (7, 9), they can also be engulfed by phagocytes, which triggers the secretion of cytokines or growth factors, such as vascular endothelial growth factor (VEGF) (10). In response to DNA damage or hypoxia and in the context of arterial injury, the CXC chemokine CXCL12 is involved in the recruitment, through its receptor CXCR4, of progenitor cells (which can be identified by the presence of Sca-1 and the absence of lineage markers on their surface) from the bone marrow for tissue repair or angiogenesis and in the defense and survival programs that counteract processes related to apoptosis (1115). Although the underlying mechanisms remain unclear, smooth muscle cell (SMC)–derived apoptotic bodies mediate the secretion of CXCL12 by vascular SMCs (13). Here, we investigated the sequence of events involved in the production of CXCL12 by apoptotic bodies in vitro and determined their role in atherosclerosis and plaque stability in vivo.


Endothelial apoptotic bodies induce the expression of CXCL12 in recipient cells

Apoptotic bodies can be incorporated by endothelial progenitor cells and can alter their function (7). Likewise, human umbilical vein endothelial cells (HUVECs) took up annexin V–labeled apoptotic bodies derived from serum-starved apoptotic HUVECs. Uptake involved a rearrangement of cytoskeletal fibers, as visualized after 2 hours by confocal microscopy and costaining for F-actin (Fig. 1A). In response to endothelial apoptotic bodies, the transcription of CXCL12 was increased, as determined by real-time reverse transcription–polymerase chain reaction (RT-PCR) analyses of HUVECs, SMCs, and mouse aortic endothelial cells (ECs) and by luciferase reporter assays in HUVECs (Fig. 1, B and C). Induction of CXCL12 expression was time dependent, observed as early as 6 hours after uptake of the apoptotic bodies (Fig. 1D), and resulted in a marked increase in the abundance of intracellular and secreted CXCL12 protein compared with that of vehicle-treated HUVECs, as determined by enzyme-linked immunosorbent assay (ELISA) (Fig. 1E).

Fig. 1

Apoptotic bodies (AB) convey functional RNA to ECs to induce the expression of CXCL12. (A) Confocal microscopic analysis of F-actin fibers (red) in HUVECs incubated with annexin V+ AB (green). (B) Real-time RT-PCR analysis of CXCL12 expression in HUVECs (n = 6), SMCs (n = 2), and mouse ECs (mEC, n = 3) that were untreated or treated with AB and in HUVECs (n = 2) that were untreated or treated with AB derived from SMCs (SAB). (C) CXCL12-luciferase promoter activity in HUVECs that were untreated or treated with AB (n = 4). (D) Representative RT-PCR analysis of the abundance of CXCL12 mRNA in HUVECs that were untreated or treated with AB for the indicated periods (n = 3). (E) The secretion of CXCL12 protein from HUVECs that were untreated or treated with AB was measured by ELISA (n = 3). (F) Representative RT-PCR analysis of the abundance of CXCL12 mRNA in AB, apoptotic HUVECs, untreated HUVECs, and HUVECs incubated with medium conditioned with AB, lipid extracts from AB, or sonicated AB pretreated as indicated (n = 3). (G) Representative flow cytometric analysis of untreated HUVECs and of HUVECs treated for the indicated times with AB derived from HUVECs that were transfected with FITC+ oligonucleotides (H) and of pEGFP-transfected HUVECs that were untreated or treated with AB (1× or 3×) from HUVECs transfected with siRNA specific for EGFP (n = 3). Data represent the mean ± SEM. MFI, mean fluorescence intensity. *P < 0.05; **P < 0.01; ***P < 0.005.

Cells undergoing apoptosis can secrete chemoattractants, for example, the phospholipid lysophosphatidylcholine, or eat-me signals for phagocytes (3, 16). The increase in the abundance of CXCL12 messenger RNA (mRNA) in treated HUVECS was seen after treatment with sonicated apoptotic bodies but not with apoptotic body–conditioned supernatants, indicating that apoptotic bodies do not secrete crucial mediators but contain them instead (Fig. 1F). We then investigated the constituents of apoptotic bodies that were required for the induction of CXCL12 expression. Treatment of cells with lipid extracts of apoptotic bodies did not increase their expression of CXCL12. In addition, pretreatment of apoptotic bodies with deoxyribonuclease (DNase) or proteinase K did not interfere with their ability to induce CXCL12 expression in recipient cells. Together, these data exclude lipids, DNA, and protein components of apoptotic bodies as inducers of CXCL12 expression (Fig. 1F). In contrast, apoptotic bodies pretreated with ribonuclease (RNase) failed to increase the expression of CXCL12 in recipient cells (Fig. 1F), indicating that an RNA component was the responsible mediator. Substantial amounts of CXCL12 transcripts were detectable in HUVECs and apoptotic HUVECs but not in apoptotic bodies, revealing that CXCL12 mRNA was not directly transferred to the recipient cells by the apoptotic bodies but was induced by them (Fig. 1F).

Horizontal transfer of whole genes and mRNA through exosomes or progenitor cell–derived microvesicles does occur, for example, in the activation of angiogenic programs in ECs (9, 1719). We verified by flow cytometry that fluorescein isothiocyanate (FITC)–labeled oligonucleotides could be conveyed to HUVECs by apoptotic bodies derived from HUVECs transfected with an FITC+–oligo-22-mer (Fig. 1G). Moreover, treatment of HUVECs transfected with a plasmid encoding an enhanced green fluorescent protein (pEGFP) with apoptotic bodies collected from HUVECs transfected with a pEGFP-specific short inhibitory RNA (siRNA) dose-dependently reduced their fluorescence (Fig. 1H), showing the transfer of functional RNA from apoptotic bodies to recipient cells.

MicroRNA-126 in endothelial apoptotic bodies mediates the induction of CXCL12 expression

MicroRNAs (miRNAs) constitute a class of highly conserved noncoding RNAs that control gene expression by either inhibiting the translation of mRNAs or destabilizing them, and participate in the regulation of diverse cellular processes, including apoptosis, proliferation, hematopoiesis, and angiogenesis (2022). To investigate whether apoptotic bodies might contain miRNAs that could account for the increased expression of CXCL12 after their uptake by ECs, the prevalence of miRNAs in HUVEC-derived apoptotic bodies was analyzed by chip arrays, which revealed a distinct expression profile for several miRNAs (Fig. 2A). Notably, microRNA-126 (miR-126) was the most abundant miRNA in apoptotic bodies, which was confirmed by quantifying absolute copy numbers by real-time RT-PCR (Fig. 2, A and B). Adding to studies that showed the presence of miR-126 in HUVECs (23), our data revealed that miR-126 was among the few miRNAs enriched in apoptotic bodies (log2 ratio >1) compared to untreated HUVECs (Fig. 2A). In contrast, let-7c was highly abundant in apoptotic bodies, but was not enriched in apoptotic bodies compared to HUVECs (Fig. 2A). Notably, treatment of HUVECs with endothelial apoptotic bodies markedly (by twofold) increased the abundance of miR-126 in HUVECs (Fig. 2C), confirming the efficient transfer of miR-126 to recipient cells.

Fig. 2

Apoptotic bodies (AB) carry miR-126, which represses the expression of RGS16 and enhances CXCR4 signals in recipient cells. (A) Profile of miRNAs found in AB with greater than fourfold increase in abundance over baseline (log2 >2) and ratios for those miRNAs that were greater than twofold enriched in abundance (log2 >1) in AB compared to HUVECs (right column) as analyzed by bioarray chips. AU, arbitrary units. (B) Quantification of the absolute amounts of miRNAs by real-time RT-PCR. (C) Abundance of miR-126 in untreated HUVECs and in HUVECs treated with AB for 1 hour (n = 3). (D) Luciferase activity in HUVECs cotransfected with pre–miR-126 and pMIR-Report plasmids carrying miR-126–binding sites found in the 3′UTR of indicated target genes downstream of luciferase is reduced compared to that in HUVECs transfected with empty pMIR-control (n = 3). (E and F) The abundance of CXCL12 mRNA and CXCL12 protein in HUVECs transfected with pre–miR-126, anti–miR-126, let-7c, or scrambled control sequences (scr), as determined by real-time RT-PCR and ELISA (n = 2 to 5). (G) The abundance of RGS16 protein in HUVECs transfected with pre–miR-126, anti–miR-126, let-7c, or scr miRNAs or in HUVECs that were untreated or treated with AB from anti–miR-126– or mock-transfected HUVECs was detected by Western blotting and compared to that of β-actin (densitometric data of n = 2 to 5 experiments). (H and I) Real-time RT-PCR analysis of CXCL12 mRNA expression (n = 2 to 5) in HUVECs transfected with siRNA against RGS16 or SPRED1 or with pCMV-RGS16 compared to that in HUVECs transfected with mock sequence controls (H) and in untreated HUVECs or in HUVECs treated with AB from anti–miR-126– or mock-transfected HUVECs, HUVECS treated with CXCL12 or AMD3100, or in HUVECS transfected with pre–miR-126 and pCMV-RGS16 (I). (J) Western blotting analysis of the abundance of total ERK1/2 and CXCL12-induced pERK1/2 in HUVECs treated with AB or AMD3100 or transfected with pre–miR-126 or scr miRNA (densitometric data of n = 3 experiments). Data represent the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

Surveying bioinformatic databases (for example, miRanda) based on thermodynamic binding efficiency (24) identified the mRNAs of CXCL12, vascular cell adhesion molecule–1 (VCAM-1), and members of the regulator of G protein (heterotrimeric guanine nucleotide–binding protein) signaling (RGS) family, including RGS3 and RGS16, as possible molecular targets of miR-126. Human and mouse miR-126 are identical and have complementary binding sites in RGS16 and Rgs16 mRNAs (fig. S1). In addition to VCAM-1 (23), sprouty-related protein 1 (SPRED1) is also a target of miR-126 (25, 26). To investigate whether miR-126 controls the expression of these targets, we cloned their predicted miR-126–binding sites (or mutated control sequences) into the 3′ untranslated region (UTR) downstream of luciferase in the pMIR-Report plasmid. Cotransfection of HUVECS with pre–miR-126 and individual pMIR-Report plasmids containing miR-126–binding sites repressed the expression of luciferase where the miR-126–binding sites of SPRED1, VCAM-1, RGS16, and CXCL12 were used (Fig. 2D), which indicated that these genes were targets of miR-126. Cotransfection with reporter plasmids containing antisense sequences of RGS16 or CXCL12 had no effect on the expression of luciferase (fig. S2). HUVECs transfected with pre–miR-126, but not with anti–miR-126, scrambled control miRNA, or let-7c showed increased expression of CXCL12 and production of CXCL12 protein in supernatants (Fig. 2, E and F).

miR-126 targets RGS16 to enable induction of CXCL12 expression through CXCR4

Our data indicated that targets of miR-126 other than CXCL12 itself, for example, RGS16, were relevant for the expression of CXCL12. Accordingly, apoptotic bodies derived from SMCs, which contained trace amounts of miR-126 (table S1), induced a more moderate increase in the abundance of CXCL12 mRNA in HUVECs than did apoptotic bodies from ECs (Fig. 1B), which may be mediated by signals or mechanisms unrelated to the transfer of endothelial-specific miR-126. Moreover, RGS16, a negative regulator of the CXCL12 receptor CXCR4 (27), was considerably more abundant than RGS3 in HUVECs, which suggested that RGS16 was the predominant RGS family member in HUVECs. In contrast, RGS3 and RGS16 were similarly abundant in Jurkat T cells (fig. S3).

As determined by Western blotting analysis, the abundance of RGS16 in HUVECs was reduced by transfection with pre–miR-126 but not with anti–miR-126, scrambled control miRNA, or let-7c. The abundance of RGS16 in HUVECs was also reduced by treatment with apoptotic bodies, but was increased by treatment with apoptotic bodies derived from HUVECs that were transfected with anti–miR-126 (Fig. 2G). Of note, siRNA-mediated knockdown of RGS16, but not of SPRED1, enhanced the expression of CXCL12 in HUVECs (Fig. 2H). Increasing the abundance of RGS16 in transfected HUVECs (fig. S4) almost completely eliminated CXCL12 mRNA, indicating that RGS16 acted as a strong suppressor of the expression of CXCL12 (Fig. 2H). These data implied that RGS16 might control CXCL12 homeostasis through its actions on CXCR4. Indeed, increasing the abundance of RGS16 in HUVECS or blocking CXCR4 with the small-molecule antagonist AMD3100 inhibited the increased expression of CXCL12 that was induced by apoptotic bodies or pre–miR-126, which suggested the involvement of RGS16-restricted, CXCR4-mediated signals in the regulation of CXCL12 expression (Fig. 2I). Apoptotic bodies derived from HUVECs transfected with anti–miR-126 but not from mock-transfected HUVECs did not induce, but reduced, the expression of CXCL12 compared to that in control cells, suggesting that the transfer of miR-126 underlay the effect of apoptotic bodies on the expression of CXCL12 in recipient cells (Fig. 2I). In addition, CXCL12 induced the increased expression of CXCL12 in HUVECs through CXCR4, as shown by inhibition of the effects of CXCL12 by AMD3100 (Fig. 2I), thus establishing an autoregulatory feedback loop. This was supported by our demonstration that CXCL12-induced phosphorylation of extracellular signal–regulated kinases (ERKs), which was mediated by CXCR4 and inhibited by AMD3100, was amplified by treatment with apoptotic bodies or pre–miR-126 and by silencing the expression of RGS16 but not that of SPRED1 (Fig. 2J and fig. S5). Reciprocal effects of RGS16 and SPRED1 on the activity of ERKs in response to CXCL12 and VEGF (25, 26) may be due to differences in the signaling pathways used by their receptors. These data suggest that RGS16 is the relevant target of miR-126 in HUVECs.

Endothelial apoptotic bodies induce the expression of Cxcl12 and the recruitment of progenitor cells in mice with atherosclerosis

Low concentrations of CXCL12 in the plasma are associated with unstable coronary artery disease, which suggests that the anti-inflammatory or plaque-stabilizing properties of CXCL12 may play a role in human atherosclerosis (28). Indeed, the CXCL12-CXCR4 axis exerts direct atheroprotective effects in apolipoprotein E–deficient (Apoe−/−) mice by controlling the homeostasis of myeloid cells (29). Endothelial damage is thought to precede the development of atherosclerotic lesions at predilection sites (5, 6, 8, 30) and is associated with formation and release of annexin V–containing apoptotic bodies. We performed scanning electron microscopy of HUVECs with immunogold-labeled annexin V, which revealed the presence of annexin V–containing detachments forming from cytoplasmic buds and of protrusions in apoptotic HUVECs but not in untreated control cells (fig. S6). Indeed, an increased concentration of circulating apoptotic microparticles correlates with impaired endothelial function in coronary artery disease (31). Feeding Apoe−/− mice a high-fat diet increased the amounts of circulating annexin V+CD31+ apoptotic bodies at 6 weeks (and even further at 30 weeks) compared to that in Apoe−/− mice fed normal chow (Fig. 3A). Given the link between endothelial apoptotic bodies and the production of CXCL12, which mobilizes progenitor cells from the bone marrow, we tested whether intravenous injection of apoptotic bodies could mobilize progenitor cells to peripheral blood. Indeed, flow cytometry revealed that the number of circulating Sca-1+ lineage progenitor cells increased in Apoe−/− mice 1 day after they were injected with apoptotic bodies (Fig. 3B). The numbers of monocytes and neutrophils in the blood were unaltered by injection with apoptotic bodies (fig. S7), indicating the specificity of apoptotic bodies in mobilizing progenitor cells.

Fig. 3

Mobilization of progenitor cells and induction of CXCL12 by apoptotic bodies (AB) in vivo. (A) Flow cytometric analysis of the number of circulating AB in Apoe−/− mice on normal chow or on a high-fat diet (n = 5) and (B) the number of circulating Sca-1+ lineage progenitor cells after injection of mice with PBS (control) or AB (n = 5). (C to E) Apoe−/− mice were fed a high-fat diet for 6 weeks, injected twice weekly with PBS or AB, and treated with or without AMD3100 (n = 7 to 8). (C) Quantification of Sca-1+ cells and (D) representative images of double-immunofluorescence staining for Sca-1 and CD31 in the aortic root of control or AB-treated Apoe−/− mice. (E) Quantification and representative images of immunofluorescence of Cxcl12 in aortic root plaques; nuclei were detected by DAPI. Data represent the mean ± SEM. *P < 0.05; **P < 0.01.

Repetitive injections of apoptotic bodies into Apoe−/− mice twice weekly while they were being fed a high-fat diet for 6 weeks increased the luminal incorporation of Sca-1+ cells into aortic root plaques in a CXCR4-dependent manner, because continuous application of AMD3100 by osmotic minipumps inhibited the process (Fig. 3C). Double immunofluorescence staining revealed the colocalization of Sca-1 with luminal CD31+ ECs (Fig. 3D) but the presence of few Mac-2+ macrophages (fig. S8) in aortic root lesions, which suggested that apoptotic bodies primarily recruited CD31+ endothelial progenitor cells for repair during early plaque growth. Accordingly, prolonged treatment with apoptotic bodies induced the expression of Cxcl12 foremost in the luminal cells of aortic root plaques after 6 weeks of high-fat diet in Apoe−/− mice (Fig. 3E).

Endothelial apoptotic bodies protect mice from atherosclerosis

Prolonged treatment with apoptotic bodies reduced atherosclerotic plaque size in the aortic root of Apoe−/− mice, which was associated with decreased numbers of macrophages and apoptotic cells compared to those of control-treated mice (Fig. 4, A and B). These effects were reversed by the application of AMD3100 (Fig. 4, A and B). To verify that the antiatherogenic mechanism exerted by HUVEC-derived apoptotic bodies could be translated into similar effects by apoptotic bodies generated in atherosclerotic mice or patients, we isolated endothelial apoptotic bodies carrying miR-126 (fig. S9) from the atherosclerotic plaques of patients (32) undergoing endarterectomy of the carotid arteries. Systemic treatment of Apoe−/− mice on a high-fat diet for 6 weeks with atherosclerotic plaque–derived apoptotic bodies resulted in the reduction of plaque area and macrophage content in the aortic root in comparison to those of control-treated mice (Fig. 4, C and D), indicating that endothelial apoptotic bodies in general serve as a compensatory signal to confer atheroprotection.

Fig. 4

Apoptotic bodies (AB) protect against the early formation of atherosclerotic lesions. (A and B) Apoe−/− mice fed a high-fat diet for 6 weeks and injected twice weekly with PBS or AB were untreated or were treated with AMD3100 (n = 7 to 8). (A) Representative images of aortic roots and quantification of the (Oil Red O)+ atherosclerotic lesion area and (B) quantification of the numbers of macrophages and TUNEL+ apoptotic cells in plaques. (C and D) Apoe−/− mice fed a high-fat diet for 6 weeks and injected once weekly with PBS or human plaque–derived AB (n = 7 to 9). (C) Representative images and quantification of (Oil Red O)+ atherosclerotic plaque areas and (D) macrophage content. Data represent the mean ± SEM. *P < 0.05; **P < 0.01.

Collectively, our data support the notion that apoptotic bodies released by apoptotic ECs may function as a paracrine alarm system to mobilize progenitor cells for endothelial repair, evoking a protective response in diet-induced atherosclerosis, which may be related to the induction of CXCL12 as an antiapoptotic survival factor (33). Similar to the effects of CXCL12, treatment with apoptotic bodies protected HUVECs from apoptosis and induced their proliferation (fig. S10), similar to findings in endothelial progenitor cells treated with HUVEC-derived apoptotic bodies (7). In contrast, apoptotic bodies had no effects on apoptosis or proliferation of SMCs or Jurkat T cells (fig. S8), which suggested that there was a cell type–specific susceptibility to endothelial apoptotic bodies or that apoptotic body–induced expression of CXCL12 was a predominant response of ECs, which may be related to the endothelial origin of miR-126. In addition, miR-126 inhibits the expression of VCAM-1 in ECs, thereby limiting the recruitment of inflammatory cells (23). This may contribute to the reduced influx of monocytes in apoptotic body–treated Apoe−/− mice, which might explain the partial modulation of the plaque phenotype by blockade of CXCR4.

miR-126 mediates the atheroprotective effects of endothelial apoptotic bodies

To scrutinize the local effects of apoptotic body–conveyed miR-126 on the extent and quality of advanced atherosclerosis, we used a model of collar-induced accelerated plaque formation mediated by proximal flow reduction (34). Carotid arteries of Apoe−/− mice were transduced with miR-126, scrambled control miRNA, or let-7c embedded in pluronic gel directly after collar placement. Carotid arteries treated with miR-126, but not let-7c, displayed a marked reduction in plaque area and an increase in the expression of Cxcl12 compared to those treated with control miR (Fig. 5, A and B). As in mice that underwent long-term treatment with apoptotic bodies, this outcome was associated with reduced numbers of macrophages and apoptotic cells in miR-126–treated arteries and with an increased number of intimal SMCs compared to those of control miR–treated arteries (Fig. 5C), which may be attributable to a CXCL12-mediated recruitment of progenitor cells (13). In addition, collagen content, as evident by Sirius red staining, was higher in miR-126–treated than in control miR–treated arteries (Fig. 5D), consistent with a more stable plaque phenotype.

Fig. 5

Transfer of miR-126 or apoptotic bodies (AB) carrying miR-126 to mice reduces the extent of plaque formation. (A to D) Collar-induced plaque formation in carotid arteries of Apoe−/− mice 4 weeks after transduction with scrambled control (scr), pre–miR-126, or let-7c (n = 7 to 8). (A) Representative images and quantification of pentachrome-stained plaque areas are shown and (B) the relative abundance of Cxcl12 mRNA and 18S rRNA in carotid arteries was determined by real-time RT-PCR. (C) Quantification of the macrophage, TUNEL+ apoptotic cell, and SMC content of plaques and (D) representative images of Sirius red staining for collagen. Data represent the mean ± SEM. *P < 0.05; **P < 0.01. (E to G) Carotid arteries of Apoe−/− mice fed a high-fat diet for 10 weeks were intraluminally incubated with or without AB isolated from miR-126+/+ or miR-126−/− mice 5 weeks before being killed (n = 5 to 6). (E) Representative images and quantification of pentachrome-stained plaque areas and (F) of the macrophage and SMC content. (G) The relative abundance of Cxcl12 mRNA and 18S rRNA as determined by real-time RT-PCR and the abundance of Rgs16 protein as analyzed by Western blotting and densitometry are shown under the indicated conditions. Data represent the mean ± SEM. *P < 0.05; **P < 0.01; ns, not significant.

To study the effects of local apoptotic body–mediated delivery of miR-126 in the context of diet-induced atherosclerosis, we developed a model for local intraluminal exposure of carotid arteries of Apoe−/− mice to endothelial apoptotic bodies isolated from miR-126+/+ or miR-126−/− mice (fig. S9). The results from these experiments indicated that miR-126–carrying but not miR-126–deficient apoptotic bodies conferred a protection against diet-induced atherosclerosis in the carotid artery (Fig. 5E), which was associated with a reduced infiltration of macrophages and an increased number of SMCs (Fig. 5G). A marked induction of Cxcl12 expression in carotid arteries incubated locally with miR-126–carrying, but not miR-126–deficient, mouse apoptotic bodies was accompanied by a reduction in the abundance of RGS16 (Fig. 5F). Collectively, these data establish that the atheroprotective effects of apoptotic bodies are mediated by miR-126, which targets RGS16.


Our results unravel a previously uncharacterized mechanism by which apoptotic bodies derived from ECs convey alarm signals to surviving cells in the vicinity to induce production of the chemokine CXCL12, which acts as a signal to mobilize progenitor cells and as an anti-apoptotic factor. The underlying process involved the removal of a block on an autoregulatory, self-amplifying feedback loop, by which CXCL12, through CXCR4, triggered the expression of CXCL12 and the secretion of CXCL12 protein. This loop was set in motion by the transfer of miR-126, which was enriched in endothelial apoptotic bodies and caused knockdown of the negative regulator RGS16 to “unlock” CXCR4, thus driving expression of CXCL12. In addition, endothelial miR-126 governs vascular integrity and angiogenesis by regulating the responses of ECs to VEGF. Targeted deletion of miR-126 in mice and its knockdown in zebrafish causes vascular leakage during embryonic development and impaired neovascularization after infarction due to defects in the proliferation and migration of ECs and angiogenesis. In a related mechanism, miR-126 enhances the VEGF pathway by repressing SPRED1, which is an intracellular inhibitor of angiogenic kinase signaling (25, 26).

In the context of atherosclerosis, apoptotic bodies are released into the circulation, promote the mobilization and incorporation of Sca-1+ progenitor cells during plaque formation, and limit atheroprogression in a CXCR4-dependent process. Further, apoptotic bodies carrying miR-126 (or local transfer of miR-126) reduce the macrophage and apoptotic cell content of plaques but increase the expression of Cxcl12 and the number of SMCs, which leads to smaller lesions with a less inflammatory phenotype overall. Notably, apoptotic bodies that carried miR-126 repressed arterial expression of Rgs16, whereas apoptotic bodies deficient in miR-126 failed to reduce the abundance of RGS16 and could not increase the expression of CXCL12. These findings establish the mechanistic relevance of miR-126–mediated knockdown of RGS16 in vivo.

In conclusion, we propose a mechanism for an autoregulatory model of the atheroprotective production and function of CXCL12 (Fig. 6). This protective mechanism may not only contribute to vascular repair by taming atherosclerosis, but may also extend to other forms of tissue damage or homeostasis in health and disease that require the influx of progenitor cells. Beyond the findings that miR-126 governs vascular integrity during embryogenesis, our data identify a fundamental adaptive principle for postnatal tissue repair and homeostasis and epitomize the importance of regulatory miRNA functions (2022).

Fig. 6

Mechanism of action of apoptotic body–induced recruitment of progenitor cells to atherosclerotic plaques. Apoptosis involves fragmentation of DNA, membrane blebbing, and the shedding of apoptotic bodies. Endothelial cell–derived apoptotic bodies can be transferred to recipient cells to induce the expression of CXCL12. This is mediated through miRNA-126, which is enriched in the apoptotic bodies and acts by knocking down the negative regulator RGS16 and enabling CXCR4 to stimulate an autoregulatory feedback loop that enhances the phosphorylation of ERK1/2 and induces the production of more CXCL12. Apoptotic bodies promote the mobilization and incorporation of Sca-1+ progenitor cells to the plaque. Transfer of apoptotic bodies or miR-126 causes a reduction in diet-induced atherosclerosis and collar-induced plaque formation. miR-126 also reduces the expression of VCAM-1 and SPRED1 in ECs (23, 25, 26).

Materials and Methods

Cell culture, collection of apoptotic bodies, and pretreatments

Human aortic smooth muscle cells (PromoCell) were cultured in SMC-2 growth medium (PromoCell) as described (35). HUVECs and aortic ECs isolated from Apoe−/− mice (36) were cultured in EC growth medium (PromoCell) and were used in experiments when they were between passages 4 and 5. Apoptotic bodies were isolated by sequential centrifugation steps (7) from supernatants of apoptotic HUVECs that had been starved of serum and growth factors for 24 hours. To collect apoptotic bodies after the transfection of HUVECs, apoptosis was induced after 4 hours. Apoptotic bodies were treated with DNase (Qiagen), RNase (Fermentas), or proteinase K (Sigma) for 30 min after they had been sonicated for 20 s. Lipids were extracted from apoptotic bodies according to Bligh and Dyer (37) with two sequential extraction steps in methanol-chloroform, and the lipids were collected by centrifugation of the lower chloroform phase after evaporation with N2 at 4000g for 10 min at 10°C. HUVECs (5 × 105) were treated with apoptotic bodies derived from 2 × 106 apoptotic HUVECs or with CXCL12 (100 ng/ml; R&D Systems) or AMD3100 (1 μg/ml) (29) for 6 hours unless indicated otherwise. Apoptotic bodies were also collected from human atherosclerotic plaque material (32) and mouse whole blood by FACS sorting. Human carotid artery endarterectomy specimens were rinsed in sterile phosphate-buffered saline (PBS) with streptomycin-penicillin (100 U/ml), minced with fine scissors, and vigorously resuspended. After centrifugation at 800g for 10 min at 20°C to remove debris, apoptotic bodies were pelleted at 16,000g for 20 min at 4°C before incubation with the appropriate antibodies. EDTA-buffered whole blood obtained from wild-type (WT) or mi-R126−/− mice (26) was centrifuged at 800g for 10 min at 20°C to obtain apoptotic body–rich plasma for antibody staining. Samples were diluted in staining buffer (BD Bioscience) and incubated with FITC-conjugated annexin V (BD Biosciences) and with allophycocyanin (APC)–conjugated antibody against human CD31 (eBioscience) or phycoerythrin (PE)-Cy7–conjugated antibody against mouse CD31 (Abcam). Annexin V+ CD31+ apoptotic bodies were sorted with a FACSAria and analyzed with FACSDiva software (BD Biosciences).

Chip array

Total RNA was isolated from apoptotic bodies or HUVECs, and miRNAs were purified with the mirVana miRNA Isolation Kit (Ambion). miRNAs obtained from 10 μg of total RNA were labeled with the mirVana miRNA Labeling Kit (Ambion) and fluorescent Cy3 (Molecular Probes), and hybridized to the Ambion mirVana miRNA Bioarray (1566 v.1; Hybridized mirVana miRNA Bioarrays were scanned and quantified with ImaGene 5.5.4 (Bio Discovery). Microarray data have been uploaded to the GEO (Gene Expression Omnibus) database ( Resulting signal intensities were corrected for background and normalized by variance stabilization (VSN package, Bioconductor project). To calculate the relative abundance of miRNAs in apoptotic bodies, an average baseline was calculated with values of the empty spots. Sample spots with fourfold or higher intensities over baseline (log2 ≥2) were selected for illustration in the graph. Ratios of miRNAs enriched in apoptotic bodies with greater than twofold intensity (log2 ≥1) compared to that in HUVECs are depicted in the right column.

RT-PCR and real-time RT-PCR

The expression of specific miRNAs in apoptotic bodies was analyzed by quantitative miR stem-loop RT-PCR technology (Ambion). Highly target-specific stem-loop structure and RT primer sets (miR-126, AM30023; miR-let7i, AM30004; miR-let7c, AM30002; and U6, AM30303, all from Ambion) were used for the amplification of miRNAs, enabling only mature miRNA targets to form RT primer–mature miRNA chimeras extending the 5′ end of the miRNA. The absolute expression of human mature miR-126, let-7c, and let-7i miRNAs was determined by obtaining standard curves with different concentrations of mature miRNA oligonucleotides (Sigma) as templates. With specific RT and real-time primer sets, the standard curve was generated and CT values and oligonucleotide concentrations were plotted. The copy number was calculated from the oligonucleotide concentration and the molecular mass of the transcript (38). The efficiencies of primers for miR-126, miR-let7i, and miR-let7c were similar, as calculated according to the equation E = 10(−1/slope) with the slope obtained from the standard curves. Isolated RNA (Qiagen) was reverse-transcribed into complementary DNA (cDNA) with Mo-MLV RT (Invitrogen). PCR was performed with 20 ng of cDNA, Taq polymerase (Promega), and specific primers (aldolase, 5′-AGCTGTCTGACATCGCTCACG-3′, 5′-CACATACTGGCAGCGCTTCAAG-3′; CXCL12, 5′-CGCCACTGCCTTCACCTCCTC-3′, 5′-GGCATACATAGGCTTCAGAGGCAATC-3′). Products were separated by agarose gel electrophoresis. Real-time RT-PCR was performed with the QuantiTect SYBR-Green PCR kit (Opticon MJ Research) or Taqman Gene Expression Master Mix (Applied Biosystems) and specific primer pairs (18S rRNA, Hs03003631_g1, 4333760F; CXCL12, Hs00171022_m1 and Mm00445552-m1, all Applied Biosystems; and RGS16, 5′-AGGGCACACCAGATCTTTGA-3′, 5′-GTCTGCAGGTTCATCCTCGT-3′).

Transfection of HUVECs and luciferase assays

Transfection of HUVECs was performed with the HUVEC Nucleofector Kit (Amaxa) according to the manufacturer’s protocol. To investigate the transfer of functional RNA by apoptotic bodies, HUVECs (2 × 106) were transfected with silencer EGFP-specific siRNA (1.5 μg, AM4626, Ambion). After 6 hours of recovery, apoptosis was induced by serum starvation and apoptotic bodies were isolated and added to HUVECs (0.5 × 106 to 1 × 106) transfected with pEGFP (3 μg, Clontech). Gain- and loss-of-function assays were performed in HUVECs (0.5 × 106 to 1 × 106) transfected with 200 pmol (~2.5 μg) of pre–miR-126, anti–miR-126, let-7c, or a scrambled miR control (Ambion), RGS16 siRNA, SPRED1 siRNA, or a scrambled control siRNA (Santa Cruz), or with 2 μg of pCMV-scrambled or pCMV-RGS16 after 48 to 72 hours of recovery. RGS16 cDNA was amplified from total RNA of HUVECs and cloned into a pCMV vector at the Kpn I and Sac I restriction sites (Clontech). To study the expression of CXCL12 in HUVECs, 5 × 105 cells were transfected with a reporter plasmid encoding luciferase (pGL-4-SDF) under the control of the promoter of CXCL12 (39) or with mock-pGL-4 (5 μg), and luciferase activity was determined after 12 hours with the Bright-Glo luciferase kit (Promega). Predicted binding sequences for miR-126 were identified by miRBase ( in the 3′UTRs of the following target genes: SPRED1 sense, 5′-TTTAACTAAATGTAAGGTACGA-3′; VCAM-1 sense, 5′-TGTATAGTACTGGCATGGTACGG-3′; RGS16 sense, 5′-GCCAGTGTTTTTTGTGGTATGA-3′; RGS16 mutated/antisense, 5′-GCGTGATAATGTGACCATGCT-3′; CXCL12 sense, 5′-TGCATTTATAGCATACGGTATGA-3′; CXCL12 mutated/antisense, 5′-GCGTTAATAAGGATTGCCA TTT-3′. These sequences were inserted into the multiple cloning site of the pMIR-Report vector (Ambion) and were thus located in the 3′UTR downstream of the luciferase reporter cDNA. HUVECs (5 × 105) were cotransfected with individual pMIR-Report-based vectors (3 μg each) and pre–miR-126 precursor oligonucleotide (200 pmol), and luciferase activity was detected with the Bright-Glo luciferase kit (Promega) 48 hours after transfection.

Flow cytometry

HUVECs were trypsinized, fixed with Cytofix (BD Biosciences), permeabilized with methanol, and incubated with an antibody against ERK1/2 (Cell Signaling) followed by an FITC-conjugated secondary antibody (Sigma), an Alexa Fluor 647–conjugated antibody against phosphorylated EKR1/2 (pERK1/2, Cell Signaling), or with the corresponding isotype control antibodies (Cell Signaling) according to the manufacturer’s protocol. Whole blood obtained from the retro-orbital plexus of mice was buffered with EDTA, subjected to red-cell lysis (Pharmlyse BD Biosciences), and washed in Dulbecco’s modified Eagle’s medium containing 2 mM EDTA and 0.5% bovine serum albumin (BSA). Blood leukocytes were incubated with combinations of specific antibodies against Gr-1, CD11b, CD3, CD19, and CD45 (BD Biosciences), or against CD115, CD4, and CD8 (eBiosciences) with a mouse anti-lineage biotin-antibody cocktail (Miltenyi), followed by incubation with a FITC-conjugated secondary antibody and with a PE-conjugated antibody against Sca-1 (BD Biosciences) in Hanks’ balanced salt solution with 0.3 mM EDTA and 0.1% BSA. To analyze the content of circulating annexin V+CD31+CD41 apoptotic bodies, 100 μL of EDTA-buffered whole blood was centrifuged at 800g for 10 min at to obtain plasma enriched in apoptotic bodies, which was incubated with a PE-Cy7-conjugated antibody against CD31, a FITC-conjugated antibody against annexin V, and a PE-conjugated antibody against CD41 (31). Apoptosis of HUVECs was quantified with the annexin V–FITC Apoptosis Detection Kit I (BD Biosciences). Samples were analyzed after appropriate fluorescence compensation and gating strategies with a FACSCanto-II or a FACSAria flow cytometer and analyzed with FACSDiva (BD Biosciences) and FlowJo software (Treestar).

Western blotting and ELISA

HUVECs were lysed in CytoBuster (Novagen) or in 20 mM tris-HCl (pH 8.0), 2 mM EDTA, 150 mM NaCl, 1% NP-40, supplemented with 1× complete protease inhibitor cocktail (Roche). To analyze phosphorylated proteins, HUVECs were lysed in 1× PBS (pH 7.6), 1% Triton X-100, 2.5 mM EDTA, and 2.5 mM EGTA, supplemented with phosphatase inhibitors (100 mM NaF, 5 mM Na3VO4, and 30 mM Na4P2O7) and 1 × complete protease inhibitor cocktail. Insoluble material was removed by centrifugation at 18,000g for 5 min at 4°C. Total protein was resuspended in 2 × Laemmli loading buffer, boiled for 10 min, supplemented with β-mercaptoethanol, resolved by 12% SDS–polyacrylamide gel electrophoresis (SDS-PAGE), and transferred to polyvinylidene difluoride membranes. Specific proteins were detected with antibodies against RGS3 (Santa Cruz), RGS16 (Abcam), pERK1/2, total ERK1/2 (Cell Signaling), and β-actin (Sigma), with horseradish peroxidase–conjugated secondary antibodies (Pierce, Santa Cruz) and enhanced chemiluminescence (GE Healthcare). Concentrations of CXCL12 in HUVEC lysates and supernatants were determined with the Duo-Set ELISA Development Kit (R&D Systems).

Mouse models of atherosclerosis

Apoe−/− mice (C57BL/6 background) were injected intravenously (iv) with apoptotic bodies obtained from 1 × 106 apoptotic HUVECs or with PBS as a negative control, and blood was collected from the retro-orbital plexus after 24 hours. Apoe−/− mice were fed a high-fat diet (21% fat, 0.15% cholesterol; Altromin) for 6 weeks for the assessment of early diet-induced atherosclerosis. These mice were injected twice per week with PBS (200 μl iv) or with apoptotic bodies obtained from 1 × 106 apoptotic HUVECs (in 200 μL of PBS). Some Apoe−/− mice were continuously treated with AMD3100 (120 μg/day; Sigma) administered by osmotic minipumps (Alzet, model 2006), and some mice were injected once per week for 6 weeks with 3 × 105 apoptotic bodies sorted from human endarterectomy plaque material or with PBS. Bilateral perivascular nonconstrictive Silastic collars were placed around carotid arteries in Apoe−/− mice that had received a high-fat diet for 2 weeks, which was immediately followed by the application of pre–miR-126, let-7c, or scrambled control miRNA (10 μg per carotid artery) mixed with pluronic gel [25% (w/v), Pluronic F127, Sigma] (34). The intraluminal incubation of common carotid arteries of Apoe−/− mice fed a high-fat diet for 10 weeks was performed by injecting 104 apoptotic bodies collected from wild-type or miR126−/− mice (in 100 μl of PBS) through a flexible tubing (Cornstar) inserted into the external carotid artery while the proximal common carotid artery and the distal internal and external branches were closed with sutures. After 30 min, the suspension of apoptotic bodies was retrieved, the closed arteries were flushed once with PBS, and blood flow was reestablished in the common carotid artery through the internal carotid artery. Mice were killed after 5 more weeks of a high-fat diet. Animal experiments were approved by local authorities and complied with German animal protection law.

Morphometry and immunohistochemistry

Mouse hearts, aortas, and carotid arteries were harvested by in situ perfusion fixation with a solution of 4% paraformaldehyde, 20 mM EDTA, and 5% sucrose. The extent of atherosclerosis was assessed in aortic roots by detecting lipid deposition with Oil Red O (29, 40, 41) in 5-μm transversal sections. After paraffin embedding, neointimal and medial areas were quantified in Movat’s pentachrome–stained serial 5-μm sections of the right common carotid artery within a standardized distance (100 μm) distal and proximal to the maximal, collar-induced plaque volume. Immunofluorescence staining was performed with antibodies against MOMA-2 (MCA519, Serotec), Mac-2 (CL8942AP, Cedarlane), α-smooth muscle actin (1A4), CD3ɛ (Abcam, Cambridge), CD31 (M-20, Santa Cruz Biotechnology), and Sca-1 (Ly-6A/E, E13-161.7, BD Biosciences) or with the appropriate isotype control antibodies (Santa Cruz or BD Biosciences) and proteins were visualized with FITC- or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch). Apoptotic nuclei were detected by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL in situ cell death detection kit, Roche). Collagen was detected with Sirius red (Polysciences). TUNEL+ cells were quantified as positive nuclei relative to total cell nuclei in the aortic root. Nuclei were counterstained by 4′,6-diamidino-2-phenylindole (DAPI). Images were recorded with a Leica DMLB fluorescence microscope and charge-coupled device camera, and image analysis was performed with Diskus Software (Hilgers). RNA isolation from paraformaldehyde-fixed and paraffin-embedded carotid tissue (50 5-μm sections) was performed with the RecoverAll Total Nucleic Acid Isolation Kit (Ambion). Recovery of high-quality RNA was verified (RNA Pico chip, Agilent 2100 Bioanalyzer) before reverse transcription assays were performed (Sensiscript Reverse Transcription kit, Qiagen). Protein was extracted from paraformaldehyde-fixed and paraffin-embedded carotid tissue (~60 10-μm sections) by Qproteome FFPE Tissue Kit (Qiagen) according to the manufacturer’s instructions.

Annexin V–FITC gold labeling and scanning electron microscopy

Untreated or serum-starved HUVECs were fixed with 1% glutaraldehyde and 2% paraformaldehyde. Unreacted aldehydes were quenched with 0.1 M glycine. Fixed cells were stained with annexin V–FITC followed by incubation with an antibody against FITC that was conjugated to gold particles (10 nm, Ted Pella). After postfixation with 1% glutaraldehyde, dehydrated samples were processed by critical point drying with CO2 and sputter-coating with carbon. Scanning electron microscopy images (FEI/Philips ESEM XL30 FEG) were acquired in secondary electron detection, back-scattered electron detection, or mixed detection mode.


HUVECs were grown on glass chamber slides (Nunc). Apoptotic bodies were stained with annexin V–FITC (BD Biosciences) for 30 min and incubated with subconfluent HUVECs for 1 hour. After washing and fixation or permeabilization, filamentous actin (F-actin) of adherent cells was visualized with Alexa 568 phalloidin (Invitrogen) as previously described (7).

Statistical analysis

Data represent the mean ± SEM and were analyzed by Student’s t test, ANOVA (analysis of variance) with Newman-Keuls or Dunnett’s multiple comparison test, nonparametric Mann-Whitney test or Kruskal-Wallis test with Dunn’s post hoc test (Prism 4.0 software, GraphPad), as appropriate. Statistical significance was considered where P < 0.05.


We thank M. Garbe, S. Roubrocks, and S. Wilbertz for excellent technical assistance and M. Bovi for expert help with scanning electron microscopy. This work was supported by the Deutsche Forschungsgemeinschaft (FOR809, WE1913/7-2, and 10-1, ZE827/1-1) and the Interdisciplinary Centre for Clinical Research “BIOMAT” within the Faculty of Medicine at the RWTH Aachen University (VV-B113). The authors declare no conflict of interests.

Supplementary Materials

Fig. S1. Identification of the targets of miR-126.

Fig. S2. Repression of luciferase reporter activity by RGS16, CXCL12, and mutated controls.

Fig. S3. Comparative analysis of the abundance of RGS3 and RGS16.

Fig. S4. Increased abundance of RGS16 in transfected HUVECs.

Fig. S5. CXCL12-mediated phosphorylation of ERK1/2 in HUVECs.

Fig. S6. Generation of apoptotic bodies by apoptotic HUVECs.

Fig. S7. Apoptotic bodies do not mobilize monocytes or neutrophils to peripheral blood.

Fig. S8. Sparse colocalization of Sca-1 with macrophages in atherosclerotic lesions of mice treated with apoptotic bodies.

Fig. S9. Apoptotic bodies isolated from human plaques and from mouse blood contain miR-126.

Fig. S10. Apoptotic bodies specifically inhibit apoptosis and induce the proliferation of HUVECs.

Table S1. The expression of miRNAs in SMCs.

References and Notes

View Abstract

Stay Connected to Science Signaling

Editor's Blog

Navigate This Article