Exosomal microRNA transfer into macrophages mediates cellular postconditioning de Couto: Exosomal RNA transfer modulates macrophages

Abstract

Background

Cardiosphere-derived cells (CDCs) confer cardioprotection in acute myocardial infarction (MI) via distinctive macrophage (Mϕ) polarization. Here we demonstrate that CDC-secreted exosomes (CDC_exo) recapitulate the cardioprotective effects of CDC therapy known as cellular postconditioning.

Methods

Rats and pigs underwent MI induced by ischemia-reperfusion prior to intracoronary infusion of CDC_exo, inert fibroblast exosomes (Fb_exo; control), or vehicle. Two days later, infarct size was quantified. Macrophages were isolated from cardiac tissue or bone marrow for downstream analyses. RNA-sequencing was used to determine exosome content and alterations in gene expression profiles in Mϕ.

Results

Administration of CDC_exo, but not Fb_exo, after reperfusion reduces infarct size in rat and pig models of MI. Furthermore, CDC_exo reduce the number of CD68+ Mϕ within infarcted tissue and modify the polarization state of Mϕ so as to mimic that induced by CDCs. CDC_exo are enriched in several miRNAs (including miR-146a, miR-181b, and miR-126) relative to Fb_exo. Reverse pathway analysis of whole-transcriptome data from CDC_exo-primed Mϕ implicated miR-181b as a significant (p=1.3×10⁻²¹) candidate mediator of CDC-induced Mϕ polarization and protein kinase C δ (PKCδ) as a downstream target. Otherwise-inert Fb_exo loaded selectively with miR-181b alter Mϕ phenotype and confer cardioprotective efficacy in a rat model of MI. Adoptive transfer of PKCδ-suppressed Mϕ recapitulates cardioprotection.

Conclusions

Our data support the hypothesis that exosomal transfer of miR-181b from CDCs into Mϕ reduces PKCδ transcript levels and underlies the cardioprotective effects of CDCs administered after reperfusion.

INTRODUCTION

Acute myocardial infarction (MI) elicits a robust innate immune response that mobilizes a large population of neutrophils and macrophages to the myocardium. This response, which plays out over the first few days following injury, occurs in a phasic manner whereby a rapid infiltration of neutrophils is succeeded by a heterogeneous influx of monocytes/macrophages¹. The secondary phase of monocyte/macrophage infiltration, which is implicated in the clearance of necrotic cellular debris, deposition of matrix, and maturation of scar tissue, critically determines ultimate infarct size. Depletion of circulating monocytes/macrophages² or selective depletion of M2-like healing macrophages³ not only exacerbates tissue injury, but also impairs cardiac function post-MI. Although there is macrophage heterogeneity within the myocardium at steady state⁴ and post-MI⁵, it is unclear which environmental cues confer the distinct polarization profile(s) observed in vivo⁶.

Cardiosphere-derived cell (CDC) therapy has been utilized as a strategy to treat ischemic heart disease in both small and large animals, as well as in humans. Despite limited retention within the host myocardial tissue following intracoronary or direct myocardial injection⁷, the beneficial effects of cell therapy are long-lasting⁸. Our lab has recently demonstrated that exosomes (nanosized lipid bilayer vesicles; ~30–150nm) are a potent secretory component of CDC therapy following MI⁹, as confirmed by Barile et al.¹⁰. These vesicles are secreted by many cell types throughout the body for local or remote cell-to-cell communication; they contain a specific payload of small RNAs (e.g., miRNA, snoRNA, Y-RNA, etc.)¹¹ and protein (e.g., transcription factors, transmembrane receptors, integrins, etc.)¹², and are readily taken up by numerous cell types including macrophages¹³, fibroblasts¹⁴, and endothelial cells¹⁵. Here, we extend our recent work on cellular postconditioning by CDCs^16–18 by identifying CDC exosomes (CDC_exo)¹⁹ as key elements of the cellular secretome which confer cardioprotection in MI. We further demonstrate that miR-181b within CDC_exo is a critical mediator of macrophage polarization in vitro and cardioprotection in vivo.

METHODS

Experimental Protocol, Animals, & Surgical procedures

All studies were performed at Cedars-Sinai Medical Center in accordance with the Institutional Animal Care and Use Committee (IACUC) guidelines.

Rat

In vivo experimental protocols were performed on 7–10 week old female Wistar-Kyoto (WKY) rats (Charles River Labs, Wilmington, MA). To induce ischemia-reperfusion (IR) injury, rats were provided general anesthesia and then a thoracotomy was performed at the 4^th intercostal space to expose the heart and left anterior descending (LAD) coronary artery. A 7–0 silk suture was then used to ligate the LAD, which was subsequently removed after 45 minutes to allow for reperfusion. Twenty minutes later, cells (5×10⁵; in 100µl PBS) or exosomes (35µg, 105µg, or 350µg; in 100µL PBS) were injected into the left ventricular cavity with an aortic cross-clamp, over a period of 20 seconds.

Pig

In vivo experimental protocols were performed on adult female Yucatan mini-pigs. To induce IR injury, pigs were provided general anesthesia and then an angioplasty balloon was threaded into the mid-LAD just after the first diagonal branch and inflated for 90 minutes. Thirty minutes following reperfusion, animals were allocated to receive CDC_exo (7.5mg) or placebo (Iscove’s Modified Dulbecco’s Medium, IMDM). To rule out any complications with retention of CDC_exo, we performed a direct intra-myocardial injection for delivery. This procedure was performed in an open-chest model (median sternotomy) using a 25G needle. The infarcted area was identified (grey and stiff zone) and ten sequential injections (0.75mg) were administered at the perimeter of the infarct (border zone).

Cardiac Function Measurements

Rat

Transthoracic echocardiography (Vevo770, VisualSonics) was performed on rats at baseline (preischemia) and prior to euthanization (48 hours or 2 weeks). Two-dimensional short and long axes were visualized. Three representative cycles were captured for each animal/time point, and measurements for LV end-systolic dimension, LV end-diastolic dimension, and fractional area of change were obtained and averaged.

Pig

To measure LV function, left ventriculograms were performed before infusion (after MI) and prior to euthanize (48 hours) using a pigtail catheter inserted into the LV. Two incidences were studied: a 45° left anterior oblique projection and a 30° right anterior oblique projection 20° cranial projection. LVEF was determined by averaging those 2 measurements.

Infarct Size Evaluation

Rat

Two days following IR injury, hearts were arrested in diastole following intraventricular injection of 10% KCl. Hearts were then excised, washed in PBS, and cut into serial sections of ~1mm in thickness (from apex to basal edge of infarction). Sections were incubated with TTC (2,3,5-Triphenyl-2H-tetrazolium chloride, 1% solution in PBS) for 20 minutes in the dark, washed with PBS, then imaged and weighed. Infarcts were delineated from viable tissue (white versus red, respectively) and analyzed using ImageJ software as previously reported¹⁶. Infarct mass, viable mass, and LV mass were calculated by extrapolating for infarct and non-infarct volumes (based on the areas calculated from both sides of a tissue section) and weight of the tissue. Percentage infarct mass was calculated using (Infarct Mass/Viable Mass) × 100%.

Pig

Two days after infusion, animals underwent lateral thoracotomy. A catheter was inserted into the left appendage and Thioflavin T (50mL, 2%) was injected into the left atrium to assess microvascular obstruction (MVO). Then an angioplasty balloon was inflated at the same position used for the infarct and Gentian violet (50mL, 1.8%) was injected to assess area at risk (AAR). Animals were then euthanized and the heart were explanted and sectioned into 1cm thick slices. TTC staining was performed to determine infarct size (IS). IS, MVO and AAR were then determined as previously described¹⁷. AAR was reported as a percent of total LV mass and both IS and MVO were reported as percent of the AAR¹⁷.

In vitro cell culture

Bone marrow-derived Mϕ

Bone marrow was isolated as previously described¹⁶. Briefly, bones from rat (femur) or pig (rib) were sterilized with ethanol (70% in H₂O) and flushed with PBS (containing 1% FBS, 2mM EDTA) and filtered through a 70µm filter. Red blood cells were lysed with ACK buffer (Invitrogen) then resuspended in IMDM (containing 10% FBS, 10ng/mL m-CSF). The resulting cell suspension was plated and grown for 7 days to obtain Mϕ. Polarization of Mϕ toward M1 (100ng/mL LPS and 50ng/mL IFNγ; Sigma-Aldrich and R&D Systems, respectively), M2 (10ng/mL IL-4 and IL-13; R&D Systems), or M_CDCexo (105ug) was performed on the night between days 7 and 8 (~18hrs).

Cardiac Mϕ

Hearts were collected and processed to isolate cardiac Mϕ following ischemic injury, as described¹⁶. Briefly, 2 days following MI hearts from animals treated with Fb_exo or CDC_exo were perfused with PBS and collected. Cardiac tissue was minced and digested, and the resulting suspension filtered (70µm) to generate a single-cell suspension. Mononuclear cells were separated by centrifugation using a density gradient (Histopaque-1083; Sigma), and then resuspended in RPMI (supplemented with 1% FBS) and plated (37°C, 5% CO₂). Two hours later, cardiac Mϕ were washed with PBS and used for downstream analyses.

Neonatal Rat Ventricular Myocytes

Cells were isolated and cultured as previously described¹⁶. Human Umbilical Vein Endothelial Cells (HUVECs). Cells were purchased (ATCC) and cultured according to the distributor’s recommendations.

Statistics for non-genomic data

Results are expressed as mean ± SEM. Linear regression analysis was performed to test the linear relationship between groups. Each replicate value was considered as an individual point to determine R². To determine differences between groups at a single time point, data were tested using either two-tailed, unpaired, Student’s t-test or 1-way ANOVA followed by Tukey’s multiple comparisons test (outliers were excluded if Z-score was >3 or <−3). To determine differences between groups at different time points/conditions, data were tested using repeated measures ANOVA followed by Sidak’s multiple hypothesis correction, 2-way ANOVA followed by Holm-Sidak’s multiple hypothesis correction, or multiple t-tests followed by Holm-Sidak’s multiple hypothesis correction (as denoted within the figure legends). To test for linear trend, we performed a 1-way ANOVA followed by a post test for linear trend. All analyses were performed using Prism 5 software (GraphPad) and only differences with a p<0.05 were considered statistically significant.

Refer to the online-only Data Supplement for additional materials and methods.

RESULTS

Characterization of CDC exosomes

CDC exosomes (CDC_exo) were isolated from serum-free media conditioned by human CDCs for 5 or 15 days in culture (Supplemental Figure IA). The conditioned media (CM) was then collected, or concentrated by ultrafiltration by centrifugation (UFC, Millipore), to assess particle number and size (Nanoparticle Tracking Analysis, Malvern Nanosight), as well as protein concentration and protein marker expression (Figure 1 and Supplemental Figure IB). The relationship between particle number and protein content was linear in CDC_exo from each of two distinct human lines (Figure 1C). Concentration of exosomes by UFC reduced the suspension volume (>10-fold, Figure 1A) without a measurable loss of particles (Supplemental Figure ID). Human dermal fibroblast exosomes (Fb_exo; shown to be functionally inert in vivo⁹) served as a negative control and were isolated and characterized in parallel, under equivalent conditions (Supplemental Figure IC). The mode particle size for CDC_exo was ~150nm in diameter, expressing typical exosome markers (including CD63, Alix, and HSP70), as well as the CDC surface marker CD105. Although this size distribution biases higher than what is generally quoted for exosomes on the basis of desiccation-prepared electron microscopy (30–100 nm)²⁰, the observed range is typical of hydrated exosomes measured by dynamic light scattering or nanoparticle tracking analysis²⁰. For uniformity, given the elevated particle number, we performed all subsequent experiments using CDC_exo isolated from 15-day conditioned media.

Figure 1. Exosome isolation and characterization.

(A), Nanoparticle tracking analysis revealing particle size distribution and concentration within the conditioned media (CM), and following ultrafiltration by centrifugation (UFC), in human CDC lines (155: ¹⁵⁵CDC_exo; 220: ²²⁰CDC_exo). (B), Representative protein immunoblots of exosome (CD63, CD81, Alix, HSP70, and MHCII) and CDC (CD105) markers in human CDC exosomes. (C), Correlation between particle number and protein amount in serially diluted samples of exosomes. Graphs depict mean +/− SEM.

CDC_exo administered following reperfusion in a rat model of AMI mimic the cardioprotective effects of CDCs

To examine whether CDC_exo could recapitulate the cardioprotective effects observed with CDCs, we delivered CDC_exo in vivo. First, we demonstrated that CDC_exo were retained within the heart by pre-labeling CDC_exo with DiI (Invitrogen), performing a modified version of our in vivo protocol (Figure 2A) by sacrificing rats after 2 hours of reperfusion and examining exosome distribution by fluorescence imaging (Xenogen, Supplemental Figure IIA). Second, in dose-finding studies, rats underwent 45 minutes of ischemia followed by 20 minutes of reperfusion and then received vehicle (PBS), rat allogeneic CDCs (rCDCs), rCDCs with an exosome inhibitor (rCDC + GW4869), or one of several doses of CDC_exo (35µg, 105µg, or 350µg). To improve efficacy after intracoronary delivery under aortic cross clamp, exosomes were precipitated with polyethylene glycol prior to delivery (Supplemental Figure IIB). Forty-eight hours later, animals were euthanized and hearts excised for infarct size (IS) determination by TTC staining. Confirming our earlier report¹⁶, rCDCs (5×10⁵ cells) reduced IS. Pretreatment of rCDCs with GW4869 abrogated the cardioprotective efficacy of rCDCs, consistent with the idea that exosome secretion is required for rCDC bioactivity. We further tested this idea by applying human CDC_exo directly in lieu of cells. CDC_exo reduced IS to a similar level as rCDCs at doses of 350µg and 105µg, but not at 35µg (Figure 2B and Supplemental Figure IIC). Based on these data we proceeded to test the potency of CDC_exo versus inert Fb_exo^{9, 11}. Third, utilizing 350µg protein as our standard dose, we show that CDC_exo from two human CDC lines (220 and 155) reduced IS to a similar level as rCDCs, in contrast to Fb_exo and PBS controls (Figure 2, C & D). These data were further validated with 3 additional human CDC_exo lines (Supplemental Figure IID). The data presented here represent findings of exosomes produced by one primary human CDC line (220, denoted herein as CDC_exo); confirmatory findings using exosomes from another human line are found in the supplemental material. Finally, we tested the durability of the cardioprotective effects of CDC_exo. When follow-up was extended to 2 weeks (Supplemental Figure IIIA), CDC_exo still reduced IS (Supplemental Figure IIIB) and preserved cardiac function (Supplemental Figure III C–D) relative to Fb_exo controls. Thus, the functional and structural benefits of CDC_exo extend beyond 48 hours, as did the benefits of CDCs^{16, 18}.

Figure 2. Intracoronary infusion of CDC exosomes (CDC_exo) in rats reduces ischemic injury when delivered 20 minutes following reperfusion.

(A), Protocol schematic for infusion of exosomes and tissue harvest. (B), Quantification of infarct mass at 48 hrs: dose-responsive effects, as measured by protein concentration, of CDC_exo (versus vehicle [PBS], rCDCs, and rCDC + GW4869). rCDC: allogeneic rat CDC; GW4869: exosome inhibitor. n=4–6/group. (C), Representative TTC-stained hearts from animals sacrificed 48hrs after infusion of exosomes from either of two human CDC lines (²²⁰CDC_exo, ¹⁵⁵CDC_exo) or fibroblasts (Fb_exo; 350µg). (D), Quantification of infarct mass in hearts from groups defined in C (n=5/group, mean +/− SEM). Statistical significance was determined using 1-way ANOVA followed by Tukey’s multiple corrections test. *P<0.05 vs. PBS and Fb_exo, respectively.

Intramyocardial injection of CDC_exo administered following reperfusion in a pig model of AMI mimics the cardioprotective effects of CDCs

To confirm these findings, we validated our work in the porcine model (Figure 3A). Briefly, Yucatan mini-pigs underwent 90 minutes of ischemia followed by 30 minutes of reperfusion before direct open-chest intramyocardial injection of CDC_exo (7.5mg) or placebo. Cardiac function was assessed at 48 hours, with subsequent euthanasia and IS determination by TTC. After treatment with CDC_exo, LVEF was better preserved (Figure 3B; Supplemental Figure IV, A & B), microvascular occlusion (MVO/AAR) was reduced, and infarct size (IS/AAR) was attenuated relative to placebo (Figure 3, C & D). Furthermore, we noted a reduction in CD68+ macrophage (Mϕ) infiltration into the infarct of CDC_exo-treated versus placebo animals (Figure 3, E & F). Together, these in vivo porcine data support the hypothesis that CDC_exo are key mediators of CDC-induced cardioprotection, not only in rodents but also in a clinically-relevant large-animal model.

Figure 3. Direct intramyocardial injection of CDC_exo in pigs reduces ischemic injury when delivered 30 minutes following reperfusion.

(A), Schematic of protocol. (B), LV ejection fraction (LVEF), measured by echocardiography, immediately following reperfusion (baseline; post-reperfusion, but before treatment) and after 48 hrs. (C), Representative TTC-stained placebo- and CDC_exo-treated pigs hearts at 48 hrs. Area-at-risk (AAR; red dotted line), microvascular occlusion (MVO; black dotted line), and infarct size (IS; white dotted line) (D), Quantification of AAR, MVO, and IS in pigs treated with placebo or CDC_exo. (E) Representative immunohistochemical stain of CD68⁺ Mϕ within the infarct region of placebo- and CDC_exo-treated pigs. Scale bar: 50µm. (F) Quantification of CD68⁺ Mϕ described in (E). Graphs depict mean +/− SEM (Placebo: n=5; CDC_exo: n=4). Statistical significance was determined using Student’s t-test or repeated measures ANOVA followed by Sidak’s multiple corrections test. *P<0.05.

Implication of uniquely-polarized macrophages as effectors of cardioprotection

The first evidence that inflammation might be involved in cellular postconditioning came from our initial studies in rats¹⁶, where we found that CDCs modulate Mϕ toward a cardioprotective phenotype. Mimicking the effects of CDCs¹⁶, CDC_exo significantly reduced CD68+ Mϕ accumulation within the infarct border zone relative to Fb_exo control as measured by immunohistochemistry (Figure 4, A & B; Supplemental Figure II, E & F) and flow cytometry (Figure 4, C–E). Next, we harvested hearts from animals treated with either CDC_exo or Fb_exo 48 hours after MI then isolated cardiac Mϕ (Figure 4F), as described¹⁶. Expression of proinflammatory genes Nos2 and Tnf was significantly attenuated in cardiac Mϕ isolated from CDC_exo-treated animals (Figure 4G).

Figure 4. CDC_exo effects on Mϕ infiltration and polarization.

(A), Representative confocal images of CD68⁺ Mϕ within the infarct border zone 48 hrs after reperfusion. (B), Quantification of CD68⁺ Mϕ in (A) (n=4/group). (C & D), Representative flow cytometry plots showing the gating strategy used to determine CD45+CD11b+CD68+ cardiac Mϕ. (C), Leukocytes were gated (FSC/SSC) and dead cells removed (DAPI−), prior to identification by CD45+ and CD11b+. (D), Representative histograms of CD68+ cells from Fb_exo and CDC_exo-treated hearts as a result of gating according to (C). Blue line: Negative control. (E), Quantification of flow cytometry data in (D) (n=4/group). (F), Representative image of isolated cardiac Mϕ denoting CD68+ by immunohistochemistry. (G), Gene expression profiles from cardiac Mϕ isolated from infarcted hearts treated with Fb_exo or CDC_exo (n=3/group). Scale bar: 50µm. Graphs depict mean +/− SEM. Statistical significance was determined using Student’s t-test or multiple t-tests followed by Holm-Sidak’s multiple corrections test. *P<0.05.

In vitro, rat bone marrow (BM)-derived Mϕ exposed to CDC_exo (Figure 5A) exhibit a distinctive shift in Mϕ polarization (not observed with Fb_exo), including elevation in anti-inflammatory gene expression levels of Arg1, Il4ra, Tgfb1, and Vegfa, without any significant change in Tnf (Figure 5B). These effects occur dose-dependently (Figure 5C and Supplemental Table 1), are consistent among donor lines (Supplemental Figure VA), and are stable following lyophilization and reconstitution (Supplemental Figure VA) or sequential freeze-thaw cycles (Supplemental Figure V, B & C). Pig BM-derived Mϕ exposed to CDC_exo exhibited a similar polarization shift (Supplemental Figure VIA). An assay of phagocytosis revealed enhanced uptake of fluorescent beads in the CDC_exo-primed Mϕ group relative to M1, M2, or Fb_exo-primed Mϕ (Figure 5, D & E; Supplemental Figure VIB), again recapitulating the effects of CDCs¹⁶. Together, these data support the notion that CDC_exo are crucial in mediating the Mϕ responses underlying cellular postconditioning¹⁶.

Figure 5. CDC_exo uniquely modulate bone marrow-derived Mϕ polarization.

(A), Schematic of in vitro assay protocol, including bone marrow (BM) harvest, Mϕ differentiation, and treatment. (B), Gene expression changes in response to Fb_exo and CDC_exo. (C), Dose-response gene expression changes in BM-derived Mϕ 18hrs after exposure to CDC_exo. (D), Representative flow cytometry plots and quantification of fluorescent microsphere bead uptake by BM-derived Mϕ following polarization with M1 (LPS & IFNγ), M2 (IL-4 & IL-13), Fb_exo, and CDC_exo. Blue line: Negative control. (E), Quantification of flow cytometry data in (D). Graphs depict mean +/− SEM with n=3–5/group. Statistical significance was determined using 2-way ANOVA followed by Holm-Sidak’s multiple corrections test, 1-way ANOVA followed by Tukey’s multiple corrections test, or multiple t-tests followed by Holm-Sidak’s multiple corrections test. *P<0.05.

CDC_exo have a distinct miRNA signature

Exosomes are intriguing as signaling intermediates with cargo rich in small RNA molecules^{11, 12}. The composition of exosomes is not determined randomly, but rather uniquely regulated by cell source and environmental stressors²¹. To understand how CDC_exo confer Mϕ polarization we performed total RNA-sequencing (RNA-seq) on exosomes isolated from CDCs (CDC_exo) and Fbs (Fb_exo), then aligned them to a database of both short and long RNA annotations. To identify the most biologically-relevant changes, we performed one set of biological replicates using the Illumina RNA-seq platform and a second using the Ion Torrent platform (Figure 6)^{22, 23}. Globally, the number of reads (and thus the RNA content) mapping to different classes of RNAs were similar between CDC_exo and Fb_exo; the greatest number of reads mapped to protein coding genes (RefSeq) and tRNAs, while fewer reads mapped to short RNAs (e.g., miRNAs and piRNAs; even fewer mapped to scaRNAs and snoRNAs) (Supplemental Figure VII). However, the largest and most consistent proportional change was observed in the miRNA class, where CDC_exo produced 5- and 8-fold greater reads than Fb_exo in the Illumina and Ion Torrent libraries, respectively (Figure 6, A & B). Next, we focused on the reads that aligned to miRNA annotations and normalized both sets of libraries using DESeq 2 to identify the miRNAs with consistent changes in expression between CDC_exo and Fb_exo (Figure 6C). While the two platforms (Illumina and Ion Torrent) had technical differences (Pearson correlation coefficient ~0.6), several miRNAs stood out as having significant differential expression. The miRNAs that were uniquely enriched in Fb_exo included miR-199, miR-3676, and miR-3648, while those uniquely enriched in CDC_exo included miR-126 and miR-181b; in addition, miR-181a was enriched just below the threshold of significance in CDC_exo. Interestingly, miR-181a/b stood out as having large numbers of differentially expressed target genes in CDC_exo-treated Mϕ (marked as red in Figure 6C, and described in detail below).

Figure 6. RNA-sequencing analysis of CDC_exo and CDC_exo-primed Mϕ identify miR-181 as the top miRNA exosome candidate.

(A), Total RNA-sequencing analysis depicting the change between Fb_exo and CDC_exo in the number of total RNA-seq reads for different RNA classes, shown as a fold-change ratio of Fb_exo/CDC_exo. RNA categories with a total of less than 1,000 reads aligned to all transcripts are omitted. The libraries sequenced using Illumina (blue) and Ion Torrent (red) sequencing platforms are shown separately, and the RNA classes are ordered by the average fold-change across both platforms. (B), Raw read counts for each of the four samples for the miRNA class (the other classes are shown in Supplemental Figure VII). (C), Volcano plot showing relative expression changes of 677 miRNAs between Fb_exo and CDC_exo, with normalization and significance testing performed by DEseq2, using the Illumina libraries as one replicate and the Ion Torrent libraries as a second replicate (statistical significance reflects the concordance between the two platforms). Dashed horizontal line: uncorrected p-value of 0.05; Filled points: miRNAs having predicted target genes present in the miRecords database, and further analyzed in D–H; Filled red points and red labels: miRNAs with targets downregulated in CDC_exo-polarized Mϕ (detailed in D). (D), The 7 miRNAs with the most significant downregulation of predicted targets in CDC_exo-polarized Mϕ, ordered by significance. Top panel: fold-change for all predicted miRNA gene targets; Bottom panel: statistical significance (hypergeometric test, see Supplemental Methods) of the association between predicted miRNA targets and downregulated genes (−log₁₀ p-values); Annotations at the bottom: miRNAs with overexpression in CDC_exo compared to Fb_exo, from C. (E), Comparison of untreated Mϕ (rn-control) and CDC_exo-primed Mϕ (rn-M_CDCexo) gene expression to published datasets^{25, 26}. Only genes with homology between rat (rn), human, and mouse (mm) are shown, while those with undetectably low expression in all our Mϕ samples are filtered out. Genes are sorted by differential expression in rn-M_CDCexo, from the most downregulated (left, blue) to the most upregulated (right, red). Gene expression values are shown as relative (log-transformed) fold change, relative to other samples in the same study. (F), Venn diagrams highlight the highly significant associations between genes sets of rn-M_CDCexo to published miR-181a/b KO (miR-181 KO) thymocytes²⁷; significance values were calculated using a hypergeometric test (see Supplemental Methods). Downregulated genes in our model (+CDC_exo) were associated with genes upregulated in miR-181 KO thymocytes (upper left quadrant, p=2.6E-147), and vice-versa (lower right quadrant, p=5.2E-79). These findings were not true for genes regulated in the same direction (upper right and lower left quadrants). (G), Predicted miR-181 target genes from miRecords are shown in the same order as (E), and are concentrated in those genes downregulated rn-M_CDCexo and upregulated miR-181 KO thymocytes. (H) Venn diagrams show the strong association of predicted miR-181 targets and those genes downregulated in rn-M_CDCexo (p=1.3E-21). Significance values were calculated using a hypergeometric test (see Supplemental Methods). Refer to the Supplemental Methods for sample size and additional methods.

miR-181 as a candidate effector of CDC_exo-mediated Mϕ polarization

To investigate the effect of CDC_exo miRNAs on Mϕ, we performed RNA-seq on two biological replicates of unprimed (control) or CDC_exo-primed (CDC_exo) rat BM-derived Mϕ. Then, we used gene annotations from the NextBio functional annotation database (Illumina Inc.)²⁴ to investigate the functions of genes with altered expression following exposure to CDC_exo. Specifically, we compared differentially-regulated genes to predicted miRNA targets from the miRecords database (reverse pathway analysis^{25, 26}), and found that target genes for several miRNAs enriched in CDC_exo were predominantly downregulated in CDC_exo-treated Mϕ (Figure 6D). This downregulation of miRNA target genes was not observed for randomly permuted gene sets of the same size (Supplemental Figure VIIIA). The fact that miRNAs with proportionally higher expression in Fb_exo (e.g., miR-30 and let-7) were also downregulated in CDC_exo-primed Mϕ may reflect the higher absolute levels of these miRNAs in CDC_exo (Figure 6A). Consistent with this interpretation, we found a significant downregulation of predicted targets for all highly-expressed miRNAs examined, whether they had proportionally higher expression in CDC_exo or Fb_exo (Supplemental Figure VIIIB). However, the three most significant miRNAs (miR-181, miR-27, and miR-19) were all proportionally higher in CDC_exo, and thus represented strong candidates for conferring the functional changes observed in Mϕ (Figure 6D). miR-181 also had > 10-fold higher proportional abundance in CDC_exo vs. Fb_exo (Figure 6C), compared to miR-19 and miR-27 (both had < 2-fold higher abundance). These data pointed to the miR-181 family as a strong candidate for the key regulatory cargo contained in CDC_exo.

To further explore transcriptional changes in CDC_exo-treated Mϕ, we performed an unbiased comparison of our RNA-seq dataset with two published rodent immune cell RNA-seq datasets (Figure 6E). First, we compared expression changes in our CDC_exo-treated cells to a mouse Mϕ RNA-seq dataset²³. This revealed a similar pattern of gene expression changes, with our unprimed rat Mϕ resembling the mouse M0 state, and CDC_exo-primed rat Mϕ resembling the mouse M2 state (Figure 6E). Since complete concordance was not observed between the two datasets, these findings support the notion¹⁶that CDC_exo-primed Mϕ are not ordinary M2 Mϕ, but lie somewhere within (or outside) the M1-M2 spectrum. We next compared a mouse miR-181a/b knockout (KO) thymocyte dataset²² with CDC_exo-primed Mϕ (Figure 6, E & F). This particular perturbation came up as the most strongly related among thousands of gene perturbation datasets within the NextBio database, and our results demonstrated a very strong degree of concordance (Figure 6E). Overlapping the genes with statistically significant expression changes across two replicates revealed that those downregulated in our model (+CDC_exo) were significantly associated (p=2.6E-147) with genes upregulated in miR-181a/b KO thymocytes (Figure 6F, upper left quadrant), and vice-versa (Figure 6F, lower right quadrant, p=5.2E-79). Further, when grouped based on their Gene Ontology (GO) annotation, these genes clustered in biological processes associated with increased translation, metabolism, and oxidative phosphorylation (Supplemental Table 2). Genes downregulated in our model had no statistical association with genes downregulated in miR-181a/b KO thymocytes (Figure 6F, lower left quadrant), nor did the converse (Figure 6F, upper right quadrant). This association is even stronger than the association with predicted miR-181 targets (p=1.3E-21, Figure 6, G & H), further supporting the notion that miR-181 may be a critical factor regulating CDC_exo function, and by far the most promising single candidate mediator of the transcriptomic changes observed in CDC_exo-primed Mϕ.

miR-181b in exosomes confers cardioprotection

The miR-181 family comprises four known members (miR-181a, −181b, −181c, and −181d), which have been implicated as modulators of myeloid leukemia and stem cell differentiation²⁷. To better understand the functional effects of miR-181, we assessed the gene expression pattern of miR-181 family members in CDCs and CDC_exo (Figure 7A; miR-181c and miR-181d were undetectable by qPCR and thus not presented). Levels of miR-181a, but not miR-181b, were higher in human CDCs relative to Fbs (Figure 7A). CDC_exo, on the other hand, were highly enriched for miR-181b, but not miR-181a, indicative of highly selective transfer of miRNA cargo into CDC_exo (Figure 7A and Supplemental Figure IXA), as has been seen with other miRNAs in other cell types²⁸. The expression levels of miR-181a and miR-181b differed in various isolated cardiac cell types (e.g., cardiomyocytes, endothelial cells, Mϕ; Supplemental Figure IXB) and were not changed within the hearts of untreated animals prior to or following ischemic injury (Figure 7C).

Figure 7. Exosomal transfer of miR-181b reduces infarct size following IR injury.

(A), Human CDCs have elevated expression of miR-181a (but not miR-181b), while CDC_exo have enriched miR-181b (but not miR-181a) relative to Fb and Fb_exo, respectively (n=3/group). (B), Representative TTC stained hearts 48hrs following IR injury. Fb_{exo(Scramble)}: Fb_exo with miR-181b scramble; Fb_exo(Mimic): Fb_exo with miR-181b mimic; CDC_{exo(Scramble)}: CDC_exo with miR-181b scramble; CDC_{exo(Antagomir)}: CDC_exo with miR-181b antagomir. (C), Expression of miR-181a and miR-181b in tissue from non-infarcted and infarcted rats, normalized to liver. LV: left ventricle; RV: right ventricle; BZ: border zone; IZ: infarct zone; NZ: normal zone (n=4/group). (D), Quantification of percent infarct mass in (C) (n=4–8/group). (E), Quantification of CD68+ Mϕ in (Supplemental Figure XI). Graphs depict mean +/− SEM. Statistical significance was determined using 2-way ANOVA followed by Holm-Sidak’s multiple corrections tests or 1-way ANOVA followed by Tukey’s multiple corrections test. *P<0.05.

We tested the effects of miR-181b on various cardiac cell types, including endothelial cells and Mϕ, in vitro. Consistent with its known targets²⁹, treatment with miR-181b, but not miR-Scramble, reduced transcript levels for E-selectin, VCAM-1, and ICAM-1 in endothelial cells (Supplemental Figure XA). Interestingly, these effects were reproduced in a similar manner following CDC_exo or Fb_exo with miR-181b, but not Fb_exo, treatment (Supplemental Figure XB). Next, we performed a scratch assay to examine whether miR-181b modulates endothelial migration. While CDC_exo enhanced scratch closure in vitro, miR-181b alone had no effect (Supplemental Figure X, C & D).

To further validate the functional efficacy of exosomal miR-181b in vivo, we overexpressed miR-181b in Fb_exo (naturally low in miR-181b) or antagonized miR-181b in CDC_exo and measured infarct size in our rat model (Figure 2A). Briefly, Fb_exo were transfected with miR-181b mimic (Fb_exo(Mimic)) (Supplemental Figure IXC) and CDC_exo were transfected with an antagomir to miR-181b (CDC_{exo(Antagomir)}); miR-scramble was used as a transfection control in both types of exosomes (Fb_{exo(Scramble)} and CDC_{exo(Scramble)}). After 48 hours of reperfusion, Fb_exo(Mimic) and CDC_{exo(Scramble)} reduced infarct size and Mϕ infiltration, although not to the same extent as CDC_exo, relative to Fb_exo, Fb_{exo(Scramble)}, and CDC_{exo(Antagomir)} (Figure 7, B, D, & E; Supplemental Figure XI). Thus, miR-181b reproduces the cardioprotective effects of CDC_exo, while blocking miR-181b blunts those effects.

miR-181b inhibits the expression of protein kinase C δ (PKCδ)

To better understand how miR-181b may lead to cardioprotection, we honed in on the target genes for miR-181b within CDC_exo-treated Mϕ. Among known miR-181b targets, protein kinase C δ (PKCδ; prkcd) was one of the most significantly downregulated (4.82E-30) (Figure 7F). Consistent with previous reports²⁷, expression levels of prkcd were reduced in Mϕ transfected with miR-181b compared to miR-scramble control (Figure 8A). PKCδ enhances inflammatory gene expression (e.g., TNFα)³⁰, as well as cell differentiation, growth, and death³¹, and PKCδ inhibitors are known to induce, or enhance, cardioprotection^{32, 33}. We thus hypothesized that PKCδ acts as a downstream effector of CDC_exo-mediated cardioprotection, via miR-181b–induced suppression of prkcd in Mϕ. To test this hypothesis, we transfected Mϕ with a specific small interfering RNA (siRNA) which effectively suppressed prkcd (Figure 8B). Given our previous demonstration that adoptive transfer of CDC-conditioned Mϕ mimicked cardioprotection by CDCs, we modified our rat model of IR (Figure 2A) to adoptively transfer Mϕ transfected with either PKCδ siRNA (Mϕ_{(PKCδ siRNA)}) or scramble siRNA (Mϕ_{(scramble siRNA)}). Adoptive transfer of Mϕ_{(PKCδ siRNA)} reduced infarct size (Figure 8, C & D), mimicking the effects of CDC_exo (Figure 2 & Figure 7). Thus, our data as well as published work are consistent with the hypothesis that PKCδ inhibition by miR-181b in CDC_exo underlies the cardioprotection induced by CDCs (Figure 8E).

Figure 8. Adoptive transfer of Mϕ with reduced levels of PKCδ (prkcd) is cardioprotective.

(A), Mϕ treated with miR-181b mimic have reduced prkcd expression (n=3–4/group). (B), Mϕ treated with PKCδ siRNA (10nM) have reduced prkcd expression (n=3/group). (C), Representative images of TTC stained hearts 48hrs following IR injury and adoptive transfer of Mϕ as indicated: Mϕ_{(PKCδ siRNA)}: Mϕ pre-treated with PKCδ siRNA; Mϕ_(Scramble): Mϕ pre-treated with scramble siRNA. (D), Quantification of percent infarct mass in (C) (n=5/group). (E), Schematic of our working hypothesis: (i), CDCs secrete exosomes (CDC_exo) that are cardioprotective (evidence in Figures 1–3); (ii), CDC_exo, rich in miR-181b, alter the polarization state of Mϕ (Figures 4–6); (iii), enrichment of miR-181b within exosomes recapitulates cardioprotection (Figure 7); (iv), downregulation of PKCδ in Mϕ by miR-181b is responsible, at least in part, for cardioprotective response of CDC_exo (Figure 8). Graphs depict mean +/− SEM. Statistical significance was determined using Student’s t-test. *P<0.05.

Despite the compelling evidence pinpointing PKCδ as a likely downstream effector, it is clear that exosomes and their miRNAs have protean effects, which may be synergistic; picking a single candidate may oversimplify the actual biology. Supplemental Table 3 lists other miR-181b targets that have been implicated in Mϕ polarization, and which were downregulated in CDC_exo-primed Mϕ. Conversely, we observed upregulation (and downregulation in miR-181a/b KO thymocytes; Figure 6F, lower right quadrant) in several genes associated with Mϕ polarization, oxidative phosphorylation, and phagocytosis (Supplemental Table 4). Together, these data point to a dramatic shift in Mϕ polarization by CDC_exo that reduces proinflammatory signaling and enhances phagocytosis to promote a cardioprotective response in vivo.

DISCUSSION

Myocardial IR injury leads to necrosis and apoptosis of the underlying myocardium (myocytes, endothelial cells, etc.)³⁴. The acute inflammatory response that follows IR injury is a physiologically archetypal sequence of events that begins with the mobilization and infiltration of neutrophils and follows with Mϕ influx³⁵. Several studies have highlighted the importance of Mϕ in models of myocardial tissue injury, including Mϕ depletion studies in MI^{2, 16, 36}, regenerative models of MI⁴, or apical myocardial resection³⁷. It is also becoming clearer that Mϕ heterogeneity (differences in polarization state in a spectrum typically defined by M1 and M2 states at the extremes³⁸) is a nuanced process that is highly dependent upon microenvironmental cues (protein-dependent³⁸, extracellular vesicles³⁹, or pH/ion imbalance⁴⁰).

Over the past decade, cell therapy has demonstrated some success in the treatment of ischemic disease, despite limited cell retention⁴¹. Our group has recently demonstrated in pigs¹⁷ and rats¹⁶ the existence of a novel cardioprotective process known as cellular postconditioning: CDCs delivered within 30 minutes of reperfusion dramatically reduce infarct size. Macrophages are required for cellular postconditioning, and appear to work by adopting a unique highly-phagocytic polarization state (non-M1, non-M2) in response to CDCs¹⁶. We have also shown that CDC exosomes (CDC_exo) increase viable mass, promote angiogenesis, and reduce scar size in established MI^{9, 14}.

Exosomes are intriguing paracrine signals that can shuttle payloads from cell to cell¹². Not surprisingly, exosomes are extremely diverse in their contents and vary greatly among cell types. Their characteristic traits are reflected in their vesicular composition (lipid, protein, small RNA, etc.), heterogeneity (totality of loaded and unloaded vesicles), and concentration (number of particles secreted) in vivo and in vitro²¹, which are further modulated in time and space by environmental cues⁴². Our lab^{9, 14} and at least 2 others^{10, 43} have demonstrated the bioactivity of exosomes derived from cardiac stem cell populations.

Here, we demonstrate that CDC_exo are a unique population of exosomes that have a distinct protein expression pattern (CD63, HSP70, and CD105) and miR profile (including high expression of miR-181b). In contrast to exosomes secreted by cardiac fibroblasts, CDC_exo do not exacerbate tissue injury⁴⁴; instead, they confer cardioprotection when delivered following reperfusion by localizing to the ischemic tissue, reducing infarct size, reducing the total number of CD68+ Mϕ, and shifting the polarization state of Mϕ toward a distinct phenotype. Interestingly, CDC_exo appear to specifically home to the site of injury within infarcted myocardium. Localization to this area of interest may confer a local restructuring of the microenvironment and polarize recruited monocytes/Mϕ toward a reparative phenotype. Although the mechanisms for CDC_exo localization remain to be defined, homing signals, such as angiotensin II type I receptor (targets exosomes to cardiomyocytes, skeletal muscle, and mesenteric resistance arterioles)⁴⁵, may be responsible for exosome retention at the site of injury. Furthermore, CDC_exo enhance the endogenous phagocytic capacity of Mϕ, a characteristic trait that may enhance clearance of necrotic cell debris and alleviate excessive proinflammatory stress within the infarcted heart³⁵ (a salutary process known as efferocytosis⁴⁶).

This work further expands our understanding of how CDCs confer cardioprotection and Mϕ polarization¹⁶. By reverse pathway analysis on transcriptomic data from both CDC_exo (cargo) and CDC_exo-primed Mϕ (recipient), we were able to pinpoint one key RNA component through which CDC_exo confer these functional benefits. Probing our own sequencing data against publicly-available gene expression and miRNA target prediction databases, we identified miR-181b as our top candidate; this miRNA species displayed the highest relative abundance within CDC_exo and the strongest correlation in its downregulation of target genes in CDC_exo-primed Mϕ. miR-181b has been implicated in the attenuation of NF-κB signaling in endothelial cells^{29, 47} and suppression of myeloid differentiation in acute myeloid leukemia²⁷, consistent with our findings that CDC_exo-primed Mϕ reduce proinflammatory gene expression and polarize Mϕ away from an M1 phenotype. Furthermore, the observed suppression of the miR-181b target PKCδ and concomitant increases in genes involved in translation and oxidative phosphorylation hint that CDC_exo shift the metabolic profile of Mϕ⁴⁸ toward a cardioprotective phenotype. Adding a miR-181b mimic to otherwise-inert Fb_exo qualitatively reproduces CDC_exo cardioprotection, while blocking miR-181b in CDC_exo blunts cardioprotection. Although the evidence points to the functional efficacy of miR-181b in reducing infarct size and Mϕ polarization, the lack of complete congruence with our CDC_exo data is not surprising. CDC_exo contain a multitude of small RNAs and proteins, not only miR-181b. Many of the other components undoubtedly are bioactive, exerting the overall functional benefits as an ensemble.

Our data provide compelling evidence for the role of CDC_exo in polarizing Mϕ to a phenotype that improves the healing response following injury. Specifically, we demonstrate that miR-181b is a key component within CDC_exo that confers cardioprotection by targeting PKCδ within Mϕ to induce a distinctive polarization state. These data add to the list of cardioprotective/cardioregenerative properties previously attributed to CDC_exo (anti-fibrotic effects; reduction of cardiomyocyte death and hypertrophy; myocardial regeneration; improved cardiac function; and angiogenesis)^{9, 11, 14}. Although we define here miR-181b as one important component within CDC_exo, the remaining composition (other noncoding RNAs, as well as proteins and lipids) of CDC_exo requires further examination⁴⁹ to unravel the details of synthesis, release, homing, pathway regulation, and cellular uptake of CDC_exo in vitro and in vivo. Likewise, while we have implicated PKCδ as one downstream contributor to CDC_exo-mediated cardioprotection, other plausible effectors remain to be investigated. Such redundancy would be consistent with the notion that exosomes contain a plethora of bioactive molecules which can target multiple signaling pathways synergistically⁵⁰.

CLINICAL PERSPECTIVE.

What Is New?

• Here we discover that exosomes mediate the cardioprotective effects of cardiosphere-derived cells (CDCs) administered after reperfusion in rats with acute myocardial infarction.

• Treatment with either CDCs or their secreted exosomes reduces infarct size and improves functional recovery.

• A specific microRNA species within CDC-derived exosomes, acting on macrophages, is implicated as a key mediator of the cardioprotective benefits.

What Are the Clinical Implications?

• Allogeneic CDCs are already in advanced clinical testing for reduction of chronic scar when administered months after myocardial infarction (ALLSTAR trial; ClinicalTrials.gov Identifier NCT01458405).

• The present findings give new reason to test the idea that allogeneic CDCs may be efficacious in preventing scar formation and improving cardiac function when given in the early reperfusion period adjunctive to percutaneous coronary intervention.