Therapeutic manipulation of angiogenesis with miR-27b

Background Multiple studies demonstrated pro-angiogenic effects of microRNA (miR)-27b. Its targets include Notch ligand Dll4, Sprouty (Spry)-2, PPARγ and Semaphorin (SEMA) 6A. miR-27 effects in the heart are context-dependent: although it is necessary for ventricular maturation, targeted overexpression in cardiomyocytes causes hypertrophy and dysfunction during development. Despite significant recent advances, therapeutic potential of miR-27b in cardiovascular disease and its effects in adult heart remain unexplored. Here, we assessed the therapeutic potential of miR-27b mimics and inhibitors in rodent models of ischemic disease and cancer. Methods We have used a number of models to demonstrate the effects of miR-27b mimicry and inhibition in vivo, including subcutaneous Matrigel plug assay, mouse models of hind limb ischemia and myocardial infarction and subcutaneous Lewis Lung carcinoma. Results Using mouse model of myocardial infarction due to the coronary artery ligation, we showed that miR-27b mimic had overall beneficial effects, including increased vascularization, decreased fibrosis and increased ejection fraction. In mouse model of critical limb ischemia, miR-27b mimic also improved tissue re-vascularization and perfusion. In both models, miR-27b mimic clearly decreased macrophage recruitment to the site of hypoxic injury. In contrast, miR-27b increased the recruitment of bone marrow derived cells to the neovasculature, as was shown using mice reconstituted with fluorescence-tagged bone marrow. These effects were due, at least in part, to the decreased expression of Dll4, PPARγ and IL10. In contrast, blocking miR-27b significantly decreased vascularization and reduced growth of subcutaneous tumors and decreased BMDCs recruitment to the tumor vasculature. Conclusions Our study demonstrates the utility of manipulating miR-27b levels in the treatment of cardiovascular disease and cancer. Electronic supplementary material The online version of this article (doi:10.1186/s13221-015-0031-1) contains supplementary material, which is available to authorized users.

Growth Factor (VEGF). The success of anti-angiogenics is most striking in patients with age-related macular degeneration where they normalize the vasculature, curtail vascular leakage/edema and restore visual acuity [ 4].
Bevacizumab is FDA-approved for metastatic cancers, including colon, lung, ovaries and glioblastoma multiforme of the brain, where it is used with standard chemotherapy and significantly improves the outcomes. However, the use of VEGF-targeting agents is associated with severe systemic toxicities, which stem from its role in cell survival and include severe hypertension and thromboembolism [ 5], renal failure [ 6], intestinal perforation [ 7,8], and neuropathies [ 9]. Thus, the need remains for the new, safer anti-angiogenic treatments [ 10].
On the other hand, pathologic conditions including peripheral arterial disease (PAD) and myocardial infarction (MI), in which ischemia causes irreversible tissue damage could benefit from enhanced angiogenesis and subsequent accelerated reperfusion and improved tissue homeostasis [ 11]. Gene therapy with angiogenic factors such as VEGF, basic fibroblast growth factor (bFGF) and hepatocyte growth factor (HGF) was very effective in preclinical models of critical limb ischemia (CLI) and MI [ 12-15]. However, randomized, double blinded placebo controlled clinical studies showed limited or no improvement [ 16,17]. Another approach to revascularization of ischemic tissue is represented by stem cell-based therapies, which show benefits in clinical and preclinical studies [ 18,19]. However, larger trials are needed to confirm the improvement of the clinically relevant outcomes. Here, we provide experimental evidence that therapeutic manipulation of miR-27b can be beneficial in angiogenesisdependent disease. Limited local application of miR-27b mimic was sufficient to restore angiogenesis and improve tissue function in two models of ischemic disease, hind limb ischemia, which mimics CLI, and coronary artery ligation, a MI model. Importantly, miR-27b increased the recruitment of bone marrow derived cells (BMDCs) to the neovasculature, as was shown in mice reconstituted with GFP-tagged bone marrow. In both models, miR-27b significantly decreased inflammation as was evidenced by decreased macrophage recruitment at the site of ischemic injury. These effects were due, at least in part, to the decreased expression of Notch ligand Dll4, PPARγ and of inflammatory cytokines. In a contrasting approach, blocking miR-27b by systemic administration of an anti-miR oligonucleotide significantly decreased microvacular density (MVD) and delayed the growth of highly aggressive subcutaneous tumors (Levis Lung Carcinoma, LLC-1). Surprisingly, treatment with anti-miR-27b also reduced the recruitment and activation of tumor-associated macrophages (TAMs). Together, our results demonstrate potential utility of manipulating miR-27b levels in the treatment of cardiovascular disease and cancer as well as associated inflammatory processes and indicate several molecular mechanisms involved.

Matrigel plug assay
Matrigel assay was performed as described previously [ 29]. Ice-cold matrigel (BD Biosciences, San Jose, CA) was supplemented with Heparin 60U/ml, with or without VEGF (200 ng/ml). miR27b and NC mimic (20 μg/mouse) were formulated with in vivo-JetPEI (PolyPlus Transfection, France) in 5 % glucose, according to manufacturer's instructions, and added, where indicated. Matrigel (0.4 ml/mouse) was injected subcutaneously in the median abdominal area of anesthetized nude mice and allowed to solidify. After 10 days, the plugs were excised and snapfrozen for further analysis.

Induction of hind-limb ischemia (HLI)
Ischemia was induced surgically as described previously [ 30]. FVB mice (10-12-weeks old) were anesthetized with Avertin (tribromoethanol, 250 mg/kg, Sigma, St. Louis, MO). Ischemia was induced by ligation of the femoral artery at the site of bifurcation, the epigastric artery and at the bifurcation of the femoral artery, and the popliteal artery followed by the removal of the entire femoral artery. The animals were followed for 14 days with visual inspection and laser Doppler imaging [ 31]. On day 14, the animals were euthanized and the quadriceps and gastrocnemius muscle harvested and frozen in TissueTek OCT compound (Fisher).

Laser doppler imaging (LDI)
Blood flow in the hind legs of anesthetized animals was measured under standardized conditions, with Laser Doppler Imager (Moor Instruments, UK). The mice were kept on a heating pad (37 °C) to minimize variation between flow measurements. The perfusion ratio between ligated and intact leg was calculated for each time point.

Induction of myocardial infarction by coronary artery ligation (CLI)
C57/Bl6 mice (10 weeks old) were anesthetized with Avertin (250 mg/kg) and orally intubated with a 22-gauge catheter, for artificial ventilation with a respirator (Harvard Apparatus). A left intercostal thoracotomy was performed and the ribs were retracted with 5-0 polypropylene sutures to open the pericardium. The left anterior descending (LAD) branch of the left coronary artery was ligated under dissecting microscope (Leika), distal to the bifurcation between LAD and diagonal branch with 8-0 polypropylene sutures. After positive end-expiratory pressure was applied to fully inflate the lung, the chest was closed with 7-0 polypropylene sutures [ 32].

Physiological assessment of left-ventricular function
Trans-thoracic 2-dimensional measurements were performed using a high-resolution echocardiography system (VEVO 770™, VisualSonics Inc., Toronto, Canada) equipped with a 30-MHz transducer. Measurements were taken14 and 28 days post-MI on mice anesthetized with a mixture of 1.5 % isoflurane and oxygen (1 L/min). M-mode tracings were used to measure LV wall thickness, end-systolic diameter (LVESD), and end-diastolic diameter (LVEDD). Systolic and diastolic left-ventricular areas were determined by M-mode in long-axis configuration and fractional shortening (FS) was measured at the mid-ventricular level. The left-ventricular chamber volumes in diastole and systole were derived from their respective measured 2D areas using a LV volume algorithm within the Vevo770 echo software. Cardiac ejection fraction was determined offline by the equation: [EF = (Diastolic Volume − Systolic Volume/Diastolic Volume) × 100 %].

Irradiation of mice and bone marrow transplantation
C57/Bl6 female mice (6-8 weeks old) were subjected to sub-lethal whole body irradiation (500 rads/mouse). Bone marrow was isolated by flushing the femurs and tibias of isogenic mice, Tg (CAG-EGFP) 13OsbLeySopj, with a 26½-gauge needle and passing through a 70-um mesh sterile filter. After 1-day rest, the mice were transplanted with the bone marrow cells (3 × 10 cells/mouse in 200 μl), by tail vein injection through a 27½-gauge needle. The recipient mice were treated with 1.6 mg/ml Sulphametoxazole/Trimethoprim (Bristol-Meyers-Squibb, New York, NY) in drinking water and allowed 8 weeks recovery.

Tumor implantation
C67Bl6 mice (12-14 weeks) reconstituted with the GFP-tagged bone marrow were injected subcutaneously in the right hindquarters with of 0.5 × 10 Lewis Lung Carcinoma cells (LLC1, ATCC, Manassis, VA) in 0.1 mL PBS. When tumors reached 5 mm in diameter, animals were randomized into groups ( n = 5) and treated with miR-27b mimic, inhibitor and negative control complexes prepared in sterile 5 % glucose using in vivo-jetPEI (PolyPlus Transfection, France).

Treatment of mice
After hind limb surgery or MI, mice were randomized into groups ( n ≥ 6). In HLI model, mice were treated by intramuscular injections of miR27b and NC mimics into the quadriceps muscle (10 μg/mouse/day, 2 injections in a total of 50 μl, via 27-gauge needle). miRNA were formulated with jetPEI in 5 % glucose. For the myocardial infarction (MI) model, miR-27b and NC mimics formulated as above were administered at the time of surgery by intracardiac injections (10 μg/mouse, two 10 μl injections), using a 10 μl Hamilton syringe with a 30-gauge needle.
Mice with LLC tumors (5-7 mm diameter) were randomized into groups ( n = 5) and treated with intraperitoneal injections of NC, miR-27b mimic or inhibitor, formulated as above (100 μg/100 μl/mouse). At the endpoint mice were sacrificed and tissues harvested for analysis.

Immunohistochemistry
All antibodies and dilutions used for this procedure are listed in Table 1below. To detect blood vessels, 5-μm thick matrigel plug or tumor cryosections were stained with rat anti-mouse CD31 antibody (1:100, BD Pharmingen, San Jose, CA) followed by donkey anti-rat Rhodamine Red-X conjugated secondary antibodies (1:1000, Jackson Immunoresearch). Dll4 in frozen sections was detected using rabbit anti-mouse antibody (1:100, Abcam, Cambridge, MA) followed by goat anti-rabbit Alexa Fluor 488 conjugate (Jackson Immunoresearch, West Grove, PA). The nuclei were visualized with 4′,6-diamidine-2-phenyl indole (DAPI). Macrophages were detected with rat anti-mouse antibody against F4/80 cell surface antigen (1:100, Affymetrix, Santa Clara, CA) followed by donkey anti-rat Rhodamine conjugate (Jackson Immunoresearch). GFP on frozen sections was visualized with rabbit anti-mouse antibody (1:1000, Abcam, Cambridge, MA) and donkey anti-rabbit Alexa Fluor 488 conjugate (Life Technologies, Grand Island, NY). Digital images of immunostained sections were obtained with Nikon Diaphot 2000 epifluorescence. Paraffin-embedded tissues were processed and stained at Northwestern University Pathology Core to detect endothelial cell marker CD31, smooth muscle actin (SMA), Collagen (Col) I, macrophage marker F4/80 and with Masson Trichrome reagent to detect fibrotic areas. Electroporation of cells RAW 264.7 macrophages were transfected by electroporation using D-032 Nucleofector device with Amaxa Cell Line Nucleofector kit V (Lonza, Walkersville, MD) following manufacturer's instructions. For every 2 × 10 30 nM miR-27b mimic, inhibitor, NC RNAi or 2 μg GFP plasmid (transfection control) were in added. After electroporation, the cells were grown 48 h and total RNA for RT-PCR was extracted as described above.

Morphometric and statistical analysis
Images were analyzed in a blinded fashion with NIS-Elements BR software (Nikon Inc., USA). GraphPad Prism software (Version 5.01, La Jolla, USA) was used for statistical analyses. For quantitative traits, all datasets were subjected to descriptive statistical analysis to establish normality, using Kolmogorov-Smirnov and D'Agostino-Pearson normality tests. Normally distributed datasets were compared pairwise or as a group using 2-tailed Student's t test or Tukey's multiple comparisons test. Non-normally distributed datasets were compared using Mann-Whitney test for single-time measurements. For multiple time points, two-way ANOVA was used followed by Bonferroni posttest for normally distributed data and Kruskal-Wallis for non-normal distribution. P < 0.05 was considered significant.

Quantitative real-time polymerase chain reaction (Q-PCR)
Total RNA was extracted with RNeasy isolation kit (Qiagen, Valencia, CA), per manufacturer's instructions. Dll4, Spry-2, Col I and SMA were amplified with Sso Advanced SYBR Green Supermix kit (Bio-Rad, Hercules, CA) with the primers specific for mouse genes ( Table 2).

Table 2
Primers used in the study   are shown. P value is calculated using Wilcoxon Signed Rank test. Note elevated BMDC recruitment in response to miR-27b. ( e) Dual immunofluorescence for the endothelial marker, CD31 ( red ) and GFP-positive BMDCs ( green ). Note minimal co-localization between CD31 and GFP ( yellow ) Neovasculature can be assembled from the endothelial cells formed by local proliferation of pre-existing endothelium, or by incorporation of the circulating precursors that originate in the bone marrow (bone marrow derived cells, BMDCs) [ 33]. To assess the contribution of the BMDCs to the pro-angiogenic function of miR-27b we used Matrigel plug assay in mice transplanted with GFP-tagged bone marrow from Tg (CAG-EGFP) 13OsbLeySopj mice. miR-27b significantly increased recruitment of GFP-positive BMDCs to the site of VEGF-induced neovascularization (Fig. 1c , d ). However, in contrast with previously published results [ 34], the majority of the GFP-positive BMDCs cells neither assumed endothelial fate nor were incorporated in the neovessels, as was evidenced by the lack of co-localization between GFP and the endothelial marker CD31 (Fig. 1e ).

miR-27b enhanced angiogenesis and improves perfusion in the ischemic hind limb
We then tested whether miR-27b similarly accelerates angiogenesis in the ischemic muscle. Hind Limb Ischemia (HLI) in FVB mice was induced by femoral artery ligation. After surgery, the mice received intramuscular (IM) injections of miR-27b mimic or negative control (non-silencing RNAi) for three consecutive days. Treatment with miR-27b mimic significantly improved tissue vascularization, as was shown by immunofluorescence for the endothelial marker, CD31 (Fig. 2a -c ). In agreement, miR-27b mimic caused a statistically significant improvement in tissue perfusion by day seven post-ligation, as measured by Doppler Laser Imaging (DLI) (Fig. 2d , e ).

miR-27b preserved angiogenesis and protected tissue function in the infarcted heart
In the mouse model of myocardial infarction (MI) caused by coronary artery ligation (CLI), a single application of miR-27b mimic had strong protective effect on the microvasculature as was determined by IHC for CD31, with 1.6fold higher MVD in the infarcted area compared to control group (Fig. 3a , b ). Improved vascularization was associated with a significant degree of tissue protection in the animals treated with miR-27b mimic. MiR-27b significantly alleviated fibrosis, as was measured by Masson Trichrome staining (Fig. 3c , d ), and Collagen 1 deposition and expression as was determined by IHC, and RT-PCR, respectively (Fig. 3e , h ). Similar trend was observed for smooth muscle actin (Additional file 1: Figure S1). Importantly, these changes caused by miR-27b mimicry resulted in improved cardiac function, with an approximately 2-fold increase in ejection fraction over control by day 14 (Fig. 3f ) and a trend towards increased fractional shortening (Fig. 3g ).

miR-27b inhibitor suppresses cancer growth and angiogenesis
Angiogenesis is a rate-limiting factor in tumor progression. We therefore examined the effect of miR-27b inhibitor on tumor angiogenesis and associated growth rates. We used Lewis Lung carcinoma, a particularly aggressive mouse tumor cell line, in a syngeneic subcutaneous model. Systemic treatment with miR-27b inhibitor, 27b(I) at 5 mg/kg caused a significant reduction of tumor weight at the endpoint (Fig. 4a c , Additional file 1: Figure S5). In contrast, miR-27b mimic, 27b(M) had only moderate, albeit statistically significant effect on tumor growth compared to control groups (untreated mice and mice treated with control RNAi). In agreement, treatment with anti-miR-27b significantly reduced tumor MVD, while miR-27b had only modest effect on already robust angiogenesis in the LLC tumors, (Fig. 4d , e ).  Figures S2 and S5). IF staining showed elevated Dll4 levels in ischemic calf muscle, which was significantly reduced in the presence of miR-27b mimic (Fig. 5a , b ). Similar decrease in Dll4 mRNA was observed by real-time PCR in cardiac tissue of mice treated with miR-27b mimic (Additional file 1: Figure S3). Spry-2, another known target of miR-27b was not significantly altered (Additional file 1: Figure S4). In subcutaneous LLC tumors, miR-27b had disparate effects on the total and vascular Dll4 expression. LLC tumors showed high baseline levels of Dll4 as was measured by immunofluorescence, and these levels were modestly but significantly increased by 27b(M) and moderately reduced in response to 27b(I) (Fig. 5d , e ). In contrast, colocalization analysis with the vascular endothelial marker, CD31, showed a significant reduction of Dll4 specifically on tumor vasculature caused by miR-27b mimic; in agreement, miR-27b inhibitor clearly augmented Dll4 expression by the tumor vasculature (Fig. 5f , g )

miR-27b alters macrophage infiltration in ischemic and tumor tissues
miR-27b was previously implicated in the regulation of macrophage responses [ 39]. We therefore analyzed macrophage infiltration in ischemic and tumor tissues where miR-27b was added (mimic) or blocked (anti-miR). miR-27b significantly attenuated macrophage infiltration to the infarcted area, as was evidenced by IHC for macrophage marker, F4/80 (Fig. 6a ). In LLC tumors F4/80, immunofluorescence showed no changes in macrophage infiltration in response to 27b(M). Surprisingly, there was a significant decrease in F4/80 positivity upon treatment with 27b(I). In LPS-treated macrophages, miR-27b was previously shown to target PPARγ and alter the downstream regulation of inflammatory cytokines [ 39]. In agreement, electroporation of RAW 264.7 macrophages with 27b(M) caused an approximately 2-fold reduction of PPARγ mRNA (Fig. 6e ). On the other hand, treatment with 27b(I) caused a robust increase of PPARγ message (Fig. 6e ), which was consistent with decreased macrophage migration/recruitment to the tumor (Fig. 6b , d ). Importantly, miR-27b blockade also altered macrophage polarization, which was reflected by the alterations in relative abundance of the M1 and M2 markers, IL12 and IL10 (Fig. 6g , f ).

Discussion
Therapeutic manipulation of angiogenesis in cancer and cardiovascular disease has been explored in pre-clinical models with considerable success. However, toxic effects of current anti-angiogenic therapies necessitate the development of safer options [ 40], and the development of pro-angiogenic therapies is still in progress [41][42][43].
miRNA regulators of angiogenesis present an attractive therapeutic option, because they act via relatively subtle regulation of multiple target genes within the same or related pathways [ 44,45]. Such a mechanism is more likely to yield sufficient therapeutic benefits with lesser toxicities, because each single target is not fully inactivated. Here, we explore the possibility of therapeutic manipulation of angiogenesis by using mimicry or inactivation, as appropriate, of a pro-angiogenic microRNA, miR-27b. Interestingly, miR-27b deficit could contribute to PAD indirectly, via lipid metabolism. Studies support the role of miR-27b as cholesterol hub in the liver [ 53]. Hepatic miR-27b is increased in response to hyperlipidemia and, in turn, regulates several key genes in control of lipid metabolism (Angptl3, Gpam) [ 54]. Moreover, hepatic miR-27b is downregulated and its targets elevated in mouse models of dyslipidemia/atherosclerosis [ 54]. In addition, miR-27b is decreased in differentiating adipocytes and inhibits their proliferation and accumulation via PPARγ/RXRα controlled pathways [ 55,56]. This function of miR-27b could indirectly benefit PAD patients.
In our short-term study, we observed increased vascularization of the muscles of the heart and extremities subjected to ischemic stress, concomitant with preserved tissue integrity and function, suggestive of the therapeutic utility of miR-27b for acute treatment of critical limb ischemia and myocardial infarction. There are similar findings regarding other members of the miR-23/27/24 cluster distinct from miR-27b [ 57]. It is unclear, however, whether long-term application of miR-27b and/or other miRNA from this cluster could be harmful for cardiomyocytes and require additional targeting to the vascular endothelium, or if there would be long-term beneficial effects on PAD via attenuated lipid metabolism and fat tissue development.
Earlier studies have demonstrated the ability of miR-27b to enhance/restore angiogenic function of the BMDCs in diabetic mice (Db/Db), causing accelerated wound healing [ 58]. Similarly, we observed the increased recruitment of GFP-tagged bone marrow derived cells to the vasculature. However, the lack of conversion of myeloid BMDCS to the endothelial (CD31-positive) phenotype in vivo suggests that the majority of BMDCs recruited in response to miR-27b treatment have an accessory role where they serve as the source of pro-angiogenic and/or survival factors for the neovasculature.
In cancer, the role of miR-27b is controversial, with studies lending support to its tumor-promoting and tumorsuppressive roles. miR-27b has been identified as a classical oncomir in cell-based assays [  Previous studies indicate that miR-27b contributes towards inflammatory processes by destabilization of PPARγ mRNA and protein [ 39]. On the other hand, it can block inflammatory responses by targeting NF-κB [ 67] or TGF-β target, Gremlin-1 [ 52]. In contrast with published studies, PPARγ expression in cultured macrophages was only modestly affected by mir-27b mimicry, however, it was significantly upregulated by miR-27b inhibitor. On the other hand, miR-27b altered cytokine profile of the cultured macrophages with increased IL-10 and decreased IL-12 levels suggestive of alternative (non-inflammatory) mode of activation. These findings suggest that miR-27 inhibitors may have beneficial effects in cancer setting by blocking the recruitment and activation of tumor-associated macrophages. The reduced inflammatory infiltrates in ischemic tissue may be secondary to anti-fibrotic action of miR-27b. Alternatively, they could be due to the down-modulation of the pro-inflammatory NF-κB. Since miR-27b itself is repressed by TGF-β [ 68], its reconstitution with mimics may be especially pertinent in the context of inflammation and fibrosis.
Together, our findings demonstrate potential utility of manipulating miR-27b levels in cardiovascular disease and cancer and confirm its targets as Dll4/Notch axis, PPARγ and its downstream effectors.

Conclusions
The main non-surgical therapy for ischemic disease involves mechanical stents to maintain blood flow through the diseased vessels. The use of drug-eluting stents was not yet successful. Several experimental strategies aim to improve vascularization in the lower extremities and in the heart by promoting the formation of collateral vessels. They employ stem cell and gene therapies. In stem cell based approaches, induced pluripotent stem cells or bone marrow monocytes are used to enhance local angiogenesis. Gene therapy approaches to enhance local expression of several key growth factors, including VEGF, bFGF and HGF, are subject to the common limitations of gene therapies. None of these approaches have yet reached clinic. We offer targeting the miR-27b-regulated pathways with miR-27bbased drugs for the treatment of cardiovascular disease. Manipulations of microRNA offer safer and more effective alternative to gene therapy. In addition, local use of oligonucleotide mimics of miR-27b facilitates the recruitment of endogenous bone marrow-derived stem cells to the newly forming vasculature at the target site. Thus we provide an alternative to cell-based cardiovascular therapies.