A MicroRNA93-IRF9-IRG1-Itaconic Acid Pathway Modulates M2-like-Macrophage Polarization to Revascularize Ischemic Muscle

Abstract

Background

Currently no therapies exist for treating, and improving outcomes in patients with severe peripheral arterial disease (PAD). MicroRNA93 (miR93) has been shown to favorably modulate angiogenesis and reduce tissue loss in genetic PAD models. However, the cell specific function, downstream mechanisms or signaling involved in miR93 mediated ischemic muscle neovascularization is not clear. Macrophages were best known to modulate arteriogenic response in PAD and the extent of arteriogenic response induced by macrophages is dependent on greater M2 to M1-activation/polarization state. In the current study, we identified a novel mechanism by which miR93 regulates macrophage-polarization to promote angiogenesis and arteriogenesis to revascularize ischemic muscle in experimental-PAD.

Methods

In vitro (macrophages, endothelial cells, skeletal muscle cells under normal and hypoxia serum starvation (HSS) conditions) and in vivo experiments in preclinical-PAD models (unilateral femoral artery ligation and resection)) were conducted to examine the role of miR93-interferon regulatory factor-9 (IRF9)-immune responsive gene-1 (IRG1)-itaconic acid pathway in macrophage-polarization, angiogenesis, arteriogenesis and perfusion recovery.

Results

In vivo, compared to wild type (WT) controls, miR106b-93-25 cluster deficient mice (miR106b-93-25^−/−) showed decreased angiogenesis and arteriogenesis correlating with increased M1-like-macrophages following experimental-PAD. Intra-muscular delivery of miR93 in miR106b-93-25^−/− PAD mice increased angiogenesis, arteriogenesis, the extent of perfusion which correlated with more M2-like-macrophages in the proximal and distal hind-limb muscles. In vitro, miR93 promotes and sustains M2-like-polarization even under M1-like-polarizing conditions (HSS). Delivery of bone marrow derived macrophages from miR106b-93-25^−/− to WT ischemic-muscle decreased angiogenesis, arteriogenesis and perfusion, while transfer of wild-type macrophages to miR106b-93-25^−/− had the opposite effect. Systematic analysis of top-differentially upregulated genes from RNA-sequencing between miR106b-93-25^−/− and WT ischemic-muscle showed that miR93 regulates IRG1 function to modulate itaconic acid production and macrophage-polarization. 3′UTR luciferase-assays performed to determine whether IRG1 is a direct target of miR93 revealed that IRG1 is not a miR93 target but IRF9 that can regulate IRG1-expression is a miR93 target. In vitro, increased expression of IRF9, IRG1 and itaconic acid treatment significantly decreased endothelial angiogenic potential.

Conclusion

We conclude that miR93 inhibits IRF9 to decrease IRG1-itaconic acid production to induce M2-like-polarization in ischemic muscle to enhance angiogenesis, arteriogenesis and perfusion recovery in experimental-PAD.

INTRODUCTION

Peripheral arterial disease (PAD) is a major atherosclerotic complication where occlusions in arteries to organs other than the heart and brain cause tissue ischemia and legs are most frequently affected⁽¹⁾. Severe PAD often results in pain at rest and non-healing ulcers that can lead to amputation⁽¹⁾. Due to the atherosclerotic occlusions, blood flow to the distal tissue becomes dependent on the degree of vascular remodeling (arteriogenic and angiogenic) in the ischemic-muscle. MicroRNAs (miR) are short non-coding RNA sequences (~18–22 nucleotide length) that are classically thought to regulate expression of a single, or several functionally related, target mRNAs by specific base-pairing with the binding sites in their 3′ untranslated region^(2,3,4). miR93 is one of the microRNAs in miR106b-93-25 cluster that has been demonstrated to promote tumor angiogenesis⁽⁵⁾. We have previously shown that miR93 improved the extent of angiogenesis in the distal ischemic-muscle and perfusion in experimental-PAD⁽⁶⁾. Furthermore, miR106b-93-25 cluster has been shown to play an important role in regulating bone-marrow derived stromal cells recruitment into ischemic-muscle to promote perfusion⁽⁷⁾. However, the downstream mechanisms or the signaling involved in miR93 mediated perfusion recovery are not completely understood.

A number of different resident and mobilized cells are active participants in forming new blood vessels. In experimental-PAD, monocyte/macrophage recruitment into the occluded vessel can induce a potent arteriogenic response^(8,9,10), however limited information is available on the role of macrophage activation state in regulating this process. Classically activated macrophages (M1-polarized) in response to lipopolysaccharide (LPS) and/or interferon-γ are pro-inflammatory and cytotoxic⁽¹¹⁾ and alternatively activated macrophages (M2-polarized), seen in response to IL-4 and IL-13, are anti-inflammatory and reparative⁽¹¹⁾. Though, clearly, intermediate states exist between the M1 and M2, these macrophage subtypes⁽¹²⁾ are considered to be dominant modulators in several cardiovascular pathologies⁽¹³⁾. The role of macrophage-polarization in modulating an angiogenic response in distal ischemic-muscle to regulate neovascularization is not clear. While a few genes including Syndecan-4⁽¹⁴⁾, Interleukin-19⁽¹⁵⁾, Protease activated receptor-2⁽¹⁶⁾ and Prolyl hydroxylase domain protein-2^(10,17) were shown to regulate macrophage-polarization in experimental-PAD, data on whether microRNAs can regulate macrophage-polarization in ischemic skeletal muscle is extremely limited. Knowing that miR93 can inhibit inflammatory cytokine production in LPS stimulated macrophages⁽¹⁸⁾, we sought to determine the role of miR93 in macrophage-polarization, arteriogenesis and angiogenesis, as well as establishing the potential genes regulated by miR93 in macrophages. Macrophages that infiltrate the ischemic-muscle are polarized to M1-phenotype due to the existing cytokine milieu in ischemic tissue and an increased M1-macrophages and/or sustained M1-phenotype can significantly lessen the arteriogenic response to decrease the perfusion recovery^(19,20). Furthermore, due to the reversible nature of macrophage-polarization (M2-phenotype to M1-phenotype and vice versa)⁽²¹⁾ it has been a challenge to pharmacologically sustain M2-macrophage-polarization to promote vascular remodeling and perfusion-recovery in ischemic muscle. Thus, there is significant interest in identifying therapies that can promote and sustain M2-phenotype in ischemic cardiovascular diseases including PAD.

METHODS

Expanded methods can be found in the online-only data supplement

Mice

Homozygous miR106b-93-25^−/− (on C57Bl/6 background) and C57Bl6 (WT) mice were bred in the University of Virginia (UVa) vivarium. Animal studies were approved by the University of Virginia-Institutional Animal Care and Use Committee.

Animal model of Hind limb ischemia (HLI) and perfusion-recovery

Unilateral femoral artery ligation and excision was performed on age and sex matched 12–16 week animals^(22,23,24).

Human samples

The Institutional Review Boards (IRB) at Duke University, University of Colorado and University of Virginia approved the research protocols with human samples.

Cell Transfections

microRNA transfections were performed by SiPORT NeoFX reagent and plasmid transfections were performed by Lipofectamine-3000 according to manufacturers’ instructions.

Statistics

GraphPad Prism-6 was used to determine the statistical significance of the data. Equality of Variance was checked by Brown-Forsythe test and Welch’s correction was applied to the T-test analysis of the data that did not pass the equal variance test. Statistical test for each individual experiment was provided in the Figure legend.

RESULTS

Decreased vascular-density in miR106b-93-25^−/− ischemic-muscle correlates inversely with pro-inflammatory M1-like-macrophage numbers

Following experimental-PAD, laser Doppler perfusion imaging showed significantly less blood flow in ischemic-foot of miR106b-93-25^−/− mice (Supplemental Figure-1) vs. WT (p<0.05, d21-post HLI-WT:79.2± 5.5 vs. miR106b-93-25^−/−:55.7±6.0, Fig-1A). In this HLI model, shear stress in the occluded in-flow vessels (in adductor-muscle) induces arteriogenesis and ischemia in the distal leg (gastrocnemius-muscle) induces angiogenesis^{(25,26,27,28)}. Hence, we examined arteriogenic induction in the adductor-muscle by α-smooth muscle actin (SMA) immunostaining and both arteriogenic and angiogenic induction in gastrocnemius-muscle by α-SMA and CD31 immunostaining. α-SMA immunostaining in the ischemic adductor-muscle (IAM) from miR106b-93-25^−/− vs. WT showed significantly lower (~2X) arteriolar density (%average of α-SMA⁺ vascular structures >50μm/image, p=0.002, Fig-1B). Furthermore, immunostaining of α-SMA and CD31 (%average of CD31⁺ cells/muscle fiber area⁽²⁹⁾) in ischemic gastrocnemius-muscle (IGA) from miR106b-93-25^−/− vs. WT showed a significant decrease in arteriogenesis (~3X, p=0.0001, Fig-1C) and angiogenesis (p>0.03, Fig-1D).

Fig-1.

miR106b-93-25^−/− ischemic muscle has lower perfusion, angiogenesis and arteriogenesis compared to WT ischemic muscle. A) Perfusion-recovery was measured non-invasively by quantifying microvascular blood flow by laser Doppler in WT-wild type (C57Bl6) and miR106b-93-25^−/− (homozygous mice deficient in miR106b~25 cluster) HLI mice. n=7, *P<0.05 indicates time points that are significantly different from d0 baseline perfusion within the same group by repeated measures ANOVA with Dunnetts post-test. #P<0.05 indicates the specific time points that are significantly different between 2 groups by Unpaired T-test. B) Arteriolar density (α-smooth muscle actin+ vessels bigger than >50uM) in WT (C57Bl6) and miR106b-93-25^−/− ischemic-muscle d21 post HLI. n=7, P<0.05 considered significant. Unpaired T-test. IAM-Adductor-muscle for ischemic leg. C) Arteriolar density (α-smooth muscle actin+ vessels bigger than >50uM) in WT (C57Bl6) and miR106b-93-25^−/− ischemic-muscle d21 post HLI. n=7, P<0.05 considered significant. Unpaired T test. IGA-Ischemic gastrocnemius-muscle. D) Microvascular density (CD31+ vessels/muscle fiber area) in WT (C57Bl6) and miR106b-93-25^−/− ischemic-muscle d21 post HLI. n=7. *P<0.05 considered significant. Unpaired T test. Data presented as Mean ± SEM.

We examined if decreased vascular-density in miR106b-93-25^−/− ischemic-muscle correlates with differences in macrophage numbers and/or with their polarization state by flow cytometry (gating strategy presented in Supplemental Figure-2). In non-ischemic-muscle there was no significant difference in the total macrophage numbers (CD11b⁺F4/80⁺), pro-inflammatory (M1-like, CD11b⁺F4/80⁺CD80^high) or anti-inflammatory (M2-like, CD11b⁺F4/80⁺CD206⁺) macrophages in miR106b-93-25^−/− non-ischemic gastrocnemius (NGA) or non-ischemic adductor (NAM) muscle vs. WT (Supplemental Figure-3A–F). Analysis of tissue-infiltrating^{(30,31,32,33)} macrophages showed significantly lower M2-like-macrophages (F4/80⁺CX3CR1⁺CD206⁺) in both NGA (~3X, p=0.009, Supplemental Figure-3I) and NAM (~2X, p=0.01, Supplemental Figure-3L) of miR106b-93-25^−/− vs. WT (Supplemental Figure-3G–L). No significant differences in total, M1-like (F4/80⁺CX3CR1⁻CD80^high) or M2-like (F4/80⁺CX3CR1⁻CD206⁺) tissue-resident macrophages in NGA or NAM in miR106b-93-25^−/− vs. WT was observed (Supplemental Figure-3M–R).

In ischemic-muscle no significant difference in total macrophage numbers (IGA: Fig-2A and IAM: Fig-2D) between miR106b-93-25^−/− vs. WT was observed. However, an increase in M1-like-macrophages (IGA:~2X) and a significant decrease in M2-like-macrophages (IGA:~3X, p=0.009; IAM:~2X, p=0.01) in miR106b-93-25^−/− vs. WT was observed (Fig-2A–F). Analysis of tissue-infiltrating macrophages showed no significant differences in total macrophage numbers (IGA: Fig-2G and IAM: Fig-2J) between miR106b-93-25^−/− vs. WT. However, a significant decrease in M2-like tissue-infiltrating macrophages (IGA:~3X, p=0.008; IAM: ~5X, p=0.003) in miR106b-93-25^−/− vs. WT was observed (Fig-2G–L). Analysis of tissue-resident macrophages showed no significant differences in total or M1-like-macrophages but a significant decrease in M2-like-macrophages (IGA:~2X, p<0.023; IAM:~4x, p=0.005) in miR106b-93-25^−/− vs. WT (Fig-2M–R). Representative flow charts for Fig-2 are presented in Supplemental Figure-4.

Fig-2.

miR106b-93-25^−/− ischemic muscle has lower M2-like tissue resident and tissue infiltrating macrophages compared to WT ischemic muscle. A) % total live F4/80⁺ macrophages in WT and miR106b-93-25^−/− ischemic gastrocnemius-muscle (IGA). *P<0.025 considered significant. Unpaired T test. B, C) %CD80^high ((B), M1-like-polarized) and CD206⁺ ((C), M2-like-polarized) CD11b⁺F4/80⁺ macrophages in WT and miR106b-93-25^−/− IGA. *P<0.025 considered significant. Unpaired T test. D) % total live F4/80⁺ macrophages in WT and miR106b-93-25^−/− adductor-muscle from ischemic leg (IAM). *P<0.025 considered significant. Unpaired T test. E, F) % CD80^high ((E), M1-like-polarized) and CD206⁺ ((F), M2-like-polarized) CD11b⁺F4/80⁺ macrophages in WT and miR106b-93-25^−/− IAM. *P<0.025 considered significant. Unpaired T test. G) % infiltrating CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in WT and miR106b-93-25^−/− IGA. *P<0.025 considered significant. Unpaired T test. H, I) %CD80^high ((H), M1-like-polarized) and CD206⁺ ((I), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in WT and miR106b-93-25^−/− IGA. *P<0.025 considered significant. Unpaired T test. J) % infiltrating CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in WT and miR106b-93-25^−/− IAM. *P<0.025 considered significant. Unpaired T test. K, L) %CD80^high ((K), M1-like-polarized) and CD206⁺ ((L), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in WT and miR106b-93-25^−/− IAM. *P<0.025 considered significant. Unpaired T test. M) % resident CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in WT and miR106b-93-25^−/− IGA. *P<0.025 considered significant. Unpaired T test (Fig-2M-Unpaired T test with Welch’s correction). N, O) %CD80^high ((N), M1-like-polarized) and CD206⁺ ((O), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in WT and miR106b-93-25^−/− IGA. *P<0.025 considered significant. Unpaired T test with Welch’s correction. P) % resident CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in WT and miR106b-93-25^−/− IAM. *P<0.025 considered significant. Unpaired T test. Q, R) %CD80^high ((Q), M1-like-polarized) and CD206⁺ ((R), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in WT and miR106b-93-25^−/− IAM. *P<0.025 considered significant. Unpaired T test. n=4. Data presented as Mean ± SEM. Dot plots and histograms of the flow data are presented in Supplement-25B.

We next examined whether miR93 contributes to differences in bone-marrow niche by pre-disposing bone-marrow monocytes to pro-inflammatory M1-like-phenotype. While no significant differences in the total monocytes (CD11b⁺CD115⁺) or monocytes predisposed to M1-like phenotype (CD11b⁺CD115⁺CD80^high) were observed between miR106b-92-25^−/− non-ischemic leg bone-marrow vs. WT non-ischemic leg bone-marrow (Supplemental Figure-5A,B); or miR106b-93-25^−/− ischemic leg bone-marrow vs. WT ischemic leg bone-marrow (Supplement-5D,E), a significant decrease in monocytes predisposed to M2-like-phenotype (CD11b⁺CD115⁺CD206⁺) was observed between miR106b-93-25^−/− non-ischemic leg bone-marrow vs. WT non-ischemic leg bone-marrow (~6X, p=0.001, Supplemental Figure-5C) and miR106b-93-25^−/− ischemic leg bone-marrow vs. WT ischemic leg bone-marrow (~8X, p=0.01, Supplemental Figure-5F).

miR93 induces M2-like-macrophage polarization even under M1-like-polarizing conditions in vitro

We next determined which microRNA deficiency in the miR106b-93-25^−/− plays a role in modulating angiogenesis and macrophage-polarization. In vitro angiogenesis experiments revealed that miR93-mimic significantly induced endothelial-branching on growth factor reduced matrigel (GFRM, P<0.05, Supplemental Figure-6) vs. scrambled-mimic. While miR106b-mimic significantly decreased endothelial-branching, miR25-mimic showed no significant difference in endothelial-branching on GFRM vs. scrambled-mimic (P<0.05, Supplemental Figure-6), suggesting that lack of miR93 in miR106b-93-25 cluster might be responsible for decreased angiogenesis and arteriogenesis in the miR106b-93-25^−/− IGA vs. WT.

We next transfected macrophages (Raw264.7, a macrophage cell line) with either miR106b, miR93, miR25-mimics or scrambled-mimics (control, same concentration as mimics, Supplemental Figure-7A-C) under normal and HSS-conditions and examined for their polarization state by quantitative-PCR analysis of inducible nitric oxide synthase ((iNos), M1-polarization marker) and arginase1 ((Arg1), M2-polarization marker). qPCR analysis showed that macrophages treated with miR93-mimic significantly increased Arg1-expression (~6X, p=0.01, Fig-3A) vs. scrambled-mimic. miR106b-mimic or miR25-mimic did not induce any significant differences in Arg1 or iNos-expression vs. scrambled-mimic in macrophages under normal conditions (Supplemental Figure-8A, B).

Fig-3.

miR93 induces macrophage M2-like polarization in vitro. A, C) qPCR analysis of Arg1 and iNos-expression normalized to hypoxanthine phosphoribosyltransferase ((Hprt), house-keeping gene) expression in macrophages transfected with (A) scrambled mimic (Scrbd Mim, control) or miR93 Mimic (miR93 Mim); (C) scrambled antagomir (Scrbd Antg, control) or antagomir93 under normal conditions. *P<0.05 considered significant. Unpaired T test. B, D) qPCR analysis of Arg1 and iNos expression normalized to Hprt in macrophages transfected with (B) Scrambled Mimic or miR93 Mimic; (D) scrambled antagomir or antagomir93 under hypoxia serum starvation. *P<0.05 considered significant. n=6. Unpaired T test. E) In vitro capillary like tube formation on growth factor reduced matrigel assay of mouse endothelial cells (EOMA) treated with conditioned medium from scrambled mimic or miR93 mimic transfected macrophages. n=6, *P<0.05 considered significant. Unpaired T test. F) In vitro capillary like tube formation on growth factor reduced matrigel assay of mouse endothelial cells (EOMA) treated with conditioned medium from scrambled antagomir or antagomir93 transfected macrophages. n=6, *P<0.05 considered significant. Unpaired T test. G) Flow cytometry analysis CD80 and CD206 expression by human peripheral blood monocytes (HPBMNCs) gated on CD14 transfected with scrambled mimic or miR93 mimic under normal conditions. *P<0.05 considered significant. Unpaired T test. H) Flow cytometry analysis CD80 and CD206 expression by human peripheral blood monocytes gated on CD14 transfected with scrambled mimic or miR93 mimic under HSS conditions. n=4. *P<0.05 considered significant. Unpaired T test. I, J) qPCR analysis of CD80 and CD206 expression by HPBMNCs transfected with scrambled mimic or miR93 mimic under (I) normal or (J) HSS conditions. *P<0.05 considered significant. Unpaired T test. K, L) qPCR analysis of CD80 and CD206 expression by HPBMDMs transfected with scrambled mimic or miR93 mimic under (K) normal or (L) HSS conditions. n=4. *P<0.05 considered significant. Unpaired T test. Data presented as Mean ± SEM. M) Purity of CD14⁺ monocytes isolated from human peripheral blood, representative dot plots before and after positive selection. N) Human peripheral blood monocyte derived macrophages (HPBMDM) day 10 after culture. Representative dot plots of macrophage marker expression (CD68), green-isotype control, blue-CD68 staining.

We next determined whether miR93 (or miR106b and miR25) has the capacity to sustain M2-like-polarization under M1-like-polarizing conditions relevant for tissue ischemia found in PAD. Hence, we examined the status of macrophage-polarization under hypoxia serum starvation (HSS). qPCR analysis showed that HSS significantly induced M1-like-polarization, evident by a significant increase in iNos-expression and significant decrease in Arg1-expression (Supplemental Figure-9). qPCR analysis showed that miR93-transfection significantly induced Arg1-expression (p=0.01, Fig-3B) with no changes in iNos-expression in HSS-macrophages vs. scrambled-mimic. Increased M2-like-phenotype in macrophages under normal and HSS transfected with miR93-mimic inversely correlated with significantly decreased reactive oxygen species levels (ROS, measured by CellROX-green that emits photostable fluorescence upon oxidation by ROS) under normal (p=0.03, Supplemental Figure-10A) and HSS-conditions (p=0.01, Supplemental Figure-10B). While miR106b-mimic did not induce any changes in Arg1 or iNos-expression in HSS-macrophages compared to scrambled-mimic (Supplemental Figure-8C), miR25-mimic significantly decreased the expression of both Arg1 (p=0.01) and iNos (p=0.002) (no difference in the ratio of Arg1:iNos) compared to scrambled-mimic in HSS-macrophages (Supplemental Figure-8D).

We next inhibited miR93 in macrophages under normal and HSS-conditions by transfecting them with antagomir93 (miR93-inhibitor, Supplemental Figure-6D) to confirm the role of miR93 in regulating macrophage-polarization. Macrophages treated with scrambled-antagomir (control, same concentration as antagomir93) were used as control. qPCR analysis showed that antagomir93 significantly decreased Arg1-expression (~5X, p=0.01, Fig-3C, left-panel) with a concomitant increase in iNos-expression (p=0.008, Fig-3C, right-panel) vs. scrambled-antagomir under normal-conditions. Antagomir93 significantly decreased Arg1-expression (p=0.001, Fig-3D) with no changes in iNos-expression vs. scrambled-antagomir in HSS-macrophages.

In order to determine whether M2-like-polarization induced by miR93 in macrophages can enhance angiogenic-capacity of endothelial cells, we next treated endothelial cells (EOMA) with conditioned media (CM) from macrophages transfected with miR93-mimic, antagomir93 or respective scrambled controls in vitro. Endothelial cells treated with CM from miR93-mimic demonstrated significantly more capillary-like tube formation (p<0.05, Fig-3E) and CM from antagomir93 treatment significantly less (p=0.002, Fig-3F) capillary-like tube formation vs. respective controls. This data strongly suggested that increased miR93-expression induces M2-like-polarization in macrophages to promote angiogenesis.

We next transfected freshly isolated HPBMNCs with miR93-mimic or scrambled-mimics under normal or HSS-conditions. Flow cytometry showed that miR93-mimic significantly increased the numbers of CD206⁺ HPBMNCs vs. scrambled-mimic under normal (~2X, p=0.008, Fig-3G) and HSS-conditions (p<0.05, Fig-3H). qPCR analysis of HPBMNCs transfected with miR93 also showed a significant increase in CD206-expression vs. scrambled-mimic under normal (p<0.05, Fig-3I) and HSS-conditions (p=0.002, Fig-3J). Furthermore, qPCR analysis of HPBMDMs transfected with miR93 also showed a significant increase in CD206-expression vs. scrambled-mimic under normal (p=0.04, Fig-3K) and HSS-conditions (p=0.02, Fig-3L).

Intra-muscular injection of miR106b-93-25^−/− bone marrow derived macrophages (BMDM) decrease vascular density and improves perfusion-recovery in WT ischemic-muscle

We next examined the polarization state of BMDMs from WT and miR106b-93-25^−/− mice cultured under normal and HSS-conditions in vitro. qPCR analysis showed no significant differences in either Arg1 or iNos-expression under normal-conditions in miR106b-93-25^−/−-BMDM vs. WT-BMDM (Fig-4A). However, HSS significantly induced iNos-expression (~5X, p=0.01) with a concomitant decrease in Arg1-expression (~7X, p=0.0006) in miR106b-93-25^−/−-BMDM vs. WT-BMDM (Fig-4B).

Fig-4.

miR106b-93-25^−/− BMDM impair perfusion, decrease angiogenesis and arteriogenesis in WT ischemic muscle. A) qPCR analysis of Arg-1 and iNos expression normalized to Hprt in WT (wild type, C57Bl6) and miR106b-93-25^−/− bone-marrow derived macrophages (BMDM) under normal growth conditions. Unpaired T-test, n=6. p<0.05 considered significant. B) qPCR analysis of Arg-1 and iNos expression normalized to Hprt in WT (wild type, C57Bl6) and miR106b-93-25^−/− BMDM under hypoxia serum starvation. Unpaired T-test, n=6. p<0.05 considered significant. C) Laser Doppler analysis of microvascular blood flow in WT-wild type (C57Bl6) HLI mice that received miR106b-93-25^−/− BMDM and miR106b-93-25^−/− HLI mice that received WT BMDM. n=7, *P<0.05 indicates time points that are significantly different from d0 baseline perfusion within the same group by repeated measures ANOVA with Dunnetts post-test. #P<0.05 indicates the specific time points that are significantly different between 2 groups by Unpaired T-test. D) Arteriolar density (α-smooth muscle actin+ vessels bigger than >50uM) in ischemic gastrocnemius-muscle (IGA) from WT-wild type (C57Bl6) HLI mice that received miR106b-93-25^−/− BMDM and miR106b-93-25^−/− HLI mice that received WT BMDM, d21 post HLI. n=7, P<0.05 considered significant. Unpaired T test. E) Microvascular density (CD31+ vessels/muscle fiber area) in IGA from WT-wild type (C57Bl6) HLI mice that received miR106b-93-25^−/− BMDM and miR106b-93-25^−/− HLI mice that received WT BMDM, d21 post HLI. n=7, P<0.05 considered significant. Unpaired T test. Data presented as Mean ± SEM.

The function of miR106b-93-25^−/−-macrophages in modulating perfusion-recovery was examined by direct intra-muscular injection of miR106b-93-25^−/−-BMDM to ischemic WT gastrocnemius-muscle, and vice versa. Intramuscular delivery of miR106b-93-25^−/−-BMDM (1×10⁶) into WT ischemic-muscle (IGA) significantly decreased perfusion in WT vs. miR106b-93-25^−/−-HLI mice that received WT-BMDM (d21:WT-BMDM to miR106b-93-25^−/−-HLI 65.8±2.0% vs. miR106b-93-25^−/−-BMDM to WT-HLI 48.9±4.7%, p=0.001, Fig-4C). Immunohistochemistry of α-SMA and CD31 in the IGA of WT-HLI mice that received miR106b-93-25^−/−-BMDM showed significantly decreased arteriolar-density (α-SMA:~3X, p<0.001, Fig-4D) and microvascular-density (CD31:p<0.03, Fig-4E) vs. miR106b-93-25^−/−-HLI mice that received WT-BMDM.

Additional controls that included WT-BMDM delivery to WT-HLI gastrocnemius-muscle and miR106b-93-25^−/− BMDM to miR106b-93-25^−/−-HLI mice showed a significant increase in WT-HLI mice that received WT-BMDM vs. miR106b-93-25^−/−-HLI mice that received miR106b-93-25^−/− BMDM (d21:WT-BMDM to WT-HLI 73.8±3.3% vs. miR106b-93-25^−/−-BMDM to miR106b-93-25^−/−-HLI 58.4±1.2%, Supplemental Figure-11).

Re-introducing miR93 rescues impaired perfusion in miR106b-93-25^−/− HLI mice by increasing anti-inflammatory macrophages and angiogenesis

To confirm the role of miR93 in modulating macrophage-polarization and perfusion, we next re-introduced miR93 (using a miR93 expressing-plasmid) into miR106b-93-25^−/− mice gastrocnemius-muscle (Supplemental Figure-12). Delivery of miR93 into miR106b-93-25^−/− HLI mice significantly improved perfusion vs. miR106b-93-25^−/− HLI mice that received scrambled (control)-plasmid (d21 post-HLI: miR106b-93-25^−/− HLI+miR93-plasmid 76.1±6.6% vs. miR106b-93-25^−/− HLI+control-plasmid 50.3±3.8%, p=0.007, Fig-5A). Immunostaining for α-SMA and CD31 in the IGA of miR106b-93-25^−/− HLI+miR93-plasmid showed a significant increase in arteriolar-density (α-SMA:~2X, p=0.004, Fig-5B) and microvascular-density (CD31:p<0.03, Fig-5C). Furthermore, while a significant increase in cell death (total and endothelial) in miR106b-93-25^−/−+control-plasmid (Total~5X, endothelial~3X, p<0.04) vs. WT was observed, miR93-delivery in miR106b-93-25^−/− ischemic-muscle significantly decreased cell death (total and endothelial) vs. miR106b-93-25^−/−+control-plasmid (Supplemental Figure-13). Since, miR106b-93-25^−/− mice have significantly lower M2-like-macrophages, we next examined whether re-introducing miR93 in miR106b-93-25^−/− mice can induce M2-like-polarization by flow cytometry.

Fig-5.

miR93 is sufficient to improve perfusion, arteriogenesis and angiogenesis in miR106b-93-25^−/− ischemic muscle. A) Laser Doppler analysis of microvascular blood flow in miR106b-93-25^−/− HLI mice that received scrambled plasmid (control) and miR106b-93-25^−/− HLI mice that received miR93 expressing plasmid. n=7, *P<0.05 indicates time points that are significantly different from d0 baseline perfusion within the same group by repeated measures ANOVA with Dunnetts post-test. #P<0.05 indicates the specific time points that are significantly different between 2 groups by Unpaired T-test. B) Arteriolar density (α-smooth muscle actin+ vessels bigger than >50uM) in ischemic gastrocnemius-muscle (IGA) from miR106b-93-25^−/− HLI treated with scrambled plasmid and miR106b-93-25^−/− HLI mice that received miR93 expressing plasmid, d21 post HLI. n=7, P<0.05 considered significant. Unpaired T test. C) Microvascular density (CD31+ vessels/muscle fiber area) in IGA from miR106b-93-25^−/− HLI treated with scrambled plasmid and miR106b-93-25^−/− HLI mice that received miR93 expressing plasmid. n=7, P<0.05 considered significant. Unpaired T test. Data presented as Mean ± SEM.

In non-ischemic-muscle there was no significant difference in the numbers of total (Supplemental Figure-14A-F), tissue-infiltrating (Supplemental Figure-14G–L), tissue-resident (Supplemental Figure-14M–R) macrophages or their M1-like or M2-like states between miR106b-93-25^−/−+miR93-expressing-plasmid and miR106b-93-25^−/−+control-plasmid.

In the ischemic-muscle, no significant differences in total or M1-like-macrophages between miR106b-93-25^−/−+miR93-expressing-plasmid and miR106b-93-25^−/−+control-plasmid was observed (IGA: Fig-6A,B; IAM: Fig-6D,E). However, a significant increase in M2-like-macrophages (IGA:~2X, p=0.0004; IAM:~2X, p=0.022) was observed in miR106b-93-25^−/−+miR93-expressing-plasmid vs. miR106b-93-25^−/−+control-plasmid (Fig-6C,F). Analysis of tissue-infiltrating macrophages showed no significant difference in total macrophage numbers but a significant decrease in M1-like-macrophages (IGA:~2X, p=0.009; IAM:p=0.01) with a concomitant increase in M2-like-macrophages (IGA:~2X, p=0.0004; IAM:~4X, p<0.0001) in miR106b-93-25^−/−+miR93-expressing-plasmid vs. miR106b-93-25^−/−+control-plasmid(Fig-6G–L). Analysis of tissue-resident macrophages showed no significant differences in either total or M1-like-macrophages but a significant increase in M2-like-macrophages (IGA:~2X, p=0.002; IAM:~2X, p=0.0001) in miR106b-93-25^−/−+miR93-expressing plasmid vs. miR106b-93-25^−/−+control-plasmid (Fig-6M–R). Representative flow charts for Fig-6 are presented in Supplemental Figure-15.

Fig-6.

miR93 induces M2-like macrophages in miR106b-93-25^−/− ischemic muscle. A) % total live F4/80⁺ macrophages in ischemic gastrocnemius-muscle (IGA) of miR106b-93-25^−/− mice treated with control plasmid (con (scrambled) plmd) or miR93 expressing plasmid (miR93 Plmd)). *P<0.025 considered significant. Unpaired T test. B, C) %CD80^high ((B), M1-like-polarized) and CD206⁺ ((C), M2-like-polarized) CD11b⁺F4/80⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IGA. *P<0.025 considered significant. Unpaired T test. D) % total live F4/80⁺ macrophages in adductor-muscle of ischemic leg (IAM) from miR106b-93-25^−/− mice treated with Con or miR93 plasmid. *P<0.025 considered significant. Unpaired T test with Welch’s correction. E, F) % CD80^high ((E), M1-like-polarized) and CD206⁺ ((F), M2-like-polarized) CD11b⁺F4/80⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IAM. *P<0.025 considered significant. Unpaired T test with Welch’s correction. G) % infiltrating CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IGA. *P<0.025 considered significant. Unpaired T test with Welch’s correction. H, I) %CD80^high ((H), M1-like-polarized) and CD206⁺ ((I), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IGA. *P<0.025 considered significant. Unpaired T test with Welch’s correction. J) % infiltrating CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IAM. *P<0.025 considered significant. Unpaired T test. K, L) %CD80^high ((K), M1-like-polarized) and CD206⁺ ((L), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁺ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IAM. *P<0.025 considered significant. Unpaired T test. M) % resident CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IGA. *P<0.025 considered significant. Unpaired T test with Welch’s correction. N, O) %CD80^high ((N), M1-like-polarized) and CD206⁺ ((O), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IGA. *P<0.025 considered significant. Unpaired T test with Welch’s correction. P) % resident CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IAM. *P<0.05 considered significant. Unpaired T test with Welch’s correction. Q, R) %CD80^high ((Q), M1-like-polarized) and CD206⁺ ((R), M2-like-polarized) CD11b⁺F4/80⁺ CX3CR1⁻ macrophages in miR106b-93-25^−/− mice treated with Con or miR93 plasmid IAM. *P<0.025 considered significant. Unpaired T test with Welch’s correction. n=4. Data presented as Mean ± SEM. Dot plots and histograms of the flow data is presented in Supplement-25C.

In the bone-marrow (Supplemental Figure-16A–F) or in circulation (Supplemental Figure-16G–L), no significant differences in total monocyte numbers, or monocytes predisposed to M1 or M2-like phenotypes were observed between miR106b-93-25^−/− HLI mice treated with miR93-expressing-plasmid and scrambled plasmid.

RNA-sequencing between miR106b-93-25^−/− and WT ischemic-muscle demonstrates miR93 modulates IRG1-expression to regulate macrophage-polarization

To determine the downstream targets of miR93, we next performed RNA-Sequencing in miR106b-93-25^−/− and WT ischemic-muscle. RNA-sequencing demonstrated that macrophage inflammatory protein-2 (CXCL2), immunoresponsive gene-1 (IRG1) and GM3893 were the top-3 upregulated genes in miR106b-93-25^−/− ischemic-muscle vs. WT ischemic-muscle (Supplemental-Table-1) that are closely associated with immune function; and C1q and tumor necrosis factor related protein-3 (C1QTNF3), myosin Heavy Chain-3 (MYH3) and cardiac muscle α-actin-1 (ACTC1) are the top-3 downregulated genes in miR106b-93-25^−/− ischemic-muscle vs. WT ischemic-muscle (Supplemental-Table-2) associated with muscle function. Based on this data we hypothesized that increased M1-like-macropahges in miR106b-93-25^−/− ischemic muscle paracrinally aggravated skeletal muscle injury. To confirm our hypothesis, we examined the role of miR93 in modulating the paracrine effects of macrophages in regulating skeletal muscle (C2C12) cell survival, in vitro. TUNEL analysis showed that 1) CM from antagomir93 transfected macrophages (M1-like-phenotype) induced significantly higher muscle cell death vs. scrambled-antagomir (p=0.01, Supplemental Figure-17A); 2) CM from miR93-mimic transfected macrophages under HSS (M2-like-phenotype) significantly decreased skeletal muscle cell death vs. scrambled-mimic (p=0.03, Supplemental Figure-17B); 3) CM from miR93-mimic transfected macrophages (M2-like-phenotype) significantly decreased cell death in C2C12 cells under HSS vs. scrambled-mimic (p=0.01, Supplemental Figure-17C). These data suggested that miR93-deficiency results in M1-like-polarization that contributes to aggravated skeletal muscle injury through paracrine effects. Cytokine/chemokine analysis of the CM from HSS-macrophages transfected with miR93-mimic vs. scrambled-mimic showed significantly lower levels of cytokines that are consistent with the M1-phenotype including TNF-α, IL-1β, IL-6, IFN-γ in addition to IL-1α, IL-7, Eotaxin and KC (CXCL1) (p<0.05, Supplemental Figure-18). Furthermore, a significant increase in MIP-2, MCP-1, IP-10 and G-CSF levels were observed in CM from HSS-macrophages transfected with miR93-mimic vs. scrambled-mimic (p<0.05, Supplemental Figure-18). These data confirmed that miR93 inhibits M1-cytokine production by macrophages subject to HSS to increase skeletal muscle cell survival. We also examined the role of miR93 in regulating paracrine effects of skeletal muscle on macrophage-polarization in vitro. qPCR analysis showed that macrophages treated with CM collected from muscle cells (C2C12) transfected with miR93-mimic demonstrated increased Arg1-expression (p=0.007) without differences in iNos-expression when compared to CM from muscle cells transfected with scrambled-mimic, indicating miR93 results in paracrine effects to increase an M2-like-phenotype in macrophages (Supplemental Figure-17D) under normal conditions.

We next systematically examined the role of top upregulated genes in modulating 1) macrophage function and 2) if the genes are miR93-targets in vitro. qPCR analysis of CXCL2 showed significantly higher CXCL2-expression in miR106b-93-25^−/− ischemic-muscle (~18X, p<0.05, experimentally confirming RNA-Sequencing data) vs. WT; and miR106b-93-25^−/− ischemic-muscle treated with miR93-expressing-plasmid decreased CXCL2-expression comparable to WT (Supplemental Figure-19A,B). In vitro, CXCL2 treatment significantly decreased Arg1-expression in macrophages under normal and HSS conditions vs. respective untreated controls indicating that CXCL2 induces M1-like-polarization (Supplemental Figure-19C–F). However, antagomir93 treatment did not show any significant differences in CXCL2-expression under normal or HSS-conditions vs. respective controls indicating that CXCL2 is not regulated by miR93 in macrophages (Supplemental Figure-19G,H).

We next examined if IRG1 is modulated by miR93 in experimental-PAD. qPCR analysis of IRG1 showed no significant differences IRG1 levels between WT and miR106b-93-25^−/− non-ischemic-muscle (Fig-7A). However, miR106b-93-25^−/− ischemic-muscle had significantly higher IRG1-expression (3d post HLI, ~85X, p<0.05, confirming RNA-Sequencing data) vs. WT (Fig-7B); and miR106b-93-25^−/− ischemic-muscle treated with miR93-expressing-plasmid had decreased IRG1-expression comparable to WT (Fig-7B). We next examined whether miR93 regulates IRG1-expression in macrophages. Antagomir93 did not change IRG1-expression in macrophages under normal-conditions vs. control (Fig-7C) but under HSS-conditions, antagomir93 induced a significant increase in IRG1-expression (p=0.006) vs. control (Fig-7D). In addition, a significant increase in IRG1-expression was also observed in HSS-miR106b-93-25^−/− BMDM vs. HSS-WT-BMDM (~2X, p<0.03, Fig-7E). qPCR analysis of endothelial cells transfected with miR93-mimic showed significant decrease in IRG1-expression under normal (p=0.02, Supplemental Figure-20A) but not under HSS-conditions vs. scrambled-mimic (Supplemental Figure-20B). No changes in IRG1-expression were observed in C2C12 cells transfected with miR93-mimic under normal or HSS-conditions vs. scrambled-mimic (Supplemental Figure-20C,D).

Fig-7.

miR93 targets IRF9 to decrease IRG1-Itacanoic acid production in murine macrophages to induce a M2-like phenotype. A) qPCR analysis of IRG1 expression in non-ischemic gastrocnemius-muscle (NGA) normalized to Hprt in WT (wild type, C57Bl6) and miR106b-93-25^−/−. *P<0.05 considered significant. n=5, Unpaired T test. B) qPCR analysis of IRG1 expression in ischemic gastrocnemius-muscle (IGA) normalized to Hprt in WT, miR106b-93-25^−/− and miR106b-93-25^−/− mice treated with miR93 expressing plasmid. *P<0.05 considered significant. n=5, Unpaired T test. C, D) qPCR analysis of IRG1 expression in macrophages transfected with scrambled antagomir (Scrbd-Antg) or antagomir93 (Antg93) under normal or hypoxia serum starvation (HSS) conditions. n=6, *P<0.05 considered significant. Unpaired T test. E) qPCR analysis of IRG1 expression in HSS bone-marrow derived macrophages (BMDM) from WT (C57Bl6) and miR106b-93-25^−/− mice. n=6, *P<0.05 considered significant. Unpaired T test. F) qPCR analysis of miR93 expression normalized to Sno202 (house-keeping micro RNA) in macrophages under normal and HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. G) qPCR analysis of IRG1 expression normalized to Hprt in macrophages under normal and HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. H) Ratio of Arg-1:iNos expression in macrophages transfected with control (con, scrambled) or IRG1 expressing plasmid (IRG-1 Exp-Plmd) under normal growth conditions. n=6, *P<0.05 considered significant. Unpaired T test. I) Ratio of Arg-1:iNos in macrophages transfected with control (Con, scrambled) plasmid and IRG1 Exp-Plmd under HSS. n=6, *P<0.05 considered significant. Unpaired T test. J) Ratio of Arg-1:iNos expression in macrophages transfected with Scrbd-Antg + Scrambled plasmid (Con Plmd) or antagomir93 (Antg93) + IRG1 Crispr-cas9 knock out plasmid (IRG1 KO plmd) under normal growth conditions. n=6, *P<0.05 considered significant. Unpaired T test. K) Ratio of Arg-1:iNos expression in macrophages transfected with Scrbd-Antg + Scrambled plasmid (Con Plmd) or antagomir93 (Antg93) + IRG1 crispr-cas9 knock out plasmid (IRG1 KO plmd) under normal growth conditions. n=6, *P<0.05 considered significant. Unpaired T test. L) qPCR analysis of IRF9 expression in macrophages transfected with scrambled mimic (Scrbd Mim) or miR93 mimic (miR93 Mim) under normal and HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. M) qPCR analysis of IRF9 expression in macrophages transfected with Scrbd Antg or Antg93 under normal and HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. N) Luciferase activity in macrophages transfected with IRF9-3′UTR-luciferase plasmid followed by transfection with scrbd mim or miR93 mim under normal or HSS conditions. n=4, *P<0.05 considered significant. Unpaired T test. O) qPCR analysis of IRG1 expression in macrophages transfected with IRF9-expressing plasmid (IRF9 Plmd) or Control Plmd under normal or HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. P) qPCR analysis of Arg1 and iNos expression in macrophages transfected with IRF9 Plmd or Con Plmd under normal conditions. n=6, *P<0.05 considered significant. Unpaired T test. Q) qPCR analysis of Arg1 and iNos expression in macrophages transfected with IRF9 Plmd or Con Plmd under HSS conditions. n=6, *P<0.05 considered significant. Unpaired T test. Data presented as Mean±SEM

Furthermore, qPCR analysis showed a significant decrease in miR93-levels in macrophages under HSS vs. control conditions (p<0.0001, Fig-7F) with a concomitant increase in IRG1-levels in HSS-macrophages vs. control conditions (p<0.0001, Fig-7G). Decreased miR93-levels and increased IRG1-levels in macrophages under HSS vs. control conditions correlated with increased M1-like phenotype in macrophages. We next examined the function of IRG1 in modulating macrophage-polarization. In vitro, IRG1 over-expression (Supplemental Figure-21A) in macrophages significantly decreased Arg1:iNos-ratio under normal (p=0.001, Fig-7H) and HSS-conditions (p=0.01, Fig-7I).

We next wanted to determine whether miR93 targeting IRG1 regulates macrophage-polarization. Since, miR93-inhibition induced IRG1-expression and correlated with increased iNos-expression, we predicted that IRG1-inhibition in antagomir93 transfected macrophages will result in decreased iNos-expression and/or increased Arg1-expression; confirming that IRG1 regulates miR93 mediated macrophage-polarization. Hence, we transfected macrophages with IRG1-Crispr/Cas9-knock-out (IRG1-KO) plasmid followed by antagomir93 transfection (Supplemental Figure-21B). Macrophages treated with equal concentrations of scrambled-plasmid and scrambled-antagomir served as controls. qPCR analysis of Arg1 and iNos-expression in macrophages transfected with IRG1-KO-plasmid and antagomir93 under normal and HSS-conditions showed a significant increase in Arg1:iNos-expression under normal-conditions (p=0.0004, Fig-7J) but no significant difference in Arg1:iNos-expression under HSS-conditions (Fig-7K). These data showed that miR93 regulates IRG1-expression to modulate macrophage-polarization in experimental-PAD.

miR93 targets IRF9 to regulate IRG1-expression and macrophage-polarization

To examine whether IRG1 is a direct miR93-target, macrophages were transfected with IRG1-3′UTR luciferase plasmid and later with miR93-mimic or scrambled-mimics and tested for differences in IRG1-3′UTR luciferase-activity. No significant differences in luciferase-activity in macrophages expressing IRG1-3′UTR between miR93-mimic and scrambled-mimic either under normal or HSS-conditions (Supplemental Figure-22) were observed. Since, IRG1 did not appear to be a direct miR93-target, we predicted that miR93 targets another gene that has the ability to modulate IRG1-expression. Recent report has shown that interferon regulatory factor (IRF)-1 can transcriptionally regulate IRG1-expression⁽³⁴⁾ and miRanda algorithm predicted that IRF1 and IRF9 are miR93-targets. qPCR analysis showed miR93-mimic did not decrease IRF1-expression (Supplement-23A,B); antagomir93 did not increase IRF1-expression (Supplemental Figure-23C,D) in normal or HSS-macrophages vs. controls.

We next examined whether IRF9 is a miR93-target in macrophages. qPCR analysis showed that miR93-mimic significantly decreased IRF9-expression in normal (p<0.05, Fig-7M, left-panel) and HSS (p=0.02, Fig-7M, right-panel) macrophages vs. scrambled-mimic. Furthermore, antagomir93 significantly induced IRF9-expression in both normal (p-0.005, Fig-7M, left-panel) and HSS (p=0.01, Fig-7M, right-panel) macrophages vs. scrambled-antagomir suggesting miR93 regulates IRF9-expression in macrophages. qPCR analysis showed that increased IRF9-expression in HSS conditions inversely (p<0.0001, Supplemental Figure-24) correlated with decreased miR93 levels (Fig-7F). qPCR analysis of miR93-mimic transfected endothelial cells showed a significant decrease in IRF9-expression under normal (p=0.01) but not HSS-conditions vs. scrambled-mimic (Supplemental Figure-25A,B). No changes in IRF9-expression were observed in C2C12 cells transfected with miR93-mimic vs. scrambled-mimic under normal or HSS-conditions (Supplemental Figure-25C,D).

To examine whether IRF9 is a direct miR93-target, macrophages were transfected with IRF9-3′UTR luciferase expressing plasmid and later with miR93-mimic or scrambled-mimic and tested for differences in luciferase-activity. miR93-mimic significantly decreased luciferase-activity in macrophages expressing IRF9-3′UTR vs. scrambled-mimic both under normal (p=0.003, Fig-7N, left-panel) and HSS-conditions (p<0.05, Fig-7N, right-panel) indicating that IRF9 is miR93-target in macrophages. Furthermore, qPCR analysis showed that IRF9-overexpression (Supplemental Figure-26) in macrophages significantly induced IRG1-expression both under normal (~16X, p=0.001, Fig-7O, left-panel) and HSS-conditions (p=0.007, Fig-7, right-panel) vs. scrambled-plasmid. IRF9-overexpression significantly induced iNos-expression (~10X) without significant changes in Arg1-expression under normal (p=0.002, Fig-7P, right-panel) and HSS-conditions (p<0.05, Fig-7Q, right-panel) vs. scrambled-plasmid. These data indicated that miR93 directly targets IRF9 to decrease IRG1-expression in macrophages to induce a M2-like-phenotype under normal and HSS-conditions.

In vitro, endothelial cells transfected with IRF9 expressing-plasmid showed significantly lower endothelial-branching on GFRM vs. scrambled-plasmid (p=0.003, Supplemental Figure-27A). IRG1 expressing-plasmid also significantly decreased endothelial-branching on GFRM vs. scrambled-plasmid (p=0.001, Supplemental Figure-27B).

In consideration of future miR93 therapeutics, we sought to determine the potential for modulation of the miR93-IRF9-IRG1 axis in humans with PAD. We compared muscle biopsies from a subgroup of PAD and age- gender matched control subjects from a prior NIH study⁽³⁵⁾. qPCR analysis showed lower levels of miR93 in PAD muscle-biopsies compared to controls (~3X, P=0.04, Fig-8A). Lower levels of miR93 in PAD muscle biopsies was associated with higher levels of IRF9 (~2X, p=0.03, Fig-8B) and IRG1-expression (~12X, p=0.04, Fig-8C) compared to normal. This suggests that miR-93 over-expression could promote anti-inflammatory macrophages to induce angiogenesis and/or arteriogenesis in humans with PAD.

Fig-8.

Decreased miR93 levels correlate with activation of IRF9-IRG1-Itaconic acid pathway in human PAD and human monocyte/macrophages. A) qPCR analysis of miR93 expression in normal and PAD muscle-biopsies normal (NL) n=8, PAD n=8, *P<0.05 considered significant. Unpaired T test. B) qPCR analysis of IRF9 expression in normal and PAD muscle-biopsies normal (NL) n=8, PAD n=8, *P<0.05 considered significant. Unpaired T test. C) qPCR analysis of IRG1 expression in normal and PAD muscle-biopsies normal (NL) n=8, PAD n=8, *P<0.05 considered significant. Unpaired T test. D) qPCR analysis of IRF9 expression in HPBMNCs transfected with Scrbd Mim or miR93 Mim under normal and HSS conditions. n=4, *P<0.05 considered significant. Unpaired T test. E) qPCR analysis of IRG1 expression in HPBMNCs transfected with Scrbd Mim or miR93 Mim under normal and HSS conditions. n=4, *P<0.05 considered significant. Unpaired T test. F) qPCR analysis of IRF9 expression in HPBMDMs transfected with Scrbd Mim or miR93 Mim under normal and HSS conditions. n=4, *P<0.05 considered significant. Unpaired T test. G) qPCR analysis of IRG1 expression in HPBMDMs transfected with Scrbd Mim or miR93 Mim under normal and HSS conditions. n=4, *P<0.05 considered significant. Unpaired T test. H) qPCR analysis of Arg-1 and iNos expression in macrophages treated with itaconic acid under normal growth conditions. n=6, *P<0.05 considered significant. One-way ANOVA with Dunnetts post-test. I) qPCR analysis of Arg-1 and iNos expression in macrophages treated with itaconic acid under HSS. n=6, *P<0.05 considered significant. One-way ANOVA with Dunnetts post-test. J) LC-MS/MS analysis of Itaconic acid levels in macrophages transfected with scrambled or antagomir93 under normal or HSS conditions. n=4, *P<0.05 considered significant. Unpaired T-test. Data presented as Mean±SEM. K) Schematic of miR93 regulation of macrophage polarization in PAD.

Subsequently, we transfected freshly isolated HPBMNCs as well as HPBMDMs with miR93-mimic or scrambled-mimic under normal or HSS-conditions and examined differences in IRF9 and IRG1-expression. qPCR analysis showed that miR93-mimic significantly decreased normal and HSS-HPBMNCs IRF9 (normal-p=0.02, HSS-p=0.03, Fig-8D) and IRG1 (normal-~5X, p=0.01, HSS-~2X, p=0.003, Fig-8E) expression; and normal and HSS-HPBMDMs IRF9 (normal-p=0.03, HSS-~5X, p=0.04, Fig-8F) and IRG1 (normal-p=0.02, HSS-p=0.0002, Fig-8G) expression vs. scrambled-mimic confirming that miR93-IRF9-IRG1 pathway is active in human monocyte/macrophages.

miR93 decreases IRG1 and itaconic acid production in macrophages to induce M2-like-polarization

Based on the previously published data demonstrating IRG1 regulates itaconic acid production in LPS treated macrophages to modulate macrophage function⁽³⁶⁾, we next examined whether increased itaconic acid production due to increased IRG1-expression (upon miR93-inhibition) in macrophages regulates macrophage-polarization. Macrophages treated with 5mM and 10mM (concentrations chosen based on itaconic acid levels in LPS treated macrophages⁽³⁶⁾) showed a significant decrease in Arg1-expression and a significant increase in iNos-expression (p<0.05, Fig-8H) vs. untreated controls under normal-conditions. Under HSS-conditions, while 5mM of itaconic acid significantly decreased Arg1-expression, no significant difference in Arg1-expression was observed with 10mM of itaconic acid treatment (p<0.05, Fig-8I, left panel). However, both 5mM and 10mM of itaconic acid treatments significantly induced iNos-expression (p<0.05, Fig-8I, right panel) vs. untreated HSS-controls. These data indicated that increased itaconic acid levels in macrophages can induce a pro-inflammatory phenotype. In vitro, Itaconic acid significantly decreased endothelial-branching in a dose dependent manner vs. untreated controls (p>0.05, Supplemental Figure-25C). These data indicated that the induction of IRF9-IRG1-itaconic acid pathway also has the ability to decrease angiogenesis.

To obtain a direct correlation between miR93 and itaconic acid production, we next examined whether miR93-inhibition can induce itaconic acid levels by liquid chromatography-tandem mass spectroscopy (LC-MS/MS) analysis. Correlating with increased IRG1-levels by miR93-inhibition, antagomir93 transfection induced a significant increase in the levels of itaconic acid production in normal and HSS-macrophages vs. scrambled antagomir (normal p=0.02, HSS p=0.01 Fig-8J).

DISCUSSION

The extent of skeletal muscle perfusion-recovery post-HLI depends on favorable synergistic function among multiple cell types^{(19,20,21,15)} and is related to the degree of both angiogenic and arteriogenic induction following experimental-PAD^(28,37,26). The ability of macrophages to favorably modulate these processes is dependent on their polarized state, especially in diseased tissue^(19,9,17). M2-like-macrophages play critical roles in inflammation resolution by secreting anti-inflammatory cytokines including IL-10, TGF-β that helps in tissue repair^(38,39) and by secreting growth factors^(40,41) that induces arteriogenesis and angiogenesis. Based on these functions, it is well accepted that increased M2-like-macrophages are beneficial for the recovery of ischemic-muscle^(20,19). However, the first wave of macrophages that infiltrates the ischemic-muscle are polarized to a proinflammatory-phenotype due to the highly inflammatory and ischemic-environment existing in the PAD-muscle thereby contributing to aggravation of tissue damage and/or inhibition of the recovery/adaptive processes^(20,19). Therapies that are aimed at increasing anti-inflammatory-macrophage population in ischemic-muscle are extremely limited and are often met with unsuccessful outcomes due to the reversible nature of M2 to M1-like-polarized states and vice-versa^{(20,19,42,13)}. Hence, therapies that not only increase M2-like-macrophages but can also sustain the M2-like-phenotype even in M1-like-polarizing conditions are much more likely to be successful as a therapeutic in PAD.

General studies aimed at understanding the molecular mechanism in macrophage-polarization include LPS/IFN-γ treatments for M1-like-polarization and IL-4/IL-10 for M2-like-polarization⁽¹²⁾. However, we show that hypoxia serum starvation, a plausible in vitro counterpart to in vivo ischemic challenge⁽⁶⁾, can itself induce a distinct proinflammatory-phenotype without any external cytokine effect, thus enabling us to study the role of miR93 in modulating macrophage activation in more pathology relevant condition. An important finding of our study is the finding that macrophages, specifically M2-like-macrophages, have the ability to induce angiogenesis in the ischemic gastrocnemius-muscle. Several previous studies have demonstrated that monocytes/macrophages recruited into the adductor-muscle at the site of vascular occlusion induce collateralization/arteriogenesis^{(10,43,44,45,46)}. However, a specific role of M2-like-macrophages in inducing angiogenesis to improve the microvascular perfusion in the ischemic gastrocnemius-muscle is not clear. In our current study we demonstrate the ability of M2-like-macrophages to induce ischemic-muscle angiogenesis and arteriogenesis by adoptive-transfer of BMDM experiments which showed that the miR106b-93-25^−/− ischemic gastrocnemius-muscle that received WT-BMDM have significantly improved vascular-density compared to WT ischemic-muscle that received miR106b-93-25^−/− BMDM. These data not only show a novel aspect of cell-specific miR93 function but also point towards the translational potential of miR93 in PAD.

In our current study, genetic deficiency of miR93 (along with mir106b and miR25) significantly decreased perfusion⁽⁷⁾ suggesting that miR93 also plays a role in regulating perfusion consistent with our earlier publication⁽⁶⁾. Reintroduction of miR93 in miR106b-93-25 deficient mice improved perfusion to similar levels as WT (C57Bl/6, d21 post-HLI: C57BL6-79.23±5.5% vs. miR106b-93-25^−/−-55.78±6% vs. miR106b-93-25^−/−+miR93-expressing plasmid-76.17±6.6%) indicating that miR93 (among miR106b-93-25 cluster) is sufficient to promote perfusion-recovery in experimental-PAD. Our in vitro experiments showing anti-angiogenic function of miR106b and no angiogenic effect of miR25 points the significance of miR93 in regulating angiogenesis. In addition to increased angiogenesis following miR93 delivery, recovery was also associated with increased arteriogenesis, decreased cell death, and significantly increased anti-inflammatory resident and infiltrating-macrophages. Increased anti-inflammatory-macrophages in miR93 treated ischemic leg correlated with increased angiogenesis, arteriogenesis and decreased cell death which were well known to be downstream phenotype changes induced by M2-like-macrophages^(8,19,17). These results suggest that miR93 mediated macrophage-polarization plays a major role in regulating endothelial function to modulate vascular remodeling (including angiogenesis and arteriogenesis) and perfusion-recovery.

Ischemia in the skeletal muscle induces a potent inflammatory response resulting in the infiltration of immune cells including monocytes/macrophages that critically regulate the next stages of tissue recovery^(19,20). While several reports focused on the role of the monocytes that infiltrate the diseased tissue, differentiate to M1 or M2-phenotype to modulate tissue recovery^{(47,48,49,15)}, the status of the resident-macrophages which can be considered as the first immune cell population to respond to injury was overlooked in ischemic skeletal muscle^(19,20). Due to their location in the ischemic skeletal muscle tissue, resident macrophages play critical roles in either aggravating or attenuating the extent of tissue damage in addition to infiltrating monocyte/macrophages. Our data demonstrate that while under normal conditions, miR93 deficiency does not affect the resident-macrophage population, lack of miR93 in ischemic-muscle dramatically decreased anti-inflammatory tissue-resident macrophage population as well as tissue-infiltrating macrophages. These data suggest that miR93 not only modulates M2-like-polarization of tissue-infiltrating but also tissue-resident macrophages.

RNA-Sequencing of WT and miR106b-93-25^−/− ischemic gastrocnemius-muscle identified potential miR93-targets that can regulate macrophage-polarization. While, the top 3 genes induced in miR106b-93-25^−/− ischemic-muscle (compared to WT) CXCL2^(50,51), IRG1^(34,52) and GM3893 were intimately associated with immune responses (with the exception of GM3893, a long non-coding RNA) the top 3 downregulated genes C1QTNF3⁽⁵³⁾, MYH3^(54,55) and ACTC1^(56,57) were closely associated with skeletal muscle function. These data clearly indicated an inverse correlation between genes that regulate immune cell function and skeletal muscle perfusion-recovery in miR106b-93-25^−/− ischemic-muscle. Experiments conducted to understand miR93 regulation of skeletal muscle-macrophage interactions showed that miR93-induction in macrophages can have paracrine effects that decrease skeletal muscle injury/cell death. Furthermore, cytokine bead array showing miR93 decreases TNF-α, IL-1β and IFN-γ secretion (M1-cytokines well known to aggravate tissue injury) from ischemic-macrophages suggests that increased M1-cytokine levels in miR106b-93-25^−/− ischemic-muscle also contributed to aggravation of skeletal muscle injury. Further experiments will be needed to fully dissect the role of miR93 in regulating paracrine effects of macrophages in experimental-PAD.

Sequential analysis of top upregulated immune-related genes from RNA-Sequencing from miR106b-93-25^−/− vs. WT mice showed that IRG1 can induce proinflammatory-macrophage polarization and is also modulated by miR93 in experimental-PAD. IRG1 is highly expressed in macrophages during inflammation⁽⁵⁸⁾ and has been demonstrated to play important roles in regulating macrophage bactericidal activity by inducing the itaconic acid production^(52,59). Macrophages treated with LPS were shown to have significantly higher (~7–8mM) intracellular itaconic acid levels⁽³⁶⁾. Itaconic acid (at 5mM and 10mM, based on published report⁽³⁶⁾) showed a significant induction in iNos-expression but no changes in Arg1-expression indicating that increased itaconic acid production in macrophages can induce M1-like-macrophage polarization. Limited information is available on molecules that can regulate IRG1-expression. In our current report we show that while IRG1 is not a direct target of miR93, IRF9 that has the ability to induce IRG1 is a direct target of miR93 indicating that a complex miR93-IRF9-IRG1-itaconic acid pathway in macrophages regulates macrophage-polarization and perfusion in experimental-PAD.

While our results do not exclude additional effects of macrophage-polarization in modulating angiogenesis and arteriogenesis beyond immune cell function, we conclude that absolute, or relatively, less miR93 in hypoxia-serum-starved macrophages results in increased IRF9-IRG1-expression leading to increased itaconic acid production which promotes proinflammatory M1-like-polarization and contributes to impaired neovascularization (including angiogenesis and arteriogenesis) and decreased perfusion (Fig-8G) in experimental-PAD.

Clinical Perspective.

What is New?

• Within the miR106b-93-25 cluster knock-out, miR93 over-expression alone is sufficient to enhance angiogenesis, arteriogenesis and perfusion in ischemic muscle via increased M2-like-macropaghes.

• miR93 targets Interferon Regulatory Factor-9 to inhibit immune response gene-1 and Itaconic acid generation in macrophages to induce M2-like-macrophage polarization.

• miR93 over-expression modulates a paracrine effect on macrophages to induce angiogenesis and skeletal muscle recovery under hypoxic conditions in vitro.

• We also provide the first evidence that induction of IRF9-IRG1-Itaconic acid pathway in endothelial cells can inhibit angiogenesis in vitro.

What are the Clinical Implications?

• No current medical therapies exist that are capable of promoting both arteriogenesis and angiogenesis in PAD patients.

• miR93 increased endothelial angiogenic potential and in macrophages increased M2-like-phenotype that contributed to skeletal muscle recovery and perfusion in preclinical PAD

• Our data demonstrate that miR93 induces beneficial effects in multiple cells that can enhance perfusion in ischemic limb and identifies miR93 as a putative therapeutic target for clinical PAD