MiR-409-3p targets a MAP4K3-ZEB1-PLGF signaling axis and controls brown adipose tissue angiogenesis and insulin resistance

Abstract

Endothelial cells (ECs) within the microvasculature of brown adipose tissue (BAT) are important in regulating the plasticity of adipocytes in response to increased metabolic demand by modulating the angiogenic response. However, the mechanism of EC-adipocyte crosstalk during this process is not completely understood. We used RNA sequencing to profile microRNAs derived from BAT ECs of obese mice and identified an anti-angiogenic microRNA, miR-409-3p. MiR-409-3p overexpression inhibited EC angiogenic properties; whereas, its inhibition had the opposite effects. Mechanistic studies revealed that miR-409-3p targets ZEB1 and MAP4K3. Knockdown of ZEB1/MAP4K3 phenocopied the angiogenic effects of miR-409-3p. Adipocytes co-cultured with conditioned media from ECs deficient in miR-409-3p showed increased expression of BAT markers, UCP1 and CIDEA. We identified a pro-angiogenic growth factor, placental growth factor (PLGF), released from ECs in response to miR-409-3p inhibition. Deficiency of ZEB1 or MAP4K3 blocked the release of PLGF from ECs and PLGF stimulation of 3T3-L1 adipocytes increased UCP1 expression in a miR-409-3p dependent manner. MiR-409-3p neutralization improved BAT angiogenesis, glucose and insulin tolerance, and energy expenditure in mice with diet-induced obesity. These findings establish miR-409-3p as a critical regulator of EC-BAT crosstalk by modulating a ZEB1-MAP4K3-PLGF signaling axis, providing new insights for therapeutic intervention in obesity.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-021-03960-1.

Introduction

Obesity is primarily caused by an imbalance between energy intake and expenditure, leading to increased body weight and overall adipose tissue enlargement [1]. This increase in obesity parallels type 2 diabetes (T2D) with approximately 1 in 10 adults diagnosed with T2D in the US accompanied by a marked increase in healthcare expenditures [2, 3].

There are two main adipose tissue depots: white adipose tissue (WAT), which primarily stores energy, and brown adipose tissue (BAT), which consumes energy through thermogenesis [4]. The presence of active BAT in human adults opened a new avenue of inquiry for the development of obesity and insulin resistance in adults [5–7]. Both WAT and BAT are highly vascularized. Studies in mice show that visceral adipose tissue can double in size after the first week of high-fat diet (HFD), but will undergo significant reduction following 24 h of fasting [8]. This plasticity is orchestrated by endothelial cells (ECs), fibroblasts, and immune cell subsets, that release various cytokines, growth factors, and adipokines, leading to angiogenesis and remodeling of the extracellular matrix [9].

Impaired angiogenesis in adipose tissue is linked to diet-induced obesity (DIO) and insulin resistance [10]. A corresponding maladaptive consequence is the whitening of BAT, which attenuates the thermogenic response, alters glucose metabolism, and decreases BAT mass in obese subjects [11]. Furthermore, targeted inhibition of neovascularization in advanced adipose tissue halts obesity progression; whereas, enhancing neovascularization in developmentally early-stage adipose tissue leads to healthy fat pads that confer favorable metabolic properties [12]. This is consistent with the absence of small vessel formation during DIO when migration, proliferation, and adhesion of ECs become impaired [13, 14].

There is a delicate balance between pro- and anti-angiogenic factors that regulate angiogenesis. They target ECs to alter migration and proliferation, and are secreted by adipose tissue, suggesting an autoregulatory function [15–17]. Of these factors, vascular endothelial growth factor-A (VEGF-A), a known contributor to vasculogenesis, angiogenesis, and tissue remodeling, facilitates pro-angiogenic activity. BAT-specific expression of VEGF-A in obese mice was associated with increased vascularity, improved BAT function and insulin sensitivity [18]. However, other studies have shown that VEGF-A is increased in obese mice, demonstrating the complexity of angiogenesis in obesity and the necessity for a better understanding of endothelial-adipocyte interactions [19, 20].

Placental growth factor (PLGF) is a member of the VEGF family and binds to fms-like tyrosine kinase-1 (FLT1) [21]. Through the activation of FLT1, PLGF induces signaling pathways different from those induced by VEGF via inducing the phosphorylation of distinct tyrosine residues of FLT1 [22]. Many cell types including ECs produce PLGF under pathological conditions. In the vascular system, PLGF was shown to induce angiogenesis by promoting proliferation and migration of endothelial cells [23]. In adipose tissue, PLGF inhibition reduced the formation of de novo fat pad without affecting adipose tissue development [24] thereby suggesting a role in the early stages of adipogenesis. On the other hand, PLGF deficiency reduced the fraction of brown adipocytes while stimulating white adipocyte hypertrophy in mice fed a high-fat diet [25].

MicroRNAs (miRNAs) are evolutionarily conserved small non-coding RNAs that inhibit gene expression at the post-transcriptional level, thereby serving as physiological “fine-tuners” with therapeutic potential [26, 27]. Distinct miRNA expression signatures have been observed in obese mice and humans when compared to their lean counterparts [28–30]. For example, expression of the anti-inflammatory miR-181b was reduced in adipose tissue ECs and delivery of miR-181b to the microvasculature significantly improved insulin resistance in mice [12]. Knockdown of the Let-7 family of miRNAs improved glucose tolerance in DIO mice [31]. Similarly, miR-143 and miR-145 were upregulated in the liver of obese mice, while mice deficient in the miR-143–145 cluster were protected from obesity-induced insulin resistance [32]. Furthermore, silencing of miR-103 and miR-107 led to improved glucose homeostasis [33]. These miRNAs illustrate their potential to impact, either positively or negatively, the development and progression of insulin resistance. However, the role of miRNAs in the regulation of the microvasculature of BAT remains poorly understood.

In this report, we identified that miR-409-3p plays a key regulatory role in EC-BAT crosstalk by modulating a MAP4K3—ZEB1—PLGF signaling axis, impacting angiogenesis, obesity, and insulin resistance. Our findings reveal that targeting miR-409-3p may represent a new therapeutic approach in DIO by accelerating the browning of adipose tissues and improving energy metabolism.

Methods

Cell culture and transfection

Human umbilical vein endothelial cells (HUVECs) cultured in EGM^®-2 (Lonza) were transfected using Lipofectamine^™ 2000 (Invitrogen). For reporter studies, HUVECs were transfected with 400 ng of the indicated reporter and either 30 nM miR-409-3p mimic or 100 nM miR-409-3p inhibitor (Thermo Scientific) [34, 35]. Gene knockdown experiments were conducted with control (Dharmacon) or target of interest siRNA specified for each experiment.

Isolation of stromal-vascular fraction and ECs from BAT

BAT was excised from the intrascapular region of mice (8 mice/group). Stromal vascular fraction was isolated by digesting the tissue in 1 mg/ml collagenase and dispase solution in DMEM/F12 at 37 °C for 40 min followed by neutralization with DMEM/F12 with 10% FBS. Cell suspension was filtrated through 70 μm strainer (BD Falcon) and centrifuged at 500 g for 10 min at 4 °C. Resulting pellet was suspended in RBC lysis solution (Ebioscience) for 5 min and neutralized with DMEM/F12 medium with 10% FBS. The suspension was then filtered through a 40 μm strainer (BD Falcon) and the pellet was suspended in PBS with 0.1% BSA and 2 mM EDTA. For EC isolation, the stromal-vascular fraction was incubated with sheep anti-rat IgG Dynabeads coated with PECAM-1 antibodies (BD, 557,355). The bead-bound cells were collected using a magnet and washed 3 times [36].

Luciferase reporter assay

HUVECs were transfected with 400 ng of MAP4K3 or ZEB1 3’-UTR (Genecopoeia) per well followed by 30 nM miR-409-3p mimic, 200 nM miR-409-3p inhibitor, or equivalent non-specific controls 24 h later. Luciferase Reporter Assay (Promega) was used and normalized to the total protein in each well (Thermo Scientific).

Differentiation of 3T3-L1 pre-adipocytes into mature brown-like adipocytes

3T3-L1 cells (ATCC) were cultured in maintenance media (MM), induced using induction media (IM) and the time point was marked as Day 1 (D1). On D3, media was switched to a differentiation media (DM) and EC conditioned media co-culture started at a 1:1 ratio. Both the differentiation and the EC conditioned media were replenished every 2 days. On D7-D8, cells were harvested. MM was composed of DMEM with F12 and HEPES supplemented with 10% Fetal Calf Serum (FCS) (GeminiBio), and 1% P/S. IM was prepared with MM supplemented with 10 uM/L human insulin, 0.5 mmol/L IBMX, 0.25 umol/L dexamethasone, and 1 nM T₃ [37]. DM was prepared with MM supplemented with 10 uM/L human insulin, 1 nM T₃, 1 uM rosiglitazone [38, 39].

Gene expression analysis

RNA was harvested in TRIzol^® reagent. Reverse transcription was performed using miScript Reverse Transcription Kit (Qiagen). QuantiTect SYBR Green RT-PCR Kit (Qiagen, 204,243) or miScript SYBR Green PCR Kit (Qiagen) were used for RT-qPCR studies (Agilent Technologies) using gene-specific primers (Supplemental Table 1) normalized to HPRT or GAPDH. Mature miRNA sequences were amplified with Hs_miR-409-3p_1 (Qiagen, MS00006895) and normalized to Hs_RN5S1_11 (Qiagen, MS00007574). Fold changes were calculated by the ΔΔCt method.

Western blotting

Total protein was isolated in RIPA buffer (Boston BioProducts) with 1X Halt^™ protease inhibitor (Thermo Scientific) and quantified by BCA (Thermo Scientific). Lysates were subjected to Western blotting using Abcam antibodies against ZEB1 (ab124512), MAP4K3 (ab173308), UCP1 (ab155117), CIDEA (ab8408), CST antibodies against α-Tubulin (2144), β-actin (4970), goat anti-rabbit (7074) or anti-mouse antibodies (7076). ECL assay was performed (GE Healthcare) and quantification was conducted (Image J).

Tube-like network formation on Matrigel

Matrigel (BD Bioscience) assay was performed as we previously described [40, 41].

Chemotaxis assays

Migration assay was performed using ChemTX multiwell system (Neuro probe). The number of cells migrating to the lower chamber (EC growth media with 50 ng/ml VEGF or bFGF as indicated) was counted after 6–8 h [40].

BrdU assay

BrdU ELISA assay (Roche, 116,472,290,001) was performed, as we previously described [41].

MiRNP immunoprecipitation (MiRNP-IP)

MiRNP-IP was performed as we previously described [42].

Scratch assay

Scratch assay was performed as we previously described [41] and cells were imaged using Eclipse TE2000-U inverted microscope (Nikon).

Human adipose organoids

Full-thickness circular (3 mm) organoids were taken from sWAT post abdominal surgery and cultured in maintenance media (MM) referred to as day 0 (D0). On D1, media was changed to induction media (IM). On D3 and D5, the media was replaced with a 1:1 ratio of differentiation media (DM including rosiglitazone to induce browning [43, 44]: EC conditioned media (transfected 72 h prior with indicated NS controls or miR-409 mimic or inhibitors). Organoids were harvested on D7 using aforementioned methods. MM was composed of Dulbecco’s Modified Eagle’s Medium (DMEM) with 4.5 g/L glucose and with L-glutamine (Lonza) supplemented with 10% FBS and 1% P/S. IM was composed of MM with 0.5 mmol/L IBMX, 0.25 umol/L dexamethasone, 1 nM T₃ and 10 uM/L human insulin. DM was composed of MM with 10 uM/L human insulin, 1 nM T₃, and 1 uM rosiglitazone. The viability of cultured explants was validated by histologic evaluation performed on Days 0, 3, 5 and 7 for adipose density and structure by H&E and perilipin 1 (PLIN1) stainings (Data not shown). Angiogenesis was analyzed by CD31 staining (28,364; Abcam), DAPI (H21492; Invitrogen) for nuclear staining and Alexa 647 conjugated donkey anti-rabbit antibody for secondary antibody (711–605-152; Jackson ImmunoResearch). Relative CD31 expression was measured in the entire cross-section and quantified using Image J.

Metabolic studies

Mice fasted for 12 h received intra-peritoneal (i.p.) injection of D-glucose (1U/kg, Sigma, G8270) for glucose tolerance test (GTT). For the insulin tolerance test (ITT), mice were fasted for 6 h and i.p. injected with 0.75U/kg insulin. Blood glucose measurements were taken at indicated time points (OneTouch, LifeScan). Whole-body energy expenditure was measured at ambient temperature (~ 22 °C) using Columbus Comprehensive Lab Animal Monitoring System.

Histological and immunohistological examinations

Tissues were fixed with 10% formalin solution, embedded in paraffin wax, cut into 10um sections, and then deparaffinized. Images were acquired on an upright Carl Zeiss LSM 510 confocal microscope. Data were analyzed in a blinded fashion using the Volocity Software (Quorum Technologies). For Oil-red O staining, cells were fixed with 10% formaldehyde for 15 min at room temperature and stained using the Oil-red O working solutions (5 g/l in isopropanol) and 4 ml ddH₂O for 30 min. After staining, the cells were washed with 60% isopropanol and images were acquired on an Eclipse TE2000-U inverted microscope (Nikon).

Animal care and use

Animal studies were approved by the Institutional Animal Care and Use Committee. For DIO, male, 4 weeks old, C57BL/6 mice (Charles River) were placed on high-fat diet (HFD) containing 60 kcal% fat (Research Diets), received weekly tail-vein injections of scrambled control LNA-anti-miR or LNA-anti-miR-409 (10 mg/kg, Exiqon) for 11 weeks.

Statistical analysis

Data are presented as mean ± SEM. Sample sizes for mouse and human organoid experiments were chosen based upon pilot or similar well-characterized studies in the literature. There were no inclusion or exclusion criteria used. Data were subjected to unpaired two-sided Student’s t test or one-way ANOVA with Bonferroni correction for multiple comparisons, and p < 0.05 was considered statistically significant.

Data and resource availability

The data generated from this study and the associated resources are available from the corresponding authors upon reasonable request.

Results

To identify miRNAs in the microvasculature of brown fat that might impact metabolic dysfunction in obesity, ECs were isolated from C57BL/6 mice (Supplemental Fig. 1A) after 8 weeks of chow or HFD and subjected to miRNA-seq profiling (fold-change > 2; FDR p < 0.05). MiR-409-3p was identified as a top significantly regulated miRNA that increased in ECs of mice fed with HFD (Fig. 1B and Supplemental Fig. 1B). Expression of miR-409-3p was highest in multiple EC cell types, including HUVECs and human umbilical artery endothelial cells (HUAECs) compared to non-EC cell lines, such as human pre-adipocytes, human adipocytes, mouse spleen and bone marrow monocytes, and human fibroblasts (Fig. 1C). Furthermore, separation of ECs from non-ECs from the brown fat of mice fed HFD for 8 weeks demonstrated a twofold increased expression of miR-409 in the EC fraction (Supplemental Fig. 1C). Overall, this data demonstrates that miR-409-3p is highly expressed in mouse brown fat ECs under diet-induced obesity conditions.

Fig. 1.

MicroRNA-409-3p (MiR-409-3p) discovery in diet-induced obesity in mice and its regulation of EC growth in vitro. A Workflow of genome-wide RNA-seq profiling for the identification of miR-409-3p using ECs from BAT. B MiR-409-3p expression in BAT of mice placed on chow or HF diet for 8 weeks. C Expression of miR-409-3p in HUAECs, pre-adipocytes, adipocytes, splenic (Sp) derived monocytes, bone marrow (BM) derived monocytes, and fibroblasts compared to human umbilical vein endothelial cells (HUVEC). D MiR-409-3p overexpressed or (E) knocked down in HUVECs and subjected to BrdU (5-bromo-2’-deoxyuridine) cell proliferation assay. Data representative of n = 6 per condition unless indicated otherwise. All statistics calculated by Student’s t test except for two-way ANOVA in 1c. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

MiR-409-3p is a highly conserved microRNA (Supplemental Table 2). To examine the functional role of miR-409-3p, in vitro gain- and loss-of-function studies were performed. Compared to non-specific controls (NS_m), overexpression of miR-409-3p in HUVECs using miR-409-3p mimics (miR-409_m) demonstrated a 35% reduction in proliferation as assessed by BrdU assay (Fig. 1D) and a 50% diminished wound closure rate using EC scratch assays (Fig. 2C). HUVECs transfected with miR-409_m showed a 50% reduction in cumulative sprout length (Fig. 2E) and 29% reduction in the number of tubes as quantified by network tube formation in Matrigel (Fig. 2G). In accordance with these findings, miR-409-3p reduced the number of cells per well by 37% when assessed by transwell migration (Fig. 2A). In contrast, when miR-409-3p was blocked using miR-409-3p inhibitors (miR-409_i), it triggered a 1.5-fold increase in proliferation as assessed by BrdU assay (Fig. 1E) and a threefold increase in wound closure rate by scratch assay compared to non-specific controls (NS_i) (Fig. 2D). In line with these findings, miR-409-3p inhibition also increased EC sprout length by 1.4-fold (Fig. 2F), network tube formation by 2.75-fold in Matrigel (Fig. 2H), and the number of migrated cells by 1.5-fold, as assessed by transwell migration (Fig. 2B). These findings demonstrate that miR-409-3p inhibits angiogenic responses; whereas, its inhibition promotes angiogenic properties in vitro.

Fig. 2.

MiR-409-3p regulates EC migration. HUVECs transfected with non-specific control (NS_m) or miR-409-3p mimic (miR-409-3pm) (A, C, E, G) or miR-negative inhibitor control (NS_i) and miR-409-3p inhibitor (miR-409-3p_i) (B, D, F, H) were subjected to (A, B) EC migration in transwell Boyden chambers; C, D scratch assay (Scale bars, 100 μm); E, F spheroid formation assay (Scale bars, 100 μm); G, H matrigel tube formation assay (Scale bars, 40 μm). Data representative of n = 6–8 experiments. All statistics calculated by unpaired Student’s t test or one-way ANOVA, based on a comparison with respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

To explore whether the miR-409-3p-mediated angiogenic responses could potentially impact brown fat, we utilized an in vitro model in which 3T3-L1 fibroblasts were differentiated to a brown-like phenotype (Supplemental Fig. 2A) and co-cultured with supernatant from HUVECs transfected with either NS_m and miR-409_m (overexpressing miR-409-3p) or NS_i and miR-409_i (inhibiting miR-409-3p) (Fig. 3A). Co-culture with supernatant from HUVECs transfected with miR-409-3 pm compared with non-specific control (NS_m) revealed reductions in expression for uncoupling protein 1 (UCP1) by 78%, cell-death inducing DNA fragmentation factor-like effector A (CIDEA) by 50% (Fig. 3B), peroxisome proliferator-activated receptor gamma co-activator 1-alpha (PGC1α) by 25% (Supplemental Fig. 2B), and a 2.5-fold increase in ADIPOQ expression (Supplemental Fig. 2C), as assessed by RT-qPCR. Additionally, co-culture from supernatants from miR-409-3p overexpression in HUVECs reduced UCP1 and CIDEA protein expression by 25% and 37%, respectively (Fig. 3D). This decreased “browning” phenotype was further supported by a 75% reduction in Oil Red O (ORO) staining for lipid droplets (Fig. 3F). In contrast, co-culture of 3T3-L1 cells with supernatants from HUVECs transfected with miR-409_i increased mRNA expression of browning markers: UCP1 (~ 2.16-fold; Fig. 3C), CIDEA (~ 1.46-fold; Fig. 3C), PGC1α (1.54-fold; Supplemental Fig. 2D), and decreased Adiponectin (ADIPOQ) expression (21%; Supplemental Fig. 2E). Furthermore, protein expression increased for UCP1 by 1.74-fold and for CIDEA by 1.56-fold (Fig. 3E). The increased “browning” phenotype exhibited by these differentiated adipocytes was further corroborated by a 1.4-fold increase in lipid droplets by ORO staining (Fig. 3G). Interestingly, when 3T3-L1 cells were transfected directly with miR-409-3pm_, we did not observe a significant reduction in UCP1 expression (Supplemental2F). Overall, these findings indicate that miR-409-3p may play a significant role in the differentiation state of adipocytes towards white or brown fat.

Fig. 3.

MiR-409-3p differentiates 3T3-L1 fibroblasts into brown-like adipocytes. Supernatant from HUVECs 3 days post-transfection with NS_m/miR-409-3 pm or NS_i/miR-409-3p_i were cultured with 3T3-L1 fibroblasts for 4 days, followed by a RT-qPCR analysis of brown fat gene expression (UCP1 and CIDEA) with either (B) mimic or (C) inhibitor, with results normalized to α-tubulin. Western blot analyses for UCP1, CIDEA, and α-tubulin protein expression (D) and (E), as well as Oil-red O staining of 3T3-L1 cells (F) and (G). Data representative of n = 4 experiments. All statistics calculated by unpaired Student’s t test or one-way ANOVA, based on a comparison with respective non-specific control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

To ascertain if altering miR-409-3p expression could similarly regulate browning of human adipocytes, we used a complementary ex vivo approach, in which human punch biopsies were obtained from discarded subcutaneous adipose tissue after surgical removal and placed in culture for 7 days as organoids (Supplemental Fig. 3A). During this differentiation process, adipocyte samples were co-cultured with supernatants of ECs transfected with either NS_m and miR-409-3pm or NS_i and miR-409-3p_i. When co-cultured with miR-409-3pm supernatants, adipocytes demonstrated a reduction in brown fat mRNA markers, UCP1 (53%), CIDEA (71%), and PGC1-α (49%) (Supplemental Fig. 3B), and reduced expression of UCP1 (54%) and CD31 (47%) by immunofluorescence staining (Supplemental Fig. 3D). In contrast, organoids co-cultured with miR-409_i supernatants increased browning mRNA markers for UCP1 (1.93-fold), CIDEA (2.85-fold), and PGC1-α (3.1-fold) (Supplemental Fig. 3C). This phenotype was further supported by increased expression of UCP1 (3.9-fold) and CD31 (1.9-fold) using immunofluorescence staining (Supplemental Fig. 3E). Taken together, this data shows that miR-409-3p regulates adipocyte browning in both mouse and human cells.

We utilized an in silico approach to identify the target genes of miR-409-3p through the use of prediction algorithms (miRWalk, MicroT4, RNAhybrid, and TargetScan). Expression of the predicted candidate genes targeting the 3’-UTR of miR-409-3p was validated on the mRNA and protein levels. From 127 genes that were predicted by 5 prediction algorithms, the mRNA of 12 genes were decreased in HUVECs overexpressing miR-409-3p, and only 2 genes, ZEB1 and MAP4K3, showed consistently reduced gene expression by Western Blot and enrichment in the Myc-AGO2 complex (Fig. 4A and data not shown). Overexpression of miR-409-3p significantly decreased the mRNA expression of ZEB1 (~ 54%) and MAP4K3 (~ 66%) (Fig. 4B) in HUVECs and in human sWAT organoids (Supplemental Fig. 4A). We also observed a significant decrease in protein expression of ZEB1 (~ 42%) and MAP4K3 (~ 48%) (Fig. 4C) in HUVECs. Conversely, inhibition of miR-409-3p increased the protein expression of ZEB1 (~ 1.34-fold) and MAP4K3 (~ 1.38-fold) (Fig. 4D). MiR-409-3p overexpression also inhibited ZEB1 by 43% and MAP4K3 by 45% in 3’-UTR reporter assays (Fig. 4E). In contrast, miR-409-3p inhibition led to a 1.28-fold increase in ZEB1 and a 1.22-fold increase in MAP4K3 3’-UTR reporter activities (Fig. 4F). Additionally, we observed a significant increase in mRNA expression of ZEB1 and MAP4K3 in human sWAT organoids in response to miR-409-3p inhibition (Supplemental Fig. 4B). Binding sites to ZEB1 and MAP4K3 3’UTR for miR-409-3p were mapped by Rna22 (Supplemental Fig. 5A, B). To further verify that miR-409-3p directly targets ZEB1 and MAP4K3 in ECs, we performed Argonaute2 (AGO2) microribonucleoprotein IP (miRNP-IP) studies to assess whether ZEB1 and MAP4K3 mRNA is enriched in the RNA-induced silencing complex following miR-409-3p overexpression in HUVECs. We observed 1.37-fold enrichment of ZEB1 and 1.56-fold enrichment of MAP4K3 mRNA following AGO2 miRNP-IP as compared to the miRNA-negative control (Fig. 4G). Finally, AGO2 miRNP-IP did not enrich SMAD1, a gene that was not found to be a target of miR-409-3p (Supplemental Fig. 5C).

Fig. 4.

ZEB1 and MAP4K3 are potential targets of miR-409-3p. A Workflow used for target identification. HUVECs were either transfected with (B, C, E, G) NS_m and miR-409-3pm or D and F NS_i and miR-409-3p_i, followed by (B) RT-qPCR analysis of ZEB1 and MAP4K3 mRNA expression. Total of n = 3 experiments. C and D Western blot analyses for ZEB1, MAP4K3, and β-actin protein expression. Protein blots representative of n = 3–4 experiments. E and F Luciferase activity (RLU) of ZEB1 and MAP4K3 3’-untranslated regions (UTRs) normalized to total protein. Data representative of n = 3 experiments. G RNA co-precipitation (co-IP) conducted using ZEB1 and MAP4K3, normalized to Myc and HPRT. n = 3 per condition. All statistics calculated by unpaired Student’s t test or one-way ANOVA, based on a comparison with respective non-specific control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

To evaluate whether neutralization of ZEB1 and MAP4K3 can “phenocopy” miR-409-3p functional effects in ECs, we undertook siRNA-mediated knockdown approaches (Supplemental Fig. 5D, E). Silencing of either ZEB1 or MAP4K3 significantly reduced EC proliferation by ~ 47% and ~ 35%, respectively, as measured by BrdU incorporation (Fig. 5A and D), transwell migration of ZEB1 by ~ 50% and MAP4K3 by ~ 25% (Fig. 5B and E) and wound closure by ~ 43% and ~ 53%, respectively, in scratch assays (Fig. 5C and F). Interestingly, concurrent silencing of ZEB1 and MAP4K3 showed cooperative effects on EC proliferation (27% compared to ZEB1_siRNA or MAP4K3siRNA alone) (Fig. 5G) and migration (85% compared to ZEB1_siRNA or MAP4K3siRNA alone) (Fig. 5H). In the absence of ZEB1 or MAP4K3, miR-409-3p overexpression impaired EC proliferation demonstrating that miR-409-3p-mediated effects are partially dependent on ZEB1 and MAP4K3 (Fig. 5I). Altogether, this data demonstrates ZEB1 and MAP4K3 are bona fide targets of miR-409-3p in ECs, where increased levels of miR-409-3p in ECs may potentially act as a molecular switch inhibiting EC growth and angiogenesis.

Fig. 5.

SiRNA-mediated knockdown of zinc finger E-box-binding homeobox 1 (ZEB1) and mitogen-activated protein kinase kinase kinase kinase 3 (MAP4K3) recapitulates miR-409-3p functional effects in endothelial cells (ECs). Human umbilical vein ECs (HUVECs) were transfected with siRNA to (A–C, G, H, I) ZEB1, D–I MAP4K3, or scrambled control (ctrl). EC proliferation was then determined by (A, D, G, I) BrdU assay. Migration of ECs was quantified by either (C, F) scratch assay or (B, E, H) transwell Boyden chambers. G and H SiRNA to ZEB1 and MAP4K3 was included to assess for combinatorial effects. I NS_m and miR-409-3pm were transfected to assess dependency. Scale bars, 100 μm. Data representative of n = 4 experiments. All statistics calculated by unpaired Student’s t test or one-way ANOVA, based on a comparison with indicated control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

To assess whether miR-409-3p impacts adipocyte browning in a paracrine manner, we used multiplex ELISA from the conditioned media of ECs overexpressing or deficient in miR-409-3p and identified that the secretion of PLGF was significantly regulated by miR-409-3p. Overexpression of miR-409-3p decreased PLGF secretion by ~ 46% (Fig. 6A); whereas, its neutralization increased PLGF secretion by sevenfold (Fig. 6B). In contrast, EC secretion of EGF, BMP-9, or VEGF-A were not regulated under these conditions (Fig. 6A, B). Interestingly, we found that PLGF stimulation of 3T3-L1 cells during differentiation significantly increased UCP1 expression (Fig. 6C). Moreover, siRNA-mediated knockdown of FLT1 in 3T3-L1 cells (Supplemental Fig. 6A) significantly reduced PLGF induced UCP1 expression (Supplemental Fig. 6B, C). Consistent with these findings, PLGF stimulation increased glucose uptake of differentiated 3T3-L1 cells (Supplemental Fig. 6D). Additionally, knockdown of miR-409-3p coupled with PLGF stimulation demonstrated greater UCP1 expression compared to PLGF stimulation alone (Fig. 6D). SiRNA-mediated knockdown of miR-409-3p target genes, ZEB1 or MAP4K3, phenocopied miR-409-3p effects on PLGF secretion and reduced EC secretion of PLGF (Fig. 6E). Taken together, this data suggests that the miR-409-3p-mediated regulation of ZEB1 and MAP4K3 regulates PLGF release from ECs and browning of 3T3-L1 adipocytes through FLT1.

Fig. 6.

MiR-409-3p regulates the secretion of PLGF and 3T3-L1 browning. A and B Conditioned media from HUVECs were transfected with (A) NS_m/miR-409-3 pm or (B) NS_i/miR-409-3p_i and subjected to multiplex ELISA. A and B Secreted PLGF, EGF, BMP9, and VEGF-A were quantified using the multiplex ELISA. C Stimulation of 3T3-L1 cells with 100 or 150 ng/mL PLGF or vehicle control increased UCP1 expression as measured by RT-qPCR. D Conditioned media from HUVECs transfected with NS_i or miR-409-3p_i were added to 3T3-L1 cells in the presence or absence of PLGF (200 ng/mL) and UCP1 expression was quantified by RT-qPCR. E Conditioned media from HUVECs transfected with scrambled control siRNA (Cntrl_siRNA), ZEB1_siRNA, or MAP4K3siRNA was subjected to ELISA to measure secreted PLGF release. All RT-qPCR results normalized to 36B4 as the housekeeping gene. All statistics calculated by (A–C, E) unpaired Student’s t test or (D) one-way ANOVA, based on a comparison with indicated control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

To evaluate whether neutralization of miR-409-3p regulates angiogenesis and insulin resistance in obese mice, LNA-miR-409-3p_inh was delivered intravenously (i.v.) weekly in mice fed a HFD (Fig. 7A). LNA-miR-409-3p_inh decreased the expression of its targets ZEB1 and MAP4K3 significantly in ECs of BAT, sWAT, and eWAT ECs (Supplementary Fig. 7A, B) while decreasing miR-409-3p expression ~ 22% in the EC of BAT following only two i.v. injections compared to the LNA-NS_inh control group (Supplementary Fig. 7C). Although body weights and composition did not significantly change between LNA-NS_inh and the LNA-miR-409-3p_inh groups over 15 weeks of HFD (Fig. 7B and Supplemental Table 3), insulin (ITT) (Fig. 7C) and glucose (GTT) (Fig. 7D) tolerance and energy expenditure (Fig. 7E and Supplementary Fig. 7D) were significantly improved in response to miR-409-3p neutralization. There were no significant differences in locomotor activity, food intake (Supplemental Fig. 7E, F) or the whole body fat distribution (Supplemental Table 3), while the respiratory exchange rate (RER) was significantly decreased (Supplemental Fig. 7G). Remarkably, inhibition of miR-409-3p induced angiogenesis as measured by CD31 ~ 2.4-fold (Fig. 7F) and increased UCP1, while decreasing adipocyte size (Fig. 7G). Collectively, these findings indicate that miR-409-3p is not only an anti-angiogenic miRNA, but may also contribute to the activation of a metabolic program in mice with obesity.

Fig. 7.

MiR-409-3p neutralization stimulates BAT angiogenesis and improves insulin resistance and energy expenditure in obese mice. A and B C57BL/6 J mice fed high-fat diet (HFD) were intravenously (i.v.) injected with vehicle non-specific control (LNA-NS_i) or LNA-miR-409-3p inhibitor (LNA-409-3p_i) at 10 mg/kg for 15 weeks. C At 13 weeks, mice were subjected to insulin tolerance test (ITT) via i.v. injection with 0.75 U/kg insulin. D At 14 weeks, glucose tolerance test (GTT) was performed using 1 U/kg glucose. E–G Metabolic caging studies over a two-day period (2 light and 2 dark cycles) were performed after 15 weeks and E energy expenditure (EE) quantified. Immunofluorescence staining for (F) CD31/DAPI, G UCP1/perilipin (PLIN1)/DAPI was quantified in brown adipose tissue from the indicated mice. Total of 10 animals per group were used. All statistics calculated by unpaired Student’s t test based on a comparison with respective control group. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate ± SEM

Discussion

The function of BAT is dependent on its proper vascularization that is tightly regulated by a range of cytokines and angiogenic factors. There are many pathological changes that occur in obesity, including EC dysfunction and impaired angiogenesis in adipose depots. Long-term HFD triggers ‘whitening’ of BAT that is characterized by reduced expression of mitochondria-associated genes, such as UCP1, increased lipid deposition, enlargement of lipid droplets, elevated mitochondrial ROS production, and membrane depolarization [18, 45]. In this study, we delineate the molecular mechanisms by which miR-409-3p deficiency regulates EC-driven angiogenesis and, in turn, improves BAT browning, insulin resistance, and metabolic parameters in obesity.

MicroRNAs contribute to BAT functionality. In response to prolonged cold exposure, expression of miR-193a/b [46], miR-365 [46], miR-30b/c [47] and miR-455 [48] were upregulated both in human and mouse BAT. Overexpression of these miRNAs in mice promoted brown adipocyte cell differentiation and browning of sWAT. Conversely, the miR-27 family negatively regulated BAT both in mice [49] and human subjects [50]. Here, we demonstrate that neutralization of miR-409-3p improves angiogenic responses and metabolic functionality in BAT through EC-adipocyte crosstalk. Interestingly, the metabolic benefits we observed in this study, including improved glucose and insulin tolerance and increased energy expenditure, did not correlate to a lower body weight in obese mice with miR-409-3p neutralization. It is likely that systemic neutralization of miR-409-3p and associated stimulation of angiogenesis may have contributed to the expansion of the WAT in addition to restoring BAT function and thereby inhibiting weight loss in response to miR-409-3p neutralization. Body composition analyses showed a non-significant increasing trend in total weight and visceral and subcutaneous adipose weights (Supplemental Table 3). Whether inhibition of miR-409-3p promotes angiogenesis in other peripheral tissues that could potentially contribute to the improved insulin sensitivity observed in these mice remains a possibility and will be of interest to explore in future studies.

In the overall response to metabolic changes, cell-to-cell communication plays a fundamental role [51]. Through in vitro co-culture techniques [52, 53], healthy adipocytes exhibit decreased insulin sensitivity and elevated inflammatory markers when exposed to ECs from obese patients; thereby, revealing the importance of cytokines, such as IL-6 and IL-1β, in the crosstalk between adipocytes and visceral adipose ECs [54]. Emerging studies demonstrate that decreasing inflammation in the microvasculature may favorably delay the development of insulin resistance [36]. Our findings build upon these paracrine mechanisms, demonstrating how miR-409-3p deficiency stimulated EC secretion of the pro-angiogenic factor, PLGF (Fig. 8). It has been reported that PLGF can activate BAT. PLGF treatment of mature adipocytes showed increased UCP1 and OXPHOS levels, mitochondrial function, and higher oxygen consumption rates in vitro through AMPKα phosphorylation, while systemic PLGF administration improved insulin sensitivity in diabetic mice [55]. Here, we report that PLGF released from ECs regulate adipocyte browning through FLT1 that is expressed in 3T3-L1 cells in vitro (Supplemental Fig. 6). Additional studies are needed to further understand the downstream mediators of the PLGF/FLT1 signaling axis that may be involved in adipose browning.

Fig. 8.

Proposed mechanism of miR-409-3p regulation in obesity. (Top panel) Chow fed mice or in response to miR-409-3p deficiency, ZEB1 and MAP4K3 expression are increased; thereby, promoting the release of PLGF, binding to FLT1, browning of adipose tissue (increased brown fat markers UCP1 and CIDEA) and improving glucose tolerance and insulin resistance. In contrast (lower panel), under high-fat diet conditions, miR-409-3p expression is increased in endothelial cells and impairs angiogenesis. MiR-409-3p targets ZEB1 and MAP4K3, which in turn decreases the release of PLGF, inhibits browning of white adipose tissue (decreased brown fat markers UCP1 and CIDEA) through a PLGF/FLT1 signaling axis, and increases glucose intolerance and insulin resistance. Created with BioRender.com

In addition to the pro-angiogenic PLGF, we identified other cytokines to be significantly modulated by miR-409-3p (Supplemental Table 4). Overexpression of miR-409-3p in ECs significantly increased CXCL5 and CXCL9 release while EC secretion of IL4 and IL22 were significantly decreased. Although the role of CXCL5 and CXCL9 in EC- adipose cross talk or adipose browning has not been studied yet, several lines of evidence implicate them in obesity and insulin resistance [56, 57]. Increased expression of IL-4 as a result of a Fas mutation was shown to reduce adipose tissue inflammation, positively altering the lipid metabolism and promoting thermogenic activity and browning in adipose tissue [58]. Similarly, IL-22, a member of the IL-10 family, was shown to reduce the epididymal fat-pad mass, decrease glucose intolerance and insulin resistance and improve heat production in vivo [59]. Additional studies are needed to understand whether these cytokines may play a role in adipose tissue browning and insulin resistance in the setting of EC-adipocyte cross talk.

We have primarily used venous endothelial cells in our study. While microvascular endothelial cells could reflect a more relevant model system than the HUVECs used in our co-culture studies, microvascular ECs from BAT are difficult to culture, rendering these studies not feasible. Alternative microvascular ECs from other tissues could be used. However, there is considerable heterogeneity in the transcriptomic, epigenomic, and functional landscape across tissue vascular beds as well [60]. In addition, although in vitro aspects of our study mostly focused on the modulation of browning during pre-adipocyte differentiation our ex vivo studies with human sWAT organoids and in vivo studies in obese mice gave insights into the ability of miR-409-3p to regulate browning in mature adipocytes.

MAP4K3 has a well-established role in cell growth, proliferation, and migration in various cell lines. Overexpression of MAP4K3 in HeLa cells induces the activation of mTOR downstream signaling molecules, S6K and 4E-BP1, in response to changes in nutrient and energy levels [61] and induces cell proliferation through the activation of the NF-κB pathway in primary human hepatocytes [62]. Whole-body MAP4K3 transgenic mice exhibited higher migration of primary lung epithelial cells that promoted metastasis through IQGAP1 phosphorylation [63]. Our findings demonstrating how the knockdown of MAP4K3 has an inhibitory effect on EC growth and migration are in agreement with the literature findings (Fig. 5). We also demonstrated a novel role for MAP4K3, where its modulation in ECs regulates PLGF secretion, contributing to adipose browning (Fig. 6).

ZEB1 knockdown also significantly inhibited EC growth, migration, and endothelial PLGF release (Figs. 5 and 6). ZEB1 is central transcriptional component of fat cell differentiation with higher expression levels in pre-adipocytes compared to the mature adipocytes [64]. It controls adipogenesis in committed 3T3-L1 cells and in mesenchymal stem cells (MSCs) through cooperating with C/EBPβ [64]. Although our study focused on the regulation of ZEB1 through miR-409-3p in ECs in vitro and in vivo, we cannot completely rule out the systemic effect of miR-409-3p neutralization and ZEB1 regulation in our in vivo studies.

MiR-409-3p was shown to regulate metastasis in human breast cancer cells through targeting ZEB1 [65]. Contradictory to these findings, in osteosarcoma cells, miR-409-3p acts as a tumor suppressor through negatively regulating ZEB1 [66]. Our study brings in a novel component where miR-409-3p-MAP4K3-ZEB1 signaling axis regulates PLGF release from ECs; therefore, linking PLGF to downstream regulation of adipose browning (Fig. 8). Interestingly, ZEB1 expression activates VEGFA transcription by recruiting SP1 to its promoter and activating PI3K and p38 signaling pathways. Furthermore, ZEB1 expressing breast cancer cells increase VEGFA production and stimulate tumor growth and angiogenesis via a paracrine mechanism [67]. Mechanistically, whether miR-409-3p-MAP4K3-ZEB1 modulation of PLGF release from ECs and its effects on adipocytes activate similar signaling pathways remains to be determined.

In summary, we identified that miR-409-3p expression is increased in BAT ECs of obese mice. MiR-409-3p serves as a negative regulator of EC angiogenesis and adipose browning through targeting ZEB1 and MAP4K3; thereby, in turn, regulating the EC secretion of PLGF. Additionally, inhibition of miR-409-3p markedly increased BAT angiogenesis, improved insulin resistance, and stimulated adipose browning in a mouse model of DIO. Due to systemic delivery of miR-409-3p inhibitor in our studies, it is possible that in addition to ECs, there can be contributions from non-ECs present in BAT and sWAT such as adipocytes. Although we showed that miR-409-3p neutralization results in higher expression levels of its targets ZEB1 and MAP4K3 in major adipose depots in ECs compared to adipocytes, targeted delivery of miR-409-3p inhibitor to ECs can help understand contributions from non-ECs and the overall phenotype in a more detailed manner. Therapies focusing on neutralization of miR-409-3p, or its downstream signaling pathways, may offer a new strategy to promote BAT neovascularization and improve metabolic phenotypes in obesity.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

SiRNA

Small interfering RNA

3’-UTR

3’-Untranslated region

ZEB1

Zinc Finger E-Box Binding Homeobox

MAP4K3

Mitogen-activated protein kinase kinase kinase kinase 3

UCP1

Uncoupling Protein 1

Author contributions

MWF and BI designed research; BI, DBG, HL, DPC, WW, FB, CEN and DOz, carried out the experiments; DO, contributed critical reagents; DBG, HL, DPC, WW, FB, MWF and BI analyzed and interpreted the data; and DBG, MWF and BI wrote the manuscript.

Funding

This work was supported by the American Diabetes Association grant #1-16-JDF-046 (to B.I.), a Behrakis Junior Faculty Development Award (to B.I.), a Watkins Discovery Award (to B.I.), a National Institutes of Health (HL149999 to B.I.), a National Institutes of Health (HL115141, HL134849, HL148207, HL148355, HL153356 to M.W.F), an American Heart Association grant (18SFRN33900144 and 20SFRN35200163 to M.W.F.), the Arthur K. Watson Charitable Trust (to M.W.F).

Availability of data and material

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials.

Code availability

Not applicable.

Declarations

Conflict of interest

The authors have declared that no conflict of interest exists with this work.

Ethics approval

All animal protocols were approved through the Brigham & Women’s Hospital Institutional Animal Care and Use Committee. All human cell lines were obtained from indicated life science company or post-operative discarded tissue.

Consent to participate

Not applicable.

Consent for publication

Not applicable.