In Lyl1-/- mice, adipose stem cell vascular niche impairment leads to premature development of fat tissues.

Lyl1 encodes a hematopoietic- and endothelial-specific bHLH transcription factor. Lyl1-deficient mice are viable, but they display mild hematopoietic and vascular defects. Specifically, LYL1 is required for the maturation and stabilization of blood vessel endothelial adherens junctions. Here, we report that young adult Lyl1-/- mice exhibit transient overweight associated with general expansion of adipose tissue, without signs of metabolic disorder and unrelated to food intake. The increased fat tissue development in Lyl1-/- mice resulted from earlier differentiation of adipose stem cells (ASCs) into adipocytes through noncell autonomous mechanisms. Specifically, we found that in Lyl1-/- mice, the adipose tissue vascular structures are immature, as indicated by their high permeability, reduced coverage by pericytes, lower recruitment of VE-cadherin and ZO1 at cell junctions, and more prone to angiogenesis. Together, our data show that in Lyl1-/- mice, the impaired vascular compartment of the adipose niche promotes ASC differentiation, leading to early adipocyte expansion and premature ASC depletion. Our study highlights the major structural role of the adipose tissue vascular niche in coordinating stem cell self-renewal and differentiation into adipocytes.

The transcription factor lymphoblastic leukemia‐derived sequence 1 (LYL1) is required for the maturation and stabilization of blood vessels. Depending on their permeability properties, blood vessels can support either stem cell quiescence or activation. This study showed that in Lyl1‐deficient mice, the vascular structures of adipose tissues are permeable, poorly covered by pericytes and proangiogenic, thus promoting adipose stem cell differentiation. This study highlights for the first time the role of LYL1 in the vascular niche maintenance in adipose tissue.

INTRODUCTION

Endothelial cells (ECs) are the major component of the vascular network that spreads into every organ of the body. Besides the conventional role of blood vessels in the transport of gases, nutrients, waste products and cells, ECs have been functionally linked to a wide range of physiological and pathological processes, including barrier formation, selective transport, scavenging, thrombosis, wound healing, and inflammation. In the last decade, it has been shown that ECs are crucial regulators of tissue morphogenesis through secretion of growth factors and presentation of molecular signals that act on neighboring cell populations in an angiocrine fashion. , , Moreover, blood vessels provide protective and niche‐supporting microenvironments for different adult stem and progenitor cell types, such as neural, , muscle and hematopoietic stem cells (HSCs) , , , as well as hepatic , and adipocyte progenitors. New vessel formation includes different sequential steps: basement membrane remodeling, perivascular cell detachment, EC proliferation and alignment, lumen formation, vessel maturation with adherens junction closure, and finally pericyte coverage that stabilizes the vascular structures. Most of the mechanisms involved in the latest steps of vessel maturation contribute to the maintenance of mature vessel integrity.

The transcription factor lymphoblastic leukemia‐derived sequence 1 (LYL1) is a member of the basic helix loop helix (bHLH) family. Its expression is restricted to hematopoietic cell and EC lineages during embryonic life and in adulthood. Lyl1‐deficient (Lyl1 ) mice develop normally without gross histological abnormalities. However, others and we highlighted the presence of hematopoietic and vascular system defects. Specifically, disruption of LYL1 activity in mice impairs the long‐term hematopoietic reconstitution capacity and maintenance of early T lineage progenitors, and partially blocks B lymphocyte differentiation. , More recently, Chiu et al demonstrated that Lyl1 can maintain primitive erythropoiesis and compensate for the loss of Scl (another bHLH family member) in megakaryopoiesis. , LYL1 is also required for the maturation and stabilization of endothelial adherens junctions in newly formed vessels. , Consequently, blood vessels in the lung of young adult Lyl1 mice cannot form a functional endothelial barrier, leading to vascular leakiness. In Lyl1 mice grafted with cancer cells, tumor vessels are highly angiogenic, fully immature, and poorly covered by mural cells.

During these previous studies, we noticed that young adult Lyl1 −/− mice were often overweight. While obesity classically results from white adipose tissue (WAT) expansion, Lyl1 absence affected all adipose depots: brown adipose tissue (BAT) and also subcutaneous and visceral WAT. Here, we show that the earlier development of these three adipose tissues in Lyl1 mice results from noncell autonomous mechanisms. Specifically, we found that the premature development of vascular structures in Lyl1 early adipose organs accelerates mature adipocyte formation. Our data show that immature vascular structures in Lyl1 adipose tissues prevent adipocyte stem cell (ASC) maintenance in the niche vessel walls, accelerating their differentiation into adipocytes and causing premature stem cell depletion.

2.1.1. Stromal vascular fraction isolation and culture

Stromal vascular fraction (SVF) cells were isolated from BAT and epididymal (eWAT) and inguinal WAT (ingWAT) of 8‐day‐old puppies and 12‐week‐old males, respectively. Fat pads were excised, finely cut with scissors and incubated in digestion medium containing 2% BSA, 1 M CaCl2, Dispase II (2.4 U/mL for BAT and ingWAT, and 1.2 U/mL for eWAT) and collagenase D (2.5 mg/mL for BAT and ingWAT and 1.25 mg/mL for eWAT) (Roche Life Science, France) at 37°C for 40 minutes for BAT and ingWAT and 30 minutes for eWAT with vortexing every 10 minutes. Floating adipocytes were separated from the SVF by centrifugation at 500g for 5 minutes. SVF cells were sequentially filtered through 100‐ and 40‐μm filters to generate single‐cell suspensions and then plated in Petri dishes with DMEM‐F12 and Glutamax (Thermo Fisher Scientific, France) supplemented with 10% fetal bovine serum (FBS) and 5 ng/mL bFGF. Culture medium was changed every 2 days.

2.1.2. Flow cytometry and cell sorting

SVF cells were then stained with anti‐CD45 PerCP (BD Biosciences; 557 235), ‐CD31 PE‐Cy7 (BD Biosciences; 561 410), ‐CD34 eFlour 660 (eBioscience, 50‐0341), ‐SCA‐1 APC‐Cy7 (BD Biosciences, 560 654), ‐CD24 FITC (eBioscience, 11‐0241), and ‐CD140a PE (eBioscience, 12‐1401) antibodies at 4°C for 30 minutes. Samples were then washed and centrifuged at 300g for 5 minutes. ASCs and preadipocytes were analyzed with a BD FACS‐Quanto II or sorted with a BD FACS Aria and data analyses were performed using the BD FACS Diva software.

2.1.3. Stromal vascular particulate isolation, culture, and staining

eWAT and ingWAT from 12‐week‐old male mice were cut in 5 to 10 pieces and incubated with 1 and 2 mg/mL of collagenase D (Roche Life Science, France), respectively, at 37°C for 2 hours with gentle agitation on a Belly Dancer (30° angle). Mixtures were passed through 300 μm mesh to remove big fragments and then through 30 μm mesh to remove single cells. The remaining stromal vascular particulates (SVPs) on the 30 μm mesh were washed off with DMEM‐F12/Glutamax (Thermo Fisher Scientific, France) supplemented with 10% FBS and seeded on coverslips coated with 2% gelatin. Microtubules were allowed to grow from SVPs in DMEM‐F12/10% FBS for 5 days before staining with an anti‐CD31 antibody (BD Biosciences, 557 355). Images of microtubule outgrown from SVPs were acquired with a Leica SP5‐SMD confocal microscope. The angiogenic response was determined in each individual SVP sample by measuring the length of the growing microtubules with the Image J software. Results are presented as the mean ± SD (SD) of the tubule length and analyzed with the Mann‐Whitney test and GraphPad Prism 5.0 (MacKiev).

See Supplemental Materials for Mice, Adipocyte size and beige fat area, Metabolic parameter survey, RNA preparation, mRNA expression analysis, Immunohistochemistry, Clonogenicity test and Adipogenic differentiation, Oil Red O staining, Blood vessel immunofluorescence staining, Extravasation of albumin‐Evans blue within tissues, Whole‐mount confocal microscopy, Primary endothelial cell description, proliferation, Migration and tube formation, and Statistical analysis.

RESULTS

Young adult Lyl1 −/− mice are overweight and display an overall increase of fat mass due to early expansion of all adipose tissues

During our phenotypic analysis of Lyl1 −/− mice, we noticed that young adults were consistently overweight. Compared with wild type (WT) littermates, Lyl1 −/− males showed a significantly and transitory body weight increase from week 8 to week 18 postpartum. This difference reached 14% at week 14 postpartum (Figure 1A). At week 22 postpartum, body weight reached a comparable plateau in both WT and Lyl1 mice. Echo magnetic resonance imaging analysis of 10‐week‐old animals showed a significantly higher percentage of total fat mass in Lyl1 than WT males, whereas the lean body mass fraction was comparable between groups (Figure 1B). Food intake monitoring for 6 weeks showed no significant difference between genotypes (Figure S1A).

FIGURE 1

Expansion of adipose tissue in young adult Lyl1 −/− mice. A, Body weight changes in 3 to 24‐week‐old wild type (WT) (n = 6‐12) and Lyl1 −/− (n = 4‐6) mice. B, Body composition in 10‐week‐old mice was assessed by nuclear magnetic resonance using an EchoMRI whole body composition analyzer. Body fat and lean mass are indicated as percentage of body weight. C, Epididymal WAT (eWAT), inguinal WAT (ingWAT) and BAT to body weight (bw) ratios were determined in 1‐, 3‐, 6‐, 12‐, and 22‐week‐old WT (n = 5‐10) and Lyl1 −/− (n = 5‐10) mice. ND: Not Detected. Results are the mean ± SD of adipose tissue mass (mg)/bw (g) (‰). *P < .05; **P < .01; ***P < .001

There are two major adipose tissues: BAT that plays a central role in energy expenditure to produce heat, and WAT that is specialized in energy storage. As total fat mass increase is associated particularly with WAT expansion, we examined eWAT and ingWAT, as examples of visceral and subcutaneous WAT respectively. Compared with WT mice, the ratios of eWAT and ingWAT to body weight were significantly higher in 12‐week‐old Lyl1 mice (Figure 1C). The BAT to body weight ratio also was significantly higher in 12‐week‐old Lyl1 mice (Figure 1C).

We assessed whether overweight was associated with metabolic syndrome. Lyl1 mice did not show any change in glucose tolerance, insulin secretion in response to glucose load, insulin tolerance, and hepatic glucose production (assessed with the pyruvate tolerance test) compared with WT animals (Figure S1B). To determine whether overweight in Lyl1 mice was caused by deregulated energy homeostasis, we housed adult males in individual metabolic cages. Lyl1 and WT mice showed similar energy expenditure and globally comparable exchanges of oxygen and carbon dioxide (respiratory exchange ratio, RER). However, the significant nonreduction of RER at the end of the light cycle in Lyl1 mice (when animals were fasting) suggested a defect in lipid oxidation, which is usually supported by BAT (Figure S1C). These findings showed that young Lyl1 mice exhibited transient overweight associated with a general expansion of adipose tissues, without signs of metabolic disorder, and unrelated to food intake.

Increased adiposity of all fat tissues in young Lyl1 mice

To determine whether fat pad mass increase was caused by enhanced adipocyte growth (hypertrophy) and/or number (hyperplasia), we analyzed paraffin‐embedded adipose tissue sections stained with hematoxylin‐eosin. Lipid droplets were bigger in Lyl1 than in WT eWAT samples at 12 weeks, but not at 3, 6, and 22 weeks of age (Figure 2A). Similarly, histological analysis of BAT sections from 3‐, 6‐, 12‐, and 22‐week‐old mice showed that lipid droplets were larger in BAT from 12‐week‐old Lyl1 than WT animals (Figure 2B), a phenomenon known as BAT whitening. In 22‐week‐old mice, lipid droplets were relatively heterogeneous in size and did not significantly differ between genotypes. Likewise, analysis of the number of brown‐like adipocytes by measuring the beige fat areas in ingWAT sections showed a marked reduction of beige fat zones associated with a global enlargement of lipid droplets in 12‐week‐old Lyl1 ingWAT compared with WT samples (Figure 2C). In agreement, the expression of the beige/brite adipocyte‐specific genes Tmem26 and Cd137 was reduced in 12‐week‐old Lyl1 ingWAT compared with WT tissue (Figure S2A). Beige zone areas were increased in 3‐week‐old Lyl1 ingWAT compared with WT, suggesting an early functionality of this tissue in Lyl1 mice. The contribution of hyperplasia to fat mass expansion was excluded because the number of Ki67‐positive cells (a proliferative marker) was comparable in eWAT (Figure S2B), BAT and ingWAT (data not shown) tissue sections from 6 and 12‐week‐old WT and Lyl1 mice. These data indicate that the weight increase in Lyl1 mice is due to fat mass expansion and global enlargement of lipid droplets in all adipose tissues.

FIGURE 2

Increased adiposity in all adipose tissues of young adult Lyl1 −/− mice. Left panels: Lipid droplet size in eWAT, A, and BAT, B, was measured as described in the Materials and Methods section. Results are presented as the median lipid droplet area (whiskers = min to max) and analyzed with the Mann‐Whitney test. *P < .05, **P < .01. The beige fat zone in IngWAT, C, was measured with the NDP.view software and is presented relative to the total surface of the ingWAT section. Results are the mean ± SD of the beige fat zone percentage and were analyzed with the Mann‐Whitney test. Right panels: Paraffin‐embedded adipose tissue sections from 3‐, 6‐, 12‐, and 22‐week‐old WT and Lyl1 −/− mice (n = 5‐8 per age and genotype) were stained with hematoxylin–eosin and images visualized with a NanoZoomer slide scanner controlled by the NDP.view software. Scale bar: 50 μm

Early activation of BAT thermogenic potential and premature eWAT aging in young adult Lyl1 mice

Due to the concomitant BAT whitening and decrease in beige fat zones in ingWAT samples from 12‐week‐old Lyl1 mice (two signs of aging), we asked whether LYL1 absence alters the thermogenic potential. Immunostaining showed a significant decrease of UCP1, a protein that mediates heat generation, in BAT of 12‐week‐old Lyl1 compared with WT mice (Figure 3A), in agreement with BAT whitening. Moreover, Cidea mRNA expression was higher in BAT samples from 12‐week‐old Lyl1 than from WT mice (Figure 3B). This is consistent with lipid droplet enlargement (see Figure 2B), given CIDE‐A role in the control of lipid storage and droplet enlargement. Conversely, in very young animals (6 weeks and younger), UCP1 expression was significantly higher in Lyl1 BAT than in controls, suggesting that the thermogenic program might be active earlier in Lyl1 −/− mice. This was confirmed by the increase of beige fat area already in ingWAT from 3‐week‐old Lyl1 mice (see Figure 2C). Similarly, Adrb3 expression was higher in 1‐ and 3‐week‐old Lyl1 than WT BAT, suggesting a potential stronger response to sympathetic β3‐adrenergic stimulation (Figure 3B). The higher Cidea expression in Lyl1 than WT 1‐week‐old puppies Figure 3B suggested early fat browning, a process involved in the acquisition of the multilocular morphology characteristic of functional brown adipocytes. Lastly, mRNA expression of Prdm16, a transcription factor required for the maintenance of brown adipocyte identity and function in adult mice, was lower in BAT from 12‐week‐old Lyl1 than WT mice (Figure 3B), further suggesting BAT dysfunction at that age. Interestingly, as seen for UCP1, Prdm16 expression was higher in 6‐week‐old Lyl1 mice, implying that the thermogenic program might be active earlier in Lyl1 −/− mice.

FIGURE 3

Early activation of BAT thermogenic potential and premature aging features in eWAT of young adult Lyl1−/− mice. A, Quantification with Aperio ImageScope (left panel) of the UCP1 signal in paraffin‐embedded BAT sections from WT (n = 3‐7 per age) and Lyl1 −/− (n = 4‐8 per age) mice. Results are relative to the total surface of the BAT section. BAT tissue sections from WT and Lyl1 −/− mice were immunostained for UCP1 protein and images visualized with a NanoZoomer slide scanner controlled by the NDP.view software (right panels). Scale bar: 20 μm. B, Total RNA was extracted from BAT samples of 1‐, 3‐, 6‐, 12‐, and 22‐week‐old WT and Lyl1 −/− mice (n = 3‐6 per group). Expression of Adrb3 and Cidea was quantified by RT‐qPCR and normalized to Actb. C, Intraperitoneal temperature monitoring in 12‐week‐old mice housed individually in metabolic cages with a temperature of 22°C for 2 d (three mice for each genotype). Representative data of one of experiments three performed in the same conditions. Inset: The area under the curve (AUC) was calculated using the trapezoidal rule. D, Total RNA was extracted from eWAT samples from 12‐ and 22‐week‐old WT and Lyl1 −/− mice (n = 6‐7 per group). Expression of Lep, Srebf1, Pparg, and Cebpa was quantified by RT‐qPCR and normalized to Actb. E, Expression of the genes encoding the proinflammatory cytokines IL‐6 and TNFα was quantified by RT‐qPCR in total RNA from 22‐week‐old WT and Lyl1 −/− eWAT samples (n = 3 per group) and normalized to Actb expression. *P < .05; **P < .01

To evaluate the consequence of BAT and ingWAT whitening on tissue function, we implanted temperature‐sensitive transmitters intraperitoneally in 12‐week‐old mice that were housed in individual metabolic cages at 22°C to 24°C for 3 days. Continuous monitoring of the internal temperature showed significantly lower temperature in Lyl1 −/− mice than in WT controls (Figure 3C), as validated by the calculation of the area under the curve (Figure 3C, inset). Given the major role of BAT and subcutaneous WAT in thermogenesis, we assessed the behavior of 12‐week‐old WT and Lyl1 mice upon exposure to 9°C in individual metabolic cages for 24 hours (Figure S3). Although the difference was not significant, the internal body temperature was lower in Lyl1 −/− mice than in WT controls, suggesting that Lyl1−/− mice control temperature less efficiently in response to cold stress.

In 22‐week‐old mice, lipid droplet size in eWAT samples was comparable between genotypes (see Figure 2A), suggesting that accumulation of large adipocytes in eWAT declines in older Lyl1 males. This adipocyte size reduction is typically observed in white adipose tissues of aged mice. , High levels of circulating leptin in plasma and high leptin gene (Lep) expression in adipocyte tissues also are correlated with aging in rodents and humans. , , , Therefore, we analyzed the mRNA expression of Lep and Srebp1, its downstream target encoding the transcription factor SREBP1 that controls lipogenic genes in eWAT. Compared with WT animals, Lep mRNA expression was increased by more than 2‐fold in 22‐week‐old Lyl1 mice, while Srebp1 was downregulated (Figure 3D). Consequently, the mRNA levels of the lipogenic genes Acaca, Fasn and Scd1 also were reduced in 22‐week‐old Lyl1 eWAT (data not shown). The downregulation of Cebpa and Pparg, two adipogenic genes, in 22‐week‐old Lyl1 eWAT (Figure 3D) further supported the hypothesis of the premature eWAT maturation in Lyl1 mice. As upregulation of proinflammatory cytokines is classically observed in adipose tissues of aged mice, we also evaluated Il‐6 and Tnfa expression in 22‐week‐old eWAT. Compared with WT samples, Il‐6 was significantly upregulated in Lyl1 eWAT, but not Tnfa, although its expression tended to be higher in Lyl1 eWAT (Figure 3E).

These data indicate that WAT and BAT develop and mature earlier in Lyl1 mice, leading to early activation of the thermogenic program of BAT and beige adipocytes but also acceleration of aging‐like processes (ie, BAT whitening and brown‐like adipocytes within ingWAT), classically observed in older adult mice. ,

Early development of adipose tissues leads to premature decline of the adipogenic potential of the SVF of young adult Lyl1 mice

The differentiation of immature progenitor cells into fully mature adipocytes involves the sequential activation of transcriptional programs. To investigate the mechanisms leading to the adipocyte faster maturation in young Lyl1 mice, we assessed the expression of genes encoding transcription factors involved in adipogenic processes in BAT and WAT (Figure 4A). Compared with WT, the expression of the early‐acting genes Zfp423 and Klf4 decreased earlier in Lyl1 BAT and ingWAT. Accordingly, the late‐acting genes Pparg, Cebpa and Srebp1 were activated earlier in Lyl1 than WT BAT and ingWAT. In addition, the expression of Cebpa and Srebp1, normally downregulated upon aging, was prematurely reduced in 12‐week‐old Lyl1 BAT and ingWAT samples. These data show that the adipogenic differentiation program is accelerated in Lyl1 mice.

FIGURE 4

Premature onset of adipogenic differentiation in Lyl1 −/− BAT and ingWAT results in reduced adipogenic potential of stromal vascular fractions (SVFs). A, Total RNA was extracted from BAT and ingWAT samples from 1‐, 3‐, 6‐, and 12‐week‐old WT and Lyl1 −/− mice (n = 4‐7). Expression of adipocyte differentiation genes (Zfp423, Klf4, Cebpa, Pparg, and Srebf1), normalized to Actb, was quantified by RT‐qPCR. B and C, SVFs isolated from ingWAT samples (B) of 12‐week‐old mice (n = 3 per group) and BAT samples (C) of 8‐d‐old mice (n = 3 per group) were differentiated into adipocytes for 6 and 5 d, respectively, and then expression of adipogenic transcription factors (Cebpa, Pparg, and Srebf1) and adipocyte‐related markers (Fabp4, Pgc1a, Ucp1, Dio2, and Plin) was quantified by RT‐qPCR and normalized to 36B4. Upper panels: Representative images of SVFs from ingWAT at day 0 and day 6 of adipocyte differentiation (B) and of SVFs from BAT at day 0 and day 5 of differentiation (C). Scale bar: 100 μm. *P < .05; **P < .01; ***P < .001

Adipocyte progenitors reside in the SVF, with ECs, pericytes, fibroblasts and immune cells. To compare the adipogenic potential of fat tissues of young adult WT and Lyl1 males, we cultured equal numbers of SVF cells from 12‐week‐old ingWAT samples before induction of adipocyte differentiation when they reached confluence. After 6 days in adipogenic differentiation medium, we observed fewer mature adipocytes in Lyl1 than in WT ingWAT‐SVF cultures (Figure 4B). Expression analysis of genes encoding proadipogenic transcription factors (Cebpa, Srebp1, and Pparg) and mature adipocyte markers (Fabp4 and Plin) confirmed the lower adipogenic potential of Lyl1 SVF cells. Similarly, upon adipogenic differentiation, we observed fewer mature adipocytes in SVF cultures from BAT of 8‐day‐old Lyl1 than WT mice, in agreement with the lower expression of Pparg, Fabp4 and BAT‐specific markers (Pgc1a, Dio2, and Ucp1) (Figure 4C). We then compared the capacity of progenitors in WT and Lyl1 ingWAT‐SVF cultures to generate clones that can differentiate into mature adipocytes (Figure S4) by clonogenic assay. The number of hematoxylin‐positive clones was lower in Lyl1 than WT ingWAT‐SVF cultures (64.5 ± 17.7 vs 125.5 ± 3.5), and fewer could differentiate into Oil Red O‐positive mature adipocytes (17% vs 45%).

These data indicate that SVF cells derived from Lyl1 adipose tissues contains fewer immature progenitors that can produce mature adipocytes, leading to a global reduction of the adipogenic potential of ingWAT and BAT in young Lyl1 mice.

Adipocyte progenitors are less numerous in Lyl1 adipose tissue SVF and poorly associated with vasculature

These findings prompted us to investigate how Lyl1 might influence SVF adipogenic capacity and specifically the number of available functional immature progenitors. In subcutaneous WAT, two different cell progenitors (uncommitted ASCs and committed preadipocytes) have been identified based on the expression of specific markers. Accordingly, by incubating SVF cells from ingWAT and BAT samples of 3‐, 6‐, and 12‐week‐old mice with antibodies against the surface markers CD31 (ECs), CD45 (hematopoietic cells), CD34 and SCA‐1 (immature cells), CD140a and CD24, we could isolate and quantify ASCs (CD45− CD31− CD34+ SCA‐1+ CD140a+ CD24+) and preadipocytes (CD45− CD31− CD34+ SCA‐1+ CD140a+ CD24−) (Figure 5A). The ASC fraction was significantly smaller in ingWAT‐ and BAT‐SVF from 6‐ and 12‐week‐old Lyl1 than WT mice (Figure 5B, upper panels). Preadipocytes also were markedly reduced in ingWAT‐SFV cultures from 6‐week‐old Lyl1 −/− compared with WT mice (Figure 5, lower panels). ASC visualization in eWAT samples from 12‐week‐old WT mice by immunofluorescence showed the presence of several CD31− CD140a+ CD24+ ASCs (asterisks in Figure 5C) that were associated with CD31+ structures or were in the tissue parenchyma. Conversely, ASCs were rare in Lyl1 eWAT samples.

FIGURE 5

Adipose stem cells (ASCs) are less numerous in ingWAT and BAT from in Lyl1 −/− mice, and Lyl1 is not expressed in the adipocyte lineage. A, Dot plots showing FACS analysis of SVFs isolated from ingWAT of a 6‐week‐old WT mouse and the gating strategy to isolate preadipocytes (CD140a+ CD24−, green) and ASCs (CD140a+ CD24+, blue). B, ASCs and preadipocytes were quantified by flow cytometry in SVFs isolated from ingWAT and BAT of 3‐, 6‐, and 12‐week‐old WT and Lyl1 −/− mice. Results are presented as the relative number of ASCs or preadipocytes in SVFs (ie, number of cells divided by the mean number in the WT condition × 100). The fraction of ASCs and preadipocytes in adipose tissues varied from 0.05% to 0.20% and from 10% to 35% of live cells, respectively; n = 3‐6 mice/age/genotype. C, ASCs were visualized in whole‐mount 12‐week‐old eWAT as CD24+ CD140a+ CD31− cells (asterisks). Scale bar: 20 μm. D, Analysis of Lyl1, Lmo2, CD31, CD45, and Pparg mRNA expression in the indicated cell fractions and primary cells by RT‐qPCR. Total RNA was extracted from cells derived from 12‐week‐old WT mice (n = 3). SVFs were prepared from ingWAT, and mature adipocytes (Mat Adipo) were isolated from eWAT. ASCs and preadipocytes (Pre‐Adipo) were FACS‐sorted from ingWAT‐SVFs using specific markers, and mouse lung endothelial cells (mLEC) were prepared from dissociated lungs, as previously described (Pirot et al ). cDNAs were amplified in triplicate with specific mouse primers (Table S1) and normalized to Actb; data are the mean ± SD

Lyl1 is not expressed in the adipocyte lineage

We next wanted to determine the mechanisms by which Lyl1 regulates the number of ASCs and progenitor cells. We checked Lyl1 expression in FACS‐isolated ASCs and preadipocytes compared with mature adipocytes and endothelial cells (Figure 5D). Lyl1 was expressed in SVF cells, in agreement with the presence of many CD31+ ECs and CD45+ hematopoietic cells known to express Lyl1. Lyl1 expression was low in purified ASCs and preadipocytes (15.6% and 11.7%, respectively), compared with purified mouse lung ECs (mLEC). Moreover, LYL1 is unlikely to be active and functional in ASCs and preadipocytes because these cells do not expressed Lmo2, a gene that encodes an obligate LYL1 partner in ECs and hematopoietic cells. The weak Lyl1 expression in the mature adipocyte fraction could be due to the faint contamination by Lyl1‐expressing cells, such as ECs and hematopoietic cells. Furthermore, Lyl1 was not expressed in committed 3T3‐L1 preadipocytes at any stage of adipocyte differentiation (Figure S5A). Importantly, equal numbers of ASCs and preadipocytes isolated from ingWAT‐SVF of 12‐week‐old WT and Lyl1 mice reached confluence at the same time, suggesting comparable proliferative rates (data not shown). Moreover, when stimulated with the adipogenic cocktail, they underwent similar robust mature adipocyte differentiation (Figure S5B). This indicates that Lyl1 absence has no effect on the intrinsic adipogenic potential of immature progenitors. Together, these data demonstrate that Lyl1 exerts a noncell autonomous function on adipocyte development.

As Lyl1 is expressed in the myeloid lineage and obesity has been associated with accumulation of newly recruited macrophages in adipose tissue, , we assessed the presence of F4/80+ macrophages in eWAT and ingWAT of 6‐ and 12‐week‐old mice. Accumulation of macrophages, before the onset of overweight (6 weeks) or during fat expansion (12 weeks), was comparable in WT and Lyl1 adipose tissues (Figure S6).

Blood vessels in Lyl1 adipose tissues are immature and prone to angiogenesis

Lineage tracing studies showed that within fat tissues, the capillary vascular wall is the niche of adipocyte progenitors. , Furthermore, neovascularization and adipogenesis are temporally and spatially linked throughout life.

To precisely pinpoint Lyl1 role in vessel growth, we first investigated its role in the early steps of angiogenesis: EC proliferation and migration. Therefore, we performed gain and loss experiments in two primary human EC lines (UCB‐EC and HUVEC). Neither Lyl1 overexpression by Ad‐Lyl1 transduction nor Lyl1 downregulation by siRNA transfection affected EC proliferation and migration (Figure S7A and B). Similarly, Lyl1 silencing in primary human ECs did not disturb 3D‐tubulogenesis at 24, 48, and 72 hours (Figure S7C).

As we previously reported that Lyl1 is required for the maturation and stabilization of newly formed vessels, , we investigated whether Lyl1 absence also affected the vascular compartment of the adipose tissue niche. To evaluate vessel coverage in BAT vascular structures, we double‐stained BAT cryosections with anti‐CD31 (ECs) and anti‐NG2 (pericytes) antibodies. Quantification of the NG2‐positive area relative to the CD31‐positive surface showed that blood vessels in BAT were 66% less covered by pericytes in Lyl1 than in WT samples (Figure 6A). This poor blood vessel coverage by pericytes was confirmed with PDGFRβ, a second pericyte marker (Figure S8B). As VE‐cadherin is a major actor in establishing functional endothelial barriers and in maintaining their integrity, we analyzed its localization in ingWAT endothelium of 12‐week‐old mice previously injected in vivo with FITC‐labeled lectin to visualize blood vessels. VE‐cadherin and ZO1 staining at cell‐cell junctions were significantly decreased in Lyl1 (by about 54% and 37.5%, respectively) compared with WT ingWAT samples (Figure 6B and Figure S8C). The poor vessel coverage by mural cells and the defective recruitment of VE‐cadherin and ZO1 at EC junctions strongly suggested that the vascular barrier integrity was affected in Lyl1 adipose tissue, as already observed in lungs of young Lyl1 mice. Evans blue dye extravasation was increased by 2.3 and 1.4‐fold in ingWAT and BAT, respectively, from 12‐week‐old Lyl1 mice compared with WT animals, confirming the vessel structure leakiness and instability in Lyl1 mice (Figure 6C). Moreover, in Lyl1 ingWAT samples, NG2‐positive pericytes were loosely attached compared with the packed and regular coverage of vessels in WT controls (Figure S8A). We then isolated SVPs from adipose tissues of 12‐week‐old mice, using a procedure that maintain the native stromal vascular structure while removing all mature adipocytes, and compared their angiogenic potential. New endothelial tubes were more numerous and longer in SVP cultures from Lyl1 than WT ingWAT (Figure 6D) and eWAT (Figure S8D) samples, revealing that Lyl1‐deficiency increases SVP angiogenic potential.

FIGURE 6

Blood vessels in Lyl1 −/− adipose tissues are immature and prone to angiogenesis. A, Reduced pericyte coverage of BAT vessels in Lyl1 −/− mice. Left panels: Microphotographs showing the severe reduction of pericyte coverage of BAT blood vessels in Lyl1 −/− mice. Scale bars: 30 μm. Right panel: Quantification of NG2‐positive cell coverage as the ratio between the NG2‐positive area and the CD31‐positive area. Data are the mean ± SD. B, Lyl1 −/− ingWAT vessels show reduced VE‐cadherin recruitment to endothelial cell junctions. FITC‐labeled lectin (green) was retro‐orbitally injected in 12‐week‐old WT and Lyl1 −/− mice to visualize blood vessels. After 30 minutes, ingWAT tissues were collected and 2 mm3 fragments stained with anti‐VE‐cadherin antibodies (red). Left panels: Microphotographs illustrating the strong reduction of VE‐cadherin staining in Lyl1 −/− compared with WT ingWAT. Scale bar: 30 μm. Right panel: Quantification of VE‐cadherin as the ratio between VE‐cadherin‐positive area and lectin‐positive area. Data are the mean ± SD. C, Increased leakiness of Lyl1 −/− blood vessels in ingWAT and BAT. Evans blue dye was intravenously injected in 12‐week‐old WT and Lyl1 −/− mice. After 30 minutes, dye extravasation was measured in ingWAT and BAT, as described in the Methods section. D, Lyl1 deficiency increases the angiogenic potential of ingWAT stromal vascular particulates (SVPs). Representative images of three experiments. The angiogenic response in each SVP samples was determined by measuring the length of the growing microtubules with Image J and analyzed with the Mann‐Whitney test. Scale bar: 80 μm. **P < .01; *** P < .001

In summary, the vascular structures of Lyl1 adipose tissues exhibit impaired VE‐cadherin and ZO1 recruitment at EC junctions, reduced pericyte coverage, increased leakiness and are more prone to angiogenesis. This is similar to the immature phenotype previously observed in tumor and lung vessels. ,

Angiogenesis is required for triggering early adipogenesis starts in Lyl1 mice

Sprouting angiogenesis is an essential event to trigger adipogenesis during early postnatal and adult adipose tissue development. To determine whether the impaired adipose tissue vasculature of Lyl1 mice could be the starting event, we evaluated angiogenesis before the beginning of the sequential activation of transcriptional factors leading to adipocyte differentiation. We analyzed the vascular network in ingWAT from 1‐week‐old mice and BAT from 17.5 day‐postcoitum embryos (E17.5), the time points that precede adipogenic differentiation onset (see Figure 4A). IngWAT samples were double‐stained with anti‐CD31 and anti‐NG2 antibodies to visualize vessel structures. The CD31‐positive surface and vessel branching were significantly higher in Lyl1 than WT IngWAT samples, revealing that sprouting angiogenesis was strongly increased in Lyl1 ingWAT (Figure 7A). Furthermore, as observed in 12‐week‐old Lyl1 ingWAT, NG2‐positive pericytes were less packed in 1‐week‐old Lyl1 ingWAT compared with the packed and regular coverage of vessels in WT ingWAT (Figure 7B). Likewise, angiogenesis (CD31 staining) was increased in E17.5 BAT from Lyl1 compared with WT animals (Figure 7C). To show that ongoing angiogenesis was an inherent characteristic of adipose tissue vessels and still active just before the appearance of the body weight difference at 12 week of age, we assessed CD31 expression in BAT and ingWAT from 6‐week‐old mice. Angiogenesis was significantly increased in Lyl1 BAT and ingWAT compared with WT samples (Figure S8E and F). These data demonstrate that in Lyl1 immature adipose tissues, angiogenesis starts earlier and causes their premature and faster development.

FIGURE 7

Angiogenesis is required for early triggering of adipogenesis in Lyl1 −/− mice. A, Increased angiogenesis in 6‐d‐old Lyl1 −/− ingWAT samples. Left panels: Microphotographs illustrating the increased vessel development in ingWAT of Lyl1 −/− mice. IngWAT cryosections were stained with antibodies against CD31 (green). Scale bars: 30 μm. Right panel: Quantification of CD31 staining and vessel branching. Results are presented as the median (whiskers = min to max) and analyzed with the unpaired t test with Welch's correction. *P < .05; n = 5‐7 mice/genotype. B, NG2‐positive pericytes (red) are loosely attached to CD31‐positive endothelial cells (green) in Lyl1 −/− ingWAT vessels. Scale bars: 15 μm. C, Increased vascular network development in 17.5E Lyl1 −/− BAT. Left panels: Microscopy images illustrating the increased vessel development in BAT of Lyl1 −/− mice. BAT paraffin‐embedded sections were stained with antibodies against CD31 (green). Scale bars: 20 μm. Right panel: The CD31‐positive area was quantified with the ImageJ software. Results are presented as the median (whiskers = min to max) and analyzed with the unpaired t test with Welch's correction. ***P < .001. n = 3‐5 embryos/genotype

DISCUSSION

This study objective was to investigate the mechanisms of the transient overweight observed in young Lyl1 −/− adult mice. We found that overweight was linked to the faster development of adipose tissues, due to the earlier differentiation of immature progenitors. Consequently, in Lyl1 mice, the premature adipogenic potential decline was not linked to cell‐autonomous defects of progenitor cells, but to the reduced availability of stem cells. Given the importance of the adipose tissue vascular niche, we found that in Lyl1 fat tissues, the vascular structures are immature and highly permeable, and therefore could accelerate adipocyte differentiation, as already described for the HSC niche.

In Lyl1 mice, mature adipocytes develop earlier in the three fat tissue types. The difference with WT is visible at 1 and 3 weeks of age for BAT and ingWAT, respectively, and at 6 weeks of age for eWAT, in agreement with the development kinetics of these adipose tissues. Consequently, in young adult Lyl1 mice, adipose tissues exhibit prematurely an aging‐like phenotype, such as BAT and ingWAT whitening, as illustrated by lipid droplet enlargement and loss of mitochondrial UCP1 expression. , In 22‐week‐old Lyl1 mice, lipid storage in eWAT starts to decrease, as shown by the smaller adipocytes and decreased expression of several lipogenic genes. Moreover, Lyl1 eWAT expresses higher levels of Lep and genes encoding inflammatory molecules, as generally observed in older animals. , , , , ,

Aging has detrimental consequences on the adipose tissue specific functions (ie, nonshivering thermogenesis for BAT and ingWAT, and lipid storage for eWAT). , In agreement with the lower UCP1 expression and loss of beige fat zones, young adult Lyl1 mice have a lower internal temperature at room temperature. However, this premature aging‐like phenotype does not appear to be detrimental because in our animal facility, Lyl1 mice can be maintained for 2 years without major physical problems.

Our data indicate that the early adipose expansion in Lyl1 mice occurs at the adipocyte progenitor levels through noncell autonomous mechanisms. ASCs and preadipocytes isolated from WT and Lyl1 mice show a similar adipogenic potential, in agreement with the absence of Lyl1 expression in the adipocyte lineage. Conversely, the adipogenic potential of SVFs from Lyl1 BAT and ingWAT declines prematurely due to the reduction in cell number that affects particularly the ASC compartment. As it is the case for several tissues, , , , , the adipose tissue vessel wall represents a reservoir of stem cells that might differentiate into preadipocytes and adipocytes. In the adipose vascular niche, ASCs function as perivascular pericytes, , , and it has been suggested that some ASCs may derive from specialized ECs. These studies highlighted the close relationship between adipocyte formation and endothelium. Indeed, sprouting angiogenesis precedes both postnatal and adult adipose tissue development. , , As we previously showed in aortic ring assays, the vascular structures from Lyl1 eWAT and ingWAT are more prone to ex vivo angiogenesis than WT samples. Consistently, we observed a highly developed vascular network in Lyl1 immature adipose tissues as early as E17.5 in BAT and 1 week after birth in ingWAT, but not in WT samples. This excessive angiogenesis of Lyl1 vascular structures might cause the premature and faster hypertrophy of adipose tissues in Lyl1 mice, because WT and Lyl1 ASCs and progenitors showed the same differentiation capacity.

Similarly to Lyl1 lung blood vessels, Lyl1 vascular structures in adipose tissues display leakiness due to impaired recruitment of VE‐cadherin to adherens junctions and poor vessel coverage by pericytes. The connection between LYL1 and VE‐cadherin was previously studied in newly formed vessels in tumors and in the lung endothelial barrier maintenance. Precisely, LYL1 silencing in primary human ECs does not impair VE‐cadherin mRNA expression, but reduces the expression of genes encoding two guanine nucleotide exchange factors that activate the small GTPase Rap1, essential for VE‐cadherin‐dependent adhesion and for the stabilization of established endothelial junctions. These results were confirmed in primary ECs isolated from the lungs of Lyl1 mice. Thus, LYL1 does not control the expression of VE‐cadherin, but rather modulates its cellular localization at adherens junctions. The decreased pericyte coverage observed in Lyl1 vessels could be the consequence of the permanent immaturity of Lyl1 vessels. Indeed, the final step of “vessel maturation” involves successively VE‐cadherin recruitment at adherens junctions, pericyte attraction and EC‐pericyte cell contacts to trigger the Notch signaling required for the final attachment of pericytes. As VE‐cadherin localization is impaired in Lyl1 vessels, leading to destabilized vessels, we suggest that in these mice the close EC‐pericyte contacts, which are necessary for the final attachment of pericytes, are compromised.

According to their permeability properties, blood vessels could support either stem cell quiescence or activation. , In the HSC niche, less permeable arterial blood vessels maintain HSC quiescence, whereas more permeable sinusoids promote HSC activation and differentiation. Specifically, in Endo mice, endothelial‐specific deletion of fibroblast growth factor receptor 1 and 2 impairs endothelial integrity of bone barrow vessels, as shown by the increased Evans blue diffusion and decreased VE‐cadherin and ZO‐1 expression. This increased permeability leads to a significant reduction of hematopoietic stem and progenitor cells (HSPCs). Thus, we propose that the immature and angiogenic vascular structures of Lyl1 adipose tissues prevents ASC maintenance and self‐renewal at the vessel wall, thereby accelerating their differentiation into adipocytes and causing premature stem cell depletion. Importantly, the adipose tissue niche might not be the only affected vascular niche in Lyl1 mice. Lyl1 adult bone marrow contains an excess of myeloid cells (our unpublished data) and this could be correlated with the presence of fewer HSCs. Indeed, several studies demonstrated that any HSC niche alteration skews HSC differentiation toward the myeloid lineage. , , , Interestingly, following transplantation, the differentiation of HSPCs derived from Endo bone marrow is skewed toward the myeloid lineage.

CONCLUSION

We show here that Lyl1 plays a crucial role in ASC maintenance in a noncell autonomous manner by controlling the adipose tissue endothelial niche. Our study confirms the major structural role of the vascular niche in coordinating self‐renewal and differentiation of stem cells within different tissues. It emphasizes that vascular alterations of the niche may profoundly affect the fate of tissue‐residing stem cells, thereby disturbing postnatal tissue homeostasis.

CONFLICT OF INTEREST

The authors declared no potential conflicts of interest.

AUTHOR CONTRIBUTIONS

A.H., V.D., L.K.: collection and/or assembly of data, data analysis and interpretation; H.T., Y.G.: collection and/or assembly of data; N.P.: conception and design; D.M.: conception and design, manuscript writing, final approval of manuscript; V.P.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

Appendix S1: Supporting Information

Click here for additional data file.

Figure S1 Lyl1 −/− mice display no obvious metabolic disorder classically linked to obesity. A, Food intake is presented as the quantity of ingested food normalized to 20 g of body weight. Food was weighted once per week for 12 consecutive weeks for WT (n = 5) and Lyl1 −/− (n = 4) mice. B, For the glucose tolerance and pyruvate tolerance tests, after overnight fasting, male mice received a peritoneal injection of glucose or sodium pyruvate (2 g/kg), respectively. For the insulin tolerance test, after fasting for 4 hours, males received a peritoneal injection of insulin (0.75 U/kg). Glucose concentration was measured with a blood glucometer (Accu‐Chek, Roche). Data are representative of 2 or 3 experiments performed in the same conditions. C, Male mice were housed in individual cages with precisely controlled temperature and light/dark cycle, and ad libitum access to standard chow diet and water for 24 hours, prior to a 48 hours period of automated recordings. For each mouse, the volume of oxygen (VO2) and of CO2 (VCO2) were recorded every 13 minutes during the entire experiment. The respiratory exchange ratio (RER) was calculated as the VCO2 vs VO2 ratio (VCO2/VO2). Inset: The area under the curve (AUC) corresponding to the end of the light cycle (2‐8 pm) was calculated using the trapezoidal rule. Results were analyzed with the unpaired t test. **P < .01. Each curve shown is a representative experiment out of 3.

Click here for additional data file.

Figure S2 A, Expression of Tmem26 and Cd137 (two beige fat‐specific markers) was quantified by RT‐qPCR in total RNA from 12‐week‐old WT and Lyl1 ingWAT samples (n = 6 per group) and normalized to Actb expression. *P < .05. B, Lyl1 and WT adipose tissues display similar proliferative status. Upper panels: eWAT tissue sections from WT and Lyl1 mice were co‐immunostained with antibodies against Ki67 (proliferative marker; brown) and CD31 (endothelial cells; purple) and images visualized with a NanoZoomer slide scanner controlled by the NDP.view software. Scale bar: 20 μm. Lower panels: Ki67‐positive cells with a round nucleus (red arrowhead) were manually counted, whereas Ki67‐positive cells with an elongated nucleus (asterisks, proliferating endothelial cells) were excluded. For each sample, a minimum of three areas of 3‐5 mm2 were counted. WT (n = 4‐7 per age) and Lyl1 (n = 4‐8 per age).

Click here for additional data file.

Figure S3 Lyl1 mice respond to cold stress as efficiently as WT mice. Intraperitoneal temperature was monitored in 12‐week‐old mice housed individually in metabolic cages with a temperature of 9°C for 24 hours (three mice for each genotype; representative experiment of two performed in the same conditions). The area under the curve (AUC) was calculated using the trapezoidal rule.

Click here for additional data file.

Figure S4 Lower clonogenic capacity of Lyl1 SVF‐IngWAT. A, The same number of SVF cells from WT and Lyl1 IngWAT was seeded and when clones reached 50‐100 cells, adipocyte differentiation was induced by exposure to an adipogenic cocktail for 2 days. After 7 days, clones differentiated into adipocytes were stained with Oil Red O, while hematoxylin staining revealed the presence of all clones. Representative images of three experiments. B, Oil Red O stained clones were counted using the ImageJ software and their number reported to the total number of clones; **P < .01.

Click here for additional data file.

Figure S5 A, Lyl1 is not expressed in 3T3‐L1 cells. Confluent mouse 3T3‐L1 cells were incubated with the adipogenic differentiation cocktail and total RNA was extracted at day 0, 3, and 7 postadipogenesis induction. Analysis of Lyl1, CD31 (endothelial cell marker) and Pparg mRNA expression was performed by RT‐qPCR. B, ACSs and preadipocytes derived from Lyl1 and WT mice exhibit similar adipogenic potential. SVFs were prepared from 12‐week‐old WT and Lyl1 IngWAT. ASCs (upper panels) and preadipocytes (lower panels) were sorted using a BD FACS Aria cell sorter. Equal number of cells were plated and induced to differentiate with an adipogenic cocktail. Representative images of WT and Lyl1 ASCs at day 7 of differentiation (upper panels) and of preadipocytes at day 6 of differentiation (lower panels). Scale bar: 100 μm.

Click here for additional data file.

Figure S6 Comparable macrophage infiltration in WT and Lyl1 adipose tissues. Macrophages were immunostained with the anti‐F4/80 antibody in paraffin‐embedded ingWAT and eWAT sections from 12‐week‐old WT (n = 5) and Lyl1 (n = 5) mice. Cells were manually counted in at least three areas/tissue per mouse with the NDP.Viewer software. Scale bar: 20 μm.

Click here for additional data file.

Figure S7 Proliferation and migration of endothelial cells and tube formation are not altered after LYL1 expression modulation. A, Proliferation of endothelial cells was quantified using a classical MTT Cell Proliferation Assay (2‐4 times in triplicate, depending on the conditions). B, Migration of endothelial cells was assessed with a Boyden chamber assay (2‐5 times, depending on the conditions). C, Differentiation of endothelial cells was monitored by in vitro 3D‐tubulogenesis in collagen gel. Photographs are representative of 2‐4 experiments, depending on the cell type. Scale bar: 500 μm.

Click here for additional data file.

Figure S8 A, Lyl1 vessels are disorganized. FITC‐labeled lectin (green) was retro‐orbitally injected in 12‐week‐old WT and Lyl1 mice to visualize blood vessels. After 30 minutes, ingWAT tissues were collected and 2 mm3 fragments stained with an anti‐NG2 antibody (red). Perivascular pericytes are less adherent (asterisks) in Lyl1 IngWAT samples. B, Reduced pericyte coverage of BAT vessels in Lyl1 mice. Quantification of PDGFRβ‐positive cell coverage as the ratio between the PDGFRβ‐positive area and the CD31‐positive area. Data are presented as the mean ± SD (n = 8‐10 mice/genotype). C, Lyl1 ingWAT vessels show reduced ZO1 recruitment to endothelial cell junctions. Quantification of ZO1 staining as the ratio between the ZO1‐positive area and the CD31‐positive area. Data are presented as the mean ± SD; n = 7 mice/genotype. D, Lyl1 deficiency increases the angiogenic potential of eWAT stromal vascular particulates (SVPs). eWAT samples from 12‐week‐old WT and Lyl1 mice were mildly digested to maintain the vascular structure integrity. After 5 days of growth, structures were incubated with an antibody against CD31 (endothelial cell marker; green). Representative images of three experiments. Scale bar: 80 μm. ***P < .001. quantification? E and F, Vessel density is increased in BAT (E) and ingWAT (F) of 6‐week‐old Lyl1 mice. Left panels: Tissue sections were immunostained for CD31 (purple). Scale bar: 20 μm. Right panels: CD31‐positive cells were manually counted. For each samples a minimum of four areas of 1‐2 mm2 each were counted. WT (n = 3‐4 per tissue) and Lyl1−/− (n = 4‐5 per tissue). Results were analyzed with the unpaired t test. ***P < .001; ****P < .0001.

Click here for additional data file.

Table S1 Sequences of the Mus musculus primers used in RT‐qPCR.

Click here for additional data file.

Table S2 Sequences of the siRNA used in HUVEC transfections.

Click here for additional data file.