|Home | About | Journals | Submit | Contact Us | Français|
Inflamed atherosclerotic plaques can be visualized by non-invasive PET-CT imaging with 18FDG, a glucose analog but the underlying mechanisms are poorly understood.
Here, we directly investigated the role of Glut1-mediated glucose uptake in ApoE−/− mouse model of atherosclerosis.
We first show that the enhanced glycolytic flux in atheromatous plaques of ApoE−/− mice was associated with the enhanced metabolic activity of hematopoietic stem and multi-potential progenitors (HSPCs) and higher Glut1 expression in these cells. Mechanistically, the regulation of Glut1 in ApoE−/− HSPCs was not due to alterations in hypoxia-inducible factor 1α (HIF1α) signaling or the oxygenation status of the bone marrow but was the consequence of the activation of the common β subunit of the granulocyte macrophage colony-stimulating factor/interleukin-3 receptor driving glycolytic substrate utilization by mitochondria. By transplanting BM from WT, Glut1+/−, ApoE−/− and ApoE−/−Glut1+/− mice into hypercholesterolemic ApoE deficient mice, we found that Glut1 deficiency reversed ApoE−/− HSPC proliferation and expansion, which prevented the myelopoiesis and accelerated atherosclerosis of ApoE−/− mice transplanted with ApoE−/− BM and resulted in reduced glucose uptake in the spleen and aortic arch of these mice.
We identified that Glut1 connects the enhanced glucose uptake in atheromatous plaques of ApoE−/− mice with their myelopoiesis through regulation of HSPC maintenance and myelomonocytic fate and suggest Glut1 as potential drug target for atherosclerosis.
Atherosclerosis is a chronic, hypercholesterolemia-driven inflammatory disease that is initiated by the deposition of cholesterol-rich lipoproteins in the artery wall, leading to monocyte-macrophage recruitment. Hypercholesterolemia and/or defective cholesterol efflux have also been documented to induce myelopoiesis, which contributes to atherosclerotic lesion formation by fueling plaques with monocytes and neutrophils.1,2 The monocyte count, in particular, independently predicts risk for coronary artery disease after adjustment for conventional risk factors.3,4
Hematopoietic stem cells (HSCs) are quiescent in the bone marrow (BM) niche and are the source of all hematopoietic stem and multi-potential progenitors (HSPCs) and differentiated cells that are critical for the maintenance and replenishment of peripheral leukocytes in adult life, particularly during emergency hematopoiesis. However, we and others have recently shown that chronic cholesterol accumulation in HSPCs due to hypercholesterolemia and/or defective apolipoprotein (Apo)-mediated cholesterol efflux promotes pathogenic HSPC expansion and proliferation leading to uncontrolled myelopoiesis.5–8 For instance, in the ApoE−/− mouse model of atherosclerosis, the progressive HSPC expansion that drives myelopoiesis,6 contributed to provide the inflamed atherosclerotic lesions with neutrophils and monocytes.9–11 Although recent research has focused on elucidating the roles of cytokines and the microenvironment in the proliferation, mobilization and commitment of HSPCs in preclinical model of atherosclerosis,1,2 the cellular metabolic pathways that regulate these processes remain unknown.
Lessons from various mutant mice displaying a wide range of bioenergetic defects in vivo have pointed to a central role for the mitochondrial energy metabolism in HSC stemness.12–17 Mounting evidence also suggests that HSC quiescence requires a hypoxic environment,18,19 to maintain glycolysis-biased metabolic activity instead of mitochondrial oxidative phosphorylation (OXPHOS).20,21 By limiting mitochondrial respiration and ATP production, this could indeed prevent HSCs from producing reactive oxygen species (ROS) to avoid their differentiation and exhaustion.22–24 In contrast, funneling glucose to the mitochondria for Krebs cycle utilization is required when the HSCs become proliferative or undergo differentiation, most likely due to the high energy demand of these cellular processes.21,25 More recently, Oburoglu et al., have also reported that glucose utilization can dictate the myeloid lineage commitment in human HSCs.26 Intriguingly, increased hematopoietic metabolic activity can be visualized by non-invasive PET-CT imaging with 18FDG, a glucose analog, not only in inflamed atherosclerotic plaques,27–30 but also in the spleen of patients with cardiovascular diseases,31,32 reflecting most likely an extramedullary hematopoiesis.33 However, the relevance of these observations as well as the underlying mechanisms are not fully understood.
In an attempt to better understand the relationship between the enhanced hematopoietic glycolytic activity, HSPC proliferation, myelopoiesis and the development of atherosclerotic lesions, we first showed that an enhanced hematopoietic glycolytic activity in the aortic arch, the BM and the spleen of ApoE−/− BM transplanted mice was associated with an enhanced Glut1 expression in ApoE−/− HSPCs. Mechanistic studies showed that the up-regulation of Glut1 in ApoE−/− HSPCs was not due to an alteration of the oxygenation status of the BM niche but rather was dependent on Ras signaling downstream of the granulocyte macrophage colony-stimulating factor/interleukin-3 receptor driving glycolytic substrate utilization by mitochondria. Finally, we carried out BM transplantation from mice with single or combined deficiencies of ApoE or the glucose transporter Glut1 into ApoE−/− mice. Consistent with our hypothesis, ApoE−/− mice that had received ApoE−/−Glut+/− BM showed reduced HSPC proliferation and expansion, myelopoiesis and atherogenesis compared to mice that had received ApoE−/− BM. Thus, we propose a causal relationship between the enhanced hematopoietic glycolytic activity in ApoE−/− mice and their myelopoiesis through regulation of HSPC expansion and fate, offering novel therapeutic perspectives.
Materials and additional methods are available in Supplementary material.
Glut1+/− mice (kindly provided by Dr. De Vivo, Columbia University) have been crossed to C57BL/6J for more than 12 generations within our colony. ApoE−/− (B6.129P2-Apoetm1Unc/J), Mx1-cre (B6.Cg-Tg(Mx1-cre)1Cgn/J) and HIF1αfl/fl (B6.129-Hif1αtm1Kats/J) mice on a C57BL/6J background were obtained from The Jackson Laboratory. ApoE−/− (B6.129P2-Apoetm1Unc/J) mice expressing Ly5.1 (CD45.2) were crossed to WT mice expressing Ly5.2 (CD45.1) to generate ApoE−/− mice expressing Ly5.2 (CD45.1). For each experiment, littermate controls were generated. For the neutralizing antibody experiment, WT and ApoE−/− mice were i.v injected with IgG control or IL-3Rβ AF549 antibody (R&Dsystems). Animal protocols were approved by the Institutional Animal Care and Use Committee of the French Ministry of Higher Education and Research and the Mediterranean Center of Molecular Medicine (Inserm U1065) and the were undertaken in accordance with the European Guidelines for Care and Use of Experimental Animals. Animals had free access to food and water and were housed in a controlled environment with a 12-h light-dark cycle and constant temperature (22°C).
BM transplantation was performed as previously described,5 using BM from 12 to 14 weeks old WT, Glut1+/−, ApoE−/−, and ApoE−/−Glut1+/ littermates with no significant variation in peripheral leukocyte counts (WT, 6.1±0.5×106/mL; Glut1+/− 5,6±0.6×106/mL; ApoE−/− 7.1±0.6×106/mL; ApoE−/−Glut1+/− 6,7±0.5×106/mL) or BM leukocyte counts (WT, 76±12×106; Glut1+/− 75±11×106; ApoE−/− 94±21×106; ApoE−/−Glut1+/− 88±16×106). Briefly, studies were conducted in 12wks old female ApoE−/− mice fed a Western type diet (TD88137) from Harlan Teklad (Madison, WI) for 12wks. Mice were allowed to recover for five weeks after irradiation and BM transplantation before diet studies were initiated. Body weight was recorded at indicated time points. Mice were euthanized in accordance with the American Veterinary Association Panel of Euthanasia. Spleen weights were determined at the time of sacrifice.
For identification of peripheral blood leukocytes, 100μl of blood was collected into EDTA tubes before red blood cell lysis (BD Pharm Lyse; BD Biosciences), filtration and staining for 30min on ice. Cells were stained with a cocktail of antibodies against CD45, Ly6C/G, CD115, B220, TCRβ and CD8 as previously described.5 Briefly, monocytes were identified as CD45+CD115+, neutrophils as CD45+CD115−Ly6C/Ghi, B-lymphocytes as CD45+B220+, T-lymphocytes as CD45+TCRβ+ and further subdivided into CD4+, CD8+ and CD8+Ly6C/G+ subsets.
BM cells were collected from leg bones, lysed to remove RBCs and filtered before use. Freshly isolated BM cells were stained with the appropriate antibodies for 30min on ice. For haematopoietic subsets, the following lineage antibodies were used: a cocktail of antibodies to lineage committed cells (CD45R, CD19, CD11b, CD3e, Ter-119, CD2, CD4, CD8 and Ly6C/G) and the following stem cell markers: c-Kit, Sca-1, Flt3 (also known as CD135), CD150 (Slamf1), CD34 and FcgRII/III as previously described.5 Briefly, HSPCs were identified as lin−Sca1+c-Kit+ (LSKs) and HSPC subsets were identified from the most quiescent as long-term LT-HSC (CD34−CD150+Flt3−) to the most cycling as ST-HSC (CD34+CD150+Flt3−) and multipotential progenitors (CD34+CD150−Flt3− > CD34+CD150−Flt3+). Hematopoietic progenitor cells were identified as common myeloid progenitors (CMP, Lin−Sca1−c-Kit+CD34intFCgRII/IIIint), granulocyte monocyte progenitors (GMP, Lin−Sca1−c-Kit+CD34intFCgRII/IIIhi) and megakaryocyte erythrocyte progenitors (MEP, Lin−Sca1−c-Kit+CD34loFCgRII/IIIlo). For DNA content analysis, HSPC stained bone marrow cells were fixed in 1% paraformaldehyde in PBS, washed, and stained with 4′,6-diamidino-2-phenylindole (5μg/mL DAPI, Molecular Probes). Cell surface expression of Glut1 was quantified using Glut1 FAB1418 antibody from R&Dsystems or Glut1 RBD ligand (Metafora Biosystems). To assess the uptake of 2-NBDG, prestained cells were incubated with 10 μM 2-NBDG (Invitrogen) for 30 min as previously described.34 The mitochondrial membrane potential and ROS production were analyzed with 25 nM fluorescent TMRE (AnaSpec) and 5μm CM-H2DCFDA stainings for 30 min on prestained BM cells. In addition, autophagy was monitored in live cells using Cyto-ID autophagy detection kit (Enzo Life Sciences, Lyon, France) according to manufacturer instructions. Viable cells, gated by light scatter or exclusion of CD45− cells, were analyzed on a four-laser BD Canto cell analyzer or sorted on a BD FACSAria Cell Sorter both running with DiVa software (BD Biosciences). Data were analyzed using FlowJo software (Tree Star Inc.)
Data are shown as mean ± SEM. Statistical significance was performed using Prism t-test or ANOVA were performed according to the dataset. Results were considered as statistically significant when P<0.05.
To monitor the metabolic activity of hematopoietic cells, we first investigated the uptake of the radiolabeled D-glucose analogue 2-deoxy [14C] glucose in organs isolated from irradiated ApoE−/− recipient mice transplanted with either WT or ApoE−/− bone marrow (BM). A more than 2-fold increase in 2-deoxy [14C] glucose uptake was observed not only in the aortic arch of ApoE−/− BM transplanted mice compared to controls but also in their BM and spleen (Fig. 1A). An approximately 3.5-fold increase in 2-deoxy [14C] glucose uptake was also consistently observed in colony forming unit assays with the multipotential progenitors (CFU-GEMM) and granulocyte macrophage progenitors (CFU-GM) from the ApoE−/− mice (Fig. 1B). The oxygen consumption was 1.3-fold higher in ApoE−/− lineage marker-positive (Lin+), ApoE−/− Lin− BM cells and Lin−Sca1+ progenitors; these cell types represent mature leukocytes or a mix of HSPCs, respectively (Fig. 1C). The higher oxygen consumption seen in ApoE−/− cells was most likely maintained by mitochondrial oxygen consumption, as treatment with oligomycin, which inhibits mitochondrial ATP synthase, clearly suppressed their respiration (Fig. 1D). Quantification of the citric acid metabolites by LC-MS showed higher citrate, fumarate and malate but not succinate in ApoE−/− leukocytes (Fig. 1E). This was associated with a 1.7-fold increase in succinate dehydrogenase activity in ApoE−/− leukocytes (WT, 13.7±1.9. vs. ApoE−/−, 24.2±4.1 OD/min/mg protein, respectively). Consistent with these findings, an approximately 1.3-fold increase in the mitochondrial membrane potential was observed in ApoE−/− Lineage marker-positive (Lin+) and Lin− BM cells by flow cytometry using a fluorescent tetramethylrhodamine ethyl ester (TMRE) dye (Fig. 1F). An analysis of the different populations within the HSPCs showed that the mitochondrial potential of the CD34− long-term hematopoietic stem cells (LT-HSCs) and CD34+ HSPCs,35,36 was increased to the same extent as the Lin− BM cells in the ApoE−/− mice (Fig. 1F). This was associated with an increase in reactive oxygen species (ROS) staining in these cells (data not shown). To test whether this phenotype was caused by modulation of the glycolytic pathway in HSPCs, we next used the fluorescent D-glucose analog 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxy-D-glucose (2-NBDG) as a tool to examine the glucose uptake in these cells.34 The NBDG staining was increased by approximately 1.5-fold in the ApoE−/− Lin+ and Lin− BM cells and CD34+ HSPCs, but not in the more primitive fractions of the CD34− LT-HSCs (Fig. 1G). We next assessed the cell surface expression of the glucose transporter 1 (Glut1) in these cells by flow cytometry. An approximately 1.25-fold increase in the Glut1 cell surface expression was observed in the ApoE−/− Lin+ and Lin− BM cells and CD34+ HSPCs but not CD34− LT-HSCs (Fig. 1H). This was similar to the NBDG pattern. Together, these observations suggest that the enhanced Glut1 expression and associated glycolytic activity in organs from ApoE−/− BM transplanted mice could reflect the metabolic state not only of leukocytes but also HSPCs.
We next set out to better understand the mechanism leading to Glut1 regulation in the ApoE−/− HSPCs. Studies have proposed that the hypoxia inducible factor 1α (HIF1α) up-regulates Glut1,37 and HIF1α contributes to HSPC homeostasis.18,20 We first assessed the hypoxic state of the BM cells isolated from irradiated ApoE−/− recipient mice transplanted with either WT or ApoE−/− BM by flow cytometry with a fluorescein (FITC)-conjugated anti-Pimonidazole (Pimo) antibody at 90 min after intravenous Pimo administration. We did not observe significant changes in Pimo staining in either the ApoE−/− Lin+ and Lin− BM cells or CD34+ HSPCs and CD34− LT-HSCs (Supplemental Fig. IA). HIF1α protein was also barely detectable in WT and ApoE−/− BM cell lysates under normoxic culture conditions and cell lysates from ApoE−/− BM cells showed amounts of HIF1α protein similar to those of WT cells under hypoxia (Supplemental Fig. IB). Even though, the increase in lactate dehydrogenase A (Ldha) mRNA expression, a HIF1α target gene, upon hypoxia was similar between WT and ApoE−/− BM cultures, Glut1 gene expression was higher under both normoxic and hypoxic culture conditions in ApoE−/− BM cultures (Supplemental Fig. IC). Thus, to directly test the contribution of HIF1α in the regulation of Glut1-dependent glucose metabolism in ApoE−/− HSPC in vivo, we next generated an inducible, hematopoietic-specific HIF1αknockout (Mx1-cre HIF1αfl/fl) on a WT or ApoE−/− background. The BM from these mice was transplanted into irradiated ApoE−/− mice and after a recovery period, the recipients were then placed on a high fat diet (HFD) for 12 weeks to exacerbate their HSPC expansion (Fig. 2A).6 HIF1αwas deleted from hematopoietic cells before the start of the diet by sequential polyl:polylC (PIpC) injections, which efficiently excised the HIF1αgene from the BM cells (Fig. 2B and and2C).2C). Ldha mRNA expression was also significantly reduced in the BM of these mice, but Glut1 was only marginally regulated (Fig. 2C). HIF1αdeficiency also did not alter the cell surface expression of Glut1 in CD34+ HSPCs and CD34− LT-HSCs (Fig. 2D) or the frequency of these cells (Fig. 2E). Furthermore, quantification of the blood myeloid cells in these mice revealed that HIF1αdeficiency further increased the neutrophil, monocyte and eosinophil counts in these mice (Fig. 2F). Together, these findings suggest that HIF1α does not mediate the upregulation of Glut1 in ApoE−/− HSPC or their expansion and minimally contributed to myelopoiesis under hypercholesterolemic conditions.
Glut1 can be up-regulated by growth hormone-dependent activation of oncogenes, such as Ras or Src.38,39 Therefore, we investigated the expression of Glut1 in WT and ApoE−/− BM cultures in response to various growth hormones. The Glut1 mRNA levels in WT BM cells were increased upon stimulation with GM-CSF and IL-3, but not Flt3L or TPO, and this response was further increased in the ApoE−/− BM cells and blunted by a farnesyl transferase inhibitor that blocks Ras activation (Supplemental Fig. IIA). These responses were not observed for the HIF1α or Ldha mRNAs (Supplemental Fig. IIA). Flow cytometry analysis confirmed an increase in Glut1 cell surface expression in ApoE−/− HSPCs upon IL-3 and GM-CSF stimulation compared to WT HSPCs (Supplemental Fig. IIB). These effects were abrogated by blocking the IL3Rβ signaling pathway and downstream Ras activation with a farnesyl transferase inhibitor, but not by the Jak2 inhibitor, AG490 or the AMP-activated protein kinase (AMPK) activator, metformin (Supplemental Fig. IIB). Removing plasma membrane cholesterol with cyclodextrin also prevented the enhanced Glut1 expression in ApoE−/− HSPCs confirming the central role of cholesterol in this regulation (Supplemental Fig. IIB), To directly test the relevance of these observations in vivo, an IL3Rβ blocking antibody was next injected into the WT and ApoE−/− mice. Consistent with earlier work,40 we first showed that this antibody efficiently reduced myelopoiesis in the ApoE−/− mice and had no effect in WT mice over a 24 h period (Fig. 3A). An analysis of the genes in the glycolytic pathway in the BM cells at the end of the study period revealed no significant changes in the HIF1α or Ldha mRNAs, but the Glut1 mRNA was down-regulated after treatment with the IL3Rβ blocking antibody in the ApoE−/− BM (Supplemental Fig. IIC). Quantification of the HSPCs in the BM of these mice by flow cytometry also revealed reduced numbers of CD34+ but not CD34− HSPCs in the ApoE−/− mice but not the WT mice (Fig. 3B), which correlated with reduced cycling of these cells (Fig. 3C). This was associated with the reduced cell surface expression of Glut1 in the ApoE−/− CD34+ HSPCs (Fig. 3D). These results revealed that the metabolic requirements for proliferation and expansion of the ApoE−/− HSPCs is associated with the IL-3/Glut1 axis and not the HIF1α/Glut1 axis in vivo.
To determine the contribution of mitochondrial OXPHOS on ApoE−/− HSPC proliferation and lineage specification upon IL-3 and GM-CSF treatment, we next artificially suppressed various enzymes that are intricately involved in the regulation of the TCA cycle using pharmacological inhibitors (Fig. 4A). We first validated our in vitro BM culture assay by showing that inhibition of the IL-3Rβ signaling pathway (IL3Rβ blocking antibody), inhibition of Ras signaling (farnesyl transferase inhibitor FTI-277) and plasma membrane cholesterol depletion (cyclodextrin CD) prevented ApoE−/− HSPC expansion (Fig. 4B) and the generation of CD11b+Gr-1+ myeloid cells upon IL-3 and GM-CSF treatment (Fig. 4C). In contrast, inhibition of lactate dehydrogenase (LDH) and acetyl-CoA carboxylase (ACC) using oxamate and Tofa, respectively or activation of AMPK with metformin did not alter ApoE−/− HSPC expansion (Fig. 4B) or their myeloid fate (Fig. 4C). Surprisingly, inhibition of mitochondrial respiratory chain complex I with rotenone was also not sufficient to dampen ApoE−/− HSPC expansion and myeloid commitment (Fig. 4B and and4C).4C). Given the enhanced succinate dehydrogenase (SDH) activity observed in ApoE−/− BM cells, we next evaluated the contribution of the mitochondrial complex II. Inhibition of SDH with 3-nitropropionic acid (3-NPA) specifically prevented the myeloid fate of ApoE−/− HSPCs but not their expansion (Fig. 4B and and4C).4C). We next assessed whether inhibition of the conversion of pyruvate for entry to the TCA cycle with a pyruvate dehydrogenase (PDH) inhibitor (CPI-613) and/or a pyruvate carboxylase (PC) inhibitor (chlorothricin) could alter the expansion and myeloid fate of ApoE−/− HSPCs. Although ApoE−/− HSPC expansion required both inhibition of PDH and PC (Fig. 4B), their myeloid fate was actually suppressed by inhibiting either PDH or PC (Fig. 4C). This revealed that the conversion of both succinate and pyruvate into the TCA cycle are central metabolic checkpoints for ApoE−/− HSPC myeloid lineage specification and to some extent ApoE−/− HSPC expansion. Interestingly, the conversion of succinate to fumarate and pyruvate to oxaloacetate (OAA) converge to the malate-aspartate shuttle, known to maximize the number of ATP molecules produced in glycolysis.41 To address the contribution of this pathway, we suppressed transaminases including glutamate oxaloacetate transaminases (GOTs) using aminooxyacetic acid (AOA). This molecule has been shown to prevent the mitochondria from utilizing glycolytic substrates by inhibiting the malate-aspartate shuttle.42 Remarkably, treatment with AOA not only prevented ApoE−/− HSPC expansion (Fig. 4B) but also their myeloid fate upon IL-3 and GM-CSF treatment (Fig. 4C). These data collectively suggest that mitochondrial metabolic reprogramming of ApoE−/− HSPCs is required for their expansion and commitment to the myeloid lineage.
To bolster our observations, we next generated single or combined knockout of ApoE (ApoE−/−) and Glut1 (Glut1+/−) mice. The increased oxygen consumption observed in ApoE−/− BM cells in response to IL-3 and GM-CSF stimulation was severely reduced by Glut1 deficiency (Fig. 5A). Remarkably, Glut1 deficiency also prevented the enhanced mitochondrial respiration of ApoE−/− Lin−Sca1+ BM progenitors cultured for 2h after isolation (Fig. 5B). Thus, we next isolated Lin− bone marrow cells (containing predominantly HSPCs) that were placed in vitro in medium either alone or supplemented with IL-3 or GM-CSF. We found that Glut1 deficiency led to significantly decreased HSPC expansion either in WT Lin− cultures after IL-3 and GM-CSF stimulation or in ApoE−/− Lin− cultures under both unstimulated and stimulated conditions (Fig. 5C and and5D).5D). As a consequence, Glut1 deficiency prevented not only the expansion of the number of cells per well (Supplemental Fig. IID), but also the generation of CD11b+Gr-1+ myeloid cells both in response to IL-3 and GM-CSF or in ApoE−/− Lin− cultures (Fig. 5E and and5F).5F). This mirrored the ROS production and mitochondrial membrane potential assessed by flow cytometry in HSPCs at the end of the culture period (Fig. 5G and and5H).5H). Mechanistically, we next tested whether Glut1 may mediate the effect of IL-3 on autophagy,43 since autophagy has recently emerged to regulate HSPC maintenance and a bias toward myelopoiesis.44,45 Western blot analysis of LC3-II protein levels, an hallmark of autophagy, revealed that Glut1 deficiency prevented the decrease of LC3-II expression in WT and ApoE−/− BM cells under basal and IL-3 stimulated conditions (Supplemental Fig. IIIA). To analyze the autophagic flux of HSPCs, we next used the Cyto-ID probe allowing analysis by multicolor flow cytometry. Remarkably, Glut1 deficiency prevented the reduced Cyto-ID staining induced by IL-3 in HSPCs isolated from WT Lin− cultures and restored the autophagic flux of ApoE−/− HSPCs to the level of control cells (Supplemental Fig. IIIB and IIIC). These data identify that Glut1 is a key metabolic sensor mediating the growth-regulatory effects of IL-3 through autophagy-dependent modulation of HSPC expansion and myeloid commitment in vitro.
To directly test the in vivo physiological relevance of Glut1 on ApoE−/− HSPCs, we transplanted the BM of single or combined knockout of ApoE (ApoE−/−) and Glut1 (Glut1+/−) into irradiated ApoE−/− mice. After a recovery period, the recipients were placed on a high fat diet (HFD) for 12 weeks using the similar experimental design described earlier (Fig. 6A). At the end of the study period, we confirmed the efficiency of the transplantation procedure by showing an approximately 1.3-fold and 1.5-fold decrease in Glut1 expression in mice receiving Glut1+/− or ApoE−/−Glut1+/− BM, respectively using Glut1 antibody and Glut1 RBD ligand by flow cyotmetry (Fig. 6B). An analysis of the different populations within the HSPCs of these mice was next performed (Supplemental Fig. IV). As shown in Figure 1F, low Glut1 cell surface expression was observed in the CD34− LT-HSCs (Fig. 6C and Supplemental Fig. IVD), which were also characterized by CD150+ and Flt3− markers (Supplemental Fig. IVA–C).35,36 Further analysis of the different populations within the HSPCs revealed higher Glut1 expression in the CD34+CD150+Flt3− multi-potent progenitors (MMP2) compared to the CD34+CD150−Flt3− and CD34+CD150−Flt3+ multi-potent progenitors (MMP3 and MMP4, respectively)(Fig. 6C). Consistent with these findings, the Glut1 cell surface expression was decreased by an approximately 1.4-fold in the MMP2 of mice receiving either Glut1+/− or ApoE−/−Glut1+/− BM (Fig. 6C and Supplemental Fig. IVD). Interestingly, the 2-NBDG staining quantified by flow cytometry suggested Glut1-independent glucose utilization in different populations within the HSPCs, but confirmed an approximately 1.35-fold decrease in 2-NBDG staining in the CD34+CD150+Flt3− MMP2 of mice receiving either Glut1+/− or ApoE−/−Glut1+/− BM compared to their respective controls (Fig. 6D and Supplemental Fig. IVE). We next investigated the relationship between the proliferation capacity of the MMP2 and the Glut1-dependent glucose utilization. Although there was no significant decrease in the S/G2M fraction in the CD34+CD150+Flt3− MMP2 and other populations within the HSPCs of mice receiving Glut1+/− BM, a significant 1.3-fold decrease in the S/G2M fraction was observed in the CD34+CD150+Flt3− MMP2 and downstream CD34+CD150−Flt3− MMP3 of mice receiving ApoE−/−Glut1+/− BM compared to mice receiving ApoE−/− BM (Fig. 6E and Supplemental Fig. IVF). Quantification of the bone marrow HSPCs confirmed an approximately 1.4-fold decrease in the frequency and absolute number of the CD34+CD150+Flt3− MMP2 and downstream MMPs in mice receiving ApoE−/−Glut1+/− BM compared to the controls (Fig. 6F and and6G).6G). Similar findings were also observed in chow-fed ApoE−/−Glut1+/− BM-transplanted mice compared to the ApoE−/− BM-transplanted mice (data not shown). Together, these findings reveal that Glut1-dependent glucose utilization was required for ApoE−/− MMP2 proliferation and downstream MMP expansion.
While working on this manuscript, Pietras et al., elegantly showed that the CD34+CD150+Flt3− MMP2 and downstream CD34+CD150−Flt3− MMP3 exhibited a myeloid-biased multipotential progenitor phenotype.36 This prompted us to test whether the decreased MMP2 and MMP3 expansion observed in mice receiving ApoE−/−Glut1+/− BM could be associated with a defective myeloid fate specification. The common myeloid progenitor (CMP), granulocyte macrophage progenitor (GMP) and megakaryocyte-erythroid progenitor (MEP) populations were analyzed by flow cytometry (Supplemental Fig. IVG) and were not significantly reduced in the mice receiving Glut1+/− BM despite a trend towards CMPs (Fig. 6H). Nevertheless, the CMP numbers were significantly decreased by more than 1.2-fold in mice receiving ApoE−/−Glut1+/− BM compared to controls receiving ApoE−/− BM (Fig. 6H). We noticed that the splenomegaly in the mice receiving the ApoE−/− BM was rescued by Glut1 deficiency (Table S1), and the spleen represents an important reservoir of myeloid cells through extramedullary hematopoiesis in ApoE−/− mice.6,33 Therefore, the hematopoietic progenitors were next quantified in this organ. Similar to the BM, we observed a 1.6-fold decrease in the frequency of splenic CMPs in mice receiving ApoE−/−Glut1+/− BM and, to some extent, a 1.3-fold decrease in the percentage of GMPs, but no changes in the MEP population (Fig. 6I). Consistent with these findings, the platelet and red blood cell counts, mean platelet volume (MPV) and hematocrit were unchanged in these mice (Table SI). Peripheral T− and B-cell numbers were also not affected in these mice (Supplemental Fig. VA and VB). In contrast, the blood counts indicated that the leukocytosis, monocytosis, neutrophilia and eosinophilia of mice receiving ApoE−/− BM in response to feeding a high fat diet were rescued by Glut1 deficiency (Fig. 6J to to6M).6M). These data indicate that Glut1-dependent glucose utilization is required at the early stage of ApoE−/− HSPC commitment to the myeloid lineage.
To test whether this phenotype was caused by cell autonomous effects of Glut1 within the myeloid-biased HSPCs or involved a cell extrinsic effect, we performed a competitive BM transplantation experiment with equally mixed BM cells from CD45.1 ApoE−/− mice and either CD45.2 ApoE−/− BM or CD45.2 ApoE−/−Glut1+/ BM into irradiated WT recipients. After BM reconstitution, we found that the frequency of CD45.1 ApoE−/− HSPCs, particularly the CD34+ HSPCs, were not affected by the presence of CD45.2 ApoE−/−Glut1+/ BM cells, despite the reduced frequency of the CD45.2 ApoE−/−Glut1+/ HSPCs (Fig. 7A). These findings mirrored the reduced S/G2M fraction in the CD45.2 ApoE−/−Glut1+/ HSPCs without altering the S/G2M fraction in the mixed CD45.1 ApoE−/− HSPCs (Fig. 7B). Consistent with these findings on BM HSPCs, there was a preferential accumulation of CD45.1 ApoE−/− vs. CD45.2 ApoE−/−Glut1+/− blood monocytes and neutrophils (Fig. 7C and and7D),7D), indicative of a cell autonomous proliferative disadvantage of Glut1 deficiency.
We next explored the in vivo relevance of reducing ApoE−/− HSPC proliferation and myelopoiesis through Glut1 deficiency on the development of atherosclerosis. This was tested in ApoE−/− recipient mice that received ApoE−/− or ApoE−/−Glut1+/− BM fed a high fat diet for 12 weeks (Fig. 6A). As shown in Table S1, the body weight, plasma LDL and HDL cholesterol or plasma glucose were not significantly different with regard to Glut1 deficiency. However, ApoE−/− mice receiving ApoE−/−Glut1+/− BM showed an approximately 1.4-fold decrease in the development of atherosclerosis in their proximal aortas (Fig. 8A). Immunohistochemical staining of the aortic root plaques revealed that this phenotype was associated with a massive decrease in the F4/80+ macrophages in the ApoE−/−Glut1+/−BM-transplanted mice (Fig. 8B). We also examined the uptake of the radiolabeled D-glucose analog (2-[14C]-DG) after ex vivo incubation of the aortic arch and spleen from the ApoE−/− mice that received ApoE−/− or ApoE−/−Glut1+/− BM. ApoE−/−Glut1+/− BM-transplanted mice showed a significant 1.4-fold and 1.3-fold decrease in total uptake of 2-[14C]-DG in the aortic arch (Fig. 8C) and the spleen (Fig. 8D), respectively compared to controls. We next analyzed Ly6Chi monocyte recruitment into atherosclerotic plaques using fluorescent-labeled latex beads as previously described.10 Fig. 8E reveals than 2 days after monocyte labeling, there was an approximately 1.8-fold decrease in the number of latex+ monocytes in atherosclerotic lesions of ApoE−/− mice receiving ApoE−/−Glut1+/− BM compared to mice receiving ApoE−/− BM. This reduced recruitment was confirmed by analysis of latex+ monocytes in the aortic arch by flow cytometry (Fig. 8F). This paralleled the reduced incorporation of the fluorescent beads in blood monocytes of ApoE−/− mice receiving ApoE−/−Glut1+/− BM compared to controls 2 days after labeling (data not shown) reflecting their reduced monocytosis (Fig. 6K). Thus, we showed that Glut1 connects the enhanced glucose uptake in atheromatous plaques of ApoE−/− mice,27–30 with their myelopoiesis through regulation of HSPC maintenance and myelomonocytic fate.
Previous studies have shown that inflamed atherosclerotic plaques can be visualized by non-invasive PET-CT imaging with 18FDG, a glucose analog, which correlates with macrophage accumulation and inflammation.27–30 However, a recent study has called into question the relevance of these observations, as macrophage-specific overexpression of Glut1 did not aggravate atherosclerosis in mice compared to Glut1 sufficient controls,46 reflecting the need for a better understanding of the underlying mechanisms. Our study provides direct evidence that Glut1 connects the enhanced glucose uptake in atheromatous plaques of ApoE−/− mice with their myelopoiesis through Glut1-dependent regulation of HSPC maintenance and myelomonocytic fate.
Recent studies have suggested that HSPC expansion and the associated myelopoiesis could underlie the hypercholesterolemia-induced atherosclerosis in mice.1,2 However, the rate of ATP generation required for cell proliferation and differentiation cannot be directly explained by cholesterol and requires alternative sources of energy.47 Our observations indicate that the leukocytes and HSPCs from hypercholesterolemic ApoE−/− mice exhibited an increased Glut1-dependent glucose uptake that was associated with increased mitochondrial potential suggesting that the influx of glycolytic metabolites in these cells fuel the mitochondria for oxidative phosphorylation and ATP generation.21 The low expression of Glut1 in CD34− LT-HSCs compared to other CD34+ HSPCs was first counterintuitive, as we initially speculated that the presence of the LT-HSCs in the ‘hypoxic BM niche’ could favor the expression of Glut1 by HIF1α.37 However, increasing glucose metabolism through translocation of Glut1 to the cell surface is thought to be crucial for the active cells and cell cycle entry rather than quiescence.48,49 During aerobic respiration, the ATP yield is linked to NAD+-dependent oxidative steps, including oxidative decarboxylation of pyruvate, that requires metabolic shuttle systems to convey reducing equivalents from cytosol to mitochondria.41 Our findings indicate that both the oxidative decarboxylation of pyruvate and the transamination reactions of the malate-aspartate shuttle were essential for HSPC expansion and commitment to the myeloid lineage. Thus, it is probably not surprising that we did not observe an accumulation of succinate in ApoE−/− BM cultures due to higher succinate dehydrogenase activity favoring fumarate and malate production. The absence of succinate accumulation in ApoE−/− BM cells could also contribute to the lack of HIF1α activation in these cells.50 Consistent with these observations, we did not observe modulation of the oxygenation status of ApoE−/− LT-HSCs in chronic hypercholesterolemia, and HIF1α deficiency in WT or ApoE−/− hematopoietic cells did not alter Glut1 expression or the HSPC frequency. In fact, HIF1α deficiency in hematopoietic cells rather led to increased myeloid expansion, which could contribute to the role of hypoxia in the development of atherosclerosis.51 In contrast, and consistent with the alternative regulation of Glut1 by growth hormone-dependent activation of Ras or Src,38,39 the enhanced Glut1 expression in proliferating ApoE−/− HSPCs was prevented by IL3Rβ blockade. This metabolic regulation was critical for the expansion of ApoE−/− HSPCs and the associated IL-3-dependent downregulation of autophagy,43 which is most likely required to limit intracellular lysosomal degradation and fulfill the high-energy demand of these cells for proliferation.
Very recently, increased splenic activity in patients with cardiovascular diseases has been demonstrated by non-invasive PET-CT imaging with 18FDG,31,32 which could reflect the metabolic activity of extramedullary hematopoiesis required to generates monocytes that infiltrate atherosclerotic plaques.33 However, these observations do not prove causality. Intriguingly, inhibition of glucose uptake with a 2-deoxyglucose (2-DG) analog has recently ben shown to inhibit myelopoiesis in human HSCs,26 but the relevance to atherosclerosis has not been tested. The present study clearly establishes that the increased Glut1-dependent glucose utilization in the ApoE−/− HSPCs could divert these cells to a myelomonocytic fate leading to extramedullary myelopoiesis and subsequent macrophage deposition-dependent atherosclerotic plaque formation. Indeed, we now provide direct in vivo evidence that Glut1 deficiency can significantly reduce the number of CMPs in the ApoE−/− BM as well as the number of CMPs and GMPs in the ApoE−/− spleen and this was associated with reduced splenic glucose uptake. This led to inhibition of the monocytosis, neutrophilia and eosinophilia in the ApoE−/− mice transplanted with the ApoE−/− BM. Consistent with the lack of effect of Glut1 deficiency on resting T-cells,51 we also did not observe variations in the number of lymphocytes in our models. While working on this manuscript, Pietras et al., elegantly showed that the CD34+CD150+Flt3− MMP2, which is now showed to express the most Glut1, and downstream CD34+CD150−Flt3− MMP3 exhibited a myeloid-biased multipotential progenitor phenotype,36 offering an alternative explanation to the role of Glut1 in favoring myelopoiesis that is independent of a change in other lineage commitments. Together, these findings reveal that the mechanism by which defective ApoE-dependent cholesterol efflux pathways skew hematopoietic stem cells towards myelopoiesis,5,6 relies on the regulation of Glut1-dependent glucose uptake by the IL-3Rβ signaling pathway.
The metabolic phenotype of ApoE−/− HSPCs outlined here could be relevant to the adaptability of HSPCs to cholesterol overload and may indicate that the glycolytic phenotype of HSPCs is not merely a product of their hypoxic environment. Thus, the existence of different molecular mechanisms underlying the different glycolytic phenotypes in HSPCs may suggest strategies for specifically modulating the pool of HSPCs that are committed to the myeloid lineage under stressed conditions, such as in myeloproliferative disorders,36 sepsis,53 myocardium infarction,54 or chronic atherosclerosis, as shown in the present study. Inhibition of glucose uptake by a Glut1 inhibitor that does not cross the blood-brain barrier could ultimately provide a novel therapeutic approach to prevent myelopoiesis-driven diseases such as atherosclerosis.
The enhanced metabolic activity visualized by non-invasive PET-CT imaging with 18FDG, a glucose analog, in inflamed atherosclerotic plaques and spleen of patients with cardiovascular diseases suggests a link between hematopoietic activity and atherosclerosis. We evaluated this hypothesis by investigating the contribution of Glut1 in the hematopoietic compartment to the development of atherosclerosis in ApoE−/− mice. We found that hematopoietic Glut1 deficiency decreased atherosclerosis by preventing hematopoietic stem and progenitor cell proliferation, myelopoiesis and the recruitment of myeloid cells in atherosclerotic lesions independent of plasma lipid profile. In ApoE−/− hematopoietic stem and progenitor cells, Glut1 serves as a key metabolic sensor for the high-energy demand of these cells for proliferation favoring glycolytic substrate utilization by mitochondria. These results provide direct evidence showing that 1) Glut1 connects the enhanced glucose uptake in atheromatous plaques and spleen of ApoE−/− mice with their myelopoiesis and 2) the activation of Glut1 in hematopoietic stem and progenitor cells of preclinical model of atherosclerosis is proatherogenic. Thus, inhibition of glucose uptake by a Glut1 inhibitor that does not cross the blood-brain barrier may be useful in the treatment of atherosclerosis.
We thank Dr. Fréderic Labret for assistance with flow cytometry and Dr. Véronique Corcelle for assistance in animal facilities.
SOURCES OF FUNDING
This work was supported by grants to L.Y.C from INSERM ATIP-AVENIR, the Fondation de France (201300038585) and Agence Nationale de la Recherche (ANR).
Animal Models of Human Disease
Lipids and Cholesterol
The authors have declared that no conflict of interest exists.