ABSTRACT
Oxygen is a key molecule for efficient energy production in living organisms. Although aerobic organisms have adaptive processes to survive in low-oxygen environments, it is poorly understood how lipolysis, the first step of energy production from stored lipid metabolites, would be modulated during hypoxia. Here, we demonstrate that fasting-induced lipolysis is downregulated by hypoxia through the hypoxia-inducible factor (HIF) signaling pathway. In Caenorhabditis elegans and mammalian adipocytes, hypoxia suppressed protein kinase A (PKA)-stimulated lipolysis, which is evolutionarily well conserved. During hypoxia, the levels of PKA activity and adipose triglyceride lipase (ATGL) protein were downregulated, resulting in attenuated fasting-induced lipolysis. In worms, HIF stabilization was sufficient to moderate the suppressive effect of hypoxia on lipolysis through ATGL and PKA inhibition. These data suggest that HIF activation under hypoxia plays key roles in the suppression of lipolysis, which might preserve energy resources in both C. elegans and mammalian adipocytes.
INTRODUCTION
Regulation of lipid metabolism is crucial for the maintenance of energy homeostasis. In most organisms, excess energy resources are stored in the form of triglycerides in lipid droplets. In order to utilize stored lipid metabolites, triglycerides need to be hydrolyzed into fatty acids and glycerol, a process called lipolysis. For energy balance, lipolysis is sophisticatedly modulated by various hormones in response to nutritional status through coordinated actions of lipases and lipid droplet-associated proteins (1–4). Given that dysregulated lipolysis and subsequent accumulation of lipid metabolites can exert detrimental effects, such as lipotoxicity, the precise regulation of lipolysis is crucial throughout evolution (5, 6).
Protein kinase A (PKA) signaling plays an essential role in the initiation of lipolysis in eukaryotic cells. In the basal state, the lipid droplet-binding protein perilipin 1 (PLIN1) covers the surface of lipid droplets and promotes lipid accumulation in the lipid droplets in mammalian adipocytes (7). Upon hormonal stimulation, activated PKA induces the phosphorylation and activation of lipases and lipid droplet proteins, thus stimulating lipolysis. For example, PKA phosphorylates PLIN1 to recruit hormone-sensitive lipase (HSL) to lipid droplets (8, 9) and to release α/β-hydrolase domain-containing 5 (ABHD5; CGI-58), a coactivator of adipose triglyceride lipase (ATGL) (10–12). ATGL is a major and primary lipase to break down triglycerides in most organisms (13–16). In the absence of ATGL, lipolysis does not occur properly despite the activation of PKA signaling, leading to massive triglyceride accumulation (17).
Oxygen is an essential factor in lipid oxidation in eukaryotic cells because it acts as an electron acceptor in mitochondria for efficient ATP production. Given the importance of the availability of oxygen, it is not surprising that living organisms have developed adaptive responses for low-oxygen environments (18–20). One of the key factors that mediate adaptive responses under low-oxygen levels is hypoxia-inducible factor (HIF). HIF is a heterodimeric DNA-binding protein complex composed of HIFα (HIF1α, HIF2α, and HIF3α in mammals) and HIF1β (ARNT). The level of HIFα protein is strictly controlled by the oxygen concentration (21, 22). Under normoxic conditions, HIFα is hydroxylated at the proline residues by prolyl hydroxylase domain proteins (PHD1, PHD2, and PHD3) or the asparagine residue by factor inhibiting HIF (FIH) and polyubiquitinated by von Hippel-Lindau (VHL) tumor suppressor protein, an E3 ubiquitin ligase complex, leading to proteasomal degradation (23–26). However, under hypoxic conditions, PHD activity is inhibited (27), and the level of HIFα protein is upregulated without degradation. Thus, HIF stimulates the expression of its target genes, which is required for survival under low-oxygen conditions (22, 28–30).
The process of energy metabolism needs to be fine-tuned to survive under hypoxia. In mitochondria, the anaerobic catabolic process is stimulated instead of oxidative phosphorylation (OXPHOS). During hypoxia, activation of the HIF pathway modulates energy metabolism toward increased glycolysis and reduced OXPHOS by promoting the expression of glycolytic enzymes (31, 32). It has been suggested that hypoxia would exert a negative effect on lipolytic activity (33–36). In contrast, several studies have shown that lipolysis can be induced under mild hypoxic conditions (5 to 10% O2) (37, 38). However, it is largely unknown how hypoxia modulates lipolysis at the molecular level.
As an excellent genetic model organism, Caenorhabditis elegans has well-conserved mechanisms in response to various stress conditions, including hypoxia (20, 39–41). The C. elegans genome has hif-1, vhl-1, and egl-9, which encode proteins homologous to the mammalian HIFα subunit, VHL, and PHD, respectively. In worms, hif-1 regulates the expression of various target genes and is involved in the adaptation of energy metabolism to hypoxic conditions (28, 42–45). In natural environments, the soil nematode C. elegans is often exposed to hypoxic conditions when it rains and the soil becomes saturated with water (46–48). During the course of soil hardening after rain, soil aggregation can block the oxygen pores, resulting sometimes in a long-lasting hypoxic situation. Thus, C. elegans has to develop systems to adapt to hypoxic conditions, accompanied by lowering the metabolic rate and oxygen consumption, for efficient use of energy to survive (19).
In this study, we demonstrate that hypoxia and HIF suppress PKA-mediated lipolysis in an evolutionarily conserved manner. Using C. elegans, we delineated the genetic components that mediate the antilipolytic effect of hypoxia. Stabilization of HIF protein was sufficient to suppress fasting-induced lipolysis, with reduced PKA activity and ATGL protein level. In mammalian adipocytes, hypoxia also had an antilipolytic effect. Elevation of HIF1α or HIF2α proteins in adipocytes downregulated ATGL protein and decreased PKA-induced lipolysis. Collectively, our data suggest that HIFα is a key factor that suppresses ATGL and lipolysis under hypoxia, which would provide a negative feedback mechanism to prevent futile lipolysis when the oxygen supply is limited.
RESULTS
Hypoxia suppresses fasting-induced lipolysis in C. elegans.C. elegans is an effective model organism to study lipid metabolism because its metabolic pathways are well conserved, and lipid droplets in the intestine are easily detectable (49). We have previously reported that fasting induces a significant decrease in intestinal lipid contents in C. elegans (16, 50). In an investigation of environmental factors that can modulate fasting-induced lipolysis, in the current study, we found that exposure of worms to hypoxia (1% O2) substantially attenuated the decrease in oil red O (ORO)-stained lipid droplets upon fasting (Fig. 1A and B). In C. elegans, a 1% O2 hypoxic condition was sufficient to stimulate the well-known hypoxia marker nhr-57 (42) (Fig. 1C). Interestingly, soaking of worms in isotonic (M9) buffer, which we used to mimic drowning caused by rain in natural environments, also induced significant blockage of lipid droplet breakdown in the fasted state. However, when the tubes containing worms in M9 buffer were shaken to provide aeration, the suppression of fasting-induced lipolysis was restored to a degree comparable to that in the normoxia fasting state (Fig. 1A and B).
Hypoxia suppresses fasting-induced lipolysis in Caenorhabditis elegans. (A and B) Young adult worms were placed in a hypoxic chamber or 1.5-ml tube filled with isotonic (M9) buffer. Representative images and quantitation data of oil red O (ORO) staining in young adult worms under feeding and at 8 h of fasting under hypoxic conditions (incubated in a 1% O2 hypoxia chamber or drowned with or without aeration by shaking). Anterior intestinal parts were subjected to quantification of ORO staining. Scale bars, 50 μm. (C) Confocal microscopic images of iaIs7[nhr-57p::gfp; unc-119(+)] fed and fasted (4 h) worms under hypoxic conditions (1% O2, 6 h). Scale bars, 50 μm. (D) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] fed and fasted (4 h) young adult worms under hypoxia (1% O2, 6 h). Scale bars, 20 μm. (E) mRNA levels of atgl-1 and fasting-responsive genes such as fil-1 and cpt-3 were measured by quantitative reverse transcription-PCR and normalized to the level of act-1/act-3 mRNA. Data are expressed as means plus SD (**, P < 0.01).
Next, we examined the level of ATGL-1, the major lipase that mediates fasting-induced lipolysis, using a green fluorescent protein (GFP)-fusion reporter system. Under hypoxic conditions, ATGL-1::GFP signal was decreased in well-fed worms and was not increased by fasting, in contrast to results under the normoxic condition (Fig. 1D). In contrast, the atgl-1 mRNA level was not reduced by hypoxia, implying that ATGL-1 would be regulated at the posttranscriptional level under this condition (Fig. 1E). However, the expression of other fasting-responsive genes, such as fil-1 and cpt-3, was downregulated by hypoxia. These results suggest that reduced oxygen availability could suppress fasting-induced lipolysis as well as ATGL-1 protein in C. elegans.
Hypoxia inhibits PKA-mediated lipolysis in C. elegans.It has been well established that various fasting cues activate PKA to promote lipolysis (2). To investigate whether PKA-dependent lipolysis might be affected by hypoxia, we tested kin-2(ce179) mutant worms as a PKA-hyperactive mutant model. kin-2(ce179) worms have an R92C mutation in the pseudosubstrate domain of the PKA regulatory subunit KIN-2. Consistent with a previous report (51), the level of neutral lipid metabolites was decreased in kin-2(ce179) worms under normoxia (Fig. 2A and B). However, hypoxia prevented the PKA-induced decrease in neutral lipid accumulation in both wild-type and kin-2(ce179) mutant worms. We next examined whether ATGL-1 might be altered by hypoxia in PKA-activated worms. To activate PKA in worms, kin-2 gene expression was suppressed via RNA interference (kin-2 RNAi). PKA activation by kin-2 RNAi enhanced ATGL-1::GFP levels under normoxia. However, in hypoxic worms, kin-2 suppression did not induce ATGL-1::GFP (Fig. 2C). Moreover, in kin-2(ce179) worms, increases in fil-1 and cpt-3 mRNA expression were repressed by hypoxia (Fig. 2D). These data suggest that PKA-mediated lipolysis in C. elegans would be inhibited by hypoxia.
Hypoxia inhibits PKA-induced lipolysis in C. elegans. (A and B) Representative images and quantitation data of ORO staining in a PKA-hyperactive kin-2 mutant strain (ce179 strain) after hypoxia (1% O2, 8 h). Marked areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. (C) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] young adult worms after control (L4440; con) RNAi or kin-2 RNAi under hypoxia (1% O2, 6 h). Scale bars, 20 μm. (D) mRNA levels of atgl-1 and fasting-responsive genes such as fil-1 and cpt-3 were measured by quantitative reverse transcription-PCR and normalized to the level of act-1/act-3 mRNA. Data are expressed as means plus SD (*, P < 0.05; **, P < 0.01).
Hypoxia attenuates stimulated lipolysis in mammalian adipocytes.To test the hypothesis that the antilipolytic effect of hypoxia might be conserved in mammalian adipocytes, we examined lipolytic activity in 3T3-L1 adipocytes upon hypoxia. When the lipolytic activity was determined by glycerol release, the effects of the PKA-activating drugs isoproterenol (ISO; β-adrenergic agonist), 3-isobutyl-1-methylxanthine (IBMX; phosphodiesterase inhibitor), and forskolin (FSK; adenylyl cyclase activator) on lipolysis were dampened under hypoxic conditions (Fig. 3A). In accordance with these, the changes in lipid droplet size and morphology induced by PKA activation were less dramatic in hypoxic adipocytes (Fig. 3B). To elucidate which factor(s) in the lipolytic pathway might be affected by hypoxia, we investigated lipolysis-related proteins. Hypoxia decreased the level of ATGL protein in both the basal and the ISO-stimulated states (Fig. 3C and D). However, the levels of Atgl mRNA were not altered by hypoxia (Fig. 3E). To test whether hypoxic adipocytes might have an altered PKA signaling cascade, we examined the phosphorylation level of PKA-downstream target proteins. Upon hypoxia, the levels of pHSL and pPKA substrate were downregulated by PKA activators (Fig. 3F). Then, to examine whether overall PKA signaling would be decreased by hypoxia, cellular cAMP levels were determined under hypoxic conditions in the absence or presence of PKA-activating chemicals (Fig. 3G). In hypoxic adipocytes, cAMP levels were decreased in ISO- and IBMX-treated cells but not in FSK-treated cells, implying that the biochemical process of cAMP production, but not adenylyl cyclase itself, might be impeded by hypoxia. Similarly, in C. elegans, the level of fasting-induced cAMP was downregulated by hypoxia (Fig. 3H). Furthermore, in adipocytes the mRNA levels of β3-adrenergic receptor (Adrb3) and adenylyl cyclase 6 (Adcy6) were decreased by hypoxia (Fig. 3I). These data indicate that hypoxia could reduce cellular cAMP production and decrease lipolytic activity, leading to a decrease in PKA activity.
Hypoxia attenuates stimulated lipolysis in mammalian adipocytes. (A) Glycerol concentration in culture medium from differentiated adipocytes. 3T3-L1 adipocytes were treated with ISO (1 μM), FSK (10 μM), or IBMX (520 μM) after hypoxia (1% O2, 6 h). (B) Representative images of differentiated adipocytes upon treatment with ISO (1 μM, 9 h) under hypoxia (1% O2, 24 h). BODIPY 493/503 staining after fixation is shown. (C) Western blotting of lipolysis-related proteins in differentiated adipocytes upon treatment with ISO (1 μM, 1 h) after hypoxia (1% O2, 6 h). β-Actin protein was used as a loading control. (D) Representative images of immunocytochemistry (ICC) analysis of PLIN1 (green) and ATGL (red) with ISO (1 μM, 9 h) under hypoxia (1% O2, 24 h). (E) mRNA levels in 3T3-L1 adipocytes with ISO (1 μM, 3 h) under hypoxic conditions (1% O2, 8 h). mRNA levels were normalized to the level of cyclophilin mRNA. (F) Western blotting of lipolysis-related proteins in differentiated adipocytes with ISO (1 μM, 1 h), FSK (10 μM, 1 h), IBMX (520 μM, 1 h), or db-cAMP (0.5 mM, 1 h) after hypoxia (1% O2, 6 h). β-Actin protein was used as a loading control. (G) Intracellular cAMP levels were measured in adipocytes. Differentiated adipocytes were treated with ISO (1 μM), FSK (10 μM), or IBMX (520 μM) for 15 min after hypoxia (1% O2, 6 h). (H) Intracellular cAMP levels were measured in C. elegans after 4 h of fasting under hypoxic conditions (1% O2, 8 h). (I) mRNA levels in 3T3-L1 adipocytes under hypoxic conditions (1% O2, 8 h). mRNA levels were normalized to the level of 36b4 mRNA. Data are expressed as means + SD (*, P < 0.05; **, P < 0.01).
To test whether downregulation of the cAMP level is the only mechanism to regulate lipolysis by hypoxia, we examined lipolytic activity under hypoxia in PKA-activated adipocytes. Previously, we revealed that suppression of the PKA regulatory subunits Iα and IIβ in adipocytes results in increased PKA activity and potentiated lipolysis without cAMP elevation (51). By using the above-mentioned PKA-activated adipocyte model without cAMP alteration, we determined whether hypoxia represses lipolysis in the PKA-downstream cascade. As shown in Fig. 4A, suppression of the PKA regulatory subunits Iα and IIβ by small interfering RNAs (siRNAs; siRIα and siRIIβ, respectively) led to PKA activation in adipocytes, resulting in increased glycerol release via stimulated lipolysis. However, the induction of glycerol release and changes in lipid droplet morphology in PKA-activated adipocytes were attenuated under the hypoxic condition. (Fig. 4A and B). The levels of lipolysis-related proteins such as ATGL, pHSL, and pPKA substrates were reduced by hypoxia (Fig. 4C and D). Collectively, these data suggest that suppression of PKA-stimulated lipolysis by hypoxia cannot be explained merely by reduced cAMP production. Rather, it would be more plausible to conclude that hypoxia might have an additional mechanism(s) to repress the phosphorylation of PKA target proteins and the ATGL protein level.
Hypoxia suppresses PKA-induced lipolysis in mammalian adipocytes. (A) Glycerol concentration in culture medium from differentiated adipocytes 24 h after siRNA transfection with siRIα and siRIIβ. After siRNA transfection, cells were subjected to hypoxia (1% O2, 6 h). (B) Representative images of adipocytes after siRNA transfection with siRIα and siRIIβ. After siRNA transfection, cells were subjected to hypoxia (1% O2, 24 h). BODIPY 493/503 was used for staining after fixation. siNC, siRNA targeting the negative control. (C) Western blotting of lipolysis-related proteins in differentiated adipocytes with siRNA transfection. Hypoxic conditions (1% O2, 6 h) were induced after siRNA transfection. β-Actin protein was used as a loading control. (D) Representative images of ICC analysis of PLIN1 (green) and ATGL (red) at 48 h after siRNA transfection with siRIα and siRIIβ. The hypoxic condition (1% O2, 6 h) was induced after siRNA transfection. Data are expressed as means plus SD (*, P < 0.05; **, P < 0.01).
HIF-1 is crucial to suppress fasting-induced lipolysis in C. elegans.Among several factors that mediate hypoxic responses, HIF-1 is a well-known transcription factor that regulates target genes involved in energy metabolism. To determine whether HIF-1 might be involved in the antilipolytic effect of hypoxia, we examined the hif-1 negative-regulator mutants, such as egl-9(sa307) and vhl-1(ok161). When we analyzed ORO staining in vhl-1(ok161) and egl-9(sa307) mutants under feeding and fasting conditions (Fig. 5A to D), both HIF-1-enhanced mutant worms revealed decreased fasting-induced lipolysis. On the other hand, double mutant worms with hif-1 mutation, such as vhl-1(ok161); hif-1(ia04) and egl-9(sa307); hif-1(ia04), maintained fasting-induced lipolysis to the levels of wild-type N2 fasted worms (Fig. 5A to D). Compared to levels with an RNAi control, the level of ATGL-1::GFP protein was downregulated by vhl-1 RNAi (Fig. 5E). Moreover, the mRNA level of the fatty acid oxidation-related gene cpt-3 was decreased in vhl-1(ok161) and egl-9(sa307) mutants upon fasting, whereas vhl-1(ok161); hif-1(ia04) and egl-9(sa307); hif-1(ia04) double mutants showed normal cpt-3 induction (Fig. 5F). Similarly, worms expressing a hydroxylation-deficient, constitutively active form of HIF-1 [iaIs34; hif-1::hif-1(P621G)::myc] showed blunted fasting-induced lipolysis and fasting-induced gene expression (Fig. 5G to I). These data suggest that elevated HIF would be a key factor for reduced lipolysis in hypoxic worms.
HIF-1 is sufficient and necessary to suppress fasting-induced lipolysis in C. elegans. (A to D) Representative images and quantitation data of ORO staining in fed and fasted (8 h) young adult worms of the indicated strains. Marked areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. (E) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] fed and fasted (4 h) worms after control (L4440) RNAi or vhl-1 RNAi. Scale bars, 20 μm. (F) mRNA level of cpt-3 was measured by quantitative reverse transcription-PCR and normalized to the level of act-1/act-3 mRNA. (G and H) Representative images and quantitation data of ORO staining in fed and fasted (8 h) young adult N2 and iaIs34[hif-1p::hif-1a (P621G)::tag + unc-119(+)] worms. Marked areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. (I) mRNA level of cpt-3 was measured by quantitative reverse transcription-PCR and normalized to the level of act-1/act-3 mRNA. (J and K) Representative images and quantitation data of ORO staining in fed and fasted (8 h) young adult N2 and hif-1(ia04) worms under hypoxia (1% O2, 8 h). Scale bars, 100 μm. (L) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] fed and fasted (4 h) worms under hypoxia (1% O2, 8 h) after control (L4440) RNAi or hif-1 RNAi. Scale bars, 20 μm. (M) mRNA level of cpt-3 was measured by quantitative reverse transcription-PCR and normalized to the level of act-1/act-3 mRNA. Data are expressed as means plus SD (*, P < 0.05; **, P < 0.01; n.s., P > 0.05).
Next, we asked whether lipolysis might be altered in the absence of HIF-1. The hif-1 loss-of-function mutant hif-1(ia04) was subjected to ORO staining upon hypoxia (Fig. 5J and K). Fasted hif-1(ia04) mutant worms showed reduced lipid droplets comparable to levels in fasted wild-type N2 worms under normoxia. However, unlike wild-type worms, hif-1(ia04) worms retained active fasting-induced lipolysis upon hypoxia (Fig. 5J and K), implying that HIF-1 would be essential for the antilipolytic effect of hypoxia. In accordance with these observations, hypoxia did not suppress the increased ATGL-1::GFP level induced by fasting with hif-1 RNAi (Fig. 5L). Further, the level of cpt-3 mRNA, which was attenuated by hypoxia, was partially rescued in the hif-1(ia04) mutant (Fig. 5M). Together, these genetic data clearly suggest that HIF-1 would be both sufficient and necessary for hypoxia-induced antilipolysis in C. elegans.
HIF-1 inhibits PKA activity in C. elegans.To test whether increased HIF-1 might repress PKA activity, we performed an in vitro PKA activity assay with total protein extracts from HIF-1-enhanced mutant worms. As shown in Fig. 6A and B, PKA enzymatic activities in vhl-1(ok161) and egl-9(sa307) mutant worms were partially but significantly decreased compared to the level in the wild-type control. In addition, the levels of dibutyryl-cAMP (db-cAMP)-stimulated PKA activity were reduced in both mutant worms compared to the level in the wild-type control. In contrast, HIF-defective vhl-1(ok161); hif-1(ia04) and egl-9(sa307); hif-1(ia04) mutant worms recovered basal and stimulated PKA activities (Fig. 6A and B), indicating that hif-1 would play a crucial role in the decrease in PKA activity in vhl-1(ok161) and egl-9(sa307) mutant worms. Moreover, when we introduced vhl-1 RNAi to stabilize HIF-1 protein in PKA-hyperactive kin-2(ce179) mutant worms, the reduced level of ORO-positive lipid droplets was reversed (Fig. 6C and D). Moreover, while vhl-1(ok161) mutant worms with kin-2 RNAi were resistant to PKA-mediated lipolysis, vhl-1(ok161); hif-1(ia04) double mutants were sensitive to kin-2 RNAi (Fig. 6E and F). Together, these data indicate that an increased level of HIF-1 could suppress PKA enzymatic activity and PKA-mediated lipolysis in C. elegans.
HIF-1 restrains PKA activity in C. elegans. (A and B) PKA activity assays using the Kemptide substrate and total protein extracts obtained from young adult worms of the indicated strains in the absence or presence of dibutyryl-cAMP (db-cAMP). (C and D) Representative images and quantitation data of ORO staining in a PKA-hyperactive kin-2 mutant strain (ce179 strain) after control (L4440) RNAi or vhl-1 RNAi. Marked areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. (E and F) Representative images and quantitation data of ORO staining in N2, vhl-1(ok161), and vhl-1(ok161); hif-1(ia04) worms with control (L4440) RNAi or kin-2 RNAi. Boxed areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. Data are expressed as means plus SD (*, P <0.05; **, P <0.01).
Hypoxia-induced HIF1 downregulates ATGL-1 protein.In C. elegans, ATGL-1 serves as the major lipase in fasting-induced lipolysis (16). ATGL-1-overexpressing worms (hjIs67) exhibited reduced lipid contents (Fig. 7A and B). However, in the hypoxic state, the degree of lipid storage in ATGL-1-overexpressing worms was partially increased. Similar to hypoxia, genetic induction of HIF-1 via vhl-1 RNAi blocked the effect of ATGL-1 overexpression on lipid storage (Fig. 7C and D). Because the protein level of ATGL-1 was reduced by hypoxia (Fig. 1D), we hypothesized that hypoxia could regulate ATGL-1 protein, probably through protein stability control. In addition, we have previously shown that ATGL-1 protein stability can be modulated by PKA activation (16). Under hypoxia or an HIF-1-inducing condition, such as vhl-1 knockdown, ATGL-1::GFP signal was decreased under both feeding and fasting conditions (Fig. 7E and F). However, when worms were pretreated with the proteasome inhibitor MG132, the level of ATGL-1 protein was rescued under both the hypoxic and HIF-1-induced conditions. These data suggest that proteasomal degradation of ATGL (ATGL-1) might be regulated by the hypoxia-HIF-1 pathway.
Regulation of ATGL-1 protein is mediated by HIF-1. (A and B) Representative images and quantitation data of ORO staining in N2 and hjIs67[atgl-1p::atgl-1::gfp] worms under hypoxic conditions (1% O2, 8 h). (C and D) Representative images and quantitation data of ORO staining in N2 and hjIs67[atgl-1p::atgl-1::gfp] worms after control (L4440) RNAi or vhl-1 RNAi. For ORO staining, boxed areas were subjected to quantitation of ORO staining. Scale bars, 100 μm. (E) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] fed and fasted (4 h) worms under hypoxic conditions (1% O2, 6 h). MG132 (100 μM) was used for pretreatment for 1 h before hypoxia. (F) Confocal microscopic images of hjIs67[atgl-1p::atgl-1::gfp] fed and fasted (4 h) worms after control (L4440) RNAi or vhl-1 RNAi. MG132 (100 μM) was used for treatment under the fasting condition. Scale bar, 20 μm. Data are expressed as means plus SD (**, P <0.01).
Mammalian HIFα inhibits PKA-induced lipolysis through the regulation of ATGL protein and PKA activity.To investigate whether the inhibitory role of HIFα in stimulated lipolysis might be evolutionarily conserved, we examined the levels of glycerol release upon HIFα induction in mammalian adipocytes. When HIF1α or HIF2α was overexpressed using adenovirus (Ad-HIF1α and Ad-HIF2α, respectively) in differentiated adipocytes, elevated HIF1α and HIF2α levels partially suppressed PKA-induced lipolysis (Fig. 8A). In addition, adenoviral overexpression of HIF1α or HIF2α decreased the levels of ATGL and PLIN1 protein compared to the level in the mock-infected (Ad-Mock) group (Fig. 8B and C). Moreover, the phosphorylation levels of HSL and PKA substrates were downregulated by Ad-HIF1α and Ad-HIF2α. In accordance with the results in adenoviral overexpression experiments, the levels of PKA-induced lipolysis and ATGL protein were partially attenuated by cobalt chloride (CoCl2; PHD inhibitor) (Fig. 8D and E). When the levels of cellular cAMP were measured, HIF1α- or HIF2α-overexpressing adipocytes showed levels of cAMP similar to those in control (Ad-Mock) adipocytes (Fig. 8F). Next, to investigate whether HIF might modulate the downstream steps of cAMP production during PKA signaling cascade, HIF1α was overexpressed in PKA-activated adipocytes. As shown in Fig. 8G and H, overexpression of HIF1α downregulated the levels of lipolysis, ATGL protein, and PKA target phosphorylation, indicating that HIF1α would suppress PKA activity, independent of cAMP. Moreover, to test whether transcriptional activity of HIF might be involved in these, we examined the effects of two different truncated forms of HIF1α with deletion in the N-terminal transactivation domain (ΔNAD) or C-terminal transactivation domain (ΔCAD). Compared to wild-type HIF1α overexpression, HIF1α ΔNAD overexpression, at least partly, restored the phosphorylation levels of HSL and PKA substrate in the presence of ISO (Fig. 8I). Also, the level of ATGL protein was slightly rescued by overexpression of the truncated forms of HIF1α with ISO treatment. To verify whether HIFα mediates antilipolytic activity upon hypoxia, HIF1α or HIF2α expression was suppressed by an siRNA in adipocytes, and lipolytic activities were determined under the hypoxic condition. As shown in Fig. 8J, the suppression of HIF1α or HIF2α in adipocytes slightly but substantially relieved the antilipolytic effect of hypoxia. Then, to test whether ATGL protein might be regulated by the proteasomal degradation pathway, adipocytes were treated with MG132. As shown in Fig. 8K, hypoxia-induced ATGL downregulation was reversed by treatment with MG132. These data imply that HIFα would be one of the key factors that can reduce the ATGL protein level and PKA activity for decreased lipolysis during hypoxia.
Mammalian HIFα inhibits PKA-induced lipolysis through ATGL regulation. (A) Glycerol concentration in culture medium from differentiated adipocytes treated with ISO (1 μM) after 72 h of infection with adenovirus expressing GFP (Ad-Mock), HIF1α (Ad-HIF1α), or HIF2α (Ad-HIF2α). (B and C) Western blotting in differentiated adipocytes treated with ISO (1 μM, 1 h) after 72 h of infection with adenovirus expressing GFP (Ad-Mock), HIF1α (Ad-HIF1α), or HIF2α (Ad-HIF2α). β-Actin protein was used as a loading control. (D) Glycerol concentration in culture medium from differentiated adipocytes. 3T3-L1 adipocytes were treated with ISO (1 μM) after CoCl2 treatment (200 μM, 24 h). (E) Western blotting in differentiated adipocytes with ISO (1 μM, 1 h) after CoCl2 treatment (200 μM, 24 h). (F) Intracellular cAMP levels were measured in adipocytes after 72 h of infection with adenovirus expressing GFP (Ad-Mock), HIF1α (Ad-HIF1α), or HIF2α (Ad-HIF2α). Adipocytes were treated with ISO (1 μM) for 15 min. (G) Glycerol concentration in culture medium from differentiated adipocytes 24 h after siRIα and siRIIβ transfection. Adipocytes were infected with adenovirus expressing GFP (Ad-Mock) or HIF1α (Ad-HIF1α) 48 h before transfection. (H) Western blotting in differentiated adipocytes 24 h after siRIα and siRIIβ transfection. Adipocytes were infected with adenovirus expressing GFP (Ad-Mock) or HIF1α (Ad-HIF1α) 48 h before transfection. β-Actin protein was used as a loading control. (I) Western blotting in differentiated adipocytes treated with ISO (1 μM, 1 h) 48 h after transfection of HA-HIF1α, HA-HIF1α ΔNAD, or HA-HIF1α ΔCAD. β-Actin protein was used as a loading control. (J) Glycerol concentration in culture medium from differentiated adipocytes in the absence or presence of ISO (1 μM) under hypoxia (1% O2, 6 h). siRNA was transfected 48 h before glycerol release was measured. (K) Western blotting in differentiated adipocytes upon hypoxia (1% O2, 6 h). Adipocytes were pretreated with MG132 (20 μM) 1 h before hypoxia. β-Actin protein was used as a loading control. Data are expressed as means plus SD (*, P < 0.05; **, P < 0.01; n.s., P > 0.05).
DISCUSSION
Oxygen availability is important in the regulation of energy homeostasis in that oxygen is an essential molecule for efficient energy production from lipid metabolites. For adaptation to hypoxic environments, precise modulation of lipolysis is required to protect organisms against detrimental effects. Our findings in this study provide genetic evidence that hypoxia suppresses lipolysis through the HIF pathway, accompanied by downregulation of PKA activity and ATGL protein levels. This regulatory mechanism seems to be evolutionarily well conserved in both C. elegans and mammalian adipocytes.
Several decades ago, it was reported that hypoxia appears to block lipolysis in animals (33, 34, 52). However, the mechanism underlying the antilipolytic effect of hypoxia was not thoroughly understood. Here, several lines of evidence suggested that HIF would play key roles in mediating hypoxia-induced antilipolytic activity. First, either hypoxia or genetic ablation of the HIF negative regulators egl-9 and vhl-1 resulted in a decrease in lipolytic activity in worms (Fig. 1 and 5). Second, loss of function of the hif-1 gene in C. elegans prevented the effects of hypoxia on lipolysis (Fig. 5). Third, knockdown of HIF negative-regulator genes alleviated antilipolytic activity upon PKA activation (Fig. 6). Last, HIFα overexpression attenuated stimulated lipolysis in adipocytes. As overexpression of HIF1α with transactivation domain deletion partly attenuated PKA suppression, it is possible to speculate that transcriptional activity of HIF1α would be important to regulate PKA activity. In contrast, knockdown of HIFα in adipocytes partially rescued the antilipolytic effect of hypoxia (Fig. 8). In accordance, it has been reported that stimulated lipolysis in adipose tissue was suppressed in an adipocyte-specific PHD2 knockout mouse model (53). Collectively, these in vivo and in vitro data suggest that HIF has crucial roles in the regulation of lipolysis upon hypoxia.
PKA is a cAMP-dependent protein kinase that is important for lipolysis. In adipocytes, PKA activity is controlled by various signals that stimulate cAMP production or degradation (2–4). Current data suggest that decreased PKA activity upon hypoxia could be associated, at least partially, with decreased cAMP levels. In adipocytes, hypoxic conditions downregulated the cAMP level in both the basal and the stimulated states (Fig. 3). Unlike other PKA-stimulating drugs, FSK could overcome the reduction in cellular cAMP upon hypoxia, implying that the enzymatic activity of adenylyl cyclase might be inhibited under hypoxic conditions. Consistent with these findings, fasting-induced cAMP was diminished in C. elegans upon hypoxia (Fig. 3). Furthermore, the mRNA levels of Adrb3 and Adcy6 were decreased by hypoxia. These results indicate that hypoxia could attenuate the stimulation of lipolysis, probably through downregulation of the cAMP level, thereby reducing PKA activity. In our experiment, however, HIF1α or HIF2α overexpression in adipocytes did not significantly affect the cellular level of cAMP (Fig. 8). In addition, when PKA was activated by db-cAMP treatment or PKA-regulatory subunit knockdown, the levels of pHSL or pPKA substrates were still downregulated by hypoxia (Fig. 3). These data imply that modulating the cAMP level may not be the only mechanism of hypoxia to suppress PKA activity. Since HIF was also able to regulate PKA activity in a cAMP-independent manner (Fig. 6 and 8), it is feasible that HIF, at least, might have another pathway(s) in the regulation of PKA activity, such as dephosphorylation of PKA substrates. Therefore, the molecular mechanisms by which HIF could suppress phosphorylation of PKA target proteins remain to be elucidated.
In C. elegans, ATGL-1 is the key lipase in whole-body lipid mobilization, and its expression level is critical for lipolytic activity. Previously, we have shown that PKA phosphorylates and stabilizes ATGL-1 protein in fasted worms (16). In mammals, ATGL protein could also be phosphorylated by PKA to mediate stimulated lipolysis (54). Our data suggest that the reduction in ATGL-1 protein by hypoxia or overexpression of HIF protein would be due, at least partially, to decreased PKA activity. Moreover, downregulation of ATGL protein upon hypoxia or HIF overexpression was reversed when proteasomal degradation was inhibited (Fig. 7 and 8). Together, these data suggest that the hypoxia-HIF pathway would promote ATGL protein degradation, leading to the suppression of lipolysis. Although it has been reported that hepatic ATGL is ubiquitinated by E3 ligase COP1 (55), there is no ortholog of COP1 in C. elegans. Thus, it is plausible to speculate that an unidentified E3 ligase of ATGL or ATGL-1 might be a potential target of HIF in mammalian adipocytes or C. elegans.
We noticed that the antilipolytic effect of hypoxia seemed to be slightly stronger than that of HIF modulation. For example, fasting of vhl-1(ok161) and egl-9(sa307) mutants under normoxia dampened fasting-induced lipolysis (Fig. 5), while fasting under hypoxia nearly completely blocked fasting-induced lipolysis (Fig. 1). Although the current data suggest that the majority of hypoxic responses of lipolysis and PKA signaling would be mediated by HIF activity, it is possible that some other factor(s) also contributes to alter lipid metabolism under hypoxia. One of the potential pathways that may reduce lipolysis during fasting includes reduced mitochondrial oxidation activity (56, 57). In addition, increased reactive oxygen species from mitochondria (58) or AMP-activated protein kinase (AMPK) activation (59, 60) upon hypoxia may also affect lipolytic activity. Since precise control of energy metabolism in response to the oxygen level is crucial, it is important for organisms to have multiple regulatory mechanisms of lipolysis. On the basis of our data, HIF1, a protein activated by direct sensing of a reduction in oxygen availability, would be a key player in suppressing stimulated lipolysis during hypoxia. While the other pathways mentioned above that could act in parallel with or dependently on HIF are still elusive, it will be interesting to investigate these in future studies.
In this study, we elucidated novel roles of HIF in alleviating lipolysis through suppression of ATGL and PKA upon hypoxia (Fig. 9). To adapt to oxygen-limiting states, C. elegans appears to turn on defensive mechanisms by inducing HIF-1 stabilization to restrain lipolysis. Moreover, this inhibitory effect of HIFα on lipolysis is able to limit inefficient lipid breakdown in mammalian adipocytes. Therefore, our data provide a clue to understand at the molecular level how local hypoxic areas, such as expanding adipose tissue in obesity, ischemic tissue, and solid tumor with poor vascularization, could alter energy homeostasis to reserve energy resources such as lipid metabolites for survival.
Proposed model.
MATERIALS AND METHODS
Nematode culture and RNAi.Worms of all strains were grown at 20°C on nematode growth medium (NGM) plates seeded with Escherichia coli OP50. The N2 Bristol strain was used as the wild-type strain. The Caenorhabditis Genetics Center (CGC) provided hjIs67[atgl-1p::atgl-1::gfp], vhl-1(ok161), vhl-1(ok161); hif-1(ia04), egl-9(sa307), egl-9(sa307); hif-1(ia04), and iaIs34[hif-1p::hif-1a (P621G)::tag + unc-119(+)] worms, and iaIs7[nhr-57p::gfp; unc-119(+)] worms were kindly provided by Seung-Jae V. Lee (POSTECH, Pohang, South Korea). For feeding RNAi experiments, kin-2, vhl-1, and hif-1 RNAi clones were obtained from the Ahringer and Vidal RNAi libraries. Synchronized worms were cultured on RNAi plates until they reached the young adult stage. To expose worms under the hypoxic condition, a hypoxia chamber (Modular Incubator Chamber MIC-101 [Billups-Rothenberg, Inc.] or ProOX C21 [BioSpherix]) was used. The hypoxia chamber was set 1% O2 with nitrogen gas.
Cell culture and transfection.Differentiation of 3T3-L1 adipocytes and siRNA transfection were performed as described previously (51). Hemagglutinin (HA)-HIF1α, HA-HIF1α ΔNAD (lacking amino acids [aa] 513 to 595), and HA-HIF1α ΔCAD (lacking aa 781 to 826) were used for overexpression of HIF1α.
Lipid staining.Oil red O (ORO) staining in C. elegans was performed as previously reported (16, 61). To visualize lipid droplets in 3T3-L1 adipocytes, cells were fixed with 1% paraformaldehyde. BODIPY 493/503 stock solution (d-3922; Molecular Probes) (1 mg/ml in methanol) was diluted 1:1,000 in phosphate-buffered saline (PBS) before being added and incubated with cells for 30 min. Mounting medium (Vectashield with 4′,6-diamidino-2-phenylindole [DAPI] H-1200; Vector Laboratories) was used for fluorescence imaging. Adipocytes were observed and imaged using an LSM 700 confocal microscope (Zeiss).
Fluorescence imaging and immunocytochemistry.Images of iaIs7[nhr-57p::gfp; unc-119(+)] worms were captured using an Axioplan 2 microscope (Zeiss). Images of hjIs67[atgl-1p::atgl-1::gfp] worms were captured using an LSM 700 confocal microscope. Tetramisole hydrochloride (50 mM; Sigma) was used as an anesthetic. To visualize ATGL and PLIN1 proteins, differentiated adipocytes were fixed with 1% paraformaldehyde. Antibodies against ATGL (2138S; Cell Signaling) (at 1:400) and PLIN1 (20R-pp004; Fitzgerald Industries) (at 1:500) were used for immunohistochemistry (ICC). Mounting medium (Vectashield with DAPI H-1200) was used for fluorescence imaging. Images of adipocytes were acquired using an LSM 700 confocal microscope.
cAMP measurement.cAMP concentrations were measured using a direct cAMP enzyme-linked immunosorbent assay (ELISA) kit (catalog number 25-0114; Enzo Life Sciences) according to the manufacturer’s protocol. C. elegans or differentiated adipocytes were resuspended and lysed in 0.1 M HCl with 0.1% Triton X-100 to inactivate phosphodiesterase. After centrifugation, the supernatants of cell extracts were subjected to ELISA. The results were analyzed using four-parameter logistic curve-fitting models. cAMP concentrations were normalized to the level of the total protein contents.
PKA activity assay.PKA activity in young adult worms was measured as described previously (51).
Western blotting.Cells were lysed on ice with modified radioimmunoprecipitation assay (RIPA) buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1% (vol/vol) Triton X-100, 0.5% (wt/vol) sodium deoxycholate, 0.1% (wt/vol) SDS, 5 mM NaF, 1 mM Na3VO4, and a protease inhibitor cocktail (catalog number P3100; GeneDEPOT). Antibodies against ATGL (2138S; Cell Signaling) (1:1,000), HSL (4107S; Cell Signaling) (1:1,000), pHSL Ser563 (4139S; Cell Signaling) (1:1,000), pPKA substrate (9624S; Cell Signaling) (1:1,000), PLIN1 (62) (1:5,000), HIF1α (3716S [Cell Signaling] and NB100-479 [Novus]) (1:1,000), HIF2α (NB100-132; Novus) (1:1,000), and β-actin (A5316; Sigma) (1:2,000) were used for Western blotting as described previously (63).
Glycerol release assay.Glycerol release from adipocytes was measured using free glycerol reagent (F6428; Sigma) according to the manufacturer’s protocol.
Adenovirus infection.HIF1α- and HIF2α-expressing adenoviruses were generously provided by Jang-Soo Chun (Gwangju Institute of Science and Technology [GIST], Gwangju, South Korea). As a negative control (mock), green fluorescent protein (GFP)-labeled adenovirus was used. Adenoviral infection was carried out as described previously (64).
Statistical analysis.Values are shown as the means plus standard deviations (SD). Mean values were compared and evaluated with Student's t test or two-way analysis of variance with a Bonferroni posttest. P values of <0.05 were considered significant.
ACKNOWLEDGMENTS
We thank Jang-Soo Chun (GIST, Gwangju, South Korea) for adenoviruses and Seung-Jae V. Lee (POSTECH, Pohang, South Korea) for iaIs7[nhr-57p::gfp; unc-119(+)] worms. We also thank Jong-Wan Park (Seoul National University, Seoul, South Korea) for HA-HIF1α, HA-HIF1α ΔNAD, and HA-HIF1α ΔCAD constructs.
This study was supported by the National Research Foundation (NRF) funded by the South Korean government (MISP 2011-0018312, to J. B. Kim) and Basic Science Research Program through the NRF (NRF-2018R1A6A3A11043137, to S. S. Choe). J. S. Han, J. Kong, Y. Ji, and J. Kim were supported by the BK21 program.
We declare that we have no conflicts of interest.
J. S. Han and J. H. Lee designed and conducted the study and wrote the paper. J. Kong, Y. Ji, J. Kim, and S. S. Choe analyzed the experiments and wrote the paper. J. B. Kim supervised the whole study and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.
FOOTNOTES
- Received 1 August 2018.
- Returned for modification 24 August 2018.
- Accepted 22 October 2018.
- Accepted manuscript posted online 5 November 2018.
- Copyright © 2019 American Society for Microbiology.
