Metabolism under hypoxia is significantly different from that under normoxia. It has been well elucidated that HIF-1 (hypoxia-inducible factor-1) plays a central role in regulating glucose metabolism under hypoxia; however, the role of HIF-1 in lipid metabolism has not yet been well addressed. In the present study we demonstrate that HIF-1 promotes LDL (low-density lipoprotein) and VLDL (very-LDL) uptake through regulation of VLDLR (VLDL receptor) gene expression under hypoxia. Increased VLDLR mRNA and protein levels were observed under hypoxic or DFO (deferoxamine mesylate salt) treatment in MCF7, HepG2 and HeLa cells. Using dual-luciferase reporter and ChIP (chromatin immunoprecipitation) assays we confirmed a functional HRE (hypoxia-response element) which is localized at +405 in exon 1 of the VLDLR gene. Knockdown of HIF1A (the α subunit of HIF-1) and VLDLR, but not HIF2A (the α subunit of HIF-2), attenuated hypoxia-induced lipid accumulation through affecting LDL and VLDL uptake. Additionally we also observed a correlation between HIF-1 activity and VLDLR expression in hepatocellular carcinoma specimens. The results of the present study suggest that HIF-1-mediated VLDLR induction influences intracellular lipid accumulation through regulating LDL and VLDL uptake under hypoxia.
The response to hypoxia is certainly the most well-understood aspect of the ability of HIF (hypoxia-inducible factor) to regulate metabolism . HIF is a heterodimeric protein that is composed of a constitutively expressed β subunit [HIFβ, also called ARNT (aryl receptor nuclear translocator)] and an oxygen-regulated α subunit (HIFα) [2,3]. There are three different HIFα peptides in higher metazoans. HIF-1α and HIF-2α have been the most extensively studied. They have a similar domain architecture, bind to the same core motif and are also regulated in a similar manner. HIF-3α is an inhibitor of HIF-1 that may be involved in feedback regulation because its expression is transcriptionally regulated by HIF-1 . In well-oxygenated cells, HIFα is hydroxylated at one (or both) of two highly conserved prolyl residues by a family of dioxygenases [EGLNs (EGL nine homologues), also called PHDs (prolyl hydroxylase domain proteins)], which use oxygen and α-oxoglutarate as substrates in a reaction that generates carbon dioxide and succinate as byproducts [5–8]. Prolyl-hydroxylated HIFα is bound by VHL (von Hippel–Lindau tumour suppressor protein), which recruits an E3-ubiquitin ligase that targets HIFα for proteasomal degradation. Under hypoxic conditions, hydroxylation is inhibited and HIFα rapidly accumulates, dimerizes with HIFβ, binds to the core DNA-binding sequence 5′-RCGTG-3′ (R is purine A or G) in target genes, recruits co-activators and activates transcription.
The role of HIF-1 in glucose/energy metabolism has been well studied. Recently studies have reported a role for HIF-1 in glycogen accumulation [9,10]. Hypoxia can also promote lipid droplet formation and lipid accumulation [11–13], and some reports have indicated that HIF-1 may play an important role in lipid accumulation [12,14–16]. Krishnan et al.  have reported that HIF-1 promotes lipid accumulation through inducing PPARγ (peroxisome-proliferator-activated receptor γ), which promotes non-esterified (free) fatty acid uptake and triacylglycerol production . All of these results support the notion that HIF-1 plays an important role in lipid accumulation under hypoxia.
The tumour microenvironment suffers from hypoxia, and lipid accumulation is often observed in solid tumours [16,17]. ADFP (adipophilin) and HIG2 (hypoxia-inducible protein 2), which are lipid droplet proteins, were induced under hypoxia in a HIF-1dependent manner, and they are overexpressed in tumours [16,17]. Overexpression of ADFP and HIG2 may contribute to tumour lipid accumulation. It has been reported that hypoxia significantly increases LDL (low-density lipoprotein) uptake and enhances lipid accumulation in arterial SMCs (smooth muscle cells), exclusive of LDLR (LDL receptor) activity . VLDLR [VLDL (very-LDL) receptor] is a receptor that could be involved in LDL and VLDL uptake. Some studies have reported that VLDLR could be induced under hypoxia [19–21]. We hypothesize that HIF-1 regulates VLDLR to promote LDL and VLDL uptake and lipid accumulation. In the present study we have demonstrated that HIF-1 induces VLDLR directly and promotes lipid accumulation through increasing LDL and VLDL uptake. In addition, we also found that VLDLR and HIF-1 are up-regulated and expressed in the same areas of tumour in hepatocellular carcinoma.
Cell culture, hypoxic exposures, and LDL and VLDL treatment
MCF7 and HeLa cells were grown in DMEM (Dulbecco's modified Eagle's medium; Gibco) supplemented with 10% FBS (fetal bovine serum), 100 units/ml penicillin and 100 μg/ml streptomycin. HepG2 cells were grown in MEM (modified Eagle's medium; Gibco) supplemented with 10% FBS, 100 units/ml penicillin and 100 μg/ml streptomycin. Normoxic cells were maintained at 37°C in a 5% CO2 and 95% air incubator. Cells were exposed to hypoxia in a three gas incubator (YCP-50S) in 5% CO2, 94% N2 and 1% O2 at 37°C. For LDL treatment, the culture medium was changed to fresh medium (10% FBS) with or without LDL at a final concentration of 200 μg/ml. For VLDL treatment, the culture medium was changed to fresh medium (10% FBS) with or without VLDL at a final concentration of 100 μg/ml. Then the cells were moved to normoxic or hypoxic incubators, cultivated for 24 h and then fixed or recollected for analysis. Human LDL and VLDL were purchased from Peking Union Bio Company.
The promoter region of VLDLR was amplified from human genomic DNA by PCR and cloned into the luciferase reporter vector pGL3-Basic (Promega). As a positive control, the nucleotide sequence containing the identified HRE (hypoxia-response element) of EPO (erythropoietin)  was also cloned into the pGL3-promoter.
Western blot analysis
Cell lysates were subjected to SDS/PAGE (12% gel) and transferred on to a PVDF membrane. Primary antibodies against the following proteins were used: HIF-1α (sc-53546, Santa Cruz Biotechnology), HIF-2α (NB100-122, Novus Biologicals), β-actin (60008-1-1g, Proteintech) and VLDLR (sc-18824, Santa Cruz Biotechnology). HRP (horseradish peroxidase)-conjugated secondary antibodies were used. Signals were detected using an ECL (enhanced chemiluminescence) kit (Millipore).
Transient transfection and luciferase assays
MCF7 cells were plated into 24-well plates to reach approximately 50–70% confluence on the following day. The cells were co-transfected with pGL3-basic-based construct and pRL-TK plasmid DNAs using Lipofectamine™ Plus (Invitrogen). The transfection medium was replaced with complete medium after 6 h. The cells were incubated under normoxia and hypoxia for an additional 24 h. Then cells were lysed with PLB (passive lysis buffer, from the dual-luciferase reporter assay kit; Promega) and reporter gene expression was assessed using the dual-luciferase reporter assay system (Promega).
RNA isolation and PCR analysis
Cells were dissolved in TRIzol® reagent (Invitrogen) and total RNA was extracted according to the manufacturer's protocol. Total RNA was converted into cDNA using the MMLV (Moloney murine leukaemia virus) reverse-transcription system (Invitrogen) in the presence of oligo(dT)18. The cDNA was used for conventional PCR or qrt-PCR (quantitative real-time PCR) with specific gene primers (Supplementary Table S1 at http://www.BiochemJ.org/bj/441/bj4410675add.htm). SYBR green PCR mix (Transgen) was used for qrt-PCR. The relative abundance of the VLDLR transcript was quantified using the comparative Ct method with β-actin being used as an internal control.
ChIP (chromatin immunoprecipitation)
MCF7 and HepG2 cells were plated on to 15 cm plates and grown to approximately 70% confluence. Then the cells were exposed to hypoxia for 24 h and ChIP was performed as described previously . Primary antibodies against the following proteins were used: HIF-1α (ab2185, Abcam), HIF-2α (NB100-122, Novus Biologicals) and rabbit IgG (ab2410, Abcam). The precipitated DNA was amplified by PCR using the primers shown in Supplementary Table S1. As a positive control, the VEGF (vascular endothelial growth factor) promoter region that contains a known HIF-1-binding site was amplified using the primers, as described previously .
Cell transfection with siRNA (small interfering RNA)
At 1 day before transfection, cells were plated on to six-well plates. The cells were grown to 50% confluence and then transfected with 25 nM (final concentration) siGENOME non-targeting siRNA2, human HIF1A (the α subunit of HIF-1) siGENOME SMART pool or human HIF2A (the α subunit of HIF-2) siGENOME SMART pool, or human VLDLR siGENOME SMART pool (Thermo Fisher Scientific) using DharmaFECT1 transfection reagent according to the manufacturer's protocol. After 24 h incubation under normoxia, the transfection medium was replaced with complete medium and the cells were incubated under hypoxia for another 24 h. Total RNA and cell lysates were collected for qrt-PCR and Western blot analysis respectively.
Flow cytometric analysis
For flow cytometric analysis, cells were placed into six-well plates. After incubation and treatment, the cells were digested with 0.25% trypsin-EDTA, collected and washed three times with PBS. Cells were then fixed with 3.7% formaldehyde for 30 min. Cells were immediately rinced twice with PBS. Then the cells were stained with 1 ml of Nile Red working solution for 20 min at 37°C. The cells were rinsed twice with PBS, resuspended in PBS and analysed immediately using an Accuri C6 flow cytometer system (Accuri Cytometers). The Nile Red stock solution was purchased from Genmed Scientifics. The Nile Red working solution was prepared by diluting 0.5 μl of the stock solution in 1 ml of PBS, and mixing well.
Coverslips were placed into a 24-well plate and then MCF7 cells were plated. After incubation and treatment, medium was removed and the cells were fixed with 3.7% formaldehyde fixative for 30 min. Cells were immediately rinsed three times with deionized water for 2 min each. Slides were allowed to air dry for a few minutes at room temperature (25°C). Cells were stained with freshly prepared Oil Red O working solution for 15–20 min. Cells were then immediately rinsed four times with deionized water for 5 min each. After slides had air-dried completely, the stained cells were covered with glass slides using mounting medium, DAPI (4′,6-diamidino-2-phenylindole) fluoromount G (SouthernBiotech). Fluorescence images were acquired with a TCS SP2 confocal microscope (Leica) using Leica Confocal Sofware. Oil Red O stock solution was prepared by adding 0.5 g of Oil Red O (Sigma–Aldrich) to 100 ml of 1,2-propanediol (Sigma–Aldrich), mixing well, and crushing larger pieces with a stir bar. The solution was then heated gently until the solution reached 95–100°C. The solution was filtered through coarse filter paper while still warm. The Oil Red O working solution was prepared by diluting 3 ml of stock solution with 2 ml of distilled water, mixing well, and filtering with a 0.22-μm filter unit. Working solution was freshly made up from the stock solution each time.
Intracellular distribution of lipid droplets
Cells were placed on to 24-well plates. After incubation and treatment of the cells, an intracellular distribution assay of lipid droplets was performed as described in the Steatosis Colorimetric Assay Kit (Cayman Chemicals).
The histological sections of paraffin-embedded tissue containing tumour and adjacent tissue were obtained from the Cancer Institute and Hospital, Chinese Academy of Sciences. The samples were obtained from the operated patients with informed consent and the research involving the samples has been approved by the Research Ethics Committee in the Cancer Institute and Hospital and the Institute of Basic Medical Sciences. Serial sections were mounted on pre-treated glass slides, deparaffinized, rehydrated and heat-induced antigen retrieval was performed. Endogenous peroxidase was quenched using 3% H2O2 for 15 min; then slides were washed in PBS (pH 7.5) and incubated overnight at 4°C with primary antibody. Antibodies against the following proteins were used: HIF-1α (Santa Cruz Biotechnology; 1:50 dilution) and VLDLR (Santa Cruz Biotechnology; 1:50 dilution). After being washed, the sections were incubated with HRP-conjugated secondary antibody (Maixin Biotechnology) for 30 min at room temperature. Then the sections were developed in DAB (diaminobenzidine) (Maixin Biotechnology), counterstained with haematoxylin, dehydrated and coverslipped. Images were taken with a BX51 microscope (Olympus).
Hypoxia induces VLDLR in an HIF-1-dependent manner
To examine whether VLDLR expression is oxygen-regulated in human cells, MCF7, HeLa and HepG2 cells were incubated under normoxia (21% O2), hypoxia (1% O2) or medium containing 100 μM DFO (deferoxamine mesylate salt) for 24 h. Both conventional RT (reverse transcription)–PCR (Figure 1A) and qrt-PCR (Figure 1B) revealed an increased VLDLR mRNA level in the cells under hypoxic or DFO treatment conditions. An increased VLDLR protein level accompanying increased HIF-1α and HIF-2α protein levels was also observed in the cells under hypoxic or DFO treatment conditions (Figure 1C). These results indicated that VLDLR is a hypoxia-inducible gene. In agreement, a significantly decreased VLDLR mRNA (Figures 2A and 2B) and protein (Figures 2C) levels were detected in the MCF7 and HepG2 cells transfected with siRNA targeting HIF1A. Taken together, these results demonstrated that VLDLR could be induced in cells by hypoxia and this response is regulated by HIF-1.
VLDLR is induced under hypoxia
Hypoxia induces VLDLR in an HIF-1-dependent manner
Identification and validation of an HRE in human VLDLR
We searched the promoter region of human VLDLR for matches to a consensus HRE sequence as described previously . Thirteen putative HREs (Figure 3A) were identified. To determine which one was the functional HRE, we performed a series of dual-luciferase reporter assays. The promoter region of VLDLR was amplified and inserted into the pGL3-basic luciferase reporter plasmid. As shown in Figure 3(B), the presence of the region from −826 to +1145 dramatically increased luciferase activity in MCF7, HeLa and HepG2 cells under hypoxia. Analysis of deletion constructs suggested that the HRE in exon 1 of VLDLR was required for maximal induction of VLDLR promoter activity in response to hypoxia (Figure 3C). Mutation of the HRE in exon 1, but not other HREs, dramatically impaired the induction of luciferase activity by hypoxia (Figure 3D), which suggested that the HRE in exon 1 is the functional HRE. The mutant sequences of the HREs are shown in Supplementary Figure S1B (at http://www.BiochemJ.org/bj/441/bj4410675add.htm). Of note, the functional HRE is involved in protein coding. The sequences are near the translational start site ‘ATG’ and span the codons from second to fourth. The HRE in exon 1 was conserved between human, rat and mouse (Supplementary Figure S1A).
Identification of a functional HRE in the VLDLR gene
To further demonstrate that HIF-1 binds to the HRE in exon 1 within living cells, we performed ChIP assays using antibodies against HIF-1α and HIF-2α (IgG as a negative control) in normoxic- and hypoxic-cultured MCF7 and HepG2 cells. The amplicon containing the HRE in exon 1 was detected from chromatin fragments immunoprecipitated with anti-HIF-1α, anti-HIF-2α and anti-IgG (Figure 3E). As a positive control, we demonstrated binding of HIF-1 to a known HIF-1 target gene, VEGF. As a negative control, an amplicon in the VLDLR promoter from −539 to −292 was detected with the primers P1 and P2. In summary, these results showed that VLDLR is directly induced by HIF-1 under hypoxia.
Hypoxia promotes LDL and VLDL uptake
VLDLR is a receptor that could be involved in LDL and VLDL uptake. It has been reported that hypoxia significantly increases LDL uptake and enhances lipid accumulation in arterial SMCs, but this was independent of LDLR activity . On the basis of this we hypothesized that hypoxia increases LDL and VLDL uptake through induction of VLDLR. To confirm this hypothesis, MCF7 cells were treated with LDL or VLDL, exposed to normoxia or hypoxia for 24 h, then fixed or recollected for analysis. Flow cytometry, confocal microscopy and a steatosis colorimetric assay kit were used to analyse intracellular lipid content. All of these results demonstrated that hypoxia increases LDL and VLDL uptake and intracellular lipid accumulation (Figure 4). Similar results were also observed in HeLa and HepG2 cells (Supplementary Figure S2 at http://www.BiochemJ.org/bj/441/bj4410675add.htm).
Hypoxia promotes LDL and VLDL uptake and lipid accumulation in MCF7 cells
HIF-1 regulates LDL and VLDL uptake through inducing VLDLR
To test the role for HIF in hypoxia-induced lipid accumulation, MCF7 cells were transfected with siRNAs specifically targeting HIF1A and HIF2A. Knockdown of HIF1A, but not HIF2A, attenuated the hypoxia-induced LDL and VLDL uptake and lipid accumulation (Figure 5). These results demonstrated that HIF-1α, but not HIF2α, is necessary for the hypoxia-induced LDL and VLDL uptake, and lipid accumulation.
LDL and VLDL uptake and lipid accumulation is attenuated by knockdown of HIF-1α under hypoxia
To further test the role of VLDLR, MCF7 cells were transfected with siRNA specifically targeting VLDLR. In comparison with the cells transfected with non-targeting siRNA (sicontrol), a reduction in LDL and VLDL uptake and lipid accumulation (Figures 6A–6C), accompanying a reduction in both VLDLR mRNA and protein levels (Figures 6D and 6E), were observed in the cells where VLDLR was knocked down. These results suggest that HIF-1 regulates LDL and VLDL uptake and lipid accumulation through inducing VLDLR under hypoxia.
Hypoxia-induced LDL and VLDL uptake and lipid accumulation is attenuated by knockdown of VLDLR
Up-regulation and co-expression of VLDLR and HIF-1α in the same areas of hepatocellular carcinoma
If VLDLR is a HIF-1α-mediated gene product, its expression should occur in the same cells and regions of a tumour where HIF-1α is stabilized. We investigated this phenomenon in vivo by performing immunohistochemistry in hepatocellular carcinoma samples. We observed a correlation between expression of HIF-1α and VLDLR (Figure 7A). Among 31 hepatocellular carcinoma samples, 19 cases were strongly positive for HIF-1α, 11 cases were positive for HIF-1α and one case was negative for HIF-1α. Strong positive expression of VLDLR was associated with the strong positive expression of HIF-1. The more strongly positive HIF-1 was expressed, the more strongly positive was VLDLR expressed (Figure 7B). In addition, we also observed that the VLDLR signals surrounded the angular empty spaces, which probably represent lipid droplets (the neutral lipid, the main content of lipid droplets in the tissues, was dissolved in organic reagents during the immunohistochemical experiments) (Figure 7A).
Co-presence of VLDLR and HIF-1α in hepatocellular carcinoma
Metabolism under hypoxia is significantly different from that under normoxia. HIF-1α plays an important role in this process. It has been well elucidated that HIF-1α plays a central role in regulating glucose metabolism under hypoxia. It has been reported that hypoxia promotes lipid body formation and lipid accumulation [11–13]. Some reports have also indicated that HIF-1 may play an important role in lipid metabolism [12,14–16]. Wada et al.  have reported that hypoxia significantly increases LDL uptake and enhances lipid accumulation in arterial SMCs, exclusive of LDLR activity. Some studies have also reported that VLDLR could be induced under hypoxia [19–21]. On the basis of this, we hypothesized that hypoxia promotes LDL uptake through VLDLR. VLDLR is the receptor of VLDL, so increased expression of VLDLR could promote VLDL uptake. We also analysed the promoter region of human VLDLR and found many putative HREs. On the basis of this we hypothesized that HIF may play a role in hypoxia-induced LDL and VLDL uptake by inducing VLDLR. The results of the present study support this hypothesis.
VLDLR is a member of the LDLR family and is most abundantly expressed in the heart, skeletal muscle and adipose tissue, but not in the liver . LDLR displays high-affinity binding of LDL-containing apoB (apolipoprotein B), as well as of apoE (apolipoprotein E)-rich lipoproteins. VLDLR binds VLDL with high affinity, whereas LDL particles containing apoB are weakly bound to VLDLR . The results of the present study confirmed that VLDLR is directly induced by HIF-1 under hypoxia, and hypoxia increases LDL and VLDL uptake through HIF-1 inducing VLDLR. During the preparation of the present paper, Castellano et al.  also found that hypoxia increased VLDL uptake in cardiomyocytes, which may partially be dependent on up-regulating VLDLR expression . In another paper, Perman et al.  found that VLDLR was regulated by HIF-1 under hypoxia in mice cardiomyocytes . Their results also support our present hypothesis. LDLR, VLDLR and LDLRAP1 (LDLR adaptor protein 1) can affect LDL and VLDL uptake . Thus we detected the protein levels of LDLR and LDLRAP1 under normoxia and hypoxia, and found that hypoxia did not cause an increase in the protein levels (Supplementary Figure S3 at http://www.BiochemJ.org/bj/441/bj4410675add.htm).
The tumour microenvironment suffers from hypoxia, and lipid accumulation is often observed in solid tumours . Where does the lipid come from? Some studies have reported that VLDLR is highly expressed in lots of tumours, such as gastric, breast and lung tumours [29,30]. Thus lipid accumulation in solid tumours may at least partially be dependent on VLDLR. What is the function of lipid accumulation to cells? Lipid droplets are the places that not only store lipid, but they also produce lipid signals. Dysregulated lipid signalling has been found in cancer . PGE2 (prostaglandin E2)-mediated signalling and the enzymes regulating its biosynthesis play a pivotal role in cancer development . COX-2 (cyclo-oxygenase-2) and PTGES (prostaglandin E synthase) are enzymes involving in the production of PGE2, and these are up-regulated in cancers and regulated by HIF-1 [33,34]. In addition, one study has reported that COX-2 is localized in lipid droplets . On the basis of this, we hypothesize that lipid accumulation in cancer may involve lipid signal production. What is more, lipid accumulation might have different functions in different cells. DCs (dendritic cells) with lipid accumulation were not able to effectively stimulate antitumour immune responses . Further studies regarding the function of lipid accumulation in tumour cells are needed.
Besides LDL and VLDL, VLDLR binds numerous other ligands, including LPL (lipoprotein lipase) , RAP (receptor-associated protein) , thrombospondin-1 [39,40], thrombospondin-2 , uPA–PAI-1 complex (urokinase plasminogen activator–plasminogen activator inhibitor-1 complex)  and Reelin [42,43]. Many of these VLDLR–ligand interactions may account for diverse biological functions of the VLDLR. A signalling function for VLDLR has also been recognized. VLDLR is essential for the proper transmission of the signalling cascade mediated by Reelin, a secreted glycoprotein that modulates neuronal migration, neurodevelopment and other physiological processes in the CNS (central nervous system) . Some reports have indicated that VLDLR might play an important role in cell proliferation and migration [44–47]. We have carried out some experiments to examine the effects of VLDLR knockdown on cell-cycle arrest and survival under hypoxia. Our present results suggest that knockdown of VLDLR did not significantly affect cell-cycle arrest and survival in MCF7 cells cultured in conditions of 1% O2 (results not shown). The uPA–PAI-1 complex stimulates pro-proliferative signalling events via a high-affinity interaction with VLDLR on MCF-7 cells . Thrombospondin-1 binds to VLDLR and functions in postnatal neuronal migration . He et al.  have reported that the higher VLDLR expression is associated with lymph node metastasis, distant metastasis and advanced TNM (tumour, node, metastasis) stage in breast and gastric cancer . Does VLDLR play a role in signal transduction or cell migration under hypoxia? Does VLDLR function through an interaction with different ligands? All of these questions, and the mechanisms of VLDLR in cancer progression, require further study.
deferoxamine mesylate salt
fetal bovine serum
hypoxia-inducible protein 2
LDLR adaptor protein 1
quantitative real-time PCR
small interfering RNA
smooth muscle cell
- uPA–PAI-1 complex
urokinase plasminogen activator–plasminogen activator inhibitor-1 complex
vascular endothelial growth factor
Guo-Min Shen designed and performed the experiments, analysed the data and wrote the paper; Ying-Ze Zhao performed the immunohistochemistry analysis; Ming-Tai Chen, Feng-Lin Zhang and Xiao-Ling Liu helped with the experiments; Yi Wang, Chang-Zheng Liu and Jia Yu supplied the hepatocellular carcinoma tissue samples; Jun-Wu Zhang designed the experiments, analysed the data and wrote the paper.
This work was supported by the National Basic Research Program of China [grant number 2006CB504100 (to J.-W. Z)].