Contents lists available at ScienceDirect

BBA - Molecular and Cell Biology of Lipids

journal homepage: www.elsevier.com/locate/bbalip

Expression of AGPAT2, an enzyme involved in the glycerophospholipid/triacylglycerol biosynthesis pathway, is directly regulated by HIF-1 andpromotes survival and etoposide resistance of cancer cells under hypoxia

Eleni-Anastasia Triantafylloua, Eleni Georgatsoub, Ilias Mylonisb, George Simosb,c,Efrosyni Paraskevaa,⁎

a Laboratory of Physiology, Faculty of Medicine, University of Thessaly, Larissa, Greeceb Laboratory of Biochemistry, Faculty of Medicine, University of Thessaly, Larissa, GreececGerald Bronfman Department of Oncology, Faculty of Medicine, McGill University, Montreal, Canada

A R T I C L E I N F O

Keywords:AGPAT2LPAAT-betaHIF-1HypoxiaTriglyceride synthesisLipid droplets

A B S T R A C T

Hypoxia inducible factor-1 (HIF-1) supports survival of normal cells under low oxygen concentration and cancercells in the hypoxic tumor microenvironment. This involves metabolic reprogramming via upregulation ofglycolysis, downregulation of oxidative phosphorylation and, less well documented, effects on lipid metabolism.To investigate the latter, we examined expression of relevant enzymes in cancer cells grown under hypoxia. Weshow that expression of acylglycerol-3-phosphate acyltransferase 2 (AGPAT2), also known as lysophosphatidicacid acyltransferase β (LPAATβ), was upregulated under hypoxia and this was impaired by siRNA-mediatedknockdown of HIF-1α. Moreover, a sequence of the AGPAT2 gene promoter region, containing 6 putativeHypoxia Response Elements (HREs), activated transcription of a reporter gene under hypoxic conditions or innormoxic cells over-expressing HIF-1α. Chromatin immunoprecipitation experiments confirmed binding of HIF-1α to one of these HREs, mutation of which abolished hypoxic activation of the AGPAT2 promoter. Knockdownof AGPAT2 by siRNA reduced lipid droplet accumulation and cell viability under hypoxia and increased cancercell sensitivity to the chemotherapeutic etoposide. In conclusion, our findings demonstrate that AGPAT2, whichis mutated in patients with congenital generalized lipodystrophy and over-expressed in different types of cancer,is a direct transcriptional target of HIF-1, suggesting that upregulation of lipid storage by HIF-1 plays an im-portant role in adaptation and survival of cancer cells under low oxygen conditions.

1. Introduction

Organisms and cells have evolved a series of adaptations critical forthe survival in low oxygen concentrations (hypoxia) under physiolo-gical or pathological conditions, like embryogenesis, development,ischemia, inflammation and cancer. Exposure of cancer cells, in parti-cular, to the hypoxic microenvironment of solid tumors, elicits induc-tion of angiogenesis and changes in their metabolism that promotesurvival, proliferation, migration and invasiveness [1, 2]. This meta-bolic switch is largely achieved through changes in gene expressionmediated by two members of the Hypoxia Inducible Factors (HIFs) fa-mily, HIF-1 and HIF-2 [3, 4].

Both HIF-1 and HIF-2 are heterodimeric transcription factors, con-sisting of unique, oxygen regulated HIFα subunits and the common,constitutively expressed HIFβ subunit (or aryl hydrocarbon receptortranslocator; ARNT). The protein levels of the HIFα subunits are

regulated mainly at the post-transcriptional level, through the action ofspecific prolyl hydroxylases (PHDs) [5]. Under physiological oxygenlevels, PHDs modify HIFα that, as a consequence, is polyubiquitinatedby the von Hippel-Lindau (VHL) protein and degraded at the protea-some. When oxygen levels become limited, PHDs become inactive,HIFα is stabilized, is imported into the nucleus, dimerizes with ARNTand the active HIF heterodimer binds to hypoxia-response elements(HREs) at the promoter or enhancer of HIF target genes and inducestheir transcription [6]. HIF expression and activity are also controlledby oxygen-independent mechanisms. In this context, HIF-1 is regulatedby many intracellular signaling pathways, including NF-κB, PKR-STAT3, PI3K-mTOR, ROS, MAPK and CK1δ [7–11].

Whereas the effect of HIFs on the expression of genes involved incarbohydrate metabolism has been extensively studied [3], the hypoxicregulation of lipid metabolism remained, until recently, poorly in-vestigated. Recent findings demonstrate that HIF-1 is implicated in lipid

https://doi.org/10.1016/j.bbalip.2018.06.015Received 13 February 2018; Received in revised form 22 May 2018; Accepted 12 June 2018

⁎ Corresponding author at: Laboratoty of Physiology, Faculty of Medicine, University of Thessaly, Panepistimiou 3 (BIOPOLIS), 41500 Larissa, Greece.E-mail address: fparaskeva@med.uth.gr (E. Paraskeva).

BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

Available online 15 June 20181388-1981/ © 2018 Elsevier B.V. All rights reserved.

T

metabolism through lipid droplet (LD) accumulation [12, 13], increaseof fatty acid and lipid synthesis [13–15], decrease of fatty acid β-oxi-dation [16] and upregulation of fatty acid uptake [17, 18], while HIF-2promotes lipid storage [19]. Importantly, HIF-mediated fatty acid ac-cumulation under hypoxia may contribute to the development of me-tabolic disorders, such as liver steatosis, and hepatocellular carcinoma[20]. However, few direct targets of HIFs have been identified so far,among the many genes that code for lipogenic enzymes.

Fatty acids can be converted to triglycerides, neutral lipids that arestored in the form of LDs. The triglyceride (TG) synthesis pathway relieson the consecutive action of glycerol-3-phosphate acyltransferase(GPAT), acylglycerol-3-phosphate acyltransferase (AGPAT), phospha-tidic acid phosphatase (PAP or lipin) and diacylglycerol acyltransferase(DGAT) enzymes, each one of which exists in multiple isoforms [21].TG synthesis is important for metabolic homeostasis and cellulargrowth since LDs function as lipid stores providing both metabolicsubstrates and components for membrane synthesis [22]. Under hy-poxia, storage of fatty acids as neutral lipids in LDs, protects hypoxiccancer cells from lipotoxicity and ROS production [23], while HIF-1driven TG synthesis and LD accumulation is critical for cancer cellsurvival and proliferation [24]. In particular, HIF-1 mediates the hy-poxic induction of the lipin 1 (but not lipin 2) isoform, the enzyme thatcatalyzes the penultimate step in TG biosynthesis [13].

AGPATs, also known as lysophosphatidic acid acyltransferases(LPAATs), the enzymes preceding lipin-1, catalyze the conversion oflysophosphatidic acid (LPA) to phosphatidic acid (PA). Several AGPATisoforms are known, that are encoded by different genes. Most but notall AGPAT isoforms reside in the ER, and cannot complement eachother, possibly due to different substrate specificities [25]. In additionto their role in TG synthesis, generation of PA by AGPATs stimulatesphospholipid production and regulates mTOR activity [26]. AGPAT1and AGPAT2 are the earlier identified and better-studied isoforms.AGPAT1 is ubiquitously expressed, while AGPAT2 expression is mostlydetected in adipose tissue, liver, heart and pancreas. AGPAT2 mutationsare the cause of type 1 congenital generalized lipodystrophy (CGL)[27]. CGL patients lack adipose tissue, develop lipid accumulation inliver and muscle and exhibit severe metabolic complications [28, 29].AGPAT2 expression is upregulated in different types of cancer includingthose of breast, pancreas, lung, bladder, ovary, melanoma, uterus,renal, prostate, central nervous system, lymphoma, colorectal, me-sothelioma and leukemia [30–33]. Furthermore, AGPAT2 is highlyexpressed in osteosarcoma [34], pancreatic [26] and myeloma [35, 36]cancer cells and AGPAT2 silencing inhibits proliferation of cancer cellsand enhances their sensitivity to chemotherapeutic drugs [26, 37–40].

We report here that hypoxia induces the expression of AGPAT2 genein human hepatocellular carcinoma Huh7 and adenocarcinoma HeLacells. We also show that AGPAT2 is a direct target gene of HIF-1 andthat AGPAT2 induction is necessary for hypoxic accumulation of LDsand for sustaining viability and etoposide resistance under hypoxia.

2. Materials and methods

2.1. Materials

Tris-Base, Bromophenol Blue, etoposide, formaldehyde, Triton X-100, PMSF, Mowiol and Nile Red were purchased from Sigma-Aldrich(St Louis, MO, USA), SDS, DTT and glycine from AppliChem(Darmstadt, Germany), glycerol, NaCl and MgCl2 from Merck(Darmstadt, Germany), dimethyloxalyl glycine (DMOG) from AlexisBiochemicals (CA, USA), nitrocellulose membranes from Macherey-Nagel (Düren, Germany), Dulbecco's modified Eagle's medium (DMEM)High Glucose, fetal bovine serum (FBS)-South America Origin and pe-nicillin-streptomycin from Labtech International Ltd. (South Essex, UK)and dimethyl sulfoxide (DMSO) from Jena Bioscience (Jena, Germany).

2.2. Cell culture and treatment

Human hepatocarcinoma (Huh7), cervical adenocarcinoma (HeLa)and embryonic kidney cells (HEK293) were cultured in Dulbecco'smodified Eagle's medium (DMEM) High Glucose containing 10% fetalbovine serum (FBS)-South America Origin, 100 U/ml penicillin-strep-tomycin at 37 °C in a humidified incubator under 5% CO2. For hypoxictreatment, cells were exposed to 1% O2, 95% N2 and 5% CO2 in an INVIVO2 200 hypoxia workstation (RUSKINN Life Sciences/BakerRuskinn, Sanford, Maine, USA). When required, cells were treated forthe indicated time with DMOG (1mM), etoposide (75 μΜ), or DMSO atthe appropriate concentration as solvent control.

2.3. Plasmid construction, site directed mutagenesis, transfection andluciferase assay

The human AGPAT2 promoter (region −1013 to −140, Acc. No.NG_008090.1) was amplified by PCR from genomic DNA isolated fromHuh7 or HeLa cells using Quick-DNA Universal Kit (Zymo Research,Irvine, CA, USA), as template, specific primers (Table 1) and Taq DNApolymerase (Minotech Biotechnology, IMBB-FORTH, Heraklion, Crete,Greece). The PCR product was purified from agarose gel using Isolate IIPCR and Gel Kit (Bioline, Luckenwalde, Germany), cloned into pminiTVector (New England Biolabs. Inc., Hitchin, Hertfordshire, UK), sub-cloned into the XhoI sites of pGL3 basic Luciferase Reporter Vector(Promega Corporation, Madison, WI, USA), to construct pGL3-AGPAT2promoter and sequenced. To mutagenize HRE4 of the AGPAT2 pro-moter in the pGL3-AGPAT2 plasmid, three nucleotide substitutions(CCGTG to CAACG) were introduced by PCR using the QuikChange IISite-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA,USA) and specific primers (Table 1). The mutant construct was verifiedby sequencing (VBC-Biotech, Vienna, Austria).

HEK293 cells cultured in 24-well plates were transiently transfectedby Polyethylenimine (PEI) reagent (Polysciences Inc., Hirschberg an der

Table 1Oligonucleotide sequence list.

Name Sequence (5′-3′)

DNA primers for AGPAT2 promoter (−1013 to −140) luciferase reporter plasmidAGPAT2 promoter FOR AAAAAGAGGGGCCGTGCTCAGPAT2 promoter REV CAGGGAAGGGCTAGGTGC

DNA primers used for CHIP analysisHRE 5–6 (−1013 to −778) FOR AAAAAGAGGGGCCGTGCTCHRE 5–6 (−1013 to −778) REV GTTCACATCCGCTTGGCAGHRE 4 (−685 to −492) FOR AGACACACGCCCCAGTTGHRE 4 (−685 to −492) REV CAGAACCACAGCTCCCCAAGHRE 1–3 (−330 to −140) FOR GTAACCTGGCAGAAGGCTGTHRE 1–3 (−330 to −140) REV CAGGGAAGGGCTAGGTGC

DNA primers for quantitative-RT-PCRhLPIN1 FOR TTTCCACGTCCGCTTTGGGhLPIN1 REV GTGGCCAGGTGCATAGGGhGPAT3 FOR TTCCCTTCGCAGAGGTGAGThGPAT3 REV CGACTCCTTTGGGGTTCCTThGPAT4 FOR CACGTCCTGACGCAGCTTGhGPAT4 REV TATCTGGCCCCATCCTCCTChAGPAT1 FOR CCCCCACCATTCCTACCGChAGPAT1 REV CCTGGCCACAAATCCATTCThAGPAT2 FOR TGTGGATCTATCCCGAGGGThAGPAT2 REV GAAGAGTACACCACGGGGAChDGAT1 FOR CCTCGGGCGCTACGAACChDGAT1 REV ACACCAGTTCAGGATGCCAChDGAT2 FOR TCACCTGGCTCAATAGGTCChDGAT2 REV TCGAAAGTAGCGCCACACAGhβ-ACTIN FOR CCAACCGCGAGAAGATGAhβ-ACTIN REV CCAGAGGCGTACAGGGATAG

DNA primers for site directed mutagenesisHRE4 Mut forward primer CCAGGCCGGTGGCCGTTCAAAGGGCCGCCGGHRE4 Mut reverse primer CCGGCGGCCCTTTGAACGGCCACCGGCCTGG

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1143

Bergstrasse, Germany) with plasmids (1 or 0.25 μg) wild-type or mutantpGL3-AGPAT2 promoter, pGL3-Lipin-1 promoter [13] or empty pGL3vector and the Renilla luciferase expressing plasmid pCI-Renilla(0.05 μg). When required, cells were also co-transfected with pEGFP-C1-HIF-1α (0.75 μg) [11] or pEGFP-N3 (0.75 μg) (BD Biosciences, SanJose, CA, USA). 24 h post transfection, cells were incubated undernormoxia or hypoxia for the indicated times and luciferase activity wasdetermined according to the Dual Luciferase Reporter Assay Systemprotocol (Abnova, Neihu District. Taipei City, Taiwan).

2.4. siRNA-Mediated silencing

siRNA AGPAT2/LPAATβ (25 nM, Santa Cruz Biotechnology, Dallas,TX, USA), siRNA HIF-1α (20 nM, Qiagen, Venlo, Netherlands), orAllStars negative control siRNA (20 nM, Qiagen, Venlo, Netherlands)was introduced in cells using Viromer Blue (Lipocalyx, Halle-Saale,Germany).

2.5. RNA extraction and quantitative real-time PCR

Total RNA was isolated by NucleoZol Reagent (Macherey-Nagel,Duren, Deutschland) or by Direct-Zol RNA MiniPrep (Zymo Research,Irvine, CA, USA). Total RNA (0.5 μg) was reverse transcribed withPrimeScript RT Reagent Kit-Perfect Real-Time (Takara Clontech,Mountain View, CA, USA). Quantitative real-time PCR was performedin a Miniopticon RT-PCR system Version 3.1 (Bio-Rad Laboratories,Inc., Hercules, CA, USA) using the KAPA SYBR FAST qPCR Kit (KapaBiosystems, Wilmington, MA USA) and primers for human LPIN1,GPAT3, GPAT4, AGPAT1, AGPAT2, DGAT1 or DGAT2 (Table 1). mRNAlevels were normalized to human β-actin mRNA. Relative gene ex-pression was calculated using the ΔΔCt method.

2.6. Total cellular protein extraction and western blot analysis

Cells were lysed in 20mM Tris-Cl pH 8.0, 150mM NaCl, 1 mMMgCl2, 1% Triton X-100, 0.5mM DTT and 0.2mM PMSF, incubated onice for 10min and cell lysates were centrifuged at 10,000g for 15min at4 °C. For AGPAT2 detection cells were lysed in 62.5 mM Tris-Cl(pH 6.8), 2.3% SDS, 10% glycerol, 0.05% (w/v) Bromophenol Blue,25 mM DTT (loading Buffer), incubated on ice for 10min, sonicated,boiled for 5min and centrifuged at 10,000g for 10min at 4 °C. Totalprotein extracts (40–80 μg) were resolved by SDS-PAGE, transferred tonitrocellulose membranes and analyzed by Western blot. Membraneswere incubated with anti-LPAATβ rabbit polyclonal antibody (1:500dilution, Santa Cruz Biotechnology, Dallas, TX, USA), or anti-AGPAT2(D8W9B) rabbit monoclonal antibody (1:500 dilution, Cell SignalingTechnology, Leiden, The Netherlands), or anti-HIF-1α rabbit polyclonalantibody (1:1000 dilution in PBS-Tween20 containing 5% milk, anti-body raised, prepared and purified as described in [41]) or anti-β-actinmouse monoclonal antibody (1:5000, Sigma-Aldrich, St Louis, MO,USA), followed by horseradish peroxidase-conjugated anti-rabbit IgG oranti-mouse IgG (1:3000 dilution, Cell Signaling Technology Inc., Dan-vers, MA, United States). Signals were visualized by the EnhancedChemiluminescence kit Luminata™ Forte/Crescento Western HRP Sub-strate (Millipore Corporation, Billerica, USA). Images and quantifica-tion of protein band intensities were obtained using an Uvitec Cam-bridge Chemiluminescence Imaging System with the help of AllianceSoftware (ver. 16.06).

2.7. Chromatin immunoprecipitation

Huh7 cells incubated under normoxia (21% O2) or hypoxia (1% O2)for 8 h, were treated with 1.1% formaldehyde for 30min, quenchedwith 125mM glycine for 5min and collected by centrifugation.Following steps were performed as previously described [13, 42] usinga previously developed polyclonal anti-HIF-1α antibody [41] or rabbit

IgG. Immunoprecipitated chromatin or input samples were amplifiedwith primers specific for regions [−1013 to −778], [−685 to −492],or [−330 to −140] of the hAGPAT2 promoter (Table 1). Ampliconswere analyzed by agarose gel electrophoresis or quantified by real-timePCR. The Ct (threshold cycle) of immunoprecipitated samples (withspecific anti-HIF-1a antibody or with non-specific IgG) was normalizedto the Ct of total input DNA (ΔCt). Results obtained with specific anti-HIF-1α antibody were then normalized to those obtained with non-specific IgG antibodies (ΔΔCt).

2.8. Fluorescence microscopy

HeLa cells, seeded on glass coverslips in 24-well plates were trans-fected with siRNA against AGPAT2 or control siRNA for 24 h and thenincubated further in normoxia (21% O2) or hypoxia (1% O2) for 24 h.Cells were fixed in PBS-3% formaldehyde for 5min at room tempera-ture, washed in PBS, and stained with Nile Red (0.1 μg/ml in PBS) for15–20min at room temperature in the dark, under agitation, in order tovisualize LDs. Coverslips were washed with PBS, mounted on slidesover 2.5 μl Mowiol mixed with DAPI, for counterstaining and viewed onan a Zeiss Axio Imager.Z2 microscope equipped with an AxioCam MRmsensor and a 100× oil-immersion lens (Zeiss Zen Software).Quantification of fluorescence was performed with the public domainsoftware for image analysis “ImageJ”. Specifically, the microscopeimage was inverted so that dark pixels become light and vice versa andthe cell region to be measured was selected. Next, the measurementthreshold was determined and the different parameters of LDs (parti-cles) analyzed were the number of LDs per cell (“count”), the percen-tage of cell area covered by LDs (“% area”) and the average LD size percell (“average size”).

2.9. Cell viability assay

HeLa cells (5000 cells/well) seeded into 96-well plates were trans-fected with siRNA against AGPAT2 or control siRNA, and 24 h laterincubated under normoxia (21% O2) or hypoxia (1% O2) for another24 h. When required, cells were treated with etoposide (75 μΜ) or withDMSO as solvent control for 5 h before the end of the incubation period,after which living cell number was determined using the “CellTiter 96Aqueous One Solution Cell Proliferation Assay” kit (PromegaCorporation, Madison, WI, USA).

2.10. Cell death assay

HeLa cells were seeded into 96-well plates (2.500 cells/well) andtransfected with siRNAs targeting AGPAT2, HIF-1α or control siRNA.Cells were cultured in DMEM supplied with 1% FBS for the duration ofthe experiment and 24 h post-transfection were incubated for the in-dicated periods of time under hypoxia (1% O2). To induce apoptosiscells were treated with 75 μΜ Etoposide for 5 h or with DMSO as sol-vent control, before the end of the incubation period. Cell death wasdetermined using the “LDH Cytotoxity Detection kit” (Takara Clontech,Mountain View, CA, USA).

2.11. Statistical analysis

Data are presented as mean ± standard deviation (SD) ormean ± standard error of the mean (SEM), as indicated. Statisticaldifferences between two groups of data were assessed using the un-paired t-test in the SigmaPlot v. 9.0 software (Systat); P < 0.05 wasconsidered to be significant.

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1144

3. Results

3.1. AGPAT2 expression is induced in cancer cell lines under hypoxia

To investigate the effect of hypoxia on the expression of genes en-coding for members of the enzyme families operating along the lipidbiosynthesis pathway, we examined the mRNA levels of representativeGPAT (GPAT3 and GPAT4), AGPAT (AGPAT1 and AGPAT2) and DGAT(DGAT1 and DGAT2) isoforms. As a positive control, we examined themRNA levels of LPIN1, a known HIF-1 target gene [13]. To this end,human hepatocellular carcinoma Huh7 cells were incubated undernormoxia (21% O2) or hypoxia (1% O2) for 8 or 24 h and mRNA levelsof TG synthesis enzymes were analyzed by quantitative-RT-PCR. Nosignificant effect was observed on the expression of GPAT or DGATisoforms under hypoxic conditions (Fig. 1S). In contrast, the LPIN1 andAGPAT2, but not AGPAT1, mRNA levels significantly increased afterincubation for 8 or 24 h under 1% O2 (Fig. 1A). Similar results wereobtained when LPIN1, AGPAT1 and AGPAT2 mRNA levels were ana-lyzed after incubation of Huh7 cells for 24 h with DMOG (1mM), aprolyl hydroxylase inhibitor that mimics hypoxia (Fig. 1B). Further-more, analysis by Western blot showed that, hypoxic treatment of Huh7

cells increased AGPAT2 protein levels in parallel to the expected in-duction of HIF-1α (Fig. 1C and Fig. 2SA). The induction of AGPAT2expression under hypoxia is not limited in hepatocellular carcinomacells, as it was also observed in human cervical adenocarcinoma HeLacells (Fig. 1D-E and Fig. 2SB).

3.2. HIF-1 mediates the hypoxic upregulation of AGPAT2

We then tested the involvement of HIF-1 in the hypoxic induction ofAGPAT2 mRNA expression, using siRNA-mediated repression of HIF-1αexpression in Huh7 cells incubated under normoxia (21% O2) or hy-poxia (1% O2). Successful knockdown of HIF-1α was confirmed bywestern blot analysis (Fig. 2A, upper panel and Fig. 2SC). Suppressionof HIF-1α expression reduced the hypoxic induction of AGPAT2 protein(Fig. 2A, lower panel) and AGPAT2 mRNA levels (Fig. 2B) in hypoxia,while a control siRNA had no effect, suggesting that HIF-1 mediates thehypoxic induction of AGPAT2 gene transcription.

In silico analysis showed that 6 putative HREs, containing theconsensus sequence NCGTG (where N is an A, G or C), are presentwithin a 1 kb AGPAT2 promoter region (Fig. 3A). To test their role inthe AGPAT2 promoter induction under hypoxia, an 873 bp AGPAT2

C

BA

E

HIF-1α

AGPAT2

β-actin

24 0

1% O2

(hs)

0

1

2

3

4

Rela

tive m

RN

A levels

8 24 0

***

***

1% O2

(hs)

***

**

+ -

Rela

tive m

RN

A levels

DMOG

0

1

2

3

*

D

HIF-1α

β-actin

AGPAT2

1% O2

(hs)

8 24 0

AGPAT1AGPAT2LPIN1

AGPAT1AGPAT2LPIN1

**

24 0

1% O2

(hs)

0

1

2

3

Rela

tive m

RN

A levels

AGPAT1AGPAT2

Fig. 1. The expression of AGPAT2 is induced under hypoxic conditions in cancer cells.(A) & (B) Quantitative real-time PCR determination of AGPAT1, AGPAT2 and LPIN1 mRNA levels in Huh7 cells incubated under hypoxia (1% O2) for 0–24 h (A) or inthe presence of DMOG (1mM) for 24 h (B), as indicated.(C) Western blotting analysis of Huh7 cells incubated at 1% O2 for 0–24 h using antibodies against the indicated proteins.(D) Quantitative real-time PCR determination of AGPAT1 and AGPAT2 mRNA levels in HeLa cells incubated under hypoxia (1% O2) for 24 h.(E) Western blotting analysis of HeLa cells incubated at 1% O2 for 24 h using antibodies against the indicated proteins.In (A), (B) and (D) values represent the mean ± SD of 3 experiments performed in duplicates (*P < 0.05, **P < 0.01 and ***P < 0.001 compared to cellsincubated for 0 h at 1% O2).

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1145

promoter region (corresponding to bases −1013 to −140) includingthe potential HREs was cloned to the pGL3 luciferase reporter plasmid.ΗΕΚ293 cells were transfected with the AGPAT2 promoter reporterplasmid, a reporter plasmid harboring the LPIN1 promoter region, aknown target of HIF-1 [13], or the empty pGL3 plasmid as a negativecontrol (Fig. 2C). In normoxia, the AGPAT2 promoter plasmid showedhigher transcriptional activity compared to the empty vector. More-over, incubation of cells under hypoxia increased significantly the ac-tivity of the AGPAT2 promoter compared to normoxia. This increasewas comparable to that of the LPIN1 promoter, which was used as apositive control (Fig. 2C), suggesting the HRE containing AGPAT2promoter is indeed stimulated under hypoxic conditions. Importantly, asimilar significant increase of AGPAT2 promoter activity was alsoachieved after overexpression of HIF-1α in HEK293 cells under nor-moxia (Fig. 2D), suggesting that AGPAT2 is a direct HIF-1 target.

The direct binding of HIF-1α to the AGPAT2 promoter was ex-amined by chromatin immunoprecipitation (ChIP). Huh7 cells werecultured under hypoxia or normoxia for 8 h, cross-linked and DNA wasimmunoprecipitated with anti-HIF-1α or rabbit IgG antibodies. Theimmunoprecipitated DNA was amplified by PCR with primers specificfor the AGPAT2 promoter regions harboring HREs 1–3, HRE 4 or HREs5–6 (Fig. 3A). Only the AGPAT2 promoter region containing HRE 4[−685 to −492] was specifically enriched in immunoprecipitates ofDNA-protein complexes isolated from cells grown under hypoxia, using

an anti-HIF-1α antibody, in comparison to rabbit IgG im-munoprecipitates or anti-HIF-1α immunoprecipitates from normoxiccells (Fig. 3B). This result was corroborated, quantitatively, by RT-PCRusing specific primers for the HRE 4 region (Fig. 3C).

To examine whether HRE 4 present within the [−685 to −492]AGPAT2 promoter region is indeed responsible for HIF-1 mediated in-duction of AGPAT2, the consensus CGTG HRE sequence was mutated toAACG in plasmid pGL3-AGPAT2 (Fig. 3D). ΗΕΚ293 cells were trans-fected with wild type or mutant AGPAT2 promoter reporter plasmids orthe empty pGL3 plasmid as a negative control. In normoxia, the mutantAGPAT2 promoter showed comparable transcriptional activity to thewild type AGPAT2 promoter. However, when cells were incubatedunder hypoxia, the activity of the mutant AGPAT2 promoter did notchange compared to normoxia and was, subsequently, significantlylower than the activity of the wild type AGPAT2 promoter (Fig. 3D).Therefore, HIF-1α not only activates but also physically associates withthe AGPAT2 promoter by binding to HRE 4. Overall our data show thatAGPAT2 is a novel direct transcriptional target of HIF-1.

3.3. Knockdown of AGPAT2 impairs LD formation, viability and etoposideresistance of cancer cells under hypoxia

Hypoxia has been previously shown to increase LD formation in aHIF-1-dependent manner [13]. To investigate the contribution of

A

C

B

siHIF-1α

siControl

21% O2

1% O2

- + - +

+ - + -

HIF-1α

β-actin

AGPAT2

AGPAT2

mR

NA

levels

01234567

***

**

***

D

Luciferase a

ctivity

Fo

ld in

cre

ase

prom

AGPAT2prom

LPIN1-

***

***

0

10

20

30

4021% Ο

2

1% Ο2

Luciferase a

ctivity

Fo

ld in

cre

ase

* **

prom

AGPAT2prom

LPIN1-

0

5

10

15 GFP

GFP-HIF-1α

21% Ο2

1% Ο2

siHIF-1α siControl

Fig. 2. The hypoxic induction of AGPAT2 expression is HIF-1 dependent.(A) & (B) Huh7 cells transfected with HIF-1α or control siRNAs (20 nM), were incubated 24 h post-transfection under normoxia (21% O2) or hypoxia (1% O2) foranother 24 h as indicated. They were then analyzed by Western blotting analysis using the indicated antibodies (A) or by quantitative real-time PCR for thedetermination of AGPAT2 mRNA levels (B). Values in (B) represent the mean ± SD of 3 experiments performed in duplicates (**P < 0.01 and ***P < 0.001).(C) Luciferase activity in extracts of HEK293 cells transfected with the empty pGL3 plasmid (−) or a pGL3 construct of the human LPIN1 promoter (promLPIN1) or apGL3 construct of the human AGPAT2 promoter (promAGPAT2) and a pCI-Renilla reporter and incubated under 21% or 1% O2 for 16 h.(D) Luciferase activity in extracts of HEK293 cells co-transfected with a GFP or a GFP-HIF-1α overexpressing plasmid and the reporter plasmids described in (C) andincubated under normoxia.Values in (C) and (D) are expressed in relation to the empty pGL3 vector and represent the mean ± SEM of 4 experiments performed in triplicates. (*P < 0.05,**P < 0.01, ***P < 0.001).

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1146

AGPAT2 upregulation in this metabolic adaptation, we measured LDaccumulation in HeLa cells treated with siRNA to AGPAT2 and in-cubated under normoxia (21% O2) or hypoxia (1% O2). To confirmsuppression of AGPAT2 expression, AGPAT2 mRNA levels were de-termined by quantitative-RT-PCR (Fig. 4A) and AGPAT2 protein levelsby western blot analysis (Fig. 4B), respectively. Under the same con-ditions, we did not detect any changes of the AGPAT1 mRNA levels,both under normoxia or hypoxia (Fig. 3S), suggesting that regulation ofAGPAT1 expression is independent of AGPAT2 expression, at least inHeLa cells. As expected, fluorescence microscopy of cells stained with

Nile Red to visualize neutral lipids, showed a profound accumulation ofLDs under hypoxic treatment in cells transfected with a control siRNA(Fig. 4C). However, LD accumulation under hypoxia was significantlyimpaired after AGPAT2 knockdown. Quantitative analysis showed thatsilencing of AGPAT2 significantly reduced LD number per cell (Fig. 4D,upper panel), the average cell area covered by LDs (Fig. 4D, middlepanel) and mean LD size under hypoxia (Fig. 4D, lower panel). Thesefindings show that AGPAT2 is required for the hypoxic induction of LDaccumulation.

We then examined whether this function of AGPAT2 is important

A

CB

AGPAT2 promoter

D

CHIP primers

HRE 1-3HRE 4HRE 5-6

-685 -491-1013 -778 -329 -140

6 5 4 3 2 1 AGPAT2 HRE

1 % O2

21 % O2

anti-HIF-1α

IgG

INPUT ChIP

+- -+ - +

- -+ + + -

- - +- +-

- - -+ + -

HRE 4

HRE 5-6

HRE 1-3

Fo

ld e

nric

hm

en

t

IgG anti-HIF-1α

***

05

10152025

HRE 4

21% Ο2

1% Ο2

-563 ..CCTTT GCGTG GCCAC.. -548

-563 ..CCTTT GAACG GCCAC.. -548

wt

mut

HRE4 *

Luciferase a

ctivity

Fo

ld in

cre

ase

prom

AGPAT2mut

-

NS

21% Ο2

1% Ο2

NS

prom

AGPAT2wt

0

5

10

15

20

*

Fig. 3. HIF-1α binds to and activates the AGPAT2 promoter.(A) Schematic representation of a 1000 bp region upstream of the human AGPAT2 gene. The positions of the three set of primers (arrows) used for the ChIP assay withthe corresponding amplicons and the potential HREs (triangles) are indicated.(B) Chromatin immunoprecipitation (ChIP) from Huh7 cells incubated for 8 h under normoxia (21% O2) or hypoxia (1% O2), using anti-HIF-1α or rabbit IgGantibodies (negative control). Input and immunoprecipitated DNA were amplified by PCR using primers specific for indicated areas of the AGPAT2 promoter andanalyzed by gel electrophoresis.(C) Quantitative real-time PCR analysis of the ChIP described in (B) using primers specific for the promoter region containing HRE 4. The data represent themean ± SEM of 2 experiments performed in triplicates (***P < 0.001).(D) Left: Sequence of the human AGPAT2 promoter region surrounding HRE 4. The sequence of HRE 4 and its mutated form are depicted in a box. Right: Luciferaseactivity in extracts of HEK293 cells transfected with the empty pGL3 plasmid (−) or pGL3 constructs of the human wild type (prom AGPAT2 wt) or mutant (promAGPAT2 mut) promoter of AGPAT2 and a pCI-Renilla reporter and incubated under 21% or 1% O2 for 16 h. Values are expressed in relation to the empty pGL3 vectorand represent the mean ± SEM of 3 experiments performed in triplicates (*P < 0.05, **P < 0.01, NS; non-significant).

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1147

A

C

AGPAT2

mRN

A le

vels

0.0

0.5

1.0

1.5

2.0

1% O221% O2

***

***

***

*

D

siCon

trol

Nile Red DAPI

siAG

PAT2

siCon

trol

siAGP

AT2

LDs/

cell

1% O221% O20

10

20

30

40

50

**

*** ******

% C

ell A

rea

1% O221% O20

1

2

3

4 ***

NS

******

Aver

age

LD si

ze

1% O221% O20

50

100

150

200***

NS

NS***

-ac�n

siAGPAT2siControl

HIF-1

21% O2 1% O2

- + - ++ - + -

AGPAT2

BsiControlsiAGPAT2

siControlsiAGPAT2

Fig. 4. AGPAT2 silencing impairs hypoxic induction of LD accumulation and growth.HeLa cells were transfected with siRNAs against AGPAT2 (siAGPAT2) or control (siControl) and incubated 24 h post-transfection for another 24 h under normoxia(21% O2) or hypoxia (1% O2).(A) Quantitative real-time PCR determination of AGPAT2 mRNA levels. Values represent the mean ± SD of 4 independent experiments performed in duplicates(*P < 0.05, ***P < 0.001).(B) Western blotting analysis of HIF-1α and AGPAT2 protein levels in cells treated as above.(C) Fluorescence microscope images of cells treated as above and stained with Nile Red to visualize LDs. Cell nuclei were stained with DAPI.(D) Quantitative analysis of fluorescence microscope images of cells presented in (C) showing mean number of LDs per cell (upper panel), average percentage of cellarea occupied by LDs (middle panel) and mean LD size (lower panel). Data represent the mean ± SEM of 100 cells. (**P < 0.01, ***P < 0.001, NS non-sig-nificant).

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1148

for cancer cell survival under hypoxia. AGPAT2 silencing significantlyimpaired the expression of AGPAT2 protein both under normoxia andhypoxia (Fig. 5A). Analysis of HeLa cell numbers after AGPAT2 silen-cing showed that suppression of AGPAT2 expression reduced cell via-bility under hypoxia compared to control siRNA-treated cells, althoughviability of normoxic cells under the same conditions was not sig-nificantly affected (Fig. 5B and Fig. 4S). Moreover, silencing of AGPAT2

increased the sensitivity of cancer cells to the apoptotic agent etopo-side, an effect that was much more pronounced under hypoxic condi-tions (Fig. 5B and Fig. 4S). To evaluate the contribution of AGPAT2 toHIF-1-dependent survival under hypoxia, we measured cell death afterAGPAT2 or HIF-1α knockdown (Fig. 5C). Silencing of AGPAT2 in-creased cell death under normoxia as well as under hypoxia in both theabsence and presence of etoposide. As expected, knockdown of HIF-1α

B

% C

ell s

urviv

al

0

20

40

60

80

100

120

140

160

21% O2

1% O2

21% O2

1% O2

- Etop

NS

NS

***

+ Etop

*

NS

***

***

***

0

10

20

30

40

50

% C

ell d

eath

0h 24h

- Etop

1% O2

***

NS

***

***

24h

+ Etop

***

*

***

**

***

C

Α

siControl

siAGPAT2

siHIF-1α

siControl

siAGPAT2

siControl

siAGPAT2

siHIF-1α

β-actin

HIF-1α

AGPAT2

siAGPAT2

-- + -

siControl

21% O2

1% O2

-+

-

siHIF-1α

-

+-

- -

-

- +

-+

+

AG

PA

T2

/ β

actin

ra

tio

1% O2

21% O2

NS

**

**

****

***

***

HIF

-1

α/

β a

ctin

ra

tio

1% O2

NS

***

0

0.4

0.8

1.2

1.6

0

0.5

1

1.5

2

2.5

3

Fig. 5. Silencing of AGPAT2 impairs cancer cell viability and etoposide resistance under hypoxia.(A) Left: HeLa cells transfected with siRNAs against AGPAT2 (siAGPAT2), HIF-1α (si HIF-1α) or control (siControl) and incubated 24 h post-transfection for another24 h under normoxia (21% O2) or hypoxia (1% O2) were analyzed by western blotting with the indicated antibodies. Right: Quantitative analysis of HIF-1α andAGPAT2 protein levels from the western blot analysis shown on the left. Values represent the mean ± SD of 3 experiments (**P < 0.01, ***P < 0.001, NS non-significant).(B) Viability of HeLa cells transfected with siRNAs against AGPAT2 (siAGPAT2) or control siRNAs (siControl) and, 24 h post-transfection, incubated for another 24 hunder normoxia (21% O2) or hypoxia (1% O2). Where indicated, etoposide (Etop, 75 μM) was added to the cell medium for the last 5 h of the hypoxic incubation.Values represent percentage of cells in relation to siControl/21% O2/−Etop cells and are shown as the mean ± SD of 4 experiments performed in triplicate(*P < 0.05, ***P < 0.001, NS non-significant).(C) HeLa cells were transfected with siRNAs against AGPAT2 (siAGPAT2), HIF-1α (si HIF-1α) or control siRNAs (siControl) and cell death was determined 24 h post-transfection at normoxia (0 h) or after incubation for another 24 h under hypoxia (24 h). Where indicated, etoposide (Etop, 75 μM) was added to the cell medium forthe last 5 h of the hypoxic incubation. Values represent percentage of cells in relation to siControl/0 h cells and are shown as the mean ± SD of 4 experimentsperformed in duplicate. (*P < 0.05, **P < 0.01, ***P < 0.001, NS non-significant).

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1149

increased cell death only under hypoxia. The effects of HIF-1α knock-down under hypoxia were more pronounced than those of AGPAT2knockdown, as also to be expected by the much wider role of HIF-1 incellular adaptation to hypoxia. Nevertheless, comparison of the datasuggests that AGPAT2 contributes substantially to HIF-1-dependentsurvival under low oxygen conditions.

Taken together, our data suggest that HIF-1-mediated induction ofAGPAT2 facilitates metabolic adaptation, survival and chemoresistanceof cancer cells under hypoxic conditions (Fig. 6).

4. Discussion

Several studies have previously reported that AGPAT2 expression iselevated in cancer cell lines or human tumors and is associated withincreased cancer cell proliferation and migration, tumor growth oraggressiveness and worse prognosis [30, 32, 34]. However, despite theimportant role of AGPAT2 in TG biosynthesis and its link to patholo-gical conditions, the mechanisms responsible for regulation of AGPAT2gene transcription have so far remained largely unknown. We now re-port for the first time that AGPAT2 is a direct target of HIF-1 andAGPAT2 induction under hypoxia is required for cancer cell survivaland chemoresistance. HIF-1α is often upregulated in solid tumors as aresult of the hypoxic microenvironment. In addition, tumor HIF-1αlevels can also be increased by oxygen independent mechanisms underthe influence of growth factors and cell signaling pathways that arestimulated in cancer cells [9]. Therefore, our finding that HIF-1 is re-sponsible for the activation of AGPAT2 gene transcription, offers aplausible explanation for the observed upregulation of AGPAT2 ex-pression in tumors and cancer cells. Our in vitro data are also inagreement with previous reports showing strong correlation of AGPAT2expression with the expression of heme oxygenase 1 (HO-1), a knownHIF-1 target gene, in a comparative analysis of human tumors [33].Moreover, kaempferol, a flavonoid that was previously shown to inhibitHIF-1 activity under hypoxia [43], was also found to decrease AGPAT2expression in 3 T3-L1 preadipocytes [44].

AGPAT2 belongs to a family of several isoforms that catalyze theconversion of LPA to PA in the triglyceride and phospholipid bio-synthesis pathway [25]. We have shown that, unlike AGPAT2, the ex-pression of its closest homologue AGPAT1 is not increased under hy-poxia or after treatment with the PHD inhibitor DMOG. Similarselectivity has also been shown regarding the hypoxic induction of li-pins, the enzymes acting downstream of AGPAT2 in the TG biosynthesis

pathway, as only lipin 1, but not its isoform lipin 2, is upregulated byHIF-1 and plays a role in hypoxic adaptation [13, 24]. AGPAT1 andAGPAT2 exhibit diverse expression levels in different tissues, showdistinct knockout phenotypes and cannot complement each other [28].Importantly, AGPAT1, which is more ubiquitously expressed thanAGPAT2 and appears to serve important functions in the physiology ofmany organ systems [45–47], is not overexpressed in tumors and cancercell lines [48], emphasizing the specificity of the HIF-1-mediated up-regulation of AGPAT2 and its significance for cancer cell adaptationunder hypoxia. The importance of AGPAT2 for cancer cells is furtherunderlined by the fact that AGPAT2 inhibitors and miRNAs targetingAGPAT2 inhibit cell proliferation and migration and increase cell sen-sitivity to cytotoxic drugs [31, 35–40, 49, 50]. This is consistent withour data showing that AGPAT2 induction under hypoxia sustains cancercell survival and increases their chemoresistance.

An interesting question is how induction of AGPAT2 supports cancercell adaptation to low oxygen conditions. We have shown that upre-gulation of AGPAT2 is required for LD accumulation under hypoxia. Itis well established that regulation of lipid metabolism and storage arevery important for cancer cells and the related pathways represent validtherapeutic targets [51]. Moreover, HIF-1 driven TG synthesis and LDaccumulation is critical for cancer cell survival and proliferation underhypoxia [24], as storage of fatty acids as neutral lipids in LDs canprotect hypoxic cancer cells from lipotoxicity and ROS production [23].Indeed, hypoxia induces additional genes directly involved in lipidstorage and LD formation. These include, besides lipin 1 [13], the nonenzymatic proteins HIG2 [12] and fatty acid binding proteins FABP3and FABP 7 [17], which are regulated by HIF-1, as well as perilipin 2,which is induced by HIF-2 [19]. Uptake of fatty acids, synthesis ofneutral lipids and their storage in LDs under hypoxia are processesrequired for avoiding endoplasmic reticulum stress [19] and damage byROS [17] or for providing energy after reoxygenation [17], and, assuch, they may be essential for cancer cell proliferation in the hostilemicroenvironment of solid tumors.

An alternative role of AGPAT2 induction under hypoxia may berelated to the production of PA. PA, apart from its role as an inter-mediate in TG and phospholipid synthesis, is also involved in cell sig-naling via the activation of the mTOR pathway [52]. The mTOR sig-naling pathway affects the cell response to hypoxia as active mTORC1increases synthesis of HIF-1α. On the other hand, hypoxia inhibitsmTORC1 activation [4]. It is, thus, also possible that the effect ofAGPAT2 silencing on cell survival under hypoxia reflects a reduction of

AGPAT2

HIF-1

AR

NT

HIF

-1

Cell survival

Etoposide resistanceLD accumulation

HYPOXIA

ER

Nucleus

LPA

PA

DAG

TAG

G-3-P + FA-CoA

GPAT

DGAT

AGPAT2

LIPIN 1

HRE

LPIN1AR

NT

HIF

-1

HRE4

-558/-554

Fig. 6. Model of regulation of AGPAT2 expressionand LD formation by hypoxia and HIF-1.Members of the GPAT, AGPAT, lipin and DGAT en-zyme families reside at the ER and mediate TAGbiosynthesis (adapted from [21]). Up-regulation ofexpression of AGPAT2 (this work) and LPIN1 [13] byHIF-1 promotes LD accumulation and may supportcell viabiity and chemoresistance of cancer cellsunder hypoxic conditions.HRE: hypoxia response elements; ER: endoplasmaticreticulum; G-3-P: glycerol-3-phosphate; FA-CoA:fatty acyl-CoA; LPA: lysophosphatidic acid; PA:phosphatidic acid; DAG: diacylglycerol; TAG: tria-cylglycerol; LD: Lipid Droplet.

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1150

cellular PA concentration and, subsequently, mTOR activity, as waspreviously shown in pancreatic cancer cells under normoxia [26].However, PA can be produced in cells through the action of two ad-ditional enzymes, phospholipase D (PLD) and diacylglycerol kinase-ζ(DGKζ), which can compensate for the downregulation of AGPAT2[52].

5. Conclusion

In conclusion, AGPAT2 is specifically and directly upregulated byHIF-1 in cancer cells and promotes their survival and chemoresistanceunder hypoxia. Its identification, following that of lipin 1, as a secondTG synthesis enzyme under HIF-1 transcriptional control, further es-tablishes the strong and multifaceted involvement of HIF-1 in lipidsynthesis and storage and their connection to cancer cell physiology(Fig. 6). Additionally, it suggests that these lipogenic enzymes may beused as prognostic markers and/or novel therapeutic targets of hypoxictumors.

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbalip.2018.06.015.

Funding

This work was partially supported by grants provided by the“ARISTEIA ΙΙ” Action of the “OPERATIONAL PROGRAMME EDUCAT-ION AND LIFELONG LEARNING” co-funded by the European SocialFund (ESF) and National Resources (Code: 3129, HYPOXYTARGET, toG.S.) and by the Research Committee of the University of Thessaly(Contract Nr 5309.17.03, to E.P.). E.-A.T. was supported by a fellowshipprovided by the Hellenic State Scholarship Foundation (IKY), in thecontext of the scholarship program for postgraduate studies using re-sources of the OP “Human Resources Development, Education andLifelong Learning 2014-2020” co-financed by the ESF and the GreekState.

Transparency document

The Transparency document associated with this article can befound, in online version.

References

[1] E.B. Rankin, A.J. Giaccia, Hypoxic control of metastasis, Science 352 (2016)175–180.

[2] G.L. Semenza, Hypoxia-inducible factors: mediators of cancer progression andtargets for cancer therapy, Trends Pharmacol. Sci. 33 (2012) 207–214.

[3] M.C. Brahimi-Horn, G. Bellot, J. Pouyssegur, Hypoxia and energetic tumour me-tabolism, Curr. Opin. Genet. Dev. 21 (2011) 67–72.

[4] L. Schito, S. Rey, Cell-autonomous metabolic reprogramming in hypoxia, TrendsCell Biol. 28 (2017) 128–142.

[5] C.W. Pugh, P.J. Ratcliffe, New horizons in hypoxia signaling pathways, Exp. CellRes. 356 (2017) 116–121.

[6] V.L. Dengler, M. Galbraith, J.M. Espinosa, Transcriptional regulation by hypoxiainducible factors, Crit. Rev. Biochem. Mol. Biol. 49 (2014) 1–15.

[7] A. Kalousi, I. Mylonis, A.S. Politou, G. Chachami, E. Paraskeva, G. Simos, Caseinkinase 1 regulates human hypoxia-inducible factor HIF-1, J. Cell Sci. 123 (2010)2976–2986.

[8] T. Kietzmann, D. Mennerich, E.Y. Dimova, Hypoxia-inducible factors (HIFs) andphosphorylation: impact on stability, localization, and transactivity, Front. CellDev. Biol. 4 (2016) 11.

[9] S. Koyasu, M. Kobayashi, Y. Goto, M. Hiraoka, H. Harada, Regulatory mechanismsof hypoxia-inducible factor 1 activity: two decades of knowledge, Cancer Sci. 109(2017) 560–571.

[10] A.I. Papadakis, E. Paraskeva, P. Peidis, H. Muaddi, S. Li, L. Raptis, K. Pantopoulos,G. Simos, A.E. Koromilas, eIF2{alpha} kinase PKR modulates the hypoxic responseby Stat3-dependent transcriptional suppression of HIF-1{alpha}, Cancer Res. 70(2010) 7820–7829.

[11] I. Mylonis, G. Chachami, E. Paraskeva, G. Simos, Atypical CRM1-dependent nuclearexport signal mediates regulation of hypoxia-inducible factor-1alpha by MAPK, J.Biol. Chem. 283 (2008) 27620–27627.

[12] T. Gimm, M. Wiese, B. Teschemacher, A. Deggerich, J. Schodel, K.X. Knaup,T. Hackenbeck, C. Hellerbrand, K. Amann, M.S. Wiesener, S. Honing, K.U. Eckardt,C. Warnecke, Hypoxia-inducible protein 2 is a novel lipid droplet protein and a

specific target gene of hypoxia-inducible factor-1, FASEB J. 24 (2010) 4443–4458.[13] I. Mylonis, H. Sembongi, C. Befani, P. Liakos, S. Siniossoglou, G. Simos, Hypoxia

causes triglyceride accumulation by HIF-1-mediated stimulation of lipin 1 expres-sion, J. Cell Sci. 125 (2012) 3485–3493.

[14] E. Furuta, S.K. Pai, R. Zhan, S. Bandyopadhyay, M. Watabe, Y.Y. Mo, S. Hirota,S. Hosobe, T. Tsukada, K. Miura, S. Kamada, K. Saito, M. Iiizumi, W. Liu, J. Ericsson,K. Watabe, Fatty acid synthase gene is up-regulated by hypoxia via activation of Aktand sterol regulatory element binding protein-1, Cancer Res. 68 (2008) 1003–1011.

[15] R.C. Sun, N.C. Denko, Hypoxic regulation of glutamine metabolism through HIF1and SIAH2 supports lipid synthesis that is necessary for tumor growth, Cell Metab.19 (2014) 285–292.

[16] T. Huang, X. Li, L. Li, L. Zhang, X. Sun, X. He, D. Zhong, L. Jia, G.L. Song,P. Semenza, H. Zhang Gao, HIF-1-mediated suppression of acyl-CoA dehy-drogenases and fatty acid oxidation is critical for cancer progression, Cell Rep. 8(2014) 1930–1942.

[17] K. Bensaad, E. Favaro, C.A. Lewis, B. Peck, S. Lord, J.M. Collins, K.E. Pinnick,S. Wigfield, F.M. Buffa, J.L. Li, Q. Zhang, M.J. Wakelam, F. Karpe, A. Schulze,A.L. Harris, Fatty acid uptake and lipid storage induced by HIF-1alpha contribute tocell growth and survival after hypoxia-reoxygenation, Cell Rep. 9 (2014) 349–365.

[18] B. Hu, Y. Guo, W.G. Garbacz, M. Jiang, M. Xu, H. Huang, A. Tsung, T.R. Billiar,S.K. Ramakrishnan, Y.M. Shah, K.S. Lam, M. Huang, W. Xie, Fatty acid bindingprotein-4 (FABP4) is a hypoxia inducible gene that sensitizes mice to liverischemia/reperfusion injury, J. Hepatol. 63 (2015) 855–862.

[19] B. Qiu, D. Ackerman, D.J. Sanchez, B. Li, J.D. Ochocki, A. Grazioli,E. Bobrovnikova-Marjon, J.A. Diehl, B. Keith, M.C. Simon, HIF2alpha-dependentlipid storage promotes endoplasmic reticulum homeostasis in clear-cell renal cellcarcinoma, Cancer Discov. 5 (2015) 652–667.

[20] T. Suzuki, S. Shinjo, T. Arai, M. Kanai, N. Goda, Hypoxia and fatty liver, World J.Gastroenterol. 20 (2014) 15087–15097.

[21] H. Wang, M.V. Airola, K. Reue, How lipid droplets “TAG” along: glycerolipid syn-thetic enzymes and lipid storage, Biochim. Biophys. Acta 1862 (2017) 1131–1145.

[22] T.C. Walther, R.V. Farese Jr., Lipid droplets and cellular lipid metabolism, Annu.Rev. Biochem. 81 (2012) 687–714.

[23] W. Yoo, K.H. Noh, J.H. Ahn, J.H. Yu, J.A. Seo, S.G. Kim, K.M. Choi, S.H. Baik,D.S. Choi, T.W. Kim, H.J. Kim, N.H. Kim, HIF-1alpha expression as a protectivestrategy of HepG2 cells against fatty acid-induced toxicity, J. Cell. Biochem. 115(2014) 1147–1158.

[24] M. Kourti, G. Ikonomou, N.N. Giakoumakis, M.A. Rapsomaniki, U. Landegren,S. Siniossoglou, Z. Lygerou, G. Simos, I. Mylonis, CK1delta restrains lipin-1 induc-tion, lipid droplet formation and cell proliferation under hypoxia by reducing HIF-1alpha/ARNT complex formation, Cell. Signal. 27 (2015) 1129–1140.

[25] R.M. Bradley, R.E. Duncan, The lysophosphatidic acid acyltransferases (acylgly-cerophosphate acyltransferases) family: one reaction, five enzymes, many roles,Curr. Opin. Lipidol. 29 (2018) 110–115.

[26] M.A. Blaskovich, V. Yendluri, H.R. Lawrence, N.J. Lawrence, S.M. Sebti,G.M. Springett, Lysophosphatidic acid acyltransferase beta regulates mTOR sig-naling, PLoS One 8 (2013) e78632.

[27] A.K. Agarwal, E. Arioglu, S. De Almeida, N. Akkoc, S.I. Taylor, A.M. Bowcock,R.I. Barnes, A. Garg, AGPAT2 is mutated in congenital generalized lipodystrophylinked to chromosome 9q34, Nat. Genet. 31 (2002) 21–23.

[28] A.K. Agarwal, Lysophospholipid acyltransferases: 1-acylglycerol-3-phosphate O-acyltransferases. From discovery to disease, Curr. Opin. Lipidol. 23 (2012)290–302.

[29] N. Patni, A. Garg, Congenital generalized lipodystrophies—new insights into me-tabolic dysfunction, Nat. Rev. Endocrinol. 11 (2015) 522–534.

[30] S. Niesporek, C. Denkert, W. Weichert, M. Kobel, A. Noske, J. Sehouli, J.W. Singer,M. Dietel, S. Hauptmann, Expression of lysophosphatidic acid acyltransferase beta(LPAAT-beta) in ovarian carcinoma: correlation with tumour grading and prog-nosis, Br. J. Cancer 92 (2005) 1729–1736.

[31] G.M. Springett, L. Bonham, A. Hummer, I. Linkov, D. Misra, C. Ma, G. Pezzoni, S. DiGiovine, J. Singer, H. Kawasaki, D. Spriggs, R. Soslow, J. Dupont, Lysophosphatidicacid acyltransferase-beta is a prognostic marker and therapeutic target in gyneco-logic malignancies, Cancer Res. 65 (2005) 9415–9425.

[32] C.S. Diefenbach, R.A. Soslow, A. Iasonos, I. Linkov, C. Hedvat, L. Bonham, J. Singer,R.R. Barakat, C. Aghajanian, J. Dupont, Lysophosphatidic acid acyltransferase-beta(LPAAT-beta) is highly expressed in advanced ovarian cancer and is associated withaggressive histology and poor survival, Cancer 107 (2006) 1511–1519.

[33] S. Tauber, A. Jais, M. Jeitler, S. Haider, J. Husa, J. Lindroos, M. Knofler,M. Mayerhofer, H. Pehamberger, O. Wagner, M. Bilban, Transcriptome analysis ofhuman cancer reveals a functional role of heme oxygenase-1 in tumor cell adhesion,Mol. Cancer 9 (2010) 200.

[34] F. Rastegar, J.L. Gao, D. Shenaq, Q. Luo, Q. Shi, S.H. Kim, W. Jiang, E.R. Wagner,E. Huang, Y. Gao, J. Shen, K. Yang, B.C. He, L. Chen, G.W. Zuo, J. Luo, X. Luo, Y. Bi,X. Liu, M. Li, N. Hu, L. Wang, G. Luther, H.H. Luu, R.C. Haydon, T.C. He,Lysophosphatidic acid acyltransferase beta (LPAATbeta) promotes the tumorgrowth of human osteosarcoma, PLoS One 5 (2010) e14182.

[35] T. Hideshima, D. Chauhan, T. Hayashi, K. Podar, M. Akiyama, C. Mitsiades, M.I. N,B. Gong, L. Bonham, P. de Vries, N. Munshi, P.G. Richardson, J.W. Singer,K.C. Anderson, Antitumor activity of lysophosphatidic acid acyltransferase-betainhibitors, a novel class of agents, in multiple myeloma, Cancer Res. 63 (2003)8428–8436.

[36] P. La Rosee, T. Jia, S. Demehri, N. Hartel, P. de Vries, L. Bonham, D. Hollenback,J.W. Singer, J.V. Melo, B.J. Druker, M.W. Deininger, Antileukemic activity of ly-sophosphatidic acid acyltransferase-beta inhibitor CT32228 in chronic myelo-genous leukemia sensitive and resistant to imatinib, Clin. Cancer Res. 12 (2006)6540–6546.

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1151

[37] L. Song, P. Duan, Y. Gan, P. Li, C. Zhao, J. Xu, Z. Zhang, Q. Zhou, MicroRNA-340-5pmodulates cisplatin resistance by targeting LPAATbeta in osteosarcoma, Braz. J.Med. Biol. Res. 50 (2017) e6359.

[38] L. Song, P. Duan, Y. Gan, P. Li, C. Zhao, J. Xu, Z. Zhang, Q. Zhou, SilencingLPAATbeta inhibits tumor growth of cisplatin-resistant human osteosarcoma in vivoand in vitro, Int. J. Oncol. 50 (2017) 535–544.

[39] L. Song, J. Yang, P. Duan, J. Xu, X. Luo, F. Luo, Z. Zhang, T. Hou, B. Liu, Q. Zhou,MicroRNA-24 inhibits osteosarcoma cell proliferation both in vitro and in vivo bytargeting LPAATbeta, Arch. Biochem. Biophys. 535 (2013) 128–135.

[40] T. Hideshima, D. Chauhan, K. Ishitsuka, H. Yasui, N. Raje, S. Kumar, K. Podar,C. Mitsiades, H. Hideshima, L. Bonham, N.C. Munshi, P.G. Richardson, J.W. Singer,K.C. Anderson, Molecular characterization of PS-341 (bortezomib) resistance: im-plications for overcoming resistance using lysophosphatidic acid acyltransferase(LPAAT)-beta inhibitors, Oncogene 24 (2005) 3121–3129.

[41] A. Lyberopoulou, E. Venieris, I. Mylonis, G. Chachami, I. Pappas, G. Simos,S. Bonanou, E. Georgatsou, MgcRacGAP interacts with HIF-1alpha and regulates itstranscriptional activity, Cell. Physiol. Biochem. 20 (2007) 995–1006.

[42] G.G. Braliou, M.V. Verga Falzacappa, G. Chachami, G. Casanovas,M.U. Muckenthaler, G. Simos, 2-Oxoglutarate-dependent oxygenases control hep-cidin gene expression, J. Hepatol. 48 (2008) 801–810.

[43] I. Mylonis, A. Lakka, A. Tsakalof, G. Simos, The dietary flavonoid kaempferol ef-fectively inhibits HIF-1 activity and hepatoma cancer cell viability under hypoxicconditions, Biochem. Biophys. Res. Commun. 398 (2010) 74–78.

[44] U.H. Park, J.C. Jeong, J.S. Jang, M.R. Sung, H. Youn, S.J. Lee, E.J. Kim, S.J. Um,Negative regulation of adipogenesis by kaempferol, a component of rhizomapolygonati falcatum in 3T3-L1 cells, Biol. Pharm. Bull. 35 (2012) 1525–1533.

[45] A.K. Agarwal, S. Sukumaran, V.A. Cortes, K. Tunison, D. Mizrachi, S. Sankella,R.D. Gerard, J.D. Horton, A. Garg, Human 1-acylglycerol-3-phosphate O-acyl-transferase isoforms 1 and 2: biochemical characterization and inability to rescue

hepatic steatosis in Agpat2(−/−) gene lipodystrophic mice, J. Biol. Chem. 286(2011) 37676–37691.

[46] A.K. Agarwal, K. Tunison, J.S. Dalal, S.S. Nagamma, F.K. Hamra, S. Sankella,X. Shao, R.J. Auchus, A. Garg, Metabolic, reproductive, and neurologic abnormal-ities in Agpat1-null mice, Endocrinology 158 (2017) 3954–3973.

[47] V.A. Cortes, D.E. Curtis, S. Sukumaran, X. Shao, V. Parameswara, S. Rashid,A.R. Smith, J. Ren, V. Esser, R.E. Hammer, A.K. Agarwal, J.D. Horton, A. Garg,Molecular mechanisms of hepatic steatosis and insulin resistance in the AGPAT2-deficient mouse model of congenital generalized lipodystrophy, Cell Metab. 9(2009) 165–176.

[48] L. Bonham, D.W. Leung, T. White, D. Hollenback, P. Klein, J. Tulinsky, M. Coon,P. de Vries, J.W. Singer, Lysophosphatidic acid acyltransferase-beta: a novel targetfor induction of tumour cell apoptosis, Expert Opin. Ther. Targets 7 (2003)643–661.

[49] M. Coon, A. Ball, J. Pound, S. Ap, D. Hollenback, T. White, J. Tulinsky, L. Bonham,D.K. Morrison, R. Finney, J.W. Singer, Inhibition of lysophosphatidic acid acyl-transferase beta disrupts proliferative and survival signals in normal cells and in-duces apoptosis of tumor cells, Mol. Cancer Ther. 2 (2003) 1067–1078.

[50] J.M. Pagel, C. Laugen, L. Bonham, R.C. Hackman, D.M. Hockenbery, R. Bhatt,D. Hollenback, H. Carew, J.W. Singer, O.W. Press, Induction of apoptosis usinginhibitors of lysophosphatidic acid acyltransferase-beta and anti-CD20 monoclonalantibodies for treatment of human non-Hodgkin's lymphomas, Clin. Cancer Res. 11(2005) 4857–4866.

[51] T. Mashima, H. Seimiya, T. Tsuruo, De novo fatty-acid synthesis and relatedpathways as molecular targets for cancer therapy, Br. J. Cancer 100 (2009)1369–1372.

[52] D.A. Foster, D. Salloum, D. Menon, M.A. Frias, Phospholipase D and the main-tenance of phosphatidic acid levels for regulation of mammalian target of rapa-mycin (mTOR), J. Biol. Chem. 289 (2014) 22583–22588.

E.-A. Triantafyllou et al. BBA - Molecular and Cell Biology of Lipids 1863 (2018) 1142–1152

1152