Coordinate regulation of stress signaling and epigenetic events by Acss2 and HIF-2 in cancer cells

Survival of cancer cells in the harsh tumor microenvironment, characterized by oxygen and glucose deprivation, requires rapid initiation of cytoprotective measures. Metabolites whose levels change during stress are ideal signaling cues, particularly if used in post-translational modifications of stress-responsive signal transducers. In cancer cells exposed to oxygen or glucose deprivation, there is an increase in cellular levels of acetate, a substrate for acetate-dependent acetyl CoA synthetase 2 (Acss2) that also stimulates translocation of Acss2 from the cytosol to the nucleus. Nuclear, but not cytosolic, Acss2 promotes acetylation of the stress-responsive Hypoxia Inducible Factor 2α (HIF-2α) subunit by the acetyltransferase/coactivator Creb binding protein (Cbp), a process that facilitates stable Cbp/HIF-2α complex formation. In addition to promoting de novo transcription, Cbp and HIF-2α act in concert to regulate local histone 3 epigenetic marks. Exogenous acetate augments Acss2/HIF-2 dependent cancer growth and metastasis in cell culture and mouse models. Thus, an acetate switch in mammals links nutrient intake and stress signaling with tumor growth and metastasis.


Introduction
The ability to sense and respond to external stress is a requisite property of all living organisms. Diverse environmental stresses encountered in vivo oftentimes impinge upon specific genetic regulators to promote cell survival. These genetic regulators, in turn, are frequently controlled by post-translational modifications induced by environmental stress. Linking changes in cellular metabolism to signal transduction via stress-dependent post-translational modifications of genetic regulators allows for a prompt response to environmental stress at the gene expression level.
F1541A) carboxy terminal c-myc tagged mouse CBP expression constructs were prepared by PCR cloning.
Based upon similar constructs used for efficient expression of shRNA [12] or for efficient knockdown [13], we used pLenti6/V5-GW/lacZ (Invitrogen, Life Technologies, Grand Island, NY) to generate lentiviral (LTV) CMV promoter-driven expression vectors that harbor a DsRed (for knockdown only cell studies) or firefly luciferase (for knockdown mouse flank tumor studies) cDNA in place of the parental lacZ-encoding cDNA, followed by a polylinker containing a concatamer of four different shRNAs for the indicated gene. The control expression cDNA used in the LTV knockdown experiments encodes DsRed (for knockdown only cell studies) or firefly luciferase (for knockdown mouse flank tumor studies) cDNA, followed by a tandem pair of two control shRNAs that generates a concatamer of four shRNAs.
For knockdown/rescue experiments, pLenti6 derivatives were constructed that generate bicistronic cDNA with a human HIF-2α or mouse Acss2 cDNA, which were modified to encode an mRNA resistant to human HIF-2α-selective or human Acss2-selective siRNA or shRNA, placed upstream of an internal ribosome entry site (IRES) and followed by the luciferase-shRNA multimer of interest as described above. The control expression cDNA used in the LTV knockdown/rescue experiments was generated by zipper PCR and produces an Acss2 cDNA with deletions of sequences corresponding to coding exons 3 through 6, which encodes a truncated and non-functional Acss2 protein.

Lentivirus stable transduction
Lentiviruses were generated by co-transfection of the expression vector of interest with packaging plasmids psPAX2 and pMD2G. The day before transduction, HT1080 cells were trypsinized and 2 x 10 5 cells per well plated in 1 mL complete culture medium in a 6-well plate for overnight incubation at 37˚C. On the day of transduction, media was removed and replaced with 1 ml of complete medium with 10 μg/ml polybrene (Cat. No. 107689, Sigma, St. Louis, MO). Lentiviral particles were thawed to room temperature, mixed gently, and added to the HT1080 cells. After gently swirling to mix, cells were incubated overnight. After 12 hr, culture medium was replaced with 2 ml of complete medium containing 10 μg/ml blasticidin S (Cat. No. ant-bl, Invivogen, San Diego, CA), which was replaced every 2 days until one week after all control cells had died. Positive cells were propagated in 1 μg/ml blasticidin S for two weeks, and then were frozen down. For experiments, cells were thawed and allowed to grow for three passages before use. Ectopic HIF-2α acetylation HT1080 cells expressing amino-terminal S-peptide (SP) epitope tagged and carboxy-terminal hemagglutinin A (HA) epitope tagged wild-type (WT) HIF-2α were pretreated with sirtinol (Cat. No. 510 8474, Chembridge Corporation, San Diego, CA; 5 μM final concentration) and nicotinamide (NAM) (Cat. No. N0636, Sigma; 10 mM final concentration) for 6 hr prior to initiation of stress or to acetate addition. Ectopic HIF-2α was purified using SP-agarose from whole cell extracts, prepared with a kit (Cat. No. 40010, Active Motif, Carlsbad, CA) supplemented with 1× protease inhibitor cocktail (Cat. No. P8340, Sigma), 1 mM PMSF, 10 mM NAM, and 5 μM sirtinol, and subjected to immunoblot analyses as previously described [8]. Similarly, amino-terminal V5-peptide (V5) epitope tagged wild-type K3 or mutant R3 (K385R, K685R, K741R) HIF-2α expressed in HT1080 cells was purified using V5-agarose before immunoblot analysis.
Endogenous HIF-2α acetylation HT1080 cells in a single 100 mm plate for each time-point were pretreated for 6 hr in complete medium with 5 μM sirtinol and 10 mM NAM, and then cultured in the appropriate medium under the indicated conditions. Whole cell extracts were prepared using a kit (Cat. No. 40010, Active Motif) supplemented with 1× protease inhibitor cocktail, 1 mM PMSF, 10 mM NAM, and 5 μM sirtinol. Endogenous HIF-2α was incubated with a monoclonal human HIF-2α antibody (Cat. No. NB100-132, Novus Biologicals) for 1 hr and then was immunoprecipitated using protein G magnetic beads (Cat. No. 54002, Active Motif). Aliquots were immunoblotted for endogenous HIF-2α or acetyl lysine as described [9].

Endogenous protein lysine acetylation
Stable HT1080 cells expressing shRNA directed against human Acss2 or expressing control non-targeting shRNA [14] maintained under blasticidin S selection (1 μg/mL medium) were plated in a single 100 mm plate and grown until 90% confluency, then were changed to the appropriate medium without blasticidin S for growth under the indicated conditions. Cells were treated for the last 6 hr of the treatment period with 5 μM sirtinol and 10 mM NAM. Whole cell extracts were prepared using a kit (Cat. No. 40010, Active Motif) supplemented with 1× protease inhibitor cocktail, 1 mM PMSF, 10 mM NAM, and 5 μM sirtinol. Aliquots (15 μg) of whole cell extracts were electrophoresed on a 10% PAGE gel, transferred to a PVDF membrane (Cat. No. IPVH00010, EMD Millipore), and immunoblotted with antibody recognizing acetylated lysine residues (1:1,000 dilution; Cat. No. 9441, Cell Signaling Technology) after pre-blocking. Parallel samples were analyzed by immunoblotting to detect endogenous human Acss2 (1:1,00 dilution; Cat. No. D19C6, Abcam) and α-tubulin (1:5,000 dilution; Cat. No. T9026, Sigma).

In vitro immunoprecipitation
For the in vitro immunoprecipitation assays using purified proteins, we first transfected SP/ HA tagged K3 or R3 human HIF-2α [8], c-myc tagged WT or HAT mouse CBP, or V5 tagged WT or MUT mouse Acss2 expression constructs into HEK293T cells; waited 48 hr; prepared whole cell extracts; incubated the ectopic HIF-2α, CBP, or Acss2 transfected cell extracts with HA (Cat. No. H9658, Sigma), c-myc (Cat. No. sc-789, Santa Cruz Biotechnology), or V5 (Cat. No. 46-0705, Invitrogen) antibodies, respectively; and then purified the ectopic proteins with protein A/G agarose (Cat. No. sc-2003, Santa Cruz Biotechnology). The ectopic proteins were eluted with 10 μg/mL working solutions of soluble competitor HA peptide (Cat. No. I2145, Sigma), c-myc peptide (Cat. No. M2435, Sigma), or V5 peptide (Cat. No. V7754, Sigma). For the in vitro immunoprecipitation assays, each reaction mixture contained 60 mM potassium phosphate (pH 7.5), 0.1 mM CoA, 4 mM MgCl 2 , 1 mM DTT, without or with 3 mM ATP plus 0.24 mM sodium acetate where indicated, in addition to purified CBP, Acss2 and HIF-2α ectopic proteins as specified. Where indicated, acetyl CoA was added to the final concentration instead of ATP and sodium acetate. After addition, we incubated the mixture for 30 min at 30˚C, performed pulldown, and subjected the bound proteins to immunoblot analyses as indicated.

Immunofluorescence microscopy
HT1080 cells were plated on poly-L-lysine coated coverslips and placed in sterile 24-well plates after overnight incubation under control tissue culture conditions with complete media. The next morning, plates were maintained under control, hypoxic (1% oxygen, 4 hr), low glucose (1 mM, 24 hr), or acetate (0.5 mM, 4 hr) conditions. To visualize ectopic Acss2, paraformaldehyde-fixed and Triton X-100 permeabilized cells were blocked with normal goat serum, and incubated with an anti-V5 epitope tag antibody (1:500 dilution; V5-FITC conjugate, Cat. No. R963-25, Invitrogen, Life Technologies). Nuclei were then stained with Hoechst dye (Hoechst 33258, Cat. No. 23491-45-4, Sigma). After mounting, coverslips were viewed and photographed using a fluorescent microscope (Olympus BX51, Olympus America Inc., Melville, NY). Image comparisons were made in Adobe Photoshop using identical settings for each image.

Subcellular fractionation
To prepare subcellular fractions for knockdown/rescue HT1080 cell lines expressing WT, CYT, SV40-WT, or SV40-CYT Acss2, we used CytoBuster protein extraction reagent (Cat. No. 71009, Novagen, Gibbstown, NJ) followed by use of NE-PER nuclear and cytoplasmic extraction reagents (Cat. No. 78833, Pierce, Rockford, IL) to prepare subcellular fractions before and after the indicated treatment as previously described [10]. Samples were subjected to immunoblot analyses for fractionation. Whole cell extracts prepared in parallel were analyzed for ectopic HIF-2α acetylation as described.
14 C-Acetate lipid synthesis determinations HT1080 cells were plated as triplicates in 60 mm plates at 60% confluence and allowed to attach overnight under standard growth conditions. The next morning, cells were incubated under control (21% oxygen, 25 mM glucose), hypoxic (1% oxygen), or low glucose (1 mM glucose) conditions. Twenty-four hours later, media containing 14 C-acetate [acetic acid, sodium salt, (1,2-14 C); Cat. No. NEC553050UC, Perkin Elmer, Santa Clara, CA] were added and the cells grown under the same conditions for an additional 24 hr. At the time of harvest, we aspirated media, rinsed cells twice with 1x PBS, and added Triton-X 100 (0.5% in ddH 2 O) to solubilize cells. After transferring the lysis to a microfuge tube, the cellular debris was pelleted at 14,000 rpm, 4˚C, 10 min, and the supernatants were transferred to a microfuge tube. We removed an aliquot for total protein determinations. For the remainder of the supernatants, we performed lipid extractions with sequential addition of methanol, chloroform, and water with vortexing and centrifugation after each step to resolve the aqueous and organic phases. After the last centrifugation, the organic phase was transferred to a new tube and evaporated to dryness under air in a hood. Extracts were resuspended in 100 μl chloroform and counted with Ecolume liquid scintillation cocktail (Cat. No. 0188247001, MP Biomedicals, Santa Ana, CA). Counts were normalized to protein concentration for each sample.

Real-time PCR analyses
The expression of endogenous MMP9, PAI1, VEGFa, GLUT1, PGK1, and cyclophilin B in HT1080 cells were determined by reverse transcription of total RNA followed by quantitative real-time PCR analysis (qRTPCR) on an Applied Biosystems ABI Prism 7000 thermocycler using Power SYBR Green Master Mix following the manufacturer's protocol as previously described [8]. The results of triplicate experiments, with each sample measured as triplicates, were expressed as 2 -(gene-of-interest number of cycles-cyclophilin number of cycles) as previously described [8]. We used the following human (forward, reverse) primer pairs: MMP9:

Chromatin immunoprecipitation (ChIP) assays
For HT1080 cells used in ChIP experiments, we seeded 2 × 10 6 HT1080 cells (150 mm plates) 48 hr prior to use, then exposed the cells to normoxia, hypoxia, or low glucose (1mM glucose), and finally harvested the cells for whole cell protein or RNA preparations. VEGFa, PAI1, MMP9, GLUT1, and PGK1 induction after hypoxia or low glucose exposure was confirmed by real-time RT-PCR.
Sequential chromatin immunoprecipitation assays were performed using the Re-ChIP-IT™ magnetic chromatin re-immunoprecipitation kit (Cat. No. 53016, Active Motif). The antisera for the first chromatin immunoprecipitation reaction was normal mouse IgG (Cat. No. 2027, Santa Cruz Biotechnology) or anti-human EPAS1 antiserum (Cat. No. NB 100-132, Novus Biologicals). The antisera for the second chromatin immunoprecipitation reaction was antihuman CBP (Cat. No. 7389, Cell Signaling Technology). After sequential ChIP, the precipitated genomic DNA was analyzed by quantitative PCR as described above.
Histone H3 analyses HT1080 cells were harvested, washed twice with ice-cold PBS supplemented with 5 mM sodium butyrate, and counted. Then 10 7 cells were resuspended and lysed in 1 mL Triton extraction buffer [TEB: PBS containing 0.5% Triton X 100 (v/v), 2 mM phenylmethylsulfonyl fluoride (PMSF), 0.02% (w/v) NaN 3 ] on ice for 10 minutes with mixing by gentle inversion. Cell lysates were centrifuged at 2000 rpm for 10 min at 4˚C, and pellets were washed once by centrifugation in 500 uL TEB. The nuclear pellets were resuspended in 250 uL 0.2 N HCl and rotated overnight at 4˚C. The supernatant containing histones was removed after centrifugation and 1/5 volume NaOH was added to neutralize the pH. The protein concentration was determined by Bradford assay.
Histones were evaluated by immunoblot analysis using antibodies for the following antigens: pan histone H3

Cell proliferation assays
For cell proliferation assays, 1 x 10 3 HT1080 cells/well were seeded in a 96-well plate with each cell line in 8-well replicate sets. After 24 h, cells were exposed for 1 week to 1% oxygen with complete (25 mM glucose) media, or exposed to 21% oxygen with either complete media, low glucose (1 mM glucose) media, or complete media supplemented with acetate (5 mM). Media were changed every 48 hr with comparable media. Cell proliferation was detected every day with the CellTiter 96 AQueous Non-Radioactive Cell Proliferation Kit (Cat. No. G5421, Promega, Madison, WI).

Colony formation assays
For colony formation assays, 5 x 10 2 HT1080 cells seeded in triplicate 100 mm plates were allowed to attach for 24 hr in complete medium. After 24 hr, medium was changed and cells were cultured for 10 days under control (21% O 2 , complete medium), hypoxia (1% O 2 , complete medium), low glucose (21% O 2 , low glucose medium) or acetate supplemented (21% O 2 , complete medium with 5 mM acetate) conditions. Media was not changed throughout the experiment. Colonies were stained with 1% crystal violet in ethanol/PBS (15%/85%). Cells were imaged and colony number determined using ImageJ software.

Cell migration and cell invasion assays
For cell migration assays, HT1080 cells were serum-starved in 0.5% FBS/DMEM media overnight. After 12 hr, 1.5 x 10 5 HT1080 cells in serum-free media were transferred into a transwell insert. For cells maintained under normal conditions, cells were incubated with complete media and exposed to 21% oxygen for 4 h. For cells maintained under hypoxic conditions, cells were incubated with complete media and exposed to 1% oxygen for 4 hr. For cells exposed to low glucose conditions, cell were incubated with low glucose (1 mM) media at 21% oxygen for 24 hr and compared to control cells maintained under standard glucose (25 mM) conditions for 24 hr. For acetate-supplementation assays, cells were incubated with complete media supplemented with acetate (5 mM) and exposed to 21% oxygen for 4 h. Cell migration was detected in triplicates for each treatment after crystal violet staining. The absorbance was recorded at 560 nm with a microplate reader.
For cell invasion assays, we used a commercially available kit containing wells pre-filled with Matrigel (CytoSelect 24-Well Cell Invasion Assay Kit; Cat. No. CBA-110, Cell Biolabs, San Diego, CA). HT1080 cells were serum-starved in 0.5% FBS/DMEM media overnight. After 12 hr, 1.5 x 10 5 HT1080 cells in serum-free media were transferred into the transwell insert as above after pre-incubating the transwell insert for 1 hr with serum-free media at room temperature. Cell migration was determined from triplicates for each treatment according to the manufacturer's protocol.

In vivo nude mice flank tumor experiments
All animal experiments were approved by the UTSWMC Institutional Animal Care and Use Committee. Mice were anesthetized by isoflurane inhalation prior to flank tumor injections. Female nude mice obtained from NCI were injected subcutaneously on the left dorsal flank with 5×10 6 luciferase-expressing stably transformed HT1080 cells suspended in 0.5 ml DMEM, that were grown using 3% FBS to increase subsequent in vivo tumor cell growth and seeding efficiency. Tumor sizes for volume estimates were measured using calipers every other day beginning on the fourth day after cell injections. Beginning six days after injection, mice were administered vehicle (PBS, 0.01 mL/g body weight) or glyceryl triacetate (GTA; 90 μL/25 gm body weight; Cat. No. W200700, Sigma-Aldrich Chemicals, Saint Louis, MO) by oral gavage once per day [15]. All mice were harvested at the same day following tumor implantation when tumor volumes reached~1.5 cm 3 for at least 25% of the mice in the largest tumorbearing set. Mice were monitored on a daily basis following tumor implantation for signs of distress. Any mice meeting euthanasia or tumor size criteria were anesthetized by isoflurane inhalation and euthanized by bilateral thoracotomy prior to lung as well as tumor excision.

Statistical analyses
We compared results obtained from the indicated experimental groups by unpaired Student's t-Test with Welch's correction for groups of equal sample size or by z-Test for groups of unequal sample size. One-tailed or two-tailed analyses were performed as indicated. We assumed equal variances for experimental groups. We used one-way or two-way ANOVA analyses for multiple comparisons with Dunnett's multiple comparison posthoc test. The statistical analyses were performed using StatPlus (AnalystSoft Inc.) and Prism 7 (GraphPad Software, Inc.). All P values less than or equal to 0.05 ( Ã ) or 0.10 ( ÃÃ ) are reported for the stated comparisons.

Acss2-generated acetyl CoA controls Cbp/HIF-2α interactions
Cbp/HIF-2α complex formation can be induced in vitro by addition of Acss2 substrate and cofactor, acetate and ATP, or by addition of Acss2 product, acetyl CoA, when wild-type Acss2 is present [16]. Cbp/HIF-2α acetylation requires HIF-2α lysines residues, Cbp acetyltransferase activity, and the acetyl CoA generator Acss2 [8,9,14,16]. We asked whether alterations in any of these factors would affect Cbp/HIF-2α complex formation in this assay.
Consistent with HIF-2α acetylation and complex formation Cbp being linked processes, K3, but not R3, HIF-2α complexes with Cbp upon addition of acetate/ATP or acetyl CoA (Fig  2A). Similarly, enzymatically active Cbp is required for stable Cbp/HIF-2α complex formation as wild-type (WT), but not histone acetyltransferase mutant (HAT), Cbp complexes with K3 HIF-2α upon addition of acetate/ATP or acetyl CoA ( Fig 2B). Finally, stable Cbp/HIF-2α complexes form in the presence of wild-type (WT), but not enzymatically inactive mutant (MUT), Acss2 when supplied with acetate/ATP, whereas the addition of the Acss2 end product acetyl CoA results in stable Cbp/HIF-2α complex formation in the presence of either WT or MUT Acss2 (Fig 2C).

A mutation in a putative nuclear localization signal impairs Acss2 translocation
We hypothesized that Acss2, which transits to the nucleus under stress conditions [14,16], produces acetyl CoA used for nuclear HIF-2α acetylation and Cbp/HIF-2α complex formation. We identified a putative nuclear localization signal (NLS) in a region of wild-type (WT) mouse Acss2 containing basic residues. We generated a substitution mutant (CYT) using residues present in this region of a prokaryotic Acss2 homologue. CYT Acss2 does not translocate to the nucleus following exposure to hypoxia, glucose deprivation, or exogenous acetate as assessed by immunofluorescence (Fig 3A and S1 Fig). WT and CYT Acss2 are not detected in the nucleus of cells maintained under basal conditions ( Fig 3A).
We also used cell fractionation to determine if CYT Acss2 can transit to the nucleus. Similar to the immunofluorescence results, CYT Acss2 is not found in the nucleus following stress induced by hypoxia or glucose deprivation as well as following addition of exogenous acetate ( Fig 3B). Forced nuclear targeting of WT or CYT Acss2 under basal as well as stress conditions is conferred by fusion of the SV40 large T antigen nuclear localization signal (SV40-NLS) to the Acss2 amino terminus (Fig 3B).

Cytosol-restricted Acss2 is enzymatically active
We hypothesize that HIF-2α acetylation requires enzymatically active Acss2 be present in the nucleus when acetate levels are increased. We therefore asked whether CYT Acss2 facilitates HIF-2α acetylation. WT, but not CYT, Acss2 confers HIF-2α acetylation following exposure to hypoxia, glucose deprivation, or exogenous acetate (Fig 4A). Despite constitutive localization, HIF-2α acetylation by SV40 NLS-Acss2 fusion proteins is only evident under stress conditions or upon acetate addition (Fig 4A). The CYT mutation likely has selective effects on The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors CYT Acss2 retains acetyl CoA generating capacity when force-translocated into the nucleus. To assess whether CYT Acss2 retains acetyl CoA generating capacity in the cytosol, we also measured acetate-dependent lipid synthesis rates under control, hypoxia, or glucose The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors deprivation conditions. Consistent with CYT Acss2 being only defective in nuclear localization and not in acetyl CoA generating capacity, all four rescue Acss2 forms (WT, CYT, SV40-WT, SV40-CYT) retain acetate-dependent cytosolic lipid synthesis capacity, which is otherwise blunted by Acss2 knockdown (Fig 4B). The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors Maximal HIF-2 signaling during stress requires HIF-2α acetylation Specific HIF-2α lysine residues, which are acetylated by Cbp during hypoxia [9], are required for stable Cbp/HIF-2α complex formation induced by hypoxia or glucose deprivation (Fig 1). We measured HIF target gene induction in stably transformed knockdown/rescue HT1080 cells expressing ectopic K3 or R3 HIF-2α under control or stress conditions (hypoxia or glucose deprivation). K3 HIF-2α efficiently induces HIF-2 target genes associated with tumor growth and metastasis (MMP9, PAI1, VEGFa, GLUT1) during hypoxia ( Fig 5A) and low glucose (Fig 5B) stress conditions. In comparison, HIF-2 target gene induction is blunted in R3 HIF-2α knockdown/rescue and HIF-2α knockdown cells. Induction of the HIF-1 target gene PGK1 is unaffected in K3 or R3 HIF-2α knockdown/rescue cells as well as in HIF-2α knockdown cells.

Maximal HIF-2 signaling during stress requires nuclear Acss2
We reasoned that HIF-2 stress signaling requires Acss2 be present in the nucleus when acetate levels are increased. We measured HIF target gene induction in stably transformed knockdown/rescue HT1080 cells expressing ectopic WT, CYT, SV40-WT, or SV40-CYT Acss2 under control or stress conditions (hypoxia or glucose deprivation). Only CYT Acss2 knockdown/rescue cells are impaired in their ability to induce HIF-2 target genes under hypoxia (Fig 6A) or glucose deprivation (Fig 6B) conditions, which are also blunted in Acss2 knockdown cells. Induction of HIF-2 target genes in WT, SV40-WT, or SV40-CYT Acss2 knockdown/rescue cells is evident only under stress conditions, consistent with signaling induced coincident with increased acetate levels.
Consistent with WT Acss2 being required for stable Cbp/HIF-2α complex formation, efficient recruitment of endogenous HIF-2α to promoter regions of HIF-2 target genes (MMP9, PAI1, VEGFa, GLUT1), as assessed by single ChIP after either hypoxia or low glucose exposure, is evident when wild-type (WT), but not cytosol-restricted (CYT), Acss2 is expressed (Fig 8A and S6A Fig). Furthermore, sequential ChIP reveals that endogenous Cbp is also recruited with endogenous HIF-2α when WT, but not CYT, Acss2 is expressed (Fig 8A and  S6A Fig).

Acetylated HIF-2α regulates in vitro tumor cell properties
Specific HIF-2α lysine residues are required for stable Cbp/HIF-2α complex formation (Fig 1), induction of HIF-2 target genes (Fig 5), and chromatin remodeling (Fig 7) in cells exposed to hypoxia or glucose deprivation. To assess whether HIF-2α acetylation is required for maintenance of in vitro tumor cell properties, stably transformed knockdown/rescue HT1080 cells expressing ectopic K3 or R3 HIF-2α were examined. Under stress conditions (hypoxia or low glucose exposure) or following acetate supplementation, R3 HIF-2α was impaired in its ability to rescue cell proliferation, colony formation, cell migration, and cell invasion, tumor cell processes also blunted by endogenous HIF-2α knockdown (Fig 9A-9D).

Nuclear Acss2 regulates in vitro tumor cell properties
Enzymatically active Acss2, which is present in the nucleus following stress or acetate exposure (Fig 3), is required for HIF-2α acetylation (Fig 4), induction of HIF-2 target genes (Fig  6), and chromatin remodeling (Fig 8) in cells exposed to hypoxia or glucose deprivation. Acss2 is necessary for effective growth in tumor cells [14]. We reasoned that tumor growth and metastasis depends upon Acss2 signaling in the nucleus rather than Acss2 cytosolic lipid synthesis. Therefore, we measured cell survival, migration, invasion, and colony formation of stably transformed HT1080 cells lacking endogenous Acss2 and expressing ectopic WT, CYT, SV40-WT, or SV40-CYT Acss2 under control, hypoxia, low glucose, or acetate supplemented conditions. Only CYT Acss2 is impaired in its ability to rescue tumor cell processes (Fig 10A-10D).

Acetylated HIF-2α regulates tumor cell growth and metastasis
We asked if stably transformed acetylation-deficient HIF-2α knockdown/rescue HT1080 cells have impaired in vivo growth and metastatic potential. Primary tumor weight and luciferase activity as well as metastatic lung luciferase activity are similar in mice with control shRNA The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors Cell proliferation of stably transformed knockdown/rescue HT1080 cells expressing wild-type (WT) Acss2, cytosol-restricted mutant (CYT) Acss2, or the same two proteins with the SV40 large T antigen nuclear localization signal fused to the amino terminus (SV40-WT, SV40-CYT). Cells were exposed to control, hypoxic, or low glucose conditions, or were supplemented with acetate (5 mM) (n = 8/treatment/day). (B) Colony formation, (C) Cell migration, or (D) Cell invasion after ten days of the same cells assessed under identical conditions (n = 3/treatment). Comparison of samples within a given condition on the day of interest was performed by two-way (cell proliferation) or one-way (colony formation, cell migration, cell invasion) ANOVA followed by Dunnett's multiple comparisons test using control shRNA knockdown/control rescue as reference with decreased samples noted (*, P<0.05). For proliferation data, the indicated samples (*) differ from control cells for at least five of the last six days in this seven-day protocol. Other samples have no or one difference during this same period. Values indicated are means with SD. https://doi.org/10.1371/journal.pone.0190241.g010 The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors knockdown or acetylation-intact (K3) HIF-2α knockdown/rescue flank tumors (Fig 11A-11C). In contrast, mice with flank tumors derived from HIF-2α shRNA knockdown or acetylation-defective (R3) HIF-2α knockdown/rescue cells have blunted primary tumor weight and luciferase activity as well as metastatic lung tumor activity. Triacetin, an oral acetate compound, augments growth and metastasis of control shRNA knockdown or K3 HIF-2α knockdown/rescue cells, but not HIF-2α shRNA knockdown or R3 HIF-2α knockdown/rescue cells.

Nuclear Acss2 regulates tumor cell growth and metastasis
Next, we asked if cytosolic restricted Acss2 knockdown/rescue HT1080 cells have impaired in vivo growth and metastatic potential. Primary tumor weight and luciferase activity as well as metastatic lung luciferase activity are similar in mice with control shRNA knockdown, WT Acss2 knockdown/rescue, or SV40-CYT Acss2 knockdown/rescue flank tumors (Fig 12A-12C). In contrast, mice with flank tumors derived from Acss2 shRNA knockdown or CYT Acss2 knockdown/rescue cells have blunted primary tumor weight and luciferase activity as well as metastatic lung tumor activity. Triacetin augments growth and metastasis of control shRNA knockdown, WT Acss2 knockdown/rescue, or SV40-CYT Acss2 knockdown/rescue cells, but not of Acss2 shRNA knockdown or CYT Acss2 knockdown/rescue cells.

Discussion
Although described in lower organisms [26] and postulated in higher metazoans [27], an acetate switch in mammals had eluded detection. Acetate increases during oxygen or glucose deprivation, is converted into acetyl CoA by acetate-dependent acetyl CoA synthetases [27]. We recently reported that a mammalian acetate switch plays a context-dependent role in stress signaling by one HIF family member, HIF-2 [14,16]. We propose that Acss2, similar to its prokaryotic homologue Acs [26,28] or yeast homologue Acss2p [29,30], functions as an acetateactivated switch to alter gene expression patterns, which includes de novo genetic (transcriptional) as well as epigenetic (altered histone marks) events (Fig 13).
The acetate-dependent acetyl CoA generator Acss2 was initially investigated as a cytosolic acetyl CoA generator for lipid biosynthesis [31]. Indeed, Acss2 knockdown substantially reduces acetate-derived lipid synthesis as well as flank tumor growth [14], consistent with previous observations implicating Acss2 in this process [32]. However, the molecular basis for impaired growth of tumors bearing mutant Acss2 is not primarily due to effects on anabolic processes. The cytosol-restricted Acss2 examined in this study retains lipid synthesis capacity, yet is markedly impaired in tumor cell function, indicating Acss2 possesses other properties that regulate tumor growth and metastasis.
Instead of existing in equilibrium [33], cytosolic and nuclear acetyl CoA pools in eukaryotic cells are likely sequestered [29,30]. We hypothesize that dynamic changes in nuclear pools of acetyl CoA, the other substrate beside HIF-2α lysines in the Cbp-mediated HIF-2α acetylation reaction, regulate HIF-2α acetylation, Cbp/HIF-2α complex formation, and hence HIF-2 signaling. This signaling function of Acss2 requires that it transit from the cytosol to the nucleus. However, nuclear localization of Acss2-as occurs with hypoxia, low glucose, or acetate exposure-is necessary, but not sufficient, to augment HIF-2 signaling. Increased acetate levels that follow stress exposure trigger Acss2 nuclear translocation and acetyl CoA production.
The mutations used to generate a cytosol-restricted Acss2 are natural sequence variants found in a prokaryotic acetyl CoA synthetase (ACS) [34]. Although nuclear translocation is not relevant for prokaryotic ACS proteins, eukaryotic Acss2 stress signaling share some commonalities with prokaryotic ACS stress signaling. Acetate in E. coli increases during hypoxia [35] and is used by ACS to generate acetyl CoA [26]; acetate in eukaryotic cells increases in  assessed by luciferase activity in nude mice carrying flank tumors derived from luciferase-expressing stably transformed control or Acss2 knockdown HT1080 cells, or stably transformed knockdown/rescue HT1080 cells expressing wild-type (WT) Acss2, cytosolrestricted mutant (CYT) Acss2, or CYT Acss2 with the SV40 large T antigen nuclear localization signal fused to the amino terminus (SV40-CYT). Luciferase activity was assessed by triplicate measurements of tissue extracts. Mice in each group were treated with control or oral acetate (triacetin) delivered by oral gavage. Comparison within each shRNA knockdown/control rescue pair treated with vehicle or acetate (triacetin) was response to hypoxia (or glucose deprivation) and is likewise used by Acss2 to generate acetyl CoA [14,16]. Sir2, a stress-responsive genetic regulator, modulates ACS activity [36]; Sirt1, the eukaryotic Sir2 homologue, also may play a putative role in regulation of Acss2 activity. However, there are some differences in acetate and hypoxia stress signaling in prokaryotes and eukaryotes. Acetate signaling by ACS is controlled by dynamic acetate-responsive acetylation/ deacetylation modifications [37]. It is not known whether Acss2 is regulated in a likewise manner. Hypoxia sensing by ArcA, a prokaryotic functional homologue of HIF-1, is activated by limiting oxygen states [38], but operates independent of ACS [39,40]. In eukaryotes, acetate and hypoxia signaling are linked through the actions of Acss2, Cbp, and HIF-2.
Since the initial reports of Acss2 in stress signaling [14,16], additional studies have expanded our knowledge of the proposed role that acetyl CoA, acetate, and Acss2 play in epigenetic regulation during metabolic stress states [41][42][43]. Consistent with its earlier reported role as a biochemical flare [14,16], acetate increases in hypoxic cancer cells and induces hyper-acetylation of histone 3 in a partially Acss2-dependent manner [44]. We confirm these findings with respect to effects of Acss2 depletion on histone acetylation, although we note that loss of Acss2 results in global depletion of H3K27, but not H3K9, in addition to global depletion of H3K18 acetylation marks. We propose a mechanism whereby HIF-2 directly recruits Cbp to hypoxia-induced target genes, which facilitates Cbp acetylation of histone 3 at these target genes.
Acss2-mediated acetyl CoA generation in both the cytosol and nucleus are important in cancer cells. Reduced levels of Acss2 result in acetyl CoA depletion, protein deacetylation, and autophagy with the latter likely a result of loss of cytosolic acetyl CoA. However, the impact of Acss2 knockdown on protein acetylation is not likely due to reduced diffusion of acetyl CoA from the cytosol to the nucleus. We find that Acss2-generated acetyl CoA is used by Cbp in acetylation of HIF-2α as well as histone 3, but only if Acss2 generates a specific nuclear acetyl CoA pool in the nucleus under stress. While this manuscript was in preparation, a study also reported that Acss2 nuclear translocation is required for histone 3 acetylation, specifically at lysosomal and autophagosomal gene promoter regions, during glucose deprivation [45]. However, we find that this stress-induced, nuclear Acss2-dependent acetyl CoA generation and histone acetylation occurs during hypoxia as well as glucose deprivation.
The specific association of Acss2 with Cbp and subsequent effects on histone acetylation also has been shown in a recent study examining the role of Acss2 in neuronal memory [46]. An increase in specific histone 3 acetylation marks, H3K9 and H3K27, occurs during neuronal differentiation, which is associated with a transition from cytosolic to nuclear Acss2 localization. Similar to our results, Acss2 depletion results in a global decrease in acetylated H3K27 levels. In contrast, these investigators also note a decrease in global acetylated H3K9 levels, which may reflect differences in cell types used in the respective study (primary hippocampal neurons versus fibrosarcoma cells). Acetylated H3K18 levels were not assessed in that study. The investigators also noted physical association of Acss2 and Cbp evident by immunoprecipitation studies, which may explain why Acss2 and Cbp are enriched in chromatin from neuronal tissue. Thus, the combined action of Acss2 and Cbp on histone acetylation in specific regulatory regions may be dictated in part by interactions of Acss2 with Cbp, but likely are also due to Acss2-dependent interactions of Cbp with transcriptional regulators such as HIF-2, which require the specific subcellular pool of acetyl CoA generated by Acss2 in the nucleus. performed by one-tailed unpaired t test with Welch's correction for groups of equal sample size or by onetailed Mann-Whitney test for groups of unequal sample size with reductions indicated (*, P<0.05). Values indicated are means with SD (weights) or SEM (luciferase measurements). https://doi.org/10.1371/journal.pone.0190241.g012 The Acss2/Cbp/Sirt1/HIF-2α signal transduction axis coordinates stress signaling in tumors An acetate switch controls Cbp/HIF-2α interactions during stress, which is activated by increased endogenous acetate generated in response to stress. Oral acetate also activates the acetate switch to stimulate tumor cell function, which potentially links tumor growth and metastasis with nutritional cues. Nuclear-localized Acss2 is the molecular mediator of the acetate switch and supplies Cbp with a specific pool of nuclear acetyl CoA used in the acetylation of HIF-2α. Cbp and HIF-2α act in concert as a downstream effector of the acetate switch to regulate genetic as well as epigenetic events. Sirt1 regenerates deacetylated HIF-2α, which undergoes repetitive acetylation as long as the nuclear acetyl CoA pool from Acss2 is present.
In this study, we show that Acss2 regulates dynamic HIF-2α acetylation, Cbp/HIF-2α complex formation, Cbp/HIF-2 signaling, epigenetic remodeling, and tumor cell growth and metastasis (Fig 13). Ties between intermediary metabolism and cancer biology through signal transduction pathways are becoming increasingly evident [41][42][43]. The acetate/Acss2 switch acts in concert with HIF-2 during hypoxia and glucose deprivation, stresses that are frequently encountered in solid tumors. Oral acetate substantially affects flank tumor outcomes in an Acss2 and HIF-2 dependent manner. Serum acetate originates from endogenous and exogenous sources, the latter dictated by dietary and gastrointestinal bacterial interactions. Defining the inner workings of the mammalian acetate switch in normal as well as in transformed cells will inform us how intermediary metabolism and stress recognition are exquisitely linked to the prosurvival cellular response. Manipulation of this pathway may define a novel approach for intervening in cancer and possibly other disease states where the acetate switch is relevant. Single and sequential chromatin immunoprecipitation (ChIP) assays in stably transformed HT1080 cells expressing ectopic wild-type (WT) or cytosol-restricted (CYT) mutant Acss2 protein maintained under control (Con), hypoxia (Hyp), or low glucose (LG) conditions. The single and first stage of the sequential ChIP was performed with antibodies recognizing endogenous HIF-2α. The second stage of the sequential ChIP was performed with antibodies recognizing endogenous Cbp. The amplicons detect chromatin containing HIF-responsive elements (HRE) in regulatory regions of the HIF-2 target genes VEGFa and GLUT1. (B) Single ChIP assays in same cells and with same amplicons as in (A) as well as with amplicons recognizing the HIF-1 selective target gene PGK1 and a non-HIF regulated gene RPL13A, but using antibodies recognizing specific marks in histone 3 proteins acetylated by Cbp, H3K18ac and H3K27ac, as well as a histone 3 mark not modified by Cbp, H3K9ac. Comparison of samples within a given condition was performed by one-tailed unpaired t test with significantly decreased samples noted ( Ã , P<0.05; ÃÃ , P<0.10). Values indicated are means with SD. (TIF) S1 File. Annotated data. An Excel file containing raw data and annotations for all material presented in this study. (XLSX)