Acetylation-mediated phase control of the nucleolus regulates cellular acetyl-CoA responses

The metabolite acetyl-CoA serves as an essential element for a wide range of cellular functions including ATP production, lipid synthesis and protein acetylation. Intracellular acetyl-CoA concentrations are associated with nutrient availability, but the mechanisms by which a cell responds to fluctuations in acetyl-CoA levels remain elusive. Here, we generate a cell system to selectively manipulate the nucleo-cytoplasmic levels of acetyl-CoA using CRISPR-mediated gene editing and acetate supplementation of the culture media. Using this system and quantitative omics analyses, we demonstrate that acetyl-CoA depletion alters the integrity of the nucleolus, impairing ribosomal RNA synthesis and evoking the ribosomal protein-dependent activation of p53. This nucleolar remodeling appears to be mediated through the class IIa HDAC deacetylases regulating the phase state of the nucleolus. Our findings highlight acetylation-mediated control of the nucleolus as an important hub linking acetyl-CoA fluctuations to cellular stress responses.


Introduction
1 Intracellular metabolites dynamically fluctuate in living organisms, in association with the 2 availability of their source nutrients, such as glucose, lipids and amino acids. A cell employs a 3 variety of molecular machineries to monitor the concentration of these metabolites. Upon 4 metabolic fluctuations, those machineries activate signaling pathways that modulate gene 5 expression and protein function to maintain metabolic homeostasis. Molecular links between the 6 metabolite sensing and cellular responses have been extensively studied and several dedicated 7 molecular networks have been described (Campbell and Wellen, 2018;Efeyan et al., 2015;Wang 8 and Lei, 2018). However, partly due to experimental difficulties to selectively manipulate target 9 metabolites among highly interconnected metabolic networks, the mechanisms for cellular 10 primary responses towards fluctuations for most metabolites are not understood.

11
Acetyl-coenzyme A (acetyl-CoA) is a central metabolite that integrates diverse nutritional inputs 12 into the biosynthesis of essential biomaterials including ATP, fatty acids and steroids, and

21
In mammalian cells, acetyl-CoA is compartmentalized into a mitochondrial pool, and a 22 nuclear/cytoplasmic pool (Pietrocola et al., 2015;Sivanand et al., 2018). In the mitochondrial 23 matrix, acetyl-CoA is generated by the metabolism of nutrients including glucose, fatty acids and 24 amino acids. Mitochondrial acetyl-CoA can enter the tricarboxylic acid (TCA) cycle thereby 25 generating ATP and reducing equivalents (e.g. NADH) or it can be utilized for acetylation of 26 mitochondrial proteins. Since acetyl-CoA is membrane impermeant, there is no direct exchange 1 between mitochondrial acetyl-CoA and the acetyl-CoA in the cytosol. Rather, the TCA 2 intermediate citrate can be exported from the mitochondria to the cytosol where it can become 3 the predominant source for cytoplasmic and nuclear acetyl-CoA. The mitochondrial exported 4 citrate is converted to acetyl-CoA and oxaloacetate by the enzyme ATP citrate lyase (ACLY) 5 (Wellen et al., 2009). This nucleo-cytoplasmic acetyl-CoA pool serves as a building block for lipid 6 synthesis including fatty acids, cholesterol and steroids, as well as serving as an acetyl donor for 7 acetylation of cytosolic and nuclear proteins including histones. Another source of nucleo-8 cytoplasmic acetyl-CoA comes from the metabolite acetate, which derives from various extra-and 9 intracellular sources such as food digestion, gut microbial metabolism, alcohol oxidation and 10 deacetylation of acetylated proteins (Schug et al., 2016). Acetate can be utilized for acetyl-CoA 11 synthesis through a reaction catalyzed by the enzyme acyl-CoA synthetase short chain family 12 member 2 (ACSS2). In contrast to whole body ACLY deficient mice that are embryonic lethal,

13
ACSS2 deficient mice have no apparent phenotype under normal breeding conditions (Beigneux 14 et al., 2004;Comerford et al., 2014). This would support a predominant role of the citrate-ACLY 15 pathway for the nucleo-cytoplasmic acetyl-CoA production during development, and potentially 16 under other circumstances as well. However, it should be noted that in some mouse tumor models 17 or hypoxic tumor cells, or in other conditions where mitochondrial metabolism is dampened, the 18 acetate-ACSS2 pathway can play a critical role in cell proliferation, lipid synthesis and histone 19 acetylation (Comerford et al., 2014;Gao et al., 2016;Mashimo et al., 2014;Schug et al., 2015; 20 Yoshii et al., 2009). Moreover, ACLY deficient mouse embryonic fibroblasts (MEFs) exhibit 21 upregulation of ACSS2 and exogenously added acetate can be utilized for acetyl-CoA production 22 in these cells (Zhao et al., 2016). These observations indicate a coordinated relationship between 23 these two pathways in order to ensure the requisite supply of acetyl-CoA in the nucleo-24 cytoplasmic compartment.

25
Intimate connections between nucleo-cytoplasmic acetyl-CoA levels, the status of protein 1 acetylation and cellular responses have been illustrated by multiple recent studies. In yeast cells 2 during continuous glucose-limited growth, oscillations of acetyl-CoA are observed in accordance 3 with distinct metabolic phases and an increase in acetyl-CoA is highly correlated with acetylation 4 of several proteins including transcriptional coactivators and histones (Cai et al., 2011). Also, in 5 various mammalian cell models, transcription is regulated either by acetyl-CoA abundance, or 6 acetyl-CoA producing enzymes, i.e. ACLY or ACSS2, or by their nutrient sources (Gao et al., 7 2016;Huang et al., 2018;Lee et al., 2018;Lee et al., 2014;Li et al., 2017;Mews et al., 2017). 8 Moreover, nutrient deprivation or starvation causes a rapid decline in acetyl-CoA abundance in 9 cells in culture and in some mouse tissues, which is accompanied by deacetylation of proteins 10 (Marino et al., 2014). In mammalian cells, as well as in yeast, pharmacological or genetic 11 manipulations to deplete cytosolic acetyl-CoA have been reported to induce autophagy suppressed by class IIa HDAC inhibitors. We identified multiple nucleolar proteins whose 1 acetylation levels were potentially regulated by the class IIa HDACs, suggesting that class IIa 2 HDAC-dependent deacetylation of nucleolar proteins may play an important role in regulating the 3 integrity of the nucleolus and the induction of a nucleolus-dependent stress response when acetyl-4 CoA levels decline.

Acetate-dependent control of acetyl-CoA production in ACLY KO cells 3
In order to understand cellular responses to fluctuations in nucleo-cytoplasmic acetyl-CoA 4 abundances, we sought to devise a cell system where we could selectively and robustly control 5 acetyl-CoA levels. To achieve this, we focused on establishing a cell line whose acetyl-CoA in the 6 nucleo-cytoplasmic compartment was solely synthesized through the acetate-ACSS2 axis, 7 assuming that such a cell line would enable us to manipulate acetyl-CoA levels by simply 8 modulating the amount of acetate exogenously supplemented in the culture media ( Figure 1A).

9
Hence, using the CRISPR-Cas9 system, we targeted the ACLY gene in HT1080 human

16
suggesting that acetate supplementation increased the viability or growth of ACLY deficient cells.

17
This observation was consistent with the previously reported phenotype of ACLY deficient MEFs 18 (Zhao et al., 2016). We selected two independent ACLY deficient (KO) and ACLY expressing 19 (WT) clones, respectively, that henceforth were continuously cultured in the presence of 20 mM 20 acetate for further analyses ( Figure 1B and Figure 1-figure supplement 1C). Hereafter, we refer 21 to these cells as ASA-KO and ASA-WT (acetate-supplemented ACLY KO and WT), respectively.

22
In this culture condition, ASA-KO cells did not exhibit any defect in viability when compared with 23 ASA-WT cells ( Figure 1C, (+) Acetate, 10% FBS). However, removal of acetate from the culture 24 media for 4 days significantly impaired the viability of ASA-KO but not ASA-WT cells ( Figure 1C).

25
Consistent with the observation that fetal bovine serum (FBS) contains submillimolar 26 concentration of acetate (Kamphorst et al., 2014;Zhao et al., 2016), either a decrease in FBS 1 concentration from 10% to 1% or the use of 10% dialyzed FBS (dFBS) enhanced the vulnerability 2 of ASA-KO cells, while these manipulations had no significant effect on viability in the acetate-3 supplemented conditions ( Figure 1C). Moreover, exogenous expression of ACLY in ASA-KO 4 cells rescued the sensitivity to acetate removal, indicating that the acetate dependence of these 5 cells was caused by ACLY deficiency (Figure 1D). 6 Using this cell system, we next examined whether acetate removal modulated intracellular 7 acetyl-CoA levels in ASA-KO cells. We used media containing 1% FBS during acetate removal 8 to maximize the effect of acetate withdrawal, although similar, albeit slightly milder effects were 9 seen following acetate withdrawal in the setting of 10% FBS. Quantifications of acetyl-CoA in the 10 whole cell lysates demonstrated that withdrawing acetate for 4 hours drastically decreased the 11 amount of acetyl-CoA in ASA-KO cells, while this intervention had only a marginal effect in ASA-

12
WT cells ( Figure 1E). In contrast, the amount of Coenzyme A (CoA) was reciprocally increased 13 in ASA-KO cells in response to acetate removal ( Figure 1E). Taken together, in ASA-KO cells, 14 even in the presence of other nutrient sources such as glucose and lipids, the nucleo-cytoplasic 15 acetyl-CoA levels are seemingly tunable solely by adding or removing acetate in the culture media.

17
Acetyl-CoA depletion modulates protein acetylation 18 Using the ASA-KO cell system, we profiled global cellular responses following acute acetyl-CoA 19 depletion. As cytoplasmic acetyl-CoA is mainly utilized for lipid synthesis and protein acetylation, 20 we examined whether a rapid decline in acetyl-CoA levels affected these downstream events.

21
Quantifications of total cholesterol and a series of fatty acids revealed that although some fatty 22 acids including linoleic acid and Dihomo-g-linolenic acid (DGLA) exhibited significant decreases, 23 overall alterations in total amounts of cholesterol and fatty acids after 4 hours of acetate removal 24 were modest when compared to the dramatic decline in acetyl-CoA levels at the same time point 25 (Figure 2A and Figure 1E). Thus, at least at early time points, dramatic reductions in acetyl-CoA 1 are not fully reflected by marked alterations in total cholesterol or lipid amounts.

2
In order to examine global changes in the status of protein acetylation upon acetyl-CoA 3 depletion, we conducted an acetylome analysis using an acetyl-lysine motif antibody-based 4 immunoaffinity purification. Recovered acetylated lysine-containing peptides were then detected 5 through liquid chromatography mass-spectrometry. This approach identified 1307 acetylated 6 lysine-containing peptides in ASA-KO cells cultured in the presence of acetate (Table S1). Among 7 these, 658 peptides exhibited more than a 50% decrease after 90 minutes of acetate withdrawal 8 {log2 [fold change (FC): (-) Acetate / (+) Acetate] < -1.0}, suggesting that acetyl-CoA depletion 9 caused rapid deacetylation of nearly half of all acetylated proteins ( Figure 2B and Table S1, 10 Column B). These peptides included a wide variety of proteins including molecules involved in 11 transcription, translation, ribosomal RNA (rRNA) processing, messenger RNA (mRNA) splicing, 12 nucleosome assembly and cell-cell adherence junctions ( Figure 2C). By immunoblotting using 13 pan-and site specific-anti-acetyl-lysine antibodies, we detected immediate and robust decreases 14 in acetylation of both histone and non-histone proteins upon acetate withdrawal (Figure 2D-2E

26
Intriguingly, we also found many transcripts that were significantly upregulated in these conditions 1 {365 and 899 transcripts in 10% FBS and 1% FBS conditions, respectively, log2 [FC: (-) Acetate 2 / (+) Acetate] > 1.0, q < 0.05}. These differentially expressed transcripts in the 10% and 1% FBS 3 contidions were highly overlapped (70.5% or 33.8% of downregulated transcritps in 10% or 1% 4 FBS conditions, respectively, and 65.2% or 26.4% of upregulated transcritps in 10% or 1% FBS 5 conditions, respectively, Figure 2-figure supplement 1A), suggesting that a smillar but stronger 6 transcriptional response occurs in the 1% FBS condition. Gene ontology analyses for differentially 7 expressed genes in the 1% FBS condition uncovered that several clusters of genes were similarly 8 regulated ( Figure 2G). For example, genes related to transcriptional regulation were highly 9 enriched in both up-and down-regulated transcripts. Of particular interest was that genes involved 10 in cellular stress responses such as apopototic process, cell cycle arrest and oxidation-reduction 11 process were selectively upregulated in acetate-withdrawn cells, suggesting that stress signaling 12 pathways are activated in response to acetyl-CoA depeltion.

14
Acetyl-CoA depletion alters the integrity of the nucleolus.

15
To further understand the complete cellular response after acetyl-CoA depletion, we also 16 performed quantitative proteomics analyses using a tandem mass tag (TMT) system with the 17 same experimental conditions used for our previous experiments (i.e. analysis 4 hours after 18 acetate removal in 1% FBS containing media). Although overall alterations in protein levels 19 appeared to be less drastic, we observed that the expression of 51 proteins were significantly 20 increased more than 1.2-fold and 135 proteins were decreased less than 0.8-fold in response to 21 acetate removal (Table S3). Intriguingly, we found that many proteins that are known to be 22 localized to the nucleolus were significantly altered ( Figure 3A, shown in red). Changes in these 23 nucleolar proteins were not evident on the RNA seq analysis (Figure 3A, y-axis), implying that 24 these proteins were likely regulated at a post-transcriptional level. By immunoblotting, we 25 confirmed that ribosome biogenesis factors RRP1 and BOP1 were increased while PICT1 (also 26 known as NOP53 or GLTSCR2) was decreased after acetate removal in ASA-KO cells but not in 1 ASA-WT cells (Figure 3-figure supplement 1A and 1B). These observations indicate that 2 alterations of the nucleolus might occur upon acetyl-CoA depletion. Hence, we next addressed 3 whether acetyl-CoA depletion affected nucleolar structures and functions. The nucleolus is the 4 organelle primarily responsible for ribosomal biosynthesis, where rRNA synthesis, rRNA 5 processing and assembly of rRNA and ribosomal proteins takes place (Mangan et al., 2017; 6 Nemeth and Grummt, 2018). Using a 5-Fluorouridine (FUrd) incorporation assay, we monitored 7 newly synthesized rRNA in the nucleolus and found that acetate removal rapidly reduced rRNA 8 synthesis in ASA-KO but not ASA-WT cells (Figure 3B-3C). We also utilized a rRNA specific dye 9 to stain cells in conjunction with immunostaining for nucleolar proteins. In the presence of acetate, 10 the nucleolar transcription factor UBF localized within the nucleolar region as evident by its  (Thiry and Lafontaine, 2005). We therefore performed a co-staining of rRNA 19 with marker proteins for the latter two compartments; fibrilalin (FBL) or nucleolin (NCL), 20 respectively. Interestingly, while both proteins colocalized with rRNA in the acetate-supplemented 21 cells, neither FBL nor NCL fully merged with the segregated rRNA signals in the acetate-deprived 22 ASA-KO cells (Figure 3-figure supplement 1D-1E). These observations imply that acetyl-CoA 23 depletion redistributes the rRNA containing compartment within the nucleolus. Furthermore, we 24 found that in ASA-KO cells, the nucleolar localization of a nucleolar scaffolding protein, 25 nucleophosmin (NPM1), became dispersed following acetate removal ( Figure 3F). We also noted 26 that ASA-WT cells did not exhibit any of these morphological alterations in the nucleolus in 1 response to acetate removal (Figure 3-figure supplement 1F-1G). Altogether, these 2 observations suggest that acetyl-CoA depletion markedly influences nucleolar components 3 including changes in levels and in localization, thereby structually and functionally remodeling the 4 nucleolus. 1 Moreover, small interfering (si) RNA-mediated knockdown of RPL11 or RPL5 diminished acetate-2 induced p53 upregulation, as well as PICT1 degradation, suggesting that RPL11 and RPL5 are 3 required for these processes ( Figure 4D). These observations were consistent with other 4 nucleolar stress-dependent p53 activation (Donati et al., 2013;Sloan et al., 2013). DNA damage 5 is a potent inducer of p53, but neither acetate removal nor ActD treatment induce phosphorylation  Table S1). Therefore, we tested a possible role for

24
( Figure 5B), suggesting that zinc-dependent HDACs are likely required for this process. In 25 contrast, EX-527, an inhibitor for the NAD + -dependent deacetylase Sirtuin 1, did not alter this 26 response ( Figure 5-figure supplement 1A). Using selective inhibitors for each HDAC class, we 1 sought to further address which HDAC is responsible for the acetyl-CoA depletion-induced p53 2 and PICT1 regulation ( Figure 5A). While the class I inhibitor Entinostat (also known as  3 and the class IIa inhibitor TMP195 exhibited differential inhibitory effect on deacetylation of 4 Histone H3 K9 and aTubulin K40, we found that TMP195 but not Entinostat blocked the reciprocal 5 regulation of p53 and PICT1 ( Figure 5B) (Lobera et al., 2013;Saito et al., 1999). TMP195 also   To investigate whether the effect of class IIa HDAC inhibitors on the nucleolus is mediated through 22 inhibition of deacetylation, we conducted an acetylome analysis for acetyl-CoA-depleted cells with 23 and without the class IIa inhibitor TMP195 and identified 365 acetylated peptides whose 24 acetylation levels were maintained at a level two-fold or greater by TMP195 treatment in the 25 setting of acetate removal {Table S1, Column C, Log2 [FC: (-) Acetate + TMP195 / (-) Acetate] 26 >1.0}. Importantly, among these TMP195 sensitive proteins, we identified multiple nucleolus-1 resident proteins including ribosomal proteins and NCL (Figure 5-figure supplement 2A). Using 2 the acetyl-lysine motif antibody immunoprecipitation assay, we demonstrated that acetylation of 3 RPL11 decreased upon acetate removal, and that this was suppressed by TMP195 but not by 4 the class I inhibitor Entinostat ( Figure 5F). Deacetylation of NCL by acetate removal was also 5 partially recovered by TMP195 (Figure 5-figure supplement 2B). These results suggest that the 6 class IIa HDACs mediate deacetylation of at least certain nucleolar proteins including RPL11 and 7 NCL.

8
The nucleolus is known to be formed through a process called liquid-liquid phase separation 9 (LLPS), a central mechanism for facilitating the formation of biomolecular condensates that exhibit     In this study, we demonstrated that the nucleolus rapidly responds to acetyl-CoA depletion. We 24 propose a model for this cellular response to a decline in acetyl-CoA nucleo-cytoplasmic levels 25 ( Figure 6). In cells that contain high levels of acetyl-CoA, multiple nucleolar proteins are highly acetylated and these post-translational modifications play a crucial role in maintaining the 1 nucleolar integrity, a phase separation state allowing for efficient rRNA synthesis. Once acetyl-2 CoA levels fall, class IIa HDACs deacetylate a significant number of nucleolar proteins. These

21
This study uncovered that the metabolite acetyl-CoA can be a novel modulator of the nucleolus 22 through protein acetylation. It has been reported that rRNA synthesis and rRNA processing are 23 regulated by multiple mechanisms through neutrient signaling pathways to meet the cellular 24 energy demands (Grummt, 2013). Toghether with these regulations, the cell likely employs this acetylation-mediated nucleolar stress response in order to help cope with physiological or 1 pathlogical fluctuations in the cellular nutrient status.       , 2005). The NCL-3xFLAG fragment was amplified using primers that attach the 3xFLAG at 1 NCL's C-terminus and was inserted into the EcoRI-digested pLVX-puro vector. For the NCL-2 mGFP expressing constract, the NCL-CDS was amplified from the GFP-Nucleolin plasmid and 3 monomeric GFP (mGFP) was amplified from the LAMP1-mGFP plasmid (a gift from Esteban

11
The immunoprecipitation assay was performed using an acetyl-Lysine motif antibody to detect 12 acetylated protein. The ASA-KO17 cells stably expressing 3xFLAG-RPL11 or NCL-3xFLAG were 13 plated, stimulated, and lysed as the FLAG immunoprecipitation assay described above. The

1
The gradient started with 10% of buffer B and was run under following conditions; 10 min at up to 2 40% of buffer B, 13 min at up to 68% of buffer B, 23 min at up to 72% of buffer B, 28 min at up to 3 100% of buffer B, and held for an additional 5 min. The initial condition was restored after 10 min 4 with 10% of buffer B. The flow rate was 0.6 ml/min and the detection was performed at 259 nm.    acid-PA esters were separated on a Phenomenex C8 column (2.1 X 150 mm, 5 µ pore size) using 8 H2O + 0.1% acetic acid for solvent A and ACN + 0.1% acetic acid for solvent B. The gradient 9 started at 65% B and increased linearly to 85% B at 10 min and was held for 1 min before ramping 10 to 100% B for 2min. Finally, the gradient returned to 65% B for a 2 min equilibration. Samples analyzed using the PTMScan method (Cell Signaling Technology) as previously described (Rush et al., 2005;Stokes et al., 2015). Briefly, the lysates were sonicated, centrifuged, reduced with 1 DTT, and alkylated with iodoacetamide. 15 mg of total protein for each sample was digested with 2 trypsin and purified over C18 columns for enrichment with the Acetyl-Lysine Motif Antibody 3 (#13416). Enriched peptides were purified over C18 STAGE tips. Enriched peptides were 4 subjected to secondary digest with trypsin and second STAGE tip prior to LC-MS/MS analysis.

5
Replicate injections of each sample were run non-sequentially on the instrument. Peptides were 6 eluted using a 90-minute linear gradient of acetonitrile in 0.125% formic acid delivered at 280 7 nL/min. Tandem mass spectra were collected in a data-dependent manner with a Thermo 8 Orbitrap Fusion™ Lumos™ Tribrid™ mass spectrometer using a top-twenty MS/MS method, a 9 dynamic repeat count of one, and a repeat duration of 30 sec. Real time recalibration of mass 10 error was performed using lock mass with a singly charged polysiloxane ion m/z = 371.101237.

11
MS/MS spectra were evaluated using SEQUEST and the Core platform from Harvard University.

12
Files were searched against the SwissProt Homo sapiens FASTA database. A mass accuracy 13 of +/-5 ppm was used for precursor ions and 0.02 Da for product ions. Enzyme specificity was 14 limited to trypsin, with at least one tryptic (K-or R-containing) terminus required per peptide and 15 up to four mis-cleavages allowed. Cysteine carboxamidomethylation was specified as a static 16 modification, oxidation of methionine and acetylation on lysine residues were allowed as variable 17 modifications. Reverse decoy databases were included for all searches to estimate false 18 discovery rates, and filtered using a 2.5% FDR in the Linear Discriminant module of Core.

19
Peptides were also manually filtered using a -/+ 5ppm mass error range and presence of an 20 acetylated lysine residue. All quantitative results were generated using Skyline to extract the 21 integrated peak area of the corresponding peptide assignments. Accuracy of quantitative data          Tiku, V., and A.

16
1% FBS containing media with or without acetate for 4 hours, or treated with 5 nM ActD or 1 µM 17 Camptothecin (CPT), a DNA damage inducer, for 4 hours.