Hallows et al. revealed that ACSS2 undergoes deacetylation mediated by SIRT1, a class III deacetylase (Hallows et al., 2006). Among the seven known class III deacetylases in mammalian cells (Sirt1-7), SIRT1 and SIRT2 are the major cytosolic members (Carafa et al., 2016; Imai and Guarente, 2014; Wang and Lin, 2021). Given the cytosolic localization of ACSS2, we were interested in investigating whether SIRT2 could also serve as a potential deacetylase for ACSS2. To address this, we ectopically expressed Flag-tagged ACSS2 and HA-tagged SIRT2 in HEK293T cells. We performed immunoprecipitation experiments using anti-acetyl-lysine beads to capture lysine-acetylated proteins, and the acetylation level of ACSS2 was assessed using an anti-Flag antibody. Our results demonstrated that the expression of SIRT2 led to a decrease in ACSS2 acetylation, indicating that SIRT2 is capable of deacetylating ACSS2 (Figure 1A).

To investigate whether endogenous SIRT2 regulates the acetylation levels of ACSS2, we employed lentiviral-mediated expression of short hairpin RNA (shRNA) to knock down SIRT2 in HEK293T cells. We then immunoprecipitated ectopically expressed Flag-tagged ACSS2 from both control and SIRT2 knockdown cells. However, we did not observe any significant difference in ACSS2 acetylation levels upon SIRT2 knockdown (Figure 1B). In contrast, SIRT2 knockdown was able to increase the myristoylation level of ARF6 (Figure 1—figure supplement 1), a known SIRT2 demyritoylation substrate (Kosciuk et al., 2020). We hypothesized that this could be due to the low basal levels of SIRT2 in cells, which might not yield a substantial deacetylation effect detectable by western blotting. SIRT2 has been reported to be upregulated under various stress conditions, including Golgi stress induced by shigella infection (Wang et al., 2022; Zullo et al., 2018) as well as under conditions of low glucose or amino acid deprivation (Zullo et al., 2018; Wang and Tong, 2009). To induce a stress response that would upregulate SIRT2 and potentially affect ACSS2 acetylation, we subjected cells to nutrient exhaustion by not changing the cell culture media over the course of 4 days (Ma et al., 2013). This resulted in an increase in SIRT2 levels (Figure 1—figure supplement 2) and when SIRT2 was knocked down under this condition, ACSS2 had increased acetylation compared to control knockdown cells (Figure 1B). These findings suggest that the deacetylation of ACSS2 by SIRT2 happens in the context of nutrient stress.

Our initial approach of inducing nutrient exhaustion could be due to a multitude of factors, hence we wanted to deconvolute the specific nutrient stress that led to the enhanced deacetylation of ACSS2. Because previous reports showed that SIRT2 is upregulated under conditions of low glucose or amino acid deprivation (Zullo et al., 2018; Sun et al., 2022), we examined the changes in ACSS2 acetylation in SIRT2 control and knockdown cells under these conditions. We observed that under amino acid deprivation, SIRT2 knockdown increased ACSS2 acetylation levels (Figure 1C). Notably, this effect was not observed under glucose starvation, underscoring the specificity of SIRT2’s deacetylation activity to amino acid stress (Figure 1—figure supplement 3). To further confirm the effect of SIRT2 on ACSS2 acetylation, we employed the SIRT2-specific small molecule inhibitor thiomyristoyllysine (TM) (Figure 1—figure supplement 4). TM increased ACSS2 acetylation under amino acid-starved conditions but not under normal conditions. Furthermore, the acetylation level of endogenous ACSS2 was increased by SIRT2 knockdown (Figure 1—figure supplement 5). These findings align with a previous study demonstrating that amino acid starvation triggers SIRT2 activation and subsequent deacetylation of ATG4B (Sun et al., 2022). Overall, our data indicates that SIRT2 deacetylates ACSS2 in response to nutrient stress, particularly amino acid deprivation.

We next aimed to investigate how nutrient stress-induced ACSS2 deacetylation by SIRT2 affects ACSS2 function. Consistent with the acetylation data, we found no alterations in the endogenous levels of ACSS2 when SIRT2 was knocked down under normal conditions. However, in the presence of nutrient stress, the endogenous ACSS2 levels exhibited an increase upon SIRT2 knockdown (Figure 2A). This effect was replicated under amino acid starvation (Figure 2B, Figure 2—figure supplement 1) conditions but not under glucose starvation (Figure 1—figure supplement 2). This was replicated in A549 cells as well (Figure 2—figure supplement 2). To further investigate this phenomenon, we used cycloheximide to inhibit translation and assess the stability of ACSS2 under amino acid limitation. We found that ACSS2 exhibited lower stability in the presence of amino acid limitation compared to normal culturing conditions (Figure 2—figure supplement 3). This suggests that under amino acid limitation, when SIRT2 is more active, there is an increased rate of deacetylation and degradation of ACSS2. Furthermore, our data demonstrated that under amino acid limitation, knocking down SIRT2 led to the stabilization of ACSS2 (Figure 2C, Figure 2—figure supplement 4). Importantly, ACSS2 mRNA levels remained largely unaffected by SIRT2 knockdown or amino acid starvation (Figure 2—figure supplement 5). Overall, the data suggests that SIRT2-mediated deacetylation of ACSS2 under amino acid limitation promotes its degradation, revealing a potential mechanism by which nutrient stress modulates ACSS2 function.

To elucidate the mechanism via which acetylation regulates ACSS2 protein levels, we initially examined whether it undergoes degradation via the lysosome or the proteasome. We treated cells with bafilomycin (Baf), which inhibits lysosomal acidification, and MG132, a proteasome inhibitor (Mauvezin and Neufeld, 2015; Zhang et al., 2013). Endogenous ACSS2 levels increased with MG132 treatment but not with Baf, indicating that ACSS2 is degraded through the proteasomal pathway (Figure 2D, Figure 2—figure supplement 6). Moreover, we readily observed ubiquitination of Flag-ACSS2 (Figure 2E). Given that SIRT2 knockdown increased endogenous ACSS2 levels, we further investigated whether SIRT2 regulates the ubiquitination of ACSS2. Notably, we found that knockdown or inhibition of SIRT2 in amino acid deprived media led to a decrease in ACSS2 K48-linked ubiquitination (Figure 2F and G), a hallmark of proteasomal degradation, suggesting that SIRT2 regulates the ubiquitination and degradation of ACSS2 through the proteasome pathway (Kwon and Ciechanover, 2017). The data together suggest that SIRT2 deacetylates ACSS2 to promote its ubiquitination and proteasomal degradation.

To further confirm this model and investigate the physiological significance of ACSS2 regulation by SIRT2, we aimed to identify the specific lysine residue deacetylated by SIRT2. Utilizing label-free quantification mass spectrometry, we analyzed purified Flag-tagged ACSS2 from control and SIRT2 knockdown cells. Our analysis showed that ACSS2 acetylation levels at K271 increased in SIRT2 knockdown cells (Supplementary file 1). SIRT2 knockdown failed to affect the acetylation level for the K271R mutant in amino acid deprived media, indicating that K271 is the site of deacetylation by SIRT2 (Figure 2H, Figure 2—figure supplement 7). In contrast, the acetylation level of ACSS2 K661R mutant was still regulated by SIRT2 (Figure 2—figure supplement 8). These observations collectively indicate that SIRT2 deacetylates ACSS2 at K271, which is different from previously reported SIRT1-regulated K661 deacetylation.

Interestingly, data from the global characterization of the ubiquitin-modified proteome indicated that ACSS2 can be ubiquitinated at the K271 (Kim et al., 2011). To validate this, we tested the K48-linked ubiquitination of K271R and K271Q mutants, and both mutants showed a decrease in ACSS2 ubiquitination levels (Figure 2—figure supplement 9). SIRT2 knockdown in amino acid deprived media decreased the ubiquitination of WT ACSS2, but not the ubiquitination of the K271R mutant (Figure 2I). Furthermore, the K271R mutant exhibited enhanced stability compared to the wild-type (WT) ACSS2 when cells were cultured in Earle’s balanced salt solution (EBSS) medium lacking amino acids (Figure 2J). Collectively, our findings suggest that under amino acid limitation, SIRT2 deacetylates K271, thereby exposing the site for ubiquitination and leading to subsequent degradation of ACSS2.

ACSS2 is known to be involved in DNL. Deletion of ACSS2 in mice results in reduced body weight and lipid deposition, while mutation of S236, a phosphorylation site on ACSS2, leads to increased triglyceride levels in adipocytes (Huang et al., 2018; Shaik et al., 2016). Based on our finding that ACSS2 is deacetylated and regulated by SIRT2, we next wanted to test whether this regulation could play a role in DNL.

To investigate this, we conducted a knockdown of endogenous ACSS2 in mouse 3T3-L1 preadipocytes and reintroduced either WT or the K271R mutant ACSS2 through transfection. To ensure comparable expression levels at the beginning, we adjusted the amount of transfected DNA for both WT and the K271R mutant ACSS2. Western blot analysis confirmed the successful knockdown of ACSS2 and the restoration of ACSS2 expression levels with both the WT and K271R mutant (Figure 3A). The cells were then subjected to adipocyte differentiation protocol. After 7 days, we observed that the protein level of the K271R mutant was higher than that of the WT ACSS2 (Figure 3B). This is consistent with our finding above that K271R mutation suppresses the ubiquitination and degradation of ACSS2.

We determined the lipogenesis with Oil Red O staining, which detects lipid droplets in cells. During the differentiation process, we supplemented the media with acetate to induce lipogenesis through the ACSS2 pathway. Additionally, post differentiation and before reading out the Oil Red O staining, we cultured the cells for 3 more days without changing media to have a nutrient limiting condition. In the ACCS2 knockdown cells, re-expression of WT ACSS2 increased lipid accumulation based on Oil Red O staining of differentiated adipocytes, thereby validating ACSS2’s functional role in DNL. Consistent with our above observation that ACSS2 K271R mutant is more stable than the WT, expressing the K271R mutant lead to more lipid droplets than expressing the WT ACSS2 (Figure 3—figure supplement 1).

To find out whether SIRT2-mediated deacetylation of ACSS2 K271 affect lipogenesis, we used the SIRT2 small molecule inhibitor, TM. The inhibition of SIRT2 using TM increased lipid accumulation in cells with WT ACSS2, but not in cells with the K271R mutant that cannot be deacetylated by SIRT2 (Figure 3C and D). To make sure the mutants were enzymatically active, we isolated Flag-tagged WT, K271R, and K271Q ACSS2 proteins from SIRT2 knockdown HEK293T cells. Subsequently, we examined acetyl-CoA formation from acetate and CoA using high-performance liquid chromatography (HPLC). Our findings indicate that while the WT ACSS2 exhibits slightly higher activity compared to the K271R and K271Q mutants, all variants remain functional (Figure 3—figure supplement 2). The slight reduction in acetyl-CoA formation for the K271R mutant may explain the relatively modest increase in lipid droplet formation observed in Figure 3D. These findings are consistent with the model that SIRT2 deacetylates ACSS2 on K271 to promote ACSS2 ubiquitination and degradation, and thus inhibiting lipogenesis (Figure 4).

Anti-Flag affinity gel (A2220), anti-Flag antibody conjugated with horse radish peroxidase (A8592, 1:10,000), and HA antibody (SAB4300603, 1:1000) were purchased from Sigma-Aldrich. Antibody against β-actin (sc-47778, 1:1000) was purchased from Santa Cruz Biotechnology. Antibodies against SIRT2 (#12650, 1:1000), ACSS2 (D19C6, 1:1000), Ubiquitin (E4I2J, 1:1000), K48-linkage Specific Polyubiquitin (#4289, 1:1000) were purchased from Cell Signaling Technology. PTM antibodies anti-Kac (cat. #PTM-101) were purchased from PTM Biolabs Inc (Chicago, IL, USA). Protease inhibitor cocktail, bafilomycin A1, cycloheximide, and adipocyte differentiation cocktail were purchased from Sigma-Aldrich. ECL plus western blotting detection was purchased from Thermo Scientific. Polyethyleneimine (PEI) was purchased from Polysciences (24765). MG132 was purchased from Cayman. TM was made as previously described (Jing et al., 2016).

ACSS2 expression vector with Flag tag (NM_018677) was purchased from Origene (Rockville, MD, USA) for mammalian expression. To create ACSS2 K271 mutant constructs, site-directed mutagenesis via Quick Change PCR amplification was performed using the following primers: K271R Forward: 5’-CAGTCCCCCCCAATTAGG AGGTCATGCCCAGAT-3’ Reverse: 5’- CCTAATTGGGGGGGACTGGCTGGTGGAGTCACC-3’. K271Q Forward: 5’ -CAGTCCCCCCCAATT CAG AGGTCATGCCCAGAT-3’. Reverse: 5’-CTGAATTGGGGGGGACTGGCTGGTGGAGTCACC-3’.

pRK5-HA-Ubiquitin-K48 was a gift from Ted Dawson (Addgene plasmid # 17605; http://n2t.net/addgene:17605; RRID:Addgene_17605).

SIRT2 WT and H187Y mutant expression vector with Flag tag was made as previously described (Jing et al., 2017).

ACSS2 expression plasmid was transfected into HEK293T cells using PEI transfection reagent following the manufacturer’s protocol. The pCMV-Tag4a empty vector was used as the negative control. After overnight transfection, cells were washed twice with ice-cold PBS and collected by centrifugation at 1000×g for 5 min. Cells were then lysed in 1 mL of 1% NP-40 lysis buffer (25 mM Tris-HCl, pH 7.8, 150 mM NaCl, 10% glycerol, and 1% NP-40) with protease inhibitor cocktail (1:100 dilution) by rocking at 4°C for 30 min. After centrifugation at 17,000×g for 30 min, the supernatant was collected, and protein concentration was determined with the Bradford assay (23200, Thermo Fisher). Normalized lysates were incubated with 20 μL of anti-Flag affinity gel or anti-Acetyl-Lysine (mAb mix) Affinity Beads (Cytoskeleton, Inc, AAC04-beads) at 4°C for 2 hr. The affinity gel was washed three times with washing buffer (25 mM Tris-HCl, pH 7.8, 150 mM NaCl, 0.2% NP-40). Beads were dried with gel loading tips and resuspended in 30 μL of 1× SDS loading dye. Western blot analysis was carried out according to standard methods.

HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (11965-092, Gibco) with 10% calf serum (C8056, Sigma-Aldrich). SIRT2 stable knockdown HEK293T cells were generated as previously described. Cells were washed with phosphate buffer saline (PBS) 24 hr post transfection and grown in EBSS (24010043, Thermo Fisher) with 10% calf serum for 16 hr to induce amino acid stress.

3T3-L1 (American Type Culture Collection, ATCC, Manassas, VA, USA) preadipocytes were cultured in high-glucose (400 mg/dL) DMEM (11965-092, Gibco) containing 10% fetal bovine serum (26140079, Gibco). For ACSS2 shRNA stable knockdown, lentivirus was generated by co-transfection of pLKO.1 with shRNA sequences specific to ACSS2 (TRCN0000045563, Millipore), pCMV-dR8.2, and pMD2.G plasmids into HEK293T cells. The cell medium was collected 48 hr after transfection and used to infect early passage cultures of 3T3L1 cells. After 72 hr, infected cells were treated with 1.5 μg/mL puromycin to select for stably incorporated shRNA constructs. The empty pLKO.1 vector was used as a negative control. ACSS2 WT and K271R mutant Flag-tagged expression plasmids were transfected into ACSS2 knockdown 3T3-L1 cells using PEI transfection reagent following the manufacturer’s protocol. The pCMV-Tag4a empty vector was used as the negative control. Differentiation of 3T3L1 cell lines were done according to the manufacturer’s protocol (DIF001-1KT, Sigma-Aldrich) 24 hr after transfection. Differentiation media was supplemented with 1.5 mM acetate (S5636, Sigma-Aldrich) and the cells were cultured for 3 days more post differentiation without changing media to mimic a nutrient limiting condition. Oil Red O of the differentiated cells was carried out according to standard protocol.

Overexpressed ACSS2-Flag was purified from SIRT2 control and knockout HEK293T cells by immunoprecipitation with anti-Flag affinity beads. The purified protein was digested with 1.5 mg of trypsin in a glass vial at 37°C for 2 hr, and then desalted using Sep-Pak C18 cartridge. The peptides were processed as previously described (Miller et al., 2022), and data was acquired using Xcalibur 2.2 operation software.

ACSS2 activity was monitored by detecting the acetyl-CoA formation using HPLC. A reaction buffer (60 mM potassium phosphate pH 7.5, 3 mM ATP, 0.1 mM CoA, 4 mM MgCl₂, 1 mM DTT, 0.24 mM sodium acetate) was prepared. Reactions were prepared with 0.04 μg of recombinant human ACSS2 protein. All samples were incubated at 37°C for the indicated time after which it was quenched with 10 μL of glacial acetic acid. After vortexing and centrifuging at 17,000×g for 5 min to remove the precipitated enzyme, the supernatant was loaded to HPLC with a Kinetex EVO C18 column (100×4.60 mm, 5 μM, 100 Å) for analysis. All experiments were performed in duplicate. The amount of acetyl-CoA formed over time was detected and quantified.

Statistical analysis was performed using Prism (GraphPad Software). Experimental values are shown as mean ± SEM. Statistical significance between two groups was determined using the two-tailed Student’s t-test. One-way ANOVA was applied for multigroup comparisons. p-Values<0.05 were considered significant.