CHFR negatively regulates SIRT1 activity upon oxidative stress

SIRT1, the NAD+-dependent protein deacetylase, controls cell-cycle progression and apoptosis by suppressing p53 tumour suppressor. Although SIRT1 is known to be phosphorylated by JNK1 upon oxidative stress and subsequently down-regulated, it still remains elusive how SIRT1 stability and activity are controlled. Here, we have unveiled that CHFR functions as an E3 Ub-ligase of SIRT1, responsible for its proteasomal degradation under oxidative stress conditions. CHFR interacts with and destabilizes SIRT1 by ubiquitylation and subsequent proteolysis. Such CHFR-mediated SIRT1 inhibition leads to the increase of p53 acetylation and its target gene transcription. Notably, CHFR facilitates SIRT1 destabilization when SIRT1 is phosphorylated by JNK1 upon oxidative stress, followed by prominent apoptotic cell death. Meanwhile, JNK inhibitor prevents SIRT1 phosphorylation, leading to elevated SIRT1 protein levels even in the presence of H2O2. Taken together, our results indicate that CHFR plays a crucial role in the cellular stress response pathway by controlling the stability and function of SIRT1.


CHFR ubiquitylates and promotes the proteasomal degradation of SIRT1. Given that SIRT1
is destabilized in the presence of CHFR as shown in Fig. 1A and they bind to each other, it is plausible that CHFR could act as a specific E3 Ub-ligase of SIRT1 to modulate its protein levels. In order to test this possibility, SIRT1 was transfected into HeLa cells, where CHFR is normally not expressed 4 , together with either mock-or CHFR-expression vector. SIRT1 protein levels were decreased in a CHFR dose-dependent manner and this reduction was blocked by the treatment of MG132 ( Fig. 2A), indicating that CHFR leads to the proteasomal degradation of SIRT1. On the contrary, either E3 Ub-ligase-defective CHFR-I306A 4 or SIRT1-binding-defective CHFR-Δ CR mutant failed to decrease SIRT1 protein levels compared to CHFR WT (Fig. 2B), suggesting that both an E3 Ub-ligase activity and a substrate binding ability of CHFR are necessary for SIRT1 destabilization.
We next sought to examine whether CHFR is able to ubiquitylate SIRT1 prior to its proteolytic degradation. HEK293T cells were transiently transfected with expression vectors encoding HA-Ub, FLAG-SIRT1, and MYC-CHFR WT or -I306A mutant, and treated with MG132. SIRT1 was heavily ubiquitylated by CHFR WT, but not by CHFR-I306A mutant (Fig. 2C). In addition, in vitro ubiquitylation assay under defined conditions was performed using purified E1, E2 (UbcH5b), FLAG-SIRT1, and His-CHFR (WT or I306A). CHFR WT efficiently catalyzed poly-ubiquitylation of SIRT1. However, there is no or little ubiquitylation of SIRT1 with CHFR-I306A mutant (Fig. 2D). Taken together, these results suggest that CHFR serves as a specific E3 Ub-ligase for SIRT1 ubiquitylation.
CHFR enhances p53 acetylation and its transcriptional activity. As SIRT1 is able to deacetylate p53 and suppress its transcriptional activity 12 , and CHFR facilitates SIRT1 degradation, we speculated that CHFR affects p53 functions through the inhibition of the SIRT1 activity. To test this hypothesis, p53 acetylation was monitored in HCT116 cells expressing p53, p300, SIRT1, and CHFR. We have previously reported that CHFR binds to and down-regulates HDAC1, resulting in the increase of p53 acetylation 4 . To rule out such possibility that CHFR-mediated HDAC1 destabilization influences p53 acetylation, cells were then treated with TSA, a class I/II HDAC inhibitor. As expected, p53 is deacetylated by the SIRT1 introduction (Fig. 3A, lanes 2 and 3). Ectopic expression of CHFR highly elevated the levels of p53 acetylation in accordance with the reduced SIRT1 protein levels (Fig. 3A, lanes 3 and 4). These results indicate that CHFR is able to inhibit SIRT1 function not to deacetylate p53. To further validate the biological consequences of CHFR-induced SIRT1 degradation, we examined the effect of CHFR on the p53 transcriptional activity using p53 response element-containing luciferase genes, i.e., PG13-luc and p21-luc. Consistent with a previous finding, SIRT1 inhibited p53-driven gene expression and this decrease was restored by CHFR co-expression (Fig. 3B,C). These data suggest that CHFR enhances the p53 transcriptional activity by destabilizing SIRT1 (as shown in the bottom panel of Fig. 3B) and inhibiting its deacetylase activity, which are illustrated in Fig. 3D. CHFR is responsible for SIRT1 degradation under oxidative stress conditions. As SIRT1 is known to be down-regulated upon oxidative stress 28 and we have shown thus far that SIRT1 is destabilized by CHFR, we investigated whether CHFR is involved in this oxidative stress-induced SIRT1 destabilization. To test this hypothesis, we utilized HeLa-CHFR stable cells to assess endogenous SIRT1 protein levels and found that the treatment of 1 mM H 2 O 2 for 6 h in cells was sufficient to decrease SIRT1 proteins ( Supplementary Fig. S1A). The turn-over rate of SIRT1 in the presence of H 2 O 2 was further determined using HeLa-control and HeLa-CHFR stable cells, which are identical except for expressing CHFR. Upon oxidative stress, SIRT1 was quickly destabilized only in HeLa-CHFR cells, indicating that CHFR is responsible for H 2 O 2 -induced SIRT1 degradation (Fig. 4A). These reduced SIRT1 protein levels were restored by the co-treatment of MG132 with H 2 O 2 , suggesting that SIRT1 is degraded by the ubiquitin-proteasome system upon oxidative stress (Fig. 4B). We have then performed the co-immunoprecipitation assay in both endogenous and transiently transfected conditions to examine whether H 2 O 2 treatment affects the interaction between CHFR and SIRT1. Since SIRT1 is destabilized in the presence of CHFR upon H 2 O 2 treatment, we have utilized either CHFR-stable cells or E3 Ub ligase-defective CHFR-I306A mutant and analysed the binding differences under the H 2 O 2 -induced oxidative stress conditions. While the binding degree of CHFR to SIRT1 was slightly increased upon H 2 O 2 treatment, apparently, it did not seem the all-or-none differences in their interaction ( Fig. 4C and Supplementary Fig. S1B). Given that SIRT1 is phosphorylated by JNK1 under oxidative stress conditions 24,25 , we investigated whether JNK signaling is linked to CHFR-mediated SIRT1 turn-over upon oxidative stress. When cells were treated with H 2 O 2 , the JNK pathway was activated, which was validated by the induction of phosphorylated JNK and phosphorylated c-Jun, and consequently, SIRT1 was destabilized. On the other hand, the treatment of JNK inhibitor SP600125 in cells together with H 2 O 2 inactivated the JNK pathway and simultaneously blocked SIRT1 destabilization (Fig. 4D). Since the treatment of hydrogen peroxide is more likely to induce acute and instant damage to cells due to its quick removal by cells, we have reiterated the H 2 O 2 -driven SIRT1 destabilization under chronic oxidative stress conditions by the glucose oxidase (Gox) enzyme (Fig. 4E). Although glucose oxidase-induced chronic oxidative stress resulted in much stronger and prolonged damage to cells, consistent with the previous results shown in Fig. 4A, SIRT1 protein levels were decreased only in the presence of CHFR upon oxidative stress regardless of the type of oxidative stress triggers. Therefore, these results indicate that phosphorylated SIRT1 by JNK1 under oxidative stress conditions is destabilized by CHFR.
CHFR promotes oxidative stress-induced cell death by destabilizing SIRT1. Given that CHFR is able to negatively regulate SIRT1 by ubiquitylation-mediated proteasomal degradation, we aimed to explore the biological outcomes of SIRT1 destabilization by CHFR, especially under oxidative stress conditions. In line with our previous results shown in Fig. 4, SIRT1 protein levels were significantly lower in H 2 O 2 -treated HeLa-CHFR cells compared to mock-treated cells. On the contrary, there was not much difference of SIRT1 in between mockand H 2 O 2 -treated HeLa-control cells, indicating that CHFR is the underlying cause of SIRT1 degradation upon oxidative stress (Fig. 5A, top panel). This was further illustrated by the apoptosis assay measuring Annexin V and propidium iodide (PI) fluorescence in H 2 O 2 -treated HeLa-CHFR cells. Cell death was increased in HeLa-CHFR cells compared to control cells, and much greatly augmented by the H 2 O 2 treatment (Fig. 5A, bottom panel). Next, we took a closer look at apoptotic events to further delineate how CHFR affects H 2 O 2 -driven cell death in   HCT116-CHFR stable cells. Stained cells were sub-divided into four quadrants according to Annexin V and PI positivity. As cells were treated with H 2 O 2 for 6 h, viable (Annexin V− /PI− ) cells were decreased, while apoptotic (Annexin V+ ) and necrotic (PI+ ) cells were increased. Notably, H 2 O 2 -treated HCT116-CHFR cells showed the highest early apoptotic (Annexin V+ /PI− ) cell death among all tested (Fig. 5B). Moreover, Annexin V and PI fluorescence microscopy in HCT116-CHFR cells also revealed that CHFR is responsible for increased cell death upon oxidative stress (Fig. 5C). We have then investigated the effect of chronic oxidative stress-induced by Gox on apoptotic cell death along with SIRT1 protein levels. Cell death was much greatly augmented according to the strength of oxidative stress in CHFR-expressing cells (Fig. 5D,E) and similarly, SIRT1 protein levels were decreased in line with this increased cell death (Supplementary Fig. S1C). This was further validated by the cell viability assay, indicating that chronic oxidative stress by Gox leads to massive cell death (Fig. 5F).
Collectively, our data highlight that SIRT1 stability and function were negatively regulated by CHFR-mediated ubiquitylation and subsequent proteolysis. The inhibition of SIRT1 in human cancer cells by CHFR expression leads to elevated acetylation of p53 and simultaneous trans-activation of p53-driven target genes to elicit apoptosis in response to oxidative stress (Fig. 5G).
It is worthy of note that not only SIRT1 but also CHFR were destabilized by H 2 O 2 in CHFR-stable cells. Since CHFR is known to be regulated by its own auto-ubiquitylation activity 29 , it would be of interest to study whether the E3 Ub-ligase activity of CHFR is controlled by the JNK signaling pathway. As both CHFR and SIRT1 have been implicated in cell cycle control and tumorigenesis, it would also be of particular interest to investigate how and when CHFR and/or SIRT1 respond to diverse cellular stresses during tumour progression. Here, we aimed to add a new line of evidence how CHFR contributes to tumour suppression. Especially, we have shown that CHFR is able to suppress not only HDAC1-class I HDAC 4 , but also SIRT1-class III HDAC. CHFR is often epigenetically inactivated in various cancer cells 27 , and reduced CHFR expression in normal cells leads to tumorigenic phenotypes 30 . Accordingly, such CHFR malfunction may lead to SIRT1 stabilization, which in turn represses p53 and other tumour suppressors to accelerate tumour initiation and metastasis. We have shown thus far that CHFR elevates the p53 activity by destabilizing SIRT1 when CHFR was re-introduced into cancer cells. Since p53 is widely regarded as "the guardian of genome" 31,32 , it is plausible that CHFR becomes a part of the watchman to keep cells under surveillance. This reinforces the role of CHFR as a tumour suppressor. CHFR also acts as a cell cycle checkpoint 33 , therefore, CHFR helps to maintain the cellular integrity against harmful stimuli and the threshold for apoptosis and cell senescence.

Methods
Cell culture, Transfection, and Reagents. HCT116, HeLa, and HEK293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 100 U ml −1 penicillin, 100 μ g ml −1 streptomycin, and 10% FBS (Gibco) at 37 °C in a humidified 5% CO 2 condition. Transient and stable transfections were carried out using either lipofectamine 2000 (Invitrogen) or polyethylenimine (Sigma) according to the manufacturer's instructions. Following chemical reagents used in the study were obtained from Sigma or otherwise stated: TSA, hydrogen peroxide, glucose oxidase, and SP600125.
Immunoprecipitation and Immunoblotting. For immunoprecipitation, cells were lysed in TNET buffer (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1 mM EDTA, and 0.2% Triton X-100) and 1x protease inhibitor cocktail (Roche). Cell lysates were incubated with anti-FLAG M2 affinity resin (Sigma) for 2 h at 4 °C. Resins were collected by centrifugation and washed three times with TNET buffer. Bound proteins were eluted, resolved by SDS-PAGE, and immunoblotted with appropriate antibodies. The following antibodies were used: anti-SIRT1, anti-MYC, anti-GST, anti-GAPDH, anti-HA, and anti-p53 (Santa Cruz Biotechnology); anti-acetyl p53 (Millipore); anti-FLAG and anti-β -actin (Sigma); anti-Xpress (Invitrogen); peroxidase-conjugated AffiniPure goat anti-rabbit and anti-mouse IgGs (Jackson ImmunoResearch); anti-CHFR antiserum was raised against a recombinant His-CHFR. Relative protein levels in the immunoblot figures were quantitated by ImageJ and normalized to either β -actin or GAPDH levels. Values are plotted as the mean ± SEM of at least three independent experiments.
GST pull-down assay. GST-SIRT1 was purified from Escherichia coli and His-CHFR was purified from Sf9 insect cells. GST-SIRT1 (1 μ g) and His-CHFR (1 μ g) were incubated with Glutathione Sepharose 4 Fast Flow (GE Healthcare) for 1 h at 4 °C. After incubation, bound proteins were eluted, resolved by SDS-PAGE, and analysed by immunoblotting with anti-CHFR and anti-GST antibodies.
Reporter assay. HCT116 cells were transfected with indicated plasmids with β -gal constructs and treated with 0.5 μ M TSA for 6 h before harvest. Luciferase activity was measured in a luminometer with a luciferase system (Promega) and normalized to β -galactosidase activity. Values were expressed as mean ± SEM from three independent experiments. Apoptosis assay. Cells were treated with either hydrogen peroxide or glucose oxidase for indicated times to induce oxidative stress, stained with either Alexa Fluor ® 488 Annexin V/Dead Cell Apoptosis Kit (Molecular Scientific RepoRts | 6:37578 | DOI: 10.1038/srep37578 Probes) or FITC Annexin V Apoptosis Detection Kit (BD Biosciences) according to manufacturers' instructions, and analysed using the TaLi ® image-based cytometer (Invitrogen) or visualized under the EVOS TM cell imaging system (Thermo Fisher Scientific). DAPI was used to counterstain the nuclei.
Cell viability assay. Cells were seeded at a density of 10 4 cells/well in 100 μ L of culture medium in a 96-well plate and treated with either hydrogen peroxide or glucose oxidase for 6 h to induce oxidative stress. Cells were treated with 10 μ L of CellVia (water-soluble tetrazolium salt, WST-1; Young In Frontier) and incubated for an additional 2 h at 37 °C. Cell viability was measured using a multiwell microplate reader at a wavelength of 450 nm along with a reference wavelength of 650 nm. The same volume of culture medium plus CellVia reagent in the absence of cells were used as a blank control to subtract the background absorbance. Values were expressed as mean ± SEM from three independent experiments.