SIRT2 suppresses expression of inflammatory factors via Hsp90‐glucocorticoid receptor signalling

Abstract SIRT2 is a NAD+‐dependent deacetylase that deacetylates a diverse array of protein substrates and is involved in many cellular processes, including regulation of inflammation. However, its precise role in the inflammatory process has not completely been elucidated. Here, we identify heat‐shock protein 90α (Hsp90α) as novel substrate of SIRT2. Functional investigation suggests that Hsp90 is deacetylated by SIRT2, such that overexpression and knock‐down of SIRT2 altered the acetylation level of Hsp90. This subsequently resulted in disassociation of Hsp90 with glucocorticoid receptor (GR), and translocation of GR to the nucleus. This observation was further confirmed by glucocorticoid response element (GRE)‐driven reporter assay. Nuclear translocation of GR induced by SIRT2 overexpression repressed the expression of inflammatory cytokines, which were even more prominent under lipopolysaccharide (LPS) stimulation. Conversely, SIRT2 knock‐down resulted in the up‐regulation of cytokine expression. Mutation analysis indicated that deacetylation of Hsp90 at K294 is critical for SIRT2‐mediated regulation of cytokine expression. These data suggest that SIRT2 reduces the extent of LPS‐induced inflammation by suppressing the expression of inflammatory factors via SIRT2‐Hsp90‐GR axis.

For example, SIRT2 deacetylates and activates PGAM (glycolytic enzyme phosphoglycerate mutase), which results in a decrease in cellular NADPH levels. 12,13 Moreover, SIRT2 can deacetylate G6PD (glucose-6-phosphate dehydrogenase), in which deacetylation of K403 results in the formation of an active dimer, leading to production of cytosolic NADPH. 14 SIRT2 has also been shown to deacetylate HIF-1α under hypoxic conditions, leading to HIF-1α hydroxylation and ubiquitination followed by degradation. 15 More recently, there has been increasing evidence that indicates SIRT2 is involved in inflammatory processes. In animal models of collagen-induced arthritis, SIRT2 inhibits the inflammatory response and relieves physical impairment. 16 Remarkably, expression level of SIRT2 mRNA was dramatically decreased in extracellular plasma of rheumatoid arthritis patients compared with the healthy controls. 17 Moreover, a cell permeative fusion protein, PEP-1-SIRT2, significantly blocks the expression of cytokines under lipopolysaccharide (LPS) stimulation. 18 However, some results are controversial. For example, repression of SIRT2 activity has been shown to down-regulate LPS-induced inflammatory response via TNF-α and IL-6, 19 and knockdown of SIRT2 has also been shown to repress microglial activation induced by LPS. 20 Overall, there is a tendency for data to support a role for SIRT2 in the repression of an excessive immune activation.
SIRT2 has been shown to deacetylate histone H3K18 in the immune response against bacterial infection 21 and, in turn, modulate the expression of anti-inflammatory genes. Other recent reports demonstrate that SIRT2 can repress the inflammatory response via deacetylation and inhibition of NF-kB, using transgenic (SIRT2 -/-) mice, in models of sepsis or colitis. [22][23][24] Repression of SIRT2 expression via forced expression of miR-339 increases the acetylation level of NF-kB, which may participate in the stress-induced immune response. 25 However, other than H3 and NF-kB, SIRT2 exhibits the ability to antagonize inflammation by previously undetermined mechanisms.
In the present study, we identified Hsp90 as a novel substrate of SIRT2. Hsp90 is known to regulate GR translocation to nucleus, modulating the expression of downstream genes. Under resting conditions, Hsp90 and GR form a complex within the cytoplasm.
Typically, glucocorticoid stimulation leads to the dissociation of the GR from Hsp90. This allows GR to form homodimers and translocate to the nucleus, where it forms heterocomplex with NF-kB, preventing the transcriptional activation of inflammatory genes. 26 Inhibition of Hsp90 also attenuates activation of NF-kB and inhibits expression of its downstream genes, 27,28 indicating a regulatory role for Hsp90 in the NF-kB-dependent inflammatory response. Accordingly, secreted Hsp90 alpha (eHsp90α) into the extracellular stromal matrix promotes an inflammatory reaction in the tumour microenvironment in an MEK-ERK-and NF-kBdependent manner. 29 Acetylated rat Hsp90α at K294 (Scroggins BT et al, but site corresponds to K295 in new version of rat Hsp90α: NP_786937) promotes its binding to its interacting partners 30 ; therefore, we speculated that SIRT2 may repress immune reaction via Hsp90-GR signalling. Our results provide new insights into the regulation of the inflammation via SIRT2.

| Hsp90 mutation and plasmid construction
The coding sequence of rat SIRT2 (rSIRT2) was amplified from a previously constructed SIRT2-pIRES2-EGFP vector 11 with the upstream and downstream primers 5′-GGATCCATGGACTTCCTACGGAATTT-3′ and 5′-CTCGAGCTAGTG TTCCTCTTTCTCTT-3′, respectively. The product was cloned into T-easy vector (Promega) for verification of sequencing, and then, the coding region was subcloned into pGEX-6P-1 vector via restriction sites EcoR I and Xho I. Then, GST-SIRT2 fusion protein was expressed in E coli using the SIRT2-pGEX vector upon IPTG induction.
An overlap PCR was performed to yield an acetylation-mimic mutation and acetylation-null mutation of rat Hsp90α (rHsp90α) at K294. In brief, the upper and lower segments of rHsp90α were amplified by PCR from cDNA clone (OriGene, Rockville, MD). Both segments are purified from an acrylamide gel and mixed together with Phusion DNA polymerase (Bio-Rad, CA), PCR buffer and dNTP.

| Pull-down and Co-Immunoprecipitation (Co-IP) with western blot
The GST-SIRT2 fusion protein was purified as described above.
B104 cell was treated with or without 10 ng/mL LPS for 12 hours prior to harvesting. Cultured B104 cells were lysed in RIPA buffer on ice for 10 minutes and vortexed intermittently. The sample was

| Isolation of nuclei from cytoplasm
Nuclei were isolated from B104 cells with overexpression or knocked-down of SIRT2, as described previously. 33  to SDS-PAGE, and purity was assessed by antibodies against lamin B1 (nuclei) or β-actin (cytoplasm) as described under 'Western blot'.

| GRE-driven dual-luciferase reporter assay
The binding ability of GR to glucocorticoid response element (GRE) was examined using a luciferase reporter gene assay as previously described. 34 The reporter plasmids, negative control plasmids and positive control plasmids were transfected into B104 cells according to manufacturer instructions of Cignal CRE Reporter Assay Kit (Qiagen). In addition, the reporter plasmids were transfected into cells infected with lentiviral particles for SIRT2 overexpression or SIRT2 knocked-down.  After washing 3 times, protein bands were visualized using ECL immunoblot detection reagents, as previously described. 11 All assays were repeated at least three separate times yielding similar results, and a representative result blots are shown.

| Statistical analysis
Fluorescence value and qPCR data are presented as mean ± standard error of the mean (SEM). Statistical analysis was performed with Student's t test in GraphPad Prism. Significance was set at P < 0.05(*) or P < 0.01(**) as indicated.

| Expression of GST-SIRT2 in E coli with low solubility
GST-SIRT2 and GST were expressed in E coli, and the solubility was assessed by SDS-PAGE. Coomassie staining indicates that GST-SIRT2 predominantly remained in inclusion bodies with only a small proportion was soluble in the supernatant ( Figure 1A). In comparison, GST remains largely soluble. Only the soluble fraction of GST-SIRT2 fusion protein was purified for use in the following experiments. In brief, both GST-SIRT2 and GST are highly purified as shown in Figure 1A.

| Hsp90α identified as a novel interacting partner of GST-SIRT2
In the pull-down assay, a new band was present in the GST-SIRT2 lane in the Coomassie stained gel with a molecular weight around 100 kD ( Figure 1B). There was no band visible at the corresponding molecular weight in either the glutathione bead or GST lanes, indicating a novel protein that interacts with SIRT2. Subsequently, this band was excised and sent for protein identification via mass spectrometry. A Mascot search of the NCBInr database identified this protein as rat Hsp90α (NP_786937) based on matched peptides corresponding to 48% sequence coverage ( Figure 1C).

| SIRT2 associates with Hsp90α in B104 cells
To confirm a physical association between SIRT2 and Hsp90α, GST-SIRT2 fusion protein was fixed on glutathione beads to pull-down Hsp90α in B104 cells with or without LPS stimulation. GST-SIRT2 pulled down Hsp90α from cell lysate in both stimulated and nonstimulated cells compared with GST alone (Figure 2A,B); however, the association between SIRT2 and Hsp90α appears to increase in cells stimulated with LPS in the lane of GST-SIRT2 ( Figure 2B compared to A). Next, binding of endogenous SIRT2 to Hsp90α was examined by Co-IP. As indicated in Figure 2C-F, SIRT2 and Hsp90α can be detected in Co-IP complexes. This was shown using anti-Hsp90α for IP followed by detection of SIRT2 by immunoblot ( Figure 2C), or using anti-SIRT2 for IP followed by detection of Hsp90α by immunoblot ( Figure 2D).

| SIRT2 deacetylates Hsp90α in B104 cells
To determine the functional relationship between SIRT2 and Hsp90α, B104 cells were infected by SIRT2 overexpression and F I G U R E 1 A, Rat SIRT2 (rSIRT2) coding region was cloned into pGEX-6P-1 expression vector. Analysis of expression and purification revealed that GST-rSIRT2 predominantly expressed in inclusion bodies (pellet), but GST is mainly in soluble fraction (supernatant). Subsequently, both GST-rSIRT2 and GST were purified from supernatant of the lysate. B, In a pull-down assay, a new protein band was isolated with GST-rSIRT2 compared with GST alone or glutathione beads. This unknown protein has an approximate molecular size of 100 kD. C, The new protein band was excised for identification by mass spectrometry, and results showed 48% sequence coverage with rat Hsp90α has been knocked-down.

| Deacetylation of Hsp90 via SIRT2 facilitates GR translocation to nuclei
Previous reports have shown Hsp90, a molecular chaperone, interacts with the GR receptor. 37 To investigate the role SIRT2 in Subsequently, the effect of SIRT2 on GR subcellular location was explored using nuclei and cytoplasm fractionation of cell lysates. SIRT2 overexpression facilitated GR translocation to nuclei, where GR was predominantly detected ( Figure 4C, left lanes labelled +). In contrast, when SIRT2 expression was knocked-down, GR was mainly localized in the cytoplasm ( Figure 4C, right lanes labelled kD). Thus, SIRT2 expression altered the proportion of GR in nuclei versus cytoplasm. Taken together, these data indicate that SIRT2 deacetylates Hsp90 inducing its disassociation with GR, leading to GR translocation to nuclei.
Nuclear translocation of GR would result in binding to the GRE (glucocorticoid response element) on genomic DNA and transcriptional activation or repression of downstream genes. To test effects of SIRT2 on the association of GR with GRE in B104 cells, a GR-driven luciferase reporter assay was performed ( Figure 4D). The results showed that SIRT2 overexpression enhanced reporter gene expression compared with control cells. Moreover, the fluorescence value is also greater than SIRT2 knocked-down, suggesting that SIIRT2 effectively increase F I G U R E 4 Co-IP assays was carried out with anti-Hsp90 in B104 cells in which SIRT2 was overexpression or knocked-down, and glucocorticoid receptor (GR) was detected in the IP complex by immunoblot (A). SIRT2 overexpression resulted in decreased interaction between Hsp90 and GR compared with SIRT2 knock-down. Co-IP assays were also performed with anti-GR in SIRT2 overexpression, SIRT2 knocked-down or control cells, and Hsp90 was detected in the IP complex by immunoblot (B; top panel). Co-IP with anti-Hsp90 under the same conditions, followed by detection of acetylated Hsp90 (B; bottom panel), showed that the interaction between Hsp90 and GR corresponds to the acetylation levels of Hsp90. (C) GR subcellular location was detected in the SIRT2 overexpression or SIRT2 knockeddown cells, with β-actin and lamin B served as markers for the fractions of cytoplasm and nucleus, respectively. SIRT2 overexpression increased the relative expression of GR in the nucleus compared with SIRT2 knock-down. Nuclei isolation was repeated at least three times, and a representative immunoblot is shown. (D) To examine the effect of SIRT2 on the binding capacity of GR to glucocorticoid receptor response element (GRE), GR-driven dual-luciferase reporter assay was performed in the B104 cells, in which rSIRT2 was overexpressed or knocked-down, 50 as described in the Materials and Methods. When rSIRT2 was overexpressed, expression of the reporter gene was significantly up-regulated compared with rSIRT2 was knocked-down or control cells (each experiment was performed in triplicate and represented as mean ± SEM; t test, *P < 0.05 or **P < 0.01   Figure 5H) compared with wildtype Hsp90 ( Figure 5I). Similarly, K294R also abolished the effect of SIRT2 expression ( Figure 5I). Interestingly, the effect of K294Q on cytokine expression is greater with LPS stimulation compared with control conditions ( Figure 5H). This was also observed for K294R, but only for the expression of TNFα ( Figure 5I). To examine significance of Hsp90 in SIRT2 mediating cytokine expression, Hsp90 expression was knocked-down using siRNA ( Figure 5J). We found that knock-down of Hsp90 totally reduced expression of TNFα and IL-6 in a SIRT2 independent manner ( Figure 5K). These results support a repressor role for SIRT2 in the inflammatory response. As depicted schematically in Figure 6, we propose a novel signalling pathway by which SIRT2 regulates inflammation by repressing expression of the inflammatory cytokines through Hsp90-GR signalling.  been recognized as a regulator of inflammatory reaction. 19,21,42 In the present study, Hsp90 has been identified as a substrate of SIRT2 through pull-down and mass spectrometry, showing 48% sequence coverage to the rat Hsp90α (NP_786937). Hsp90 is a known molecular chaperone, 43 and its interaction with the GR has well been characterized. 44 Hsp90 plays a critical role in regulation of GR translocation to nuclei. 44 Figure 6 depicts the molecular mechanism by which SIRT2

| D ISCUSS I ON
can regulate the inflammatory response through SIRT2-Hsp90-GRcytokine repression.
Recently, Min  Several other studies also indicate that SIRT2 deacetylase activity modulates the inflammatory reaction. 21,22 For example, SIRT2 inhibits inflammation in animal models of collagen-induced arthritis, sepsis, or colitis. 16,[22][23][24] More recently, SIRT2 mRNA expression levels of extracellular plasma was found to be reduced in 54 rheumatoid arthritis patients compared with healthy controls. 17 It is possible that a resultant decrease in SIRT2 protein expression contributed to the up-regulated expression and release of inflammatory factors. As a result, inflammation was exacerbated. NF-kB has been identified as a target substrate of SIRT2 in modulating inflammation [22][23][24] ; however, in many studies, the precise molecular mechanism of SIRT2 action remains unknown. Here, we have characterized a novel pathway by which SIRT2 influences the inflammatory response ( Figure 6). Based on our finding, we postulate that deacetylation of Hsp90 releases GR allowing it to dimerize and translocate to the nucleus. GR binding to GRE, or possibly by also forming a heterocomplex with NF-kB, 26 inhibits the transcriptional activation of inflammatory cytokines.
A mild inflammatory reaction is necessary for an organism to fight infections and remove pathogens, but excessive inflammation may be detrimental. Sirtuins may play a critical role in balancing the extent of an inflammatory reaction. A growing body of evidence suggests that SIRT2 down-regulates excessive inflammation via multiple and partially redundant mechanisms. Intriguingly, it has been reported that loss of SIRT1 led to down-regulation of inflammatory cytokine gene expression 48 while treatment with a SIRT1 agonist was able to overcome immunosuppression 49 suggesting that SIRT1 enhances the inflammatory reaction. Thus, SIRT1 and SIRT2 may have apposing effects on inflammation, adding another level of complexity to the control Sirtuin family of deacetylases exerting over cellular function. In the future, SIRT2 may have potential as a treatment target to regulate the exacerbated immune responses associated with inflammatory-related diseases. Aj. Nazarali contributed to design assays and discussed the results.

ACK N OWLED G EM ENTS
S. Ji designed the project and interpreted the data.

E TH I C A L A PPROVA L
This research article does not contain any investigations or studies with human participants or animals performed by any author.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.