Acetylation of glucokinase regulatory protein decreases glucose metabolism by suppressing glucokinase activity

Glucokinase (GK), mainly expressed in the liver and pancreatic β-cells, is critical for maintaining glucose homeostasis. GK expression and kinase activity, respectively, are both modulated at the transcriptional and post-translational levels. Post-translationally, GK is regulated by binding the glucokinase regulatory protein (GKRP), resulting in GK retention in the nucleus and its inability to participate in cytosolic glycolysis. Although hepatic GKRP is known to be regulated by allosteric mechanisms, the precise details of modulation of GKRP activity, by post-translational modification, are not well known. Here, we demonstrate that GKRP is acetylated at Lys5 by the acetyltransferase p300. Acetylated GKRP is resistant to degradation by the ubiquitin-dependent proteasome pathway, suggesting that acetylation increases GKRP stability and binding to GK, further inhibiting GK nuclear export. Deacetylation of GKRP is effected by the NAD+-dependent, class III histone deacetylase SIRT2, which is inhibited by nicotinamide. Moreover, the livers of db/db obese, diabetic mice also show elevated GKRP acetylation, suggesting a broader, critical role in regulating blood glucose. Given that acetylated GKRP may affiliate with type-2 diabetes mellitus (T2DM), understanding the mechanism of GKRP acetylation in the liver could reveal novel targets within the GK-GKRP pathway, for treating T2DM and other metabolic pathologies.


Results
GKRP is prominently acetylated at lysine 5 by p300. In humans, GKRP expression is highest in the liver 10 , and we first examined GKRP mRNA in tissues from C57B/6J mice. While GKRP mRNA levels were (expectedly) highest in the liver, unlike humans, mouse white adipose tissue also expressed considerable GKRP mRNA ( Supplementary Fig. S1).
Acetylated GKRP acetylation resists ubiquitin-mediated degradation. To assess the effect of acetylation on GKRP stability, we treated HeLa cells with the aforementioned HDACIs, NAM and TSA, resulting in increased acetylated GKRP protein levels ( Fig. 2A). Conversely, immunoblotting demonstrated that Myc-GKRP coexpression with HA-tagged full-length p300 increased, and anti-p300 siRNA decreased, GKRP protein levels (Fig. 2B,C). These results strongly suggest that GKRP acetylation by p300 might mediate its stability.
To explore regulation of GKRP stability by ubiquitin-dependent degradation, GKRP was coexpressed with HA-tagged ubiquitin in the absence or presence of MG132, a proteasome inhibitor. That assay demonstrated that downregulated proteasome activity substantially elevated ubiquitin-conjugated GKRP (Fig. 2D). To further confirm whether K5 acetylation increases GKRP stability, HA-tagged ubiquitin was co-expressed with WT or each of the two GKRP K5 mutants (K5R and K5Q), in the presence of MG132. Those specific combinations showed that indeed, GKRP ubiquitination increased in the K5R mutant, which mimics the deacetylated state, compared to the WT or the K5Q mutant, which mimics acetylated GKRP (Fig. 2E). Furthermore, GKRP degradation after cycloheximide (CHX, an inhibitor of protein biosynthesis) treatment showed that about 60% of the total GKRP protein remained after 8-hr HDACI Scientific RepoRts | 5:17395 | DOI: 10.1038/srep17395 (NAM and TSA) treatment, while GKRP degraded completely, after 4 h, in the absence of those HDACIs (Fig. 2F, p ≤ 0.001). These results support the conjecture that GKRP acetylation by p300 protects it from ubiquitin-dependent proteasomal degradation.
Next, we performed ubiquitin degradation assays using each of the acetyl-mimic and deacetyl-mimic forms of GKRP. As shown in Fig. 3A,B, GK robustly interacted with the WT and the K5Q GKRP acetyl-mimic, but significantly less with the K5R GKRP deacetyl-mimic (Fig. 3C,D, p ≤ 0.001). Moreover, acetylated GKRP showed increased interaction with GK in vitro ( Supplementary Fig. 3A,B). From these results, we speculated that GKRP acetylation promotes its interaction with GK. As a result, GK-GKRP  complex formation is believed to be critical for regulating GK activity and cytosolic glycolysis, consistent with a previous finding that GKRP acetylation caused GK nuclear retention 22 . To visualize whether GKRP K5 acetylation affects nuclear retention of the GKRP-GK complex, due to glucose "master sensors", HeLa cells were incubated in 5.5 mM glucose, and immunofluorescence microscopy then performed (  panel), consistent with other studies of this phenomenon 23 . As HDACI treatment similarly increased nuclear retention of the complex (Fig. 3E, lower panel), taken together, these results solidly suggest that GKRP acetylation increases nuclear retention of GK. GKRP acetylation decreases glycolytic flux. We next hypothesized that if GKRP acetylation increases its interaction with GK in the nucleus, cytosolic glycolysis should be reduced. Consequently, we overexpressed GK in HeLa cells and measured glycolytic flux (i.e., basal glycolysis, glycolytic capacity, and glycolytic reserve) by assessing the extracellular acidification rate (ECAR), after sequential treatment with glucose, oligomycin (an inhibitor of mitochondrial respiration), and 2-deoxyglucose (an inhibitor of glycolysis), in glucose-free Seahorse assay media ( Fig. 4A and Supplementary Fig. S4A). Under those conditions, glycolytic flux significantly increased in HeLa cells overexpressing GK ( Fig. 4B-D, p ≤ 0.001); that effect was negated by expression of the WT or K5Q mutant GKRP ( Fig. 4B-D, p ≤ 0.05 (*) or p ≤ 0.001 (***)). In contrast, there was no significant difference in glycolytic flux in GK-or GKRP-K5R mutant-overexpressing cells ( Fig. 4B-D), nor was it changed in the WT or K5Q or K5R GKRP mutants, in the absence of GK (see Supplementary Fig. S4B-D). Together, these data strongly suggest decreased glucose utilization when GKRP is acetylated.
Increased GKRP acetylation in db/db mice. To assess the functional relevance of GKRP acetylation to T2DM, we performed acetylation studies in the db/db mouse model of diabetes, dyslipidemia, and obesity 24 . As shown in Fig. 5A,B, GKRP acetylation levels were elevated in db/db mice, in addition to its increased interaction with GK ( Fig. 5C,D, p ≤ 0.05). These findings suggest that increasing GKRP stability and interaction with GK suppresses GK activity, thus impairing glucose homeostasis, in db/db mice. Figs 1A and 6A, GKRP acetylation increased when HeLa cells were treated with nicotinamide (NAM), an inhibitor of the NAD + -dependent class III ("sirtuin") family of histone deacetylases (HDACs) 25 (Fig. 1A). Co-immunoprecipitation assays and immunoblots further showed sirtuin 2 (SIRT2) interaction with GKRP (Fig. 6B), although other sirtuins (1 and 3-7) interacted with GKRP (see Supplementary Fig. S5A). To determine if any other sirtuins deacetylate GKRP, HeLa cells were co-tranfected with expression vectors for Myc-GKRP, various sirtuins, and p300. As shown in Supplementary Fig. S5B, GKRP was deacetylated only by SIRT2, but no other sirtuins. Further, co-overexpression of SIRT2, but not its catalytically inactive H187Y mutant, with p300, resulted in decreased deacetylation of GKRP (Fig. 6C). In addition, GKRP deacetylation by SIRT2 was reversed by NAM, but not by TSA (Fig. 6D), which does not inhibit class III HDACs. These results suggest that GKRP deacetylation is catalyzed by SIRT2.

SIRT2 deacetylates GKRP. As shown in
We also assessed the effect of GKRP deacetylation on glycolytic flux, in the presence of WT or the SIRT2 inactive mutant H187Y (Fig. 6E). Under those conditions, glycolytic flux was significantly decreased when GK was co-expressed with GKRP, as compared to GK alone ( Fig. 6F-H, p ≤ 0.05 and p ≤ 0.001); that effect was negated by expression of WT SIRT2 (Fig. 6F-H, p ≤ 0.01 and p ≤ 0.001). In contrast, there was no significant difference in glycolytic flux in cells overexpressing GK-GKRP or the GK-GKRP-H187Y mutant (Fig. 6F-H, p ≤ 0.05). Together, these data demonstrate increased glucose utilization when GKRP is deacetylated.

Discussion
Post-translational modifications (PTMs) are an important means for transducing signals that "sense" metabolic changes in the body and restore homeostasis 18 . One such PTM, protein acetylation, often in concert with other PTMs, regulates metabolic enzymes 26,27 . However, whether acetylation impacts the metabolic effects of GK and its interacting partner, GKRP, had not yet been elucidated. This study demonstrates that GKRP Lys-5 is the principle site of acetylation by p300, as confirmed by LC-MS/MS and mutational analysis. Although Lys-5 was identified as the major acetylation site of GKRP, computational analysis predicted other acetylation sites (Supplementary Table 2). Most protein acetylation affects intracellular compartmentalization, transactivation, stability, or competition with other PTMs such as ubiquitination and phosphorylation 17,28 . For instance, acetylation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κ B), at distinct lysine residues, affects numerous biological processes, including transcriptional activation, DNA-binding affinity, and subcellular localization 29 . Likewise, other GKRP PTMs may yield different functional outcomes; this possibility requires further investigation. Based on these other examples of protein acetylation, we hypothesized that acetylation of GKRP could affect  HeLa cells transfected with expression plasmids of Myc-tagged GKRP and Flag-tagged SIRT2 were lysed, immunoprecipitated, and subjected to immunoblot with antibodies to Flag or Myc. (C) SIRT2 deacetylates GKRP in HeLa cells transfected with the indicated plasmids. Protein was precipitated with anti-Myc antibody and immunoblotted using anti-Ac-Lys or anti-Flag or anti-Myc antibodies, respectively. (D) GKRP deacetylation by SIRT2 is reversed by NAM. HeLa cells were transfected with the indicated plasmids, treated with NAM (5 mM) and TSA (1 μ M) for 6 h, and precipitates subjected to immunoblot with antibodies to Ac, Myc or Flag. β -actin, protein level were used as an internal control. Deacetylation of GKRP regulates glycolytic flux. (E-H) HeLa cells were seeded in V7 cell plates at a density of 10,000 cells/well. Glycolysis assays were performed using a glycolytic stress test kit, according to manufacturer's protocol, on a XF24 instrument (Seahorse Biosciences). (E) A representative XF24 graph showing the ECAR response to glucose, oligomycin, and 2-deoxyglucose in Seahorse glucose-free medium. (F) Basal glycolysis calculated relative to its biological function, a phenomena previously observed for the farnesoid X receptor (FXR) and p53, both known acetylation targets of p300 30,31 . Similarly, substitution of Lys (K) with Gln (Q) and Arg (R), mimicking the constitutively acetylated and deacetylated forms of p53, respectively, severely affected the various activities of that tumor suppressor, including DNA binding and gene transactivation 32,33 . Since non-synonymous GKRP variants have significantly different phenotypic effects 23,34 , it is also possible that the GKRP structural conformation may be changed by acetylation, although we did not examine this in the current study (nor do we propose its occurrence).
We also observed increased GKRP stability by inhibiting ubiquitin-proteasomal degradation during its acetylation state, as confirmed by ubiquitin and protein stability assays. In GKRP-KO mice, the stability and activity of GK is diminished, glucose levels are significantly increased, and glycemia is impaired 15 . Based on those findings, and those of our current study, we theorize that GKRP stability, regulated by acetylation, is critical to controlling homeostatic blood glucose levels.
It has been well established that GK translocates from the nucleus to the cytoplasm under conditions of high glucose 23 . Our finding, however, was that even under high glucose conditions, GK remains in the nucleus when GKRP acetylation is increased by the presence of Class III HDACIs. In addition, GK/.GKRP dimerization is decreased by ectopic expression of the deacetyl mimic of GKRP (K5R), as compared to overexpression of its WT or acetyl mimic (K5Q) forms. Indeed, glycolytic flux by GK significantly decreased in the WT, and acetyl mimic (p ≤ 0.05) GKRP forms, but was not affected by the deacetyl mimic GKRP (K5R). In consideration of GKRP activity being prominently influenced by glucose and different fructose phosphates 35 , we further propose that acetylation represents another mechanism of regulating GK activity.
Based on demonstrations that GKRP phosphorylation by AMPK inhibits both cytoplasmic translocation and activity of GK in hepatocytes 16 , we speculate that GKRP acetylation may represent a similar means of inhibiting GK activity by affecting its subcellular localization. Moreover, study of a possible interrelationship between GKRP phosphorylation and acetylation strongly warrants further investigation, akin to FOXO1 regulation by a reciprocal balance of acetylation and phosphorylation 36 .
In the presence of specific HDAC inhibitors (e.g., nicotinamide, NAM), GKRP acetylation was restored. In our experiments, NAM significantly increased acetylation, while Trichostatin A (TSA, a class I/II HDAC inhibitor) did not, suggesting that GKRP is deacetylated by the NAD + -dependent class III HDAC family. Among these, SIRTs 4, 6, 7, known ADP-ribosyltransferases, and SIRT5, an active desuccinylase, did not deacetylated GKRP. Among the remaining class III HDACs, only SIRT2 deacetylated GKRP ( Fig. 6C and Supplementary Fig. S5B), consistent with SIRT2 involvement in deacetylation of numerous other metabolic enzymes 37 . For example, SIRT2 is believed to regulate the stability of phosphoenolypyruvate carboxykinase (PEPCK), by opposing its acetylation by p300, a critical event in the short-term regulation of gluconeogenesis 38 .
Other physiologic findings supporting our conclusions regarding increased GKRP Lys-5 acetylation were revealed in db/db diabetic/dyslipidemia mice, as hyperglycemia elevated p300-mediated protein acetylation 39 , while p300-overexpressing mice show impaired glucose clearance, with elevated blood glucose 40 . Moreover, these pathologies are similarly observed in both diabetic and GKRP-KO mice 15,41 . In contrast, disruption of the mouse p300 CH1-domain, a conserved protein-binding region, improves glucose tolerance and insulin sensitivity 42 , while anti-p300 adenoviral short hairpin RNAs (shRNAs) decreases hepatic glucose production 43 . Thus, GKRP acetylation likely contributes to GK's stability and subcellular distribution, by a complicated mechanism requiring p300 and SIRT2, which subsequently affects glucose metabolism, an assumption supported by our experiments shown in Fig. 6.
Finally, hormonal effects were implicated in this phenomenon, as we observed increased GKRP acetylation in db/db mice, which have a high glucagon effect, but are also insulin-resistant 44 . During states of energy deficiency, glucagon induces the activity of AMPK, which phosphorylates GKRP, resulting in GK restraint to the nucleus 16 , thus suggesting that Lys 5 acetylation and GKRP phosphorylation could corroboratively control GK activity. Reduced GK activity is well-known to occur in diabetes, for which small molecule GK activators are actively being investigated 45 .
Taken together, this study strongly suggests that GKRP K5 acetylation is critical for its stability, affinity, and subcellular localization, all of which affect the short-term regulation of GK activity in the liver. A better understanding of how this process affects glycolytic flux may improve insight into possible mechanisms of metabolic pathologies such as hyperglycemia, obesity, and diabetes. the control after subtraction of non-glycolytic acidification. (G) After the addition of oligomycin, glycolytic capacity was calculated relative to control. (H) Glycolytic reserve (the difference between basal glycolysis and glycolytic capacity rate). A minimum number of n = 5 with 3-4 replicate wells per group was employed for all experiments. ECAR, extracellular acidification rate. 2-DG, 2-deoxyglucose. Con, Control. SIRT2 H187Y, catalytic mutant of SIRT2. NAM, nicotinamide; TSA, Trichostatin A. Data are expressed as means ± SEMs, *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. NS, not significant.
Scientific RepoRts | 5:17395 | DOI: 10.1038/srep17395 Methods Plasmids and materials. Human GKRP and PCAF cDNAs were amplified by PCR from total HepG2 and HeLa cell RNA and inserted into Myc-or FLAG-tagged pSG5 (Agilent Technologies, La Jolla, CA, USA). A pSG5-based expression plasmid encoding a FLAG-tagged version of the catalytic domain of human p300 was described previously 46 . Other expression plasmids encoding FLAG-tagged hMOF and Tip60 in pcDNA3.1, and pQE30-hGK_liver-type variant 3 (v3) were kindly provided by Drs. XJ Yang (McGill University, Montreal, Canada) and D Schmoll (Sanofi-Aventis Deutschland GmbH, Germany), respectively 47,48 . Another expression plasmid, encoding HA-tagged ubiquitin was a kindly supplied by HG Yoon (Yonsei University College of Medicine, Republic of Korea) 49 . pcDNA3.1-hp300, pRc/RSV-mCBP-HA, and pAdEasy-Flag-hGCN5 plasmids were kind gifts from WC Greene (University of California at San Francisco, USA), RH Goodman (Oregon Health Sciences University, Portland, OR, USA), and P Puigserver (Johns Hopkins University, Baltimore, MD, USA) (Addgene plasmids #23252, #16701 and #14106), respectively 29,50,51 . DNA fragments encoding general control of amino synthesis protein 5-like 2 (GCN5) and p300 were generated by PCR from pAdEasy-Flag-hGCN5 and pcDNA3.1-hp300, respectively, and inserted into FLAG-and HA-tagged pSG5 (Agilent). V5/His-tagged hGK_v3 was derived from pQE30-hGK_liver type variant 3. A FLAG-tagged human SIRT1 (hSIRT1) expression vector was described previously 46  Animal experiments. Ten nine-week-old db/m + (~25 g) and db/db (~35 g) male mice were purchased from Shizuoka Laboratory (Hamamatsu, Japan). Mice were housed under a 12-hr light/12-hr dark cycle and given unrestricted access to a standard chow diet and tap water. After three weeks of housing, all animals were sacrificed after the 12-hr dark cycle. Animals were cared for in accordance with the National Institutes of Health Guidelines for Animal Care. . The lysates were briefly vortexed and then cleared by centrifugation at 21,000 g for 20 min at 4 °C. Protein concentrations were determined using a BCA assay kit (Pierce, Rockford, IL, USA). Equal amounts of protein extracts were subjected to electrophoresis in 6-12% SDS-polyacrylamide gels and then transferred to nitrocellulose membranes (Whatman, Dassel, Germany). The membranes were blocked with milk, incubated with primary and houseradish peroxidase-conjugated secondary antibodies, and detected using SuperSignal ® West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, Waltham, MA, USA). For in vivo experiments, livers were homogenized using a tissueLyser (Qiagen) for 5 min in lysis buffer, at a frequency of 25/s.
Protein stability assay. Myc-tagged GKRP transfected HeLa cells were pretreated with NAM (5 mM) and TSA (1 μ M) 4 h before treatment with 50 μ g/ml cycloheximide (CHX) to block total protein synthesis for various times, at which GKRP protein levels were determined by western blot.
Immunofluorescence. Immunofluorescence assays were performed as previously described 46 . Briefly, HeLa cells were transfected with GK and GKRP expression constructs. The cells were then fixed with 3.7% formaldehyde (w/v), permeabilized with 0.2% Triton X-100 in PBS, and incubated with blocking buffer containing 1% bovine serum albumin (BSA) in 0.1% PBST for 1 h at room temperature. Next, the cells were incubated with antibodies containing 1% BSA in 0.1% PBST for 12 h at 4 °C, washed with 0.1% PBST, followed by a 2 h placement in the dark with Alexa Fluor 488-conjugated donkey anti-rabbit and Alexa Fluor 568-conjugated goat anti-mouse secondary antibodies (Invitrogen). After washing with 0.1% PBST, nuclei were stained with Hoechst 33342 for 2 min, mounted using Dakocytomation Fluorescent Mounting Medium (Dako, Glostrup, Denmark), and confocal micrographs taken on an LSM 700 (Carl Zeiss, Jena, Germany), and analyzed by ZEN software (Carl Zeiss).
Measurement of glycolytic flux using the XF24 analyzer. XF glucose flux assays were performed using an XF24 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA, USA). That is based upon fluorimetric detection of O 2 and H + levels via solid-state probes on a sensor cartridge 53 . HeLa cells, seeded at 10,000 cells/well in XF24 cell plates (Seahorse Bioscience), were transfected with expression vectors for GKRP (WT), GKRP (K5R), GKRP (K5Q), GK, p300, SIRT2 (WT) and/or SIRT2 (H187Y). The following day, the media was changed to DMEM (without serum, glucose or bicarbonate, but with 2 mM glutamine), and incubated 1 h before the assay in a non-CO 2 incubator at 37 °C. Injections of glucose (10 mM final), oligomycin (2.5 μ M final) and 2-deoxyglycose (0.1 M final) were then diluted in DMEM media and loaded into ports A, B, and C sequentially. Reagents were optimized using a Glycolysis Stress Test kit (Seahorse Bioscience), using the XF24 analyzer protocol and algorithm. Each assay was run in one plate with 3-4 replicates, and repeated at least 5 times.

Identification of modified GKRP peptide amino acids by tandem liquid chromatography-tandem mass spectrometry (LC-MS/MS).
Myc-tagged GKRP was immunoprecipitated from HeLa cells using an anti-Myc antibody. After immunoblot, protein bands were excised from stained one-dimensional electrophoresis gels and destained with 25 mM ammonium bicarbonate and 50% acetonitrile. In-gel digestion of dried gel pieces was performed using sequencing grade trypsin (Promega, Madison, WI, USA) in 25 mM ammonium bicarbonate buffer overnight at 37 °C. The tryptic peptides were desalted using a GELoader tip (Eppendorf, Hamburg, Germany) packed with 1.5 μ g of POROS ® 20 R2 resin (PerSpective Biosystems, Ramsey, MN, USA) and applied to a C18 RP-HPLC column (75 m × 150 mm). An Agilent 1100 Series LC system was then used to separate the trypsin-digested peptides, which were eluted with a 0-40% acetonitrile gradient for 60 min. followed by analysis using a Finnigan LCQ Deca (ThermoQuest, San Jose, CA, USA) equipped with a nanoelectrospray ion source. Spray and tube lens voltages were 1.9 kV and 40 V, respectively. The capillary temperature was maintained at 250 °C at 5 V. The individual LC-MS/MS spectra were processed using TurboSEQUEST software (ThermoQuest) and the sequences were searched in NCBI databases using MASCOT software (Matrix Science Ltd., London, UK).

Statistical analysis. Statistics were determined using a two-tailed unpaired Student's t test or
One-Way ANOVA, using Graphpad Prism software (GraphPad, La Jolla, CA, USA). Final results we are calculated as means ± standard errors of the mean (SEMs), unless otherwise indicated. Statistical significance is represented in the figures by *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.