The GCN5-CITED2-PKA signalling module controls hepatic glucose metabolism through a camp-induced substrate switch

Transcription

1 Reeived 6 Apr 216 Aepted 8 Sep 216 Pulished 22 Nov 216 DOI: 1.138/nomms13147 OPEN The GCN5-CITED2-PKA signlling module ontrols hepti gluose metolism through AMP-indued sustrte swith Mshito Ski 1, Tomoko Tujimur-Hykw 1, Tkshi Ygi 1,2, Hiroyuki Yno 1,2, Msru Mitsushim 1, Hiroyuki Unoki-Kuot 3, Ysushi Kurgi 3, Hiroshi Inoue 4, Yoshiki Kido 5,6, Msto Ksug 7 & Mihihiro Mtsumoto 1 Hepti gluoneogenesis during fsting results from gluoneogeni gene tivtion vi the glugonampprotein kinse A (PKA) pthwy, proess whose dysregultion underlies fsting hyperglyemi in dietes. Suh trnsriptionl tivtion requires epigeneti hnges t promoters y mehnisms tht hve remined unler. Here we show tht GCN5 funtions oth s histone etyltrnsferse (HAT) to tivte fsting gluoneogenesis nd s n etyltrnsferse for the trnsriptionl o-tivtor PGC-1 to inhiit gluoneogenesis in the fed stte. During fsting, PKA phosphoryltes GCN5 in mnner dependent on the trnsriptionl oregultor CITED2, therey inresing its etyltrnsferse tivity for histone nd ttenuting tht for PGC-1. This sustrte swith onomitntly promotes oth epigeneti hnges ssoited with trnsriptionl tivtion nd PGC-1medited otivtion, therey triggering gluoneogenesis. The GCN5-CITED2-PKA signlling module nd ssoited GCN5 sustrte swith thus serve s key driver of gluoneogenesis. Disruption of this module meliortes hyperglyemi in oese dieti nimls, offering potentil therpeuti strtegy for suh onditions. 1 Deprtment of Moleulr Metoli Regultion, Dietes Reserh Center, Reserh Institute, Ntionl Center for Glol Helth nd Mediine, Toym, Shinjuku-ku, Tokyo , Jpn. 2 Deprtment of Bioregultion, Nippon Medil Shool, Kosugi-mhi, Nkhr-ku, Kwski , Jpn. 3 Deprtment of Dieti Complitions, Dietes Reserh Center, Reserh Institute, Ntionl Center for Glol Helth nd Mediine, Toym, Shinjuku-ku, Tokyo , Jpn. 4 Metolism nd Nutrition Reserh Unit, Innovtive Integrted Bio-reserh Core, Institute for Frontier Siene Inititive, Knzw University, 13-1 Tkr-mhi, Knzw , Jpn. 5 Division of Medil Chemistry, Deprtment of Metolism nd Disese, Koe University Grdute Shool of Helth Sienes, Tomogok, Sum-ku, Koe , Jpn. 6 Division of Dietes nd Endorinology, Deprtment of Internl Mediine, Koe University Grdute Shool of Mediine, Kusunoki-ho, Chuo-ku, Koe 65-17, Jpn. 7 Ntionl Center for Glol Helth nd Mediine, Toym, Shinjuku-ku, Tokyo , Jpn. Correspondene nd requests for mterils should e ddressed to M.M. (emil: mmtsumoto@ri.ngm.go.jp). NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

2 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 Hepti gluoneogenesis is indued y pnreti glugon during fsting to mintin gluose homeostsis 1, proess whose dysregultion in dietes results in fsting hyperglyemi 2,3. The glugonampprotein kinse A (PKA) pthwy tivtes hepti gluoneogenesis y induing expression of the gluoneogeni genes G6p (enoding the tlyti suunit of gluose-6-phosphtse) nd Pk1 (phosphoenolpyruvte roxykinse) 4. Suh indution is thought to e medited through orhestrtion of hormone-dependent epigeneti hnges 5 nd ssemly of trnsriptionl mhinery 3,613 t the gene promoters. A key omponent of this mhinery is the CREB (AMP response element-inding protein)crtc2 (CREB-regulted trnsriptionl o-tivtor 2) omplex, whose ssemly is triggered y PKA-dependent phosphoryltion of CREB nd the dephosphoryltion of CRTC2 nd whih promotes reruitment of the histone etyltrnsferse (HAT) CBP (CREB-inding protein, lso known s Kt3) nd therey tivtes peroxisome prolifertortivted reeptor g o-tivtor1 (PGC-1) gene trnsription 6. PGC-1 then funtions with the trnsription ftors forkhed ox O1 (FoxO1) 8 nd heptoyte nuler ftor4 (HNF-4) 7,9 to medite synergisti tivtion of the gluoneogeni progrm; however, how glugon signlling triggers epigeneti hnges for effiient trnsription of gluoneogeni genes hs een unler 5. Generl ontrol non-repressed protein 5 (GCN5, lso known s Kt2) funtions not only s HAT tht links histone etyltion nd trnsriptionl tivtion 14 ut it lso possesses etyltrnsferse tivity for non-histone sustrtes 15,16. GCN5 thus etyltes nd intivtes PGC-1 nd therey inhiits PGC-1dependent gluoneogenesis This funtion of GCN5 is inhiited during fsting y CBP- nd p3-interting trnstivtor with glutmi id nd sprti idrih COOH-terminl domin 2 (CITED2), fsting-induile trnsriptionl oregultor tht inds to GCN5 nd disrupts the GCN5PGC-1 intertion, resulting in deetyltion nd tivtion of PGC-1 (ref. 2); however, the preise role of GCN5 ssoited with CITED2 in the regultion of hepti gluoneogenesis y the glugon-amp-pka xis hs remined unknown. We now show tht GCN5 serves dul funtion in the feedingto-fsting trnsition: s HAT to tivte gluoneogenesis in the fsted stte nd s PGC-1 etyltrnsferse to inhiit it in the fed stte. This dul funtion of GCN5 is reiprolly regulted y glugon-dependent sustrte swith t the feeding-to-fsting trnsition. The fsting-indued GCN5-CITED2-PKA omplex serves s speifi module for PKA-dependent phosphoryltion of GCN5 t Ser 275, whih drives the sustrte swith from PGC-1 to histone H3. This swith integrtes epigeneti hnges nd the o-tivtor tivity of PGC-1, leding to full indution of the gluoneogeni progrm. Disruption of the GCN5-CITED2- PKA signlling module meliortes hyperglyemi in oese dieti nimls nd is therefore potentil strtegy for the tretment of suh onditions. Results Hepti GCN5 expression is regulted y AMP-PKA signlling. We first investigted hepti expression of GCN5 s well s of CBP, its prlog p3 (lso known s Kt3), nd the GCN5 prlog p3/cbp-ssoited ftor (PCAF, lso known s Kt2) in mie. Among these HATs, we found tht the expression of GCN5 ws seletively inresed in the liver of two mouse models of oesity-ssoited type 2 dietes norml mie fed high-ft diet (HFD) (Fig. 1) nd d/d mie (Fig. 1,) ompred with ontrol mie. The mount of GCN5 in primry ultured mouse heptoytes ws lso inresed y tretment with ell-permele nlogue of AMP (pcpt-amp) in PKA-dependent mnner (Fig. 1d), suggesting tht glugon-amp-pka signlling inreses the hepti of GCN5 s well s tht of CITED2 (ref. 2). These findings prompted us to exmine whether GCN5 regultes gluoneogenesis through histone etyltion in ollortion with CITED2. GCN5 depletion ttenutes gluoneogenesis in vivo nd in vitro. In len mie, depletion of hepti GCN5 with short hirpin RNA (shrna) redued the level of gluoneogeni gene expression pprent fter food deprivtion for 24 h s well s the lood gluose onentrtion either fter fsting for 6 h (Fig. 2 nd Supplementry Fig. 1) or fter dministrtion of pyruvte (Fig. 2) despite the ssoited mrked ttenution of PGC-1 etyltion (Fig. 2). Knokdown of hepti GCN5 in d/d mie lso redued fsting gluoneogeni gene expression nd glyemi (Supplementry Fig. 1). In primry mouse heptoytes, GCN5 knokdown ttenuted gluoneogeni gene expression s well s gluose prodution indued y pcpt-amp (Fig. 2d nd Supplementry Fig. 1) despite the oserved downregultion of PGC-1 etyltion (Fig. 2e). These results thus suggested tht GCN5 ontriutes in mnner independent of its PGC-1 etyltrnsferse funtion to gluoneogeni gene indution vi the glugon-amp pthwy. In heptoytes, fored expression of CITED2 enhnes gluoneogeni gene expression indued y AMP (Fig. 2f nd Supplementry Fig. 1d) or y PGC-1 overexpression in the sene of AMP 2 (Fig. 2g nd Supplementry Fig. 1e). The PGC-1 etyltrnsferse tivity of GCN5 is suppressed y intertion with CITED2 under these onditions 2. We therefore exmined whether GCN5 tht interts with CITED2 might funtion other thn s PGC-1 etyltrnsferse in the regultion of gluoneogenesis y dopting loss- nd gin-of-funtion pprohes in this setting. Loss of GCN5 ttenuted the CITED2-dependent enhnement of gluoneogeni gene expression indued y AMP (Fig. 2f nd Supplementry Fig. 1d) or y PGC-1 overexpression (Fig. 2g nd Supplementry Fig. 1e) despite the downregultion of PGC-1 etyltion indued y GCN5 knokdown (Fig. 2e). GCN5 promotes gluoneogenesis in onert with CITED2. Overexpression of wild-type (WT) GCN5 suppressed gluoneogeni gene indution y AMP in primry heptoytes (Fig. 3), onsistent with the previously oserved suppression of PGC-1indued gluoneogeni gene expression y GCN5 overexpression in Fo heptom ells 17. This effet ws lso oserved with mutnt (DAT) of GCN5 tht lks etyltrnsferse tivity 21 towrd oth histone H3 nd PGC-1 (Fig. 3 nd Supplementry Fig. 2,), however, inditing tht GCN5 suppresses gluoneogeni gene expression independently of its enzymti tivity in this setting. In ontrst, oexpression of GCN5(WT), ut not tht of GCN5(DAT), with CITED2 enhned AMP-indued gluoneogeni gene expression (Fig. 3 nd Supplementry Fig. 2). We lso exmined the effets of overexpression of GCN5, CITED2, nd oth proteins in the liver of C57BL/6J mie on gluoneogenesis. As previously shown 2, mie overexpressing CITED2 in the liver mnifested inresed fsting glyemi (Fig. 3) nd hepti expression of G6p nd Pk1 (Fig. 3d nd Supplementry Fig. 2d) ompred with ontrol mie. Hepti overexpression of GCN5 lone redued oth fsting glyemi (Fig. 3) nd hepti expression of gluoneogeni genes (Fig. 3d nd Supplementry Fig. 2d), onsistent with our in vitro results (Fig. 3) s well s previous oservtions 17. Agin onsistent with our in vitro findings (Fig. 3), oexpression of GCN5 with 2 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

3 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE Reltive Gn5 mrna NC HFD NC HFD IB: Histone H1 Reltive Gn5 mrna d/m d/d IB: Histone H1 d/m d/d d/m d/d IB: PCAF d IP: GCN5 IB: CBP IB: PCAF IB: p3 3 h 6 h 6 h Reltive GCN5/H1 rtio d/m d/d Reltive PCAF/H1 rtio 1..5 Reltive CBP/H1 rtio 1..5 Reltive p3/h1 rtio 1..5 H89: Figure 1 Hepti expression of GCN5 is upregulted in oese dieti mie vi AMP-PKA signlling. (,) qrt-pcr nlysis of Gn5 mrna nd immunolot (IB) nlysis of nuler GCN5 in the liver of C57BL/6J mie mintined on norml how (NC) or HFD ()orofd/d or d/m (ontrol) mie () fter deprivtion of food for 16 h. RTPCR dt re mens±s.e.m. (n ¼ 7() or6()). Histone H1 ws exmined s loding ontrol for immunolot nlysis. () Immunolot nlysis of GCN5, PCAF, CBP nd p3 in the liver of d/d or d/m mie deprived of food for 16 h. Quntittive dt re mens±s.e.m. (n ¼ 3). (d) Primry mouse heptoytes were inuted in the sene or presene of 1 mm pcpt-amp or the PKA inhiitor H89 (2 mm) for the indited times. Cell lystes were then sujeted to immunolot nlysis of PCAF or to IP followed y immunolot nlysis with ntiodies to GCN5. Dt re representtive of t lest three independent experiments. Sttistil nlysis ws performed with the unpired Student s t-test (). Po.5 versus NC () or d/m mie (,). RTPCR, PCR with reverse trnsription. CITED2 in the liver of C57BL/6J mie inresed oth fsting glyemi (Fig. 3) nd hepti gluoneogeni gene expression (Fig. 3d nd Supplementry Fig. 2d) ompred with those pprent with CITED2 expression lone. To evlute the effets of suh ltered gluoneogeni gene expression, we performed pyruvte hllenge test. Gluose levels fter pyruvte dministrtion were signifintly higher in mie with hepti CITED2 overexpression ut were lower in those with hepti GCN5 overexpression ompred with ontrol mie (Fig. 3e). Mie with hepti oexpression of CITED2 nd GCN5 showed lood gluose onentrtions fter pyruvte dministrtion tht were higher thn those in mie with CITED2 overexpression lone (Fig. 3e). Together, these in vitro nd in vivo results suggested tht GCN5 might funtion s AMP-responsive, CITED2-dependent HAT in the regultion of gluoneogenesis. GCN5 is ruil HAT for initition of gluoneogenesis. To test this hypothesis, we exmined the moleulr signtures of gluoneogeni gene promoters y hromtin immunopreipittion (ChIP) nd quntittive PCR nlysis in primry heptoytes depleted of GCN5, CITED2 or PGC-1. The oupny of suh promoters with GCN5 nd CBP ws inresed fter exposure of ontrol heptoytes to pcpt-amp for 16 h (Fig. 4, nd Supplementry Fig. 3). The mounts of histone H3 etylted t Lys 9 (H3K9) or Lys 22 (H3K27) whih re generted predominntly y GCN5 nd CBP, respetively 22 s well s tht of trimethylted H3K4 (H3K4me3) (ref. 23), ll three of whih mrks re ssoited with tive gene trnsription 24, were lso inresed t these promoters in ells treted with pcpt-amp (Fig. 4 nd Supplementry Fig. 3). Consistent with the ssoited downregultion of gluoneogeni gene trnsription 2 (Fig. 2d), these AMP-indued inreses in the mounts of H3K9, H3K27 nd H3K4me3 were lunted y depletion of either GCN5 (Fig. 4 nd Supplementry Fig. 3,e), CITED2 (Fig. 4 nd Supplementry Fig. 3,e) or PGC-1 (Fig. 4 nd Supplementry Fig. 3,e). The AMP-indued reruitment of GCN5 nd CBP ws lso ttenuted y NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

4 NATURE COMMUNICATIONS DOI: 1.138/nomms GCN5 KD 1..5 Gluose (mg dl 1 ) Gluose (mg dl 1 ) 15 GCN5 KD Fed Fsted Fsted 6 h 24 h GCN5 KD Time (min) GCN5 shrna: IP: PGC-1α IB: PGC-1α IP: PGC-1α IB: ALys d GCN5 shrna: 3,5 3, 2,5 2, 1,5 1, 5 Gluose prodution (AU) e FLAG- PGC-1α: GCN5 shrna: IB: PGC-1α IB: ALys IP: GCN5 f CITED2: GCN5 shrna: CITED2: GCN5 shrna: PGC-1α: g Figure 2 GCN5 depletion suppresses hepti gluoneogenesis in vivo nd in vitro. (,) Effets of shrna-medited knokdown (KD) of GCN5 in the liver of C57BL/6J mie on hepti gluoneogeni gene expression under the fsted (24 h) ondition () or on plsm glyemi either under fsted (6 or 24 h) or fed onditions () or fter pyruvte dministrtion (). () IP nd immunolot nlysis of etylted (A) PGC-1 in the liver of C57BL/6J mie injeted with n denovirus for GCN5 shrna nd deprived of food for 24 h. (d) Effets of shrna-medited depletion of GCN5 on gluoneogeni gene expression nd gluose prodution in primry mouse heptoytes exposed (or not) to pcpt-amp for 16 h. (e) IP nd immunolot nlysis of etylted PGC-1 in primry heptoytes expressing FLAGPGC-1 with or without GCN5 depletion nd inuted in the sene or presene of pcpt-amp for 6 h. (f) Effets of GCN5 depletion on CITED2-dependent enhnement of gluoneogeni gene expression indued y pcpt-amp (1 mm, 6 h) in primry heptoytes. (g) Effets of GCN5 knokdown on PGC-1-indued gluoneogeni gene expression with or without CITED2 overexpression in primry heptoytes. All quntittive dt re mens±s.e.m. (n ¼ 7(,) or 3(d,f,g)). Sttistil nlysis ws performed with the unpired Student s t-test () or ANOVA followed y Bonferroni s post ho test (,d,f,g). Po.5, Po.1 ompred with ontrol or s indited. Dt in,e re representtive of t lest three independent experiments. Adenovirl vetors enoding GCN5 shrna, FLAGPGC-1 or CITED2 were used for these experiments. ANOVA, nlysis of vrine. depletion of either CITED2 (Fig. 4, nd Supplementry Fig. 3,e) or PGC-1 (Fig. 4 nd Supplementry Fig. 3,e). In ddition, the AMP-indued reruitment of CBP to gluoneogeni gene promoters ws impired y GCN5 depletion (Fig. 4 nd Supplementry Fig. 3,e), inditing tht GCN5 is required for the reruitment of CBP. On the other hnd, wheres the inresed oupny of gluoneogeni gene promoters with PGC-1 or FoxO1 oserved in immortlized murine heptoytes treted with forskolin nd dexmethsone 8 or in primry mouse heptoytes exposed to glugon for 1 h (ref. 11), respetively, ws not pprent in primry ultured mouse heptoytes fter exposure to pcpt-amp for 6 h, promoter oupny with PGC-1, FoxO1, or HNF-4 ws lso dysregulted y depletion of GCN5, CITED2 or PGC-1 lthough we did not identify unifying defet in the inding of these regultors tht might ount for the impired indution of gluoneogeni genes (Fig. 4d nd Supplementry Fig. 3,e). To lrify whether epigeneti hnges indued y GCN5 depletion re restrited to gluoneogeni gene promoters or re more widespred, we exmined the effets of GCN5 knokdown on the epigeneti sttus of the promoters of two nongluoneogeni genes: Gpdh (enoding glyerldehyde-3- phosphte dehydrogense) nd Hmgr (hydroxymethylglutrteoenzyme A redutse). Depletion of GCN5 ffeted neither etyltion of H3K9 or H3K27 or trimethyltion of H3K4 t these promoters nor the orresponding mrna (Supplementry Fig. 3d), inditing tht the epigeneti hnges ssoited with GCN5 knokdown our seletively t gluoneogeni gene promoters. We lso investigted whether these epigeneti hnges indued y GCN5 depletion re oserved in mouse liver. Consistent with our in vitro results, the fsting levels of H3K9 nd H3K27 etyltion (ut not tht of H3K4 trimethyltion) t the gluoneogeni gene promoters were redued y GCN5 depletion (Supplementry Fig. 3f). The differene in the effet of GCN5 depletion on H3K4 trimethyltion sttus etween the in vivo nd in vitro results my reflet non-ell-utonomous response to ompenste for the impirment of gluoneogeni gene expression indued y 4 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

5 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE GCN5 (WT): GCN5 (ΔAT): 5, 4, 3, 2, 1, Reltive Gn5 mrna , 1,5 1, 5 5 CITED2: GCN5 (WT): GCN5 (ΔAT): Gluose (mg dl 1 ) GCN5 CITED2 CITED2GCN5 d GCN5 CITED2 CITED2GCN e Gluose (mg dl 1 ) GCN5 CITED2 CITED2GCN Time (min) Figure 3 GCN5 promotes gluoneogenesis in n etyltrnsferse- nd CITED2-dependent mnner. () Quntittive RT-PCR nlysis of Gn5 nd gluoneogeni gene expression in primry heptoytes infeted with denoviruses for WT or DAT mutnt forms of GCN5 nd exposed to pcpt-amp for 6h. () Effets of fored expression of WT or DAT forms of GCN5 together with CITED2 on pcpt-amp-indued gluoneogeni gene expression in primry heptoytes. (e) Effets of GCN5 overexpression with or without tht of CITED2 in the liver of C57BL/6J mie on glyemi under the fsted (6 h) ondition () or fter pyruvte dministrtion (e) s well s on hepti gluoneogeni gene expression under the fsted (24 h) ondition (d). All dt re mens±s.e.m. (n ¼ 3(,), 1 () or 8(d,e)). Sttistil nlysis ws performed with ANOVA followed y Bonferroni s post ho test. Po.5, Po.1 ompred with ontrol or s indited; wpo.5, wwpo.1 versus CITED2. Adenovirl vetors enoding GCN5(WT), GCN5(DAT) or CITED2 were used for these experiments. ANOVA, nlysis of vrine; RTPCR, PCR with reverse trnsription. hepti depletion of GCN5. Colletively, these dt indited tht GCN5 is ruil HAT for the initition of AMP-indued gluoneogeni gene trnsription in onert with CITED2. AMP nd CITED2 promote GCN5 sustrte swith. Given tht glugon-indued GCN5-CITED2 intertion is medited y the AMP-PKA pthwy 2, we tested whether GCN5 tivity is regulted y AMP, CITED2 or oth in heptoytes. The HAT tivity of GCN5 mesured in vitro with histone H3 s sustrte ws inresed y tretment of AML12 ells with pcpt-amp, nd this effet ws enhned y CITED2 overexpression in n dditive mnner (Fig. 5). These dt thus indited tht the HAT tivity of GCN5 ws inresed in the presene of AMP nd CITED2. We next exmined the effet of CITED2 on the lne etween the HAT nd PGC-1 etyltrnsferse tivities of GCN5 mesured in vitro with histone H3 nd n NH 2 -terminl frgment of PGC-1 s sustrtes. In this setting, GCN5 tivity ws reiprolly regulted y CITED2: HAT tivity ws inresed, wheres PGC-1 etyltrnsferse tivity nd utoetyltion tivity were deresed (Fig. 5). These results indited tht CITED2 inds to GCN5 nd promotes sustrte swith from PGC-1 to histone H3 s mesured in vitro. To exmine whether this swith lso ours in heptoytes, we investigted the intertion of GCN5 with PGC-1 or histone H3 in the nuleus with n in situ proximity ligtion ssy (PLA) 25. Exposure of primry heptoytes to pcpt-amp disrupted GCN5PGC-1 intertion nd promoted GCN5histone H3 intertion (Fig. 5). Together with our results showing tht GCN5 reruitment to nd H3K9 mrking of gluoneogeni gene promoters re dependent on AMP nd CITED2 (Fig. 4, nd Supplementry Fig. 3), these findings supported the notion tht AMP- nd CITED2-dependent sustrte swith of GCN5 integrtes epigeneti hnges nd o-tivtor tivity in gluoneogeni gene indution. PKA phosphoryltes GCN5 within module hroring CITED2. Given tht reominnt CITED2 interted with, ut did not enhne the HAT tivity of, GCN5 immunopreipitted from NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

6 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 CITED2 KD GCN5 ChIP G6p promoter GCN5 ChIP Pk1 promoter Time (h) Time (h) GCN5 KD CITED2 KD CBP ChIP G6p promoter Time (h) Time (h) H3K9 ChIP H3K27 ChIP H3K4me3 ChIP G6p promoter G6p promoter G6p promoter Time (h) Time (h) AMP () AMP () GCN5 ChIP G6p promoter PGC-1α KD CBP ChIP G6p promoter.1 PGC-1α KD H3K9 ChIP G6p promoter PGC-1α KD H3K27 ChIP G6p promoter PGC-1α KD H3K4me3 ChIP G6p promoter PGC-1α KD d AMP () AMP () FLAG-PGC-1α ChIP HNF-4α ChIP FLAG-FoxO1 ChIP G6p promoter G6p promoter G6p promoter GCN5 CITED2 GCN5 CITED2 PGC-1α KD KD KD KD KD.1 GCN5 KD CITED2 KD PGC-1α KD Figure 4 GCN5 reruitment to gluoneogeni gene promoters is regulted y AMP in CITED2- nd PGC-1-dependent mnner in primry heptoytes. () ChIP-qPCR nlysis of the oupny of G6p nd Pk1 promoters with GCN5 in ells depleted of CITED2 nd exposed to pcpt-amp for the indited times. () ChIP-qPCR nlysis of the oupny of the G6p promoter with CBP or epigenomi mrks in ells depleted of GCN5 or CITED2 nd exposed to pcpt-amp for the indited times. () ChIP-qPCR nlysis of the oupny of the G6p promoter with GCN5, CBP or epigenomi modifitions in ells depleted of PGC-1 nd exposed to pcpt-amp for 6 h. (d) ChIP-qPCR nlysis of the oupny of the G6p promoter with FLAG-tgged PGC-1, HNF-4 or FLAG-FoxO1 in ells depleted of GCN5, CITED2 or PGC-1 nd exposed to pcpt-amp for 6 h. All dt re mens±s.e.m. (n ¼ 3). Po.5, Po.1 versus ontrol or s indited (ANOVA with Bonferroni s post ho test). Adenovirl vetors enoding GCN5, CITED2 or PGC-1 shrnas, FLAGPGC-1 or FLAG-FoxO1 were used for these experiments. ANOVA, nlysis of vrine; qpcr, quntittive PCR. 6 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

7 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE My-GCN5: FLAG-CITED2: IP: My IB: My IB: H3K9 IB: H3 IB: ALys IB: His 6 (PGC-1α(N-term)) IB: ALys (A-PGC-1α(N-term)) IB: H3K9 IB: H3 AMP() PLA AMP() FLAG-GCN5: His 6 -PGC-1α (N-term): HA-CITED2: GCN5PGC-1α intertion AMP() AMP() GCN5Histone H3 intertion Figure 5 GCN5 swithes sustrte in AMP- nd CITED2-dependent mnner. () Immunolot nlysis of the effets of FLAG-CITED2 expression or pcpt-amp tretment (for 1 h) in AML12 ells on the HAT tivity of immunopreipitted My epitope-tgged GCN5 ssyed in vitro with histone H3 s sustrte. () Effets of hemgglutinin epitope (HA)-tgged CITED2 expression in AML12 ells on the etyltrnsferse tivity of immunopreipitted FLAG-GCN5 ssyed in vitro with histone H3 nd His 6 -tgged NH 2 -terminl frgment of PGC-1 s sustrtes. () Intertion of GCN5 with PGC-1 or histone H3 ws ssessed y PLA in primry heptoytes expressing either My-GCN5 with FLAGPGC-1 (top) or FLAG-GCN5 lone (ottom) nd exposed (or not) to pcpt-amp for 1 h. PLA signls (red dots) represent proximity (o4 nm) of GCN5 nd either PGC-1 (top) or histone H3 (ottom). Nulei re stined lue with 4,6-dimidino-2-phenylindole. Sle rs, 1 mm. All dt re representtive of t lest three independent experiments. Adenovirl vetors were used for these experiments. AML12 ells overexpressing GCN5 lone (Supplementry Fig. 4,), we onluded tht n dditionl ftor tht interts with CITED2 might e neessry for the sustrte swith. We found tht, in the sene of pcpt-amp, inhiition of PKA y H89 signifintly suppressed gluoneogeni gene indution y overexpression of PGC-1 lone or in omintion with CITED2 in primry heptoytes (Fig. 6 nd Supplementry Fig. 4), inditive of the requirement for PKA tivity in this setting. PKA tivted y glugon-amp signlling medites indution of gluoneogenesis nd remodelling of the tin ytoskeleton through phosphoryltion of CREB 26,27 nd the inositol 1,4,5- trisphosphte reeptor (IP3R) 12 nd through tht of vsodiltorstimulted phosphoprotein (VASP) 28, respetively. We then tested whether GCN5 tivity is regulted y PKA-medited phosphoryltion. With the use of ntiodies speifi for phosphorylted PKA sustrtes, we found tht GCN5 phosphoryltion ws indued y pcpt-amp in AML12 ells (Fig. 6) s well s y glugon in mouse liver (Supplementry Fig. 4d). This effet of pcpt-amp ws enhned y CITED2 overexpression (Fig. 6) nd suppressed y shrna-medited CITED2 knokdown (Fig. 6). In ontrst, CITED2 knokdown did not ffet either the phosphoryltion of other PKA sustrtes suh s CREB, IP3R, nd VASP or the dephosphoryltion of CRTC2 (s ssessed on the sis of the ssoited nd moility shift) indued y pcpt-amp (Fig. 6). These results suggested tht PKA seletively phosphoryltes GCN5 in CITED2- dependent mnner, prompting us to exmine whether PKA interts with the GCN5-CITED2 omplex. The PKA holoenzyme onsists of two tlyti (PKAC) nd two regultory (PKAR) suunits, the ltter of whih serve s reeptors for AMP. The inding of AMP to the regultory suunits triggers dissoition nd tivtion of the tlyti suunits 29. Co-IP nlysis of AML12 ells reveled CITED2-PKA intertion (Supplementry Fig. 4e) s well s CITED2-dependent GCN5-PKA intertion (Fig. 6d,e) t the level of endogenous or epitope-tgged proteins, inditive of formtion of GCN5-CITED2-PKA omplex. We lso tested whether PKAC lters the HAT tivity of GCN5. Knokdown of PKAC lmost ompletely ttenuted the CITED2-indued inrese in HAT tivity (Fig. 6f), wheres fored expression of PKAC lone upregulted this tivity (Fig. 6g). Together, these results indited tht GCN5, CITED2 nd PKA form omplex tht funtions s module for PKA-dependent upregultion of the HAT tivity of GCN5. Co-IP nlysis lso reveled tht the intertion of GCN5 with PKA or CITED2 (Supplementry Fig. 4f) s well s tht of CITED2 with PKA (Supplementry Fig. 4g) were disrupted y exposure of AML12 ells to pcpt-amp for 3 min. PKA-dependent phosphoryltion of GCN5 ws lso oserved onomitntly with these effets (Supplementry Fig. 4h), nd AMP-indued disruption of the CITED2-GCN5 intertion ws loked y PKA inhiition (Supplementry Fig. 4i). Together, these results suggested tht the GCN5-CITED2-PKA signlling module disssemles in response to phosphoryltion of GCN5 y AMP-tivted PKA. GCN5 phosphoryltion t Ser 275 drives the sustrte swith. We next explored further the role of PKA-dependent phosphoryltion in the regultion of GCN5 tion within the module. In silio nlysis identified five serine or threonine NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

8 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 Reltive G6p mrna PGC-1α: H89: CITED2: FLAG-GCN5: Reltive Pk1 mrna IB: Phospho-PKA sustrte IB: Phospho-CREB IB: CREB IB: Phospho-IP3R IB: IP3R IB: Phospho-VASP IB: VASP IB: CRTC2 FLAG-GCN5: My-PKAC: HA-CITED2: FLAG-GCN5: HA-CITED2: 1 min 3 min 1 min 3 min d e CITED2 shrna: AMP (3 min): IB: β-tin f IB: PKAC IB: HA IB: H3K9 IB: H3 IB: PKAC IB: My IB: HA IB: PKAC g FLAG-GCN5: IB: Phospho-PKA sustrte FLAG-GCN5: CITED2 shrna: My-PKAC: IB: H3K9 IB: H3 IB: PKAC IB: PKAC FLAG-GCN5: HA-CITED2: PKAC sirna: Figure 6 GCN5 is phosphorylted y PKA within GCN5-CITED2-PKA signlling module. () Effets of PKA inhiition with H89 (2 mm, 6 h) on gluoneogeni gene expression indued y overexpression of PGC-1 with or without CITED2 in primry heptoytes. Dt re mens±s.e.m. (n ¼ 3). Po.1 (ANOVA with Bonferroni s post ho test). () Immunolot nlysis of the effets of HA-CITED2 expression or pcpt-amp tretment (1 or 3 min) on phosphoryltion of FLAG-GCN5 in AML12 ells s ssessed with ntiodies to phosphorylted PKA sustrtes. () Effets of CITED2 depletion on pcptamp-indued phosphoryltion of FLAG-GCN5 nd other PKA sustrtes s well s on the dephosphoryltion of CRTC2 in primry heptoytes. (d,e) IPnd immunolot nlysis of the intertion of FLAG-GCN5 with My-PKAC nd HA-CITED2 (d) s well s of the effet of CITED2 knokdown on the intertion of FLAG-GCN5 with PKAC (e) in AML12 ells. (f) Effet of sirna-medited PKAC depletion in AML12 ells on sl nd CITED2-indued HAT tivity of FLAG- GCN5 s ssessed y in vitro ssy. (g) Effet of PKAC overexpression in AML12 ells on HATtivity of FLAG-GCN5 mesured in vitro. A My-PKAC plsmid nd PKAC sirna were introdued into ells y trnsfetion, wheres denovirl vetors were used to introdue other exogenous proteins or shrnas in these experiments. Dt re representtive of t lest three independent experiments. ANOVA, nlysis of vrine; sirna; smll interfering RNA. residues onserved etween rodents nd humn s puttive PKA phosphoryltion sites in GCN5 (Supplementry Fig. 5). We therefore generted series of puttive phosphoryltion-defetive mutnts, in whih these residues re repled with lnine. Similr to PKA inhiition with H89, the S275A muttion olished sl nd AMP-indued phosphoryltion of GCN5 s ssessed with the ntiodies to phosphorylted PKA sustrtes (Fig. 7). The phosphoryltion signl intensity for GCN5(S275A) ws similr to tht for GCN5(WT) in ells exposed to oth AMP nd H89 nd my therefore e ttriutle to phosphoryltion t PKA-independent site detetle y these ntiodies. Immunolot nlysis of immunopreipittes prepred from primry heptoytes with ntiodies to the Ser 275 -phosphorylted form of GCN5 lso onfirmed tht pcpt-amp indued the phosphoryltion of GCN5 in PKA- nd CITED2-dependent mnner (Supplementry Fig. 5). In ddition, reominnt PKAC phosphorylted GCN5(WT), ut not GCN5(S275A), in vitro, suggesting tht PKA diretly phosphoryltes Ser 275 of GCN5 (Supplementry Fig. 5). Colletively, these results indited tht Ser 275 is the prinipl PKA phosphoryltion site in GCN5. We exmined whether phosphoryltion of GCN5 t Ser 275 regultes etyltrnsferse tivity with the use of the S275A mutnt nd the phosphoryltion-mimiking mutnt S275D. GCN5(S275A) mnifested redued HAT tivity (Fig. 7) nd similr PGC-1 etyltrnsferse tivity (Supplementry Fig. 5d) ompred with the WT protein, ut it ws resistnt to CITED2-indued suppression of PGC-1 etyltrnsferse tivity in AML12 ells (Fig. 7). In ontrst, GCN5(S275D) showed inresed HAT tivity, redued PGC-1 etyltrnsferse tivity, nd n ttenuted intertion with PGC-1 (Fig. 7d,e). These results suggested tht PKAdependent phosphoryltion of GCN5 t Ser 275 drives its sustrte swith. 8 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

9 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE IP: Phospho-PKA sustrte IB: FLAG IB: FLAG IB: H3K9 FLAG-GCN5(WT): FLAG-GCN5(S275A): H89: FLAG-GCN5(WT): FLAG-GCN5(S275A): IB: H3 FLAG-PGC-1α: GCN5(WT): IB: PGC-1α IB: ALys IB: PGC-1α d FLAG-GCN5(WT): FLAG-GCN5(S275D): IB: H3K9 IB: H3 FLAG-PGC-1α: GCN5(WT): GCN5(S275D): IB: PGC-1α IB: ALys GCN5(S275A): HA-CITED2: e FLAG-GCN5(WT): FLAG-GCN5(S275D): V5-PGC-1α: IB: FLAG IB: V5 IB: FLAG IB: V5 f FLAG-GCN5(WT): FLAG-GCN5(S275A): IB: FLAG IB: HA IB: FLAG IB: HA FLAG-GCN5(S275D): HA-CITED2: g FLAG-GCN5(WT): FLAG-GCN5(S275A): HA-CITED2: AMP (3 min): Figure 7 GCN5 phosphoryltion t Ser 275 drives its sustrte swith. () AML12 ells expressing FLAG-tgged GCN5(WT) or GCN5(S275A) were exposed to pcpt-amp with or without H89 for 3 min nd then sujeted to IP with ntiodies to phosphorylted PKA sustrtes followed y immunolot nlysis with ntiodies to FLAG. () Effet of the S275A muttion of FLAG-GCN5 on in vitro etyltion of histone H3 in AML12 ells treted with pcpt-amp (1 h). () Aetyltion of FLAGPGC-1 in AML12 ells expressing GCN5(WT) or GCN5(S275A) with or without HA-CITED2. (d) Effets of the S275D muttion of GCN5 on histone H3 nd PGC-1 etyltrnsferse tivities in AML12 ells. (e) Effet of the S275D muttion of FLAG-GCN5 on intertion with V5-tgged PGC-1 in AML12 ells. (f) Effets of the S275A nd S275D muttions of FLAG-GCN5 on intertion with HA-CITED2 in AML12 ells. (g) The S275A muttion of GCN5 loks the pcpt-amp-indued dissoition of HA-CITED2 from FLAG-GCN5 in AML12 ells. All dt re representtive of t lest three independent experiments. Adenovirl vetors were used for these experiments. We lso tested whether GCN5 phosphoryltion t Ser 275 ffets its intertion with CITED2. The mount of HA-tgged CITED2 tht o-immunopreipitted from AML12 ells with FLAG-tgged GCN5 ws redued for the S275D mutnt ompred with the WT or S275A forms (Fig. 7f). Exposure of ells to pcpt-amp lso redued the mount of HA-CITED2 tht o-immunopreipitted with GCN5(WT) ut not of tht o-immunopreipitting with GCN5(S275A) (Fig. 7g). Phosphoryltion of GCN5 t Ser 275 thus ppers to promote disssemly of the GCN5-CITED2-PKA signlling omplex, supporting role for this module in phosphoryltion of GCN5 t this site y PKA. NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

10 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 Phosphoryltion of GCN5 t Ser 275 promotes gluoneogenesis. We next investigted the effets of these GCN5 mutnts on gluoneogenesis in vitro. In primry heptoytes overexpressing CITED2, pcpt-amp-indued gluoneogeni gene expression ws enhned to lesser extent y GCN5(S275A) thn y GCN5(WT) (Fig. 8 nd Supplementry Fig. 6). On the other hnd, expression of the S275D mutnt without CITED2 overexpression enhned AMP-indued gluoneogeni gene expression (Fig. 8 nd Supplementry Fig. 6) in ontrst to the effet of GCN5(WT) (Fig. 3) resulting in enhned gluose prodution (Fig. 8). The seletive gluoneogeni effet of GCN5(S275D) ws ssoited with enhned AMP-indued etyltion of H3K9 nd H3K27 t gluoneogeni gene promoters, ut not t Gpdh or Hmgr promoters (Supplementry Fig. 6). GCN5(S275D) lso resued, t lest in prt, the inhiition of gluoneogeni gene expression y CITED2 depletion in pcpt-amp-stimulted heptoytes (Fig. 8d nd Supplementry Fig. 6d), suggesting tht GCN5 phosphorylted t Ser 275 funtions s gluoneogeni HAT downstrem of CITED2. These results indited tht phosphoryltion of GCN5 t Ser 275 y PKA in the GCN5-CITED2-PKA signlling module drives the sustrte swith of GCN5 nd therey promotes gluoneogenesis in vitro. We lso ssessed the phosphoryltion level of hepti GCN5 t Ser 275 in physiologil setting in vivo. The phosphoryltion level of GCN5 t Ser 275 in the liver of len mie ws inresed in the fsted stte ompred with the fed stte (Supplementry Fig. 6e). To onfirm the physiologil relevne of this finding, we exmined the time ourses of GCN5 phosphoryltion, PGC-1 etyltion, GCN5-CITED2 intertion nd gluoneogeni gene expression in the liver of len mie expressing oth FLAG-GCN5 nd HA-CITED2 t ner physiologil levels (out twie the of the endogenous proteins) during fsting fter refeeding for 12 h susequent to previous overnight fst. The 1 5 GCN5(WT): GCN5(S275A): CITED2: e Gluose prodution (AU) GCN5(S275D): GCN5(S275D): CITED2 shrna: GCN5(S275D) ,5 1, 5 d Reltive Gn5 mrna GCN5(S275D): 1 5 2,5 8 2, 1,5 1, f Gluose (mg dl 1 ) GCN5(S275D) Figure 8 Phosphoryltion of GCN5 t Ser 275 promotes gluoneogenesis. () Effets of fored expression of GCN5(WT) or GCN5(S275A) together with CITED2 on gluoneogeni gene expression in primry heptoytes exposed (or not) to pcpt-amp (6 h). (,) Effets of fored expression of GCN5(S275D) on gluoneogeni gene expression () nd gluose prodution () in primry heptoytes exposed (or not) to pcpt-amp (16 h). (d) qrt-pcr nlysis of G6p nd Pk1 expression in primry mouse heptoytes infeted with denoviruses enoding CITED2 shrna or GCN5(S275D) nd exposed (or not) to pcpt-amp for 6 h. (e,f) Effets of GCN5(S275D) expression in the liver of C57BL/6J mie on gluoneogeni gene expression (e) nd plsm glyemi (f) under the fsted (24 h) ondition. All dt re mens±s.e.m. (n ¼ 3(d)or7(e,f)). Po.5, Po.1 versus ontrol or s indited (ANOVA with Bonferroni s post ho test (d) or unpired Student s t-test (e,f)). Adenovirl vetors were used for these experiments. RTPCR, PCR with reverse trnsription. ANOVA, nlysis of vrine. 1 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

11 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE initil inrese in GCN5-CITED2 intertion (ssessed on the sis of the mount of CITED2 o-immunopreipitted with GCN5) ws followed y inreses in the level of GCN5 phosphoryltion t Ser 275 nd in PGC-1 (Supplementry Fig. 6f). Furthermore, the inrese in the level of GCN5 phosphoryltion ourred onomitntly with redution in the level of PGC-1 etyltion nd with the mximl upregultion of gluoneogeni gene expression (Supplementry Fig. 6f). In ddition, expression of GCN5(S275D) in the liver of len mie inresed gluoneogeni gene expression (Fig. 8e) nd lood gluose levels (Fig. 8f) in the fsted ondition. These dt thus suggested tht phosphoryltion of GCN5 t Ser 275 within the GCN5-CITED2-PKA signlling module during fsting drives the sustrte swith of GCN5, resulting in tivtion of PGC-1 nd of the HAT funtion of GCN5 nd onsequent promotion of gluoneogenesis. the liver of len mie (Fig. 8e,f), nd they indite tht inhiition of GCN5 phosphoryltion t Ser 275 suppresses gluoneogenesis nd therefore meliortes hyperglyemi, wheres enhned phosphoryltion of GCN5 t Ser 275 promotes gluoneogenesis nd exertes hyperglyemi, in dieti nimls. We hve previously shown tht depletion of CITED2 in the liver of d/d mie suppresses gluoneogenesis through enhnement of GCN5-dependent etyltion nd the onsequent intivtion of PGC-1 nd therey meliortes dietes 2. In this setting, phosphoryltion of GCN5 t Ser 275 ws lso inhiited (Fig. 9d), inditing tht disruption of the GCN5-CITED2-PKA signlling module y CITED2 depletion ttenutes gluoneogenesis nd meliortes hyperglyemi through suppression of GCN5 phosphoryltion t Ser 275 nd onsequent enhned GCN5- dependent etyltion nd inhiition of PGC-1 s well s redued GCN5-dependent etyltion of histone H3K9. Suppression of GCN5 phosphoryltion meliortes dietes. We lso exmined the phosphoryltion level of GCN5 t Ser 275 in the liver of dieti mie. Phosphoryltion of GCN5 (Fig. 9) s well s the mount of GCN5 protein (Fig. 1) were inresed in the liver of d/d mie nd of C57BL/6J mie fed HFD. The expression of CITED2 (ref. 2), ut not tht of PKA (Supplementry Fig. 7), ws lso inresed in the liver of dieti mie. These hnges in GCN5 nd CITED2 re onsistent with enhnement of gluoneogenesis nd prompted us to ddress whether GCN5 phosphoryltion t Ser 275 ontriutes to hyperglyemi through promotion of hepti gluoneogenesis in these nimls. Hepti expression of GCN5(S275D) inresed, wheres tht of GCN5(S275A) or GCN5(WT) deresed, lood gluose levels nd gluoneogeni gene expression in the liver of fsted d/d mie (Fig. 9, nd Supplementry Fig. 7). These effets re onsistent with those oserved in heptoytes (Fig. 8) or in Disussion Our dt show tht, in the fsted stte, glugon-amp signlling inreses the expression of CITED2 (ref. 2) nd GCN5 s well s promotes the formtion of GCN5-CITED2-PKA signlling module. Within this module, PKA tivted y AMP phosphoryltes GCN5 t Ser 275 (with the phosphoryltion of other PKA sustrtes suh s CREB, VASP nd IP3R eing unffeted), resulting in swith in the sustrte preferene of GCN5 from PGC-1 to histone H3 nd onsequent inrese in H3K9 etyltion nd the deetyltion-medited tivtion of PGC-1 t gluoneogeni gene promoters. The inrese in the mount of H3K9 promotes further epigeneti hnges nd the reruitment of trnsriptionl regultors required for initition of gene trnsription, wheres the tivted PGC-1 funtions s o-tivtor for gluoneogeni trnsription ftors suh s FoxO1 nd HNF-4. These two effets t oopertively d/m NC d/d GCN5(WT) GCN5(S275A) GCN5(S275D) IP: Phospho-Ser 275 GCN5 IP: GCN5 HFD IP: Phospho-Ser 275 GCN5 IP: GCN Gluose (mg dl 1 ) d GCN5(WT) GCN5(S275A) GCN5(S275D) Fed Fsted CITED2 shrna: IP: Phospho-Ser 275 GCN5 IP: GCN5 IP: CITED2 Figure 9 Suppression of hepti GCN5 phosphoryltion t Ser 275 meliortes dietes. () Anlysis of GCN5 phosphorylted t Ser 275 in the liver of d/d or d/m (ontrol) mie or of C57BL/6J mie fed NC or HFD. All mie were deprived of food for 16 h efore nlysis. Liver extrts were sujeted to IP with ntiodies to Ser 275 -phosphorylted GCN5 followed y immunolot nlysis with ntiodies to GCN5. (,) Effets of expression of GCN5(WT), GCN5(S275A) or GCN5(S275D) in the liver of d/d mie on plsm gluose onentrtion under fed or fsted (24 h) onditions () s well s on hepti expression of G6p nd Pk1 under the fsted (24 h) ondition (). (d) Effet of CITED2 depletion on GCN5 phosphoryltion t Ser 275 in the liver of d/d mie deprived of food for 24 h. All quntittive dt re mens±s.e.m. (n ¼ 7(,)). Po.5, Po.1 (ANOVA with Bonferroni s post ho test). Dt in,d re representtive of t lest three independent experiments. Adenovirl vetors were used for these experiments. ANOVA, nlysis of vrine; NC, norml how. NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

12 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 Feeding Glugon Fsting or dietes Glugon P P P FoxO1 AMP AMP R R C C CITED2 CITED2 CITED2 PKA R R R R C C CBP C C P GCN5 A PGC-1α etyltion Ser 275 P GCN5 CBP GCN5 PGC-1α PGC-1α etyltion PGC-1α PGC-1α PGC-1α PGC-1α HNF-4α OFF FoxO1 HNF-4α OFF FoxO1 HNF-4α A CBP H3K9 H3K27 H3K4me3 ON GCN5 HAT OFF PGC-1α AT ON HAT ON GCN5 PGC-1α AT OFF Figure 1 Model for regultion of hepti gluoneogenesis y the GCN5-CITED2-PKA signlling module through AMP-indued sustrte swith of GCN5. (left) In the fed stte, the GCN5-CITED2-PKA signlling module is not ssemled s result of the low level of CITED2 expression mintined in the sene of glugon signlling nd the insulin-indued inhiition of GCN5-CITED2 intertion. GCN5 is not phosphorylted t Ser 275 nd therefore etyltes PGC-1 rther thn histone H3, resulting in intivtion of PGC-1 nd onsequent suppression of gluoneogeni gene trnsription. (middle) In the fsted stte or dietes, glugon-amp signlling inreses the expression of CITED2 nd GCN5 nd therey promotes formtion of the GCN5-CITED2-PKA signlling module, within whih PKA tivted y AMP phosphoryltes GCN5 t Ser 275. (right) Phosphoryltion of GCN5 indues sustrte swith from PGC-1 to histone H3 nd onsequent inrese in H3K9 etyltion nd the deetyltion-medited tivtion of PGC-1 t gluoneogeni gene promoters. The inrese in the mount of H3K9 promotes further epigeneti hnges nd the reruitment of trnsriptionl regultors required for initition of gene trnsription, wheres tivted PGC-1 funtions s o-tivtor for gluoneogeni trnsription ftors suh s FoxO1 nd HNF-4. These two effets t oopertively to tivte the gluoneogeni progrm. AT, etyltrnsferse. to tivte the gluoneogeni progrm (Fig. 1). In the fed stte, ttenution of glugon signlling leds to downregultion of CITED2 nd GCN5 expression, wheres insulin signlling inhiits GCN5-CITED2 intertion 2, resulting in disruption of the GCN5-CITED2-PKA signlling module. In this ondition, phosphoryltion of GCN5 t Ser 275 is inhiited nd GCN5 therefore etyltes PGC-1 rther thn histone H3, resulting in intivtion of PGC-1 nd the onsequent suppression of gluoneogeni gene trnsription. A previous study showed tht GCN5 etyltes nd suppresses the otivtion tivity of PGC-1 (ref. 17). Overexpression of GCN5 in Fo ells thus suppressed gluoneogeni gene expression nd gluose prodution indued y overexpression of PGC-1 (ref. 17). Overexpression of GCN5 in the liver of mie lso redued hepti gluoneogeni gene expression in the fsted stte s well s glyemi fter fsting or pyruvte dministrtion 17 (Fig. 3e). In ddition, the PGC-1 etyltrnsferse funtion of GCN5 ws previously shown to e tivted y Sirt6 or the CDK4ylin D1 omplex, whose expression is downregulted y glugon-amp signlling or upregulted y insulingsk3 (glyogen synthse kinse 3) signlling, respetively 18,19. GCN5 hs therefore een thought to funtion s negtive regultor of PGC-1 nd, onsequently, of hepti gluoneogenesis (presumly in the fed stte) 182,3,31. On the other hnd, the role of GCN5 in gluoneogeni gene indution y glugonamp signlling during fsting nd its role s HAT hve remined to e eluidted. Given tht CITED2 forms omplex with GCN5 during fsting nd tivtes PGC-1 through inhiition of GCN5-dependent etyltion 2, we foused on the fsting-induile GCN5-CITED2 omplex. Our dt now provide evidene tht GCN5 funtions s HAT to tivte the gluoneogeni progrm in the fsted stte in ddition to s PGC-1 etyltrnsferse to inhiit it in the fed stte. This dul funtion of GCN5 is reiprolly regulted y glugon-dependent sustrte swith t the fed-to-fsting trnsition. The fsting-indued formtion of the GCN5- CITED2-PKA omplex llows the speifi PKA-dependent phosphoryltion of GCN5 t Ser 275, whih, in turn, drives the sustrte swith of GCN5. This newly unovered mehnism thus inorportes GCN5-medited repression of the trnsriptionl o-tivtor tivity of PGC-1 in the fed stte s well s AMP- nd CITED2-medited upregultion of suh tivity in the fsted stte, nd it therefore integrtes epigeneti hnges nd o-tivtor tivity, leding to full indution of gluoneogenesis. The GCN5-CITED2-PKA signlling module explins why the HAT tivity of GCN5 s well s oth CITED2 nd PKA re required for the gluoneogeni effet of GCN5 overexpression. However, the moleulr mehnism y whih expression of GCN5(WT) or GCN5(S275A) in the sene of CITED2 overexpression suppresses AMP- or fsting-indued gluoneogeni gene expression remins to e eluidted. It n e rgued tht hepti overexpression of GCN5(WT) lone should promote formtion of this signlling module with endogenous CITED2 nd PKA, the former of whih is upregulted during fsting, nd therey tivte gluoneogenesis (Fig. 3e). It should e noted, however, tht in this setting the of Gn5 mrna in the liver overexpressing GCN5 ws 3 times tht in the ontrol liver, wheres the mount of CITED2 protein in the liver of fsted mie showed n B2.5-fold inrese ompred with tht in refed mie 2. It is therefore likely tht the mount of endogenous CITED2 ws insuffiient to form the GCN5-CITED2-PKA signlling module with overexpressed GCN5. Given tht phosphoryltion of GCN5 y PKA is dependent on this module, the exess GCN5 would not e phosphorylted y PKA. As previously desried 17, the suppressive effet of overexpressed GCN5 in the sene of CITED2 overexpression on AMP- or fsting-indued gluoneogeni gene expression might e medited y PGC-1 intivtion through etyltion medited y the nonphosphorylted form of GCN5. However, we found tht expression of GCN5(DAT) lone lso inhiited AMP-indued gluoneogeni gene expression, suggesting tht overexpression of GCN5 suppresses gluoneogenesis through oth etyltrnsferse-dependent nd -independent mehnisms. We hve previously shown tht oexpression of CITED2 with 12 NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms

13 NATURE COMMUNICATIONS DOI: 1.138/nomms13147 ARTICLE GCN5(WT) hnges the pttern of GCN5 loliztion in the nuleus from homogeneous to spekled 2. Overexpressed GCN5 tht fils to intert with CITED2 my thus disrupt or ffet the loliztion of the GCN5-CITED2-PKA signlling module independently of its etyltrnsferse tivity, resulting in suppression of gluoneogeni gene expression. Depletion of GCN5 or expression of the phosphoryltiondefetive mutnt GCN5(S275A) in the liver of mie with oesity nd type 2 dietes suppressed gluoneogenesis nd therey improved glyemi, suggesting tht inhiition of GCN5 expression or phosphoryltion t Ser 275 my meliorte dietes. Disruption of the GCN5-CITED2-PKA signlling module (suh s tht hieved y CITED2 depletion) lso suppressed gluoneogenesis nd meliorted dietes, suggesting tht this module is promising phrmologil trget for tretment of oesity nd type 2 dietes. Methods Mie. All mouse experiments were performed ording to proedures pproved y the Institutionl Animl Cre nd Use Committee of the Ntionl Center for Glol Helth nd Mediine (Tokyo). Len C57BL/6J mle mie (CLEA Jpn) s well s d/d (C57BLKS/J Ir þ Lepr d / þ Lepr d )ndd/m (C57BLKS/J Ir þ m d / þ Lepr d ) mle mie otined from Institute for Animl Reprodution were studied t 8 weeks of ge. Reominnt denoviruses were injeted into the til vein of mie s desried previously 32 t dose of or plque-forming units in totl volume of.25 ml for C57BL/6J nd d/d mie, respetively, nd experiments were performed 4 dys fter denovirus injetion. Blood gluose onentrtion ws mesured with the use of stndrd gluose sensor (Glutest Ae, Snw Kgku Kenkyusho). For the pyruvte hllenge test, mie deprived of food for 16 h were injeted intrperitonelly with pyruvte dissolved in sline (2 g kg 1 ) s desried previously 33. For experiments with mie fed HFD, C57BL/6J mie were mintined on how ontining 3% ft y weight (14% ovine ft, 14% porine ft, 2% soyen oil; Orientl Yest) from 4 to 24 weeks of ge. For exmintion of the effets of glugon, mie were deprived of food for 3 h nd then injeted intrperitonelly with the hormone t dose of 1 mgkg 1. Liver extrts were prepred 1 h fter glugon tretment. We lloted ges of mie to the experimentl groups y rndom drw. The investigtor ws not linded to the group llotion during experiments. Plsmids. GCN5, CITED2, PGC-1, PKAC nd PKARI DNAs were isolted from C57BL/6J mouse liver nd loned into the mmmlin expression vetors pdna3 or pdna3.1 (Life Tehnologies). The DNAs for GCN5(DAT) (GYG-AYA/582Y584), GCN5(S275A) nd GCN5(S275D) were generted with the use of KOD-plus Mutgenesis Kit (Toyoo). Adenoviruses. Reominnt denoviruses were onstruted with the use of n Adenovirus Dul Expression Kit (Tkr Bio). FLAG-GCN5, My-GCN5, FLAG-GCN5(DAT), FLAG-GCN5(S275A), FLAG-GCN5(S275D), FLAG- CITED2, HA-CITED2, FLAGPGC-1, FLAG-FoxO1 nd -gltosidse (ontrol) were expressed under the ontrol of CAG promoter (ytomeglovirus enhner, hiken -tin gene promoter, nd rit -gloin gene poly(a) signl), wheres shrnas were expressed under the ontrol of U6 promoter. The shrnas for GCN5, CITED2 nd PGC-1 were sed on the sequenes 5 -TGTCAGAGG ACGAGATTAA-3 ;5 -TGACGGACTTCGTGTGCA-3 ;nd5 -GTATCTGACC ACAAACGAT-3, respetively. A negtive ontrol shrna sequene ws otined from BD Biosienes. Primry heptoytes or AML12 ells were infeted with denoviruses 1 dy fter plting. Anlyses of gene nd protein expression s well s gluose prodution ssy were performed 2 dys fter infetion. Chemils nd ntiodies. Glugon, pcpt-amp, insulin, MG132, trihosttin A nd etyl-coa were otined from Sigm; H89 nd niotinmide were from Enzo Life Sienes nd Nli Tesque, respetively; reominnt histone H3.1 ws from New Englnd Biols; nd reominnt humn CITED2 ws from Am. Antiodies to My (#2276, 1:1, dilution; #2272, 1:1,), to histone H3 (#4499, 1:1,), to H3K9 (#9649, 1:1,), to etylted lysine (#9441, 1:1,), to Ser/Thrphosphorylted PKA sustrtes (#9621, 1:1,), to CREB (#9197, 1:1,), to Ser 133 -phosphorylted CREB (#9198, 1:1,), to IP3R1 (#8568, 1:1,), to Ser phosphorylted IP3R (#8548, 1:1,), to VASP (#3132, 1:1,), to Ser 157 -phosphorylted VASP (#3111, 1:1,), to PKAC (#5842, 1:1,) nd to PKARI (#5675, 1:1,) were otined from Cell Signling. Antiodies to GCN5 (s-2698, 1:5; s , 1:5), to PCAF (s-13124, 1:5), to CBP (s-51517, 1:5; s-369), to HNF-4 (s-8987), to PGC-1 (s-67285, 1:5) nd to histone H1 (s-186, 1:4) were otined from Snt Cruz Biotehnology. Antiodies to CITED2 (18345, 1:1,), to H3K9 (4441, 1:1,) nd to H3K27 (4729) were from Am; those to p3 (5-257, 1:1,), to CRTC2 (ST-199, 1:2,), nd to H3K4me3 (5-745R) were from Millipore; those to FLAG (F184, 1:1,) nd to -tin (A5441, 1:1,) were from Sigm; those to HA ( , 1:2,) nd to His 6 ( , 1:5) were from Rohe; nd those to V5 (R96, 1:2,), to FLAG (KO62, 1:2,), to DsRed (for mcherry; , 1:1,), nd to T7 (69522, 1:1,) were from Life Tehnologies, Trnsgeni, Clonteh, nd Novgen, respetively. Rit ntiodies to Ser 275 -phosphorylted GCN5 (1:4) were generted in response to phosphopeptide ontining mino ids 271 to 28 of mouse GCN5. Cell ulture. Primry heptoytes were isolted from 8- to 12-week-old mle C57BL/6J mie fed norml how diet s desried previously 32. In rief, mie were nesthetised y intrperitonel injetion of medetomidine (.75 mg kg 1 ), midzolm (4 mg kg 1 ) nd utorphnol (5 mg kg 1 ), nd the liver ws perfused t rte of 4.5 ml min 1 first for 35 min with oxygented Hnks lned slt solution ontining 1 mm Hepes-NOH (ph 7.4) nd then for 18 min with the sme solution ontining ollgense type I (332 mg per 1 ml, Worthington) nd Protese Inhiitor Coktil CompleteEDTA Free (one tlet per 5 ml, Rohe). The heptoytes were hrvested nd purified y density grdient entrifugtion with Peroll (Sigm), nd their viility ws ssessed on the sis of trypn lue exlusion. We studied only heptoyte preprtions with viility of 49%. The ells were plted on type I ollgenoted six-well pltes ( ells per well) in Medium 199 (Life Tehnologies) supplemented with 5% fetl ovine serum, nd were inuted overnight in serum-free Medium 199 (Life Tehnologies) efore the ddition of pcpt-amp (1 mm) (ref. 34). AML12 nd HEK293 ells were otined from Amerin Type Culture Colletion. AML12 ells were ultured in 1:1 (v/v) mixture of Duleo s modified Egle s medium (DMEM) nd Hm s F12 medium tht ws supplemented with insulin (5 mgml 1 ), trnsferrin (5 mgml 1 ), selenium (5 ng ml 1 ), dexmethsone (4 ng ml 1 ), nd 1% fetl ovine serum, wheres HEK293 ells were ultured in DMEM supplemented with 1% fetl ovine serum. Primry heptoytes nd AML12 ells were tested for gluoneogeni gene indution y pcpt-amp. Where indited, primry heptoytes or AML12 ells were exposed to 2 mm H89 for 3 min efore inution in the dditionl presene of 1 mm pcpt-amp for the indited times. The ell lines were regulrly tested for myoplsm ontmintion. Gene expression nlysis. Totl RNA ws isolted from ells or pulverized liver with the use of n RNesy Mini Kit nd RNse-Free DNse Set (Qigen). For quntittive RT-PCR nlysis, DNA ws synthesized from the totl RNA with the use of rndom primers nd High Cpity DNA Reverse Trnsription Kit (Life Tehnologies) nd PCR ws then performed in triplite with the use of StepOnePlus Rel-Time PCR System nd Power SYBR Green PCR Mster Mix (Life Tehnologies). Reltive mrna ws lulted y the stndrd urve method nd ws normlized y the orresponding mount of 18S rrna. Primer sequenes re listed in Supplementry Tle 1. Protein intertion nlysis. Protein-protein intertions in ells were exmined with o-ip ssys. Epitope-tgged proteins were expressed in AML12 ells or HEK293 ells y trnsfetion with the use of Nuleofetor Kit V (Amx) or y denovirl trnsdution. The ells were then lysed in lysis uffer ontining 2 mm Tris-HCl (ph 7.5), 15 mm NCl,.5% Nonidet P-4, 2 mm EDTA, 1% glyerol, 1 mm niotinmide, 1 mm trihosttin A, 1 mm MG132, nd protese nd phosphtse inhiitors (Rohe), nd the lystes were sujeted to IP with the indited ntiodies nd either protein G-Sephrose (GE Helthre) or Dyneds Protein G (VERITAS). The immunopreipittes were frtionted y SDSpolyrylmide gel eletrophoresis (PAGE) nd sujeted to immunolot nlysis with the indited ntiodies. Unropped imges of representtive immunolots re shown in Supplementry Fig. 8. Preprtion of nuler extrts from mouse liver. The liver ws removed, rinsed in ie-old phosphte-uffered sline (PBS), suspended in 5 ml of uffer A (1 mm Hepes-NOH (ph 7.9), 25 mm KCl, 1 mm EDTA, 2 M surose, 1% glyerol, 1 mm niotinmide, 1 mm trihosttin A, 1 mm MG132 nd protese nd phosphtse inhiitors (Rohe)), nd homogenized riefly with Polytron disruptor. The homogente (15 ml) ws pssed five times through 2 G needle nd then lyered on top of 2 ml of uffer A in entrifuge tue nd entrifuged t 1,g for 1 h t 4 C. The resulting nuler pellet ws wshed with 3 ml of solution ontining 1 mm Hepes-NOH (ph 7.9), 1 mm KCl, 2 mm MgCl 2, 1 mm EDTA nd 1% glyerol nd ws then digested with Enzymtil Shering Coktil in 2 ml of Complete Digestion Buffer (Nuler Complex Co-IP Kit, Ative Motif). The retion ws terminted y the ddition of 4 ml of.5 M EDTA, nd the mixture ws then entrifuged t 14,g for 1 min t 4 C. The resulting superntnt ws olleted s the nuler extrt for o-ip experiments. Gluose prodution ssy. Primry heptoytes were ultured in serum-free Medium 199 in the sene or presene of 1 mm pcpt-amp for 16 h s previously desried 34. They were then inuted for 6 h in gluose- nd phenol red-free DMEM (ph 7.4) supplemented with sodium ltte nd pyruvte efore NATURE COMMUNICATIONS 7:13147 DOI: 1.138/nomms