PRMT5 modulates the metabolic response to fasting signals

Wen-Wei Tsaia, Sherry Niessenb, Naomi Goebela, John R. Yates IIIb, Ernesto Guccionec,d, and Marc Montminya,1

aThe Clayton Foundation Laboratories for Peptide Biology, Salk Institute, La Jolla, CA 92037; bDepartment of Chemical Physiology, The Scripps Research Institute, La Jolla, CA 92037; cInstitute of Molecular and Cell Biology, Proteos, Singapore 138673; dDepartment of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 119074

Contributed by Marc Montminy, March 11, 2013 (sent for review February 20, 2013) Under fasting conditions, increases in circulating glucagon maintain sites. The results provide a mechanism to explain how latent cy- glucose balance by promoting hepatic . Triggering toplasmic regulators such as CRTC2 may contribute to signaling of the cAMP pathway stimulates gluconeogenic expression in the nucleus through their association with mod- through the PKA-mediated phosphorylation of the cAMP response ifying enzymes. element binding (CREB) and via the dephosphorylation of the latent cytoplasmic CREB regulated transcriptional 2 Results (CRTC2). CREB and CRTC2 activities are increased in insulin resis- In mass spectroscopy studies using epitope-tagged CRTC2 to tance, in which they promote hyperglycemia because of constitu- identify relevant interacting , we recovered the protein tive induction of the gluconeogenic program. The extent to which arginine methyltransferase 5 (PRMT5) (Fig. S1 A and B). We CREB and CRTC2 are coordinately up-regulated in response to glu- confirmed the CRTC2:PRMT5 interaction in coimmunopreci- cagon, however, remains unclear. Here we show that, following its pitation studies with epitope-tagged proteins (Fig. 1A). The activation, CRTC2 enhances CREB phosphorylation through an as- CRTC2:PRMT5 association was evident under basal conditions sociation with the protein arginine methyltransferase 5 (PRMT5). In and to a greater extent following exposure to cAMP agonist turn, PRMT5 was found to stimulate CREB phosphorylation via forskolin (FSK). increases in H3 Arg2 that enhanced chromatin CRTC2 and its paralogs share a similar structure consisting of

accessibility at gluconeogenic promoters. Because depletion of an N-terminal CREB-binding domain (amino acids 1–55), central BIOCHEMISTRY PRMT5 lowers hepatic glucose production and gluconeogenic gene regulatory domain (amino acids 55–600), and C-terminal trans- expression, these results demonstrate how a chromatin-modifying activation domain (amino acids 600–692) (2). In deletion map- enzyme regulates a metabolic program through epigenetic ping studies, PRMT5 was found to associate with the central changes that impact the phosphorylation of a regulatory region of CRTC2; deletion of sequences within this in response to hormonal stimuli. domain (amino acids 56–144) disrupted the CRTC2:PRMT5 interaction (Fig. 1B). hronic elevations in circulating blood glucose contribute to the The PRMT family of arginine methylases consists of three types; Cchronic complications of type II diabetes, which include retinal they are distinguished by their ability to carry out mono- or asym- blindness, renal failure, peripheral neuropathy, and cardiovascular metric di-methylation (type I), mono- or symmetric di-methylation diseases (1). Fasting hyperglycemia represents an early sign of (type II), or exclusively monomethylation (type III) (8, 9). PRMT5 insulin resistance, which is characterized by an inability for insulin functions as the major type II symmetric arginine methyltransfer- to promote glucose uptake into muscle and to inhibit glucose ase. Consistent with a critical role during development, mice with production by the liver. aknockoutofPrmt5 have early embryonic lethality (10). PRMT5 During fasting, increases in circulating pancreatic glucagon has also been shown to function in transcriptional regulation as well stimulate the gluconeogenic program via the protein kinase A as signal transduction (11). (PKA)-mediated phosphorylation and activation of the tran- Based on the ability for PRMT5 to associate with CRTC2, we scription factor cAMP response element binding (CREB) pro- tested whether this enzyme is correspondingly recruited to CREB tein, a modification that promotes recruitment of the coactivators target in response to fasting signals. In ChIP studies, expo- P300 and CREB binding protein (CBP) (2). In parallel, increases sure of mouse primary hepatocytes to glucagon increased amounts in cAMP signaling also promote the dephosphorylation of the of CRTC2 over CREB-binding sites on the gluconeogenic glucose- CREB regulated transcriptional coactivator 2 (CRTC2) (3). Se- 6-phosphatase (G6pc) and phosphoenolpyruvate carboxykinase questered in the cytoplasm through phosphorylation-dependent (Pck1) promoters (Fig. 1C). Consistent with its ability to associate interactions with 14-3-3 proteins, CRTC2 undergoes de- with CRTC2, PRMT5 was also recruited to these CREB target Crtc2 phosphorylation in response to cAMP, where it shuttles to the genes following exposure to glucagon. Depletion of with adenovirally encoded RNAi disrupted PRMT5 recruitment in nucleus and mediates induction of cellular genes via an interaction hepatocytes exposed to glucagon (Fig. 1D). with CREB over relevant binding sites (4, 5). Supporting a functional role for PRMT5 in modulating CREB CRTC2 dephosphorylation represents an early event in the activity, exposure to glucagon increased cAMP response element cAMP response relative to CREB phosphorylation, although the (CRE)-luciferase (Luc) and G6pc-Luc reporter activities ro- extent to which the two processes are linked remains unclear (6). bustly; these effects were disrupted following RNAi-mediated The PKA-mediated phosphorylation of CREB on chromatin depletion of Prmt5 (Fig. 1E). Conversely, overexpression of Crtc2 templates appears to be generally inefficient compared with its phosphorylation on naked DNA templates; these inhibitory effects can be circumvented through the acetylation of resident nucleo- Author contributions: W.-W.T. and M.M. designed research; W.-W.T. and N.G. performed somes and subsequent increases in chromatin accessibility (7). The research; S.N., J.R.Y., and E.G. contributed new reagents/analytic tools; W.-W.T., S.N., J.R.Y., extent to which these or other chromatin modifications are per- E.G., and M.M. analyzed data; and W.-W.T. and M.M. wrote the paper. missive for CREB activation in vivo has not be addressed, however. The authors declare no conflict of interest. Here we characterize a CRTC2-associated arginine methylase 1To whom correspondence should be addressed. E-mail: [email protected]. that promotes CREB phosphorylation and target gene activation This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. in part by enhancing chromatin accessibility over CREB binding 1073/pnas.1304602110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1304602110 PNAS Early Edition | 1of6 Downloaded by guest on September 25, 2021 A B C input input IP: myc CRTC2 CRTC2

- + - + - + FSK - + + + + + myc-PRMT5 - + + + + +

CRTC2 myc PRMT5 myc-PRMT5

PRMT5 PRMT5 CREB Flag- Flag- CRTC2s CRTC2s IP: HA

PRMT5

D E F CRTC2 CRTC2 CRE-Luc

NSi Prmt5i glucagon - + - +

PRMT5

p-CRTC2 PRMT5 PRMT5 G6pc-Luc CRTC2

CREB

Hsp90

Fig. 1. PRMT5 associates with CRTC2 and is recruited to gluconeogenic genes in response to glucagon. (A) Coimmunoprecipitation assay of HEK293T cells expressing epitope-tagged PRMT5 and CRTC2. No interaction with epitope-tagged CREB as shown. Exposure to FSK (30 min) indicated. (B) Coimmunopre- cipitation assay of WT and mutant CRTC2 polypeptides with PRMT5 in cells exposed to FSK for 30 min. endpoints for each construct indicated. (C) ChIP assay of primary hepatocytes showing effect of glucagon (1 h) on recruitment of PRMT5 and CRTC2 to CREB-binding sites over gluconeogenic (Pck1, G6pc) promoters. Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. **P < 0.01; ***P < 0.001. (D) Effect of CRTC2 depletion by RNAi-mediated knockdown on PRMT5 recruitment to gluconeogenic promoters in primary hepatocytes exposed to glucagon (1 h). Each bar represents averaged results, n = 2 biological replicates, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01. (E) Effect of PRMT5 depletion on CRE-Luc and G6pc-Luc reporter activities in primary hepatocytes exposed to glucagon (4–6 h). Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. ***P < 0.001. (F) Western blot showing effect of Prmt5 RNAi on PRMT5 protein amounts and on CRTC2 dephosphorylation in primary hepatocytes exposed to glucagon (30 min). Hsp, heat shock protein 90.

and Prmt5 increased G6pc-Luc reporter activity cooperatively in exposed to glucagon (Fig. 2C). As a consequence, knockdown of transient assays of HepG2 hepatoma cells (Fig. S1C). In keeping Prmt5 reduced glucose secretion from primary hepatocytes exposed with the role of CRTC2 in promoting PRMT5 recruitment, to glucagon (Fig. 2D). Pointing to a specific effect on CREB sig- PRMT5 had no effect on G6pc-Luc reporter activity in cells naling, Prmt5 knockdown did not alter mRNA amounts for in- depleted of Crtc2 (Fig. S1D). Arguing against a direct effect on sulin-like growth factor binding protein 1 (Igfbp1), a Forkhead box cAMP signaling per se, knockdown of Prmt5 did not interfere O1 (FOXO1) target gene that is also up-regulated during fasting with the dephosphorylation and hence activation of CRTC2 in (Fig. S2D). hepatocytes exposed to glucagon (Fig. 1F). Taken together, these PRMT5 has been reported to exert both positive and negative results indicate that PRMT5 up-regulates CREB activity following effects on cellular (8), prompting us to evaluate its CRTC2-dependent recruitment to relevant genes in response relative effects of this methylase in primary hepatocytes by gene to cAMP agonist. profiling analysis. Depletion of Prmt5 down-regulated the ex- We tested whether PRMT5 is required for induction of the glu- pression of 1.5% of cellular genes, whereas it up-regulated the coneogenic program during fasting. RNAi-mediated depletion of Prmt5 in liver decreased circulating glucose concentrations and expression of about 1% of genes (Fig. 2E). By contrast, Prmt5 gluconeogenic gene expression in lean mice and to a greater extent depletion decreased the expression of 20% of glucagon-inducible in genetically obese (ob/ob) or high-fat diet fed insulin-resistant genes, many of which are direct targets for CREB that function mice (Fig. 2 A and B and Fig. S2 A–C). The effect of PRMT5 on in fasting glucose [tyrosine amino transferase (Tat), hepatic gluconeogenesis appears cell autonomous because RNAi- G6pc, Pck1, Ppargc1a (Fig. 2F). Consistent with its proposed role mediated depletion of Prmt5 also down-regulated expression of in potentiating CREB activity, knockdown of Prmt5 did not up- Pck1, G6pc, and Peroxisome Proliferator Activated Receptor regulate the expression of any glucagon-inducible gene in Gamma Coactivator 1a (Ppargc1a) genes in primary hepatocytes this setting.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304602110 Tsai et al. Downloaded by guest on September 25, 2021 A B F Prmt5 Pck1 ob/+ ob/ob

G6pc Ppargc1

C Prmt5 Pck1 D glucose secretion

G6pc Ppargc1 BIOCHEMISTRY

E

GLU

Fig. 2. PRMT5 stimulates hepatic gluconeogenesis during fasting and in diabetes. (A) Effect of hepatic Prmt5 depletion with adenovirally encoded RNAi on fasting blood glucose concentrations in genetically obese (ob/ob) mice compared with lean controls (ob/+). Each bar represents averaged results, n = 5; error bars, SD. **P < 0.01; ***P < 0.001. (B) Effect of hepatic PRMT5 knockdown on gluconeogenic gene expression in livers of fasted obese (ob/ob) mice. Each bar represents averaged results, n = 5; error bars, SD. **P < 0.01; ***P < 0.001. (C) Effect of PRMT5 knockdown on gluconeogenic gene expression in primary hepatocytes exposed to glucagon. Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01; ***P < 0.001. (D) Effect of PRMT5 depletion on glucose secretion from primary hepatocytes exposed to glucagon. Each bar represents averaged results, n = 3; error bars, SD. **P < 0.01. (E) Gene profiling analysis of primary mouse hepatocytes following RNAi-mediated knockdown of Prmt5. Pie charts show proportion of total cellular genes up or down-regulated 1.5-fold or better under basal conditions and following exposure to glucagon. (F) Heat map showing effects of Prmt5 knockdown on top 40 scoring glucagon-inducible genes in primary hepatocytes.

PRMT5 has been shown to catalyze the symmetric di-methyla- recruitment to gluconeogenic promoters (Fig. 3C). Conversely, tion of at Arg-2 (-s) and at Arg-8 (H3R8me2- Prmt5 depletion decreased both H3R2me2-s and WDR5 s); these marks have been associated with transcriptional repression amounts over these genes in primary hepatocytes following as well as activation (8, 12). In turn, H3R2me2-s has been reported glucagon treatment (Fig. 3D). Depletion of Crtc2 also de- to modulate cellular gene expression in part by promoting the re- creased H3R2me2-s amounts at Pck1 and G6pc promoters, cruitment of WD repeat protein 5 (WDR5) (12), a core subunit of demonstrating the importance of CRTC2 in mediating PRMT5 lysine methyl transferase complexes (13) and a component of the recruitment (Fig. S3B). transcriptional adaptor (Ada) 2A containing complex (ATAC) Based on the ability for PRMT5 to increase H3R2me2-s (14). We confirmed that PRMT5 catalyzes increases in H3R2me2-s amounts over CREB target genes, we considered whether this invitroandthatH3R2me2-sinturnassociateswithWDR5byGST methylase could open chromatin structure at relevant promoters pull-down assays (Fig. 3 A and B). in response to glucagon. Supporting this idea, exposure to glu- Consistent with these observations, exposure of primary hep- cagon decreased nucleosome occupancy over CREB-binding sites atocytes to glucagon increased H3R2me2-s amounts alongside in primary hepatocytes as revealed by decreases in histone H3 the active mark histone H3 trimethyl-lysine 4 at CREB-binding occupancy; these effects were blocked following RNAi-mediated sites over gluconeogenic genes (Fig. 3C and Fig. S3A). In keeping knockdown of Prmt5 or Crtc2 (Fig. 4 A and B and Fig. S4A). Taken with these increases, exposure to glucagon stimulated WDR5 together, these results indicate that PRMT5 enhances CREB

Tsai et al. PNAS Early Edition | 3of6 Downloaded by guest on September 25, 2021 activation in part by increasing H3R2me2-s amounts over CREB binding sites in response to cAMP agonist. We reasoned that PRMT5 may enhance chromatin accessibility over CREB-regulated promoters. To test this idea, we evaluated A in-vitro methylation B effects of PRMT5 on micrococcal nuclease (MNase) sensitivity in control and Prmt5-depleted cells in response to glucagon. Although GST pull-down it did not alter MNase sensitivity globally, Prmt5 depletion blocked glucagon-dependent increases in MNase sensitivity over G6pc and Pck1 promoters (Fig. 4 C and D). As a result, Ser133-phosphory- H3R2me2-s lated CREB (p-CREB) amounts over gluconeogenic genes were substantially reduced in Prmt5-depleted compared with control GST-WDR5 cells exposed to glucagon (Fig. 4E and Fig. S4B). Similarly, de- H3R2me2 pletion of Crtc2 reduced p-CREB amounts over gluconeogenic promoters in cells exposed to glucagon, reflecting in part the im- H3 portance of this coactivator in mediating PRMT5 recruitment (Fig. H3 S4C). Although Prmt5 depletion decreased p-CREB amounts in cells exposed to glucagon, it did not interfere with the phosphory- lation of cytosolic PKA substrates such as liver kinase B1 (LKB1) C H3R2me2-s H3R2me2-s (Fig. 4F). Collectively, these results demonstrate that PRMT5 modulates the expression of signal-dependent genes through increases in chromatin accessibility that are permissive for the phosphorylation of nuclear factors such as CREB by PKA and perhaps other regulatory enzymes (Fig. 4G). Discussion Under fasting conditions, increases in circulating glucagon pro- mote hepatic glucose production via the cAMP-dependent phos- phorylation of CREB and dephosphorylation of CRTC2 (2). WDR5 WDR5 CREB and CRTC2 activities are increased in diabetes, where they contribute to the attendant hyperglycemia via up-regulation of the gluconeogenic program. We found that CRTC2 stimulates gluconeogenic gene expression viaanassociationwiththeproteinargininemethyltransferase PRMT5 in response to glucagon stimulation. Following its re- cruitment to CREB-binding sites, PRMT5 promotes increases in H3R2me2-s that in turn trigger the expression of relevant CREB target genes. PRMT5-dependent increases in H3R2me2-s appeared D H3R2me2-s H3R2me2-s to stimulate cellular gene expression in part by enhancing the re- cruitment of WDR5. The CRTC2-dependent recruitment of this chromatin-modifying enzyme to CREB-binding sites appears criti- cal for the subsequent phosphorylation of CREB. In addition to effects of PRMT5, the type I PRMT family member PRMT4/coactivator-associated arginine methyltransferase 1 (CARM1) has also been found to promote gluconeogenic gene expression via an interaction with CREB and through sub- sequent increases in H3 methylation (15). Indeed, we have also detected increases in asymmetric H3R8me2 amounts at CREB- WDR5 WDR5 binding sites, potentially catalyzed by the type I enzyme PRMT2, in response to glucagon simulation. Collectively, these studies indicate that histone arginine methylation plays an important role in hepatic gluconeogenesis. Phosphorylation has been shown to modulate transcription factors and coactivators in the cytoplasm by regulating their nu- clear translocation, and in the nucleus by increasing their trans- activation potential (16). Consistent with its proximity to the Fig. 3. PRMT5 promotes H3R2 symmetric dimethylation and recruitment of signaling machinery in the cytoplasm, CRTC2 undergoes rapid and WDR5 to gluconeogenic promoters in response to glucagon. (A) Western quantitative dephosphorylation in response to glucagon. Although blot showing effect of PRMTs (5–7) on H3R2me2 and H3R2me2-s of histone the two events are closely spaced, CRTC2 dephosphorylation and H3 Arg-2 in vitro. (B) Western blot showing effect of PRMT5 and PRMT6- promoter recruitment appear to precede cAMP-dependent in- catalyzed histone H3 Arg-2 di-methylation on WDR5 binding by pull-down creases in phospho-CREB levels, suggesting a potential role for assay with GST-WDR5. (C) ChIP assay of primary hepatocytes, showing effect CRTC2 in contributing to CREB activation (6). of glucagon (1 h) on H3R2me2-s methylation and WDR5 recruitment at In reconstitution studies, CREB phosphorylation by PKA was CREB-binding sites over gluconeogenic promoters. Each bar represents av- strongly inhibited on chromatin compared with naked DNA tem- eraged results, n = 3 biological replicates, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01. (D) ChIP assay showing effect of Prmt5 plates, whereas CREB occupancy was more resistant (7). We knockdown on H3R2me2-s and WDR5 amounts over gluconeogenic pro- found that CRTC2 enhances CREB phosphorylation in response moters in hepatocytes exposed to glucagon (1 h). Each bar represents av- to glucagon via its association with PRMT5. In turn, PRMT5 po- eraged results, n = 3 biological replicates, assayed three times each; error tentiated CREB phosphorylation through increases in H3R2me2-s bars, SD. *P < 0.05; **P < 0.01. that stimulate nucleosome clearance and thereby increase

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304602110 Tsai et al. Downloaded by guest on September 25, 2021 A BC D Pck1-CRE H3 H3 NSi P5i NSi P5i NSi P5i NSi P5i NSi P5i NSi P5i - - 1x 1x 2x 2x 4x 4x 8x 8x 16x 16x MNase

1000bp G6pc-CRE H3 H3 500bp

100bp

EFG

CREB CREB NSi Prmt5i - + - + glucagon

PRMT5 BIOCHEMISTRY p-CREB

CREB p-CREB p-CREB p-LKB1

Hsp90

Fig. 4. PRMT5 promotes CREB phosphorylation at gluconeogenic promoters by increasing chromatin accessibility at CREB-binding sites. (A)Effectof glucagon (1 h) exposure on total histone H3 amounts over CREB-binding sites on gluconeogenic genes in primary hepatocytes. Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01. (B)EffectofPrmt5 knockdownonH3oc- cupancy over CREB-binding sites at gluconeogenic promoters in hepatocytes exposed to glucagon for 1 h. Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. *P < 0.05. (C) Effect of PRMT5 depletion on MNase digestion of chromatin from primary hepatocytes exposed to glucagon (1 h). Increasing amounts of MNase indicated. (D)EffectofPrmt5 RNAi on MNase sensitivity over CREB-binding sites on Pck1 and G6pc promoters in cells exposed to glucagon (1 h). Representative experiment from three independent experiments shown. Each bar represents averaged results, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01. ***P < 0.001. (E) ChIP assay showing effect of Prmt5 RNAi on CREB and p-CREB amounts over gluconeogenic promoters in primary hepatocytes exposed to glucagon (1 h). Each bar represents averaged results, n = 3 biological replicates, assayed three times each; error bars, SD. *P < 0.05; **P < 0.01. ***P < 0.001. (F) Western blot showing effect of PRMT5 depletion on amounts of p-CREB and total CREB in primary hepatocytes under basal conditions and following exposure to glucagon (30 min). Relative phosphorylation of LKB in response to glucagon shown for comparison. (G) Model for activation of gluconeogenic genes during fasting. Exposure to glucagon leads to CRTC2 activation and association with PRMT5. Following its CRTC2-mediated recruitment to relevant target genes, PRMT5 increases nucleosome clearance over CREB-binding sites, in part through increases in histone H3 Arg2 di-methylation that promote WDR5 occupancy. As a consequence of the increase in chromatin accessibility, CREB phosphorylation by PKA is up-regulated over PRMT5 occupied promoters, leading to increases in target gene expression. NSi, nonspecific RNAi; P5i, PRMT5 RNAi; p-LKB1, phospho-liver kinase B1.

accessibility to PKA. Taken together, these results indicate that Coimmunoprecipitation. HEK293T cells and cultured mouse primary hep- local changes in chromatin structure may be required for sig- atocytes were plated in six-well plates with or without FSK or glucagon treatment. HEK293T cells were transfected with indicated plasmids using naling to certain nuclear factors. Future studies should reveal ’ μ whether such changes are predictive for the selective activation of Effectene (Qiagen) following manufacturer s instructions. Typically, 1- g plasmids were used for each transfection. Protein lysate was collected 24 h target gene subsets by CREB and other phosphorylation-dependent after transfection. The treated cells were lysed in NTDT buffer (150 mM factors in response to extracellular stimuli. NaCl, 50 mM Tris·HCl, 0.1% deoxycholate, and 1% TX-100). The cellular debris was pelleted by centrifugation and the supernatant (immunoprecip- Materials and Methods itation lysate) was collected and diluted for coimmunoprecipitation experi- Cell Culture. HEK293T cells were cultured in DMEM with 10% (wt/vol) FBS and ments. The lysate was precleared with 20 μL protein A

1× Pen/Strep at 37 °C in 5% (wt/vol) CO2 and exposed to FSK (10 μM). Mouse beads for 1 h at 4 °C. Ethidium bromide was added to a final concentration primary hepatocytes were isolated as previously described and cultured in of 100 μg/mL to prevent protein–DNA binding. Precleared, ethidium bromide- Medium 199 and exposed to glucagon (100 nM) (18). treated lysate was incubated with indicated antibody-bound agarose beads

Tsai et al. PNAS Early Edition | 5of6 Downloaded by guest on September 25, 2021 overnight at 4 °C. Protein/antibody-bound beads were recovered by centri- Glucose Secretion Analyses. Mouse primary hepatocytes were plated in six- fugation and washed three times with NTP buffer (150 mM NaCl, 50 mM well plates and exposed to glucagon for 7 h. Glucose concentrations in Tris·HCl, 0.1% Nonidet P-40). The protein/antibody-bound beads were in- conditioned medium were measured as described (22). cubated with 3× SDS/PAGE loading buffer for 5 min at 95 °C and subjected fi to SDS/PAGE. Gene Pro ling Analyses. The mouse primary hepatocytes were plated in Mass spectroscopy studies were performed on precipitates from Halo- 60-mm plates and infected with adenovirus encoded with indicated shRNAs for 60 h and exposed to glucagon for 1.5 h. RNA was isolated and analyzed by tagged CRTC2 as reported (19). Affymetrix GeneChip Mouse Gene 1.0 ST Array following manufacturer’s instructions. Top 40 glucagon-induced genes were chosen for heat map Chromatin Immunoprecipitation Assay. Cultured mouse primary hepatocytes analysis using matrix2png software (23). were plated in 150-mm plates and exposed to glucagon for 1 h. Adeno- virally encoded shRNAs were added to cells for 60 h. ChIP assays were In Vitro Methylation and GST Pull-Down Assays. Recombinant H3 proteins fi fl performed as described, with minimal modi cation (20). Brie y, the (Active Motif) were methylated in vitro using purified PRMTs expressed treated hepatocytes were cross-linked with 1% formaldehyde for 15 min. either in bacteria as GST fusion proteins (GST-PRMT6, GST-PRMT7) or as Cross-linking reactions were stopped with 0.125 M glycine. Cross-linked myelocytomatosis (myc)-tagged proteins (myc-PRMT5) purified from cells were washed by PBS three times and stored at −80 °C before use. HEK293T cells as described (24). Methylated H3 proteins were then iso- Fragmented, precleared chromatin lysate was incubated overnight with lated and incubated with GST-WDR5 proteins in NTP buffer (150 mM NaCl, indicated antibodies: H3, histone H3 trimethyl-lysine 4, and PRMT5 (Active 50 mM Tris·HCl, 0.5% Nonidet P-40) overnight. GST Sepharose beads were Motif), PRMT5, and WDR5 (Millipore). addedtoproteinsamplesfor1handwashed three times. The GST-WDR5/ H3 bound beads were analyzed by SDS/PAGE. Luciferase Reporter Assay. Cultured mouse primary hepatocytes were infected with Ad-CRE-Luc or Ad-G6PC-Luc and adenovirally encoded shRNAs for 60 h MNase Digestion Analyses. Mouse primary hepatocytes were plated in six-well plates and infected with adenovirus encoded with indicated shRNAs for 60 h and exposed to glucagon for 4–6 h. Luciferase assays were performed as and exposed to glucagon for 1 h. Micrococcal nuclease (MNase) digestion described (21). analyses were performed as previously described (25), with minimal modi- fication. Briefly, the treated hepatocytes were permeabilized using 0.05% Mouse Strains and in Vivo Analyses. High-fat diet-fed ob/+ and ob/ob mice lysolecithin for 5 min, and treated with different amounts of MNase for were purchased from Jackson Laboratories. All mice were adapted to their 5min. DNA was purified and analyzed by quantitative PCR. environment for 1 wk before study. Adenovirus encoded with indicated × 8 shRNAs (3 10 pfu) were delivered to mice by eye injection. Six days after Statistical Analyses. GraphPad Prism6 software (GraphPad Software, Inc.) was infection, mice were fasted for 16 h, blood glucose concentrations were used for analysis of P values based on at least two independent experiments measured before and after fasting, and liver tissue was collected after in three independent PCR reactions. The two-tailed unpaired t test was used fasting for analysis. Animal studies were approved by the Institutional to compare the differences between two groups. P values < 0.05 were Animal Care and Use Committee (IACUC) at the Salk Institute. considered statistically significant, *P < 0.05, **P < 0.01, and ***P < 0.001.

RNA Analyses. Cellular RNA was isolated by RNeasy kit (Qiagen) from mouse ACKNOWLEDGMENTS. We thank M. Bedford for PRMT plasmids and liver tissue or from mouse primary hepatocytes. Mouse primary hepatocytes K. Ravnskjaer and J. Butler for advice and reagents. Funding for this work was provided by National Institutes of Health Grants F32 DK096778 (to W.-W.T.), were plated in six-well plates and infected with adenovirus encoded with R01 DK049777, R01 DK083834, and R01 DK091618; the Clayton Foundation indicated shRNAs for 60 h and exposed to glucagon for 1 and 2 h. mRNA levels for Medical Research; the Leona M. and Harry B. Helmsley Charitable Trust; were measured as described (20). and the Kieckefer Foundation.

1. Saltiel AR (2001) New perspectives into the molecular pathogenesis and treatment of 14. Krebs AR, Karmodiya K, Lindahl-Allen M, Struhl K, Tora L (2011) SAGA and ATAC type 2 diabetes. Cell 104(4):517–529. histone acetyl transferase complexes regulate distinct sets of genes and ATAC defines 2. Altarejos JY, Montminy M (2011) CREB and the CRTC co-activators: Sensors for hor- a class of p300-independent enhancers. Mol Cell 44(3):410–423. monal and metabolic signals. Nat Rev Mol Cell Biol 12(3):141–151. 15. Krones-Herzig A, et al. (2006) Signal-dependent control of gluconeogenic key enzyme 3. Koo SH, et al. (2005) The CREB coactivator TORC2 is a key regulator of fasting glucose genes through coactivator-associated arginine methyltransferase 1. J Biol Chem metabolism. 437(7062):1109–1111. 281(6):3025–3029. 4. Screaton RA, et al. (2004) The CREB coactivator TORC2 functions as a calcium- and 16. Hunter T, Karin M (1992) The regulation of transcription by phosphorylation. Cell – cAMP-sensitive coincidence detector. Cell 119(1):61–74. 70(3):375 387. 5. Bittinger MA, et al. (2004) Activation of cAMP response element-mediated gene ex- 17. Mayall TP, Sheridan PL, Montminy MR, Jones KA (1997) Distinct roles for P-CREB and pression by regulated nuclear transport of TORC proteins. Curr Biol 14(23):2156–2161. LEF-1 in TCR alpha enhancer assembly and activation on chromatin templates in vitro. – 6. Ravnskjaer K, et al. (2007) Cooperative interactions between CBP and TORC2 confer Genes Dev 11(7):887 899. 18. Dentin R, et al. (2007) Insulin modulates gluconeogenesis by inhibition of the co- selectivity to CREB target gene expression. EMBO J 26(12):2880–2889. activator TORC2. Nature 449(7160):366–369. 7. Michael LF, Asahara H, Shulman AI, Kraus WL, Montminy M (2000) The phosphory- 19. Qi L, et al. (2006) TRB3 links the E3 ubiquitin ligase COP1 to lipid metabolism. Science lation status of a cyclic AMP-responsive activator is modulated via a chromatin- 312(5781):1763–1766. dependent mechanism. Mol Cell Biol 20(5):1596–1603. 20. Wang Y, et al. (2010) Targeted disruption of the CREB coactivator Crtc2 increases 8. Bedford MT, Clarke SG (2009) Protein arginine methylation in mammals: Who, what, insulin sensitivity. Proc Natl Acad Sci USA 107(7):3087–3092. and why. Mol Cell 33(1):1–13. 21. Liu Y, et al. (2008) A fasting inducible switch modulates gluconeogenesis via activator/ 9. Bedford MT, Richard S (2005) Arginine methylation an emerging regulator of protein coactivator exchange. Nature 456(7219):269–273. function. Mol Cell 18(3):263–272. 22. Wang Y, et al. (2012) Inositol-1,4,5-trisphosphate receptor regulates hepatic gluco- 10. Tee WW, et al. (2010) Prmt5 is essential for early mouse development and acts in the neogenesis in fasting and diabetes. Nature 485(7396):128–132. – cytoplasm to maintain ES cell pluripotency. Genes Dev 24(24):2772 2777. 23. Pavlidis P, Noble WS (2003) Matrix2png: A utility for visualizing matrix data. Bio- 11. Bao X, et al. (2013) Overexpression PRMT5 promotes tumor cell growth and is as- informatics 19(2):295–296. sociated with poor disease prognosis in epithelial ovarian cancer. JHistochemCy- 24. Butler JS, Zurita-Lopez CI, Clarke SG, Bedford MT, Dent SY (2011) Protein-arginine tochem 61(3):206–217. methyltransferase 1 (PRMT1) methylates Ash2L, a shared component of mamma- 12. Migliori V, et al. (2012) Symmetric dimethylation of H3R2 is a newly identified lian histone H3K4 methyltransferase complexes. JBiolChem286(14):12234–12244. histone mark that supports euchromatin maintenance. Nat Struct Mol Biol 19(2): 25. Yamasaki T, et al. (2007) Assays of nucleosome assembly and the inhibition of histone 136–144. acetyltransferase activity. (11) Digestion of chromatin; and (12) purification and 13. Wysocka J, et al. (2005) WDR5 associates with histone H3 methylated at K4 and is characterization of DNA after digestion of chromatin. Protocol Exchange, 10.1038/ essential for H3 K4 methylation and vertebrate development. Cell 121(6):859–872. nprot.2007.340.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1304602110 Tsai et al. Downloaded by guest on September 25, 2021