A coactivator trap identifies NONO (p54nrb) as acomponent of the cAMP-signaling pathwayAntonio L. Amelio*, Loren J. Miraglia†, Juliana J. Conkright‡, Becky A. Mercer*, Serge Batalov†, Valerie Cavett§,Anthony P. Orth†, Jennifer Busby§, John B. Hogenesch¶�, and Michael D. Conkright*�

Departments of *Cancer Biology and §Molecular Therapeutics and ‡Translational Research Institute, The Scripps Research Institute, 5353 Parkside Drive,Jupiter, FL 33458; †Genomics Institute of the Novartis Research Foundation, 10675 John Jay Hopkins Drive, San Diego, CA 92121; and ¶Department ofPharmacology, Institute of Translational Medicine and Therapeutics, University of Pennsylvania, 3451 Walnut Street, Philadelphia, PA 19104

Edited by Joseph S. Takahashi, Northwestern University, Evanston, IL, and approved November 2, 2007 (received for review August 24, 2007)

Signal transduction pathways often use a transcriptional compo-nent to mediate adaptive cellular responses. Coactivator proteinsfunction prominently in these pathways as the conduit to the basictranscriptional machinery. Here we present a high-throughputcell-based screening strategy, termed the ‘‘coactivator trap,’’ tostudy the functional interactions of coactivators with transcriptionfactors. We applied this strategy to the cAMP signaling pathway,which utilizes two families of coactivators, the cAMP responseelement binding protein (CREB) binding protein (CBP)/p300 familyand the recently identified transducers of regulated CREB activityfamily (TORCs1–3). In addition to identifying numerous knowninteractions of these coactivators, this analysis identified NONO(p54nrb) as a TORC-interacting protein. RNA interference experi-ments demonstrate that NONO is necessary for cAMP-dependentactivation of CREB target genes in vivo. Furthermore, TORC2 andNONO complex on cAMP-responsive promoters, and NONO acts asa bridge between the CREB/TORC complex and RNA polymerase II.These data demonstrate the utility of the coactivator trap byidentification of a component of cAMP-mediated transcription.

transcription � signal transduction � cell-based screen � RNA polymeraseII � transducer of regulated cAMP response element-binding protein

Transcription is regulated by large multisubunit complexes thatcan be grouped into three general categories; DNA binding

proteins, coregulators (coactivators and corepressors), and basaltranscriptional components. DNA binding proteins recognize dis-crete sequences or response elements within promoters and ingeneral function as scaffolds that direct the recruitment of coregu-latory proteins. Coregulators in turn function as a conduit to thebasic transcriptional machinery. Coregulators may also influencegene expression via intrinsic enzymatic activity or by recruitment ofother enzyme activities (e.g., acetylation, methylation, poly ADP-ribosylation, ubiquitination, sumoylation, or ATP-dependent re-modeling complexes) capable of modifying both transcriptionalproteins and chromatin (1). Thus, determining the interactionnetworks of coregulators recruited by transcription factors and theaccompanying enzymatic activities is necessary for understandingthe complexities of gene expression.

The cAMP signal-transduction pathway activates transcriptionby stimulating interactions between cAMP response elementbinding protein (CREB) and two coactivator families, CREB-binding protein (CBP)/p300 and transducers of regulated CREB(TORCs) (2–4). CREB–CBP/p300 interaction occurs when el-evations in intracellular cAMP liberate protein kinase A (PKA)catalytic subunits (PKAc) from PKA regulatory subunits. PKAcdirectly phosphorylates serine 133 in the kinase-inducible do-main of CREB, increasing the affinity of CBP/p300 for CREB(5, 6). CBP/p300 interacts with components of the RNA poly-merase II (RNA pol II) complex to facilitate transcription andcontains intrinsic acetyltransferase activity speculated to facili-tate transcriptional activation by affecting chromatin structure(7–10).

TORC recruitment to CREB-bound promoters by PKAc isless direct. Under basal conditions, TORC1 and TORC2 arephosphorylated by AMP/SNF kinases (11, 12) and bound by14-3-3 proteins that sequester TORCs in the cytoplasm. As levelsof cAMP rise, PKAc phosphorylates SNF kinases, inhibitingtheir phosphorylation of TORCs. Dephosphorylated TORC1and TORC2 are then released from 14-3-3 proteins, translocatedto the nucleus, and bound to CREB (12, 13). Remarkably, thephosphorylation status of TORCs and cytoplasmic retention by14-3-3 can serve to integrate converging cellular signals. Forexample, hormone and energy-sensing pathways converge onTORC2 phosphorylation to modulate glucose output via CREB-mediated hepatic gene expression (11). Furthermore, in excit-able cells, TORC phosphorylation functions as a coincidencedetector that funnels cAMP and calcium-signaling pathways toCREB-dependent transcription (12).

TORCs function as robust transcriptional activators at cAMP-responsive promoters (3). Ectopic expression of TORCs by-passes normal regulatory mechanisms and activates reportergenes many hundredfold, suggesting that TORCs either haveintrinsic enzymatic activity or recruit catalytic proteins. Recentstudies demonstrate that TORC2 mediates target gene activa-tion in response to cAMP in part by cooperative interactionswith CBP (14, 15); however, CBP/p300 recruitment cannot fullyexplain how TORCs robustly activate transcription. Moreover,the contribution of each of these coactivators varies, dependingon the cAMP-responsive promoter (15).

To better understand the mechanisms by which TORCsinfluence cAMP-mediated transcription, we sought to identify amore complete cohort of functional TORC-interacting tran-scription factors. We developed the ‘‘coactivator trap,’’ a high-throughput screen with a functional readout for interactions.Here we apply our screening method to the cAMP signalingpathway and identify NONO (p54nrb) as a component of cAMPsignaling that interacts with TORCs to coordinate transcriptionby recruiting RNA pol II.

ResultsA Functional Transcription Factor Trap Identifies Proteins Interactingwith TORC Coactivators. The coactivator trap was developed toscreen for interacting partners of TORCs in a mammalian cell

Author contributions: A.L.A., L.J.M., J.J.C., B.A.M., S.B., A.P.O, J.B.H., and M.D.C. designedresearch; A.L.A., L.J.M., J.J.C., B.A.M., V.C., and M.D.C. performed research; L.J.M., J.B.H.,and M.D.C. contributed new reagents/analytic tools; A.L.A., L.J.M., J.J.C., B.A.M., V.C., J.B.,J.B.H., and M.D.C. analyzed data; and A.L.A., J.J.C., B.A.M., J.B.H., and M.D.C. wrote thepaper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

�To whom correspondence may be addressed. E-mail: hogenesc@mail.med.upenn.edu (forquestions regarding the coactivator trap) or conkright@scripps.edu (for questions regard-ing the cAMP pathway).

This article contains supporting information online at www.pnas.org/cgi/content/full/0707999105/DC1.

© 2007 by The National Academy of Sciences of the USA

20314–20319 � PNAS � December 18, 2007 � vol. 104 � no. 51 www.pnas.org�cgi�doi�10.1073�pnas.0707999105

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

environment. The premise for the assay is rooted in earlytechnology developed to screen protein–protein interactions, theyeast and mammalian two-hybrid systems, whereby functionaldomains of the yeast GAL4 transcription factor are used (16–19). We constructed and arrayed a library of 837 sequenced-verified plasmids coding for human transcription-relatedproteins fused to the GAL4 DNA binding domain. In essence,combining our library with previous methodologies allows forsimultaneous interrogation of interactions between hundreds oftranscription factors and coactivators, such as CBP, p300, andTORCs [supporting information (SI) Fig. 5]. For example, whenthe library of heterologous fusion proteins was cotransfectedinto cells with a GAL4 UAS::luciferase reporter in the presenceor absence of expression plasmids for either of the well charac-terized coactivators CBP and p300, many previously identifiedCBP/p300 partners were confirmed. These partners includeSTAT1, ATF4, FOXl1, NFE2, NeuroD1, and NR4A1 (Fig. 1 aand c) (20).

In an effort to determine a more complete cohort ofTORC-interacting transcription factors, we applied our screento TORCs 1–3. TORCs are capable of robust activation of

transcription; however, the transcriptional activity of mostclones in the library was not inf luenced by TORCs (Fig. 1b),whereas proteins known to interact with TORCs, such asCREB, ATF1, and TORC2, were found to have enhancedreporter activity (Fig. 1 a and b). The finding that TORC2 wasactivated by itself was not surprising, because TORCs havebeen shown to oligomerize (3).

To determine hits from the primary assay for later experi-mental validation, we included several positive controls, theweakest of which, Gal4 CREB, generated an average 10-foldchange in transcriptional activity upon cotransfection withTORCs. All hits that exceeded this threshold, therefore, wouldinteract with TORCs at least as strongly as Gal4 CREB. We notethat this method has the possibility of introducing a significantfalse-negative rate and that weak interactors (weaker thanCREB) will be missed. Because the assays were done in dupli-cate or more, we also filtered for those genes that had a Student’st test �0.05 when comparing vector alone (group 1) to theirperformance upon cotransfection with TORC1, TORC 2, orTORC 3 (group 2). To empirically determine the false-positiverate of this method, we reconfirmed each hit meeting our

a b

c

d

Fig. 1. Coactivator trap identifies TORC 1, 2, or 3 functional interactions with NONO (p54nrb). Shown are transient transfection assays with GAL4 UAS::luciferasereporter in HEK293T cells cotransfected with GAL4-cDNA library. (a) Heat map representation of transcriptional activities increased (red) or repressed (green)at least 10-fold by any TORC compared with pCMV-SPORT6 vector control. For comparison, the coactivators CBP and p300 expression constructs are also mapped.(b) Changes in transcriptional activity of GAL4 fusion proteins induced by cotransfection with TORCs 1, 2, or 3 plotted against a sorted rank of transcriptionalactivity of fusion constructs from left to right. (c) Changes in transcriptional activity of GAL4 fusion proteins induced by cotransfection with either p300 or CBPplotted against a sorted rank of transcriptional activity of fusion constructs. (d) Transient transfection assays with GAL4 UAS::luciferase reporter in HEK293T cellscotransfected with GAL4-NONO and indicated coactivators (n � 3 wells, mean � SEM).

Amelio et al. PNAS � December 18, 2007 � vol. 104 � no. 51 � 20315

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

filtering criteria in secondary assays. Approximately 90% of theobserved interactions were confirmed in independent assays,indicating that this method generates an �10% false positiverate. By using this criteria, several factors were found to bestrongly activated in the coactivator trap by TORCs. NONO(p54nrb) was of particular interest because a previous studyimplicated NONO in cAMP signaling (21). Furthermore, weanalyzed immunoprecipitated TORC2 cellular protein com-plexes in a parallel study by using nano-liquid chromatographycoupled to tandem mass spectrometry (nLC-MS/MS). This studyalso revealed interactions of NONO protein with TORC2 afterstimulation with the adenylate cyclase agonist forskolin, whichelevates intracellular cAMP levels (SI Fig. 6).

cAMP Signaling Stimulates TORC–NONO Complex Formation. Weperformed endogenous coimmunoprecipitation experiments fo-cusing on TORC2, which is natively expressed in HEK293T cells.Lysates from cells stimulated with forskolin or DMSO vehicle asa control were immunoprecipitated with three different anti-bodies that recognize endogenous TORC2. Complete immu-nodepletion of TORC2 from cell lysates resulted in consistentrecovery of cellular NONO in a TORC2 containing complex,

whereas a non-TORC interacting factor, NF�B/p65, was notpresent in the TORC2 immunoprecipitate. Moreover, forskolinstimulation resulted in a marked increase in NONO recovery(Fig. 2a).

The formation of an endogenous TORC–NONO complexprompted us to explore whether TORC2 and NONO werecolocalized in the cell. Under basal conditions, immunocyto-chemistry of endogenous TORC2 and NONO demonstratedlittle colocalization. However, elevated cAMP levels stimulatedby forskolin treatment resulted in nuclear colocalization ofendogenous proteins (SI Fig. 7). Similar to the endogenousproteins, GFP–NONO was present exclusively in the nucleusunder basal conditions and exhibited a marginal 27 � 8% overlapwith CFP–TORC on the basis of pixel area quantification. Incontrast, elevations in cAMP produced a robust 80 � 8% overlap(P � 0.05) between the two proteins (Fig. 2b). Forster FRET wasperformed to quantify NONO and TORC molecular proximityin response to forskolin. Quantitation of the FRET/GFP fluo-rescence ratio revealed that forskolin induced a 8.3-fold increasein the transfer of energy from the donor to the photobleachedacceptor (0.09 � 0.05 control vs. 0.71 � 0.14 forskolin), dem-onstrating an interaction between TORC and NONO (Fig. 2c).Moreover, measurement of this interaction revealed that fors-kolin stimulation brought the proteins into close proximity(2.5 � 0.08 nm), whereas unstimulated cells were near the upperlimits of our detection (10 nm) (Fig. 2c). Collectively, theseresults demonstrate that cAMP stimulates the formation of anuclear complex between TORC and NONO.

NONO Plays Essential Roles in cAMP-Dependent Transcription. IfNONO is a critical component of the CREB–TORC complex,then depletion of NONO should effectively silence TORC-dependent gene activation and, consequently, cAMP signaling.To examine the effects of NONO knockdown, four siRNAoligonucleotides targeting NONO were each tested for theirability to block cAMP activation of the CRE-containing EVX1promoter (SI Fig. 8) (22). The four siRNA molecules targetingNONO varied in their efficacy to block forskolin induction,which strongly correlated (Pearson correlation, r � 0.77) withdecreases in NONO protein levels (SI Fig. 8). Remarkably, themost potent NONO siRNA blocked cAMP induction at levelssimilar to siRNA molecules targeting CREB and TORC2 (Fig.3). None of the siRNAs blocked TNF�-mediated activation ofthe IFN promoter, demonstrating specificity of the NONOsiRNA (Fig. 3). NONO protein depletion was confirmed byWestern blot analysis (Fig. 3). Therefore, NONO is necessary forcAMP-dependent transcriptional activation.

NONO Is Recruited to cAMP-Responsive Promoters by TORC2 andTethers TORC2 to RNA Pol II. The identification of a TORC–NONOnuclear complex and our RNA interference data identifyingNONO as a necessary component for cAMP-signaling promptedus to assess whether NONO occupied endogenous cAMP-responsive promoters and whether this occupation was forskolin-dependent. To determine whether NONO was present at en-dogenous cAMP-dependent promoters, we used ChIP assayswith antibodies directed against CREB, TORC2, and NONO.Like CREB and TORC2, NONO was detected on the endoge-nous CRE-containing NR4A2 promoter in the presence offorskolin (Fig. 4a). Multiple independent ChIP experimentswere conducted and analyzed by real-time PCR to confirmresults obtained by standard PCR (Fig. 4b). Based on theseresults we hypothesized that disrupting interactions betweenTORCs and NONO should impair the regulation of endogenousCREB target genes. Indeed, knockdown of NONO attenuatedthe induction of NR4A2 and FOS mRNA in response to fors-kolin, comparable to effects observed with CREB and TORC2siRNAs (Fig. 4c). However, depletion of NONO had no effect

Fig. 2. cAMP signaling stimulates TORC–NONO complex formation. (a) (Left)Endogenous NONO coimmunoprecipitates with TORC2. TORC2 was immuno-precipitated by using affinity-purified 638A, 3363, or 3364 �-TORC2 antibody(TORC-IP) or beads alone (Control-IP) from cell lysates stimulated with fors-kolin (�) or vehicle (�) and were immunoblotted (IB) with �-NONO or �-p65/NF�B. (Right) Pre-IP of NONO and p65/NF�B protein (2% of input) is shown. (b)(Left) FRET analysis of interaction of TORC and NONO. Shown are represen-tative pseudocolored fluorescence images of cells cotransfected with CFP-TORC and GFP-NONO expression plasmids and treated with DMSO (control) orforskolin and IBMX for 1 h. (Scale bar, 10 �m.) (Right) Percentage of overlapin fluorescence energy of CFP–TORC and GFP–NONO expression plasmids incontrol and forskolin-treated cells (n � 5 experiments; mean � SEM; eachexperiment was the average of 10 measurements). (c) (Left) FRET/GFP ratio ascaptured in pseudocolored fluorescence images of transfected cells aftercontrol and forskolin treatment for 1 h. (Scale bars, 10 �m.) Shown are FRETefficiency (Center) and the distance between CFP donor and GFP acceptor(Right) between CFP–TORC and GFP–NONO in control and forskolin-treatedcells (n � 5 experiments; mean � SEM). Data are from at least five regions ofinterest (ROI) per treatment group from three independent experiments.

20316 � www.pnas.org�cgi�doi�10.1073�pnas.0707999105 Amelio et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

on YWHAH, a constitutively active gene not responsive toelevations in intracellular cAMP.

Previous studies demonstrated that NONO can bind to theC-terminal domain (CTD) of RNA pol II (23, 24). Therefore,TORC2–NONO interactions may represent uncharacterizedconduits from TORCs to the basic transcriptional machinery.ChIP assays were performed to determine whether NONObridges the CREB–TORC complex to the CTD of RNA pol IIby using antibodies that recognize the CTD of RNA pol IIregardless of its phosphorylation status. Two endogenous CRE-containing promoters, NR4A2 and FOS, demonstrated increasedRNA pol II occupancy after forskolin stimulation (Fig. 4d).Moreover, knockdown of NONO selectively impaired cAMP-dependent recruitment of RNA pol II (Fig. 4d), whereas aconstitutively active gene, YWHAH, did not have an attenuationof RNA pol II recruitment after NONO knockdown. Collec-tively, these data demonstrate NONO-dependent recruitment ofRNA pol II to cAMP-regulated promoters.

DiscussionThe Coactivator Trap Identifies New cAMP Pathway Components.TORC coactivators are integral components of the cAMPsignaling pathway and regulate numerous biological processes,including hepatic gluconeogenesis, adaptive mitochondrial bio-genesis, and long-term synaptic plasticity (11, 25, 26). Theidentification of TORC-interacting transcription factors shedslight on the mechanisms used by TORCs to influence thesediverse biological processes. To this end, we developed andapplied the coactivator trap assay to identify interacting partnersof the TORC family of CREB coactivators.

Fig. 3. NONO plays essential roles in cAMP-dependent transcription. Shownis the effect of NONO knockdown by siRNA on the cAMP-responsive EVX1luciferase (Top) versus control IFN luciferase (Middle) reporters after forskolinor TNF� stimulation, respectively, in HEK293T cells. (Bottom) Western blotshowing amounts of protein in cells treated with siRNA (n � 3 wells; mean �SEM).

Fig. 4. NONO recruits RNA pol II to cAMP-responsive promoters. NONO ispresent at a cAMP-responsive promoter along with CREB and TORC2. (a) ChIPof the NR4A2 and GAPDH promoters from HEK293T cells using anti-CREB(Left), anti-TORC2 (Center), anti-NONO (Right), or anti-GAL4-specific antiseraas a negative control. Preimmunoprecipitation (Pre-IP) control DNA is alsoshown. (b) Quantification of precipitated NR4A2 promoter by TaqMan real-time (qRT) PCR (n � 3 experiments; mean � SEM; each experiment was theaverage of three measurements). Occupancy of the target protein on theNR4A2 promoter is expressed relative to the GAPDH promoter. (c) RNA fromHEK293T cells transfected with siRNAs for CREB, TORC2, NONO, or NS controlwas quantified by quantitative RT-PCR using YWHAH, NR4A2, and FOS mRNA-specific TaqMan probes. Fold changes in endogenous YWHAH, NR4A2, andFOS mRNA levels after treatment with forskolin or DMSO vehicle for 45 minare graphed. NONO siRNA abolishes cAMP-dependent transcriptional activa-tion but not constitutive activity of YWHAH (n � 3 wells; mean � SEM). (d)Knockdown of NONO protein attenuates RNA pol II recruitment to cAMP-responsive promoters. ChIP assay of YWHAH, NR4A2, FOS, and GAPDH pro-moters from HEK293T cells using anti-RNA pol II antibody. Fold occupancy ofRNA pol II on the target promoter is expressed relative to the GAPDH promoter(n � 3 experiments; mean � SEM).

Amelio et al. PNAS � December 18, 2007 � vol. 104 � no. 51 � 20317

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

Although the coactivator trap uncovered many putative interac-tions, we chose to investigate NONO (p54nrb) because it wasstrongly activated by TORCs1–3, and an independent experimentalmethodology, mass spectrometry of coimmunoprecipitated pro-teins, also identified NONO (p54nrb) as a TORC-interactingpartner. It has been previously demonstrated that NONO is amultifunctional protein implicated in transcription and RNA pro-cessing. With respect to transcriptional regulation, NONO waspreviously shown to bind to thyroid hormone receptors (TR) andretinoid X receptors (RXR) (27) and the Spi-1/PU.1 transcriptionfactor (28) to regulate transcription from their respective targetpromoters. Moreover, the possible roles for NONO in cAMP-dependent regulation have been previously suggested. Specifically,NONO is recruited to the cytochrome p450 hCYP17 promoter bythe steroidogenic factor-1 (gene symbol NR5A1) in response toelevations in intracellular cAMP (21), but it is unclear whetherTORCs or CREB are part of this complex. Therefore, we hypoth-esized that NONO may play a role in cAMP signaling throughinteractions with TORCs.

Our studies demonstrate that NONO is an integral componentof the cAMP pathway that tethers TORCs to RNA pol II toregulate cAMP-responsive genes. First, we showed that endog-enous NONO forms complexes with TORC2 in response tocAMP and together they are assembled on endogenous CRE-containing promoters. Moreover, RNA interference experi-ments showed that NONO is required for cAMP-dependentactivation of CREB target genes in vivo. Our observation thatNONO facilitates the recruitment of RNA pol II to cAMP-responsive genes supports previous studies indicating thatNONO mediates transcription by providing a direct physical linkto the RNA pol II CTD (23, 24). Collectively, the data supporta mechanism by which NONO is recruited to cAMP-dependentpromoters by TORCs and subsequently couples TORC proteinsto the basal transcriptional machinery, thereby enabling tran-scriptional responses to elevations in intracellular cAMP. Al-though previous studies also demonstrate that NONO binding tothe RNA pol II CTD cotranscriptionally regulates premRNAprocessing, it remains to be determined whether TORC:NONOinteractions regulate cotranscriptional processing of cAMP-dependent transcripts.

Development of the Coactivator Trap. Coactivator proteins encom-pass a subset of transcriptional modulators whose actions aremediated by DNA binding proteins. For example, the coactiva-tors p300 and CBP modulate transcriptional responses to cAMPby their CREB-dependent recruitment to promoters (2). How-ever, there are also dozens of other trans factors, including Clock,HIF1�, STAT1, ATF4, FOXO1, NFE2, Neuro D, NF-Y, andmembers of the nuclear hormone receptor family that are ableto bind CBP/p300 (20). Most of these interactions were discov-ered serendipitously or by trial and error, because each factorbinds its own distinct regulatory element. To date, a genericsystem has not been available to investigate these issues. Toovercome this limitation and to parallelize the process of iden-tification of functional transcription factor/coactivator interac-tions, we devised the coactivator trap. We reasoned thatcoactivators would stimulate the transcription of a distinctcohort of these transcription factors and that this screen couldbe used to determine transcription factor/coactivator pairs.

Although, the coactivator trap assay has many benefits, suchas interrogation of transcription factors for interactions on amassively parallel scale, some bona fide protein–protein inter-actions may not function properly in this context. Although thistechnique offers a method to efficiently interrogate and identifyinteracting transcription factors on a large, parallel scale, theassay does have inherent limitations. To account for falsepositives, putative interactions have to be experimentally vali-dated using standard biochemical procedures. False-negative

errors could potentially arise from poor expression of the fusionprotein or because of missing cofactors not present in themammalian screening line. Along the lines of the latter point,many coactivators or inhibitors function by recruiting DNA-modifying enzymes, such as HATs and HDACs, that will notaffect a luciferase plasmid and therefore will not show in theassay. Therefore, although the coactivator trap represents apowerful method to identify transcription factor/coactivatorinteractions, caution should be used in interpreting the primaryscreening results, especially in regard to null observations.

Experimental ProceduresAmplification, Purification and Cloning of Transcription Factors. Human cloneswere identified by surveying the expert annotation systems, such as Panther(29) and GO annotation (by �20 categories) at the National Center forBiotechnology Information (NCBI) EntrezGene (LocusLink in 2003), combinedwith the filtered keyword searches over the NCBI annotation. The resulting listof 4,083 human Celera and public transcripts and transcript predictions wasfollowed by a similarity search in the available clone sequences by BLAST (30).We were able to select 1,428 human transcription factor clones from theMammalian Gene Collection (MGC), and, of these, we were able to success-fully subclone and sequence-verify 837 clones. The CM-GAL4 expression vectorwas created using standard cloning. Purified plasmids were normalized andspotted into 384-well plates at a concentration of 10 ng per well.

Generation of Expression Plasmids for FRET Microscopy. The C-terminallytagged expression plasmid pECFP-TORC1 was generated by PCR amplificationof the entire human TORC1-coding region by using published primer se-quences (13) and was cloned into the BamHI and BglII sites of pECFP-N1(Clontech).

High-Throughput Transfection and Reporter Assay. HEK293T cells were culturedin DMEM (Invitrogen) supplemented with 10% FBS and antibiotics (100units/ml penicillin and 100 �g/ml streptomycin). Reverse transfection wascarried out using the arrayed transcription factor library collection containing10 ng of transcription factor CM-GAL4 fusion cDNA per well. Serum-freeDMEM (20 �l), containing test cDNA (cDNAs encoding TORC1, TORC2, orTORC3), the reporter GAL4::lucifease, and Fugene 6 (Roche Diagnostics) wasallocated into each well. After a 30-min incubation at room-temperatureDMEM 20% FBS (20 �l) containing 104 293T cells was dispensed into each well.Cells were cultured for 24 h in a humidified incubator at 37°C in 5% CO2.BrightGlo (Promega) reagent (35 �l) was added to each well, and luciferaseluminescence was measured with an Acquest plate reader (LJL Biosystems).

RNA Interference. RNA interferance and luciferase activity assays were per-formed as described in ref. 22, normalizing to activity from Rous sarcoma virus(RSV)-�-galactosidase expression plasmid. Where indicated, 20 nM siRNA, 100ng of reporter vector, and 25 ng of RSV-�-galactosidase expression plasmidper well were cotransfected. Seventy-two h after transfection, cells weretreated for 4 h with 10 �M forskolin or vehicle (DMSO).

Real-Time PCR. RNA was extracted from 293T cells with TRIzol (Invitrogen), andreverse transcription was conducted with SuperScript III reverse transcriptase(Invitrogen), following the manufacturer’s protocol. PCR amplification reac-tions were run in triplicate in 96-well optical microplates in an ABI Prism 7900HT SDS instrument (Applied Biosystems). Differences in mRNA expressionlevels between the nonspecific and gene-specific siRNA-treated cells duringcontrol and forskolin stimulation were measured by relative quantificationwith the Comparative ��Ct method (Applied Biosystems) and were normal-ized against GAPDH levels. Probe sequences are listed in SI Table 1.

Antibodies. Antiserum 3363 and 3364 against human TORC2 was generatedusing peptide [H]-CAETDKTLSKQSWDSKKAG-[NH2] at Covance. Other anti-bodies included NONO (Bethyl), CREB (Upstate Biotechnology), monoclonalanti-NONO (Upstate Biotechnology), anti-RNA polymerase II (Upstate Bio-technology), anti-GAL4 (Santa Cruz Biotechnology), and tubulin (Sigma).

Coimmunoprecipitation Assays. To coimmunoprecipitate endogenous pro-teins, 293T cells were stimulated with 10 �M forskolin in culture medium for20 min and cells were washed with cold PBS and lysed with 1 ml of cold 1%Nonidet P-40 lysis buffer for 30 min at 4°C. Lysates were pelleted 10,000 � gfor 20 min at 4°C, and the supernatant was incubated with protein G Sepha-rose beads (East Coast Biologicals) for 2 h at 4°C. Precleared lysates were

20318 � www.pnas.org�cgi�doi�10.1073�pnas.0707999105 Amelio et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0

incubated with polyclonal anti-TORC2 antibodies chemically coupled to pro-tein G Sepharose (31) or Sepharose beads alone as a control, with rockingovernight at 4°C. Beads were washed three times with 1 ml of cold lysis buffer,resuspended in 2X sample buffer, and incubated at 42°C for 20 min. Beadswere pelleted at 10,000 � g for 1 min, and �-mercaptoethanol (5% finalconcentration) was added to the eluant. Western blots were performed asdescribed in ref. 22.

Tandem Nano-LC/MS/MS Analysis. The elutes from the immunoprecipitationwere loaded onto an SDS/PAGE gel for protein separation. Each sample lanewas excised for in-gel trypsin digestion via standard methods. After digestion,peptides were analyzed via nano-LC/MS/MS; raw data files were processedthrough an in-house workflow. Briefly, spectral data were extracted, filteredfor quality, and searched via a clustered version of the Sequest search engine(Thermo Fisher). Results were than parsed into Scaffold (Proteome Software)for statistical analysis and manual review of the peptide assignments.

Confocal Microscopy and Immunohistochemistry for TORC2 and NONO. 293Tcells were treated with either DMSO or 100 �M 3-isobutyl-1-methylxanthine(IBMX) with 25 �M forskolin for 1 h. Cells were fixed in 4% paraformaldehydeand permeabilized with 0.2% Triton X-100 in PBS. Cells were blocked in 5%BSA in PBS for 30 min. Coverslips were incubated with anti-TORC2 andanti-NONO antibodies, followed by incubation with Texas red goat anti-rabbitIgG and Alexa Fluor 488 goat anti-mouse IgG1. Nuclei were stained with 2�g/ml DAPI dye (Sigma–Aldrich). Coverslips were mounted onto glass slides,and cells were visualized using an Olympus 1 � 81 confocal laser scanningmicroscope. Images were acquired using Fluoview, version 1.5, imaging soft-ware (Olympus) and were imported using Adobe Photoshop CS2, version9.0.2.

Forster FRET-Based Colocalization of NONO and TORC. A549 cells were usedbecause they have a high ratio of cytoplasm–nuclear area. Cells were tran-siently cotransfected with 100 ng each of pECFP-TORC1 and pEGFP-NONOplasmids by using Lipofectamine 2000. The cells were treated with DMSO or100 �M IBMX plus 25 �M forskolin for 1 h and were fixed with 4% parafor-maldehyde. Direct interactions between TORC1 and NONO were measured bydetecting the energy transfer between pECFP-TORC1 (donor: excitation, 436nm; emission, 488 nm) and pEGFP-NONO (acceptor: excitation, 517 nm; emis-sion, 528 nm) after bleaching of NONO emission. The acceptor photobleach-ing method was performed using an Olympus 1 � 81 confocal laser scanningmicroscope.

To measure the interaction of TORC and NONO, the Acceptor Photobleachcommand was run as described in the Olympus FV-10 software package.Calculations for FRET overlap, efficiency, and distance were measured auto-matically by Olympus software and were imported manually into GraphPadPrism 4.

ChIP. ChIP assays were performed as described in ref. 22. The preimmunopre-cipitation-input sample was purified in a manner similar to the bound ChIPfraction described above. Serial dilutions of genomic 293T DNA were used asreferences to demonstrate linearity of the PCR. The CREB target promoterNR4A2 PCR primers (forward, CCCAAGCTGGCTACCAAGGTGAAC; reverse,GGC CGC CAA TGT GCC TTT GTT TAT) yield a 228-bp amplicon, whereas thenegative control GAPDH PCR primers (forward, CCTTCTTGCCTTGCTCTTGC-TAC; reverse, GCCTGCCTGGTGATAATCTTTG) yield a 192-bp amplicon.

ACKNOWLEDGMENTS. We thank Massimo Caputi, John Cleveland, RobertScreaton, Trey Sato, Kevin Hayes, and Gina Hayes for their insightful com-ments during the preparation of this manuscript. Funding and support wereprovided by the State of Florida and the Novartis Research Foundation.

1. Rosenfeld MG, Lunyak VV, Glass CK (2006) Genes Dev 20:1405–1428.2. Chrivia JC, Kwok RP, Lamb N, Hagiwara M, Montminy MR, Goodman RH (1993) Nature

365:855–859.3. Conkright MD, Canettieri G, Screaton R, Guzman E, Miraglia L, Hogenesch JB, Mont-

miny M (2003) Mol Cell 12:413–423.4. Iourgenko V, Zhang W, Mickanin C, Daly I, Jiang C, Hexham JM, Orth AP, Miraglia L,

Meltzer J, Garza D, et al. (2003) Proc Natl Acad Sci USA 100:12147–12152.5. Gonzalez GA, Montminy MR (1989) Cell 59:675–680.6. Radhakrishnan I, Perez-Alvarado GC, Parker D, Dyson HJ, Montminy MR, Wright PE

(1997) Cell 91:741–752.7. Bannister AJ, Kouzarides T (1996) Nature 384:641–643.8. Kwok RP, Lundblad JR, Chrivia JC, Richards JP, Bachinger HP, Brennan RG, Roberts SG,

Green MR, Goodman RH (1994) Nature 370:223–226.9. Ogryzko VV, Schiltz RL, Russanova V, Howard BH, Nakatani Y (1996) Cell 87:953–959.

10. Yuan W, Condorelli G, Caruso M, Felsani A, Giordano A (1996) J Biol Chem 271:9009–9013.

11. Koo SH, Flechner L, Qi L, Zhang X, Screaton RA, Jeffries S, Hedrick S, Xu W, BoussouarF, Brindle P, et al. (2005) Nature 437:1109–1111.

12. Screaton RA, Conkright MD, Katoh Y, Best JL, Canettieri G, Jeffries S, Guzman E, NiessenS, Yates JR, III, Takemori H, et al. (2004) Cell 119:61–74.

13. Bittinger MA, McWhinnie E, Meltzer J, Iourgenko V, Latario B, Liu X, Chen CH, Song C,Garza D, Labow M (2004) Curr Biol 14:2156–2161.

14. Ravnskjaer K, Kester H, Liu Y, Zhang X, Lee D, Yates JR, III, Montminy M (2007) EMBO J26:2880–2889.

15. Xu W, Kasper LH, Lerach S, Jeevan T, Brindle PK (2007) EMBO J 26:2890–28903.

16. Fields S, Song O (1989) Nature 340:245–246.17. Giniger E, Varnum SM, Ptashne M (1985) Cell 40:767–774.18. Kakidani H, Ptashne M (1988) Cell 52:161–167.19. Sadowski I, Ma J, Triezenberg S, Ptashne M (1988) Nature 335:563–564.20. Goodman RH, Smolik S (2000) Genes Dev 14:1553–1577.21. Sewer MB, Nguyen VQ, Huang CJ, Tucker PW, Kagawa N, Waterman MR (2002)

Endocrinology 143:1280–1290.22. Conkright MD, Guzman E, Flechner L, Su AI, Hogenesch JB, Montminy M (2003) Mol Cell

11:1101–1108.23. Emili A, Shales M, McCracken S, Xie W, Tucker PW, Kobayashi R, Blencowe BJ, Ingles CJ

(2002) RNA 8:1102–1111.24. Rosonina E, Ip JY, Calarco JA, Bakowski MA, Emili A, McCracken S, Tucker P, Ingles CJ,

Blencowe BJ (2005) Mol Cell Biol 25:6734–6746.25. Kovacs KA, Steullet P, Steinmann M, Do KQ, Magistretti PJ, Halfon O, Cardinaux JR

(2007) Proc Natl Acad Sci USA 104:4700–4705.26. Wu Z, Huang X, Feng Y, Handschin C, Feng Y, Gullicksen PS, Bare O, Labow M,

Spiegelman B, Stevenson SC (2006) Proc Natl Acad Sci USA 103:14379–14384.27. Mathur M, Tucker PW, Samuels HH (2001) Mol Cell Biol 21:2298–2311.28. Hallier M, Tavitian A, Moreau-Gachelin F (1996) J Biol Chem 271:11177–11181.29. Thomas PD, Campbell MJ, Kejariwal A, Mi H, Karlak B, Daverman R, Diemer K,

Muruganujan A, Narechania A (2003) Genome Res 13:2129–2141.30. Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ (1997)

Nucleic Acids Res 25:3389–3402.31. Harlow E, Lane D (1988) Antibodies: A Laboratory Manual (Cold Spring Harbor Lab

Press, Cold Spring Harbor, NY).

Amelio et al. PNAS � December 18, 2007 � vol. 104 � no. 51 � 20319

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Nov

embe

r 29

, 202

0