Activation of Myod-Dependent Transcription by Cdk9/Cyclin T2

Activation of MyoD-dependent transcription by cdk9/cyclin T2

Cristiano Simone1,2,3, Peter Stiegler2, Luigi Bagella1,2,4, Bruna Pucci2, Cristiana Bellan1,4, Giulia De Falco1, Antonio De Luca1,5, Ginevra Guanti3, Pier Lorenzo Puri6,7 and Antonio Giordano*,1,2

1Sbarro Institute for Cancer Research and Molecular Medicine, Temple University, Philadelphia, Pennsylvania, PA 19122 USA; 2Department of Pathology, Anatomy and Cell Biology, Thomas Jeerson University, Philadelphia, Pennsylvania, PA 19107 USA; 3Department of Internal Medicine and Public Medicine, Division of Medical Genetics, University of Bari, Bari 70124 Italy; 4Institute of Pathologic Anatomy and Histology, University of Siena, Siena 53100 Italy; 5Institute of Topographical Anatomy, School of Medicine, 2nd University of Naples, Naples 80138 Italy; 6Laboratory of Gene Expression, Fondazione Andrea Cesalpino, University of Rome `La Sapienza', 0061 Rome, Italy; 7Clayton Foundation Laboratories for Peptide Biology, The Salk Institute for Biological Studies, La Jolla, California, CA 92093, USA

Myogenic transcription is repressed in myoblasts by (Grana et al., 1994), which is widely expressed in human serum-activated cyclin-dependent kinases, such as cdk2 and murine tissues with high levels found in terminally and cdk4. Serum withdrawal promotes muscle-speci®c dierentiated cells (De Luca et al., 1997a; Bagella et al., gene expression at least in part by down-regulating the 1998, 2000). The regulatory units of cdk9 are the T activity of these cdks. Unlike the other cdks, cdk9 is not family cyclins (T1, T2a and T2b) and cyclin K (Fu et al., serum- or cell cycle-regulated and is instead involved in the 1999). Cyclin T1 (cycT1) and cyclin T2 (cycT2) mRNAs regulation of transcriptional elongation by phosphorylat- are expressed widely in adult human tissues with higher ing the carboxyl-terminal domain (CTD) of RNA levels in muscle, the spleen, thymus, testes, ovaries and in polymerase II. While ectopic expression of cdk2 together peripheral blood lymphocytes. In HeLa nuclear extracts, with its regulatory subunits (cyclins E and A) inhibits roughly 80% of cdk9 is complexed with cycT1 and 10% myogenic transcription, overproduction of cdk9 and its with cycT2a and cycT2b (Wei et al., 1998; Peng et al., associated cyclin (cyclin T2a) strengthens MyoD-depen- 1998). Cyclin K is ubiquitously expressed in adult mouse dent transcription and stimulates myogenic dierentiation and human tissue, but is most abundant in developing in both MyoD-converted ®broblasts and C2C12 muscle germ cells of the adult testes and ovaries (Edwards et al., cells. Conversely, inhibition of cdk9 activity by a dominant 1998). The elevated levels of cdk9 and its regulatory negative form (cdk9-dn) represses the myogenic program. subunits (cyclins T1 and T2) in terminally dierentiated Cdk9, cyclinT2 and MyoD can be detected in a multimeric cells, together with the fact that cdk9/cyclinT complexes complex in C2C12 cells, with the minimal cdk9-binding are not cell cycle-regulated, distinguish cdk9 from the region of MyoD mapping within 101 ± 161 aa of the other cdks (MacLachlan et al., 1995; De Falco and bHLH region. Finally, cdk9 can phosphorylate MyoD in Giordano, 1998). Moreover, unlike the other cdks, vitro, suggesting the possibility that cdk9/cycT2a regula- which regulate cell cycle progression and phosphorylate tion of muscle dierentiation includes the direct enzymatic histone H1, cdk9 fails to phosphorylate H1. It is instead activity of the kinase on MyoD. implicated in the regulation of transcriptional elongation Oncogene (2002) 21, 4137 ± 4148. doi:10.1038/sj.onc. via phosphorylation of the carboxyl-terminal domain 1205493 (CTD) of RNA polymerase II (Grana et al., 1994; Dahmus, 1996; Marshall et al., 1996; Zhu et al., 1997). A Keywords: cdk9; cyclins; MyoD; muscle dierentiation, kinase activity corresponding to cdk9 has been detected phosphorylation in the HIV Tat-associated kinase (TAK) complex, where cdk9 is required for Tat-dependent stimulation of transcriptional elongation (Yang et al., 1996; Zhu et Introduction al., 1997; Mancebo et al., 1997). CycT1 was the ®rst cyclin discovered to associate with cdk9 thereby forming Cyclin-dependent kinase 9 (cdk9) is a cdk1-related a complex that interacts with HIV-1-Tat protein in the serine/threonine kinase, previously named PITALRE TAK complex (Wei et al., 1998). Human cycT1 markedly stimulates Tat activation, while cycT2a and T2b fail to form a complex with Tat bound to the *Correspondence: A Giordano, Sbarro Institute for Cancer Research transactivation response RNA element (TAR) (Kwak et and Molecular Medicine, College of Science and Technology, Temple al., 1999; Wimmer et al., 1999). These data indicate University, BioLife Science Bldg. Suite 333, 1900 N 12th Street, functional dierences between cycT1 and cycT2. In Philadelphia, Pennsylvania, PA 19122 USA; addition, unlike cyclin T1 and cyclin T2, the targeted E-mail: [email protected] Received 6 December 2001; revised 1 March 2002; accepted 18 recruitment of cyclin K to promoter DNA does not March 2002 stimulate transcription in vivo (Napolitano et al., 2000). Physical and functional interactions between MyoD and cdk9 C Simone et al 4138 In human cells, cdk9 complexes are required for study we show that, unlike other cyclin/cdk complexes, transcription of major histocompatibility class II cdk9/cycT2a levels and activity are not down-regulated (MHCII) genes (Kanazawa et al., 2000) and, in during muscle dierentiation. Overproduction of cdk9/ Drosophila cells, are involved in the regulation of cycT2a enhances MyoD function and promotes myo- many, but not all, genes (Lis et al., 2000). These data genic dierentiation, while inhibition of cdk9 kinase suggest that cdk9/cyclinT complexes are not a general activity by a dominant negative form (cdk9dn) prevents basal factor, but depending on cell types, they could the activation of the myogenic program. The N- represent promoter-speci®c transcriptional elongation terminal region of cdk9 directly interacts with the complexes. This hypothesis is con®rmed by the fact bHLH region of MyoD, allowing the formation of a that TAK/P-TEFb activation exerts a dierent func- complex also containing cycT2a. Finally, our data tion in the dierentiation program of lymphocytes suggest that cdk9-mediated phosphorylation of MyoD compared to that of monocytes/macrophages (Foskett could be one of the mechanisms responsible for cdk9- et al., 2001). Activation of cdk9/cyclin T1 in peripheral dependent stimulation of myogenic transcription. blood lymphocytes (PBLs) involves an increase in mRNA and the levels of both proteins (Garriga et al., 1998); conversely, phorbol myristate acetate (PMA)- Results induced dierentiation of promonocytic cells involves only cyclin T1 protein levels through posttranscrip- The cdk9/cycT2a complex enhances the transcriptional tional modi®cation (Herrmann et al., 1998). In activity of MyoD monocyte dierentiation, cdk9 has a clear and important role; in fact, the overexpression of cdk9dn To establish a functional dierence between cdk9 and in U937 cells renders them sensitive to apoptosis after cell cycle-regulated cdks, we ®rst compared the eect of PMA induction of the dierentiation program (Foskett overproduction of cdk2, along with its regulatory et al., 2001). subunits (cyclins E and A) to that of cdk9/cycT on The dierences between cdk9 and those cdks involved MyoD-dependent transcription. The E-box-speci®c in the regulation of the cell cycle may also re¯ect a 4RE-luciferase reporter (4RE-luc) was transiently dierent function during terminal dierentiation of transfected into NIH3T3 cells and the in¯uence of muscle cells. In this context, the induction of myocyte either cdk2/cyclins E and A or cdk9, cdk9dn and T- dierentiation upon serum withdrawal is accompanied type cyclins on the transactivation potential of MyoD by the down-regulation of cyclins D1, E and A and the was assayed after 48 h in dierentiation medium (DM) up-regulation of several cdk inhibitors, which lock the (2% horse serum DMEM) (Figure 1a). Co-transfection cyclin-cdk complexes in an inactive form. This allows of either cyclin E-cdk2 or cyclin A-cdk2 reduced the exit from the cell cycle and the expression of muscle- activation of 4RE by MyoD (Figure 1a). Cdk9 and speci®c genes (Walsh and Perlman, 1997). Transcription cdk9dn alone and with cycT1, cycT2a and cycT2b did of genes that contribute to the phenotypic and not aect the basal transcriptional level of the reporter biochemical identity of dierentiated muscle cells is (Figure 1a). When co-transfected with MyoD, only the activated by myogenic bHLH transcription factors, addition of cdk9 and cycT2a enhanced MyoD- including MyoD, Myf5, myogenin and MRF4 (Wei and dependent transcription. Conversely, ectopic expression Paterson, 2001). These proteins operate by hetero- of a kinase-defective form of cdk9 with the property of dimerizing with the E2A gene products via the HLH a dominant negative (cdk9dn) inhibited MyoD trans- domain and by binding to certain recognition sites (E- activation (Figure 1a). By using the muscle creatine box) in the regulatory regions of muscle-speci®c genes kinase luciferase (MCK-luc) reporter, we con®rmed the (Murre et al., 1989; Lassar et al., 1991; Weintraub et al., ability of the combination of cdk9 and cycT2a to 1991). Overexpression of cyclin/cdks has been reported enhance MyoD-dependent transcription (Figure 1b). to inhibit the activity of MyoD and prevent myogenic Again, we found that the dominant negative form of dierentiation by dierent modalities (Puri and Sartor- the kinase strongly down-regulated MyoD-dependent elli, 2000). For instance, overproduction of cyclins E myogenic transcription (Figure 1b). Similar results and A, together with their associated kinase, cdk2, were also observed using a Gal4MyoD fusion protein impinges on MyoD function by virtue of their ability to whose activity was scored by a Gal4-luc reporter phosphorylate pRb (Skapek et al., 1996; Guo and (Figure 1c). In this assay the transcriptional activity of Walsh, 1997), an essential co-factor for MyoD- MyoD is not in¯uenced by changes in DNA binding, dependent transcription (Gu et al., 1993; Novitch et indicating that cdk9/cycT2a can stimulate MyoD al., 1996). In addition, cdk2 (and the cyclin B-associated function without altering its binding to the DNA. kinase, cdk1) can inactivate MyoD function by Gal4-MyoD was not transactivated by cdk9 alone or phosphorylating serine 200, thereby triggering MyoD by cdk9 with cycT1 or cycT2b; however, co-transfec- degradation (Song et al., 1998; Kitzmann et al., 1999). tion with cdk9 and cycT2a stimulated Gal4-MyoD Finally, cyclin D1 overexpression inhibits myogenic activity (Figure 1c). Also in this setting, we found that dierentiation by promoting the nuclear accumulation the expression of cdk9dn down-regulates the activity of of cdk4, which in turn binds the C-terminus of MyoD MyoD (Figure 1c). To test the speci®city of the eect and prevents MyoD DNA binding and activation of of cdk9/cycT2a on MyoD-mediated transcription, we myogenic transcription (Zhang et al., 1999a,b). In this used Gal4-VP16 and the Gal4-luc reporter. The same

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4139

Figure 1 Activation of MyoD-dependent transcription and myogenic conversion by cdk9-cycT2A. (a) Activation of the MyoD- dependent promoter 4RE-luc was determined in NIH3T3 cells transfected with expression vectors coding for dierent combinations of cyclins/cdks, as reported in the ®gure, transferred in DM (2% horse serum) and harvested after 48 h. The luciferase activity was measured and normalized to the expression of cotransfected b-gal-encoding plasmid and the levels of overexpressed proteins. The data presented are representative of ®ve experiments and standard deviation was calculated. (Amount of DNA transfected: 4RE-luc 0.3 mg; MyoDwt 0.3 mg; cdk2, cdk9, cdk9dn, cycE, cycA, cycT1, cycT2a and cycT2b 0.5 mg; b-gal 0.3 mg) (b) 10T1/2 ®broblasts were transfected with dierent combinations of cdk9 and its associated cyclins, transferred in DM and harvested after 48 h. The luciferase assay was performed in triplicate and normalized to the expression of co-transfected b-gal-encoding plasmid and the levels of overexpressed proteins. The data presented are representative of ®ve experiments and standard deviation was calculated. (Amount of DNA transfected: MCK-luc 0.3 mg; MyoDwt 0.3 mg; cdk9, cdk9dn, cycT1, cycT2a and cycT2b 0.5 mg; b-gal 0.3 mg). (c) Gal4-MyoDwt fusion protein, cdk9, cdk9dn and the T-family cyclins were co-transfected with a Gal4-luc reporter in NIH3T3 cells in triplicate. The cells were transfected in GM, transferred in DM and harvested after 48 h. The luciferase activity was measured and normalized to the expression of co-transfected b-gal-encoding plasmid. The data presented are representative of ®ve experiments and standard deviation was calculated. (Amount of DNA transfected: Gal4-luc 0.3 mg; Gal4-MyoDwt, 0.3 mg; cdk9, cdk9dn, cycT1, cycT2a and cycT2b 0.5 mg; b-gal 0.3 mg). Seventy per cent con¯uent 10T1/2 cells were transfected with MLC-b-gal vector and induced to dierentiate for 48 h. Cells were ®xed and stained following the b-galactosidase standard protocol. The vector was co- transfected with cdk9/cycT2a alone, with MyoDwt alone, with MyoDwt plus cdk9/cycT2a and with MyoDwt plus cdk9dn/cycT2a. (d) 10T1/2 cells were converted into myogenic cells by ectopic expression of MyoD and the eect of co-expression of either cdk9- cycT2A or cdk9dn-cycT2A was evaluated. The cells were co-transfected with a reporter expressing b-galactosidase (b-gal) fused to a nuclear localization signal and under the control of MLC promoter (MLC-b-gal) and the extent of myogenic conversion was scored by visualizing cells with more than one b-gal-positive nucleus. Cells were ®xed and stained following the b-galactosidase standard protocol. The percentage of multinucleated myotubes with b-gal-positive nuclei on the total number of b-gal-positive cells was calculated and is shown as bars. Multinucleation was considered as single cell with more than two b-gal-positive nuclei. (e) Normal light microscope (206). The ®gures presented are representative of three similar experiments. (Amount of DNA transfected: MLC- gal 0.5 mg; MyoDwt 0.5 mg; cdk9, cdk9dn and cycT2a 0.5 mg) combination as indicated in Figure 1c was used and we localized in the nucleus, under the control of the found that all T family cyclins were able to cooperate myosin light chain promoter (MLC-b-gal), which is with cdk9 to activate transcription (data not shown). transactivated during myogenic dierentiation The results obtained are in accordance with data (McGrew et al., 1996). These cells were co-transfected previously reported (Peng et al., 1998). Next, we with MyoD, cdk9, cdk9dn and cycT2a expression performed a myogenic conversion assay using the vectors in various combinations. As shown in Figure ability of MyoD to convert dierent cell lines into 1d,e, cdk9/cycT2a was capable of enhancing MyoD- multinucleated myotubes. 10T1/2 cells were transiently mediated conversion, while cdk9/cycT2b and cdk9/ transfected with a reporter expressing b-galactosidase cycT1 do not (data not shown). Cdk9/cycT2a alone

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4140 was unable to convert 10T1/2 cells into myotubes and C2C12 dierentiation by establishing C2C12 cell clones to transactivate the MLC reporter, which gives only overexpressing EGFP alone, EGFP-cdk9 and EGFP- background levels. Finally, cdk9dn/cycT2a inhibited cdk9dn (referred to as C2-EGFP, C2-EGFP-cdk9 and MyoD-mediated conversion of ®broblasts into multi- C2-EGFP-cdk9dn, respectively). The impaired activity nucleated myotubes. of endogenous cdk9 in C2C12 cells ectopically expressing cdk9dn (the C2-EGFP-cdk9dn clones) was veri®ed by kinase assay (see Figure 5b). The cells were The kinase activity of cdk9 is necessary during muscle analysed during dierentiation by monitoring three differentiation dierent markers of dierentiation: (1) morphology, (2) Based on the results reported in the preceding section, expression of muscle-speci®c genes and (3) transcrip- we hypothesized that the kinase activity of cdk9 could tional activity of a myogenic reporter. have a relevant function during myogenic dierentia- The cells were plated onto microscope slides and tion. We addressed the functional impact of cdk9 on induced to dierentiate for 5 days. We found striking

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4141 dierences between these three cell lines. The C2- dierent time points during 5 days of culture in DM EGFP-cdk9 cells started to form multinucleated (Figure 2c). The cell extracts were probed with an anti- myotubes earlier than the mock control (C2-EGFP) MHC antibody. MHC levels were signi®cantly higher (Figure 2a, middle panels), whereas the C2-EGFP- in C2-EGFP-cdk9 cells than C2-EGFP cells (Figure cdk9dn cells failed to form myotubes at all (Figure 2a, 2c). Again, overexpression of cdk9dn inhibited the right panels). The C2-EGFP-cdk9dn cells, when expression of MHC during dierentiation (Figure 2c). induced with DM, were unable to undergo the In order to further evaluate the impact of cdk9 on the dierentiation program and became sensitive to cell myogenic program, we transfected the three stable cell death (data not shown), as was demonstrated for U937 lines with a muscle-speci®c luciferase reporter to promonocytic cells after PMA induction (Foskett et al., evaluate its activity during muscle dierentiation. We 2001). To con®rm the morphological data, we analysed utilized the muscle creatine kinase promoter fused to the expression of early and late dierentiation markers the luciferase gene (MCK-luc). The activation of by immunoblot analysis. Myogenin was considered as MCK-luc was higher in C2-EGFP-cdk9 cells than an early marker and MHC as a late marker for muscle C2-EGFP cells, while the C2-EGFP-cdk9dn cells dierentiation. Extracts from the three stable cell lines always maintained the basal level (Figure 2d). These cultured in growth medium (GM) or after 6, 12, 24 and sets of experiments strongly support the role of cdk9 in 36 h in DM were collected and examined for the promoting muscle dierentiation. expression of myogenin and MHC. The C2-EGFP- cdk9 cells expressed more abundant levels of myogenin The N-terminus of cdk9 directly interacts in vitro with the than the control C2-EGFP cells (Figure 2b), while bHLH domain of MyoD myogenin was undetectable in the C2-EGFP cdk9dn cells at all points examined (Figure 2b). In a similar The myogenic program in C2C12 is initiated by the experiment we collected cells both in GM and at activation of pre-existing myogenic regulatory proteins,

Figure 2 Overexpression of cdk9 accelerates myogenic program, while its dominant negative form inhibits dierentiation in C2C12 cells. Three stable cell lines C2-EGFP (left panels), C2-EGFP-cdk9 (middle panels) and C2-EGFP-cdk9dn (right panels) were induced to dierentiate. (a) GM shows logarithmically growing C2-EGFP, C2-EGFP-cdk9 and C2-EGFP-cdk9dn cells. The same number of cells was plated and upon reaching 70% con¯uence dierentiation was induced with DMEM 2% HS (DM). The cells were maintained in DM for up to 120 h. The ®rst six rows of panels from the top of the ®gure indicate what was observed under a normal light microscope (206). The three last panels at the bottom of the ®gure indicate what was observed under a ¯uorescent microscope (406). (b) Levels of myogenin were determined by immunoblotting at 0, 6, 12, 24 and 36 h and b-tubulin was used as a loading control. (c) Levels of MHC were determined by immunoblotting at 0, 24, 48, 72, 96 and 120 h and b-tubulin was used as loading control. (d) MCK-luc reporter was transfected into C2-EGFP, C2-EGFP-cdk9 and C2-EGFP-cdk9dn cells. After transfection, the cells were placed in DM and luciferase activity was measured after 0, 24, 48 and 72 h. All the experiments were repeated at least three times. The transfections were performed in triplicate and normalized for b-gal activity

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4142 such as MyoD and Myf5. These bHLH factors are be mediated by a physical interaction with MyoD. To latent in myoblasts and can be activated by placing the test whether a direct physical interaction occurs cells in a serum-deprived medium. Thus, if cdk9 has a between cdk9 and MyoD, we ®rst investigated if the role in the induction of myogenic dierentiation by two proteins could interact in vitro. To this purpose, in serum withdrawal, MyoD could be one possible target vitro translated 35S labeled cdk9, cdk9dn and three of this kinase. Previous studies have reported the deletion mutants, cdk9 (129 ± 372), cdk9 (218 ± 372) and inhibition of MyoD by cdks, either by physical cdk9 (1 ± 195) (Figure 3a), were tested for their ability interaction (as in the case of cdk4) (Zhang et al., to interact with GST-MyoD in a pull-down assay. GST 1999a), or by direct phosphorylation (as in the case of alone was used as a negative control. An interaction cdk2 and cdk1) (Kitzmann et al., 1999; Song et al., was detected with the full-length cdk9, with cdk9dn 1998). Thus, interaction with cell cycle-regulated cdks and with cdk9 (1 ± 195) (Figure 3c,d). The same can provide an inhibitory constraint for MyoD in experiment was performed using wild-type MyoD and myoblasts. In a similar fashion, we reasoned that the three deletion mutants of in vitro translated MyoD. positive eect of cdk9 on myogenic transcription could MyoD (1 ± 100) lacks the bHLH domain and the C-

Figure 3 The cdk9/cycT2a complex directly interacts in vitro with MyoD and the N-terminus of cdk9 and the bHLH domain of MyoD are the minimal binding regions. (a,b) In vitro translated (IVT) proteins and minimal binding region between cdk9 and MyoD. (TD: transactivation domain; b: basic domain; HLH: helix ± loop ± helix domain. 167Asp?Asn is the mutation present in the cdk9dn). (c) Interaction between GST-cdk9 with in vitro translated 35S-MyoD, and between GST-MyoD with IVT-cdk9wt and cdk9dn. The GST-fusion proteins were precipitated by glutathione-agarose anity chromatography and the interaction with the 35S- labeled in vitro translated proteins was visualized by autoradiography. (d) The same experiment was performed as described in c. The interaction between GST-MyoD and dierent IVT fragments of cdk9 was tested. Note that only IVT-cdk9 (1 ± 195) is able to interact with GST-MyoD (lane 9) under stringent conditions (see Materials and methods). (e) The same experiment was performed as described in c and d. The interaction between GST-cdk9 and dierent IVT fragments of MyoD was tested. Note that only IVT- MyoD (1 ± 185) interacts with GST-cdk9 (lane 9) under stringent conditions (see Materials and methods). (f) GST-cdk9 was incubated with HisTag-MyoD full-length, N-terminus, bHLH, C-terminus and subjected to immunoblot analysis (see Materials and methods). (g) 35S-IVT-cdk9 and/or -cycT2a were incubated with His-MyoD fusion protein. The His-tagged protein was pulled down with nickel beads and the associated proteins were detected by autoradiography

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4143 terminus, MyoD (162 ± 318) lacks the N-terminus and three types of T cyclins: T1, T2a and T2b (Figure 4a). the bHLH domain and MyoD (1 ± 185) lacks most of The cycT2-speci®c serum used in the immunoblot was the C-terminus (Figure 3b). Full-length MyoD inter- raised against human cyclin T2. In preliminary acted with GST-cdk9, but not with GST alone used as experiments, we tested the ability of anti-human cyclin a negative control (Figure 3c). The binding also T2 to cross-react with mouse cycT2. The serum detects occurred with MyoD (1 ± 185), whereas the interaction a similar pattern of two bands corresponding to the was undetectable with the other two mutants (Figure two isoforms cycT2a and cycT2b (approximatively 80 3e). These data indicate that the regions mediating the and 88 kD, respectively) in mouse and human extracts. binding between cdk9 and MyoD encompass the In an additional experiment, anti-MyoD antibody was bHLH domain of MyoD (101 ± 161 aa) and the N- able to co-immunoprecipitate cdk9 and cycT2, but not terminus (1 ± 128 aa) of cdk9 (see Figure 3a). To cycT1, con®rming the transcriptional data presented in further de®ne whether the two proteins interact Figure 1 (Figure 4b). directly, or if an unknown protein present in the Because cdk9/cycT2a was the only activator of reticulocyte lysate mediates the binding, we analysed MyoD in functional assays (Figure 1a,b), since cycT2a, the in vitro interaction between bacterially puri®ed alone and in combination with cdk9, interacted with GST-cdk9 and dierent mutants of histidine-tagged MyoD in vitro (Figure 3g), and, ®nally, to con®rm the MyoD (His-MyoD). Full-length His-MyoD and dier- data presented in Figure 4b, we performed immuno- ent His-MyoD deletions (MyoD-C-ter expressing the precipitations (Ips) from proliferating C2C12 cells and carboxyl-terminal domain, MyoD-bHLH the bHLH dierentiated C2C12 cells by using anti-cycT2 anti- domain and MyoD-N-ter the amino terminal domain) body. The associated proteins were detected by were tested for their ability to bind to cdk9. In a pull- Western blot with anti-MyoD, anti-cdk9, anti-cycT2 down assay followed by immunoblot with anti-cdk9, and anti-cycT1 antibodies. As presented in Figure 4c, only full-length MyoD and the bHLH fragment were MyoD, cycT2 and cdk9 could be co-immunoprecipi- able to interact with the GST-cdk9 (Figure 3f). These tated with anti-cycT2 antibody in GM as well as early data con®rm that cdk9 directly interacts with the DM, while neither cycT2 nor cdk9 and MyoD was bHLH domain of MyoD. detected in the Ip with pre-immune rabbit serum. The Using an independent approach we set out to immunoblot analysis of the anti-cycT2 antibody Ip also investigate whether the binding between MyoD and revealed the absence of cycT1 as previously described cdk9 allows for the simultaneous recruitment of cdk9 (Peng et al., 1998). and the cdk9-associated cycT2a, thus making the formation of a multimeric complex possible. We Cdk9 phosphorylates MyoD and its kinase activity peaks performed an in vitro experiment using His-MyoD at 24 h of muscle differentiation and an in vitro translated form of 35S-cdk9 and -cycT2a. In a pull-down assay in which His-MyoD The involvement of cdk9-speci®c kinase activity in was precipitated, cycT2a and cdk9 were both recruited myogenesis and its association with MyoD prompted by MyoD (Figure 3g, lanes 5 and 6). The transcription us to analyse the possibility that MyoD might be a factor pulled down the two proteins together with the substrate for cdk9. Kinase assays were performed same eciency (Figure 3g, lane 7). These data indicate using immunoprecipitated cdk9 complexes of C2C12 that, at least in vitro, MyoD is able to independently cells. Puri®ed His-MyoD was used as a substrate. As bind cdk9 and cycT2a, and that the mutual binding shown in Figure 5a, whole cell extracts from C2C12 between the three proteins allows the formation of a myoblasts were incubated with antibodies against multimeric complex. cdk2 and cdk5, which were used as the positive and negative control, respectively (Kitzmann et al., 1999), and against cdk9. His-MyoD was eciently phos- Cdk9, MyoD and cycT2 form a multimeric complex phorylated by cdk2, but not by cdk5, as previously in vivo reported (Kitzmann et al., 1999). Immunoprecipitated To verify that the interaction described above was cdk9 could phosphorylate His-MyoD as eciently as present in vivo, we performed immunoprecipitations cdk2 (Figure 5a, lanes 3 and 4), indicating that (Ips) from proliferating C2C12 myoblasts growing in MyoD is a good in vitro substrate for cdk9. To GM or induced to dierentiate in DM. Based on the con®rm that MyoD phosphorylation depended on fact that MyoD is activated in the early phase of the cdk9 and not another protein present in the complex, dierentiation program, we collected logarithmically we performed an immunoprecipitation with anti-cdk9 growing cells and cells after 24 h in DM. Whole cell from C2-EGFP and C2-EGFP-cdk9dn cell extracts extracts were immunoprecipitated with anti-cdk9 anti- and a kinase assay with His-MyoD as a substrate. No body and subjected to immunoblotting with anti- signi®cant dierence from the negative control (Ip MyoD antibody. As presented in Figure 4, MyoD anti-NRS) was observed, con®rming that the in vitro and cdk9 could be co-immunoprecipitated with a-cdk9 phosphorylation of MyoD depended only on cdk9 antibody in GM as well as early DM, while neither kinase activity (Figure 5b). A study of the phosphor- cdk9 nor MyoD was detected in the Ip with pre- ylation of MyoD during myogenesis showed that the immune rabbit serum. Further analysis of the a-cdk9 activity of cdk9 towards MyoD peaked at 24 h in antibody immunoprecipitates revealed the presence of DM (Figure 5c). This fact is noteworthy since it

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4144

Figure 4 Cdk9, cycT2 and MyoD co-immunoprecipitate in C2C12 in GM as well as in DM. (a) C2C12 whole cell extracts were precleared and incubated with anity puri®ed anti-cdk9 and pre-immune serum (NRS) (used as a negative control). After extensive washes, the samples were resuspended in LAEMMLI-Buer, split in two parts and loaded respectively on 6% (upper panel) and 10% (lower panel) acrylamide gel. The associated proteins were detected by immunoblot by using anti-cycT1, anti-cycT2, anti-cdk9 and anti-MyoD antibodies. (b) Under the same conditions as previous experiments Ip with anti-MyoD antibody was blotted with anti-cycT1, anti-cycT2, anti-cdk9 and anti-MyoD antibody. (c) Whole cell extracts were precleared and incubated with anti-cycT2 and the pre-immune serum (NRS) (used as a negative control). The samples were split in two parts and loaded respectively on 6% (upper panel) and 8% (lower panel) gel. The associated proteins were detected by immunoblot with anti-cycT1, anti-cycT2, anti- cdk9 and anti-MyoD antibodies

raises the possibility that the phosphorylation of interfere with the stability of MyoD and suggest that MyoD by cdk9 is a positive signal that stimulates serine residues other than serine 200 can be phos- muscle-speci®c gene expression during the early phase phorylated by cdk9. of dierentiation.

Discussion Cdk9 phosphorylation of MyoD does not influence MyoD stability in vivo In this study we have shown that cdk9, in association Cdk1 and cdk2 are able to phosphorylate MyoD on with cyclinT2a, plays an important role in the Ser200, which allows MyoD degradation via the activation of the myogenic program. With the ubiquitin-proteosome pathway (Kitzmann et al., exception of cdk5, the other cdks have been reported 1999). In order to understand if cdk9 could interfere to be functionally down-regulated during muscle with the stability of MyoD, half-life studies were dierentiation and their ectopic activation impedes performed. Whole cell extracts were subjected to the induction of muscle-speci®c transcription (Walsh immunoblot analysis 0, 15, 30, 45, 60, 90 and and Perlman, 1997). Unlike these cdks, cdk9 and its 120 min after protein synthesis was inhibited by regulatory subunit cyclin T2a are not inactivated cycloexamide addition. The MyoD levels were eval- during myotube formation and are instead implicated uated in the three stable C2C12 cell lines (C2-EGFP, in the activation of MyoD. The formation of a C2-EGFP-cdk9 and C2-EGFP-cdk9dn). No signi®cant multimeric complex, containing cdk9/cycT2a and dierences were observed (data not shown). Similar MyoD is detectable in muscle cells both prior to and experiments were performed using extracts of transi- during activation of the dierentiation program. This ently transfected NIH3T3 and 10T1/2 cells. Neither raises the possibility that this complex can be ectopic expression of cdk9 nor cdk9dn had a signi®cant functionally inactivated in myoblasts by inhibitory impact on exogenous MyoD (data not shown). These signals. Alternatively, or in addition, the complex data imply that unlike cdk1 and cdk2, cdk9 does not might be present in myoblasts in a latent form due to

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4145

Figure 5 Cdk9 phosphorylates MyoD in vitro and its kinase activity peaks in the early phase of the dierentiation program. (a) Ips were performed with anti-cdk5, anti-cdk2, anti-cdk9 and with pre-immune serum in growing C2C12. The immunocomplexes were extensively washed, incubated in kinase assay reaction buer with 0.5 mg of dialyzed His-mMyoD and resolved on 10% gel. (b) Ips with anti-cdk9 antibody, and pre-immune rabbit serum, in C2-EGFP and C2-EGFP-cdk9dn, and the kinase assay were performed as previously reported. (c) C2C12 were induced to dierentiate and collected every 24 h. The extracts were immunoprecipitated with anity puri®ed anti-cdk9 and incubated with 0.5 mg of dialyzed HisTag-mMyoD in kinase reaction buer and subjected to autoradiography. The data shown are representative of three independent experiments

the failure of MyoD to form a productive complex phage homeostasis that implies, after a period of a with E2A products and to eciently bind the DNA few days during which they circulate in the blood, (Lassar and MuÈ nsterberg, 1994). In accordance with they undergo dierentiation or apoptosis (Foskett et this possibility, it was necessary to place the cells in al., 2001). After induction of the dierentiation DM in order to achieve the functional enhancement of program, cells in which the cdk9 protein is inactivated MyoD-dependent transcription by cdk9/cycT2a (Fig- cannot continue to dierentiate and they undergo ures 1 and 2). Nevertheless, we cannot exclude that apoptosis. To our knowledge and according to the engagement of cdk9/cycT2a by MyoD might be data present in the literature, the antiapoptotic cdk9 relevant for other still unknown functions of MyoD function appears to be due only to a block in the in myoblasts. dierentiation program by the cdk9dn protein. Experiments concerning overexpression indicate that Finally, the cells in which cdk9 function is impaired upon induction of muscle dierentiation, cdk9/cycT2a try uselessly to dierentiate, causing a switch toward promotes MyoD-dependent transcription and acceler- the apoptosis pathway. ates the myogenic program. Conversely, suppression While a general transactivating eect of cdk9 on of cdk9 activity by the overproduction of a dominant transcription can be predicted by its function as an negative form of cdk9 impairs MyoD activity and RNA polymerase II-activating kinase (Moncebo et al., prevents the expression of muscle-speci®c markers. 1997; Zhu et al., 1997), the role of cdk9 in modulating These results identify cdk9 as an essential kinase promoter-restricted transcription appears to be depen- during the activation of the myogenic program. As dent on the sequence-speci®c factor that recruits cdk9. previously reported, cdk9 catalytic activity is low in In fact, unlike MyoD, B-Myb function is suppressed by promonocytic cells, according to the monocyte/macro- overproduction of cdk9 (De Falco et al., 2000). In

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4146 addition, the cdk9/cyclinT1 complex is able to regulate and in vivo models (Lassar and MuÈ nsterberg, 1994; MHCII genes, but not CD96, CD25 or IL-2 genes Molkentin and Olson, 1996). (Foskett et al., 2001), con®rming a promoter-speci®c In conclusion, this study reveals a role for cdk9, in activity with respect to a general eect on basal association with cyclinT2a, in assisting MyoD during transcription. This may re¯ect an opposite function the activation of muscle-speci®c transcription. Future of cdk9/cycT complexes on distinct promoters, result- studies are awaited to further de®ne the molecular ing in synergistic eects toward a coordinated process, mechanism(s) underlying the myogenic function of such as muscle dierentiation, which requires both cdk9 to verify its biological relevance using in vivo activation and inactivation of a variety of genes. models. Engagement of cdk9/cycT by muscle-speci®c activators (e.g., MyoD-family members) would result in activation of transcription. Recruitment of the cdk9/cycT complex by proteins such as B-Myb, whose activity Materials and methods counteracts myogenic dierentiation, might prevent the transcription of anti-myogenic genes. Cell lines, stable transfections and differentiation assay The importance of cdk9 activity in sustaining Stable C2C12 cells, overexpressing EFGP were generated as MyoD-mediated transcription and myogenic dieren- EGFP-cdk9 and EGFP-cdk9dn, described elsewhere (MacLa- tiation is underscored by experiments of either cdk9 chlan et al., 1998). The stable mix-clones were selected with overexpression or, remarkably, functional inactivation G-418 (0.5 mg/ml; Sigma, MO, USA) avoiding con¯uence at by a cdk9 kinase-defective mutant. In addition, data any time. C2C12, EGFP-C2C12, EGFP-cdk9-C2C12 and presented in Figure 5 suggest the possibility that the EGFP-cdk9dn-C2C12, NIH3T3 and 10T1/2 cells were grown cdk9/cycT2a complex could phosphorylate MyoD in in DMEM supplemented with 10% FBS, L-glutamine and vivo and that phosphorylation is the key to the cdk9- antibiotics (GM). To induce muscle dierentiation, the cells mediated activation of MyoD during the muscle were cultured in DMEM supplemented with 2% HS, L- dierentiation program. Given the importance of cdk glutamine and antibiotics (DM). All cell lines were obtained from ATCC. interaction, phosphorylation, as well as other post- translational modi®cations, in regulating MyoD function during the myoblast-to-myotube transition (Puri Myogenic conversion assay and Sartorelli, 2000; Wei and Paterson, 2001), it is 10T1/2 cells were plated 70% con¯uence in 100 mm dishes, likely that cdk9 binding to MyoD plays a critical role transfected with MyoD in the absence or presence of other in the control of skeletal myogenesis. cdk/cyclin combinations, as reported in the text, and cultured How would the recruitment of cdk9/cycT2a (and in DM for 48 h. Cells were ®xed in formaldehyde/ eventually phosphorylation) augment MyoD transcrip- gluteraldehyde and stained for b-gal using the standard tional function? The activation domain of MyoD has procedure. been implicated in the recruitment of the co-activator p300 (Sartorelli et al., 1997), which is required for Plasmids MyoD-dependent transcription (Yuan et al., 1996; The constructs pcDNA3-cdk9wt-HATag and pcDNA3- Eckner et al., 1996; Puri et al., 1997a,b). Thus, it is cdk9dn-HATag expressing full-length wild-type cdk9 and its possible that cdk9-mediated phosphorylation might kinase-de®cient counterpart have been previously described regulate this interaction. Likewise, recruitment of (De Falco et al., 2000). The constructs pcDNA3-cdk9 (D1± cdk9 by MyoD can allow the phosphorylation of 129)-HATag, pcDNA3-cdk9 (D1 ± 218)-HATag and pcDNA3- MyoD-associated acetyl-transferases, such as p300 cdk9 (D195 ± 372) have also been described elsewhere (Simone itself and PCAF, to modulate their own enzymatic et al., submitted). The plasmids pcDNA3-mMyoD (D101 ± activity during myogenic transcription (Puri et al., 318), pcDNA3-mMyoD (D186 ± 318) and pcDNA3-mMyoD 1997b; Sartorelli et al., 1999; reviewed in Sartorelli and (D1 ± 161) coding for, respectively, MyoD (1 ± 100), MyoD (1 ± 185) and MyoD (162 ± 318) were constructed by PCR. The Puri, 2001). Additionally, the recruitment of cdk9/ ampli®ed products were subcloned in the BamHI site of the cycT2a by the bHLH region of MyoD can provide a pcDNA3 vector (Invitrogen). The correct sequences of all functional link between MyoD and the basal machin- these constructs were con®rmed by sequencing using the ery transcription, resulting in the stimulation of Applied Biosystem model 373A DNA sequencer. GST-cdk9 transcription elongation through the ability of cdk9 (De Falco et al., 2000), the constructs expressing HA-cyclin to phosphorylate the CTD of RNA polymerase II T1, HA-cyclin T2a, HA-cyclin T2b (Peng et al., 1998), MCK- (Simone and Giordano, 2001). In this regard, it is also luciferase, 4RE-luciferase, Gal4-luciferase and Gal4-MyoD-wt possible that cdk9 binding with MyoD could modulate (Puri et al., 2000) have been described previously. His-MyoD the potency of cdk9 in stimulating transcription fusion proteins (wt and the mutants) have been previously elongation. The complex regulation of MyoD by a described (Sartorelli et al., 1999). variety of signals and the activating network of muscle regulatory factors during myogenic dierentiation In vitro binding suggest that cdk9 phosphorylation of MyoD could be The TNT coupled reticulocyte kit was used for in vitro one of these positive signals. This explanation is translation (Promega, WI, USA), according to the manu- supported by the extremely high redundancy of facturer's instructions. All the samples were labeled using 35S- regulation of myogenic transcription in cell culture Methionine (Amersham, IL, USA). The labeled samples were

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4147 incubated with GST-fusion proteins, GST-cdk9 or GST- gal plasmid and by performing a standard b-galactosidase MyoD, using GST as a negative control, for 2 h at 48C, assay. Two mg of total DNA plasmid were transfected and rocking, for in vitro binding. These fusion proteins were transfection-luciferase assays were performed ®ve times in precipitated by glutathione-agarose anity chromatography triplicate. There was no signi®cant dierence in the transfec- and the immunoprecipitates were washed extensively in CPA tion eciencies among the various constructs as determined by buer (20 mM Tris-HCl pH 7.6, 250 mM NaCl, 5 mM MgCl2, the b-gal activity. No evidence of cdk9 altering the b-gal 0.2 mM EDTA, 10% glycerol, 0.2% NP-40) containing fresh activity was observed. The expression of overexpressed inhibitors and 1 mM DTT. Afterwards, the immunoprecipi- proteins was controlled by Western blot analysis. tates were resolved on 10% or 12% SDS ± PAGE and subjected to autoradiography. His-MyoDwt (125 ng), His- Kinase assay MyoD-Nter (125 ng), His-MyoD-bHLH (125 ng) and His- MyoD-Cter (125 ng) were incubated with eluted GST-cdk9 Cell extracts (200 mg) prepared in Lysis Buer were used for (50 ng) for 1 h at 48C, rocking, and pulled down with His- immunoprecipitation with a polyclonal anti-cdk9, anti-cdk2 bind resin (Novagen Inc.) The samples were then washed (1 : 40; De Luca et al., 1997b) and anti-cdk5 (1 : 40; Santa extensively in washing buer (CPA 30 mM imidazole) Cruz, CA) antibody (negative control: pre-immune rabbit containing fresh inhibitors and 1 mM DTT. The samples serum). The immunocomplexes were pulled down with were resolved on 10% SDS ± PAGE and subjected to protein A-sepharose and washed three times with Lysis immunoblot. Bacterially expressed His-MyoD was puri®ed Buer and two times with Lysis Buer containing 400 mM and dialyzed (CPA buer 30 mM imidazole) and after 1 h NaCl. The complexes were also equilibrated in kinase assay 48C incubation with 35S-IVT-cdk9 and/or -cycT2a His-bind buer (minus ATP) (20 mM HEPES pH 7.4, 10 mM Mg resin was added (1 h 48C). The samples were washed Acetate). The kinase assay was performed in a volume of extensively (CPA buer 30 mM imidazole), resolved on 8% 20 ml, using 5 mCi/sample of g-ATP (Amersham, IL, USA). SDS ± PAGE and subjected to autoradiography. Histidine-tagged MyoD wild-type and His-MyoDsp2 (between 0.1 and 1 mg) were added to the samples and incubated for 30 min at 308C. The reactions were stopped by adding Western blot and immunoprecipitation (Ip) 56Laemmli Buer and the samples were loaded and resolved C2C12-EGFP alone, C2C12-EGFP-cdk9 and C2C12-EGFP- on a 10% SDS ± PAGE. cdk9dn extracts (50 mg) were used for Western blot and C2C12 for Ips. These extracts were obtained by lysing the Cyclohexamide treatment cells in Lysis Buer (50 mM Tris, 5 mM EDTA, 250 mM NaCl, 50 mM NaF, 0.1% Triton, 0.1 mM Na3VO4, and NIH3T3 and 10T1/2 cells were transfected with 1 mgof 10 mg/ml aprotinin, leupeptin and phenylmethylsulphonyl CMV-MyoD and pcDNA3-cdk9-HA or pcDNA3-cdk9dn- ¯uoride (PMSF)). The protein concentration was determined HA in 100-mm-diameter dishes using kits according to the by Bradford assay (Biorad, CA, USA), following the manufacturer's instructions (Lipofectamine, Life Technology, manufacturer's instructions and by using BSA as a standard. MD, USA and Promega, WI, USA). The cells were treated Immunoprecipitations were performed after preclearing the with cyclohexamide (Sigma, MO, USA) at 15 mg/ml for the extracts with pre-immune rabbit serum. Anity puri®ed anti- indicated times and harvested for immunoblot analysis. cdk9 (1 : 50; De Falco et al., 2000), anity puri®ed anti- MyoD was detected with anti-MyoD antibody. For each MyoD (1 : 20; Santa Cruz, CA) or anti-cycT2 (1 : 100; Peng et experiment b-tubulin was used as an internal control. al., 1998) antibody incubations were performed overnight, at 48C, followed by the addition of protein A-sepharose for 60 min, rocking, at 48C. Beads were washed three times with Lysis Buer and twice with Lysis Buer 400 mM NaCl, and Acknowledgments loading buer (50 mM Tris/HCl pH 6.8, 2% SDS, 10% We thank Marie Basso for her editorial assistance in the glycerol) with or without 5% b-mercaptoethanol was added. preparation of the manuscript, Dr Margaret Kasten, The samples were resolved in 6, 8, 10 or 12% SDS ± PAA- William Eidenmuller and Dr Pier Paolo Claudio for their Gels and were transferred to a Hybond-ECL nitrocellulose comments and helpful suggestions during the preparation ®lter (Amersham Life Science Inc.) at 48C and at 70 V for of the manuscript and Dr Giuseppe Russo for his technical 2 h. To ensure equal transferring, 0.5% Ponceau Red was support. We also thank Drs Bruce M Paterson, Vittorio used. The blots were blocked with TBST containing 5% non- Sartorelli, Maureen A Harrington, David Price and Franca fat dry milk. Anti-cdk9 (1 : 200), anti-MyoD (1 : 100), anti- Cambi for generously providing reagents. This work was cycT1 (1 : 1000; Peng et al., 1998), anti-cycT2 (1 : 2000) supported in part by grants from NIH, Sbarro Institute for polyclonal antibody and anti-myogenin (1 : 100, Puri et al., Cancer Research and Molecular Medicine (A Giordana) 2000), anti-MHC (1 : 100; Puri et al., 2000), anti-b-tubulin and MURST (G Guanti). C. Simone is supported by a (1 : 100; T-4026) monoclonal antibody were used in TBST `Dottorato di Ricerca in Biologia della Riproduzione containing 5% non-fat dry milk, according to the Western Umana ed Animale' from the University of Bari. C Bellan blot conditions suggested by Santa Cruz (Santa Cruz, CA, and L Bagella are supported by a `Dottorato di Ricerca in USA). Anti-rabbit and anti-mouse peroxidase conjugated Patologia Diagnostica Quantitativa' from the University of (1 : 20000) (Amersham, IL, USA) and ECL detection system Siena. G De Falco was partially supported by an AIDS (NEN, Du Pont, MS, USA) were used for detection. fellowship from ISS and the Sbarro Institute for Cancer Research and Molecular Medicine. A De Luca is the recipient of a grant from the Second University of Naples Transient transfections and luciferase assay and a CNR grant. PL Puri is supported by Telethon, Italy, Transient transfections and luciferase assays were performed and is an Assistant Telethon Scientist. Finally, we thank using kits according to the manufacturer's instructions Letizia Lavermicocca for her encouragement and support. (Lipofectamine, Life Technology, MD and Promega, WI, We dedicate this work to the memory of Professor Angelo USA). Transfections were normalized by co-transfecting the b- Carbonara and Mr Domenico Di Domenico.

Oncogene Physical and functional interactions between MyoD and cdk9 C Simone et al 4148 References

Bagella L, MacLachlan TK, Buono RJ, Pisano MM, Molkentin JD and Olson EN. (1996). Proc. Nat. Acad. Sci., Giordano A and De Luca A. (1998). J. Cell. Physiol., 93, 9366 ± 9373. 177, 206 ± 213. MurreC,McCawPS,VaessinH,CaudyM,JanLY,JanYN, Bagella L, Stiegler P, De Luca A, Siracusa LD and Giordano Cabrera CV, Buskin JN, Hauschka SD and Lassar AB. A. (2000). J. Cell. Biochem., 78, 170 ± 178. (1989). Cell, 58, 537 ± 544. Dahmus ME. (1996). J. Biol. Chem. 271, 19009 ± 19012. Napolitano G, Majello B, Licciardo P, Giordano A and De Falco G and Giordano A. (1998). J. Cell. Physiol., 177, Lania L. (2000). Gene, 254, 139 ± 145. 501 ± 506. Novitch BG, Mulligan GJ, Jacks T and Lassar AB. (1996). J. De Falco G, Bagella L, Claudio PP, De Luca A, Fu Y, Cell. Biol., 135, 441 ± 456. Calabretta B, Sala A and Giordano A. (2000). Oncogene, Peng J, Zhu Y, Milton JT and Price DH. (1998). Genes Dev., 19, 373 ± 379. 12, 755 ± 762. De Luca A, MacLachlan TK, Bagella L, Dean C, Howard Puri PL, Avantaggiati ML, Balsano C, Sang N, Graessmann CM, Claudio PP, Baldi A, Khalili K and Giordano A. A, Giordano A and Levrero M. (1997a). EMBO J., 16, (1997a). J. Biol. Chem., 272, 20971 ± 20974. 369 ± 383. De Luca A, Esposito V, Baldi A, Claudio PP, Fu Y, Caputi Puri PL, Sartorelli V, Yang XJ, Hamamori Y, Ogryzko VV, M, Pisano M, Baldi F and Giordano A. (1997b). J. Cell. Howard BH, Kedes L, Wang JY, Graessmann A, Physiol., 172, 265 ± 273. Nakatani Y and Levrero M. (1997b). Mol. Cell., 1, 35 ± 45. Eckner R, Yao TP, Oldread E and Livingston DM. (1996). Puri PL, Wu Z, Zhang P, Wood LD, Bhakta KS, Han J, Genes Dev., 10, 2478 ± 2490. Feramisco JR, Karin M and Wang JY. (2000). Genes Dev., Edwards MC, Wong C and Elledge SJ. (1998). Mol. Cell. 14, 574 ± 584. Biol., 18, 4291 ± 4300. Puri PL and Sartorelli V (2000). J Cell Physiol., 185, 155 ± Foskett SM, Ghose R, Tang DN, Lewis DE and Rice AP. 173. (2001). J. Virol., 75, 1220 ± 1228. Sartorelli V, Huang J, Hamamori Y and Kedes L. (1997). Fu TJ, Peng J, Price DH and Flores O. (1999). J. Biol. Chem., Mol. Cell. Biol., 17, 1010 ± 1026. 274, 34527 ± 34530. Sartorelli V, Puri PL, Hamamori Y, Ogryzko V, Chung G, Garriga J, Peng J, Parreno M, Price DH, Henderson EE and Nakatani Y, Wang JY and Kedes L. (1999). Mol. Cell., 4, Grana X. (1998). Oncogene, 17, 3093 ± 3102 725 ± 734. GranaX,DeLucaA,SangN,FuY,ClaudioPP,Rosenblatt Sartorelli V and Puri PL. (2001). Front. Biosci., 6, d1073 ± J, Morgan DO and Giordano A. (1994). Proc. Natl. Acad. d1082. Sci., 91, 3834 ± 3838. Simone C and Giordano A. (2001). Front. Biosci., 6, d1024 ± Gu W, Schneider JW, Condorelli G, Kaushal S, Mahdavi V d1047. and Nadal-Ginard B. (1993). Cell, 72, 309 ± 324. Skapek SX, Rhee J, Kim PS, Novitch BG and Lassar AB. Guo K and Walsh K. (1997). J. Biol. Chem., 272, 791 ± 797. (1996). Mol. Cell. Biol., 16, 7043 ± 7053. Herrmann CH, Carroll RG, Wei P, Jones KA and Rice AP. Song A, Wang Q, Goebl M and Harrington M. (1998). Mol. (1998). J. Virol., 72, 9881 ± 9888. Cell. Biol., 18, 4994 ± 4999. Kanazawa S, Okamoto T and Peterlin BM. (2000). Walsh K and Perlman H. (1997). Curr. Opin. Genet. Dev., 7, Immunity, 12, 61 ± 70. 597 ± 602. Kitzmann M, Vandromme M, Schaeer V, Carnac G, Labbe Wei Q and Paterson BM. (2001). FEBS Lett., 490, 171 ± 178. JC, Lamb N and Fernandez A. (1999). Mol. Cell. Biol., 19, Wei P, Gender ME, Fang SM, Fisher WH and Jones KA. 3167 ± 3176. (1998). Cell, 92, 451 ± 462. Kwak YT, Ivanov D, Guo J, Nee E and Gaynor RB. (1999). Weintraub H, Davis R, Tapscott S, Thayer M, Krause M, J. Mol. Biol., 288, 57 ± 69. BenezraR,BlackwellTK,TurnerD,RuppR,Hollenberg Lassar AB, Davis RL, Wright WE, Kadesch T, Murre C, S et al (1991). Science, 251, 761 ± 766. Voronova A, Baltimore D and Weintraub H. (1991). Cell, Wimmer J, Fujinaga K, Taube R, Cujec TP, Zhu Y, Peng J, 66, 305 ± 315. Price DH and Peterlin BM. (1999). Virology, 255, 182 ± Lassar A and MuÈ nsterberg A. (1994). Curr. Opin. Cell Biol., 189. 6, 432 ± 442. Yang X, Hermann HC and Rice AP. (1996). J. Virol., 70, Lis JT, Mason P, Peng J, Price DH and Werner J. (2000). 4576 ± 4584. Genes Dev., 14, 792 ± 803. Yuan W, Condorelli G, Caruso M, Felsani A and Giordano MacLachlan TK, Sang N and Giordano A. (1995). Crit. A. (1996). J. Biol. Chem., 271, 9009 ± 9013. Rew. Eukariot. Gene Expr., 5, 127 ± 156. Zhang JM, Wei Q, Zhao X and Paterson BM. (1999a). MacLachlan TK, Sang N, De Luca A, Puri PL, Levrero M EMBO J., 18, 926 ± 933. and Giordano A. (1998). J. Cell. Biochem., 71, 467 ± 478. Zhang JM, Zhao X, Wei Q and Paterson BM. (1999b). Mancebo H, Lee G, Flygare J, Tomassini J, Luu P, Zhu Y, EMBO J., 18, 6983 ± 6993. Blau C, Hazuda D, Price D, and Flores O. (1997). Genes ZhuY,Pe'eryT,PengJ,RamanathanY,MarshallN, Dev., 11, 2633 ± 2644. MarshallT,AmendtB,MathewsMandPriceD.(1997). Marshall N, Peng J, Xie Z and Price D. (1996). J. Biol. Genes Dev., 11, 2622 ± 2632. Chem., 271, 27176 ± 27183. McGrew MJ, Bogdanova N, Hasegawa K, Hughes SH, Kitsis RN and Rosenthal N. (1996). Mol. Cell. Biol., 16, 4524 ± 4534.

Oncogene