Oncogene (1999) 18, 2916 ± 2924 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc The Myc oncoprotein: a critical evaluation of transactivation and target gene regulation

Michael D Cole*,1 and Steven B McMahon1

1Department of Molecular Biology, Princeton University, Princeton, New Jersey, NJ 08544-1014, USA

Mutations which disrupt the regulation or expression Functional domains and Myc transactivation level of the c-myc gene are among the most common found in human and animal cancers (reviewed in ref. Like most transcription factors, the Myc protein has Cole, 1986; Henriksson and Luscher, 1996; Marcu et al., two major domains. The C-terminal 90 amino acids are 1992). Ectopic expression studies de®ne numerous required for dimerization with Max and sequence- biological activities of the c-myc gene, including speci®c DNA binding (Blackwood and Eisenman, transformation, immortalization, blockage of cell di€er- 1991; Prendergast et al., 1991). Disruption of this entiation and induction of apoptosis (Askew et al., 1991; domain destroys all biological activity (Stone et al., Cole, 1986; Evan and Littlewood, 1993; Freytag et al., 1987), indicating that DNA binding is essential for 1990; Henriksson and Luscher, 1996; Marcu et al., function. The second domain encompasses the remain- 1992). Furthermore, c-myc is required for ecient ing three-fourths of the Myc protein and can be progression through the cell cycle (Goruppi et al., broadly de®ned as involved in `transactivation' (Kato 1994; Prochownik et al., 1988; Yokoyama and Imamoto, et al., 1990). Transactivation of cellular promoters by 1987), although recent studies indicate that it is not Myc is quite low in transient assays, usually about 3 ± 5 absolutely essential (Mateyak et al., 1997). This fold (see Table 1). Reiteration of Myc/Max consensus fascinating array of biological activities makes the c- binding sites in an arti®cial reporter construct can yield myc gene one of the most intriguing oncogenes and up to eightfold transactivation (Amin et al., 1993; Gu presents the challenging question of how a single gene et al., 1993; Kretzner et al., 1992), whereas fusion of can manifest so many di€erent e€ects. The c-Myc the Myc N-terminus to the GAL4 DNA binding protein exhibits sequence-speci®c DNA binding when domain has been reported to yield 20 ± 200-fold dimerized with its partner Max, and DNA binding is transactivation in various studies (Brough et al., mediated through the basic region, which recognizes the 1995; Kato et al., 1990; Resar et al., 1993). The core sequence CACGTG (Berberich et al., 1992; Black- cellular factors that contribute to this transient well et al., 1993; Blackwood and Eisenman, 1991; transactivation remain unknown, although a direct Prendergast and Zi€, 1991; Prendergast et al., 1991), interaction with the TATA binding protein has been but exhibits somewhat higher anity for the more reported (Hateboer et al., 1993; Maheswaran et al., extended sequence ACCACGTGGT (Berberich et al., 1994). The c-Myc N-terminus has short `acidic', 1992; Blackwell et al., 1993; Halazonetis and Kandil, `proline-rich' and `glutamine-rich' clusters similar to 1991). There are three closely related Myc family those associated with some transactivation domains proteins (c-Myc, N-Myc and L-Myc), each with (Kato et al., 1990), but the role of these regions in documented oncogenic potential (Birrer et al., 1988; biological activity has not been thoroughly tested. The Schwab et al., 1985; Yancopoulos et al., 1985) and glutamine-rich region of c-Myc is dispensable for similar DNA binding properties (Mukherjee et al., 1992). oncogenic activity (Stone et al., 1987). In addition, For simplicity, we will use the term Myc to refer to all the Myc N-terminus can `squelch' transactivation by three proteins, but delineate any distinct activities where the herpes virus transactivator VP16 (Resar et al., they apply. The goal of this review is to discuss Myc as a 1993), suggesting that the two proteins compete for transcriptional activator and critically evaluate the common cellular factors in transient assays. GAL4 evidence for the transactivation of speci®c target genes fusion proteins with the L-Myc N-terminus have as direct downstream e€ectors. Since excellent compre- reduced transactivation compared to fusions with c- hensive reviews on Myc have been published recently Myc, which correlates with the weaker oncogenic (Facchini and Penn, 1998; Henriksson and Luscher, potential of L-Myc in most assays (Barrett et al., 1996), we will focus on the latest observations that o€er 1992). Speci®c regions responsible for these di€erences mechanistic insight into transactivation and oncogenic have not been mapped, nor has there been a transformation. comparison of cellular promoter transactivation by c- Myc and L-Myc themselves. Keywords: Myc; oncoprotein; target gene; transactiva- Like the DNA binding domain, early reports tion; chromatin remodelling indicated that the N-terminal transactivation domain is also required for Myc's biological activities. Myc proteins containing deletions within the N-terminus have reduced abilities to cooperate with the ras oncogene, to induce foci in a rat ®broblast line, and to block the di€erentiation of adipocytes (Stone et al., 1987; Sarid et al., 1987; Freytag et al., 1990). However, these deletions with arbitrary boundaries appear to *Correspondence: MD Cole contrast with a naturally occurring variant of c-Myc Myc oncoprotein transactivation MD Cole and SB McMahon 2917 Table 1 Summary of experimental evidence implicating Myc in the regulation of potential target genes

cad ODC MrDb LDH-A prothy cdc25a ISGF3g RCC1 p53 elF-4E ECA39 E2F2 Regulation in log phase myc null + 7 7 7 7 7 7 7 nd 7 7 nd cellsa Fold regulation in serum stimulated 8 1.5 2.2 nd 7 1.8 nd nd +/7 7 nd nd myc null cells; G0?S onlya In vivo footprinting +b nd ?c nd nd nd nd nd nd nd nd nd Fold activation by MycER nd 1.7 ± 3.6d 1.8e 2.0f 4.2g 4h +i 2.2g 1.2f nd nd nd Misregulation from stable ectopic expression of Myc nd +f nd +f +g nd nd +j +k nd nd nd Transient transfections; promoter fusions vs isolated binding sites +l +m +e +n +o +h +i +j +p 19 +r +s All genes contained in the table meet the criteria set forth in the text, i.e. reported activation by Myc (rather than repression) and the presence of a Myc binding site with regulatory region. aBush et al., 1998; bBoyd and Farnham, 1997; Boyd et al., 1998; cAs noted in the text, Myc,Max dimers have been reported to bind an MrDb pseudogene in vivo Grandori et al., 1996; dTavtigian et al., 1994; Tsuneoka et al., 1997; Wagner et al., 1993; eGrandori et al., 1996; fTavtigian et al., 1994; Shim et al., 1997; gEilers et al., 1991; hGalaktionov et al., 1996; iWeihua et al., 1997; jTsuneoka et al., 1997; kReisman et al., 1993; Tavtigian et al., 1994; lMiltenberger et al., 1995; mBello-Fernandez et al., 1993; nShim et al., 1997; oGaubatz et al., 1994; pReisman et al., 1993; qJones et al., 1996; rBenvenisty et al., 1992; sSears et al., 1997. nd=not determined

called MycS. The MycS protein originates from al., 1994; MacGregor et al., 1996). Interestingly, translation initiation at either of two methionines at MbII mutants have unaltered or even enhanced amino acids 100 ± 110 in relation to the M1 position of transactivation in most reported promoter fusion the most abundant Myc2 protein, which means that constructs with one exception (Bello-Fernandez et MycS retains approximately 260 amino acids N- al., 1993; Brough et al., 1995; Desbarats et al., terminal to the DNA binding domain (Spotts et al., 1995). The MbII domain has recently been shown to 1997). Even though MycS is devoid of all transactiva- facilitate Myc binding to a novel large nuclear tion activity assayed with reporter constructs, ectopic cofactor called TRRAP (TRansactivation/tRansfor- expression of MycS can induce anchorage-independent mation-domain Associated Protein) (McMahon et growth and apoptosis as well as rescue the cell cycle al., 1998). TRRAP is a 3830 amino acid protein delay of Myc-de®cient ®broblasts (Xiao et al., 1998). with limited homology to the PI3 kinase/ATM Conversely, replacing the c-Myc N-terminus with the family, although TRRAP lacks the kinase catalytic potent transactivation domain from the herpesvirus residues present in other members of the family. VP16 protein fails to reconstitute an oncogenic TRRAP binding to the N-terminus is directly function (Brough et al., 1995). Thus, transactivation correlated with Myc oncogenic activity, since by the Myc N-terminus (as de®ned by transient deletions or mutations in Myc that disrupt TRRAP reporter assays) is neither necessary nor sucient in binding are transformation-defective and the weakly many biological assays. On the other hand, it remains transforming L-Myc protein exhibits poor TRRAP entirely possible that the transactivation of speci®c binding. Furthermore, the disruption of endogenous cellular promoters is required for Myc function and TRRAP pools using antisense and dominant that the apparent dichotomy is only a consequence of inhibitory constructs severely impairs Myc-mediated the inability of transient reporter assays to recapitulate oncogenic transformation. These data imply that the the regulation of chromosomal targets. Alternately, the recruitment of TRRAP to cellular promoters is biological functions of Myc may be linked to gene essential for Myc-mediated oncogenic transformation. repression rather than activation, although the The identi®cation of TRRAP as an essential mechanism of repression and its direct association cofactor provided an important mechanistic insight with Myc remain unclear (Claassen and Hann, 1999). into the function of the Myc N-terminal domain when TRRAP was found to be part of the SAGA complex (Saleh et al., 1998). SAGA (SPT/ADA/GCN5/Acetyl- TRRAP links Myc transactivation to chromatin transferase) is a 1.8 MDa complex containing approxi- remodeling and histone acetylation mately 20 proteins which have been implicated in transcriptional regulation, primarily through genetic The Myc N-terminus contains several discrete blocks screens in yeast (Grant et al., 1997). However, several of amino acids which have been conserved in all recent studies have demonstrated that in addition to Myc family proteins throughout evolution (Cole, TRRAP, many other components of the SAGA 1986; Henriksson and Luscher, 1996), referred to as complex are also highly conserved from yeast to Myc homology boxes. Among the three major humans (Martinez et al., 1998; Ogryzko et al., 1998; conserved regions of the N-terminus, the most Smith et al., 1998a). Among the many proteins critical for oncogenic transformation is Myc box II contained in SAGA, the only one with a clearly (MbII) (Stone et al., 1987), which is centered around de®ned biochemical function is the histone acetyltrans- amino acid 135 in human c-Myc. Small deletions or ferase GCN5 (Georgakopoulos and Thireos, 1992; single missense mutations in MbII are sucient to Marcus et al., 1994; Wang et al., 1997). Histone eliminate virtually all Myc biological activity, acetylation by transcription cofactors has frequently including oncogenic activity, apoptosis and blocking been associated with gene activation (Grant et al., di€erentiation (Askew et al., 1991; Brough et al., 1998), making this an attractive mechanism for Myc- 1995; Evan et al., 1992; Freytag et al., 1990; Li et mediated transactivation. This model is appealing since Myc oncoprotein transactivation MD Cole and SB McMahon 2918 the alternate Max heterodimeric partners, Mad and expression pattern. In addition to the speci®c Mxi, can antagonize Myc function through recruitment techniques utilized, the search for c-Myc targets is of Sin3A/Sin3B and the histone deacetylases HDAC1/2 heavily in¯uenced by underlying assumptions inherent (Hassig et al., 1997; Laherty et al., 1997; Zhang et al., in any screen. Is Myc an activator or repressor? Will 1997; Sommer et al., 1997). the expression of the target gene be higher in Myc A general model of Myc-mediated histone acetyla- transformed cells than in normal growing cells or tion as the basis of oncogenic transformation raises a simply expressed at the same level in both cell types? A number of important questions that must be compilation of all proposed Myc targets and the means addressed in the future. First, the TRRAP ortholog of identi®cation has been presented in recent compre- in S. cerevisiae (TRA1p) is essential for viability hensive reviews (Dang, 1999; Facchini and Penn, 1998; whereas the GCN5p protein is non-essential (Marcus Grandori and Eisenman, 1997). In this review, we will et al., 1994; Saleh et al., 1998). This observation focus on only the proposed targets (Table 1) that meet raises the possibility that some other component of the following criteria: the SAGA complex is a critical mediator of Myc- dependent gene regulation. One possibility is that 1 The proposed target gene is activated rather than TRRAP is associated with another histone acetyl- repressed by Myc. Although repression is an transferase such as the mammalian ortholog of the important activity of c-Myc linked to its biological essential ESA1p protein (called TIP60 in humans) functions, no repressed targets have been shown to (Kamine et al., 1996; Smith et al., 1998b). However, have direct Myc binding sites. Furthermore, the preliminary experiments fail to show an interaction Myc protein itself or fusion proteins containing between Myc and TIP60 (S McMahon, unpublished domains of Myc activate reporter constructs with results). Alternately, TRRAP itself may have a appropriate binding sites, and hence the direct Myc unique gene regulatory function that is not predicted targets are predicted to be activated. Repression by through known protein motifs. A second major Myc is reviewed elsewhere in this issue. question is which, if any, Myc target genes are 2 The proposed target gene has a binding site for Myc/ dependent on the MbII domain and hence presum- Max heterodimers in or near the promoter and this ably on TRRAP and histone acetylation. In most binding site has been implicated in regulation, assays, it is repression by Myc that is dependent on suggesting a direct interaction with Myc. MbII, not transactivation (Li et al., 1994; Penn et al., 1990). However, repression is not consistent with the It is important to note an often overlooked ®nding expected role of histone acetylation in opening from innumerable studies of proposed Myc targets. condensed chromatin. The latter question will be Ectopic expression of Myc rarely if ever leads to any discussed in more detail below. Finally, there are net upregulation of proposed target genes compared likely to be other cofactors recruited by Myc that to controls with the expression vector alone, as long contribute gene regulatory functions. For example, as the culture conditions are not biased to favor the the MycS protein does not bind to TRRAP and growth of only the Myc-expressing cells. An example cannot cooperate with H-ras to transform primary of the latter are cells deprived of growth factors, rodent cells (McMahon et al., 1998). This suggests where ectopic Myc promotes factor-independent that MycS may interact with other cofactors to growth. Under these circumstances, it becomes mediate cell cycle progression, anchorage-independent dicult to distinguish direct Myc-dependent gene growth and apoptosis. The identi®cation of additional regulation from the indirect consequences of growth. Myc cofactors may o€er new mechanistic insights For even the best candidate targets, the level of into oncogenic transformation. Other proteins re- endogenous Myc protein is not rate-limiting for ported to interact with Myc are discussed elsewhere transcriptional regulation. Some exceptions to this in this issue (B LuÈ scher and colleagues). general ®nding are discussed below. Before discussing the potential contribution of proposed Myc target genes to growth and oncogen- What are Myc's target genes? esis, it is useful to review the di€erent assays that have The single biggest impediment to understanding the been used to implicate individual genes in Myc- function of the c-Myc protein is the lack of a dependent regulation. The strengths and weaknesses comprehensive set of Myc target genes. The diculty of each assay for assigning a direct regulatory link stems in part from the fact that myc was ®rst identi®ed between Myc and target genes are noted. genetically, i.e. as the oncogenic sequence within di€erent avian viruses and as a site of chromosomal (A) Misregulation of candidate target genes in Myc- lesions in diverse tumors (reviewed in Cole, 1986). This de®cient ®broblasts. An extremely powerful new contrasts with the majority of transcription factors, tool to evaluate candidate Myc target genes is a which have in general been identi®ed through their Rat1 cell line rendered completely Myc-de®cient binding to key regulatory elements in speci®c by homologous recombination (Mateyak et al., promoters. Myc DNA binding activity is virtually 1997). These Myc null cells do not express c-Myc, undetectable in most cells, so in the absence of the N-Myc or L-Myc and, although they continue to genetic and oncogenic links the protein would have grow, they have a dramatically extended cell cycle gone undetected by most conventional approaches to time that is nearly three times longer than their transcriptional regulation. wild type parent. In these cells, direct Myc Numerous studies have attempted to identify the transactivation targets are expected to exhibit targets of the Myc oncogene, mainly using di€erential reduced expression, and Myc-dependent regula- or subtractive hybridization, or guessing based on tion can be studied in both log phase and after Myc oncoprotein transactivation MD Cole and SB McMahon 2919 growth factor deprivation/restimulation. A criti- addition promotes the dissociation of the inhibi- cism of this approach is that the cells could `adapt' tory complex and migration to the nucleus. An to the Myc-de®ciency through some compensatory important advantage of MycER is the ability to upregulation of Myc targets by other transcription activate the fusion protein in the presence of factors. Although it is not yet possible to eliminate protein synthesis inhibitors, thereby preventing this caveat entirely, we think that the stability of secondary gene regulatory cascades. Transducing the cell cycle defect, which appears immediately MycER into the Myc null cells should in theory upon targeting of the second c-myc allele, argues provide a completely hormone-inducible system that the Myc null cells have not `adapted' to the without endogenous Myc protein to complicate the Myc-de®ciency. For genes that are not upregulated analysis. However, MycER must be used with care by Myc overexpression and not downregulated by for a number of reasons. First, the MycER fusion Myc loss-of-function, in lieu of compelling is often leaky, especially with extended cell culture evidence to the contrary the principle of Occam's times, and even with a mutant form of the ER razor supports the simplest interpretation, i.e. that domain that is only responsive to tamoxifen such genes are not likely to be direct Myc targets. (Littlewood et al., 1995). For example, Myc null On the other hand, when using Myc-de®cient cells cells harboring MycER frequently appear to be to assess target genes, one must remain aware that phenotypically rescued, even in the absence of changes in gene expression that arise may be an hormone (J Sedivy, personal comm.). Similarly, indirect consequence of the slowed growth rate or cell lines expressing MycER can also exhibit c-myc other downstream pathways and not a direct autosuppression or proposed target gene activation consequence of the absence of Myc. in the absence of hormone. The successful use of (B) In vivo crosslinking/footprinting. The ultimate MycER appears to require the derivation of proof that a candidate gene is a direct Myc freshly transduced cell lines that are analysed target demands that the Myc/Max heterodimers before extended propagation in culture (L Penn, be shown to physically occupy the proposed personal comm.). A ®nal caveat with MycER is regulatory site. Although in vivo footprinting has the high level of protein that accumulates in the been available for many years, the technical cytoplasm, which then ¯oods the nucleus with diculty and the ambiguity arising from multiple non-physiological doses of a potent oncoprotein potential factors that can occupy a chromosomal upon hormone addition. Once induced with site (e.g. the unrelated transcription factor USF hormone, the MycER fusion protein is degraded binds with equal or greater anity than Myc/Max very rapidly and the level falls to barely detectable dimers to CACGTG sites (Sirito et al., 1994)) has levels within 2 h in the presence of cyclohexamide discouraged its use. The importance of demon- (L Penn, personal communication). Therefore any strating in vivo binding to a potential regulatory examination of target gene e€ects should be site is underscored by a recent unanticipated conducted at this timepoint. ®nding in studies of the E2F transcription factor (D) Inducible and constitutive Myc expression. One family, which shares the complexities of the Myc of the mainstays for the analysis of candidate family. Transient assays suggested that an E2F Myc target genes is the use of cell lines binding site in the B-myb promoter was the target engineered to express Myc from either viral or of E2F-mediated transactivation (Zwicker et al., inducible promoters. The recently developed tet 1996). However, in vivo footprinting demonstrated system appears to give tight control (Lutz et al., that the E2F site was only occupied when B-myb 1996), although unlike MycER induction must was repressed (perhaps by an Rb/E2F complex) allow cellular protein synthesis. and not occupied when E2F transactivation is (E) Transient expression of Myc with candidate target induced by Rb dissociation. Thus, transient promoters. The dissection of regulatory elements cotransfection experiments had yielded data that within promoters continues to require the analysis was completely opposite to the in vivo ®ndings. In of promoter fusions with reporter constructs. vivo crosslinking techniques have recently been However, since the location of Myc/Max binding extended to study the chromosomal sites to which sites may in¯uence inducibility (Gaubatz et al., the endogenous Myc protein is predicted to bind. 1994; Boyd et al., 1998) maintaining the promoter To date, only the cad gene has been examined for intact is far preferable to the insertion of potential an in vivo interaction with Myc/Max heterodimers regulatory elements into an arti®cial reporter. The (Boyd and Farnham, 1997; Boyd et al., 1998) classic study of Ma and Ptashne showed that (although a pseudogene of MrDb was shown to transactivation can be enhanced by simply bind Myc/Max in vivo; [Grandori et al., 1996]), tethering small acidic domains near basal promo- and a wider use of in vivo analysis should resolve ters (Ma and Ptashne, 1987), even though such many questions surrounding candidate Myc domains may have no physiological role in gene targets. regulation. Since potential Myc/Max binding sites (C) Myc-Estrogen Receptor activation. One of the presumably occur at random in the genome, most useful tools for the demonstration of a direct cotransfection of promoter fusions with a Myc regulatory interaction between Myc and proposed expression vector may yield fortuitous transactiva- target genes is activation by a fusion protein of tion. Nevertheless, site-directed mutagenesis of Myc with the hormone binding domain of the potential Myc-dependent regulatory elements estrogen receptor (Eilers et al., 1991). The fusion provides critical data to support a role for specific protein is held in an inactive, largely cytoplasmic Myc/Max binding sites in candidate promoter form in the absence of hormone, and hormone regulation. Myc oncoprotein transactivation MD Cole and SB McMahon 2920 Evaluation of proposed Myc target genes consistently downregulated in proliferating Myc-defi- cient cells, and even this downregulaton was quite What can an evaluation of proposed Myc target genes modest (*threefold). The absence of Myc also had the reveal about the mechanism of Myc-mediated transac- most severe a€ect on the cad gene during serum tivation? Are all proposed Myc targets actually stimulation (*eightfold). Reconstitution of Myc into regulated by Myc in vivo or are some genes more the null cells restores cad expression to a level at or likely to be regulated in vivo than others? Among the above that found in parental cells, con®rming that the strongest candidates, is there a unique con®guration of downregulation was indeed Myc-dependent. Further- sites that can be used to predict new Myc targets in the more, the cad promoter has been demonstrated to genome? The data derived from diverse approaches are interact with Myc and Max in vivo by crosslinking listed for all of the targets proposed to be transacti- (Boyd and Farnham, 1997; Boyd et al., 1998). Thus, vated by Myc through consensus Myc/Myc binding the cad gene appears to be uniquely dependent on Myc sites. The level of support for a direct in vivo compared to other proposed targets. interaction between Myc and the proposed targets It is useful to consider several features of the cad varies widely. We have divided the proposed targets promoter that might make it more dependent on into three groups to facilitate discussion, with the most endogenous Myc than with other genes. First, cad is weight given to the response of individual genes to the only Myc target without a TATA element in the Myc loss-of-function where determined. promoter (Kollmar et al., 1994). Second, the Myc/Max consensus binding site shown to contribute to gene regulation falls at +65 from the major mRNA The cad gene exhibits a unique Myc-dependent initiation site (Miltenberger et al., 1995). This site is regulation closer to the transcription start site than those mapped The Myc loss-of-function analysis in Rat1 ®broblasts in other promoters, although it conforms to the provided the surprising ®nding that virtually all of the frequent ®nding that the Myc/Max binding sites are proposed Myc transactivation targets were una€ected within transcribed regions rather than in the 5' ¯anking by the Myc-de®ciency in growing cells (Bush et al., sequences. Third, the Myc/Max consensus binding site 1998). Therefore, the una€ected genes must not is well conserved in both sequence and location in contribute to the stable slow growth phenotype caused di€erent mammals where it has been mapped (Boyd by the Myc knockout. Only the cad gene was and Farnham, 1997; Boyd et al., 1998). Fourth, the cad

a

Myc-level required for activation Myc oncoprotein transactivation MD Cole and SB McMahon 2921 promoter contains Sp1 sites that contribute to basal of peak cad induction (Mateyak et al., 1997), so it is expression and also may contribute cooperatively to possible that the transcriptional activity of c-Myc may cad regulation by Myc (Kollmar et al., 1994). No other be di€erentially regulated at distinct points in G1 proposed target promoters have these speci®c features, (Colman and Ostrowski, 1996; Lutterbach and Hann, although the relative importance of each to Myc- 1994). Alternately, some cofactor that parallels cad dependent regulation remains unresolved. It will be induction may synergize with Myc to activate the important to determine if cad is dependent on TRRAP promoter. and the SAGA chromatin remodeling complex or on other Myc cofactors. Preliminary experiments suggest Target genes a€ected by Myc during growth stimulation that cad expression is not dependent on MbII in log and growth suppression phase cells (M Cole, unpublished data). It is interesting to note that the time course of cad Withdrawal of essential growth factors suppresses induction does not directly parallel c-Myc protein proliferation and leads to the downregulation of expression. The cad expression peaks at 8 ± 12 h after growth-regulated genes, including c-myc and most serum stimulation, whereas Myc protein levels peak at proposed Myc targets. Sustained or inducible Myc 3 h (Bush et al., 1998). However, c-Myc continues to expression, including the activation of MycER, pro- be synthesized and levels remain high during the time motes cell growth during growth factor deprivation,

b

Figure 1 Models of gene-speci®c and global transcriptional regulation by Myc family proteins. (a) Genes speci®cally regulated by Myc proteins have been divided into three classes. Most of the currently known Myc targets fall into Class I, where e€ects of Myc are seen only following experimental manipulation of Myc protein levels. These genes are una€ected by the absence of Myc in knockout cells, and can therefore not be responsible for the profound proliferative defects seen in these cells. Evidence suggests that Myc-dependent regulation of Class I genes is observed only when Myc protein concentrations are raised to arti®cially high levels, such as following transient transfection or induction of MycER. It is possible that in certain tumor cells, Myc levels are similarly raised to very high levels, bringing these genes under the control Myc. However, in most cases upregulation in Myc-induced tumor cells has not been documented experimentally. The only known gene belonging to Class II is cad, which is strongly downregulated in cells lacking Myc. Furthermore, cad levels are restored in knockout cells which have been rescued by re-introduction of Myc and Myc proteins have been crosslinked to the cad gene in vivo. However, it is only the as yet unknown targets which comprise Class III which are dependent on the biologically critical N-terminal MbII domain of the Myc protein. These hypothetical Class III targets may also be dependent on the MbII-dependent cofactor TRRAP, which links Myc to the large transcriptional regulatory SAGA complex. (b) Myc proteins may exert global a€ects on gene expression by inducing regional chromatin remodeling in the vicinity of Myc binding sites in the genome. Because Myc binding sites occur at a high frequency (approximately every 4000 bp) in the genome, this mechanism may be dicult to detect with standard assays of gene expression. Like gene-speci®c regulation, this mechanism may rely on the ability of Myc to recruit TRRAP and the associated SAGA complex. It is important to note that the gene-speci®c and global mechanisms of Myc-dependent transcriptional regulation presented here are not mutually exclusive Myc oncoprotein transactivation MD Cole and SB McMahon 2922 sometimes making cells growth factor-independent. It have been shown to bind to both consensus has always been dicult to separate the direct e€ect of (ACCACGTGGT) and non-consensus sites in vitro, Myc on target gene expression from indirect gene and the binding anity is highest for consensus activation as a consequence of cell growth. MycER sites (Blackwood et al., 1992; Blackwell et al., 1993; activation o€ers one solution to this problem, since the Prendergast and Zi€, 1991). All of the target genes fusion protein can be activated by hormone in growth that exhibit any Myc-dependent regulation in vivo factor deprived cells in the presence of cyclohexamide. contain E-box elements which conform well to the MycER can activate prothymosin (Eilers et al., 1991), consensus binding site, and no non-consensus sites ODC (Wagner et al., 1993), LDH-A (Shim et al., 1997), which function in chromatin have been identi®ed. MrDb (Grandori et al., 1996) and cdc25A (Galaktionov Sequence variations introduced into the cad et al., 1996) under these conditions. Quantitation of promoter at the +4 and +5 positions (numbered MrDb induction with the tamoxifen-inducible MycER from the dyad center) that have increased anity fusion showed a 1.8-fold increase when normalized to for USF and decreased anity for Myc/Max are GAPDH expression (Grandori et al., 1996). also reduced for growth factor induction (Boyd and An alternate approach to the problem is to compare Farnham, 1997). In the limited set of promoters growth factor-induced gene expression in Myc null cells available, there are no common adjacent binding with that in their parental line. When normalized to sites for other transcription factors that might controls, prothymosin, ODC, MrDb and cdc25A contribute combinatorial speci®city. exhibit a small reduction (1.5 ± 2-fold) in the initial One factor that will in¯uence the accessibility of growth factor induction in Myc null cells, but the RNA potential Myc/Max binding sites in vivo is methylation. levels become equal to the parental line by 24 h (Bush The core consensus sequence contains a CpG which is et al., 1998). The latter observation is consistent with frequently methylated in mammalian cells. Methylation the equal levels of mRNA from these genes in log of the consensus site blocks Myc/Max binding, but not phase Myc null and wild type cells. The Myc- binding by USF (Prendergast and Zi€, 1991; dependent contribution in this assay is similar to that Prendergast et al., 1991). The structural basis for this observed with MycER (1.5 ± 2-fold in each case). inhibition by CpG methylation is not readily apparent Combining the two assays above with promoter from the existing crystal structure (S Burley, personal mapping that demonstrates functional Myc/Max comm.). A consequence of this methylation interfer- consensus binding sites suggests that Myc makes a ence is that only non-methylated sites should be small contribution (1.5 ± 2-fold) to the activation of available as targets of Myc transactivation during prothymosin, ODC, MrDb, LDH-A and cdc25A development. How this might in¯uence the dynamics of during the early phases of growth factor stimulation. Myc/Max heterodimer interactions with chromosomes More importantly, these studies make it clear that remains to be explored. other factors must account for the majority of the growth factor inducibility of these genes. The Myc- dependence of these genes disappears as cells reach a General models for Myc-mediated transactivation steady state of cell cycle progression, whether at a normal rate in wild type cells or a very slow rate in Despite recent advances that have identi®ed new Myc Myc null cells. For the targets in this group, it is cofactors and helped to clarify the contribution of Myc interesting to note that the contribution of Myc to to speci®c target gene transactivation, numerous growth factor inducible gene expression peaks at 3 h questions remain to be answered before we can along with the peak of Myc expression, whereas the understand how this transcription factor functions in more dramatic Myc-dependent regulation of cad peaks oncogenic transformation. Models that accommodate later. This observation suggests that transactivation by the existing data are diagrammed in Figure 1. One Myc occurs through a di€erent mechanism or through model (Figure 1a) is that Myc may function to di€erent cofactors in the di€erent classes of target transactivate 2 ± 3 di€erent tiers of target genes, genes (see Figure 1). depending on the abundance of Myc in the cell and the activity of other transcription factors. The broadest class may contain genes that are normally regulated as Proposed Myc targets with insucient characterization cells transit from a non-growing to growing state. Myc to classify may bind to these genes only when it is expressed at its The data in Table 1 shows that many proposed targets highest level, and Myc may either supplement have not been characterized as induced by Myc alone or transcriptional activation by other transcription fac- MycER in cell lines, and transient cotransfection assays tors or enhance the activity of other factors by are insucient to con®rm a regulatory interaction. For relieving general chromatin-mediated repression. A example, eIF4E and ECA39 show no misregulation in second tier of Myc targets is represented by the cad either log phase or serum stimulated Myc null cells gene, where transcription is dependent on the lower (Bush et al., 1998). A more comprehensive analysis of levels of Myc found in log phase cells and on the close these genes should be performed before they are proximity of the Myc binding site to the mRNA considered bona ®de Myc targets. initiation site. The model also proposes a speculative third tier of Myc targets that are highly dependent on functionally essential domains of Myc, such as MbII. Myc/Max binding sites These genes, as yet unidenti®ed, may be more profoundly dependent on Myc in growing cells than How does Myc choose speci®c chromosomal sites cad and these genes may account for the cell cycle as targets of transactivation? Myc/Max heterodimers defect in the Myc null cells. Myc oncoprotein transactivation MD Cole and SB McMahon 2923 The model in Figure 1a casts Myc as a conventional Conclusion transcription factor with speci®c target genes as downstream e€ectors. However, it remains possible There have been major advances in our understanding that the transactivation of select target genes is not at of Myc-dependent transcriptional activation and target the heart of oncogenic transformation at all. An genes in the last year. The derivation of Myc-de®cient alternate model (Figure 1b) is that Myc is a global cells, in vivo crosslinking, and the identi®cation of new potentiator of chromatin opening or remodeling cofactors linking Myc to previously characterized (perhaps via TRRAP and the SAGA complex), and transcriptional regulatory complexes have combined that Myc provides a subtle but signi®cant enhancement to provide a much clearer picture of the role of Myc in of transcription from hundreds or even thousands of target gene regulation. However, the enigma surround- genes simultaneously. A genome-wide increase in ing the role of Myc in oncogenic transformation transcription would be very dicult to detect due to remains intact, since no compelling direct target gene the lack of a suitable point of normalization. The that can mediate log phase cell cycle progression has models in Figure 1 are not mutually exclusive, and it emerged. The recent advances discussed here should remains possible that Myc may potentiate both gene- make it possible to identify and evaluate new Myc speci®c and global transcription simultaneously. target genes with much greater con®dence in the future.

References

Amin C, Wagner AJ and Hay N. (1993). Mol. Cell. Biol., 13, Gaubatz S, Meichle A and Eilers M. (1994). Mol. Cell. Biol., 383 ± 390. 14, 3853 ± 3862. Askew DS, Ashmun RA, Simmons BC and Cleveland JL. Georgakopoulos T and Thireos G. (1992). EMBO J., 11, (1991). Oncogene, 6, 1915 ± 1922. 4145 ± 4152. Barrett J, Birrer MJ, Kato GJ, Dosaka-Akita H and Dang GoruppiS,GustincichS,BrancoliniC,LeeWMFand CV. (1992). Mol. Cell. Biol., 12, 3130 ± 3137. Schneider C. (1994). Oncogene, 9, 1537 ± 1544. Bello-Fernandez C, Packham G and Cleveland JL. (1993). Grandori C and Eisenman RN. (1997). Trends Biochem. Sci., Proc. Natl. Acad. Sci. USA, 90, 7804 ± 7808. 22, 177 ± 181. Benvenisty N, Leder A, Kuo A and Leder P. (1992). Genes Grandori C, Mac J, Siebelt F, Ayer DE and Eisenman RN. Dev., 6, 2513 ± 2523. (1996). EMBO J., 15, 4344 ± 4357. Berberich S, Hyde-DeRuyscher N, Espenshade P and Cole GrantPA,DugganL,CoteJ,RobertsSM,BrownellJE, M. (1992). Oncogene, 7, 775 ± 779. Candau R, Ohba R, Owen-Hughes T, Allis CD, Winston Birrer MJ, Segal S, Degreve JS, Kaye F, Sausville EA and F, Berger SL and Workman JL. (1997). Genes Dev., 11, Minna JD. (1988). Mol. Cell. Biol., 8, 2668 ± 2673. 1640 ± 1650. BlackwellTK,HuangJ,MaA,KretznerL,AltFW, Grant PA, Sterner DE, Duggan LJ, Workman JL and Berger Eisenman RN and Weintraub H. (1993). Mol. Cell. Biol., SL. (1998). Trends Cell Biol., 8, 193 ± 197. 13, 5216 ± 5224. Gu W, Cechova K, Tassi V and Dalla-Favera R. (1993). Blackwood EM and Eisenman RN. (1991). Science, 251, Proc. Natl. Acad. Sci. USA, 90, 2935 ± 2939. 1211 ± 1217. Halazonetis TD and Kandil AN. (1991). Proc. Natl. Acad. Blackwood EM, Luscher B and Eisenman RN. (1992). Genes Sci. USA, 88, 6162 ± 6166. Dev., 6, 71 ± 80. Hassig CA, Fleischer TC, Billin AN, Schreiber SL and Ayer Boyd KE and Farnham PJ. (1997). Mol. Cell. Biol., 17, DE. (1997). Cell, 89, 341 ± 347. 2529 ± 2537. Hateboer G, Marc Timmers HT, Rustgi AK, Billaud M., Boyd KE, Wells J, Gutman J, Bartley SM and Farnham PJ. van'T Veer LJ and Bernards R. (1993). Proc. Natl. Acad. (1998). Proc. Natl. Acad. Sci. USA, 95, 13887 ± 13892. Sci. USA, 90, 8489 ± 8493. Brough DE, Hofmann TJ, Ellwood KB, Townley RA and Henriksson M and Luscher B. (1996). Adv. Cancer Res., 68, Cole MD. (1995). Mol. Cell. Biol., 15, 1536 ± 1544. 109 ± 182. Bush A, Dugan K, Mateyak M, Obaya A, Adachi S, Sedivy J Jones RM, Branda J, Johnston KA, Polymenis M, Gadd M, and Cole MD. (1998). Genes Dev., 12, 3797 ± 3802. Rustigi A, Callahan L and Schmidt EV. (1996). Mol. Cell. Cole MD. (1986). Ann. Rev. Gen, 20, 361 ± 385. Biol., 16, 4754 ± 4764. Claassen GF and Hann SR. (1999). Oncogene, 18, 2925 ± Kamine J, Elangovan B, Subramanian T, Coleman D and 2933. Chinnadurai G. (1996). Virology, 216, 357 ± 366. Colman MS and Ostrowski MC. (1996). Nucl. Acids Res., 24, Kato GJ, Barrett J, Villa-Garcia M and Dang CV. (1990). 1971 ± 1978. Mol. Cell. Biol., 10, 5914 ± 5920. Dang CV. (1999). Mol. Cell. Biol., 19, 1±11. Kollmar R, Sukow K, Sponagle S and Farnham PJ. (1994). Desbarats L, Gaubatz S and Eilers M. (1996). Genes Dev., 10, J. Biol. Chem., 269, 2252 ± 2257. 447 ± 460. Kretzner L, Blackwood EM and Eisenman RN. (1992). Eilers M, Schirm S and Bishop JM. (1991). EMBO J., 10, Nature, 359, 426 ± 429. 133 ± 142. Laherty CD, Yang WM, Sun JM, Davie JR, Seto E and Evan G and Littlewood TD. (1993). Curr. Opin. Genet. Dev., Eisenman RN. (1997). Cell, 89, 349 ± 356. 3, 44 ± 49. Li LH, Nerlov C, Prendergast G, MacGregor D and Zi€ EB. Evan GI, Wyllie AH, Gilbert CS, Littlewood TD, Land H, (1994). EMBO J., 13, 4070 ± 4079. Brooks M, Waters CM, Penn LZ and Hancock DC. Littlewood TD, Hancock DC, Danielian PS, Paten MG and (1992). Cell, 69, 119 ± 128. Evan GI. (1995). Nucleic Acids Res., 23, 1686 ± 1690. Facchini LM and Penn LZ. (1998). FASEB J., 12, 633 ± 651. Lutterbach B and Hann SR. (1994). Mol. Cell. Biol., 14, Freytag SO, Dang CV and Lee WMF. (1990). Cell Growth 5510 ± 5522. Di€., 1, 339 ± 343. Lutz W, Stohr M, Schurmann J, Wenzel A, Lohr A and Galaktionov K, Chen X and Beach D. (1996). Nature, 382, Schwab M. (1996). Oncogene, 13, 803 ± 812. 511 ± 517. Ma J and Ptashne M. (1987). Cell, 51, 113 ± 119. Myc oncoprotein transactivation MD Cole and SB McMahon 2924 MacGregor D, Li LH and Zi€ EB. (1996). J. Cell. Physiol., Shim H, Dolde C, Lewis BC, Wu CS, Dang G, Jungmann 167, 95 ± 105. RA, Dalla-Favera R and Dang CV. (1997). Proc. Natl. Maheswaran S, Lee H and Sonenshein GE. (1994). Mol. Cell. Acad.Sci.USA,94, 6658 ± 6663. Biol., 14, 1147 ± 1152. Sirito M, Lin Q, Maity T and Sawadogo M. (1994). Nucl. Marcu KB, Bossone SA and Patel AJ. (1992). Ann. Rev. Acids Res., 22, 427 ± 433. Biochem., 61, 809 ± 860. Smith ER, Belote JM, Schiltz RL, Yang X-J, Moore PA, Marcus G, Silverman N, Berger S, Horiuchi J and Guarente Berger SL, Nakatani Y and Allis CD. (1998a). Nucl. Acids L. (1994). EMBO J., 13, 4807 ± 4815. Res., 26, 2948 ± 2954. Martinez E, Kunda TK, Fu J and Roeder RG. (1998). J. Biol. SmithER,EisenA,GuW,SattahM,PannutiA,ZhouJ, Chem., 273, 23781 ± 23785. Cook RG, Lucchesi JC and Allis CD. (1998b). Proc. Natl. Mateyak MK, Obaya AJ, Adachi S and Sedivy JM. (1997). Acad.Sci.USA,95, 3561 ± 3565. Cell Growth Di€er., 8, 1039 ± 1048. Sommer A, Hilfenhaus S, Mankel A, Kremmer E, Seiser C, McMahon SB, VanBuskirk HA, Dugan KA, Copeland TD Loidl P and LuÈ scher B. (1997). Curr. Biol., 7, 357 ± 365. and Cole MD. (1998). Cell, 94, 363 ± 374. Spotts GD, Patel SV, Xiao Q and Hann SR. (1997). Mol. Miltenberger RJ, Sukow KA and Farnham PJ. (1995). Mol. Cell. Biol., 17, 1459 ± 1468. Cell. Biol., 15, 2527 ± 2535. Stone J, De Lange T, Ramsay G, Jakobovits E, Bishop JM, Mukherjee B, Morgenbesser SD and DePinho R. (1992). Varmus H and Lee W. (1987). Mol. Cell. Biol., 7, 1697 ± Genes Dev., 6, 1480 ± 1492. 1709. Ogryzko VV, Kotani T, Zhang X, Schiltz RL, Howard T, Tavtigian SV, Zabludo€ SD and Wold BJ. (1994). Mol. Biol. Quin J and Nakatani Y. (1998). Cell, 94, 35 ± 44. Cell, 5, 375 ± 388. Penn LJZ, Brooks MW, Laufer EM, Littlewood TD, Tsuneoka M, Nakano F, Ohgusu H and Mekada E. (1997). Morgenstern JP, Evan GI, Lee WMF and Land H. Oncogene, 14, 2301 ± 2311. (1990). Mol. Cell. Biol., 10, 4961 ± 4966. Wagner AJ, Meyers C, Laimins LA and Hay N. (1993). Cell Prendergast GC, Lawe D and Zi€ EB. (1991). Cell, 65, 395 ± Growth Di€er., 4, 879 ± 883. 407. Wang L, Mizzen C, Ying C, Candau R, Barlev N, Brownell J, Prendergast GC and Zi€ EB. (1991). Science, 251, 186 ± 189. Allis CD and Berger SL. (1997). Mol. Cell. Biol., 17, 519 ± Prochownik EV, Kukowska J and Rodgers C. (1988). Mol. 527. Cell. Biol., 8, 3683 ± 3695. Weihua X, Lindner DJ and Kalvakolanu DV. (1997). Proc. Reisman D, Elkind NB, Roy B, Beamon J and Rotter V. Natl. Acad. Sci. USA, 94, 7227 ± 7232. (1993). Cell Growth Di€., 4, 57 ± 65. Xiao Q, Claassen G, Shi J, Adachi S, Sedivy J and Hann SR. Resar LMS, Dolde C, Barrett JF and Dang CV. (1993). Mol. (1998). Genes Dev., 12, 3803 ± 3808. Cell. Biol., 13, 1130 ± 1136. Yancopoulos GD, Nisen PD, Tesfaye A, Kohl NE, Goldfarb Saleh A, Schieltz D, Ting N, McMahon SB, Litch®eld DW, MP and Alt FW. (1985). Proc. Natl. Acad. Sci. USA, 82, Yates-III JR, Lees-Miller SP, Cole MD and Brandl CJ. 5455 ± 5459. (1998). J. Biol. Chem., 273, 26559 ± 26570. Yokoyama K and Imamoto F. (1987). Proc. Natl. Acad. Sci. Sarid J, Halazonetis TD, Murphy W and Leder P. (1987). USA, 84, 7363 ± 7367. Proc. Natl. Acad. Sci. USA, 84, 170 ± 173. Zhang Y, Iratni R, Erdjument-Bromage H, Tempst P and Schwab M, Varmus HE and Bishop JM. (1985). Nature, 316, Reinberg D. (1997). Cell, 89, 357 ± 364. 160 ± 162. Zwicker J, Liu N, Engeland K, Lucibello FC and Muller R. Sears R, Ohtani K and Nevins DR. (1997). Mol. Cell. Biol., (1996). Science, 271, 1595 ± 1596. 17, 5227 ± 5235.