Oncogene (2000) 19, 2836 ± 2845 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc The chicken c-Jun 5' untranslated region directs translation by internal initiation

Anil Sehgal1,2, Joe Briggs1, Janet Rinehart-Kim1, Johnny Basso1,3 and Timothy J Bos*,1

1Department of Microbiology and Molecular Cell Biology, Eastern Virginia Medical School, PO Box 1980, Norfolk, Virginia, VA 23501, USA

The 5' untranslated region (UTR) of the chicken c-jun Because of its role in controlling gene expression, message is exceptionally GC rich and has the potential much e€ort has been expended studying the regulation to form a complex and extremely stable secondary and function of the c-Jun protein. Like other nuclear structure. Because stable RNA secondary structures can oncogenes, c-jun is tightly regulated at a number of serve as obstacles to scanning ribosomes, their presence di€erent levels. Transcriptional regulation is well suggests inecient translation or initiation through documented in response to a wide variety of growth alternate mechanisms. We have examined the role of factors (Lamph et al., 1988; Ryder and Nathans, 1988; the c-jun 5' UTR with respect to its ability to in¯uence Vogt and Bos, 1990). Post-translational regulation of translation both in vitro and in vivo. We ®nd, using Jun activity occurs through speci®c phosphorylation rabbit reticulocyte lysates, that the presence of the c-jun and dephosphorylation events which regulate DNA 5' UTR severely inhibits translation of both homologous binding as well as transcriptional regulatory activity and heterologous genes in vitro. Furthermore, transla- (Adler et al., 1992; Baker et al., 1992; Boyle et al., tional inhibition correlates with the degree of secondary 1991; Franklin et al., 1992; Pulverer et al., 1991; structure exhibited by the 5' UTR. Thus, in the rabbit Smeal et al., 1991, 1992). Activity can also be reticulocyte lysate system, the c-jun 5' UTR likely regulated by a variety of di€erent protein-protein impedes ribosome scanning resulting in inecient interactions with non-leucine zipper transcription translation. In contrast to our results in vitro, the c-jun factors (Bengal et al., 1992; Boise et al., 1993; SchuÈ le 5' UTR does not inhibit translation in a variety of et al., 1990; Stein et al., 1993a,b; Touray et al., 1991; di€erent cell lines suggesting that it may direct an Yang-Yen et al., 1990). alternate mechanism of translational initiation in vivo. The c-jun message is produced from an intronless To distinguish among the alternate mechanisms, we gene (Hattori et al., 1988) and has an interesting generated a series of bicistronic expression plasmids. structure which suggests additional levels of control. In Our results demonstrate that the downstream cistron, in addition to the protein coding region, the c-jun message the bicistronic gene, is expressed to a much higher level contains long 5' and 3' untranslated regions (UTRs). when directly preceded by the c-jun 5' UTR. In addition, The 3' UTR contains several AUUUA motifs, adding inhibition of ribosome scanning on the bicistronic an additional layer of complexity to c-jun regulation at message, through insertion of a synthetic stable hairpin, the level of mRNA stability (Aharon and Schneider, inhibits translation of the ®rst cistron but does not inhibit 1993; Curatola et al., 1995; Savant-Bhonsale and translation of the cistron downstream of the c-jun 5' Cleveland, 1992). The 5' UTR of c-jun is extremely UTR. These results are consistent with a model by which long and GC rich, ranging in size from 301 nucleotides the c-jun message is translated through cap independent in chicken to 974 bases in human (see Table 1). internal initiation. Oncogene (2000) 19, 2836 ± 2845. Typically, messenger RNAs with long GC rich 5' UTRs are translated ineciently by cap dependent Keywords: c-Jun; translation; internal initiation; 5' scanning ribosomes. Translation of these messages UTR often requires alternative mechanisms of translational initiation. Most eukaryotic messenger RNAs contain 5' UTRs Introduction less than 100 bases long and are translated by a mechanism referred to as ribosome scanning (reviewed v-Jun was originally discovered as the oncogenic in Gray and Wickens, 1998; Hershey, 1991; Kozak, component of Avian Sarcoma Virus 17 (Maki et al., 1991a). RNA secondary structures with standard free 1987). The cellular homologue, c-Jun, is a sequence energy values (DG) up to about 730 kcal/mole can speci®c DNA binding protein that binds to promoters easily be negotiated by the scanning 43S ribosome in responsive genes to activate transcription (Angel et subunit, allowing progression until the AUG start al., 1988; Bohmann et al., 1987; Vogt and Bos, 1990). codon in the proper context is found (Kozak, 1989, 1991b). When more stable structures are encountered, the 43S subunit stalls and can not proceed. RNAs that *Correspondence: TJ Bos contain these stable structures will likely not be Current addresses: 2Department of Neurological Surgery, University translated eciently by the ribosome scanning mechan- of California at San Francisco, 1855 Folsom Street, MCB 230, San ism (Kozak, 1989). Interestingly, the 5' UTRs of many Francisco, CA 94103; 3University of Ottawa, Department of Biology, growth factor and proto-oncogene messages are long Gendron Bldg., 550 Cumberland Street, Ottawa, Ontario, Canada, K1N 6N5 and have the potential to form stable secondary Received 7 January 2000; revised 23 March 2000; accepted structures (Kozak, 1991a). A growing number of these 28 March 2000 proteins have been shown to initiate translation Internal initiation of c-Jun translation A Sehgal et al 2837 Table 1 Comparison of 5' untranslated sequences of Jun family presence of motifs which have the potential to play mRNAs across species roles in RNA structural stability as well as RNA- Length Free energy protein interactions. Several other potential structures Species (bases) % GC Accession # kcal/mole within this optimal free energy range are depicted in c-Jun Figure 1c. All of these potential secondary structures Chicken 301 81 M57467 7181.4 are extremely stable with optimal free energy values Quail 354 73 X15547 7173.6 well beyond that shown to inhibit ribosome scanning. Rat 917 63 X17215 7402.1 Secondary structures with free energy values as low as Mouse 614 63 X12740 7255.8 Human 974 65 J04111 7444.6 750 kcal/mole have been shown to inhibit cap dependent translation (Kozak, 1989, 1991b). Transla- Jun B tion of messages with stable secondary structures Rat 289 70 X54686 7138.2 generally occurs through alternative mechanisms such Mouse 287 70 J03236 7127.6 Human 287 74 M29039 7135.1 as internal initiation and ribosome jumping (Gray and Wickens, 1998). From this structural analysis, we Jun D predicted that the full length c-jun message would fall Mouse 121 83 J04509 774.1 into this category. To test this hypothesis, we have Human 174 86 X56681 7103.2 analysed the mechanism of translation of the chicken c- Chicken 165 84 X60063 791.2 jun message both in vitro and in vivo.

Effect of the c-jun 5' UTR on translation from rabbit through alternate mechanisms. For example, c-Myc reticulocyte lysates in vitro (Nanbru et al., 1997; Stoneley et al., 1998), bFGF (Vagner et al., 1995), PDGF-2 (Sella et al., 1999), We utilized an in vitro transcription and translation GRP78/BIP (Macejak and Sarnow, 1991; Sarnow, system to assay the e€ect that the 5' UTR has on 1989), eIF4G (Gan and Rhoads, 1996), Apaf1 translation of c-jun.c-jun cDNA with (pG5'CJ3) and (Coldwell et al., 2000) and VEGF (Akiri et al., 1998; without (pGCJ3) the 5' untranslated region were Stein et al., 1998) are each capable of bypassing the cloned into pGEM4 such that they were under the impediment of RNA secondary structure in the 5' UTR control of the SP-6 promoter. Because 3' untranslated by translation initiation through cap independent sequences of c-jun have been implicated in mRNA internal ribosome entry. This capability extends stability, we prepared templates for transcription such another level of complexity to the regulation of the that transcripts would run o€ shortly after the stop expression of these proteins allowing their expression codon (Figure 2a). RNA levels were assayed by even under conditions where cap dependent ribosome synthesis in the presence of [32P]UTP or [32P]CTP scanning is inhibited. followed by separation on 6% urea-acrylamide gels We have examined the 301 nucleotides of the 5' and quantitated (Figure 2b, bottom panel) by UTR of the chicken c-jun message and found it to phosphorimage analysis. Equivalent values of RNA contain 81% G or C residues. Algorithmic analysis were used in subsequent translation reactions. Transla- predicts that this sequence will form a stable and tion of in vitro transcribed RNA was carried out in complex secondary structure with an optimal standard rabbit reticulocyte lysates in the presence of free energy of 7181.4 kcal/mole. The predicted [35S]methionine. Labeled proteins were then separated stability of this structure far exceeds that which can by SDS ± PAGE and quantitated by phosphorimage be melted by a scanning 43S ribosome subunit, analysis. The results are presented in Figure 2b suggesting that translation will either be extremely (compare lanes 1 and 6). Clearly, the presence of the inecient or proceed through an alternative mechan- 5' UTR has a dramatic inhibitory e€ect on translation ism of initiation. We demonstrate here that c-jun is of c-Jun in rabbit reticulocyte lysates. part of a growing list of important cellular genes whose To examine this e€ect in more detail, we generated a translation is directed through cap independent internal series of deletion mutations within the c-jun 5' UTR ribosome entry. designed to reduce the potential structural complexity and stability. These mutants are diagrammed in Figure 2a and contain UTR sequences with predicted standard free energy values ranging from 7143 to 711 kcal/ Results mole. The in¯uence of these truncated structures on translation of c-Jun is presented in Figure 2b. As Structure of the c-jun 5' UTR predicted, we ®nd that the translational eciency of c- The basic structure of the chicken c-jun message jun does not signi®cantly increase until the complexity (Nishimura and Vogt, 1988) is shown in Figure 1a. of the secondary structure falls to around 711 kcal/ The 5' untranslated region is 301 nucleotides long and mole. These results are consistent with a model by contains 81% G or C residues. The c-jun message also which the c-jun 5' UTR structure inhibits the scanning has a long 3' untranslated region which is AT rich. A ribosome. computer analysis of the 5' UTR using Zuker's mfold version 3.0 (Mathews et al., 1999; Zuker et al., 1999) Effect of the c-Jun 5' UTR on translation of a revealed that this region has the potential to form heterologous message many secondary structures within the optimal free energy range. The structure depicted in Figure 1b The results presented in Figure 2 demonstrate that the represents the most energetically favorable structure. c-jun 5' UTR can in¯uence translation of its own We have annotated this structure to indicate the message. Whether this e€ect is due solely to the 5'

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2838 UTR or requires the presence of downstream sequences demonstrated that the UTR itself was sucient to within the c-jun gene is not clear. We addressed this inhibit translation of a heterologous gene, we used a question by placing the c-jun 5' UTR in front of a similar strategy to examine its in¯uence on translation heterologous gene sequence encoding bacterial chlor- in vivo. We generated plasmids carrying the CAT amphenicol acetyl transferase (CAT) (Figure 3). The reporter gene with and without the c-jun 5' UTR driven translation eciency of CAT in the presence or by the SV40 early promoter (Figure 4a). Each of these absence of the c-jun 5' UTR is presented in Figure plasmids was transfected into primary chicken embryo 3b. These results clearly show that CAT protein ®broblasts (CEF) and assayed for CAT activity (Figure synthesis is markedly reduced when the c-jun 5' UTR 4b). In sharp contrast to our ®ndings in rabbit is present. This indicates that the 5' UTR of c-jun,by reticulocyte lysates, we found that the c-jun 5' UTR itself, is sucient to modulate translation in the rabbit did not have an inhibitory e€ect on CAT expression in reticulocyte lysate in vitro model system. vivo. In fact, we very often detected a stimulation. Similar results were found in two other ®broblast cell lines (RAT 1 and NIH3T3) (Figure 4b) as well as Effect of the c-jun 5' UTR on translation in vivo MCF7 epithelial cells (not shown) demonstrating that Figures 2 and 3 clearly demonstrate a dramatic this e€ect is likely a general phenomenon rather than a negative in¯uence on translation by the c-jun 5' UTR cell or even species speci®c response. in vitro. In these experiments, translation of c-Jun or CAT is almost completely abolished in the presence of The c-jun 5' UTR directs translation from bicistronic the 5' UTR. Paradoxically, c-Jun protein is known, in messenger RNAs vivo, to be expressed and accumulate to high levels in response to a variety of di€erent stimuli. With this in Because of the highly structured and stable nature of mind, we have examined the e€ect of the c-jun 5' UTR the 5' UTR, it is unlikely that this ecient translation on translation in vivo. Because our in vitro results response in vivo occurs through the typical mechanism

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2839 of ribosome scanning. More likely, an alternative constructions with two di€erent reporter genes (beta initiation mechanism is utilized. galactosidase and CAT) (Figure 5). pSV21acCAT A variety of alternate translation initiation mechan- contains the coding sequence for beta galactosidase isms have been described that can be categorized as and CAT in tandem. pSV21ac5'CAT contains the c-jun either 5' cap dependent or cap independent. The cap 5' UTR between the two reporter genes. Transfection dependent mechanisms are variations of the scanning of these plasmids into NIH3T3 cells results in model and include reinitiation or leaky scanning, and expression of the ®rst cistron (lac) to equivalent levels ribosome jumping (Gray and Wickens, 1998). Reinitia- (Figure 6a, lanes 1 and 3). As expected, pSV21acCAT tion is an important component to translational did not express any CAT protein from the second control of messenger RNAs that contain multiple cistron (Figure 6b, lane 1) suggesting that there was no small open reading frames upstream of the main signi®cant read through or leaky scanning occurring protein coding sequence. Ribosome jumping is the from this template. However, when the c-jun 5' UTR process by which an impeded scanning ribosome can was placed between the cistrons, a signi®cant amount bypass a stable hairpin structure and continue scanning of CAT expression was observed (Figure 6b, lane 3) downstream. Thus far, this mechanism has only been suggesting that the 5' UTR either allows read through, described in a handful of viral messages (Latorre et al., jumping or internal initiation. 1998; Yueh and Schneider, 1996). The cap independent To distinguish between these possibilities, we created mechanism is referred to as internal initiation and can additional bicistronic plasmids in which a synthetic proceed even when cap dependent translation is hairpin loop was placed in front of the ®rst cistron inhibited (Belsham and Sonenberg, 1996; Jackson and (Figure 5). Because both leaky scanning and ribosome Kaminski, 1995; Martinez-Salas, 1999). jumping initiate at the cap structure, we reasoned that The internal initiation and ribosome jumping a hairpin before the ®rst cistron would inhibit both of mechanisms allow translation to proceed despite the these mechanisms, resulting not only in a decrease in presence of highly structured untranslated regions. To lac but also CAT expression. However, if the down- test the possibility that the c-jun 5' UTR directs stream cistron is initiated by internal ribosome entry, translation initiation through one of these alternative the upstream hairpin should not inhibit translation of mechanisms, we generated a series of bicistronic gene the downstream cistron (CAT). The synthetic hairpin

Figure 1 The chicken c-jun 5' UTR is predicted to form a very stable secondary structure. (a) Schematic representation of the structure of the chicken c-jun messenger RNA. The hatched box represents the Jun protein coding region. 5' and 3' untranslated regions are indicated. The primary sequence of the 5' UTR is indicated below the schematic. Two potential in frame AUG start codons are indicated at positions 302 and 314. (b) Most energetically favorable predicted secondary structure of the chicken c-jun 5' UTR indicating the presence of sequence motifs potentially involved in RNA-RNA and RNA-protein interactions (N=G, A, T or C; R=G or A). (c) Additional secondary structures (three of possible 27) within the optimal standard free energy range as determined by Zuker's mfold version 3.0. Shown are structures that represent the magnitude of structural diversity within this range

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2840 was designed to have a stable structure with a DGof 763 kcal/mole. Structures with values greater than about 750 kcal/mole have been demonstrated to inhibit cap dependent ribosome scanning (Kozak, 1989, 1991b). Two plasmids were constructed, pSV2 631acCAT and pSV2 631ac5'CAT. These plasmids are identical to their respective parent plasmids, pSV21ac- CAT and pSV21ac5'CAT, with the exception of the hairpin loop inserted 5' of the lac gene (Figure 5). Each of these plasmids was transfected into NIH3T3 cells and assayed as before for both lac and CAT activity (Figure 6). As expected, the presence of the synthetic hairpin loop completely inhibited expression of lac from pSV2 631acCAT as well as pSV2 631ac5'CAT demonstrating inhibition of ribosome scanning from these templates by the hairpin (Figure 6a, lanes 2 and 4). Quite strikingly, pSV2 631ac5'CAT was still able to direct translation from the second cistron (CAT)

Figure 2 The chicken c-jun 5' UTR inhibits c-Jun protein expression in vitro.(a) Schematic representation of 5' UTR deletion mutants and predicted structural stability. (b) SDS ± PAGE of in vitro translated Jun protein produced from in vitro transcribed RNA containing di€erent deletions within the 5' UTR. Lane 1; pG5'CJ3 (full length); lane 2, pG5' (D140 ± 193)CJ3; lane 3, pG5' (D140 ± 249)CJ3; lane 4, pG5' (D1± 193)CJ3; lane 5, pG5' (D1 ± 259)CJ3; lane 6, pGCJ3 (no UTR). Also shown (bottom panel) are in vitro RNA transcription levels (see Materials and methods) to demonstrate equivalent template levels during translation

Figure 4 The c-jun 5' UTR does not inhibit CAT expression in vivo.(a) Schematic of monocistronic CAT reporter genes driven by the SV40 early promoter with (pSV2 5'CAT) or without (pSV2 CAT) the c-jun 5' UTR. (b) Standard CAT assay of extracts from three di€erent cell lines transiently transfected with pSV2 5' CAT (lane 1 in each cell line), pCAT (promoterless CAT reporter ± lane 2 in each cell line), or pSV2 CAT (lane 3 in each cell line)

Figure 5 Schematic of bicistronic reporter genes. Each of the bicistronic genes is driven by the SV40 early promoter. The ®rst Figure 3 The 5' UTR of c-jun inhibits translation of a cistron encodes beta galactosidase (lac) and the second cistron heterologous gene (CAT) in vitro.(a) Schematic of CAT gene encodes chloramphenicol acetyl transferase (CAT). In pSV2lac5' with (pG5' CAT) and without (pG CAT) the c-jun 5' UTR. (b) CAT and pSV2 63lac5'CAT, the c-jun 5' UTR has been inserted SDS ± PAGE of in vitro translated CAT protein from RNA between the lac and CAT genes in the native orientation. pSV2 templates with (lane 1) or without (lane 2) the c-jun 5' UTR. 63lacCAT and pSV2 63lac5'CAT contain an additional synthetic RNA levels from in vitro transcription are also shown in the hairpin with a predicted free energy of 763 kcal/mole just 5' of bottom panel the lac gene

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2841 see a single large transcript from both of these transfections indicating that CAT protein expression from the bicistronic transcript is not due to c-Jun 5' UTR induced nuclease cleavage or alternative splicing. These results demonstrate that the c-jun 5' UTR can direct translation under conditions where cap depen- dent scanning is inhibited and therefore rule out reinitiation and ribosome jumping as mechanisms to explain translation of the downstream cistron. How- ever, our results are consistent with a model by which the c-jun 5' UTR directs translation initiation through cap independent internal ribosome entry.

Structure of other Jun family messenger RNA's Although the experiments presented here were per- formed with the chicken c-Jun 5' UTR, analysis of the potential 5' UTR structural elements of c-Jun from other species suggests that alternate translation initia- tion strategies may not be unique to the chicken message. Table 1 presents the length, GC content and potential structural stability of c-Jun, JunB and JunD from multiple species. Interestingly, all of these genes encode messenger RNAs with long 5' UTRs ranging from 121 nucleotides for mouse JunD to 974 nucleotides for human c-Jun. In addition, the predicted free energy for each of these structures ranges from 774.1 to 7444.6 kcal/mole, again suggesting a signi®cant impediment to traditional ribosome scan- ning while favoring alternative translation initiation.

Discussion

The c-jun message is induced rapidly and transiently in response to a variety of extracellular growth and di€erentiation factors. Because this response occurs in the absence of new protein synthesis, c-Jun is considered an immediate early response protein. As Figure 6 The chicken c-jun 5' UTR can initiate translation of such, c-Jun is under tight regulatory controls. c-Jun bicistronic reporter genes through a mechanism consistent with along with other members of the Jun, Fos and CREB/ internal ribosome entry. NIH3T3 cells were transiently transfected with the bicistronic plasmids described in Figure 5 and assayed ATF families form a complex (AP-1) that speci®cally for beta galactosidase and CAT activities. (a) Assay for beta binds to and regulates the promoters of AP-1 galactosidase activity (®rst cistron) from cell lysates transiently responsive genes. Many of these genes are thought to transfected with pSV2lacCAT (lane 1), pSV2 63lacCAT (lane 2), play critical roles in cell growth and di€erentiation. pSV2lac5'CAT (lane 3), pSV2 63lac5'CAT (lane 4). pCH110 (lane Multiple levels of control are critical for key regulatory 5) and pSV2CAT (lane 6) serve as positive and negative controls respectively. (b) CAT activity (second cistron) from the same proteins. This is illustrated by the fact that deregulated NIH3T3 cell extracts as in (a). pCH110 (lane 5) and pSV2CAT expression of c-Jun will induce oncogenic transforma- (lane 6) serve as negative and positive assay controls. (c) Northern tion in avian and rodent ®broblasts (Bos et al., 1990; blot analysis of bicistronic transcripts. Total RNA from NIH3T3 Schutte et al., 1989). Thus, maintenance of proper cells transiently transfected with pSV2 63lacCAT (2), pSV2 63lac5'CAT (4) and pCH110 (5) was probed with a cDNA to regulation of AP-1 activity is critical in preserving the CAT gene (top panel). The bottom panel demonstrates equal normal cellular homeostasis. c-Jun is regulated at the loading by staining of 28S and 18S RNA with ethidium bromide transcriptional and post-translational levels as well as through a variety of protein-protein interactions. The results presented here describe an additional layer of (Figure 6b, lane 4) even though cap dependent complexity to the regulation of c-Jun, at the level of scanning was inhibited by the synthetic hairpin. translation. Interestingly, the level of CAT expression was actually Most cellular proteins are translated through a elevated when cap dependent translation was inhibited mechanism termed ribosome scanning. Here, transla- by the stem loop. This e€ect has also been seen by tion is initiated after cap recognition by the eIF-4F others (Jackson and Kaminski, 1995) and may re¯ect a complex and subsequent binding by the 43S ribosome localized increase in availability of translation initiation subunit. The eIF-4F complex is required for cap factors. In Figure 6c we examined the bicistronic recognition and contains at least three di€erent transcripts induced from transfection with pSV proteins: eIF-4E (the cap binding protein), eIF4G 631acCAT and pSV 631ac5'CAT by Northern blot and eIF4A (reviewed in: Gray and Wickens, 1998). The analysis using a CAT cDNA probe. As expected, we eIF4E protein is directly involved in binding to the cap

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2842 structure and is often limiting in the cell. eIF4A is a (PCBP2); and an oligo A loop which may serve as part RNA helicase necessary for unwinding secondary of an eIF4G binding site (Belsham and Sonenberg, structure near the cap. eIF4G not only acts as a 1996; Jackson and Kaminski, 1995; Kolupaeva et al., sca€old for eIF4E and 4A, but also interacts with eIF3 1998; Le and Maizel, 1997; Lopez de Quinto and in the 43S ribosome complex which facilitates its Martinez-Salas, 1997; Martinez-Salas, 1999). Signi®- binding to mRNA (Gray and Wickens, 1998). cantly, many of the potential structures resulting from Conditions a€ecting the availability or activity of mfold 3.0 analysis of the chicken c-Jun 5' UTR contain eIF4F result in a general reduction in translation of these cis acting elements including the Y shape, cap dependent messenger RNAs. A variety of di€erent polypyrimidine tract, GNRA and ACCC motifs mechanisms can in¯uence eIF4F activity. Infection (Figure 1b). However, the functional contribution of with picornaviruses such as poliovirus results in each of these in directing internal ribosome entry on cleavage of eIF4G, which uncouples its association the c-Jun mRNA has yet to be determined. with eIF4E and cap recognition (Belsham and In this report we demonstrate that the 5' UTR of c- Sonenberg, 1996; Jackson and Kaminski, 1995). Jun behaves di€erently with respect to translation Adenovirus infection results in dephosphorylation of initiation in vitro and in vivo. This e€ect is not eIF4E which reduces its cap binding anity (Gingras surprising or unusual and is likely due to di€erences and Sonenberg, 1997). Growth arrest, heat shock and in availability of speci®c trans acting factors in the hypoxia cause similar decreases in eIF4F activity either reticulocyte system compared with intact cells. This through reduction of overall levels of eIF4E or by phenomenon has been documented for a number of maintaining it in a dephosphorylated state (Feigenblum di€erent viral and cellular IRESs. In fact, the and Schneider, 1996; Gray and Wickens, 1998; Stein et functional capacity of an IRES can vary dramatically al., 1998). Interestingly, even under conditions in which from one system to another. For example, the IRES cap dependent translation is severely inhibited, protein elements derived from poliovirus, human rhinovirus synthesis from some mRNAs can still occur eciently. and hepatitis A virus direct extremely inecient Some of these messenger RNAs, such as those for the translation in rabbit reticulocyte lysates (Borman et heat shock proteins, require only small amounts of al., 1995; Jackson and Kaminski, 1995; Martinez-Salas, eIF4E and can therefore be selectively translated under 1999). In contrast, IRES elements from encephalomyo- adverse conditions (Jackson and Kaminski, 1995). carditis virus and foot-and-mouth disease virus direct Others, such as those derived from a number of ecient translation in the same system (Borman et al., di€erent viruses, proceed through cap independent 1995; Jackson and Kaminski, 1995). Internal initiation internal ribosome entry (Belsham and Sonenberg, from poliovirus and human rhinovirus, but not 1996; Gray and Wickens, 1998; Jackson and Kaminski, hepatitis A virus IRES elements can be improved with 1995; Martinez-Salas, 1999). A growing list of cellular the addition of Hela cell cytoplasmic extract (Borman genes can also initiate translation by internal ribosome et al., 1995; Jackson and Kaminski, 1995). Interest- entry including GRP78/BiP (Macejak and Sarnow, ingly, hepatitis A translation can be improved with the 1991; Sarnow, 1989), FGF (Vagner et al., 1995), addition of liver cell extracts suggesting the need for PDGF2 (Sella et al., 1999), VEGF (Akiri et al., 1998; tissue speci®c factors to direct internal ribosome entry Stein et al., 1998), eIF4G (Gan and Rhoads, 1996), (Jackson and Kaminski, 1995). Similar di€erences have Apaf1 (Coldwell et al., 2000) and c-Myc (Nanbru et been seen with many of the recently described cellular al., 1997; Stoneley et al., 1998). In this report we have IRES elements. For instance the 5' UTR of FGF-2 is demonstrated that the 5' UTR of the c-jun mRNA can inhibitory in wheat germ extracts but not in rabbit direct translation through internal ribosome entry. reticulocyte lysates (Vagner et al., 1995). In contrast, Thus, c-Jun can be added to this growing list of the 5' UTR of c-myc is inhibitory to translation in both cellular genes that utilize alternative translation wheat germ extracts as well as rabbit reticulocyte strategies as an additional means to control their lysates (Parkin et al., 1988). The di€erences in expression under a variety of di€erent conditions. translation eciency by these IRESs in di€erent Although internal ribosome entry appears to be systems is likely due to variations in the level, need utilized by a divergent group of mRNAs, the exact and anity for speci®c trans acting factors such as the mechanism by which it occurs still remains obscure. La autoantigen and PTB (Belsham et al., 1995; Kim The common requirement for internal entry is a long and Jang, 1999; Martinez-Salas, 1999; Sickinger and and highly structured 5' UTR. The complex folding of Schweizer, 1999). Consistent with this hypothesis, both the UTR creates a structure functionally de®ned as an La and PTB are high in Hela extracts but low in rabbit internal ribosome entry segment (IRES). Interestingly, reticulocyte lysates (Belsham and Sonenberg, 1996). IRESs can vary dramatically in length, primary IRES elements that do not require binding by these sequence and in their functional properties (Jackson (and other) factors would be expected to initiate and Kaminski, 1995; Martinez-Salas, 1999). The cis translation eciently in reticulocyte lysates, whereas and trans elements involved in IRES mediated IRESs that require relatively high levels of these factors translation initiation are just beginning to be identi®ed. would be translated poorly. Cis elements present in some, but not all, IRESs Our results indicate that the c-Jun 5' UTR (Figure 4) include: a Y shaped secondary structure 5' of the AUG can direct ecient translation in three di€erent start codon; a polypyrimidine tract which can serve as ®broblast cell lines. This does not rule out the a recognition site for polypyrimidine binding protein possibility that the eciency of c-Jun translation may (PTB); GNRA sequence motifs (thought to stabilize vary in other specialized cell systems. Interestingly, structure, also found in group I introns) within the dramatic di€erences in IRES driven internal initiation head of one or more loops; ACCC motifs which make have been documented in vivo. For example, the up part of a binding site for poly (rC) binding protein poliovirus IRES works well in Hela and Hep G2 cells

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2843 but not in neuroblastoma or BHK21 cells (Borman et a 1.2 kb BssHII/polI/HindIII fragment of pG5'CJ-3 into al., 1997). The IRES from foot-and-mouth disease pGEM4 digested with SmaI and HindIII. pG5'(D143-193)CJ- virus, however, works well in all four of these cell lines 3 was constructed by digesting pG5'CJ-3 with SacI and BglI. (Borman et al., 1997). Similarly, utilization of IRES The resulting plasmid fragment was religated in the presence mediated translation of PDGF2 is activated only by of a synthetic adapter called ad-1. ad-1 was prepared by annealing two synthetic sequences 5'-TAGCAGAGCT-3' and signals linked to di€erentiation (Sella et al., 1999). The 5'-CTGCTAGCGG-3'. This adapter has compatible cohesive hepatitis A IRES is inecient in many di€erent cell ends for SacI and BglI. In a separate set of reactions pG5'CJ- lines including HepG2 (Borman et al., 1997). In 3 was digested with BglI and BssHII. The resulting plasmid contrast, the c-myc IRES directs ecient translation fragment was religated in the presence of a synthetic adapter in a variety of di€erent cell lines (Nanbru et al., 1997; called ad-2 to yield pG5'(D143-259)CJ-3. ad-2 was prepared Stoneley et al., 1998). Clearly, IRES elements can by annealing two synthetic sequences 5'-TGAGACTA-3' and contribute an additional level of complexity to the 5'-CGCGTAGTCGCAGCGG-3'. This adapter has compa- expression of important regulatory proteins. tible ends for the BglI and BssHII restriction enzymes. Internal ribosome entry has been implicated as a mechanism to allow for translation of messages under Construction of CAT plasmids with and without the jun 5' UTR conditions in which overall cellular protein synthesis is pGEM 4 was digested with HindIII and BamHI and ligated compromised. For example, under hypoxic conditions, with a 1.6 kb fragment containing the chloramphenicol acetyl the angiogenic growth factor VEGF is translated by transferase (CAT) gene to yield pGCAT. The CAT gene internal ribosome entry (Stein et al., 1998). Similarly, c- fragment was obtained by digesting pSV2CAT with HindIII Myc has recently been shown to be translated by and BamHI. pG5'CAT was constructed by digestion of internal ribosome entry during apoptosis (Stoneley et pGCAT with HindIII followed by ligation with the c-jun 5' al., 2000). UTR generated by polymerase chain reaction (PCR) and cut As an immediate early response gene, c-Jun plays a with HindIII. The orientation of the 5'UTR was con®rmed by key regulatory role in a variety of cellular processes Southern blot analysis. The 5'UTR was PCR ampli®ed as including cell proliferation, di€erentiation and apopto- follows: 50 ng of pG5'CJ-3 cut with NcoI was mixed with 200 m dNTPs, 2 mM MgCl , 8% DMSO, 5 mlof106 Vent sis. Interestingly, c-Jun is also activated as part of the M 2 bu€er, 5 ml BSA (1 mg/ml), 0.5 mM Bos-as-4 primer (5'- stress response induced by UV radiation, peroxide, GCTCAAGCTTATAGAATACACGGAATTACT-3') and heat shock, and hypoxia (Bukh et al., 1990; Chihab et 0.45 mM Bos-as-5 primer (5'-ATGGAAGCTTAGAACA- al., 1998; Devary et al., 1991; Diamond et al., 1999; GAGCCCGCGGA-3')ina50ml reaction volume. Both of Gilby et al., 1997; Rupec and Baeuerle, 1995). these primers contain HindIII sites. After boiling 2 min, two Although the mechanisms involved in Jun activation units of Vent polymerase was added and PCR was initiated. by these stress signals are complex, and involve both The conditions for PCR were as follows: Two minutes at transcriptional as well as post-translational events, our 948C, 1 min at 438C and 2 min at 748C for eight cycles. The results demonstrate a mechanism by which c-Jun next 30 cycles were carried out with an annealing temperature translation can be eciently maintained so that it can of 658C instead of 438C. This method was based on a protocol described previously for PCR ampli®cation of very regulate gene activity even under these adverse high GC rich sequences (Dutton et al., 1993). conditions. In vitro transcription and translation pG5'CJ-3, pG5'(D1-193)CJ-3, pG5'(D1-258)CJ-3, pG5'(D143- Materials and methods 193)CJ-3, pG5'(D143-259)CJ-3 and pG5'3'CJ-3 were linear- ized with HindIII. pGCJ-1 was linearized with MluI. Construction of c-jun plasmids with and without the 5' UTR pG5'CAT and pGCAT were linearized with BamHI. In vitro The chicken genomic c-jun gene is contained within a transcription was carried out as follows: 1 mlof1mg/ml 3.2 kb XbaI fragment in pUCGCJ-1. The genomic sequence linearized plasmid was mixed with 6 ml DEPC (diethyl of chicken c-jun was obtained from GenBank accession pyrocarbonate) treated water, 56 transcription bu€er #M57467. The coding region of c-Jun contains two in (Promega), 2 ml of 100 mM dithiothreitol (DTT), 20 units frame and conserved AUG start codons separated by three RNasin, NTP mix 1 (10 ml each of 10 mM ATP, CTP, UTP codons (Figure 1a). For all of our analyses, we have and 1 mlof10mM GTP and 9 ml DEPC water), 2 mlof5mM assumed the ®rst AUG to be the authentic start codon. 7mGpppG cap analog and 20 units of SP6 RNA polymerase The 5' UTR of c-jun was cloned into the in vitro expression or T7 RNA polymerase (pGCAT and pG5'CAT). Incubation plasmid pGCJ-3 by a multi step process. pUCGCJ1 was was at 378C for 30 min. Four ml of NTP mix 2 (10 ml each of ®rst cut with HincII releasing two fragments 3.2 and 3.0 kb 10 mM ATP, CTP, UTP, GTP) was added and reaction in length. The 3.2 kb fragment was isolated and digested continued for another 30 min. For RNA quantitation, 2.5 ml with HindIII. The resulting 1.5 kb fragment was then [32P]UTP (10 mCi/mole) was added. RNA was brie¯y run on digested with Hinf1 releasing three fragments of 1.0, 0.2 a 6% acrylamide gel and subjected to autoradiography or and 0.3 kb in length. The 1.0 kb fragment was treated with phosphorimage analysis. For translation, 1 ml of freshly Klenow DNA polymerase 1 and then digested with NcoI. transcribed RNA product was mixed with 17.5 ml of nuclease The resulting 700 bp fragment was then ligated into an treated rabbit reticulocyte lysate, 0.5 mlof1mM amino acid EcoRI/polI/NcoI treated pGCJ-3 to yield pG5'CJ-3. Con- mix minus methionine and 15 mCi [35S]methionine (1175 Ci/ struction of pGCJ-3 has been described previously (Bos et mMole). Incubation was for 60 min at 308C. Five mlof al., 1990). translated product was mixed with 1 ml of 10 mg/ml RNase and incubated another 15 min at 308C. Five mlof26 sample bu€er was then added (125 mM Tris pH 6.8, 4% SDS and Construction of c-jun 5' UTR mutants 5.7 M 2-mercaptoethanol) to the translated product and pG5'(D1-193)CJ-3 was constructed by ligating a 1.2 kb SacI/ incubated at 908C for 5 min. After 2 min of centrifugation HindIII fragment of pG5'CJ-3 into pGEM4 digested with the at room temperature, 6 ml of protein dye (1 mg/ml same enzymes. pG5'(D1-259)CJ-3 was constructed by ligating bromophenol blue, 62.5 mM Tris pH 6.8 and 50% glycerol)

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2844 was added to the translation product and loaded on to a 10% encoding b-galactosidase. The estimated free energy for the SDS gel for separation followed by autoradiography and/or synthetic stem loop structure is 763 kcal/mole. pSV2 phosphorimage analysis. 63lac5'CAT was prepared by ligation of the 2.2 kb fragment of pSV2lac5'CAT digested with NdeI and BamHI to the 6.4 kb fragment of pSV2 63lacCAT digested with the same Computer analysis enzymes. The secondary structure of the 5' UTR of the c-jun gene, each mutation, and all of the Jun family members was predicted Cell culture, transfections and Northern analysis using Zuker's mfold version 3.0 (www.ibc.wustl.edu/*zuker) (Mathews et al., 1999; Zuker et al., 1999). Analysis was Primary chicken embryo ®broblasts were generated from 10- conducted using the following conditions: 378C, 5% sub- day-old embryos as described previously (Bos et al., 1990). optimality and default parameters for base pairing con- Rat 1 and NIH3T3 cells were grown in Improved MEM Zinc straints. Figures and tables show original, most optimal free Option Richter's Modi®cation (Gibco ± BRL)+10% FBS and energy values unless otherwise stated. Sequences of the 5' 1% penicillin/streptomycin. For transfection, cells were UTR of Jun family members were obtained from the seeded onto six well plates and allowed to reach 50% GENBANK and EMBL databases. con¯uence. Transient transfections were conducted using 3 mg of plasmid DNA and 6 ml of Fugene reagent (Boehringer Mannheim) in serum free media. Media was replaced 8 h Construction of in vivo monocistronic and bicistronic plasmids after transfection with fresh media including serum. Cells The plasmid, pSV2 5' CAT contains the c-jun 5' UTR derived were harvested 48 h post transfection by freeze thaw for from pG5'CAT cut with HindIII and inserted into assay of protein, CAT activity and beta galactosidase pSV2CATBlue digested with HindIII. pSV2CATBlue is activity. CAT assays were performed as previously described identical to pSV2CAT except that the SV40 promoter using 50 ± 70 mg of total protein from transfected cell lysates. enhancer and CAT gene have been cloned into the Bluescript To assay for beta galactosidase activity, 10 ± 14 mg of protein backbone. pSV2lac5'CAT was constructed by ligating the was added to 1 ml of chlorophenolred-b-D-galactopyrano- HindIII to BamHI fragment of pSV2CATBlue to the SgrA-1/ side, 50 mM K2HPO4/1 mM MgCl2 pH 7.8, and incubated at polI/HindIII fragment of pbgal promoter (Clontech) and a 378C for 10 min. Colorimetric changes were read at 578 nm. HindIII partial/polI/BamHI fragment of pSV2 5'CAT. pCH110 (Pharmacia) contains the lac gene under the control pSV2lacCAT was constructed by ligation of a HindIII/ of the SV40 early promoter and was used as a positive BamHI fragment of pSV2CATBlue to the SgrA-1/polI/ control for the assay. For Northern analysis, total RNA was HindIII fragment of pbgal promoter and a complete isolated as described previously (Hadman et al., 1996). The HindIII/polI/BamH1 fragment of pSV2CATBlue. For the CAT probe was constructed by digesting the pCATbasic preparation of pSV2 63lacCAT, the plasmid pSV2lacCAT (Promega) plasmid with XbaI and NcoI. The resulting CAT was linearized with HindIII, treated with calf intestinal fragment was then gel puri®ed and labeled using the Prime-a- alkaline phosphatase (CIP) and gel puri®ed. A synthetic gene labeling system (Promega) following the manufacturer's stem-loop structure was generated using the oligonucleotide protocol. 5'-TAT-ACAT GTAAGCT TGATA TCCCCTTGGACCTC- CGTACCCTGCAGG-3'. This oligonucleotide contains eight terminal nucleotides that are self-complementary, as well as HindIII and EcoRV restriction sites. The oligonucleotide was Acknowledgments boiled for 3 min and allowed to cool on ice for 20 min. It This work was supported in part by a grant from the was then added to a Klenow polI extension reaction at 168C Edmondson Cancer Fund, the Horsley Cancer Research for 60 min, digested with HindIII and gel puri®ed. The Fund, the Je€ress Memorial Trust and NIH R01 CA51982. synthetic oligo was subsequently inserted into the HindIII site We thank Melissa Johnston for excellent technical assis- of pSV2lacCAT immediately upstream of the ®rst cistron tance and Dr Casey Morrow for helpful discussions.

References

Adler V, Franklin CC and Kraft AS. (1992). Proc. Natl. Borman AM, Bailly JL, Girard M and Kean KM. (1995). Acad. Sci. USA, 89, 5341 ± 5345. Nucl. Acids Res., 23, 3656 ± 3663. Aharon T and Schneider RJ. (1993). Mol. Cell. Biol., 13, Borman AM, Le Mercier P, Girard M and Kean KM. (1997). 1971 ± 1980. Nucl. Acids Res., 25, 925 ± 932. Akiri G, Nahari D, Finkelstein Y, Le SY, Elroy-Stein O and Bos TJ, Monteclaro FS, Mitsunobu F, Ball Jr, AR, Chang Levi BZ. (1998). Oncogene, 17, 227 ± 236. CHW, Nishimura T and Vogt PK. (1990). Genes. Dev., 4, Angel P, Allegretto EA, Okino ST, Hattori K, Boyle WJ, 1677 ± 1687. Hunter T and Karin M. (1988). Nature, 332, 166 ± 171. Boyle WJ, Smeal T, De®ze LHK, Angel P, Woodgett JR, Baker SJ, Kerppola TK, Luk D, Vandenberg MT, Marshak Karin M and Hunter T. (1991). Cell, 64, 573 ± 584. DR, Curran T and Abate C. (1992). Mol. Cell. Biol., 12, Bukh A, Martinez-Valdez H, Freedman SJ, Freedman MH 4694 ± 4705. and Cohen A. (1990). J. Immunol., 144, 4835 ± 4840. Belsham GJ and Sonenberg N. (1996). Microbiol Rev., 60, Chihab R, Ferry C, Koziel V, Monin P and Daval JL. (1998). 499 ± 511. Brain Res. Mol. Brain Res., 63, 105 ± 120. Belsham GJ, Sonenberg N and Svitkin YV. (1995). Curr. Coldwell MJ, Mitchell SA, Stoneley M, MacFarlane M and Top. Microbiol. Immunol., 203, 85 ± 98. Willis AE. (2000). Oncogene, 19, 899 ± 905. Bengal E, Ransone L, Scharfmann R, Dwarki VJ, Tapscott Curatola AM, Nadal MS and Schneider RJ. (1995). Mol. SJ, Weintraub H and Verma IM. (1992). Cell, 68, 507 ± Cell. Biol., 15, 6331 ± 6340. 519. Devary Y, Gottlieb RA, Lau LF and Karin M. (1991). Mol. Bohmann D, Bos TJ, Admon A, Nishimura T, Vogt PK and Cell. Biol., 11, 2804 ± 2811. Tjian R. (1987). Science, 238, 1386 ± 1392. Diamond DA, Parsian A, Hunt CR, Lofgren S, Spitz DR, Boise LH, Petryniak B, Mao X, June cH, Wang CY, Goswami PC and Gius D. (1999). J. Biol. Chem., 274, Lindsten T, Bravo R, Kovary K, Leiden JM and 16959 ± 16964. Thompson CB. (1993). Molec. Cell. Biol., 13, 1911 ± 1919.

Oncogene Internal initiation of c-Jun translation A Sehgal et al 2845 Dutton CM, Paynton C and Sommer SS. (1993). Nucleic Rupec RA and Baeuerle PA. (1995). Eur. J. Biochem., 234, Acids Res., 21, 2953 ± 2954. 632 ± 640. Feigenblum D and Schneider RJ. (1996). Mol. Cell. Biol., 16, Ryder K and Nathans D. (1988). Proc. Natl. Acad. Sci. USA, 5450 ± 5457. 85, 8464 ± 8467. Franklin CC, Sanchez V, Wagner F, Woodgett JR and Kraft Sarnow P. (1989). Proc. Natl. Acad. Sci. USA, 86, 5795 ± AS. (1992). Proc. Natl. Acad. Sci. USA, 89, 7247 ± 7251. 5799. Gan W and Rhoads RE. (1996). J. Biol. Chem., 271, 623 ± Savant-Bhonsale S and Cleveland DW. (1992). Genes. Dev., 626. 6, 1927 ± 1939. Gilby KL, Armstrong JN, Currie RW and Robertson HA. SchuÈ le R, Rangarajan P, Kliewer S, Ransone LJ, Bolado J, (1997). Brain Res. Mol. Brain Res., 48, 87 ± 96. Yang N, Verma IM and Evans RM. (1990). Cell, 62, Gingras AC and Sonenberg N. (1997). Virology, 237, 182 ± 1217 ± 1226. 186. Schutte J, Minna JD and Birrer MJ. (1989). Proc. Natl. Acad. Gray NK and Wickens M. (1998). Ann. Rev. Cell. Dev. Biol., Sci. USA, 86, 2257 ± 2261. 14, 399 ± 458. Sella O, Gerlitz G, Le SY and Elroy-Stein O. (1999). Mol. Hadman M, Gabos L, Loo M, Sehgal A and Bos TJ. (1996). Cell. Biol., 19, 5429 ± 5440. Oncogene, 12, 135 ± 142. Sickinger S and Schweizer M. (1999). Biol. Chem., 380, HattoriK,AngelP,LeBeauMMandKarinM.(1988).Proc. 1217 ± 1223. Natl. Acad. Sci. USA, 85, 9148 ± 9152. Smeal T, Binetruy B, Mercola D, Grover-Bardwick A, Hershey JWB. (1991). Ann. Rev. Biochem., 60, 717 ± 755. Heidecker G, Rapp UR and Karin M. (1992). Mol. Cell. Jackson RJ and Kaminski A. (1995). RNA, 1, 985 ± 1000. Biol., 12, 3507 ± 3513. Kim YK and Jang SK. (1999). J. Gen. Virol., 80, 3159 ± 3166. Smeal T, Binetruy B, Mercola DA, Birrer M and Karin M. KolupaevaVG,PestovaTV,HellenCUandShatskyIN. (1991). Nature, 354, 494 ± 496. (1998). J. Biol. Chem., 273, 18599 ± 18604. Stein B, Baldwin JAS, Ballard DW, Greene WC, Angel P and Kozak M. (1989). Mol. Cell. Biol., 9, 5134 ± 5142. Herrlich P. (1993a). EMBO J., 12, 3879 ± 3891. Kozak M. (1991a). J. Cell. Biol., 115, 887 ± 903. Stein B, Cogswell PC and Baldwin JAS. (1993b). Mol. Cell. Kozak M. (1991b). J. Biol. Chem., 266, 19867 ± 19870. Biol., 13, 3964 ± 3974. Lamph WW, Wamsley P, Sassone-Corsi P and Verma IM. Stein I, Itin A, Einat P, Skaliter R, Grossman Z and Keshet (1988). Nature, 334, 629 ± 631. E. (1998). Mol. Cell. Biol., 18, 3112 ± 3119. Latorre P, Kolakofsky D and Curran J. (1998). Mol. Cell. Stoneley M, Chappell SA, Jopling CL, Dickens M, Biol., 18, 5021 ± 5031. MacFarlane M and Willis AE. (2000). Mol. Cell. Biol., Le SY and Maizel Jr, JV. (1997). Nucleic Acids Res., 25, 20, 1162 ± 1169. 362 ± 369. Stoneley M, Paulin FE, Le Quesne JP, Chappell SA and Lopez de Quinto S and Martinez-Salas E. (1997). J. Virol., Willis AE. (1998). Oncogene, 16, 423 ± 428. 71, 4171 ± 4175. Touray M, Ryan F, Jaggi R and Martin F. (1991). Oncogene, Macejak DG and Sarnow P. (1991). Nature, 353, 90 ± 94. 6, 1227 ± 1234. Maki Y, Bos TJ, Davis C, Starbuck M and Vogt PK. (1987). Vagner S, Gensac M-C, Maret A, Bayard F, Amalric F, Prats Proc. Natl. Acad. Sci. USA, 84, 2848 ± 2852. H and Prats A-C. (1995). Mol. Cell. Biol., 15, 35 ± 44. Martinez-Salas E. (1999). Curr. Opin. Biotechnol., 10, 458 ± Vogt PK and Bos TJ. (1990). Adv. Cancer Res., 55, 1 ± 35. 464. Yang-Yen H-F, Chambard J-C, Sun Y-L, Smeal T, Schmidt Mathews DH, Sabina J, Zuker M and Turner DH. (1999). J. TJ, Drouin J and Karin M. (1990). Cell, 62, 1205 ± 1215. Mol. Biol., 288, 911 ± 940. Yueh A and Schneider RJ. (1996). Genes Dev., 10, 1557 ± Nanbru C, Lafon I, Audigier S, Gensac MC, Vagner S, Huez 1567. G and Prats AC. (1997). J. Biol. Chem., 272, 32061 ± Zuker M, Mathews DH and Turner DH. (1999). RNA 32066. Biochemistry and Biotechnology, Barciszewski J and Clark Nishimura T and Vogt PK. (1988). Oncogene, 3, 659 ± 663. BFC (eds). Kluwer Academic Publishers: NATO ASI Parkin N, Darveau A, Nicholson R and Sonenberg N. series, pp. 11 ± 43. (1988). Mol. Cell. Biol., 8, 2875 ± 2883. Pulverer BJ, Dyriakis JM, Avruch J, Nikolakaki E and Woodgett JR. (1991). Nature, 353, 670 ± 674.

Oncogene