c-Jun NH2-terminal Kinases Target the Ubiquitination of Their Associated Transcription Factors*

Regulatory proteins are often ubiquitinated, depending on their phosphorylation status as well as on their association with ancillary proteins that serve as adapters of the ubiquitination machinery. We previously demonstrated that c-Jun is targeted for ubiquitination by its association with inactive c-Jun NH2-terminal kinase (JNK). Phosphorylation by activated JNK protects c-Jun from ubiquitination, thus by prolonging its half-life. In the study reported here, we determined the ability of JNK to target ubiquitination of its other substrates (Elk1 and activating transcription factor 2 (ATF2)) and associated proteins (ATF2 and JunB). We demonstrate that phosphorylation by JNK protects ATF2, but not Elk1, from JNK-targeted ubiquitination. We also show that association of inactive JNK with JunB or ATF2 is necessary to target them for ubiquitination. Unlike its targeting of c-Jun, JNK requires additional cellular components, yet to be identified, to target the ubiquitination of ATF2. Elk1 is phosphorylated by JNK, but JNK neither associates with nor targets Elk1 for ubiquitination. The implications for the dual role of JNK in the regulation of ubiquitination and stability of c-Jun, ATF2, and JunB in normally growingversus stressed cells are discussed.

The cellular response to stress activates early response proteins by both transcription and post-translational modifications which dictate the cell's ability to undergo cell cycle arrest for DNA damage repair or to initiate programmed cell death. Among the stress-modulated factors that contribute to the cell's ability to cope with stress are c-Jun and ATF2, 1 both of which are activated by their NH 2 -terminal phosphorylation via stress-activated protein kinases (c-Jun NH 2 -terminal kinases; JNK) (1)(2)(3). JNKs are proline-directed serine/threonine kinases, which are activated by a wide variety of stimuli, including physical and chemical DNA-damaging agents and inhibitors of protein synthesis as well as heat and osmotic shock (reviewed in Kyriakis and Avruch (4)). Different forms of stress utilize alternate cellular pathways for JNK activation (5). For example, JNK activation by UV irradiation requires their phosphorylation by the upstream kinase, mitogen-activated protein-kinase kinase 4 (6), the association of JNK with p21 ras (7), the presence of nuclear DNA lesions (8,9), and inactivation of a redox-sensitive inhibitor. 2 JNK were first identified and named as a Jun-associated kinases (11), reflecting their strong hydrophobic interaction with c-Jun. JNK-c-Jun association is ATP-independent and is required for efficient ATP-dependent phosphorylation of c-Jun at flanking phosphoacceptor sites (Ser 63 and Ser 73 ). The mechanism by which JNK phosphorylation confers transcriptional activities of c-Jun remains largely unknown.
One of the key mechanisms for regulating protein's activity is tight control of its stability. Many regulatory proteins are selectively degraded by the proteasome pathway at specific phases of cell growth. Polyubiquitination, i.e. covalent attachment of multiple ubiquitin residues to ⑀-lysil amino groups of lysine, serves as a marker for proteasome recognition (reviewed in Hochstrasser (12)). The ubiquitination process is regulated by several mechanisms, including degradation of inhibitors, processing of inactive precursors, and stabilization of activated proteins. For example, activation of NFB requires the ubiquitination and degradation of its inhibitor IB as well as the processing of its precursor p105 (13). Conversely, it is the DNA damaged-induced stabilization of the tumor suppressor protein p53 that acquires its activities (14).
Central to JNK's association with c-Jun is the ␦ domain of c-Jun (amino acids 30 -57), which is deleted in its oncogenic counterpart v-Jun. The ␦ domain is also essential for c-Jun ubiquitination, which explains the mechanism underlying the greater stability of v-Jun as compared with its cellular homologue (15). Using an in vitro model system, we previously demonstrated that c-Jun is targeted for ubiquitination by association with JNK. However, phosphorylation of c-Jun on Ser 73 by JNK is sufficient to protect c-Jun from ubiquitination, resulting in a prolonged half-life (16). The dual activity of JNK in targeting c-Jun ubiquitination via physical association and in protecting it from entering this pathway via phosphorylation points to the role of JNK in controlling c-Jun's stability in cells exposed to environmental stress or inflammatory cytokines. In light of finding phosphorylation-dependent targeting of c-Jun ubiquitination, in the present study, we have compared JNK target ubiquitination of its substrates and associated protein (ATF2, c-Jun) with nonassociated substrate (Elk1) and associated non-substrate (JunB). Our results provide the foundation for the model in which (i) JNK-targeted ubiquitination requires tight association and (ii) the degree of targeting is affected by the extent of phosphorylation on JNK-associated protein. (15), c-Jun his ⌬1-72 (16) and Elk1 his (17) were previously described. JunB open reading frame was amplified by PCR using the wild type JunB mammalian expression vector (18) as a template and cloned into the pET15b vector (Novagen) at the NdeI site. A BamHI digest of the same amplification product has been cloned into pET15b at the BamHI site, providing the JunB his construct, which lacks the first 44 amino acids (JunB his ⌬1-44 ). Full-length ATF2 as well as ATF2 with mutated JNK phosphoacceptor sites Thr 3 Ala 69 and Ala 71 open reading frames were amplified by PCR using pECE-ATF2 plasmids (4,19) as templates followed by unidirectionally cloning them into pET15b at NdeI/BamHI sites, resulting in ATF2 his and ATF2 his ⌬69,71 constructs, respectively. The ATF2 mutant lacking JNK binding site (ATF2 his ⌬40 -66 ) has been created using a QuickChange site-directed mutagenesis kit (Stratagene). An HA-tagged ubiquitin encoding construct was generated by PCR-mediated cloning. The sequence encoding ASYPYDVDPYASLSR followed by the second codon of ubiquitin open reading frame was used as a 5Ј primer for PCR amplification and cloned unidirectionally into pET15b at NdeI/BamHI sites. Open reading frames of all final constructs were verified by dideoxy sequencing (Sequenase kit, U. S. Biochemical Corp.).

Plasmids and Recombinant Proteins-Constructs encoding c-Jun his
Histidine-fusion proteins were expressed in the BL21(D:E3)3pLysS bacterial strain and purified by affinity chromatography using nickel resins under denaturing conditions, as recommended by the manufacturer (Qiagen). Proteins attached to the beads were refolded by excessive (20 volumes) column washings with mixtures of 8 M urea in sodium phosphate buffer (pH 8.0) and renaturation buffer (Tris-HCl, pH 7.8, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 10% EtOH, and 0.2% Nonidet P-40) at the subsequent ratios of 3:1, 1:1, 1:3, and 1:7. The columns were then washed three times with 20 volumes of renaturation buffer, followed by two additional washes using the same buffer without alcohol. The beads were then washed with 2 ϫ storage buffer (40 mM HEPES, pH 7.6, 150 mM NaCl, 10 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 0.2% Nonidet P-40) to block the remaining nonoccupied nickel binding sites, which are not protected by bound proteins. Bead-bound proteins were then resuspended in 50% glycerol and stored at Ϫ20°C.
HA-tagged ubiquitin (Ub-HA)-bound beads were washed three more times (rather than resuspended in glycerol) with thrombin cleavage buffer (40 mM Tris-HCl, pH 8.5, 150 mM NaCl, 2.5 mM CaCl 2 ) and incubated in 10 volumes of the same buffer with thrombin (Sigma; 2 units/mg of recombinant protein) for 16 h at 20°C. Resins were pelleted, and the supernatant containing Ub-HA was incubated at 90°C for 15 min, chilled on ice, and cleared by centrifugation at 15,000 ϫ g for 20 min at 4°C. The resulting supernatant was concentrated and washed in double distilled water using Ultra-free-15 centrifugation units (Sigma) with 5000 cutoff membranes to obtain pure Ub-HA (verified by silver staining of SDS-PAGE gel and immunoblotting with antibodies against ubiquitin and HA).
Cells, Lysates, and JNK Purification-4A-3T3 and NIH-3T3 mouse fibroblasts were grown in Dulbecco's modified Eagle's medium supplemented with 10% calf serum (Life Technologies, Inc.) and antibiotics at 37°C and 5% CO 2 . Calpain inhibitor LLM (Sigma) and proteasome inhibitor MG132 (Peptide International Co.) were added to the cells at 50 and 10 M (respectively) in Me 2 SO (less than 0.1% of medium volume) 16 h before harvesting. UV exposure (50 J/m 2 ) or sham irradiation was performed as described previously (16). Rabbit blood enriched with reticulocytes was purchased (Pel-Freeze). Cell lysates and reticulocyte lysates (RL) were prepared as described previously (16). Since RL were found to contain trace amounts of JNK they were immunodepleted of JNK prior to their use (16).
For JNK purification, 600 mg of protein extract from 4A-3T3 cells were prepared from UV-irradiated 4A-3T3 cells (60 J/m 2 ) 45 min after the treatment and subjected to the purification procedure as described previously (11). Briefly, the protein extract was loaded on a sizing column (Sepharose 6B; Pharmacia Biotech Inc.) and active fractions (measured in the solid phase kinase assay) (11) between 30 and 70 kDa were pooled, preincubated with glutathione S-transferase-bound beads, and loaded on the glutathione S-transferase-c-Jun (amino acids 5-89) affinity columns. After extensive washes with kinase buffer, the proteins were eluted with 3% n-octyl ␤-D-glucopyranoside (Sigma), dialyzed against kinase buffer, and loaded onto a phenyl Sepharose column (Pharmacia). The bound material was eluted with 0.2 M (NH 4 ) 2 SO 4 and separated on a SuperX gel filtration column (Pharmacia). The middle active fraction containing ϳ80% of JNK2 and 20% of JNK1 (revealed by immunoblotting) was used in in vitro ubiquitination assays (ϳ0.1 g/ assay). The substrates were phosphorylated by means of the purified JNK2 (20 ng/reaction) or active bacterially expressed JNK (Biomol; 400 ng/reaction) in the solid phase kinase reaction as described elsewhere (16).
In Vitro Ubiquitination Assay-Previously we established an in vitro system for studying the ubiquitination of c-Jun (16). In this system, after preincubation of recombinant c-Jun with whole cell extract, the unbound proteins are washed off and the targeting effects of c-Junbound proteins on its ubiquitination by rabbit RL are monitored. The extent of ubiquitination is reflected by the intensity of the multi-ubiquitin chain that appears as a smear of ubiquitin immunoreactive material produced from the position of the substrate to the top of the gel.
Using this system, we demonstrated that binding of cellular proteins from rat fibroblasts to recombinant c-Jun increases its ubiquitination. Immunodepletion of JNK reduced the targeting activity of protein extracts, which could be restored by adding immunopurified JNK (16). However, in the absence of any fibroblast proteins, the purified form of JNK was capable of mediating only a marginal increase in c-Jun ubiquitination. The latter has been attributed to low sensitivity of our detection system which relied on antibodies to ubiquitin.
To improve the sensitivity of ubiquitination detection, we generated a construct that allows expression and purification of Ub-HA. The NH 2 -terminal fusion of HA peptide was shown not to interfere with formation of polyubiquitin chains in vivo (15), and yet it enabled tracking ubiquitination of different substrates with highly specific and sensitive anti-HA antibody.
Fifty micrograms of whole cell lysates from NIH-3T3 cells (immunodepleted with normal rabbit serum or antibody against JNK as described previously (16)) or 0.1 g of purified JNK were incubated on ice with bacterially expressed substrates (1-5 g) bound to nickel beads for 45 min. After extensive washes (four times with 1 ml of kinase buffer) (16), the substrate-bound beads were equilibrated with 1 ϫ ubiquitination buffer (50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 0.5 mM dithiothreitol, 2 mM NaF, and 3 mM okadaic acid) and incubated in the same buffer supplemented with 2 mM ATP, 10 mM creatine phosphate, 0.02 unit of creatine phosphokinase, 2 g of Ub-HA, 1.5 mM ATP␥S (Sigma) and 33% RL (v/v) in a total volume of 30 ml at 30°C for 5 min. Ubiquitin-aldehyde, synthesized as described previously (21), was added at 1 M final concentration as indicated under "Results." The reaction was stopped by adding 0.5 ml of 8 M urea in sodium phosphate buffer (pH 6.3) with 0.1% of Nonidet P-40. The beads were washed three times with the stop-buffer and once with phosphate-buffered saline supplemented with 0.5% of Triton X-100, and the protein moiety was eluted with Laemmli sample buffer at 100°C. Samples were resolved on 8% SDS-PAGE and electrotransferred onto a nitrocellulose filter. When anti-ubiquitin antibody was used, a polyvinylidene difluoride membrane served as a filter. Nitrocellulose filters were boiled in double distilled water for 10 min, blocked with 5% nonfat milk, and probed with HA11 antibody. After their detection via chemiluminescence (ECL, Amersham Corp.) the blots were stripped and reprobed with antibody against the specific substrate, followed by alkaline phosphatase detection to ensure equal loading of the substrate.
In Vitro Binding Assay-To analyze the association of JNK with JunB and JunB ⌬1-44 , 250 g of lysates obtained from UV-irradiated 4A-3T3 cells were incubated with NTA bead-bound JunB proteins for 45 min on ice. After four washes with kinase buffer, proteins were eluted by boiling in Laemmli sample buffer, separated on 10% SDS-PAGE, and transferred onto a nitrocellulose filter. The filter was probed with antibodies to JNK (clone 666, PharMingen) and reprobed with polyclonal anti-JunB (Santa Cruz) antibody.

JNK Targeting of c-Jun Ubiquitination and Stability-Pre-
incubation of c-Jun his with JNK purified from 4A-3T3 mouse fibroblasts led to substantial increase in c-Jun his ubiquitination by RL immunodepleted of JNK ( Fig. 1; compare lanes 1  and 2). A further increase was noted when the whole cellular extract, immunodepleted with normal rabbit serum, was added as a source of targeting proteins (Fig. 1, lane 3). Immunodepletion of whole cellular extract with antibody to JNK substantially decreased its ability to target c-Jun his ubiquitination; adding purified JNK to the JNK-depleted protein extract restored the original level of c-Jun ubiquitination (Fig. 1). The addition of ubiquitin aldehyde to the ubiquitination reaction did not change the pattern of results (data not shown), suggesting that addition and removal of JNK affected the conjugation of ubiquitin rather than isopeptidase activity. As negative control nickel resins were incubated with protein lysates of uninduced bacterial strain BL21(D;E)3pLysS and purified, as c-Jun his did not exhibit any HA-detectable smear, providing evidence of the substrate specificity of the reaction (Fig. 1, lane  NTA). The same results were observed when an immunoblot from a parallel experiment was probed with anti-ubiquitin monoclonal antibody (not shown).
Since the physical association between JNK and c-Jun targets the latter for ubiquitination, their in vivo interaction is expected to be unstable because of c-Jun degradation. To modulate the steady state level of the c-Jun-JNK complex, mouse fibroblasts were pretreated with potent proteasome inhibitor MG132 (20). Fig. 2A demonstrates that pretreatment with MG132 substantially increased the amount of c-Jun which could be co-immunoprecipitated with antibody against JNK. MG132 treatment did not affect the JNK level (Fig. 2B) but increased the amount of c-Jun measured in the whole cell extracts (Fig. 2A). These findings suggest that the c-Jun-JNK complex in vivo is a target for proteasome activity. UV irradiation of mouse fibroblast cells neither affected the JNK-c-Jun association nor led to retardation of the electrophoretic mobility of c-Jun bound to JNK (Fig. 2A). These data provide further support for the role of JNK as a targeting molecule in c-Jun ubiquitination and, therefore, in determining its stability.
JNK Does Not Target ELK-1 Ubiquitination-In addition to its ability to bind and phosphorylate c-Jun, JNK is known to associate with JunB and ATF2 and to phosphorylate Elk1 and ATF2. We therefore examined the possible involvement of JNK in the regulation of ATF2, Elk1, and JunB ubiquitination. In all cases, the His-tagged substrates were used in our in vitro ubiquitination assay. To study Elk-1 ubiquitination we used a bacterially expressed histidine-tagged Elk-1 protein, which was previously shown to be a functional sequence specific DNA binding protein (17). As is evident from the data presented in Fig. 3, Elk1 is efficiently ubiquitinated by RL, even in the absence of targeting molecules. Preincubation with either purified JNK or with whole cell lysate did not alter the extent of ubiquitination (Fig. 3). These findings suggest that Elk1, which is not capable of association with JNK (17), cannot be targeted for ubiquitination by JNK (purified or in the content of the cell lysate). The ability of RL to mediate a high degree of Elk1 ubiquitination suggests that RL provides all necessary components for Elk1 ubiquitination, including enzymes of the ubiquitination machinery and targeting molecule(s), which were not depleted by the antibodies to JNK (16). As JNK phosphorylation of c-Jun protects it from subsequent targeting for ubiquitination (16,22), we tested whether Elk1 phosphorylation affects its degree of ubiquitination. Unlike c-Jun, extensive phosphorylation of Elk1 by JNK (data not shown) did not alter its ubiquitination (Fig. 3). JNK Targets JunB Ubiquitination-JunB preserves a ␦ domain-like sequence within its NH 2 terminus, thus enabling JNK binding (17,23). Basal levels of JunB his ubiquitination by RL were higher than those found with c-Jun (data not shown). As in the instance of c-Jun, preincubation with JNK or with whole cellular extract increased the extent of JunB his ubiquitination (Fig. 4A, compare lanes 1-3). Whole cell extract immunodepleted of JNK mediated the decreased extent of JunB his ubiquitination, whereas addition of a purified form of JNK to these extracts restored the original degree of ubiquitination (Fig. 4A, lanes 4 and 5).
To further confirm JNK's role in targeting JunB ubiquitination, we performed experiments using JunB his ⌬1-44 , which lacks the first 44 amino acids. This mutant lacks the ability to associate with JNK as suggested by available data (23) and evidenced by an in vitro binding assay (Fig. 4B). Ubiquitination of JunB his ⌬1-44 was severely impaired. Neither JNK nor cell extract could increase JunB his ⌬1-44 ubiquitination (Fig. 4C). Although JNK binds JunB, it cannot phosphorylate this transcription factor because of the absence of phosphoacceptor sites  (17,23). Indeed, preincubation of JunB with JNK and ATP in a solid phase kinase reaction followed by JNK removal (with 3% n-octyl ␤-D-glucopyranoside as described previously (16)) did not result in incorporation of [ 32 P]phosphate into JunB nor did it yield any protection from subsequent ubiquitination (data not shown).
JNK Targeting of ATF2 Ubiquitination Is Phosphorylationdependent-Transcription factor ATF2 can associate with JNK and is a substrate for JNK-mediated phosphorylation. Surprisingly, preincubation of ATF2 his with JNK2 purified from 4A-3T3 cells did not target ubiquitination of this recombinant protein (Fig. 5A, compare lanes 2 and 3). Addition of cell extract as a source of targeting molecules led to a clear increase in the extent of ATF2 ubiquitination (Fig. 5, lane 4). Immunodepletion of this protein extract with antibody to JNK substantially decreased ATF2 his ubiquitination. Conversely, reconstituting this protein extract with purified JNK restored the extent of ATF2 ubiquitination (Fig. 5, lanes 5 and 6).
To further support the role of JNK in ATF2 ubiquitination, we performed experiments using ATF2 his ⌬40 -60 , which cannot bind JNK (19,24). Although ATF2 his ⌬40 -60 exhibits some basal ubiquitination by RL, it cannot be increased by whole cell extract (Fig. A, lanes 7 and 8).
Phosphorylation of ATF2 his by JNK leads to a modest yet highly reproducible increase in the extent of its basal ubiquitination (Fig. 5B, lane 1 versus 2). However, such phosphorylation prevents an increase in the degree of ATF2 his ubiquitination by mouse fibroblast protein extracts (compare lanes 3 and  4). Additional studies to confirm the protective effect of ATF2 phosphorylation by JNK in ATF2 ubiquitination utilized an ATF2 mutant in which JNK phosphorylation sites (Thr 69 and Thr 71 ) were replaced with Ala residues. Phosphorylation of this mutant by JNK was not capable of preventing whole cell extract-targeted ubiquitination (Fig. 5B; compare lanes 3 and 4  versus 7 and 8).
ATF2 and JunB Proteins Associated with JNK Are Substrates for Degradation by Proteasomes-Treatment of mouse fibroblasts with proteasome inhibitor MG132 (but not with calpain inhibitor LLM) increased the amount of JunB co-immunoprecipitated with anti-JNK antibody (Fig. 6A). Blocking the proteasome pathway did not alter the total amount of JunB, suggesting that only a small portion of JunB is bound to JNK and susceptible for JNK-targeted ubiquitination and proteasome-mediated degradation. We cannot rule out the possibility that calcium-dependent proteases play a role in a JunB degradation in the JNK-independent manner.
Immunoblotting analysis of the same experiment with antibodies to ATF2 revealed multiple splicing variants of ATF2 and the cross-reactive ATFa (Fig. 6B) which is also known to bind JNK (25) (Fig. 6B). While the amount of full-length ATF2 (molecular mass, 68 kDa) slightly increased after MG132 treat- ment, its association with JNK was noticed only in the immunoprecipitates from MG132-treated cells when the blot was overexposed (not shown). The 41.5-kDa splicing version of mouse ATF2, which is a constitutively active transcription factor (26), was sensitive to proteasome inhibition, especially when complexed with JNK (Fig. 6B). These data demonstrate that JNK-associated JunB and ATF2 are substrates for a proteasome pathway. DISCUSSION Post-translational modifications of preexisting transcription factors play a central role in the immediate cellular response to damage and stress. Members of the AP1 and ATF families are among the stress-activated transcription factors that we studied in the present investigation. Phosphorylation of c-Jun by JNK correlates with its transactivation (27,28), providing the mechanistic link between JNK signaling and the induction of de novo expression of stress-responsive genes. Since the transactivating potential of v-Jun, which lacks the ␦ domain, seems to be independent of phosphorylation (29,30), it may be attributed to the increased stability that is due to its lack of JNK association and targeted ubiquitination (15,16). That association of JNK with c-Jun via the ␦ domain impairs this protein's ability to undergo transactivation in nonstressed cells, which has been previously proposed and is consistent with our model (31).
The data presented here, together with our previous results, suggest that JNK actively participates in the regulation of c-Jun's stability and, therefore, availability. In nonstressed cells, JNK functions as a targeting molecule for ubiquitination, an activity which depends not on the fact that it is a kinase, but on its ability to associate tightly with c-Jun. This association attracts the enzymes of ubiquitination machinery to c-Jun, thereby marking it for a proteasome-dependent degradation. The necessity of JNK for ubiquitination (and degradation) is directly supported by the data presented here.
JNK regulates c-Jun ubiquitination depending on (i) JNK association with c-Jun and (ii) c-Jun phosphorylation status. The fact that c-Jun is protected from ubiquitination upon being phosphorylated by JNK (16,22) prompted us to test the role of JNK in ubiquitination of its associated substrate ATF2 (19), its nonassociated substrate Elk1 (17) and associated yet non-substrate JunB (17,23). JNK was incapable of targeting the ubiquitination of Elk1 (Fig. 3). Moreover, phosphorylation of Elk1 by JNK did not protect this transcription factor from ubiquitination. These observations suggest that JNK requires a physical association to regulate ubiquitination.
JNK participates in the targeting of JunB for ubiquitination through its association with this protein (Fig. 4), albeit in a non-phosphorylation-dependent manner since JunB is not phosphorylated by JNK (17, 23). The JunB mutant, which lacks the JNK binding domain (Fig. 4B) (23), is poorly ubiquitinated (Fig. 4C). Interestingly, the inability to bind JNK was also documented for another member of the Jun family, JunD (17,23). JunD, which does not have a JNK binding domain, is poorly ubiquitinated compared with c-Jun (32). Together, these data strengthen JNK's role in ubiquitination of its associated proteins.
ATF2 is another member of the bZIP transcription factor superfamily which participates in the cellular response to stress/damage (19,24,33,34). The ATF2-c-Jun heterodimers were shown to mediate transactivation from the UV-responsive element (35) the motif which regulates various stress-responsive genes (i.e. c-Jun, ERCC3) (36). JNK-mediated phosphorylation of ATF2 at Thr 69 and Thr 71 has been shown to induce its transactivation potential (19,24,33). As for c-Jun, the flanking hydrophobic domain (amino acids 40 -66) was found to interact with JNK and to facilitate ATF2 phosphorylation (19,24). Here we demonstrate that although JNK alone does not target ATF2 for ubiquitination, its presence in the cellular lysate is required for such targeting (Fig. 5A). The ATF2 mutant, which lacks amino acids 40 -66, cannot be targeted for ubiquitination, possibly because of its lack of association with JNK. Thus, JNK association with ATF2 is necessary but not sufficient for targeting ATF2 ubiquitination. The additional required factors are yet to be identified.
Phosphorylation of c-Jun by JNK was shown to facilitate the recruitment of cyclic AMP response element-binding protein (37), which links cyclic AMP response element-dependent transactivation by cyclic AMP response element-binding protein with the transcription machinery (37,38). For ATF2, NH 2terminal phosphorylation is thought to abrogate the self-inhibitory intramolecular interaction (39,40). Common to both proteins is likely to be the change in their conformation due to phosphorylation, which would result in weaker association with cellular adapters and/or ubiquitin machinery enzymes.
Change in conformation would also coincide with greater stability of these proteins, as shown for their ␦ domain-deleted counterparts. The activity of various proteins is often limited by their availability, as shown for another JNK-associated protein, p53 (41). We suggest that stabilization of c-Jun and ATF2 through their protection from ubiquitination by JNKmediated phosphorylation provides an important mechanism for their activation.
In all, the emerging model from our studies suggests that JNK targeting for ubiquitination requires tight interaction with its associated protein. Phosphorylation of the associated protein decreases JNK targeting capacity, as shown for c-Jun and ATF2, probably due to altered conformation of the associated protein, which is likely to affect this binding affinity. Ubiquitination of c-Jun, JunB, and ATF2 targeted by JNK is expected to play a central role in rendering transcription factors such as these inactive through their rapid degradation in nondamaged cells. We hypothesize that JNK targeting ubiquitination of ATF2, which regulates the expression of c-Jun, as of c-Jun itself, is a functionally important mechanism that maintains a balanced expression of key regulatory proteins in normally proliferating cells. The activation of these proteins via phosphorylation and gained stability, in response to stress, has been implicated as an early signal in apoptosis (10).