Novel ubiquitin-independent nucleolar c-Myc degradation pathway mediated by antizyme 2

The proto-oncogene c-Myc encodes a short-lived protein c-Myc that regulates various cellular processes including cell growth, differentiation and apoptosis. Degradation of c-Myc is catalyzed by the proteasome and requires phosphorylation of Thr-58 for ubiquitination by E3 ubiquitin ligase, Fbxw7/ FBW7. Here we show that a polyamine regulatory protein, antizyme 2 (AZ2), interacts with c-Myc in the nucleus and nucleolus, to accelerate proteasome-mediated c-Myc degradation without ubiquitination or Thr-58 phosphorylation. Polyamines, the inducer of AZ2, also destabilize c-Myc in an AZ2-dependent manner. Knockdown of AZ2 by small interfering RNA (siRNA) increases nucleolar c-Myc and also cellular pre-rRNA whose synthesis is promoted by c-Myc. AZ2-dependent c-Myc degradation likely operates under specific conditions such as glucose deprivation or hypoxia. These findings reveal the targeting mechanism for nucleolar ubiquitin-independent c-Myc degradation.

HA-c-Myc. An immunoprecipitation assay with the swapped tags brought about essentially the same result (Fig. 1b). We further performed an in vitro immunoprecipitation assay using HA-AZ2, HA-c-Myc and HA-ODC affinity purified from 293-F cells (Fig. S1) with anti-HA antibody. The binding between c-Myc and AZ2 were also confirmed in vitro (Fig. 1c).
c-Myc is known to localize in the nucleus and nucleoli 1 . Nuclear c-Myc binds to MYC-associated factor X (MAX) forming a heterodimer through the leucine zipper and the complex binds to E-Box sequences in the promoters of various target genes 20 . Nucleolar c-Myc plays a key role in regulating ribosome biogenesis 21 . Since AZ2 is mainly localized in the nucleus 17 , we tested whether AZ2 colocalizes with c-Myc in the nucleus and/or nucleoli. AZ2 fused with enhanced cyan fluorescent protein (ECFP) and c-Myc fused with enhanced yellow fluorescent protein (EYFP) were expressed in the human pancreas carcinoma-derived cell line, Panc-1 cells and then observed under fluorescent microscopy. When expressed alone, ECFP-AZ2 dominantly distributed in the nucleus in 40% of the cells and in both cytoplasm and nucleus in the reminder (60%), whereas EYFP-c-Myc was distributed in the nucleus in more than 80% of the cells (Fig. 2a, panels and bar graphs). When ECFP-AZ2 and EYFP-c-Myc were expressed together, they colocalized in the nucleus in more than 80% of the cells (Fig. 2b, panels and bar graph). We next tested the effect of the proteasome inhibitor MG132 that is known to cause nucleolar accumulation of c-Myc 22 . As shown in Fig. 2c, when individually expressed, not only EYFP-c-Myc but also ECFP-AZ2, were largely shifted to the nucleoli in 80% of the cells 5 h after addition of MG132. In about 10% of the cells, EYFP-c-Myc was aggregated in the nucleus (N Agg in bar graph, photograph not shown). Nucleolar localization of AZ2 and c-Myc was confirmed by the nucleolar marker, Fiblillarin (Fib, orange). When ECFP-AZ2 and EYFP-c-Myc were coexpressed in the presence of MG132, the two proteins were colocalized in the nucleoli in 70% of the cells (Fig. 2d). Notably, aggregated EYFP-c-Myc was not observed when ECFP-AZ2 was coexpressed. The nucleolar accumulation of proteins by MG132 is not a general phenomenon since ODC, which is known to be degraded by proteasome, stayed in the cytosol even in the presence of MG132 ( Fig. S2 and bar graph). To elucidate the influence of endogenous c-Myc on the localization of AZ2, Panc-1 cells expressing HA-AZ2 were treated with c-Myc or control siRNA with or without MG132 and then immunostained with anti-HA antibody and fluorescent secondary antibody. In the cells treated with control siRNA, 30% of HA-AZ2 was distributed in the nucleus and 70% was distributed in both nucleus and cytoplasm (Fig. 2e, si-Cont -MG and bar graph). MG132 shifted the localization of AZ2 to the nucleoli in 90% of the cells (Fig. 2e, si-Cont + MG and bar graph) consistent with Fig. 2c. Knockdown of c-Myc dramatically changed the HA-AZ2 distribution to the cytoplasm both in the presence and absence of MG132 (Fig. 2e, si-c-Myc -MG and + MG and bar graph, methods). These results indicate that the nuclear and nucleolar localization of AZ2 is dependent of c-Myc.

AZ2 accelerates proteasomal degradation of c-Myc in a manner independent of ubiquitination.
A distinctive function of AZs is that its binding to ODC triggers ODC degradation by the proteasome without ubiquitination 10,14,15 . It has also been reported that AZ1 interacts with some proteins other than ODC to accelerate their degradation [23][24][25] . Therefore, we tested whether AZ2 accelerates c-Myc degradation in cells. AZ2 binds Cell lysates were immunoprecipitated with anti-FLAG antibody (IP: FLAG) and the resulting precipitates as well as the original cell lysates (input), were subjected to immunoblot analysis with anti-HA or anti-FLAG antibody. The experiment was repeated three times (b) The above experiment was performed with swapped tags (HA-AZ1, HA-AZ2, HA-ODC and FLAG-c-Myc). Immunoprecipitaion was performed with anti-FLAG antibody and c-Myc bound proteins were detected with anti-HA antibody. Expressed protein levels in cell lysates were checked by immunoblotting with anti-HA antibody. Experiments were repeated three times. (c) In vitro immunoprecipitation assay was performed using Human HA-AZ2, HA-c-Myc or HA-ODC purified from 293-F cells. Purified proteins were mixed in M-PER buffer, and after overnight incubation at 4 °C, HA-c-Myc was immunoprecipitated by anti-c-Myc antibody and c-Myc bound protein was detected by immunoblotting using anti-c-Myc antibody. Detailed protocol for in vitro immunoprecipitation assay indicated in Methods. to and activates, ATP citrate lyase without affecting its stability 26 , but no protein other than ODC is known to be destabilized by AZ2. HA-tagged c-Myc (HA-c-Myc) was expressed in 293-F cells alone or together with HA-AZ2, and after addition of cyclohexamide, c-Myc levels were measured by immunoblotting using anti-HA antibody. As shown in Fig. 3a HA-AZ2 clearly accelerated degradation of HA-c-Myc whereas HA-AZ1 had no effect. We next examined the effect of the polyamine putrescine, an inducer of AZ2 (Fig. 3b). With pretreatment of cells with putrescine for 60 min before addition of cycloheximide, degradation of endogenous c-Myc was accelerated, and the acceleration was inhibited by MG132. This indicates that the degradation was catalyzed by the proteasome. It is noteworthy that when putrescine and cycloheximide were simultaneously added, c-Myc stability was unaffected, suggesting that the effect of putrescine is not direct but via inducing a protein factor, likely AZ2 ( Fig. 3b) We performed AZ2-mediated degradation assays in the presence or absence of the lysosome inhibitors, leupeptin and chloroquine (data not shown) and the proteasome inhibitor, MG132 (Fig. S4). However, AZ2-mediated c-Myc degradation was inhibited only by MG132. To confirm the involvement of AZ2 in the putrescine-induced destabilization of c-Myc, we performed a knockdown of AZ2 using siRNA. Activities of the siRNAs were confirmed beforehand ( Fig. S5a-c). In putrescine-treated 293-F cells, AZ2 siRNA significantly stabilized c-Myc compared to control siRNA (Fig. 3c). In contrast, AZ1 siRNA does not stabilize c-Myc. We observed that AZ1 siRNA increased putrescine and spermidine levels up to approximately 24-and 1.8-fold, respectively, and AZ2 siRNA slightly increased putrescine and spermidine levels up to approximately 2-and 1.5-fold, respectively in 293-F cells (Fig. S6). Essentially identical results for AZ2 siRNA were obtained in Panc-1 cells, although the destabilization of c-Myc by AZ1 siRNA is more evident (Fig. 3d). We suspect that in the cells treated with AZ1 siRNA, increased polyamines induced AZ2, which caused acceleration of c-Myc degradation. Next, we compared the effect of knockdown of AZ2 and E3 ubiquitin ligase Fbxw7. In both 293-F cells (result not shown) and Panc-1 cells (Fig. 3d) Fbxw7 siRNA resulted in suppression of c-Myc degradation to the same extent as AZ2 knockdown. To further confirm c-Myc destabilization activity of AZ2 in vitro, we performed a degradation assay of 35 S-Methionine-labeled c-Myc (and ODC as a control) in rabbit reticulocyte lysates supplemented with an ATP regeneration system with or without, AZs (Fig. S7). As previously reported 14,16 , AZ1 accelerated ODC degradation in an energy-dependent and proteasome dependent manner. Meanwhile, AZ1 failed to accelerate c-Myc degradation in vitro confirming the result in cells (Fig. 3a). Under the same condition, AZ2 did not accelerate the degradation of ODC or c-Myc. Thus, the difference in AZ2's ability to accelerate degradation of target proteins between in vivo and in vitro is a common feature for both ODC and c-Myc.
Sequential phosphorylation of Ser-62 and Thr-58 is a prerequisite for the ubiquitination of c-Myc with E3 ubiquitin ligase Fbxw7 recognizing the phosphorylated Thr-58 residue of c-Myc 5 . We examined degradation of HA-tagged c-Myc in which the two phosphorylation site residues were substituted with alanine (HA-c-Myc (T58A/S62A)) in 293-F cells (Fig. 3e). HA-tagged wild-type c-Myc was destabilized when coexpressed with either AZ2 or Fbw7. The mutant HA-c-Myc (T58A/S62A) was not destabilized by Fbw7 as previously reported 5 , but still destabilized by AZ2. Acceleration of degradation of HA-c-Myc (T58A/S62A) by AZ2 was also observed in the human osteosarcoma cell line U2OS (Fig. S8). To test whether AZ2-mediated c-Myc degradation is independent of ubiquitination, a cell-permeable inhibitor of E1 ubiquitin-activating enzyme, PYR41, was added to 293-F cells. PYR41 partially inhibited c-Myc degradation, but expression of AZ2 accelerated c-Myc degradation even in the presence of PYR41 (Fig. 3f). These results indicate that AZ2-madiated c-Myc degradation by the proteasome is ubiquitin independent.
AZ2 contributes to pre-rRNA synthesis through the control of nucleolar c-Myc level. Nucleolar c-Myc plays a key role in positively regulating ribosomal RNA (rRNA) synthesis 21 , with its requisite for nucleolar localization being promoted by a nucleolar protein nucleophosmin 1 (NPM1) 27 . Overexpression of NPM1 increases the nucleolar proportion of c-Myc and stimulates c-Myc-mediated rRNA synthesis whereas knockdown of NPM1 represses it. That NPM1 also destabilizes wild type and notably a T58A mutant of c-Myc is consistent with the finding that c-Myc is mainly degraded in the nucleolus 27 . Therefore, it is reasonable to hypothesize that AZ2 is involved in the NPM1-mediated regulation of c-Myc. To address this possibility, we examined the effect of knockdown of AZ2 or NPM1 on the nucleolar localization of c-Myc and rRNA biosynthesis. Treating Panc-1 cells with AZ2 or NPM1 siRNAs increased cellular c-Myc level two-fold (Fig. 4a,b). However, the cellular distribution of c-Myc after siRNA treatment displays a sharp contrast; the proportion of cells with nucleolar c-Myc doubled with AZ2 knockdown whereas it decreased to one-fourth with NPM1 knockdown (Fig. 4c,d). Expression of pre-rRNA that reflects rRNA synthesis, was analyzed by qRT-PCR with two probes complementary to the 5′ external transcribed spacer sequences of 47 S pre-rRNA (ETS-1 and ETS-2). The expression was also increased 2-to 3-fold in cells treated with AZ2 siRNA whereas it was decreased to a half with NPM1 siRNA (Fig. 4e).
antibody, AlexaFluor555 (orange)-conjugated anti-rabbit IgG as described in methods. N Agg , nuclear dominant distribution with aggregated forms. (d) Panc-1 cells were transiently transfected with both ECFP-AZ2 and EYFP-c-Myc, and after 24 h, cells were treated with 20 µM MG132 for 5 h. Subcellular distribution of these proteins were analyzed as in (c). (e) Panc-1 cells were transfected with c-Myc siRNA or control siRNA. After 24 h, the cells were transfected with HA-AZ2 (human) and incubated for further 24 h. Then the cells were treated with MG132 for 5 h and immunostained with anti-HA antibody and secondary antibody conjugated with AlexaFluor 555. Monochrome images of AZ2 and nuclei were colored in orange and cyan, respectively. A phase-contrast image was added to confirm the location of nucleoli (siCont + MG). Bar graphs represent quantification of the cells with subcellular localization of the proteins. C, cytosolic distribution. Bar graphs on the right of each images represent quantification of 50 cells (a-d) or 100 cells (e). Data (a-e) shown represent the mean ± SD calculated from three independent experiments. Scale bars, 20 μm.  28 . The mechanism of this downregulation is different from the previously reported HIF1α-induced mechanism that requires phosphorylation at Thr-58 following its ubiquitination by Fbxw7 29 . Since the hypoxic condition reportedly increases cellular polyamine levels 30 , we hypothesized that AZ2 might mediate c-Myc degradation under the glucose-free and/or hypoxic conditions. To test this hypothesis, Panc-1 cells were cultured under hypoxic (1% O 2 ) and/or glucose-free conditions. Cellular hypoxic stress was confirmed by checking the induction of HIF1α protein (Fig. S9). When cells were cultured under the normoxic condition, the c-Myc level was slightly increased until 24 h and then decreased to 70% until 48 h, and under the glucose-free or both glucose-free and hypoxic condition, c-Myc was decreased to 40% and 50%, respectively until 24 h, and 20% and 35%, respectively until 48 h (Fig. 5a,b). Under the hypoxic condition with glucose-containing medium, c-Myc was increased to 150% until 10 h and then decreased to 60% until 48 h. The downregulation of c-Myc under these conditions except for normoxia was inhibited by the treatment of AZ2 or Fbxw7 siRNA for 48 h in Panc-1 cells (Fig. 5c). AZ1 siRNA does not inhibit c-Myc downregulation. Inversely further downregulation of c-Myc was observed only in the hypoxic condition and not in a control (Fig. 5c bottom bar graph). We measured polyamine concentration under these conditions to confirm the increase of cellular polyamine levels in Panc-1 cells (Fig. S10). Under the glucose-free condition, putrescine was increased five-fold until 10 h and spermidine was gradually increased 2.5-fold until 48 h but spermine was not changed. Under the hypoxic condition with glucose, putrescine was initially decreased, and then that increased two-fold until 48 h, and spermine was gradually increased 1.5-fold until 48 h, but inversely spermidine was decreased gradually. Under both glucose-free and hypoxic condition, only putrescine was increased 2.8-fold until 24 h, but spermidine was gradually decreased like hypoxic condition, and spermine was not changed. These results provide evidence that at least glucose-free or hypoxic conditions are environments in which AZ2 is induced and where it exerts its function by accelerating c-Myc degradation. Thus AZ2-madiated c-Myc downregulation is likely to be induced by stress condition like glucose-free or hypoxic condition (Fig. 6).

Discussion
In the present study, we show for the first time that AZ2 interacts with c-Myc and accelerates c-Myc degradation by proteasomes (Figs 1 and 3). AZ2 colocalizes with c-Myc in the nucleus and nucleolus, and both proteins accumulate in the nucleoli in the presence of a proteasome inhibitor (Fig. 2). Prior work has shown that the E3 ubiquitin ligase Fbxw7 interacts with, and polyubiquitinates, c-Myc to promote its degradation 5 . Further, a non-ubiquitinated mutant of c-Myc can be degraded by the proteasome in the nucleolus 27 . These results suggest the existence of a local ubiquitin-independent degradation pathway for c-Myc 27 . Our results that AZ2 destabilizes the mutant c-Myc (T58A) and accelerates degradation of c-Myc in the presence of the E1 inhibitor (Fig. 3e,f) provide strong evidence that ubiquitin-independent c-Myc degradation is mediated by AZ2. Both pathways (Fig. 6) seem to contribute to regulating c-Myc degradation since knockdown of AZ2 and Fbxw7 by siRNA inhibit c-Myc degradation to the similar extent (Fig. 3d). However, we cannot exclude the possibility that additional factors may be required for AZ2-madiated c-Myc degradation because in our in vitro degradation assay, AZ2 purified from bacteria could not degrade c-Myc (Fig. S8). This may mean that the nucleolar localization of c-Myc and AZ2 is essential for AZ2-mediated c-Myc degradation. Actually, knockdown of AZ2 significantly increased the level of pre-rRNA by increasing nucleolus-localizing c-Myc (Fig. 4e). This suggests that AZ2 is likely involved in regulating ribosome biogenesis through c-Myc degradation. Further studies are needed to reveal the detailed mechanism.
We also demonstrated that downregulation of c-Myc under glucose-deprivation and/or hypoxic conditions, is partially mediated by AZ2 (Fig. 5c). It has been proposed that growth deceleration through c-Myc downregulation under such conditions may be a survival strategy of cancer cells 28 . In addition, under the glucose-free condition, rapid downregulation of c-Myc and rapid increase of putrescine and spermidine were observed (Fig. 5b,c  and S10). This may also mean that AZ2-mediated c-Myc degradation is involved in malnutrition and starvation.
ODC, the key enzyme for polyamine biosynthesis, is a known transcriptional target of c-Myc 31 . Regulation of c-Myc by AZ2 and polyamines is a novel connection between the MYC signaling and polyamine metabolism, and might be a potential therapeutic target for controlling cancer. Since the molecular architecture of c-Myc is quite similar to MYCN 20 , which is critical for normal brain development, we speculate that AZ2 also controls degradation of MYCN. This could explain why AZ2-expression is correlated with the survival of neuroblastoma patients 19 .  (e) Panc-1 cells were treated with either control or AZ2 or NPM1 siRNA for 48 h and total RNA was prepared from the cells. qRT-PCR was performed using two probes of 47S pre-rRNA (ETS-1 and ETS-2) as described in Methods. Relative pre-rRNA expressions were represented with bar graph Data in b, d and e represent the mean ± SD calculated from three independent experiments. *P < 0.05, **P < 0.01 versus the control (t test). Plasmid constructions. Human cDNAs for c-Myc, Fbxw7, ODC and NPM1 were amplified from the human brain cDNA library (Clontech) by polymerase chain reaction (PCR). Human AZ1ΔT and AZ2ΔT cDNAs were prepared as described for the mouse cDNAs 17 . These cDNAs were inserted into the EcoRI and KpnI restriction sites of pCMV-HA vector (Clontech) to generate pCMV-HA-h-c-Myc, pCMV-HA-Fbxw7, pCMV-HA-h-ODC, pCMV-HA-h-NPM1, pCMV-HA-h-AZ1, pCMV-HA-h-AZ2, respectively. Mouse c-Myc and FBW7 (Fbxw7) cDNA were amplified from the mouse kidney cDNA library (Clontech) by PCR and were inserted into the EcoRI and BamHI restriction sites of pCMV-HA vector to generate pCMV-HA-c-Myc and pCMV-HA-FBW7, respectively. Mouse c-Myc cDNA was inserted into the EcoRI/BamHI restriction sites of pEYFP vector to generate pEYFP-c-Myc. Subcellular localization were analyzed using plasmid pEYFP-c-Myc, pECFP-AZ2 17 , pCMV-HA-h-AZ2 and pCMV-HA-h-ODC. Plasmid p3xFLAG-ODC 17 was used for expression of FLAG-tagged mouse ODC. To generate plasmids p3xFLAG-AZ1, p3xFLAG-AZ2, and p3xFLAG-c-Myc, DNA fragments encoding mouse AZ1ΔT, AZ2ΔT, and c-Myc were amplified by PCR and inserted into EcoRI/XbaI sites of the p3xFLAG-CMV-7.1 vector (Sigma). cDNAs for the phosphorylation site mutants of mouse (T58A/ S62A) and human (T58A) were amplified by PCR with primers corresponding to the mutant sequences and inserted into EcoRI/BamHI sites of pCMV-HA vector to generate pCMV-HA-c-Myc (T58A/S62A) and pCMV-HA-h-c-Myc (T58A), respectively. All constructs were verified by sequencing with an ABI PRISM 3700 sequencer and Big Dye terminator v3.1 cycle-sequencing kit (ABI). All plasmids for transfection were purified by Plasmid Midi Kit (Qiagen).

Methods
Antibodies and immunoblot analyses. Anti-HA polyclonal and monoclonal antibodies were purchased from Medical & Biological Laboratories (MBL) and Cell Signaling Technology, respectively. Anti-FLAG antibody (M2) (Sigma). Anti-c-Myc rabbit monoclonal antibodies were purchased from Epitomics (c-Myc N-term) and abcam (Y69, ab32072) Anti-fibrillarin monoclonal antibody and anti-β-Actin polyclonal antibody were purchased from Cell Signaling Technology. Cells were washed twice in phosphate-buffered saline (PBS) and suspended in CelLytic M Cell Lysis Reagent (Sigma) supplemented with Protease Inhibitor Cocktail Set II (Calbiochem). The cell suspension was sonicated with a small size homogenizer Sonifier 250D (Branson) for 15 s (output 6) and centrifuged at 13,000 × g at 4 °C for 20 min. Protein concentrations of the supernatants were determined using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). The supernatants containing 50-70 μg of protein were separated by SDS-PAGE and protein bands were electrotransferred to Hybond-P (GE Healthcare) or Immobilon-P PVDF membrane (Millipore). Immunodetection was carried out with anti-HA and anti-FLAG NaCl, 2 mM EDTA, 1% NP-40, 0.1% sodium deoxycholate, 0.1% SDS, and 50 mM sodium fluoride). HA-tagged c-Myc and endogenous c-Myc were detected by immunoblot analysis using anti-HA polyclonal antibody and anti-c-Myc antibody, respectively. In the case of adherent Panc-1 or U2OS cells, the cells t were cultured in φ35 mm plastic dish up to 80% confluence and were transfected with the plasmids mentioned above.
Measurement of cellular polyamines. Cells were washed twice with ice-cold PBS and suspended in RIPA buffer supplemented with protease inhibitor cocktail set II (Merck Millipore). Cell lysates were placed on ice for 15 min, sonicated for 12 s, and centrifuged at 15,000 rpm for 25 min at 4 °C. Aliquots of supernatants were used to determine protein concentrations, and the rest were used for the measurement of polyamines. Trichloroacetic acid was added at a final concentration of 4% to the cell lysates, vortexed, and centrifuged for 20 min at 4 °C. The supernatants were filtrated by Millex-LH 0.45 μm (Merck Millipore) and subjected to HPLC analysis using Shimadzu LC Solution System (LC-20AT, SIL-20AC, CTO-20A, and RF-10AXL) equipped with a cation exchange column Shim-Pack ISC-05 (Shimadzu) 32 . Polyamine concentrations were calculated from each polyamine peak area compared with the standards (putrescine, spermidine and spermine).