The Carboxyl-terminal Tail of Noxa Protein Regulates the Stability of Noxa and Mcl-1^*

Antibodies and Reagents

Antibodies used in this study were anti-Noxa (Imgenex, IMG-349A), anti-multiubiquitin (StressGen Bioreagents Corp., SPA-205), anti-Mcl-1 (Santa Cruz, Sc-819), anti-GFP (Santa Cruz, Sc-9996), and anti-β-actin (Sigma, A5441). Cycloheximide (CHX,⁵ catalog number 357420010) was purchased from Acros Organics (Thermo Fisher Scientific). MG-132 (catalog number 81-5-15) was obtained from American Peptide (Sunnyvale, CA).

Cell Lines and Cell Culture

Regular and retrovirus-infected HeLa cells stably expressing various proteins were maintained in Dulbecco's modified Eagle's medium supplemented with antibiotics and 10% fetal bovine serum.

Plasmid Construction

To generate the transient expression plasmid of wild-type Noxa, the cDNA of human Noxa was PCR-amplified with the addition of a His₆ tag to the N terminus. An XhoI site and an EcoRI site were engineered into the forward and reverse primers, respectively. Following digestion with XhoI and EcoRI, the PCR product was ligated into the XhoI- and EcoRI-digested pcDNA3.1(−) vector (Invitrogen). This construct was used as a template for the generation of Noxa BH3 mutant (L29E), different K/R mutants, and the compound mutants of BH3 and K/R by site-directed mutagenesis. The His₆-tagged ubiquitin plasmid (pMT107) is a gift from Dr. Richard Baer and Dr. Dirk Bohmann. To construct the retroviral expression plasmid for Noxa and its mutants, the Noxa sequences were amplified by PCR, with the forward primer containing an XhoI site followed by a FLAG tag, and the reverse primer containing an EcoRI site. The PCR products were digested with XhoI and EcoRI and then cloned into the XhoI-EcoRI-digested pMSCV-IRES-GFP vector (pMIG, a gift from Dr. Robert Lewis). To construct the retroviral expression plasmid for GFP-Noxa fusion protein (pMIG-GFP-Noxa), human Noxa was first cloned into the pEGFPC3 vector (Clontech) by ligation of the PCR-amplified Noxa sequence into the XhoI and EcoRI sites. Subsequently, the XhoI site in pEGFPC3-Noxa, which is between GFP and Noxa sequence, was destroyed by site-directed mutagenesis. The resulting plasmid was then used as a template to amplify the sequence of GFP-Noxa, with XhoI in the forward primer and HindIII in the reverse primer. The resulting PCR product was then digested by XhoI and HindIII and ligated into the XhoI-HindIII cut pMIG vector. The XhoI-HindIII digestion deleted a large portion of the original GFP sequence within the pMIG vector. All the C-terminal truncation mutants of Noxa, including G-Noxa1–50, G-Noxa1–45, G-Noxa1–39, and G-Noxa1–30, and GFP were cloned into pMIG using the same strategy. N-terminal His-tagged G-Noxa1–50, G-Noxa1–45, G-Noxa1–39, and G-Noxa1–30 were also cloned into pMIG by ligation using XhoI and HindIII sites. The constructs for the N-terminal truncation mutants of Noxa, including pMIG-G-Noxa16–54, pMIG-G-Noxa35–54, pMIG-G-Noxa40–54, and pMIG-G-Noxa45-54, were made by site-directed mutagenesis with pMIG-G-Noxa as the template. The alanine-scanning constructs of Noxa, including pMIG-G-Noxa40–41A, pMIG-G-Noxa42–43A, pMIG-G-Noxa44–45A, pMIG-G-Noxa46–47A, pMIG-G-Noxa48–49A, pMIG-G-Noxa50–51A, and pMIG-G-Noxa52–54A, were made by site-directed mutagenesis based on pMIG-G-Noxa. pMIG-G-Bcl-xL was made by site-directed mutagenesis by replacing Noxa sequence with Bcl-xL sequence. pMIG-G-Bcl-xL/NoxaCT20 was made by site-directed mutagenesis by replacing the sequence of Bcl-xL209–233 with the sequence of Noxa35–54 (NoxaCT20) using pMIG-G-Bcl-xL as the template. pMIG-G-NoxaL29E and pMIG-His₆-G-NoxaL29E were made by site-directed mutagenesis based on their wild-type Noxa constructs. pMIG-His₆-G-Noxa42–43A was made by site-directed mutagenesis using pMIG-His₆-G-Noxa as the template. pMIG-His₆-G-NoxaL29E/42–43A was made by site-directed mutagenesis using pMIG-His₆-G-NoxaL29E as the template. pMIG-G-NoxaCT42–43A was made by site-directed mutagenesis based on pMIG-G-Noxa40–54 (pMIG-G-NoxaCT). The constructs were verified by sequencing.

Transfections

HeLa cells were plated (at a density of 2.5 × 10⁵ cells/35-mm plate) and cultured for 20–24 h before transfection with the Effectene reagents (Qiagen) according to the manufacturer's recommendation. For each transfection in a 35-mm plate, 50 ng of plasmid of interest were used and pcDNA3.1(+) was included in the mixture to maintain the total amount of DNA at 0.5 μg/transfection. Eighteen hours after transfection, cells were collected for Western blot.

Retrovirus Production and Stable Cell Line Generation

All the retroviruses were produced by transfecting the pMIG constructs into the virus packaging cell line 293GP. The details of retrovirus production and transduction into HeLa cells were described elsewhere (14).

Assessment of Protein Stability Using Cycloheximide Chase

HeLa cells or stable cell lines of HeLa cells expressing different proteins were plated in a 35-mm plate at a density of ∼3 × 10⁵/plate 1 day before treatment. Cycloheximide (50 μm) was added for 0–4 h. At different time points, cells were harvested and then lysed in EBC buffer (0.5% Nonidet P-40, 120 mm Tris-HCl, pH 7.5, 120 mm NaCl, 1 mm EDTA) for Western blot.

Nickel Bead Pulldown Assay

HeLa cells with or without overexpression of various proteins were pelleted by centrifugation at 1000 ×g for 4 min, washed once with ice-cold PBS, and lysed in the appropriate volume of EBC buffer with protease inhibitors for 1 h at 4 °C with gentle rotation. The lysates from cell pellet of a 6-cm plate were cleared by centrifugation at 22,000 ×g for 20 min. The supernatant was incubated with 20 μl of nickel-nitrilotriacetic acid-agarose beads (Qiagen) by gentle rotation at 4 °C for 3 h. The beads were washed twice by EBC buffer containing 10 mm imidazole. SDS loading dye with 250 mm imidazole was used to elute proteins from nickel beads.

Degradation of Noxa Is Ubiquitylation-independent and Does Not Require the Association of Noxa with Mcl-1

To study the regulation of the stability of Noxa, we first measured the half-life of Noxa by Western blot following the addition of cycloheximide, an inhibitor of protein synthesis, in HeLa cells. Consistent with earlier findings, we found that Noxa is short-lived with a half-life of 45–60 min, and the degradation of Noxa is completely blocked by the proteasome inhibitor MG-132 (Fig. 1A). Because most proteins are targeted to the proteasome for degradation through ubiquitylation, we tested for the ubiquitylation of Noxa. Cell lysates from HeLa cells transiently expressing polyhistidine-tagged Noxa (His₆-Noxa) were subjected to nickel bead pulldown, and Western blots were probed with an antibody recognizing human Noxa. As shown in Fig. 1B, in addition to the monomeric Noxa, the Noxa antibody detected a ladder of larger proteins, reminiscent of ubiquitylated Noxa. Moreover, two protein bands within the 37–50-kDa range of that ladder were also detected by an antibody against ubiquitin, supporting the conclusion that Noxa is ubiquitylated (Fig. 1B).

As Mcl-1 is complexed with Noxa in the cell (13, 14), it is important to see whether the ubiquitylation of Noxa is dependent on its association with Mcl-1. A BH3 mutant of Noxa (FLAG-NoxaL29E), which lacks the ability to bind Mcl-1 (14), was transiently expressed in HeLa cells and tested for ubiquitylation. As shown in Fig. 1B (bottom panel), unlike wild-type Noxa, NoxaL29E was unable to bind endogenous Mcl-1 in the nickel bead pulldown assay. However, both wild-type Noxa and NoxaL29E displayed a similar pattern of protein species in Western blot using an anti-Noxa antibody. Further, the same two protein bands within the 37–50-kDa range were detected by the anti-ubiquitin antibody in the nickel bead pulldown for both wild-type Noxa and the NoxaL29E mutant, indicating that the ubiquitylation of Noxa is not dependent on its association with Mcl-1. As another test for ubiquitylation, we examined the presence of NoxaL29E among the ubiquitylated proteins. HeLa cells were co-transfected with two plasmids expressing His₆-ubiquitin and FLAG-NoxaL29E, respectively, and were subjected to nickel bead pulldown. The bound fraction, presumably enriched for ubiquitylated proteins, was probed with an anti-Noxa antibody. As shown in Fig. 1C, multiple larger protein species with a pattern similar to that detected in nickel bead pulldown from His₆-Noxa-transfected cells (Fig. 1B) were detected. Together, these results demonstrated that Noxa is efficiently ubiquitylated independently of Mcl-1 association.

We next sought to identify the ubiquitylation site(s) of Noxa through systematic mutagenesis, in which the lysine residues of the wild-type Noxa were changed into arginine in various combinations. His-tagged Noxa and its mutant proteins were expressed in HeLa cells, and the ubiquitylation was monitored by nickel bead pulldown. As shown in Fig. 2A, one Noxa mutant, termed C3KR, with Lys³⁵, Lys⁴¹, and Lys⁴⁸ simultaneously changed to arginine residues, showed a complete loss of ubiquitylation. On the other hand, the ubiquitylation of any other mutant remains, as long as one of the three aforementioned lysine residues is still intact, suggesting that Lys³⁵, Lys⁴¹, and Lys⁴⁸ are all ubiquitylated in vivo.

To investigate whether ubiquitylation and Mcl-1 association are necessary for the degradation of Noxa by the proteasome, Noxa mutants that are defective in ubiquitylation (F-NoxaC3KR) or binding to Mcl-1 (F-NoxaL29E), or both (F-NoxaL29E/C3KR), were cloned into transient expression plasmids as well as the retroviral vectors. As expected, in transient transfection assays, the wild-type Noxa and the C3KR mutant both bind endogenous Mcl-1, whereas the NoxaL29E and NoxaL29E/C3KR mutants both lost the ability to bind Mcl-1 (Fig. 2B). The retroviral plasmids allowed us to establish HeLa cell lines stably expressing these proteins at levels similar to that of the endogenous Noxa. As a precaution against the proapoptotic activity of Noxa due to its overexpression, Noxa or its mutant-expressing cell lines in Fig. 2C were established in the presence of Bcl-2 overexpression. However, we did not detect apoptosis in the stable cell lines of Noxa or its mutants in the absence of Bcl-2 overexpression in subsequent experiments. Because the addition of the FLAG tag causes the protein to migrate slower on the SDS-PAGE, we were able to detect both the exogenous and the endogenous Noxa, with the latter serving as an internal control, using an antibody against Noxa in Western blot. The half-lives of these proteins were measured by cycloheximide-induced degradation. As shown in Fig. 2C, all four FLAG-tagged Noxa proteins showed half-lives of less than an hour, similar to that of endogenous Noxa. These results indicate that the degradation of Noxa is independent of ubiquitylation and is not dependent on its association with Mcl-1.

The C-terminal Tail Is Responsible for the Degradation of Noxa

To investigate the mechanism of the ubiquitylation-independent degradation of Noxa, we carried out a serial truncation analysis of the Noxa protein. As human Noxa contains only 54 a. a., it is difficult to detect the expression of all the truncation mutants. We therefore resorted to GFP-Noxa fusion proteins, which enabled us to detect all the Noxa truncation mutants with an antibody against GFP. Importantly, GFP-Noxa displayed a short half-life similar to that of the endogenous Noxa (Fig. 3A). Most mutants with truncations at the N terminus generated short-lived proteins. In particular, mutants G-Noxa35–54 and G-Noxa40–54, containing only the last 20 and 15 amino acid residues, respectively, even showed a half-life less than 30 min, suggesting a loss of sequences inhibitory to the degradation of Noxa. However, a further truncation (G-Noxa 45–54) resulted in a stable protein, with a half-life longer than 2 h, suggesting that the sequence element critical for the degradation of Noxa is disrupted in this mutant.

We next generated a series of C-terminal truncation mutants. Surprisingly, all the truncation mutants with deletions in the C terminus, including a mutant that has a deletion of only the last 5 amino acid residues, became stable proteins with half-lives longer than 4 h (Fig. 3B). For unknown reasons, the G-Noxa1–45 mutant appears moderately less stable than G-Noxa 1–50 and G-Noxa 1–39. Nonetheless, these results, combined with those from the N-terminal truncation mutants, strongly suggest that the last 15 amino acid residues (40–54 a. a.) may carry a degradation signal (degron) for Noxa.

To map the amino acid residues critical for the degradation of Noxa, an alanine-scanning mutagenesis was performed on the last 15 amino acid residues on the background of full-length G-Noxa. As shown in Fig. 4A, one mutant, G-Noxa42–43A, showed a significantly higher basal level as compared with WT Noxa, suggesting that this mutant became more stable. Indeed, G-Noxa43–43A showed a significantly longer half-life than that of the WT G-Noxa (Fig. 4B). Similar results were obtained when G-Noxa42–43A was expressed in the colorectal cancer HCT116 cells (Fig. 4C). Of note, when Leu⁴² or Leu⁴³ was singly mutated to Ala, the L42A mutant showed a modest increase in half-life, whereas the L43A mutant became much more stable, indicating that although both residues contribute to the degradation, Leu⁴³ plays a more critical role in mediating the degradation of Noxa (Fig. 4D).

To examine the effect of the C-terminal tail mutation on the apoptotic activity of Noxa, we generated stable pools expressing G-Noxa42–43A, G-Noxa, or the BH3-mutant G-NoxaL29E in both HeLa and HCT116 cells by retroviral infection (Fig. 4E). The wild-type Noxa (G-Noxa) and the G-Noxa42–43A mutant displayed similar abilities to cooperate with the Bcl-2/Bcl-xL inhibitor ABT-737 to induce apoptosis in HeLa and HCT116 stable pools (Fig. 4, F and G). In addition, both WT and the G-Noxa42–43A mutant partially localized to the mitochondria (Fig. 4H). Thus, the Leu to Ala mutations on Leu⁴² and Leu⁴³ diminished the degradation of Noxa without affecting the mitochondria localization and function of the protein, suggesting that these two residues in the Noxa C-terminal tail play a critical role in the degradation of Noxa.

Noxa Contains a Degron Sequence

To examine the ability of the C-terminal region of Noxa to serve as a degron, we constructed a Bcl-xL mutant protein, replacing the C-terminal tail of Bcl-xL with that of Noxa (last 20 amino acid residues of Noxa). Although GFP-Bcl-xL showed a long half-life, the GFP-Bcl-xL-NoxaCT chimeric protein became unstable with a half-life of ∼1 h, indicating that the C-terminal region of Noxa contains a functional degron responsible for the short half-life of Noxa (Fig. 5A).

To further examine the role of the C-terminal tail (CT) as the degron of Noxa, we generated a GFP fusion protein (G-NoxaCT) consisting of GFP and the CT of Noxa (the last 15 amino acid residues). When stably expressed in HeLa cells, G-NoxaCT exists at an extremely low level (Fig. 5B). However, this level was greatly elevated following the addition of MG-132, suggesting that, similar to the full-length Noxa, Noxa CT is a labile protein degraded by the proteasome (Fig. 5B). Indeed, this fusion protein displayed a very short half-life of less than 30 min. (Fig. 5C). To test the role of Leu⁴² and Leu⁴³ in the degron function of the CT, these two residues were mutated to two alanine residues to generate G-NoxaCT42–43A. This mutant showed a significantly higher basal level than the wild-type Noxa CT and is less sensitive to stabilization by MG-132, suggesting that the degradation of the C-terminal tail is primarily mediated through leucine residues 42 and 43 (Fig. 5B). Not surprisingly, G-NoxaCT42–43A exhibited significantly longer half-lives than that of the wild-type CT (G-NoxaCT, Fig. 5C). These results strongly support the notion that Noxa C-terminal tail is a bona fide degron sequence.

The C-terminal Tail of Noxa Is Involved in the Regulation of Mcl-1 Stability

Because Mcl-1 is the major binding partner of Noxa in the cell, we examined the effect of the C-terminal tail mutations of Noxa on the level of endogenous Mcl-1 in the cells stably expressing the wild-type Noxa and its truncation mutants. As shown in Fig. 6A, whereas expression of full-length Noxa (G-Noxa) had little effect on Mcl-1, the expression of either G-Noxa1–50 or G-Noxa1–45 elevated the level of endogenous Mcl-1. This result suggested that the loss of an intact C-terminal tail caused the up-regulation of endogenous Mcl-1. However, the expression of Noxa1–39 or Noxa1–30, which both lost the C-terminal tail, failed to elevate the level of Mcl-1. Nickel bead pulldown of the His-tagged Noxa and its C-terminal truncation mutants revealed that full-length Noxa, Noxa1–50, and Noxa 1–45, but not Noxa1–39 and Noxa1–30, were able to bind endogenous Mcl-1. This correlation suggests that the elevation of endogenous Mcl-1 is dependent on the direct association with Mcl-1. Indeed, the BH3 mutation (L29E) completely abolished the interaction with Mcl-1 as well as the ability of full-length Noxa, Noxa1–50, and Noxa1–45 to up-regulate endogenous Mcl-1 (Fig. 6B).

To further test the hypothesis that loss of an intact C-terminal tail is responsible for up-regulating endogenous Mcl-1, we examined the Mcl-1 level of cells stably expressing G-Noxa42–43A. Similar to the truncations (Noxa1–50 and Noxa1–45), the point mutation of the C-terminal tail elevated the level of endogenous Mcl-1 (Fig. 6C). As expected, this effect was completely abolished when the BH3 mutation (L29E) was introduced into this mutant, indicating that the up-regulation of Mcl-1 by the Noxa C-terminal tail mutants requires a physical binding to Mcl-1 (Fig. 6C).

To investigate whether up-regulation of Mcl-1 is through protein stabilization, we measured the half-life of Mcl-1 in the cell lines stably expressing Noxa and its C-terminal tail mutants. As shown in Fig. 6D, whereas Mcl-1 in regular HeLa or G-Noxa-expressing HeLa cells is short-lived with a half-life of less than an hour, its half-life in both G-Noxa42–43A-expressing and G-Noxa42–43D-expressing cells is significantly longer (>2 h). Not surprisingly, when either mutation is combined with the BH3 mutation (Noxa42–43A/L29E or Noxa42–43D/L29E), the half-life of the compound mutants remained long. However, this mutant completely failed to prolong the half-life of endogenous Mcl-1, indicating that the physical interaction between the Noxa C-terminal tail mutant and Mcl-1 is required for the stabilization of Mcl-1 (Fig. 6D). The stabilization of endogenous Mcl-1 was also observed in HCT116 cells stably expressing G-Noxa42–43A (Fig. 6E). Taken together, these results suggest that the Noxa C-terminal tail not only plays a critical role in the degradation of Noxa, but also serves as a determinant of the stability of Mcl-1 because Noxa C-terminal tail mutants stabilize endogenous Mcl-1 through a direct interaction with Mcl-1.

Although mouse Noxa (102 a. a.) and human Noxa (54 a. a.) are different in size, they share a highly conserved C-terminal tail. We therefore examined the role of the C-terminal tail in the stability of full-length mouse Noxa in HeLa cells. Similar to the situation of human Noxa, the removal of the C-terminal tail greatly stabilized mouse Noxa as well as the endogenous Mcl-1 (Fig. 6E), suggesting that the stability of mouse Noxa, similar to that of human Noxa, is controlled by its C-terminal tail. Further, like its human counterpart, the C-terminal tail of mouse Noxa has the capacity to regulate the stability of Mcl-1.

The Ubiquitylation-independent Degradation of Noxa

Consistent with results from a recent study (16), Noxa is found to be efficiently ubiquitylated in this study. Moreover, this ubiquitylation is independent from association with Mcl-1, which is also a ubiquitylated protein. Interestingly, all three lysine residues at positions 35, 41, and 48 seem to be ubiquitylated as the single mutations were unable to abolish the polyubiquitylation. With such an efficient polyubiquitylation, it is surprising that polyubiquitylation is not required for the degradation of Noxa. Noxa therefore joins the growing number of proteins that are degraded in a polyubiquitylation-independent fashion (23, 24). This observation raised two obvious questions. First, how is Noxa recognized by the proteasome? The answer to this question may lie in part in the C-terminal tail, which serves as the degradation signal for Noxa. In most cases of ubiquitin-independent degradation of proteins, an unstructured sequence is involved in the degradation. It is therefore reasonable to predict that the Noxa C-terminal tail is unstructured. Second, what role does polyubiquitylation play in the function and regulation of Noxa? Apparently, polyubiquitylation dramatically increases the size of Noxa as multiple ubiquitins (80 a. a.) are added to Noxa (54 a. a.) through covalent bonds. However, at present, we are not able to observe any noticeable differences between the wild-type Noxa and the ubiquitin-free Noxa mutant (NoxaC3KR) as both localized to the mitochondria and both showed similar binding to Mcl-1 (data not shown).

It is worth noting that Mcl-1 was also shown to be degraded in a ubiquitin-independent fashion (21). However, because endogenous Noxa is complexed with Mcl-1 in vivo, it remains unclear whether the Noxa-Mcl-1 complex can be degraded in a ubiquitylation-independent fashion. In other words, it is still possible that degradation of Noxa-Mcl-1 complex is through the ubiquitylation of Mcl-1. A definitive answer to this question may come from lysine-free mutants of both Noxa and Mcl-1 expressed in cells that are doubly deficient for Noxa and Mcl-1. Regardless, our results demonstrated that Mcl-1-free Noxa is degraded through a ubiquitylation-independent mechanism.

Regulation of Mcl-1 Stability through the Noxa C-terminal Tail

Is endogenous Noxa a regulator of Mcl-1 stability? A recent study found that Mcl-1 stability in Noxa-deficient mouse embryonic fibroblasts remains the same as that in wild-type cells, suggesting that endogenous Noxa is not involved in the regulation of Mcl-1 stability in mice (25). Consistent with this finding, we were unable to detect a significant increase of Mcl-1 level following siRNA knockdown of Noxa in HeLa cells (data not shown). In the current study, expression of Noxa mutants with defects in the C-terminal tail was found to stabilize Mcl-1. Because expression of wild-type Noxa did not show this effect, the mutations in the C-terminal tail are apparently gain-of-function mutations as they stabilized the tail-less Noxa mutants, which bind to Mcl-1 and inhibit the degradation of Mcl-1. Therefore, our results suggest that Noxa C-terminal tail regulates Mcl-1 stability indirectly by regulating the stability of Noxa.

The tail mutants of Noxa apparently stabilize Mcl-1 by direct association as the BH3 mutation of Noxa, which abolishes the interaction with Mcl-1, completely abolished the stabilizing effect. It is possible that this binding competes with the binding of an E3 ligase for Mcl-1. A good candidate for the E3 ligase is Mule, which was shown to bind Mcl-1 through a putative BH3 domain (19). However, in light of the recent observation that a lysine-free Mcl-1 displayed a similar half-life as the wild-type Mcl-1 (21), alternative mechanisms might exist. It is also possible that the C-terminal tail of Noxa serves as the degradation signal for the Noxa-Mcl-1 complex; therefore, the removal of the Noxa tail prolonged the half-life of the Noxa-Mcl-1 complex.

The Mitochondrial Protease Responsible for Noxa Degradation

What is the nature of the protease responsible for the degradation of Noxa? Although the 26 S proteasome is apparently a candidate, the answer becomes less clear when considering the fact that the C-terminal tail of Noxa also contains a mitochondrial targeting sequence (26). Is mitochondrial localization required for the degradation of Noxa? Several Noxa C-terminal truncation mutants that lost the entire mitochondrial targeting sequence become extremely stable. Thus, it appears that mitochondrial localization is required for the degradation of Noxa. However, the L42A/L43A Noxa mutant, which retains mitochondrial localization, becomes much more stable than the wild-type Noxa, indicating that mitochondrial localization per se is not sufficient for degradation. A recent study identified a stress-response system for the degradation of mitochondrial proteins (27). However, this system is dependent on the ubiquitylation of the target proteins on the mitochondria. It is therefore possible that the Noxa C-terminal tail is associated with a ubiquitylated protein, which serves as a mediator for the degradation of Noxa through this system. Nonetheless, the mechanism of Noxa degradation remains unclear. However, with the identification of the C-terminal tail as the degron, it is possible to identify C-terminal tail interacting proteins that may serve as the link between Noxa and the unidentified protease. Alternatively, it is possible that the C-terminal tail of Noxa directly interacts with one or several of the subunits of the proteasome, therefore serving as the direct recruiter of the proteasome.

In summary, we present evidence that the C-terminal tail serves as the degradation signal for Noxa and can function indirectly to regulate the stability of Mcl-1. These results provide new insights into the regulation of stability of the Bcl-2 family proteins, which play important roles in apoptosis and drug resistance in cancer cells.