Degradation of certain inhibitor of apoptosis proteins (IAPs) appears to be critical in the initiation of apoptosis, but the factors that regulate their degradation in mammalian cells are unknown. Nrdp1/FLRF is a RING finger‐containing ubiquitin ligase that catalyzes degradation of the EGF receptor family member, ErbB3. We show here that Nrdp1 associates with BRUCE/apollon, a 530 kDa membrane‐associated IAP, which contains a ubiquitin‐carrier protein (E2) domain. In the presence of an exogenous E2, UbcH5c, purified Nrdp1 catalyzes BRUCE ubiquitination. In vivo, overexpression of Nrdp1 promotes ubiquitination and proteasomal degradation of BRUCE. In many cell types, apoptotic stimuli induce proteasomal degradation of BRUCE (but not of XIAP or c‐IAP1), and decreasing Nrdp1 levels by RNA interference reduces this loss of BRUCE. Furthermore, decreasing BRUCE content by RNA interference or overexpression of Nrdp1 promotes apoptosis. Thus, BRUCE normally inhibits apoptosis, and Nrdp1 can be important in the initiation of apoptosis by catalyzing ubiquitination and degradation of BRUCE.
The primary mediators of apoptosis (or programmed cell death) are caspases, cytosolic cysteine proteases that, once activated, initiate an irreversible cascade of events resulting in rapid cell death (Cryns and Yuan, 1998; Thornberry and Lazebnik, 1998). The only known cellular caspase inhibitors are the inhibitor of apoptosis proteins (IAPs), which are characterized by the presence of one to three tandem baculoviral IAP repeats (BIRs). Overexpression of almost all the known IAPs suppresses apoptosis. Several human IAPs (e.g., XIAP, c‐IAP1, c‐IAP2, and NAIP) and Drosophila IAP1 directly bind and inhibit caspases (Kaiser et al, 1998; Hawkins et al, 1999; Salvesen and Duckett, 2002; Wilson et al, 2002; Shiozaki et al, 2003).
In addition to BIR domain, certain IAPs contain a domain that is implicated in ubiquitination. Ubiquitination is the primary mechanism for targeting proteins for rapid degradation by the 26S proteasome (Coux et al, 1996; Hershko and Ciechanover, 1998). Conjugation of a chain of ubiquitin (Ub) molecules to the substrate involves the sequential actions of the Ub‐activating enzyme (E1), a Ub‐carrier protein (E2), and a Ub‐protein ligase (E3). The E3 binds the substrate and catalyzes the transfer of the activated Ub from a specific E2 to the substrate. Several IAPs, including c‐IAP1, XIAP, and Drosophila IAP1, contain a RING finger domain, a characteristic feature of a large family of E3s (Joazeiro and Weissman, 2000). In certain cells, these IAPs catalyze ubiquitination and degradation of themselves and other proteins important for apoptosis, such as caspase 3 and TRAF2 (Yang et al, 2000; Suzuki et al, 2001b; Li et al, 2002; Palaga and Osborne, 2002). In addition, certain IAPs can promote ubiquitination of the RHG motif‐containing proteins (e.g., Rpr in Drosophila or Smac/DIABLO in mammals), which interact with and inhibit several IAPs (Du et al, 2000; Salvesen and Duckett, 2002; Hu and Yang, 2003; Olson et al, 2003).
BRUCE/apollon is a huge membrane‐associated protein (530 kDa) that bears one BIR domain at its N‐terminal region. It is exceptional in containing a C‐terminal E2 motif, which can form thioester bonds with Ub (Hauser et al, 1998; Chen et al, 1999). BRUCE has also been proposed to function as an E3, since some E3s can form thioester bonds with Ub (Hauser et al, 1998; Jesenberger and Jentsch, 2002). Interestingly, BRUCE is upregulated in certain brain cancers (gliomas) that are resistant to certain DNA‐damaging agents, and antisense oligonucleotides against BRUCE enhance the extent of apoptosis induced by these agents (Chen et al, 1999). It remains uncertain whether BRUCE reduces the damage to DNA or inhibits apoptosis. However, overexpression of Drosophila BRUCE has been shown to suppress cell death induced by the proapoptotic factors, Rpr and Grim (Vernooy et al, 2002).
Nrdp1 (referred to as FLRF in mice) is a Ub ligase that promotes ubiquitination and proteasomal degradation of ErbB3, a member of the epidermal growth factor (EGF) receptor family (Qiu and Goldberg, 2002). Nrdp1 may play a role in the regulation of growth or development, since it is differentially expressed in mouse fetal and adult hematopoietic stem cells and progenitors (Abdullah et al, 2001). While the C‐terminal half of Nrdp1 associates with ErbB3, its N‐terminal half contains a variant of the RING finger domain (C3HC3D) that is essential for its E3 activity (Qiu and Goldberg, 2002). It seems likely that Nrdp1 targets other substrates for degradation, since an ortholog of Nrdp1 is present in Drosophila, which lacks ErbB3. In a search for other substrates in mammalian cells, we have found that the C‐terminal half of Nrdp1 also binds BRUCE. Here we present evidence (1) that Nrdp1 catalyzes ubiquitination and proteasomal degradation of BRUCE and promotes apoptosis, (2) that, in response to apoptotic stimuli, BRUCE is degraded in all cell types tested, while XIAP and c‐IAP1 are degraded only in certain cell types, and (3) that decreasing BRUCE content by RNA interference (RNAi) can induce apoptosis. Thus, BRUCE, unlike other mammalian IAPs, is essential for viability in various cell types, and Nrdp1‐mediated degradation of BRUCE represents a novel mechanism for apoptosis.
Nrdp1 associates with BRUCE in cells
In order to identify additional substrate(s) of Nrdp1, we first used co‐immunoprecipitation and mass spectrometry to isolate and identify proteins that associate with Nrdp1 in 293T cells, which do not express ErbB3 (the only previously known Nrdp1 substrate). The C‐terminal half of Nrdp1 (Nrdp1C, amino acids 134–317) was also used to isolate potential substrates because it contains the region that binds ErbB3, but cannot catalyze ubiquitination since it lacks the RING finger domain (Qiu and Goldberg, 2002). Following transfection with either Nrdp1 or Nrdp1C, both of which carried a FLAG tag on their C‐termini, 293T cell proteins were metabolically labeled with [35S]methionine. Nrdp1 or Nrdp1C and several other polypeptides were precipitated from the cell lysates by the anti‐FLAG antibody (Figure 1A). All the major polypeptides that were precipitated from the lysates of the cells transfected with full‐length Nrdp1 seemed to be bound directly to the antibody rather than to Nrdp1, because when the immunoprecipitates were denatured and re‐dissolved in SDS sample buffer, these polypeptides could be re‐precipitated with the same antibody (Figure 1A, right panel). Thus, our attempts to co‐immunoprecipitate substrates with full‐length Nrdp1 brought down insufficient amounts for detection, probably due to the rapid degradation of the substrates following ubiquitination by Nrdp1. However, immunoprecipitation of the enzymatically inactive C‐terminal half of Nrdp1 brought down significant amounts of an unusually large protein of 530 kDa (Figure 1A). Mass spectrometry identified 25 peptides from this band, which matched sequences from various parts of BRUCE (Figure 1B). Following immunoprecipitation of either the full‐length or the C‐terminal half of Nrdp1, BRUCE was also detected in the precipitates by Western blotting using an anti‐BRUCE antibody (Figure 1C and Supplementary Figure 1‐s). Thus, BRUCE, like ErbB3, associates with the C‐terminal half of Nrdp1 in vivo.
Nrdp1 promotes proteasomal degradation of BRUCE
This association with the substrate‐binding region of Nrdp1 raised the possibility that BRUCE might also be a substrate for Nrdp1. Indeed, transfection of 293T cells with Nrdp1 markedly reduced the levels of BRUCE, but did not affect the levels of two other IAP family members, c‐IAP1 and XIAP (Figure 2A). Nrdp1C, presumably by acting as a dominant‐negative form of Nrdp1, can markedly elevate the levels of transfected ErbB3 (Qiu and Goldberg, 2002). In contrast, transfection with Nrdp1C did not change the levels of BRUCE, suggesting that BRUCE is relatively stable in normal cells (also see below). Furthermore, when NIH/3T3 cells were transfected with Nrdp1 fused to green fluorescence protein (GFP), BRUCE disappeared from the cells (Figure 2B). Pulse‐chase analysis of the 293T cells transfected with Nrdp1 confirmed that Nrdp1 promoted BRUCE degradation (Figure 2C). In the cells transfected with control vectors, BRUCE was quite stable (only about 10% were lost within 4 h). However, in the cells transfected with Nrdp1, BRUCE had a half‐life of less than 2 h. In addition, the proteasome inhibitors, MG132 and lactacystin, reduced BRUCE degradation, while the lysosome inhibitor chloroquine had no effect. Thus, Nrdp1 stimulates degradation of BRUCE, apparently by the 26S proteasome.
Nrdp1 catalyzes ubiquitination of BRUCE in vitro and in vivo
Since Nrdp1 is a Ub ligase, we examined whether Nrdp1 can catalyze BRUCE ubiquitination. Immunopurified BRUCE was incubated with crude lysates of 293T cells that had been transfected with wild‐type Nrdp1. This E3 promoted the formation of high‐molecular‐weight Ub conjugates, which were immunoprecipitated with an anti‐BRUCE antibody and detected by Western blot with an anti‐Ub antibody (Figure 3A). These conjugates could be eluted from the beads with SDS sample buffer and re‐immunoprecipitated with the anti‐BRUCE antibody. Thus, the Ub is covalently conjugated to BRUCE (Figure 3B). The BRUCE from the reaction containing the wild‐type Nrdp1 ran slower on the gel than that from control extracts, presumably because of its ubiquitination (Figure 3A, middle).
When the RING finger mutant (D56V) of Nrdp1, which does not stimulate degradation of ErbB3 (Qiu and Goldberg, 2002), was transfected into cells, and a similar assay was performed in the lysates, BRUCE was not ubiquitinated (Figure 3A). The N‐terminal half of Nrdp1 contains this RING finger domain, which is believed to function as a docking site for an E2 (Joazeiro and Weissman, 2000). Even though BRUCE contains an E2 motif, the N‐terminal half of Nrdp1 did not associate with BRUCE in immunoprecipitation experiments (data not shown).
To investigate whether BRUCE may function as an E2 in its own ubiquitination, wild‐type Nrdp1 was immunopurified from the lysates, eluted from the beads, and incubated directly with immunopurified BRUCE on the beads. No BRUCE ubiquitination was detected, despite the presence of E1, Ub, and ATP (Figure 3C). Although the bacterially expressed E2 motif of BRUCE can form a thioester bond with Ub (Hauser et al, 1998), when full‐length BRUCE was incubated with 125I‐labeled Ub, E1, and ATP, thioester bond formation between BRUCE and Ub was not detected (data not shown). In contrast, UbcH7 or UbcH5c (a member of the UbcH5 family of E2 proteins) was found to form a clear 125I–Ub thioester bond. Since UbcH5c supports Nrdp1‐mediated ubiquitination of ErbB3 (Qiu and Goldberg, 2002), ubiquitination of BRUCE was assayed in the presence of UbcH5c. Addition of this E2 markedly stimulated the formation of BRUCE–Ub conjugates (Figure 3C). However, if Nrdp1 was replaced by the immunopurified RING finger mutant (D56V), no Ub conjugation of BRUCE was evident (Figure 3C). Thus, Nrdp1 catalyzes BRUCE ubiquitination, apparently by recruiting UbcH5 through its RING finger domain.
To test whether Nrdp1 also promotes ubiquitination of BRUCE in vivo, 293T cells were transfected with Nrdp1. BRUCE–Ub conjugates were detected in the cells transfected with Nrdp1 by immunoprecipitation of BRUCE and Western blot analysis using anti‐Ub antibodies (Figure 3D). As we found after ubiquitination of BRUCE in vitro (Figure 3A), a large fraction of the BRUCE immunoprecipitated from the Nrdp1‐transfected cells ran slightly slower than that from the untransfected cells (Figure 3D, second panel from top), apparently due to its conjugation to Ub. Thus, Nrdp1 promotes ubiquitination of BRUCE both in vitro and in vivo, and the E2 motif on BRUCE cannot support this process in vitro.
Apoptotic stimuli induce proteasomal degradation of BRUCE in various types of cells
Both c‐IAP1 and XIAP in thymocytes are degraded by proteasomes in response to apoptotic stimuli, including glucocorticoids and etoposide (an inhibitor of topoisomerase II), and the resulting decrease in their levels may contribute to apoptotic death (Yang et al, 2000). To examine whether BRUCE levels also fall in response to apoptotic stimuli, 293T cells were treated with etoposide (5 μM). The levels of BRUCE decreased markedly by 24 h after treatment (Figure 4A). The highly specific proteasome inhibitor lactacystin greatly reduced the etoposide‐induced loss of BRUCE (Figure 4B), suggesting that this decrease in BRUCE content is largely due to degradation by the 26S proteasome.
Although etoposide induces degradation of c‐IAP1 and XIAP in thymocytes (Yang et al, 2000), this agent did not significantly alter their levels in 293T cells (Figure 4A). In HeLa cells, the levels of BRUCE, but not of c‐IAP1 or XIAP, also decreased after exposure to etoposide for 24 h (Figure 4C). We then examined whether net degradation of BRUCE is associated with apoptosis in other cell types. In response to etoposide or camptothecin (another topoisomerase inhibitor), the levels of BRUCE also dramatically decreased in NIH/3T3, MCF‐7, and HT1080 cells (Figure 4C and Supplementary Figure 2‐s). This degradation of BRUCE was associated with apoptosis, since a marked activation of caspase 7 and cleavage of PARP were detected following these treatments (Figure 4C). While the levels of c‐IAP1 did not fall in these cell lines, the levels of XIAP dramatically decreased in MCF‐7 cells following treatment with camptothecin (Figure 4C). Thus, these apoptotic stimuli induce degradation of BRUCE in all five cell lines tested, while the degradation of other IAPs occurs only in relatively few cell types.
siRNA for Nrdp1 reduces etoposide‐induced loss of BRUCE
To investigate whether Nrdp1 may mediate BRUCE degradation induced by apoptotic stimuli, 293T cells transfected with Nrdp1 or an empty vector were incubated with or without etoposide. This agent potentiated the Nrdp1‐dependent decrease in BRUCE levels, but did not affect the levels of c‐IAP1 (Figure 5A). These observations suggest that Nrdp1 mediates the etoposide‐induced loss of BRUCE.
If Nrdp1 plays a key role in BRUCE degradation in vivo, decreasing the levels of endogenous Nrdp1 by RNAi should reduce BRUCE ubiquitination and degradation. To suppress the expression of Nrdp1, we utilized the DNA vector (BS/U6)‐based RNAi method of Sui et al (2002) to construct a DNA vector (BS/U6/Nrdp1) that expresses 21‐nt small interference RNAs (siRNAs) for the coding region of Nrdp1. In the 293T cells transfected with FLAG‐tagged Nrdp1, cotransfection of BS/U6/Nrdp1 prevented expression of transfected Nrdp1 as shown by Western blot (Supplementary Figure 3‐s). Moreover, real‐time reverse transcriptase–PCR analysis indicated that siRNA for Nrdp1 reduced the mRNA levels of endogenous Nrdp1 by 40–70% (Figure 5B). In the absence of etoposide, the siRNA for Nrdp1 increased BRUCE levels slightly, presumably because it reduced the slow degradation of this quite stable protein. However, after treatment of the cells with etoposide, the levels of BRUCE decreased markedly (Figures 4A and 5B), and the expression of Nrdp1 siRNAs greatly reduced this etoposide‐induced loss of BRUCE (Figure 5B). The levels of BRUCE in the cells expressing GFP siRNA were slightly higher than those for the empty vector, but were dramatically lower than those for Nrdp1 siRNA. In contrast, Nrdp1 siRNAs did not affect the levels of XIAP. Thus, endogenous Nrdp1 mediates specifically the degradation of BRUCE induced by etoposide.
siRNA for BRUCE promotes apoptosis
Since apoptotic stimuli trigger Nrdp1‐mediated degradation of BRUCE, we tested the possibility that this loss of BRUCE may by itself cause apoptosis. Expression of siRNAs for BRUCE markedly reduced the levels of this protein in HeLa cells as demonstrated by immunostaining (Figure 6A). Following transfection of HeLa cells with control vectors, 4.53±0.03% of the cells were apoptotic, as shown by condensed chromatin. However, following transfection of BRUCE siRNAs, the fraction of apoptotic cells increased to 16.7±1.9%, even though the transfection efficiency was only about 27% (Figure 6B). Meanwhile, expression of siRNA for BRUCE led to a clear increase in the levels of active caspase 3, especially the 19‐kDa band (Figure 6C, left panels). This increase was even larger if the cells were also treated with tumor necrosis factor α (TNFα) (Figure 6C). In 293T cells, which are known to be quite resistant to apoptotic stimuli, the expression of siRNA‐BRUCE had no effect on the levels of active caspase 3 (the 17‐kDa band) by itself, but caused a marked increase in active caspase 3 in the presence of TNFα (Figure 6D). These results indicate that decreasing the cellular content of BRUCE promotes apoptosis or enhances sensitivity to other apoptotic stimuli.
Activation of caspases follows mitochondrial damage and release of caspase activators, such as cytochrome c. This process can be inhibited by the antiapoptotic members of the Bcl‐2 family, Bcl‐2 and Bcl‐xL (Merry and Korsmeyer, 1997; Kroemer and Reed, 2000). To examine whether the apoptosis induced by BRUCE‐siRNA is resistant to Bcl‐2/Bcl‐xL, an MCF‐7/Fas cell line that overexpresses Bcl‐xL was employed. Although Bcl‐xL suppressed the cell death induced by TNFα and cycloheximide (an inhibitor of protein synthesis) as shown previously (Li et al, 1998), the cell death induced by BRUCE RNAi was not inhibited by Bcl‐xL (Figure 6E). To test whether caspase activation mediates the effect of BRUCE‐siRNA, we treated MCF‐7/Fas cells with the general inhibitor of caspases, z‐VAD, for 2 days. Although z‐VAD suppressed TNFα‐induced cell death, it had no effect on the cell death induced by BRUCE‐siRNA at 4 days after transfection (Figure 6F). In other experiments (not shown), when fresh z‐VAD was added daily from the beginning of transfection, there was still no inhibition of the cell death induced by BRUCE‐siRNA (although there was some toxicity induced by z‐VAD itself). Thus, while BRUCE‐siRNA leads to caspase activation, it also appears to promote apoptosis by a mechanism not sensitive to Bcl‐2 or inhibition of caspases.
Overexpression of Nrdp1 promotes caspase activation and apoptosis
These findings strongly suggest that degradation of BRUCE by Nrdp1 should promote apoptosis. To test this idea, we first transfected NIH/3T3 cells (which do not express ErbB3) with Nrdp1‐FLAG and selected Nrdp1‐FLAG‐positive cells by incubating them with geneticin for 11 days. The growth of the cells transfected with Nrdp1 was much less than that of cells transfected with an empty vector (Figure 7A). This effect was due to the Ub ligase activity of Nrdp1, since transfection of the Nrdp1 N‐terminal half or the inactive RING finger mutant, Nrdp1D56V, did not reduce cell growth. Thus, the ability of Nrdp1 to suppress the growth of NIH/3T3 cells requires both the RING finger domain and the substrate‐binding region. In similar experiments, overexpression of Nrdp1 also suppressed the growth of HCT116 human colon carcinoma cells (data not shown).
To determine whether Nrdp1 promotes apoptosis, HeLa cells were transiently transfected with Nrdp1 and stained with Hoechst 33342 to visualize condensed chromatin. While transfection with Nrdp1 by itself had no significant effect on the frequency of cell death within 24 h, it increased by over 100% the frequency of apoptosis induced by TNFα in the presence of cycloheximide (Figure 7B, left panel). In addition, even in 293T cells, which are normally quite resistant to apoptosis induced by TNFα, Nrdp1 transfection also led to a three‐fold increase in the low level of cell death upon TNFα treatment (Figure 7B, right panel). Following treatment with TNFα, the 293T cells overexpressing Nrdp1 also showed a marked activation of caspase 3, as demonstrated by enzymatic assays (Figure 7C, upper panel) and immunoblotting of the active fragments (Figure 7C, lower panels). Thus, TNFα appears to enhance sensitivity of the cells to overexpression of Nrdp1. However, the TNFα treatment also increased the levels of transfected Nrdp1 (Figure 7C), perhaps by increasing promoter activity of the plasmid, since the levels of transfected GFP were also elevated by this treatment (data not shown). In any case, overexpression of Nrdp1, like the suppression of BRUCE content by RNA interference or by apoptotic stimuli, promotes apoptosis or enhances sensitivity to apoptotic stimuli.
BRUCE inhibits apoptosis
The present studies have demonstrated that the Ub ligase Nrdp1 catalyzes ubiquitination and proteasomal degradation of BRUCE, and that the net loss of BRUCE can play an important role in inducing apoptosis. Antisense oligonucleotides against BRUCE enhance the sensitivity of the SNB‐78 glioma cells to apoptosis induced by DNA‐damaging agents (Chen et al, 1999). In certain cell types (HeLa and MCF‐7 cells), BRUCE is essential to prevent apoptosis, since decreasing BRUCE levels by RNAi could by itself induce apoptosis. In other cells (e.g., 293T cells), knockdown of BRUCE enhanced sensitivity to TNFα‐induced cell death. In contrast to BRUCE, no other mammalian IAPs have been shown to be essential for viability, even though overexpression of virtually any IAP suppresses apoptosis. For example, mice lacking XIAP showed no sign of increased sensitivity to apoptotic stimuli (Harlin et al, 2001).
An important question raised by these findings is how BRUCE inhibits apoptosis. BRUCE appears to act downstream of Bcl‐2/Bcl‐xL (Figure 6E), which inhibits the release of caspase activators from the mitochondria (Merry and Korsmeyer, 1997; Kroemer and Reed, 2000). Hence, one possibility is that BRUCE, like other IAPs, may inhibit caspases. On the other hand, IAPs, such as XIAP, have antiapoptotic activities unrelated to their ability to inhibit caspases (Silke et al, 2001, 2002). Since Drosophila BRUCE seems unable to block the Drosophila caspase (Dronc) activity (Vernooy et al, 2002), the mammalian BRUCE may also not inhibit caspases directly. Several IAPs can bind the mammalian RHG motif‐containing proteins, such as Smac/DIABLO and HtrA2/Omi, which promote apoptosis (Du et al, 2000; Suzuki et al, 2001a). Interestingly, Drosophila BRUCE suppresses cell death induced by the Drosophila RHG proteins, Rpr and Grim (Vernooy et al, 2002). In related studies, we have found that BRUCE interacts with both Smac/DIABLO and HtrA2/Omi (Qiu et al, in preparation). Thus, one mechanism by which mammalian BRUCE inhibits apoptosis is probably through downregulation of Smac/DIABLO and related proapoptotic proteins.
Nrdp1 promotes ubiquitination and proteasomal degradation of BRUCE in response to apoptotic stimuli
The Drosophila IAP, DIAP1, binds and inhibits Drosophila caspases, and genetic studies indicate that the cellular content of DIAP1 determines the threshold for apoptosis (Goyal, 2001). Perhaps in an analogous way, the level of BRUCE may also determine the sensitivity of certain mammalian cells to apoptotic stimuli. In fact, reducing BRUCE by overexpression of Nrdp1 also caused caspase 3 activation and promoted apoptosis. Furthermore, accelerated degradation of BRUCE appears to be a key event in apoptosis. During the etoposide‐induced apoptosis, the levels of c‐IAP1 and XIAP fall markedly in thymocytes (Yang et al, 2000), but did not change in HeLa and 293T cells (Figure 4). In MCF‐7 cells, the levels of both BRUCE and XIAP, but not of c‐IAP1, decreased sharply in response to another apoptotic stimulus, camptothecin. The levels of BRUCE also fell dramatically in HeLa, 293T, HT1080, and NIH/3T3 cells upon treatment with either etoposide or camptothecin. Thus, degradation of BRUCE appears to be a general event during initiation of apoptosis in many cell types, while degradation of other IAPs may or may not occur, depending on the nature of the cell and, perhaps, the apoptotic stimulus. Since the loss of BRUCE could be inhibited by RNAi for Nrdp1 and by lactacystin, these results indicate that Nrdp1 catalyzes proteasomal degradation of BRUCE in response to these apoptotic stimuli.
The present findings also raise the possibility that Nrdp1 or BRUCE may have to be modified to cause ubiquitination of BRUCE. Transient transfection with Nrdp1 itself for up to 3 days did not cause cell death in any cell type tested. However, cell death was triggered following either prolonged incubation (11 days) in the presence of geneticin (Figure 7A) or 1‐day incubation in the presence of TNFα (Figure 7B) after transfection with Nrdp1. One possible explanation is that most Nrdp1 may be normally present in an inactive form or that BRUCE may normally exist in an Nrdp1‐resistant form, and apoptotic stimuli (such as TNFα) activate BRUCE degradation by causing enzyme or substrate modification (e.g., by phosphorylation). Alternatively, the levels of BRUCE may need to be quite low in order to trigger apoptosis, and the prolonged expression of Nrdp1 or treatment with TNFα may trigger apoptosis, because they lower the levels of BRUCE appreciably. The latter notion is also supported by the results in Figure 6, where siRNA‐BRUCE decreased BRUCE levels sufficiently to trigger apoptosis in HeLa and MCF‐7 cells, but needed TNFα to do so in 293T cells. Although all the five lines of cells tested showed some cell death in response to the treatments that reduced BRUCE, 293T cells were relatively resistant to apoptosis. For example, while a combination of Nrdp1 transfection and TNFα treatment led to a three‐fold increase in the rate of cell death in 293T cells, the total cell death was only about 5% (in contrast to 60% in HeLa cells).
In addition to BRUCE, Nrdp1 might promote ubiquitination and degradation of other protein(s) that may also be involved in the regulation of apoptosis. The present findings were obtained in cells lacking ErbB3. In cells expressing this receptor, the Nrdp1‐catalyzed degradation of ErbB3 perhaps also contributes to apoptosis, since EGF, which activates both the EGF receptor and ErbB3, suppresses apoptosis (Zhou et al, 2000; Gilmore et al, 2002).
Nrdp1 seems to be proapoptotic by promoting degradation of BRUCE and perhaps ErbB3. However, in related experiments, we found that knockdown of Nrdp1 by RNAi enhanced the sensitivity of 293T cells to TNFα, as evidenced by the elevated levels of active caspase 3 in 293T cells transfected with siRNA for Nrdp1 (Supplementary Figure 4‐s). The simplest explanation for these findings is that although Nrdp1's effect on BRUCE can induce apoptosis, Nrdp1 can also inhibit apoptosis by promoting the degradation of additional substrate(s), which are proapoptotic. There are several instances where the same E3 seems to have very different, even opposite, physiological effects in degrading different substrates. For example, c‐IAP1 can catalyze ubiquitination and degradation of TRAF2, which is antiapoptotic, but c‐IAP1 is best known as an inhibitor of caspases and thus should retard apoptosis (Deveraux et al, 1998; Li et al, 2002). In such cases, the susceptibility of the different substrates to E3 is regulated by distinct mechanisms, and the effects of knockdown of the E3 must depend on the physiological conditions and the relative roles of the substrates. Understanding the effects of Nrdp1 knockdown must await the identification of the additional Nrdp1 substrates and elucidation of their roles in apoptosis.
BRUCE is a very unusual IAP
Unlike most human IAPs, which are E3s containing a RING finger domain and undergo autoubiquitination (Yang et al, 2000; Wilson et al, 2002), BRUCE bears an E2 motif in its C‐terminal region. Despite this E2 motif, we were unable to obtain any evidence that BRUCE may function as an E2 in its own ubiquitination. The Nrdp1‐mediated ubiquitination of BRUCE in vitro required an exogenous E2, UbcH5. Also, immunopurified BRUCE could not serve as an E2 for autoubiquitination of Nrdp1 (data not shown). Nevertheless, it remains possible that BRUCE may function as an E2 for ubiquitination of other proteins under other conditions.
Another important feature of BRUCE is its association with membranes including the Golgi apparatus (Hauser et al, 1998). Interestingly, Nrdp1 also appears to be membrane associated (Qiu and Goldberg, 2002). Transfected Nrdp1 was localized on the cytoplasmic membrane as well as in the cytosol. Although most membrane proteins are degraded by lysosomes, BRUCE, like ErbB3, appears to be degraded by proteasomes, since MG132 and lactacystin reduced the Nrdp1‐mediated degradation of BRUCE (Figure 2C). In contrast, the lysosomal inhibitor chloroquine did not reduce BRUCE degradation (Figure 2C) under conditions where it can inhibit the Cbl‐mediated degradation of the EGF receptor, a lysosomal substrate (Levkowitz et al, 1998). Since BRUCE is one of the largest proteins known (530 kDa), it is intriguing that proteasomes can digest such a long polypeptide chain.
In certain mammalian cells, apoptotic stimuli induce autoubiquitination and degradation of c‐IAP1 and XIAP, but no proteins are known to regulate degradation of mammalian IAPs. In contrast, Rpr, Grim, and Hid in Drosophila have been found to reduce the levels of DIAP1 by promoting its ubiquitination and degradation (Palaga and Osborne, 2002). Also, a Drosophila E2, UBCD1, and an F box protein (a subunit of some E3s), Morgue, can synergize with Rpr in promoting the degradation of DIAP1 (Hays et al, 2002; Ryoo et al, 2002; Wing et al, 2002). Therefore, the Nrdp1‐mediated degradation of BRUCE is the first example where degradation of a mammalian IAP is regulated by another protein and can contribute to apoptosis.
Materials and methods
293T adenovirus‐transformed human embryo kidney cells, HT1080 human fibrosarcoma cells, HeLa human epitheloid carcinoma cells, and HCT116 human colon carcinoma cells were maintained in DMEM (Gibco), supplemented with 10% fetal bovine serum and antibiotics in 5% CO2, while the DMEM was supplemented with 10% calf serum for NIH/3T3 mouse embryonic cells. All the above cell lines were obtained from the American Type Culture Collection. MCF‐7/Fas and MCF‐7/Fas/Bcl‐xL human breast cancer cells were obtained and cultured as described (Li et al, 1998). Transfection was carried out using the SuperFect transfection kit (Qiagen).
Identification of BRUCE by mass spectrometry
A total of 10 100 mm plates of 293T cells were transfected with FLAG‐tagged Nrdp1C, and incubated for 2 days. The cells were then treated with 10 μM of MG132 for 3 h and lysed in buffer containing 20 mM Tris–HCl, pH 8.0, 100 mM KCl, 0.2% Nonidet P‐40, 10% glycerol, 1 mM ZnCl2, 10 mM β‐glycerophosphate, 5 mM tetrasodium pyrophosphate, 1 mM NaF, 1 mM NaVO3, and a protease inhibitor cocktail (Roche). The lysates (about 15 mg of protein) were immunoprecipitated with anti‐FLAG resin (Sigma), eluted with 100 mM Gly, pH 3.5, resolved by 4–12% gradient SDS–PAGE, and visualized by Coomassie staining. The band at about 500 kDa was excised for analysis by mass spectrometry (Department of Cell Biology, Harvard Medical School).
To express GFP–Nrdp1 fusion protein in mammalian cells, the cDNA encoding mouse full‐length Nrdp1, obtained from Research Genetics (GenBank accession number AA692083), was amplified by PCR and subcloned into pEGFP‐C1 (Clontech) at EcoRI/BamHI sites. The construction of BS/U6/Nrdp1 and BS/U6/BRUCE was carried out as described (Sui et al, 2002). The parent plasmid BS/U6 for siRNA expression and BS/U6/GFP for GFP RNAi were kindly provided by Drs G Sui and Y Shi. Four DNA oligos, which form an inverted repeat, were synthesized based on the 21‐nt sequence in the coding region of the Nrdp1 cDNA (5′‐GGGATGATCTCGACTCCTGAT‐3′) or the BRUCE cDNA (5′‐GGGCCAGTATGGCT‐TGCTAGTA‐3′) that bears no significant homology with other genes.
Analysis of Nrdp1 mRNA by real‐time reverse transcriptase–PCR
Total RNA was isolated from 293T cells using TRIzol reagent (Invitrogen). cDNA synthesis was carried out from 0.5 μg of total RNA using ProSTAR ultra‐HF RT–PCR system (Stratagene). The cDNA from each sample was processed for relative quantification by real‐time PCR. The primers for amplifying a 321‐bp fragment of Nrdp1 cDNA are: 5′‐TGCATTAAGCAC‐CTGCGC‐3′ and 5′‐AGCAGGACAGCCACTCTCC‐3′. The primers for amplifying a 450‐bp fragment of glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) cDNA are: 5′‐CTCAGACACCATGGGGAAGGT‐3′ and 5′‐ATGATCTTGAGGCTGTTGTCATA. Templates and primers were mixed with components from the LightCycler FastStart DNA MasterPLUS SYBR Green I kit (Roche Applied Science). Reactions in triplicate were carried out in the LightCycler real‐time PCR machine (Roche Applied Science) under the following conditions: initial denaturation at 95°C for 10 min, and then 40 cycles of denaturation at 95°C for 10 s, annealing at 55°C for 5 s, and extension at 72°C for 25 s. Melting curves were obtained to examine the purity of amplified products. In addition, the products were separated on 1.2% agarose gels in the presence of ethidium bromide to ensure the right sizes of the amplified products. The relative amount of Nrdp1 mRNA in each sample was normalized to that of GAPDH mRNA.
Western blot analysis
Cells were lysed as described above for analysis of all the proteins. BRUCE, c‐IAP1, XIAP, PARP, caspase 7, active caspase 3, actin, Ub, and Nrdp1 were detected by Western blot with horseradish peroxidase‐conjugated secondary antibodies using standard procedures. Anti‐Nrdp1 serum was generated by immunizing rabbits with a C‐terminal peptide of Nrdp1 (GDDMVQEPGLVMIFAHGVEE). Other antibodies were purchased from BD Biosciences (IAPs), Sigma (caspase 7, actin, and FLAG), Cell Signaling Inc. (active caspase 3), Santa Cruz (PARP), and Zymed (Ub).
Measurement of protein degradation
293T cells were incubated in serum‐free medium for 45 min, metabolically labeled with 35S‐Met and 35S‐Cys (Amersham), and then chased in nonradioactive medium. Cell extracts were immunoprecipitated with anti‐BRUCE antibodies, separated on 4–12% gradient SDS–PAGE, and analyzed with a PhosphoImager (Bio‐Rad).
The reaction was carried out in the ubiquitination buffer containing 40 mM Tris, pH 7.5, 5 mM MgCl2, 0.5 mM DTT, and 2 mM ATP. Unless indicated otherwise, this buffer (20 μl in total volume) was supplemented with 20 ng of the purified human E1 expressed in Escherichia coli (CalBiochem), 0 or 0.1 μg of the purified UbcH5c, 0.2 μg of [125I]‐Ub, Nrdp1‐FLAG, and BRUCE. The immunopurified BRUCE on beads was incubated with the reaction mix at 30°C for 90 min, washed twice in the same buffer, resolved by SDS–PAGE, and analyzed by immunoblotting or PhosphoImager.
Spectrofluorometric assay of caspase 3
Caspase 3 activity was assayed similarly as described (Du et al, 2000). Aliquots of cell lysates were analyzed in 200 μl of a reaction mixture containing 0.1 mM HEPES (pH 7.4), 2 mM DTT, 0.1% (w/v) CHAPS, 1% (w/v) sucrose, and 20 μM of Z‐DEVD‐7‐amino‐4‐methylcoumarin at 37°C for 30 min. The release of 7‐amino‐4‐methylcoumarin from the substrates was monitored continuously at wavelength pair 380/460 nm (excitation/emission).
Supplementary data are available at The EMBO Journal Online.
Supplementary Figure 1‐s Specificity of anti‐BRUCE antibody. The lysates from human 293T cells, which were transfected with either empty vector (C) or Nrdp1‐FLAG, were separated on a 4–12% NuPAGE gel (Invitrogen).Western blot analysis using mouse monoclonal anti‐BRUCE antibody, raised against aa 372‐571 of mouse BRUCE (BD Transduction Laboratories), detected two bands at about 500 and 100 kD, respectively. The 500‐kD band represented the full‐length BRUCE, whose levels were reduced by transfected Nrdp1‐FLAG, while the levels of the 100 kD band (a cross reacting protein or a fragment of BRUCE) seemed not to be affected by transfection of Nrdp1. However, these contaminating species varied with types of cells. For example, although the antibody could still react with the 500‐kD band, the 100‐kD band was not detectable from the lysates of mouse NIH/3T3 cells (see Supplemental Figure 2‐sB).
Supplementary Figure 2‐s Etoposide or camptothecin (CPT) reduces the levels of BRUCE in HT1080 (A) and NIH/3T3 (B) cells. The cells were treated with either etoposide or CPT at concentrations indicated for 24 h. The levels of BRUCE, c‐IAP1, XIAP, β‐actin and caspase 7 were detected by western blot. β‐actin serves as a control for loading,and cleavage of caspase 7 as a marker for apoptosis.
Supplementary Figure 3‐s Nrdp1 siRNA suppresses expression of the transfected Nrdp1. 293T cells were co‐transfected with pcDNA3.1‐Nrdp1‐FLAG and either BS/U6 or BS/U6/Nrdp1, and were incubated for 4 days. In some experiments, the control BS/U6 was replaced by BS/U6/GFP, but the similar results were obtained. The levels of Nrdp1‐FLAG were analyzed by western blot with an antibody against FLAG tag.
Supplementary Figure 4‐s Nrdp1 siRNA increases the levels of active caspase 3 in the presence of TNF α. 293T cells were transfected with BS/U6, BS/U6/GFP, BS/U6/Nrdp1, or BS/U6/BRUCE, and were split at 1 to 5 at 3 days after transfection.Sixteen hours later, the cells were treated with TNF α (20 ng ml−1), and incubated for another 6 h. The levels of the active caspase 3 and β‐actin were determined by western blot.
Supplementary Figure 5 to 7
We thank Dr Lewis C Cantley for advice, Ms Dongmei Cheng and Dr Steve Gygi for mass spectrometry, Drs Guangchao Sui and Yang Shi for reagents, Dr Maria G Masucci and members of the Goldberg laboratory for helpful comments, and Nikon Imaging Center at Harvard Medical School for providing microscopes. This work was supported by grants from the NIGMS and Hereditary Disease Foundation to ALG, and by postdoctoral fellowships from the American Heart Association, New England Affiliate, and the NIH to X‐BQ.
- Copyright © 2004 European Molecular Biology Organization