Specific Cleavage of Mcl-1 by Caspase-3 in Tumor Necrosis Factor-related Apoptosis-inducing Ligand (TRAIL)-induced Apoptosis in Jurkat Leukemia T Cells*

Tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) induces programmed cell death through the caspase activation cascade and translocation of cleaved Bid (tBid) by the apical caspase-8 to mitochondria to induce oligomerization of multidomain Bax and Bak. However, the roles of prosurvival Bcl-2 family proteins in TRAIL apoptosis remain elusive. Here we showed that, besides the specific cleavage and activation of Bid by caspase-8 and caspase-3, TRAIL-induced apoptosis in Jurkat T cells required the specific cleavage of Mcl-1 at Asp-127 and Asp-157 by caspase-3, while other prototypic antiapoptotic factors such as Bcl-2 or Bcl-XL seemed not to be affected. Mutation at Asp-127 and Asp-157 of Mcl-1 led to cellular resistance to TRAIL-induced apoptosis. In sharp contrast to cycloheximide-induced Mcl-1 dilapidation, TRAIL did not activate proteasomal degradation of Mcl-1 in Jurkat cells. We further established for the first time that the C-terminal domain of Mcl-1 became proapoptotic as a result of caspase-3 cleavage, and its physical interaction and cooperation with tBid, Bak, and voltage-dependent anion-selective channel 1 promoted mitochondrial apoptosis. These results suggested that removal of N-terminal domains of Bid by caspase-8 and Mcl-1 by caspase-3 enabled the maximal mitochondrial perturbation that potentiated TRAIL-induced apoptosis.

Mcl-1) protect mitochondrial membrane integrity, while the multidomain proapoptotic proteins (e.g. Bax and Bak) facilitate the release of apoptogenic factors, such as cytochrome c, from mitochondria to initiate the caspase cascade and organelle dysfunction. Bid is a unique "BH3-only" 1 proapoptotic member of the Bcl-2 family; once activated, the resulting C-terminal domain (tBid) can translocate to mitochondria (11,12) and induce oligomerization of Bax or Bak (13,14) that results in the permeabilization of the mitochondrial outer membrane and the release of cytochrome c.
Mcl-1 was originally identified as an immediate-early gene induced during differentiation of ML-1 myeloid leukemia cells (15). Although Bcl-2 and Bcl-X L have long been thought to be the critical regulators of cytokine withdrawal and stress-induced apoptosis (16), recent data have indicated that Mcl-1 may play more profound roles than Bcl-2 and Bcl-X L in response to a variety of death stimuli. First, conditional gene targeting experiments show that early T and B lymphocyte development and mature lymphocyte homeostasis rely on cytokine-dependent inhibition of apoptosis by Mcl-1 (17). In contrast, hematopoiesis is initially normal in bcl-2-deficient (18) and bcl-X L -deficient mice (19,20). Second, induction of apoptosis by ultraviolet irradiation (21) or adenovirus infection (22) requires proteasomal degradation of Mcl-1 but not Bcl-X L or Bcl-2, which is upstream of Bcl-X L ; Bax translocation to mitochondria; and caspase activation (21). Finally specific cleavage of Mcl-1 but not Bcl-2 or Bcl-X L by caspase-3 or granzyme B is essential for spontaneous apoptosis of B cell lymphoma or cytotoxic T lymphocytes, respectively (23,24).
TRAIL belongs to the tumor necrosis factor (TNF) superfamily (25). The homotrimeric form of TRAIL initiates apoptosis through engagement of death receptors (DR4 or DR5) whose intracellular death domain recruits and forms the death-inducing signaling complex (26). Procaspase-8 is then recruited to the death domain via Fas-associated death domain protein and subsequently activated through interactions between the death effector domain (27). Caspase-8, the apical caspase, in response to the extrinsic death signal can cleave and activate caspase-3 to execute programmed cell death (28). On the other hand, caspase-8 can cleave Bid, and tBid then turns on mitochondrial dysfunction to activate caspase-3 (4,29). Therefore, caspase-8 activation of Bid bridges the extrinsic and intrinsic apoptotic pathways; however, the physiological role and specificity of this cross-talk remain elusive.
Gene targeting experiments implicate that TRAIL-mediated apoptosis most likely plays a critical role in negative selection of T cells (30). However, negative selection at least of early thymocytes is normal in mice deficient for other TNF superfamily members, such as TNF or TNF receptor (31,32), Fas or Fas ligand (33,34), and more importantly components of deathinducing signaling complex, Fas-associated death domain protein (35,36), or caspase-8 (37). Given the more important role of Mcl-1 than that of Bcl-2 and Bcl-X L in lymphocyte development (38), there might exist a functional link between TRAIL and the antiapoptotic function of Mcl-1. Indeed overexpression of Mcl-1 is reported to be responsible for TRAIL resistance in cholangiocarcinoma cells (39). In the current study, we found that TRAIL apoptosis required the specific cleavage of Mcl-1 but not Bcl-2 or Bcl-X L by caspase-3 in Jurkat T cells. We further demonstrated that the resulting C-terminal domain of Mcl-1 became a proapoptotic factor that cooperated with tBid to potentiate mitochondrial apoptosis.
Plasmid Construction-The prokaryotic and eukaryotic expression vectors of human Mcl-1 were constructed by add-on PCR and in-frame subcloning. In brief, PCR-amplified full-length Mcl-1 was inserted between EcoRI and SalI sites of pGEX-4T1 (designated pGEX-4T1-Mcl-1) and pCMV-HA (designated pCMV-HA-Mcl-1). The similar expression vectors of C-terminal domains of Mcl-1 (Mcl-1-C1 (amino acids 128 -350) and Mcl-1-C2 (amino acids 158 -350)) were constructed in parallel by subcloning of related cDNA fragments between EcoRI and SalI sites of pGEX-4T1 and pCMV-Myc. The EYFP fusion vector was made in parallel to yield pEYFP-N1-Mcl-1. The cDNAs of Mcl-1 without the transmembrane domain (encoding amino acids 1-329, 128 -329, or 157-329) were subcloned between EcoRI and SalI sites of pGEX-4T1 to yield pGEX-4T1-Mcl-1⌬TM, pGEX-4T1-C-Mcl-1-C1⌬TM, and pGEX-4T1-C-Mcl-1-C2⌬TM. The transmembrane domain of Mcl-1 was deleted to facilitate the purification of recombinant proteins. For expression of the GST fusion protein of tBid, amino acids 61-195 of human Bid was subcloned into pGEX-4T1 between BamHI and EcoRI sites. The mammalian expression vectors for Bid or tBid were made by insertion of corresponding cDNAs between EcoRI and SalI of pCMV-HA (designated pCMV-HA-Bid and pCMV-HA-tBid). All constructs were verified by DNA sequencing analysis.
Site-directed and Truncational Mutagenesis-Point mutations to the potential caspase-3 cleavage sites within Mcl-1 (D127A, D157E, and D127A/D157E double mutation) were introduced by site-directed mutagenesis (QuikChange, Stratagene) using either pGEX-4T1-Mcl-1 or pCMV-HA-Mcl-1 as the template. A series of truncation mutants and wild type Mcl-1 were subcloned between EcoRI and BamHI sites of pGBKT7 by add-on PCR. The primer sequences are available on request.
Co-immunoprecipitation and GST Pull-down Assays-For analysis of protein-protein interaction between Mcl-1-C2 and tBid, MCF-7 cells (2.5 ϫ 10 6 ) were transiently co-transfected with 2 g of pCMV-myc-Mcl-1-C2 and pCMV-HA-tBid or pCMV-HA control vector using Effectene (Qiagen). Twenty-four hours post-transfection, whole cell lysates were prepared with 0.5 ml of radioimmune precipitation assay buffer plus 1ϫ protease inhibitors (Roche Applied Science). The supernatant was precleared with 20 l of protein A/G-agarose beads (Santa Cruz Biotechnology), and the anti-HA monoclonal antibody at the final concentration of 2 g/ml was then added to 0.5 ml of lysates and incubated on a roller overnight at 4°C. Formed complexes were precipitated with 20 l of protein A/G-agarose beads and resolved by 12% SDS-PAGE, and coprecipitated Mcl-1-C2 was detected by Western blotting with anti-Myc antibody (1:1000 in phosphate-buffered saline/Tween 20 with 1% milk). To assess protein-protein interaction between Mcl-1 and various Bcl-2 family proteins and VDAC1, various GST fusion and His-tagged proteins where indicated were purified, and GST pull-down assays were performed exactly as described previously (41).
Reverse Transcription PCR-The mRNA expression levels of various Bcl-2 family genes was performed using reverse transcription (RT)-PCR after Jurkat T cells were treated or not treated with TRAIL or cycloheximide. In brief, 5 g of total RNA extracted using an RNeasy minikit (Qiagen) were utilized to synthesize the first strand cDNA according to the manufacturer's instructions (Invitrogen). Five microliters of reverse transcription products were used for the second strand of cDNA amplification with 1.0 units of Pfu polymerase (Stratagene). The yield of various cDNAs was quantitated in a 1.5% agarose gel. The sequences of specific primers for mcl-1, bid, bcl-X L , bcl-2, and ␤-actin are available on request.
Cleavage of Mcl-1 Protein-Caspase in vitro cleavage experiments were performed as described previously with modification (42). In brief, 10 g of purified GST-Mcl-1⌬TM protein was incubated with 20 ng of recombinant caspase-3 or 200 ng of caspase-8 at 37°C for 2 h in 50 l of reaction buffer. The reactions were stopped and resolved by 15% Tris/Tricine SDS-PAGE. After electrotransfer to a polyvinylidene difluoride membrane (Millipore) the cleavage products were either visualized with Coomassie Blue R-250 staining or Western blotted with Mcl-1 antibody (K-20). The Mcl-1 fragments were cut from polyvinylidene difluoride membrane for N-terminal Edman protein sequencing (Procise, ABI) according to the manufacturer's manual. Recombinant human caspase-3 and caspase-8 were prepared as described previously (12,43).
Caspase in vitro cleavage assays using apoptotic cell extracts were performed as described previously (44). Cell Death Assays-TRAIL-induced apoptosis in Jurkat cells was evaluated with a FACSCalibur (BD Biosciences) using the Annexin-V-fluorescein isothiocyanate and propidium iodine double staining method (Baosai, Beijing, China). To better correlate the relationship between apoptosis and various exogenously expressed Bcl-2 homologs, cell death assays where transfection had to be used were performed essentially as described previously (45,46). Briefly HeLa cells or 293T cells (1 ϫ 10 5 ) were seeded in 12-well plates. After 16 h, cells were transiently transfected with the plasmids as indicated together with 0.1 g of pEGFP vector as transfection marker using the Effectene method (Qiagen). Cells were then exposed to TRAIL where needed and analyzed by fluorescence microscopy (Nikon TE2000). At least 300 GFP-positive cells were counted for each transfection, and apoptotic cells based on morphological alterations (rounded, condensed, and detached from dish) were counted. The percentage of apoptosis represented the mean value from three independent experiments conducted in triplicate (mean Ϯ S.D.).
Cytochrome c Release-Mitochondria of rat livers were prepared, and cytochrome c release experiments were performed exactly as described previously (47).

Down-regulation of Mcl-1 Was Associated with TRAIL-induced
Apoptosis-To test the response of Bcl-2 family proteins to TRAIL-induced apoptosis, we performed Western blotting on the major Bcl-2 family proteins in Jurkat cells treated with recombinant FLAG-tagged TRAIL (Fig. 1A). The results indicated that, along the time course of the treatment, increased apoptosis rates (top panel) accompanied reduction of both fulllength Bid and Mcl-1, while protein levels of Bcl-2 and Bcl-X L remained constant (bottom panel). The antibody (7A3, Cell Signaling Technology) used in this experiment could not detect tBid, but other experiments have already established that Bid is cleaved by TRAIL-activated caspase-8 (11,12). It is intriguing that Mcl-1 started to decay before Bid cleavage (Fig. 1A, lanes 2 and 3), implicating different stability dynamics and maybe a more important role for Mcl-1 in response to TRAIL signaling than that for Bid.
Caspase Cleavage but Not Proteasome Degradation of Mcl-1 in Response to TRAIL-Mcl-1 is a highly regulated antiapoptotic protein, and it can be degraded by the proteasome pathway upon genotoxic stress (21) or cleaved by caspase either in spontaneous apoptosis of B cell lymphoma (24) or in Fas-mediated cell death (48). To test which pathway might be involved in TRAIL-induced down-regulation of Mcl-1, we preincubated Jurkat T cells with pan-caspase inhibitor Z-VAD-fmk (Fig. 1A, lane 5) or proteasome inhibitor lactacystin (Fig. 1B, lane 5) before addition of TRAIL. In contrast to the previous observation that TNF␣-induced Bid degradation in Jurkat cells was inhibited by proteasome inhibitors (49), in the present study lactacystin could not block the proteasomal degradation of fulllength Bid, but Z-VAD-fmk could do so. In a similar fashion, reduction of Mcl-1 by TRAIL engagement could also be ablated by pretreatment with Z-VAD-fmk but not lactacystin (Fig. 1, A and B, compare lane 5 in both), consistent with a recent report (50). In fact, lactacystin could promote Mcl-1 reduction through inhibition of proteasomal degradation of activated casapse-3 (data not shown). Therefore, these results suggested that TRAIL-induced apoptosis might require a unique step of caspase cleavage of Mcl-1. Furthermore both caspase-3 (Z-DEVD-fmk) and caspase-8 (Z-IETD-fmk) inhibitors were able to inhibit TRAIL-induced reduction of Mcl-1 (Fig. 1B, lanes 3  and 4), strongly suggesting that Mcl-1 cleavage could be a downstream event of caspase-3 activation.
Mcl-1 turnover in apoptotic cells could be regulated at both transcription and post-translation levels (51). To test whether TRAIL-mediated down-regulation of Mcl-1 could occur at the transcription level as well, we performed RT-PCR analysis of the mRNA level of Mcl-1 and that of Bcl-X L as a control (Fig.  1C). The results showed that prolonged TRAIL treatment of Jurkat T cells led to decreased transcription of Mcl-1 gene, while that of Bcl-X L remained the same. This line of evidence indicated that TRAIL signaling caused reduction of full-length Mcl-1 protein at both transcription and post-translation levels.
The de novo protein synthesis inhibitor cycloheximide (CHX) can effectively induce apoptosis and cause Mcl-1 degradation via the proteasomal pathway (52). Western blot analysis indicated that, in contrast to the effect of TRAIL on Mcl-1, reduction of Mcl-1 by CHX signaling could be completely inhibited by lactacystin (Fig. 2, A and B) but only partially abolished by Z-VAD-fmk ( Fig. 2A). Moreover this reduction was not due to decreased Mcl-1 transcription as revealed by RT-PCR analysis (Fig. 2C). Similar to TRAIL, addition of CHX did not affect the Bcl-X L level either at the post-translation (  C2 (lanes 2 and b). rhCasp-3, recombinant human caspase-3; wt, wild type. tions in other apoptotic systems reported, strongly suggested that the regulation of the Mcl-1 level rather than Bcl-2 or Bcl-X L plays a critical role in the survival of TRAIL-sensitive cells. It further suggested that specific mechanisms might be involved in down-regulation of Mcl-1 in response to different death signals.
Caspase-3 Specifically Cleaved Mcl-1-To further address the issue of whether there is any caspase specificity in the proteolysis of Mcl-1 and more importantly to define the mechanism of Mcl-1 cleavage in the TRAIL signaling pathway, we analyzed which caspase was involved in specific segmentation of Mcl-1. To do this, we first performed in vitro caspase cleavage assays utilizing active forms of recombinant human caspases and purified Mcl-1. The results indicated that caspase-3 ( Fig. 3A) but not caspase-8 (Fig. 3B)  To verify whether the in vitro cleavage pattern of Mcl-1 by caspase-3 authentically represented that of in vivo digestion, we subjected Jurkat cells to increasing concentrations of TRAIL, and Western blotting analysis showed the appearance of the Mcl-1-C1 fragment (Fig. 3C, top). The reason why we could not detect Mcl-1-C2 was because the epitope used to generated the antibody (S-19, Santa Cruz Biotechnology) was not present in Mcl-1-C2 as suggested by others as well (24,48,50). Note that a larger SDS-polyacrylamide gel with overloaded proteins should be used for S-19 to detect Mcl-1-C1, which was usually too faint to observe in the regular minigel system (Fig.  1C). To better detect both Mcl-1-C1 and Mcl-1-C2 in vivo, we then treated HEK293T cells stably transfected with pEYFP-N1-Mcl-1 with TRAIL (Fig. 3C, bottom). Western blotting analysis using GFP monoclonal antibody showed the two specific polypeptide fragments migrated at molecular weights, after that of GFP was subtracted, similar to those of in vitro digestion ( Fig. 3A) with the density of Mcl-1-C2 higher than that of Mcl-1-C1. No further cleavage was observed with Mcl-1-C2. These results therefore strongly indicated that caspase-3 cleavage sites within Mcl-1 identified in vitro represented the authentic enzymatic substrate sites in vivo.
To identify the precise cleavage sites of caspase-3 within Mcl-1, we sliced out the two fast migrating bands from the Coomassie Blue-stained polyvinylidene difluoride membrane (Fig. 3A) and performed Edman peptide sequencing analysis (data not shown). Visual inspection of the derived peptide sequences positioned the cleavage sites of caspase-3 right after Asp-127 and Asp-157, respectively (Fig. 3D). The adjacent amino acid sequences (EELD2G and TSTD2G) matched the consensus caspase-3 cleavage motif and were rather conserved in mammals.
To further confirm the specificity of caspase-3 cleavage at Asp-127 and Asp-157 of Mcl-1, we performed site-directed mutagenesis of these two amino acids (D127A and D157E) individually or simultaneously. Digestion of 35 S-labeled wild type Mcl-1 with recombinant caspase-3 yielded the typical Mcl-1-C2 fragment with Mcl-1-C1 rarely detectable (Fig. 3E, lane b). The mutant derivatives of Mcl-1 showed site-specific resistance to caspase-3 cleavage in that incubation of Mcl-1-D127A with caspase-3 yielded Mcl-1-C2 specific to cleavage at Asp-157 by caspase-3 (Fig. 3E, lane d), and Mcl-1-D157E abolished the corresponding cleavage by caspase-3 but yielded Mcl-1-C1 specific to cleavage at Asp-127 (Fig. 3E, lane f). As expected, the double mutant Mcl-1-D127A/D157E completely abolished the digestion (Fig. 3E, lane h). The appearance of the slightly faster migrating band beneath Mcl-1 (Fig. 3E, lane h) was unseen in caspase-3 digestion of wild type Mcl-1 (Fig. 3, A and E, lanes d  and f). It could be the product of caspase-3 cleavage at its cryptic site in Mcl-1 when the favorite sites at Asp-127 and Asp-157 were both mutated. Also note that, because the typical yield of 35 S-labeled Mcl-1 was in the nanogram range, the caspase-3:Mcl-1 ratio used here was much lower than that for the in vitro digestion in Fig. 3A where 10 g of Mcl-1 was used. The similar resistance to caspase-3 by Mcl-1 mutant derivatives was also observed when TRAIL-primed apoptotic Jurkat T cell extracts were applied to 35 S-labeled proteins (Fig. 3E,  right panel). The specificity of caspase-3 cleavage at Asp-127/ Asp-157 was further confirmed in vivo where the stably expressed EYFP-Mcl-1-D127A/D157E double mutant in HEK293T cells was resistant to TRAIL-activated caspase proteolysis (Fig. 3C, lane 3). These results therefore strongly suggested that Asp-127 and Asp-157 of Mcl-1 served as the specific cleavage sites for caspase-3.

The C-terminal Domain of Mcl-1 Was Proapoptotic-
Whether the C-terminal domain of Mcl-1 is a proapoptotic factor still remains controversial (24,48,50). To first determine whether the C-terminal domain of Mcl-1 is an agonist or antagonist of apoptosis, we performed a transient transfection assay in which cell viability was measured with overexpression of Mcl-1-C2 (Mcl-1-C1 yielded similar results, data not shown). Transfection of an equal molar amount of Mcl-1-C2 caused a degree of cell death similar to that caused by tBid (Fig. 4A). More interestingly, there existed cooperation between Mcl-1-C2 and tBid since co-expression of these two factors further enhanced apoptosis (Fig. 4A) Mcl-1-C2-induced apoptosis could be partially blocked by overexpression of XIAP (Fig. 4B), which is the downstream inhibitor of caspase-9 activation or upstream inhibitor of activated caspase-3 (53). Mcl-1-C2-induced apoptosis was also partially inhibited by overexpression of Bcl-X L (Fig. 4B). Therefore, it was highly likely that the C-terminal domain of Mcl-1 functioned at mitochondria (also see below) by antagonizing the protection of mitochondrial integrity by Bcl-X L . TRAILactivated caspase-3 cleaved Mcl-1 to amplify mitochondrial activation of caspase-9 in a feed forward loop.
Mcl-1-C2 Provides a Scaffold for tBid, Bak, and VDAC-Bcl-2 family proteins function through hetero-and homodimerization with intrafamily members. The concerted proapoptotic effect of Mcl-1-C2 and tBid in transient co-transfection experiments suggested the physical interaction between these two molecules. To address this issue, we transiently co-transfected MCF-7 cells with pCMV-HA-tBid and pCMV-myc-Mcl-1-C2. Co-immunoprecipitation assays indicated that there existed direct physical contact between Mcl-1-C2 and tBid (Fig. 5A,  panel 3). This interaction was further confirmed by immunofluorescent staining in which both Mcl-1 and Mcl-1-C2 colocalized with tBid in mitochondria (Fig. 5B). We then performed truncational analysis (Fig. 5C) to assess the functional domain of Mcl-1 that was required to contact tBid by GST pull-down assays. Full-length Mcl-1 readily made protein-protein interaction with GST-tBid (Fig. 5D, lane 10 PEST domain of Mcl-1 (M1 and M2 mutants) abolished its interaction with tBid (Fig. 5D, lanes 8 and 9). This suggested the requirement of the C-terminal domain of Mcl-1 for contact with tBid. Indeed the C-terminal 78 amino acids of Mcl-1 containing just the BH2 domain (M6 mutant) was sufficient to mediate protein-protein contact with tBid (Fig. 5D, lane  14), reminiscent of the finding that BH2 domain of Bcl-2 interacts with Bid (54).
To understand how heterodimerization of Mcl-1-C2 and tBid potentially regulated mitochondrial apoptosis, we performed cytochrome c release assays as the first step. Similar to transient transfection experiments in which overexpression of Mcl-1-C2 led to apoptosis (Fig. 4A-B), addition of Mcl-1-C2 to isolated mitochondria caused cytochrome c release in a dosedependent fashion (Fig. 6A). Both Bid and tBid were able to induce cytochrome c release (Fig. 6B, lanes 2 and 3) that could be completely abolished by addition of the full-length Mcl-1 (Fig. 6B, lanes 8 and 9). There apparently existed increased cytochrome c release when both Mcl-1-C2 and tBid were added to the isolated mitochondria (Fig. 6B, lane 7 compared  with lanes 2 and 4). The inhibitory N-terminal domain of Bid (55) seemed to interfere with Mcl-1-C2 because Mcl-1-C2 induced a lesser amount of cytochrome c release in the presence of Bid as compared with the Mcl-1-C2/tBid reaction (Fig.  5B, compare lanes 4 and 6 and lanes 6 and 7). These results strongly suggested that removal of N-terminal domains of both Bid by caspase-8 and Mcl-1 by caspase-3 enabled the maximal mitochondrial perturbation, and the Mcl-1-C2/tBid heterodimer potentiated TRAIL-induced mitochondrial apoptosis.
The notion that tBid induces oligomerization of Bax and Bak and mitochondrial dysfunction suggests a direct contact among these molecules. The finding that the BH2 domain of Mcl-1 contacted tBid led us to further investigate whether Mcl-1 could also contact Bax or Bak. To do this, we performed GST pull-down assays using 35 S-labeled Bax or Bak. The results showed that Bak (Fig. 7A, lane 8) but not Bax (Fig. 7A, lane 7) could be pulled down by GST-Mcl-1-C2, suggesting the specificity of protein contact. Mcl-1-C2 was able to pull down more Bak in the presence of Bax than in the absence of Bax (Fig. 7A,  compare lanes 8 and 9), indicating that Bax might facilitate the interaction between Bak and Mcl-1-C2. On the other hand, the full-length Mcl-1 most likely did not contact either Bax or Bak in this particular in vitro system (Fig. 7A, lanes 4 and 5), but it was able to pull down Bak in the presence of Bax (Fig. 7A, lane  6). The latter was in agreement with recent intracellular coimmunoprecipitation results (22). The mechanisms of the Nterminal domain of Mcl-1 potentially inhibiting protein-protein interaction with Bak and contribution of Bax to Mcl-1-Bak interaction are now under investigation.
Recent studies strongly suggest that the VDAC might control the release of apoptogenic factors from mitochondria to cytosol in activation of the apoptotic cascade, and the transmembrane potential (⌬⌿ m ) and pore permeability of VDAC could be further modulated by Bcl-2 family proteins (4,7,8,41). As a first step to test whether Mcl-1 could be involved in regulation of VDAC function, we performed GST pull-down assays using various Bcl-2 family proteins and recombinant human VDAC1. In agreement with our previous results (41), GST-Bax and GST-Bcl-X L interacted with VDAC1 (Fig. 7B, lanes 1 and 2). Interestingly Mcl-1-C2 but not the full-length Mcl-1 was able to make protein-protein interaction with VDAC1 (Fig. 7B, lanes 4  and 5). Moreover the BH3 domain of Mcl-1 in a GST fusion protein was able to pull-down VDAC1 (Fig. 7C, lane 6), suggesting that the BH3 domain of Mcl-1 was sufficient to make direct contact with VDAC1. These results suggested that the C-terminal domain of Mcl-1 might exert its proapoptotic function through regulation of the pore conformation of VDAC1. Given the interactions of Mcl-1-C2 with tBid, Bak, and VDAC1, it is reasonable to postulate that the proteolytic cleavage of Mcl-1 by caspase-3 to remove the inhibitory N-terminal domain and expose the proapoptotic C-terminal domain might be an essential checkpoint for TRAIL-induced apoptosis in Jurkat T cells.
To address the biological significance of caspase-3 cleavage of Mcl-1 in response to TRAIL signaling, we transiently overexpressed wild type Mcl-1 and mutant derivative Mcl-1-D127A/ D157E in HEK293T cells (Fig. 7C). Declining cell survival was observed along the course of TRAIL treatment with mock transfection. Surprisingly overexpression of Mcl-1 led to enhanced TRAIL-induced apoptosis. This could be attributed to increased production of proapoptotic C-terminal fragments of Mcl-1 resulting from caspase-3 cleavage of exogenously expressed Mcl-1 (data not shown). This notion was strengthened by the observation that overexpression of the Mcl-1-D127A/ D157E mutant effectively alleviated TRAIL-induced apoptosis possibly due to its failure to produce the proapoptotic C-terminal domains of Mcl-1 (Fig. 7C). Therefore, depending on whether the caspase-8/caspase-3 axis was activated, Mcl-1 either inhibited ectopic tBid-induced apoptosis, which bypassed caspase-8 activation (Fig. 4A), or promoted TRAIL killing where the caspase-8/caspase-3 cascade is turned on. To substantiate this observation, we knocked down the expression of endogenous Mcl-1 in HEK293 cells using siRNA (Fig. 7D, inset) and measured the survival rates after TRAIL engagement of cells reintroduced with Mcl-1 or Mcl-1-D127A/D157E (Fig. 7D). The results showed that down-regulation of Mcl-1 by siRNA sensitized cells to TRAIL, while complement of cells with exogenous wild type Mcl-1 further enhanced apoptosis in good agreement with the observation in Fig. 7C. However, overex- pression of Mcl-1-D127A/D157E could effectively rescue cells from TRAIL apoptosis. These results therefore support the hypothesis that TRAIL signaling required the specific removal of the inhibitory N-terminal domain of Mcl-1 by caspase-3, allowing the proapoptotic C-terminal domain of Mcl-1 to contact tBid, Bak, and VDAC1. Whether the C-terminal domain of Mcl-1 could functionally substitute Bax in a Bax/Bak oligomer remains to be investigated. DISCUSSION It has recently been shown that, for apoptosis to occur, the antiapoptotic function of Mcl-1 has to be removed through either degradation or specific cleavage. Here we showed that, besides the specific cleavage and activation of Bid by caspase-8 and caspase-3, apoptosis of Jurkat T cells induced by TRAIL required a specific cleavage of Mcl-1 by caspase-3 but not proteasomal degradation, while other prototypic antiapoptotic factors such as Bcl-2 or Bcl-X L seemed not to be involved. We further established for the first time that the physical interaction and cooperation between tBid and the C-terminal domain of Mcl-1 (Mcl-1-C2) resulted from caspase-3 cleavage in promoting TRAIL apoptosis. The finding that Mcl-1-C2 effectively targets to mitochondria and contacts both Bak and VDAC1 suggests that TRAIL might induce the formation of a tetrameric complex of Mcl-1-C2⅐tBid⅐Bak⅐VDAC1 in mediating its apoptotic signaling in Jurkat T cells.
Proteolytic cleavage by caspases of various Bcl-2 family proteins (Bcl-2, Bcl-X L , Bid, and Bax) and non-Bcl-2 family members (XIAP, receptor-interacting protein, and mitogen-activated protein kinase/extracellular signal-regulated kinase kinase kinase) in regulation of death and survival signals is not uncommon (56). We and others demonstrated that Mcl-1 could be cleaved by caspase-3 (24,48,50) in response to death ligands or chemotherapeutic agents. This line of evidence strongly indicates that proteolytic cleavage of Mcl-1 might be a general mechanism in the regulation of both extrinsic and intrinsic apoptosis. The caspase-3 substrate sites that we have identified are very intriguing. On the one hand, the Asp-127 site (EELD2G, which yields Mcl-1-C1) is highly conserved from nematode to human, while the Asp-157 site (TSTD2G, which yields Mcl-1-C2) is somehow more variable. Careful inspection of the compiled caspase substrate motifs revealed that the sequence TSTD2G was closer to caspase proteolytic sites of Bcl-2 family proteins, while EELD2G was identical to that of the transcription factor NRF2 involved in antioxidant gene expression and highly homologous to factors involved in protein synthesis, DNA binding and repair, signal transduction, and metabolism (56). Our in vitro and in vivo experiments also demonstrated that the Asp-157 site represented the more favorable caspase-3 cleavage site than Asp-127 and suggested further a general action for caspase-3 upon Bcl-2 family members in effecting apoptosis. Whether caspase-3 cleaves Mcl-1 sequentially or simultaneously remains unknown.
Mcl-1 distinguishes itself from the other prosurvival Bcl-2 family proteins because it lacks a true antiapoptotic BH4 domain (15) but possess a long N terminus containing the PEST domain with the debatable function of targeting Mcl-1 to the proteasome degradation pathway. Our data showed that dissociation of this portion from the C-terminal domain of Mcl-1 seemed necessary for TRAIL-induced apoptosis in Jurkat cells. Our in vitro data further suggested that the N-terminal domain of Mcl-1 might function as a repressive domain that interfered with the interaction between Mcl-1 and Bak and VDAC1, therefore making Mcl-1 an antiapoptotic Bcl-2 family member. On the other hand, the precise function of the C-terminal domain of Mcl-1, whose amino acid sequence is highly similar to Bax, is still controversial. We and Michels et al. (24) observed that overexpression of the C-terminal domain of Mcl-1 promotes apoptosis. This is reminiscent of previous results that removal of the BH4 domain of Bcl-2 or Bcl-X L by caspases converted it into a Bax-like, potent proapoptotic factor (57,58).
Our findings that Mcl-1, but not Bcl-2 or Bcl-X L , is proteolyzed and activated by caspase-3 in response to TRAIL in Jurkat cells is also very interesting. The protection of prosurvival Bcl-2 family proteins to TRAIL apoptosis generally depends upon both cell type specificity and whether the mitochondrial apoptotic pathway is evoked (mitochondria-independent type I cells versus mitochondria-dependent type II cells) (59). The functional link between TRAIL and Mcl-1 cleavage and activation shortcuts the requirement of Bcl-2 or Bcl-X L in Jurkat T cell death and therefore partially addresses the previous gene targeting results of TRAIL and Bcl-2 family members. Our finding that the C-terminal domain of Mcl-1 contacted tBid and Bak might delineate a higher order of specificity in signaling for cell death by Bcl-2 homologs and might provide insight into the molecular composition and mechanisms of the TRAIL-activated mitochondrial apoptosis signaling pathway.
The direct interaction and functional cooperation between Mcl-1 and Bid revealed in this work is very striking. We observed that full-length Mcl-1 could inhibit tBid-induced cytochrome c release from isolated mitochondria and apoptosis in co-transfection assays. This is reminiscent of the findings that Mcl-1 interacts with and inhibits the function of another BH3only Bcl-2 family protein, Bim (17,23,50). We further observed that the C-terminal domain of Mcl-1 and tBid had a concerted effect in promoting cytochrome c release and apoptosis. This line of evidence therefore suggested for the first time that, unlike the case in Mcl-1-Bim interaction, there might exist a collaborative pathway of caspase cleavage of Bcl-2 family proteins and activation of mitochondrial apoptosis by these factors in TRAIL signal transduction.