|Home | About | Journals | Submit | Contact Us | Français|
Mitochondria play central roles in integrating pro- and anti-apoptotic stimuli and JNK is well-known to have roles in activating apoptotic pathways. We establish a critical link between stress-induced JNK activation, mitofusin 2, which is an essential component of the mitochondrial outer membrane fusion apparatus, and the ubiquitin-proteasome system (UPS). JNK phosphorylation of mitofusin 2 in response to cellular stress leads to recruitment of the ubiquitin ligase (E3) Huwe1/Mule/ARF-BP1/HectH9/E3Histone/Lasu1 to mitofusin 2, with the BH3 domain of Huwe1 implicated in this interaction. This results in ubiquitin-mediated proteasomal degradation of mitofusin 2, leading to mitochondrial fragmentation and enhanced apoptotic cell death. The stability of a non-phosphorylatable mitofusin 2 mutant is unaffected by stress and protective against apoptosis. Conversely, a mitofusin 2 phosphomimic is more rapidly degraded without cellular stress. These findings demonstrate how proximal signaling events can influence both mitochondrial dynamics and apoptosis through phosphorylation-stimulated degradation of the mitochondrial fusion machinery.
Mitochondria are highly dynamic organelles that continually undergo fusion and fission of both their outer and inner membranes. The equilibrium between these processes determines mitochondrial morphology (Bereiter-Hahn and Voth, 1994; Sesaki and Jensen, 1999), which in general is a highly interconnected tubular network. The fidelity of mitochondrial membrane fusion and fission is required to maintain the electrochemical gradient necessary for oxidative phosphorylation, for a variety of anabolic processes, for proper DNA inheritance and for appropriate cellular responses to apoptotic stimuli (Chen et al., 2005b; Chen et al., 2010; Lee et al., 2004; Liesa et al., 2009). Mitochondrial fission and fusion are carried out by distinct machineries, both of which employ large dynamin-like GTPases (Okamoto and Shaw, 2005). In higher eukaryotes, there are two such GTPases that are integral to mitochondrial outer membrane (MOM) fusion, referred to as mitofusin 1 and 2 (Mfn1 and Mfn2) (Rojo et al., 2002).
Mfn1 and Mfn2 are essential for mitochondrial fusion and function and form both homo- and heterotypic complexes (Chen et al., 2003; Hoppins et al., 2011; Detmer and Chan, 2007). Critical to their activity are their GTPase domains, as well as heptad repeat coiled-coil domains that interact in trans with mitofusin molecules on apposing mitochondria, thereby initiating early events leading to outer membrane fusion (Koshiba et al., 2004). Mfn1 and Mfn2 exhibit ~60% amino acid identity and appear to function in part through the generation of hetero-oligomers, with wild type Mfn1 having the capacity to complement disease-associated mutations in Mfn2 (Detmer and Chan, 2007). Specific non-redundant roles have been identified, as exemplified by the finding that Mfn1 null embryos are smaller at earlier stages of development, while deletion of Mfn2 results in earlier lethality (Chen et al., 2003). Mfn1 appears to play the dominant role in mitochondrial fusion by effectively tethering mitochondria and exhibiting greater GTPase activity than Mfn2 (Ishihara et al., 2004); Mfn2 has been demonstrated to be more important in differentiated tissues. In particular, Mfn2 is crucial to mitochondrial function in Purkinje cells (Chen et al., 2007) and mutations of Mfn2, primarily in its GTPase domain, are causal for Charcot-Marie-Tooth 2A (CMT2A) neuropathy, and can be associated with optic atrophy (Casasnovas et al., 2009; Zuchner et al., 2004). Mfn2 has also been established as having functions that are not obviously related to mitochondrial membrane fusion. These include roles in sequestering p21Ras (Chen et al., 2004), maintaining the mitochondrial electrochemical gradient (Pich et al., 2005) and in generating contacts between mitochondria and the endoplasmic reticulum (de Brito and Scorrano, 2008).
The function and levels of mitofusins are interwoven with the role of the mitochondria in apoptosis. Results from studies based on Mfn2 overexpression support an anti-apoptotic role for this protein (Jahani-Asl et al., 2007; Sugioka et al., 2004). Mitochondrial integrity is critical for preventing the release of cytochrome C, and loss of mitochondrial fusion and generation of fragmented mitochondria correlates with apoptosis. However, the causal role of fragmentation in apoptosis remains unresolved in the literature (Autret and Martin, 2009; Jahani-Asl et al., 2007; Lee et al., 2004; Sheridan et al., 2008). An even more complex relationship exists between mitofusins and the pro-apoptotic members of the Bcl-2 family, Bax and Bak, which have been shown to associate with both Mfn1 and Mfn2 (Brooks et al., 2007; Karbowski et al., 2006; Neuspiel et al., 2005; Perumalsamy et al., 2010; Hoppins et al., 2011).
Transcription of Mfn2 is known to be upregulated by peroxisome proliferator-activated receptor γcoactivator-1β (PGC-1 β) in MEFs (Liesa et al., 2008). However, the post-translational regulation of Mfn2 remains less clear. Over the last 2 years, several studies have provided evidence for ubiquitination of mitofusins by Parkin as part of the process leading to autophagy (Poole et al., 2010; Ziviani et al., 2010; Glauser et al., 2011; Gegg et al., 2010). A recent study suggests that the role of Parkin at damaged mitochondria is more general and that it is involved in the remodeling of the mitochondrial outer membrane preceding autophagy by stimulating the ubiquitination and proteasomal degradation of proteins with diverse functions in mitochondrial fusion and fission as well as in solute and protein transport (Chan et al., 2011). In mammals, Mfn1 levels inversely correlate with expression of the ubiquitin ligase (E3) March V, although a direct link to degradation has not been established (Park et al., 2010). Finally, MG132, which inhibits proteasome function, has been shown to increase Mfn2 steady-state levels (Karbowski et al., 2007).
In this study we examine the relationship between Mfn2 and stress-induced apoptosis. We establish that Mfn2 is a substrate for Jun N-terminal kinase (JNK), which becomes activated in response to genotoxic stresses including doxorubicin (DR; Adriamycin) as well as other cellular stresses. Phosphorylation by JNK leads to enhanced ubiquitination and proteasomal degradation of Mfn2. Moreover, we demonstrate that the large HECT domain ubiquitin ligase (E3) Huwe1 is recruited to Mfn2 in response to this phosphorylation, and that Huwe1 recruitment is responsible for the accelerated degradation of Mfn2. Further, while loss of Huwe1 expression has been shown to have an anti-apoptotic role in response to genotoxic stress (Zhong et al., 2005), this is bypassed by loss of Mfn2 expression, consistent with a causal role for Huwe1-mediated Mfn2 ubiquitination and proteasomal targeting in stress-induced apoptosis.
It is well known that treatment of many cell types, including the sarcoma U2OS, with the genotoxic chemotherapeutic agent doxorubicin results in apoptosis. We find that doxorubicin induces fragmentation of mitochondria (Figure 1A), which first becomes detectable after ~4 hr of treatment (not shown). To begin to assess whether critical components of the mitochondrial fusion machinery might be altered by this genotoxic stress, we examined changes in levels of Mfn1 and Mfn2. Doxorubicin treatment resulted in loss of Mfn2, but not of Mfn1 (Figure 1B). Similarly, neither the integral mitochondrial protein TOM40 nor the soluble dynamin-like component of the mitochondrial fission apparatus DRP1 decreased. This selectivity excludes mitophagy as an explanation for the loss of Mfn2. Pulse-chase experiments revealed that Mfn2 degradation increases in response to doxorubicin (Figure 1C and 1D). This, together with doxorubicin-induced increase in Mfn2 ubiquitination (Figure 1E), indicates that the decrease in Mfn2 (Figure 1B) is not simply due to known inhibitory effects of doxorubicin on protein expression (Momparler et al., 1976), but rather is consistent with accelerated degradation of Mfn2 by the ubiquitin-proteasome system (UPS). HeLa cells, which do not express endogenous Parkin, an E3 required for degradation of multiple mitochondrial proteins as part of the process leading to mitophagy (Chan et al., 2011), also show enhanced degradation of Mfn2 in response to doxorubicin (Figure S1A).
The selective loss of Mfn2 suggests that this might play a causal role in apoptosis in this setting. Alternatively its loss could be secondary to apoptotic changes. To directly assess the role of Mfn2 in apoptosis, we examined caspase activation after Mfn2 knockdown. Loss of Mfn2 alone did not induce apoptosis as assessed by caspase 3 activation; however, it markedly sensitized cells to doxorubicin-induced apoptosis (Figure 1F and 1G; see Supplemental Experimental Procedures for complete list of RNAi reagents used in this study). This sensitivity is reversed by re-expression of Mfn2 (Figure S1B). Consistent with an anti-apoptotic role for Mfn2, overexpression of Mfn2 protected cells from UV irradiation comparable to a dominant negative anti-apoptotic DRP1 mutant (DRP1K38A) (Frank et al., 2001) and to the apoptosis inhibitor Bcl-2 (Figure S1C).
To explore the regulation of Mfn2, we generated a stable transfectant of U2OS that expresses Mfn2 with C-terminal HA tags (U2OSMfn2HA) at a level that approximates the endogenous protein (Figure S2A). Notably, the levels of endogenous Mfn2 are reduced in all cell lines stably expressing exogenous Mfn2, suggesting that levels of Mfn2 are tightly regulated (data not shown). We then purified Mfn2HA from U2OSMfn2HA cells treated with proteasome inhibitor for mass spectrometry (MS) analysis. Tandem MS analysis showed over 80% coverage of Mfn2, and identified multiple ubiquitination signatures throughout (data not shown). An unanticipated observation was phosphorylation of Ser27. Direct evidence of phosphorylation has not been previously described for components of the mitochondrial fusion apparatus. Using Group-based Prediction Software 2.1, the peptide sequence 25NASPLK30 scored highest for several mitogen-activated protein (MAP) kinase family members involved in stress responses (Xue et al., 2008). To confirm Ser27 phosphorylation, we immunoprecipitated Mfn2 and immunoblotted with antibody that recognizes S*PXR/K [where (*) represents Ser phosphorylation] (Ritt et al., 2010). A low level of reactivity was observed with proteasome inhibition (MG132) that increased markedly with doxorubicin, which activates multiple MAP kinases (Niiya et al., 2004) (Figure 2A). To further validate phosphorylation of Mfn2 Ser27, we generated a Mfn2 phospho-Ser27 polyclonal antibody that was validated using a non-phosphorylatable (S27A) mutant, Mfn2HAS27A (Figure 2B). Using this antibody, we found that Mfn2 phosphorylation peaks at ~2 hr of treatment with doxorubicin and decreases thereafter (not shown). We also found that stress-induced phosphorylation of Mfn2 Ser27 could be induced in response to other agents including etoposide, 17-AAG, cis-platinum, tunicamycin and MG132 (Figure S2B).
To assess which stress-activated kinases might be involved in Mfn2 phosphorylation, we examined the activation of JNK, ERK and p38. Doxorubicin activates JNK and p38 (Figure S2C). In contrast, doxorubicin fails to increase the activity of ERKs, which are constitutively activated in these cells, making these less likely to be responsible for the observed phosphorylation of Mfn2. The role of JNK and p38 in doxorubicin-induced phosphorylation of Mfn2 Ser27 was assessed by knockdown of JNK (Figure 2C). JNK knockdown resulted in a loss of Mfn2 phosphorylation, while knockdown of p38 was without effect. The role of JNK was confirmed using distinct siRNAs directed against the 3′ UTRs of JNK1 and JNK2 isoforms and re-expression of constitutively active JNK2α2 (Pimienta et al., 2007), which results in Mfn2 phosphorylation (Figure S2D and S2E). These data establish that Mfn2 is phosphorylated on Ser27 in response to a variety of cellular stresses and implicate JNK in this process. Co-immunoprecipitation was carried out to assess the interaction of Mfn2. Mfn2HA specifically associated with FLAG-JNK2α2, whereas GFP, which served as a control, showed no specific binding (Figure 2D). The functional interaction between the two proteins was confirmed by the Ser27 phosphorylation of Mfn2 by recombinant JNK2α2 in vitro (Figure 2E and Figure S2F).
To determine whether JNK has a role in doxorubicin-induced Mfn2 degradation, we monitored degradation of Mfn2 after JNK knockdown. This abrogated the doxorubicin-induced accelerated degradation of Mfn2 (Figure 3A and 3B). To investigate the significance of Ser27 phosphorylation in Mfn2 degradation, we assessed ubiquitination of wild type (Mfn2HA), Mfn2HAS27A and a Mfn2 phosphomimic (Mfn2HAS27D). The phosphomimic showed increased ubiquitination, in accord with a proteasome-targeting role for Ser27 phosphorylation (Figure 3C). Consistent with a destabilizing role for phospho-Ser27, Mfn2HAS27D exhibited a marked increase in the rate of degradation compared to Mfn2HAS27A, which was degraded with kinetics resembling Mfn2HA (Figure 3D). However, in response to doxorubicin, WT Mfn2 became less stable and degraded with kinetics resembling the phosphomimic. In contrast, Mfn2HAS27A degradation did not increase with doxorubicin (Figure 3E). Both the baseline degradation (WT and Mfn2HAS27A) and the accelerated degradation (Mfn2HAS27D) were largely prevented by the proteasome inhibitor lactacystin (Figure 3F), indicating UPS-mediated degradation.
We next assessed the effect of expressing Mfn2 and its Ser27 mutants on mitochondrial morphology after doxorubicin treatment. Transfection of Mfn2 and mutants reduced mitochondrial fragmentation compared to vector (Figure 4A and B), reinforcing the idea that doxorubicin-induced loss of Mfn2 promotes mitochondrial fission. Furthermore, consistent with its increased stability, Mfn2HAS27A reduced mitochondrial fragmentation more effectively than either wild type Mfn2HA or the relatively unstable Mfn2HAS27D (Figure 4A and 4B). In accord with these findings, knockdown of JNK1 and JNK2 resulted in a >85% decrease in mitochondrial fragmentation to doxorubicin at 6 hr (not shown). To determine the role of Ser27 phosphorylation in apoptosis, we assessed the effect of overexpressing Mfn2 Ser27 mutants on doxorubicin-induced cell death. Overexpression of Mfn2 and mutants reduced doxorubicin-induced apoptosis (Figure 4C; see Figure S3 for levels of Mfn2). Mfn2HAS27A was most protective in this assay, inhibiting apoptosis to a similar extent as a dominant negative anti-apoptotic DRP1K38A (Frank et al., 2001), and approaching that seen with overexpression of Bcl-2. In comparison, overexpression of Mfn2HAS27D was less effective (Figure 4C). A similar trend was observed with thapsigargin, which induces endoplasmic reticulum stress, activates JNK (Urano et al., 2000) and can induce cell death. Overexpression of Mfn2HAS27D was ineffective in inhibiting cell death whereas overexpression of Mfn2HAS27A was protective to a similar degree as Bcl-2 and DRP1K38A (Figure 4D). To determine whether Ser27 phosphorylation and the consequent decrease in stability plays a role in sensitizing cells to apoptosis, we examined apoptosis in cells depleted of endogenous Mfn2 but re-expressing WT or Ser27 mutants of Mfn2. Cells expressing Mfn2HAS27D showed increased caspase 3 activation when treated with doxorubicin. In contrast, cells re-expressing Mfn2HAS27A exhibited resistance to doxorubicin-induced caspase 3 activation (Figure 4E), establishing a role for Ser27 phosphorylation in sensitization to stress-induced cell death.
Specificity in ubiquitination is conferred by ubiquitin-protein ligases (E3s), which together with ubiquitin-conjugating enzymes (E2s), mediate the transfer of ubiquitin to substrates. Parkin targets mitofusins (Poole et al., 2010; Ziviani et al., 2010; Glauser et al., 2011; Gegg et al., 2010) and other mitochondrial proteins in the process leading to mitophagy (Chan et al., 2011). However, the stress-induced degradation of Mfn2 is specific for this protein and occurs in cells lacking Parkin, pointing to another E3 specific for Mfn2. Notably, our MS/MS analysis of proteins co-precipitating with Mfn2HA identified with high confidence a 4,374 aa, HECT domain E3 Huwe1 [HECT, UBA, and WWE containing protein 1; a.k.a. Mcl-1 ubiquitin ligase E3 (Mule), ARF binding protein 1 (ARF-BP1), E3Histone and HectH9; Figure S4A]. Among its targets are both pro- and anti-apoptotic substrates (see Discussion). In addition to its catalytic C-terminal HECT domain, Huwe1 includes a Bcl-2 homology domain 3 (BH3), a WWE domain and an ubiquitin-associated (UBA) domain (Figure 5A). Association of Mfn2 and Huwe1 was confirmed by co-immunoprecipitation using catalytically inactive Huwe1(FLAG-Huwe1C4341A; Figure S4B). Interestingly, Mfn2 has been shown to interact with two other BH3 containing pro-apoptotic proteins, Bax and Bak (Brooks et al., 2007; Hoppins et al., 2011; Waxman and Kolliputi, 2009). Additionally, Huwe1/Mule is implicated in apoptosis by ubiquitination of Mcl-1, to which it binds through its BH3 domain (Zhong et al., 2005). To determine whether a BH3-mediated interaction similarly exists between Huwe1 and Mfn2, we assessed binding of GST-Huwe1BH3 (aa 1904 to 2065) to in vitro translated Mfn2 (Figure 5B). In contrast to WT and S27A Mfn2, the phosphomimic, S27D, showed substantial binding to GST-Huwe1BH3. As the BH3 domain represents only about 15 of the residues in GST-Huwe1BH3, we mutated the four most conserved BH3 amino acids, including two that are essential for BH3-mediated interactions (Wang et al., 1998). This mutation (BH3m) decreased the specific interaction with Mfn2HAS27D (Figure 5C), suggesting that Ser27 phosphorylated Mfn2 recruits Huwe1 at least in part through the latter’s BH3 domain.
To further evaluate the Mfn2 Huwe1 interaction, we assessed co-immunoprecipitation of both proteins when transfected in cells (Figure 5D and 5E). WT Mfn2HA but not Mfn2HAS27A interacted with Huwe1. Importantly, the phosphomimic exhibited significantly greater association. These findings strongly suggest that phosphorylation of Mfn2 Ser27 recruits Huwe1, and that this association is mediated, at least in part, through the Huwe1 BH3 domain.
We next asked whether Huwe1 is responsible for stress-induced degradation of Mfn2. Mfn2 stability was assessed in a U2OS line expressing a doxycycline-inducible shRNA for Huwe1 (U2OSshHuwe1) (Zhong et al., 2005). Doxycycline results in a greater than 90% decrease in Huwe1 (Figure S5A). This resulted in the abrogation of doxorubicin-accelerated Mfn2 degradation (Figure 6A) as well as a marked decrease in Mfn2 ubiquitination (Figure 6B). To confirm this, we transiently depleted Huwe1 in U2OS cells with siRNAs directed against Huwe1 (Figure 6C). Doxorubicin failed to elicit accelerated degradation of Mfn2 in cells acutely depleted of Huwe1 (Figure 6D, E). Importantly, loss of Huwe1 expression did not inhibit JNK activation in response to doxorubicin (Figure S5B). To rule out off-target effects, we knocked down Huwe1 using an shRNA directed against its 3′ UTR and re-expressed either wild type or catalytically inactive Huwe1 in cells. Re-expression of WT, but not catalytically inactive Huwe1, restored doxorubicin-induced degradation of Mfn2 (Figure 6F–6H).
To evaluate the relationship between Huwe1 and mitochondrial fragmentation, we examined the effects on mitochondrial fragmentation when Huwe1 was knocked down. Loss of Huwe1 expression protected against mitochondrial fragmentation in response to doxorubicin (Fig. 7A). This is consistent with the function of Mfn2 in promoting mitochondrial fusion and the role established herein for Huwe1 in targeting of Mfn2 for degradation.
Loss of expression of Huwe1 has previously been shown to inhibit apoptosis, which is correlated with stabilization of Mcl-1 in response to cis-platinum (Zhong et al., 2005) and of degradation of Miz1 in response to TNF-α (Yang et al., 2010). We similarly found that loss of Huwe1 protect cells against doxorubicin-induced cell death, as assessed by both caspase activation and direct cellular visualization, and that this improved cell survival correlated with stabilization of Mfn2 (Figure 7B and 7C). More importantly, Figures 7B and 7C demonstrate that the anti-apoptotic effect of Huwe1 knockdown was lost when Mfn2 was also knocked down. These findings are consistent with a function of Huwe1 ‘upstream’ of Mfn2.
Mitochondria are a nexus for pro- and anti-apoptotic signaling (Autret and Martin, 2009). However, until now no mechanistic link between upstream sensors of stress and the mitochondrial fusion or fission machineries has been established. We demonstrate that a critical component of the mitochondrial fusion apparatus, the mitofusin Mfn2, is a target for phosphorylation in response to a variety of cellular stresses. We provide direct evidence that JNK mediates this phosphorylation. This results in the recruitment of the large HECT domain E3 Huwe1, which in turn ubiquitinates Mfn2 leading to its proteasomal degradation resulting in enhanced apoptosis in response to cellular stress. Despite their overall similarity, this stress-induced degradation is not observed for Mfn1, which correlates with the stability of Mfn1 under basal conditions (t1/2 ~ 40 hr) (Tsai and Weissman unpublished observations). It has been established that Mfn1 and Mfn2 can form both homo- and heterodimers and that heterodimers are more efficient in mediating mitochondrial membrane fusion (Detmer and Chan, 2007; Hoppins et al., 2011). By controlling the degradation of Mfn2 in response to stress, Huwe1 can therefore regulate Mfn1-Mfn2 heterodimers and consequently the efficiency of mitochondrial fusion. Thus, while the causal relationship between mitochondrial fragmentation and apoptosis is yet to be definitively established, the net effect of targeting Mfn2 for degradation is to promote both processes.
It is probable that other E3s are involved in the constitutive regulation of Mfn2, as evidenced by the finding that basal Mfn2 degradation proceeds with both knockdown of Huwe1 and with expression of non-phosphorylatable Mfn2 and that its degradation is largely inhibited by the proteasome inhibitor lactacystin, regardless of the status of Ser27 (Figure 3D and 3F). While Parkin can mediate the degradation of Mfn2 in response to mitochondrial damage and there is data suggesting a role for March V (Nakamura et al., 2006), it is likely that other ubiquitin ligases can associate with Mfn2. A major ubiquitin ligase for Fzo1, the sole yeast mitofusin, is SCFMdm30 (Cohen et al., 2008). There is no known mammalian ortholog for the substrate recognition element of this E3, Mdm30. However, considering that there is only 12% identity between Fzo1 and Mfn2, there may be other F-box proteins without obvious homology to Mdm30 playing a role in regulating Mfn2.
The functionally divergent set of known Huwe1 substrates includes Mcl-1, p53, C-Myc, N-Myc, Miz1, HDAC2, MyoD, TopBP1 and histones (Chen et al., 2005a; Zhao et al., 2008; Zhong et al., 2005; Zhang et al., 2011; Yang et al., 2010; Liu et al., 2005; Noy et al., 2012; Herold et al., 2008; Adhikary et al., 2005). Given this, the net effect of Huwe1, like that of the stress-activated kinases, is likely context-dependent as has been suggested (Shmueli and Oren, 2005) and likely cell-type dependent as well. In this regard, Mfn2 and Huwe1 have both been implicated in development of the cerebellum (Chen et al., 2007; D’Arca et al., 2010). As with other E3s, a variety of factors undoubtedly control the net effect of Huwe1. However, based on our findings and in the context of the literature, it would appear that a major role of Huwe1 at mitochondria is to inactivate two key anti-apoptotic molecules, Mcl-1 and Mfn2.
The size of Huwe1 (4,374 aa) makes a detailed structure-function analysis challenging. Roles for the UBA or WWE domains in Mfn2 interactions cannot be discounted, and indeed the UBA could play a role in enhancing ubiquitination as seen with ubiquitin-binding domains in other E3s (Chen et al., 2006; Morito et al., 2008; Peschard et al., 2007). Nevertheless, the Huwe1 BH3 domain mediates interactions between Huwe1 and Mfn2 in what appears to be an Mfn2 phosphorylation-dependent manner. In this regard, it is of particular interest that mitofusins interact with pro-apoptotic Bcl-2 family members, Bax and Bak. A change in the relative binding of these molecules to each mitofusin occurs in response to pro-apoptotic cellular stresses with Bak dissociating from Mfn2, and associating with Mfn1 (Brooks et al., 2007; Waxman and Kolliputi, 2009). In addition, Bax co-localizes with Mfn2 and DRP1 at sites of mitochondrial fission during apoptosis (Karbowski et al., 2002), and Bax and Bak expression correlates with the formation of discrete high molecular weight complexes containing Mfn2 and tubular mitochondrial morphology (Karbowski et al., 2006). There is also evidence that a GTPase hydrolysis-deficient mutant of Mfn2 blocks activation of Bax and apoptosis (Neuspiel et al., 2005). In C. elegans, the anti-apoptotic Bcl-2 like protein, CED-9, interacts with mitofusin and activates fusion (Rolland et al., 2009). Along these lines, it has recently been shown that in a cell free system a soluble form of Bax, but not the membrane inserted form capable of oligomerization and promoting apoptosis, can promote Mfn2-mediated fusion (Hoppins et al., 2011). As Huwe1 interacts with Mfn2 at least partially in a BH3-dependent manner and similarly targets Mcl-1 for degradation dependent on the Huwe1 BH3 domain, our findings in the context of the literature suggests that there is complex network of highly regulated BH3-dependent interactions surrounding the mitochondrial fusion apparatus that is likely to play an important role in regulating cellular fate in response to stress. The possibility should now be considered that interactions between Mfn2 and Bcl-2 family proteins are similarly regulated by the Mfn2 phosphorylation described herein and that this facilitates recruitment of Bcl-2 family members to the mitochondria. We routinely detect ~2% of Mfn2 being found phosphorylated in response to cell stress; this is likely an underestimate due to phosphatases and phospho-antibody efficiency (unpublished observations). Given the complex relationship between mitochondrial fragmentation and apoptosis, as well as between aerobic glycolysis and tumor growth (Vander Heiden et al., 2009), it now becomes of great interest to explore the role of Mfn2 phosphorylation in tumor growth and apoptosis in vivo.
Cells were transfected for 24–48 hr before metabolic labeling as previously described (Tsai et al., 2007). Briefly, cells were starved in Met and Cys free media for 30 min and labeled for either 30 min with 200 μCi/mL or 18 hrs with 20 μCi/mL (total 100 μCi) Trans 35S Label (MP Biologicals) – half-lives are indistinguishable under the two conditions, compare for example Figures 1C (18 hr) and 3A (30 min). Cells were then washed 3X with PBS pH 7.4 followed by chase in the presence of complete media supplemented with unlabeled Met and Cys. After lysis in IPB1 or IPB2 (see Supplemental Experimental Procedures) Mfn2HA was subject to IP, washed with lysis buffer and processed for autoradiography. Quantification was performed using Storm Phosphorimager and ImageQuant software (GE Healthcare, Waukesha, WI).
For the experiment shown in Figure 2E, U2OS cells stably expressing Mfn2HA were lysed in IPB2 without MG132 and the soluble fraction immunoprecipitated with HA antibody and washed extensively. Samples were resuspended in 40 μL of buffer containing 25 mM TRIS pH 7.5, 25 mM NaCl, 10 mM MgCl2, 0.1 mM Na3VO4, 1mM DTT, 0.01% Triton-X-100, 0.25 mM EGTA, 0.1 μM calyculin A, 400 μM ATP and the indicated amounts of purified recombinant 6xHis-JNK2α2. After incubation for 30 min at 30°C reactions were terminated by heating to 70°C in reducing SDS-PAGE sample buffer and immunoblotting carried out as indicated.
To assess mitochondrial morphology, cells were transfected with mito-dsRED plasmid. Images were collected by confocal microscopy (Zeiss LSM510) and analyzed with ImageJ 1.42. Fragmentation was judged based on the mitochondrial distribution of a normal cell, and cells exhibiting over 80% mitochondrial fragmentation were counted as fragmented. For each condition, at least 200 cells were counted. Quantification was carried out by an individual blinded to the conditions. To assess cell death induced by doxorubicin, cells were scored based on Annexin V or DAPI staining on an epifluorescence microscope (Zeiss Axiovert 200M, Hamamatsu Photonics ORCA-ER camera). For each time point, more than 300 cells were counted. All fragmentation and cell death experiments were performed in triplicate and data represent mean ± standard deviation (SD) for three experiments.
We thank Mickael Cohen, Gustavo Gutierrez, Antonio Iavorone, Anna Lasorella, Stephen Lockett, Deborah Morrison, Ze’ev Ronai, Daniel Ritt and Thomas Turbyville for helpful discussions; Wei Gu, Gustavo Gutierrez, J. Silvio Gutkind, Wu Ou, Ze’ev Ronai and Xiaodong Wang for invaluable reagents; and Stanley Lipkowitz, and Daniel Stringer for critical review of this manuscript. This work was supported by the National Cancer Institute (NCI), National Institutes of Health (NIH) Intramural Research Program and Contract NO1-CO-12400 (MZ and TDV), a NCI Director’s Innovation Award (YCT), by a grant from the Michael J. Fox Foundation (AMW) and a grant from the USA-Israel Bi-National Science Foundation (AMW and MHG).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.