Main

Many human cancers demonstrate de-regulation of the c-Myc (Myc) oncoprotein, a bHLH-ZIP (basic helix-loop-helix-leucine zipper) transcription factor that modulates genes transcribed by all three RNA polymerases.1, 2 Among Myc’s targets are genes whose products regulate metabolism, including enzymes involved in glycolysis and oxidative phosphorylation (OXPHOS).3, 4, 5 This property likely accounts for Myc’s capacity to promote the Warburg effect, which is defined as the generation of ATP via glycolysis despite sufficient oxygen to support OXPHOS.5, 6 However, the Warburg effect and OXPHOS are not mutually exclusive activities, and both glycolysis and OXPHOS are increased and de-regulated in Myc overexpressing cells.4, 7, 8 It is likely that these processes represent different facets of the flexible metabolic state that supports the high rates of proliferation and anabolism of transformed cells.5, 9, 10 By providing for the rapid but inefficient generation of ATP, glycolysis also promotes tumor survival in hypoxic environments generated by inadequate vascular perfusion.5, 9, 11

Additional Myc target genes include those involved in mitochondrial and ribosomal biogenesis, biomass accumulation and survival.2, 4, 12 Mitochondrial mass and interconnectivity, cellular mass and protein synthetic rates are increased by aberrant Myc expression, whereas the opposite occurs in myc−/− cells.4, 13, 14, 15 Myc ablation is also associated with ATP depletion that likely arises from a combination of atrophic mitochondria, reduced glycolysis and OXPHOS and electron transport chain (ETC) dysfunction.4 Myc’s role in survival involves both intrinsic and extrinsic apoptotic pathways, which converge at the mitochondrial level.16 For example, Myc activates certain pro-apoptotic members of the Bcl-2 family such as Bax and Bim and suppresses anti-apoptotic members, such as Bcl2 itself and Bcl-XL.16 Highly coordinated interactions among these members are needed to ensure maximal control over these survival pathways.16

Mitochondria normally engage in a dynamic interplay between fusion into large, interconnected reticular networks and fission, which produces smaller, fragmented organelles.17, 18, 19 Fusion is believed to improve respiration and prolong organelle lifespan by limiting the oxidation of mitochondrial contents as a result of their dilution with the undamaged molecules of larger and healthier organelles.17, 18 By contrast, fission reduces mitochondrial mass during periods of relative metabolic inactivity and aids in the removal of dysfunctional organelles.17, 18, 19, 20 Because of varying degrees of fitness among mitochondria at either end of the fission–fusion spectrum, these activities can exert significant effects on cell survival.17, 21, 22

The balance between fission and fusion is orchestrated by a group of proteins that localize to the outer or inner mitochondrial membrane.17, 18, 19 Among the major mammalian fission proteins, dynamin-related protein 1 (Drp1) has additional roles in maintaining mitochondrial shape, size, distribution and cristae remodeling.17, 22, 23, 24, 25 Purified Drp1 spontaneously forms oligomeric ring-like structures in vitro and reversibly localizes to sites of mitochondrial constriction during fission.26 These sites are also encircled by projections of endoplasmic reticulum that co-localize with Drp1, promote further contraction and lead to eventual mitochondrial scission.27

We have examined here how compromising the normal fission:fusion balance affects Myc’s ability to regulate metabolism and energy generation, cell size and survival. We show that constitutive Drp1 overexpression leads to a state of chronic, abortive mitochondrial hyperfission. These mitochondria structurally resemble those of myc−/− cells as do their dysfunctional OXPHOS and ETC profiles.4 Moreover, profound ATP depletion, also resembling that of myc−/− cells, likely explains the reduced cell mass and resistance to both Myc-dependent and Myc-independent apoptosis. These results underscore the necessity for maintaining normal mitochondrial dynamics and function in order for Myc to regulate several of its fundamental phenotypes. They suggest that ATP depletion protects cells from the highly pro-apoptotic consequences of Myc de-regulation, thus favoring the survival of tumor cells, particularly when nutrients and/or oxygen are limiting.

Results

Cell line characterization

Rat1a fibroblasts expressing a MycER (human Myc protein fused to the estrogen receptor hormone-binding domain) fusion protein28 were engineered to express stably a V5-epitope-tagged Drp1 protein (Drp1 cells) or the empty lentiviral vector alone (Vector cells) (Figure 1a). Staining with MitoTracker Green or 10-nonoyl-acridine orange (NAO) showed the mitochondrial mass of pooled Drp1 cells to be approximately half that of Vector cells (Figure 1b). These results were confirmed by stably expressing a Discosoma sp. red fluorescent protein fused to a mitochondrial signal peptide (DsRED-Mito). Confocal microscopy and flow cytometry also showed Drp1 cell mitochondrial mass to be reduced (Figure 1c and Supplementary Video S1). On average, these mitochondria were also smaller, less well connected than those of Vector cells and less clustered around the nuclear periphery as previously described in myc−/− cells.4 Enforced Drp1 expression is therefore sufficient to drive a reduction in overall mitochondrial mass and interconnectivity in Rat1a-MycER cells.

Figure 1
figure 1

Characterization of cell lines and mitochondria. (a) Stable expression of Drp1. Rat1a MycER cells were stably transduced with a lentiviral vector encoding a full-length, rat Drp1-V5 fusion protein (Drp1 cells) or with the empty parental vector (Vector cells). Drp1 protein expression was assessed by immunoblotting for V5 epitope expression. (b) Quantification of mitochondrial mass. Cell lines were evaluated by flow cytometry after staining with MitoTracker Green or NAO to quantify overall mitochondrial mass. Drp1:Vector mean fluorescence ratios are expressed above each flow histogram. The results shown are typical of at least 4–6 independent experiments performed at different times. (c) Representative confocal images obtained on cells stably transfected with a plasmid encoding DsRED2-Mito. Mean fluorescence intensities were also quantified by flow cytometry with the Drp1:Vector ratio being shown beneath the panel. See Supplementary Video S1 for additional images

Transmission electron microscopy confirmed that in contrast to the mitochondria of Vector cells those of Drp1 cells were smaller, less abundant, less well interconnected and contained a paucity of cristae (Figure 2a). These findings indicate that Drp1 deregulation leads to atrophic mitochondria incapable of forming highly interconnected networks. Moreover, these cumulative deficits lead to an uncompensated reduction in mitochondrial mass. In keeping with these findings, a comparison of several representative mitochondrial membrane and matrix proteins indicated that their levels in Drp1 cells were somewhat reduced (Figures 2b and c).

Figure 2
figure 2

Abnormal mitochondrial ultrastructure and mitochondrial fusion of Drp1 cells. (a) Transmission electron micrographs of Vector and Drp1 cells showing a high degree of mitochondrial interconnectivity and abundant, well-organized cristae patterns in the former cells and small, sparse mitochondria with a lack of interconnectivity and poorly formed cristae in the latter cells. (b) Immunoblotting for select mitochondrial membrane and matrix proteins in Vector and Drp1 cells. Proteins included sirtuins (sirt) 3 and 5, mitochondrial fission factor (MFF) and medium-chain acyl-CoA-dehydrogenase (MCAD). Typical results are depicted. (c) Quantification of the western blots shown in panel (b). Three separate immunoblots typical of those depicted in panel (b) were densitometrically scanned and normalized to β-actin. The values obtained for Vector cells were then arbitrarily set to 1. The results of relative mean band intensities±1 S.E. are shown. (d) Higher rates of mitochondria fusion in Drp1 cells. Vector and Drp1 cells stably expressing DsRed2-Mito or EGFP-Mito were mixed in equal numbers and fused with PEG for 1 min. After washing, cells were incubated in medium containing CHX for 2 h, fixed, stained with DAPI and examined by confocal microscopy. Mitochondrial fusion in the resultant heterokaryons was identified by the merging of green and red signals. The histogram indicates the degree of mitochondrial fusion in each cell pair and was generated by calculating the percentage of co-localized fluorescent signals from at least 20 randomly selected heterokaryons with the use of the CoLocalizer Pro software package. Fluorescence micrographs below the histogram show typical images of mitochondrial fusion in each cell type. The rapidly alternating fission and fusion activity in the Drp1 cells when compared with the Vector cells can best be appreciated in the time-lapse movies in Supplementary Video S2

To appreciate the dynamics of mitochondrial fission and fusion, we co-cultured DsRed2-Mito expressing Vector and Drp1 cells with identical cells expressing mitochondrial-localized enhanced green fluorescent protein (EGFP-Mito) and then induced plasma membrane fusion with polyethylene glycol (PEG). After fixation, heterokaryons expressing both DsRed2-Mito and EGFP-Mito were visualized for evidence of mitochondrial fusion by quantifying the co-localized red and green signals.4 As shown in Figure 2d, Drp1 cells showed a nearly 15-fold higher rate of mitochondrial fusion. These findings were confirmed and extended by concurrent live cell, confocal, time-lapse imaging of Vector and Drp1 cells stained with MitoTracker Green (Supplementary Video S2). To a greater degree than could be appreciated with the above PEG fusion studies, these experiments also revealed a much more dynamic population of mitochondria in the latter cells, even among those organelles not undergoing fission or fusion. Collectively, these results indicate that, despite their structural defects, Drp1 cell mitochondria are more actively engaged in fusion. However, the newly fused organelles appear to be unstable and subject to immediate de novo fission.

Structural and functional abnormalities in ETC components in Drp1 cells

Because the properties of Drp1 cell mitochondria were reminiscent of those from myc−/− rat fibroblasts,4, 15, 29 we compared the four mitochondrial ETC components plus Complex V (ATP synthase) in Vector and Drp1 cells using blue-native gel electrophoresis (BNGE).30, 31 Supercomplexes, comprised of Complexes I, III and IV30 and possessing higher intrinsic Complex I activity, can also be visualized by this method.4 As shown in Figure 3a, BNGE of mitochondria from Vector cells resolved each of the ETCs in proportions roughly equivalent to those we previously described for another strain of myc+/+ rat fibroblasts.4 By contrast, the BNGE profile of Drp1 mitochondria was strikingly different. Quantification of these differences (Figure 3b) showed significant reductions in Complex V (45%) and supercomplexes (38%). By contrast, Complexes II, III and IV were unaffected, whereas Complex I was reduced by 28%, although this did not reach statistical significance. A previously uncharacterized complex (Complex ‘X’), originally seen in mitochondria of myc−/− cells and shown to contain >40 subunits of Complexes II–V,4 was also increased by approximately twofold in Drp1 cells. Thus, Drp1 cell mitochondria contain abnormalities in supercomplexes, Complex V and Complex X, that closely resemble those of myc−/− cells.

Figure 3
figure 3

Aberrant ETC complexes in Drp1 cells. (a) BNGE of purified mitochondria from Vector and Drp1 cells. Each of the major complexes and supercomplexes (SC) is indicated. (b) Quantification of ETC differences in Vector and Drp1 cells. Complex abundance was determined by scanning densitometry of Coomassie blue-stained blue-native gels similar to those depicted in panel a and arbitrarily expressed relative to Complex IV, which showed no change between the two cell lines. An aberrant complex (Complex ‘X’), previously seen in myc−/− cells4 was also observed at a significantly elevated level in Drp1 cells

To investigate the significance of the above quantitative differences in ETC components, we performed in situ enzymatic activity assays for complexes I and V using the same gels used for BNGE determinations.4 Both the NADH ubiquinone oxidoreductase activity of Complex I and its associated supercomplexes (Figures 4a and b) and the ATPase activity of Complex V (Figures 4c and d) were significantly reduced in Drp1 cells. In addition, the presence of a Complex V component, termed F1, which is believed to reflect Complex V dysfunction,32, 33 was increased in Drp1 cells, although this did not reach statistical significance (P=0.07; Figures 4 c and d). That the activities of Complex I and supercomplexes in Drp1 cells are reduced out of proportion to their actual levels as measured by BNGE suggests that these complexes are dysfunctional.

Figure 4
figure 4

Abnormal in situ ETC complex activities in Drp1 cells. (a) Complex I activity. In situ staining for NADPH (nicotinamide adenine dinucleotide phosphate reduced) ubiquinone oxidase was performed as previously described6 with gels similar to those shown in panel (a). A typical result is shown. (b) Quantification of the results shown in panel (a). The mean values of three independent in situ determinations are shown±1 S.E. Note the better resolution of the individual components of the supercomplexes (SC1 and SC2) than is seen in panel (a). (c) Complex V activity. In situ ATPase activity was determined as previously described6 using the same gels used to quantify Complex I activities. (d) Quantification of the results shown in panel (c). The mean values±1 S.E. of three independent in situ determinations are shown

Metabolic defects in Drp1 cells and their consequences

To determine whether the above-described defects also affected mitochondrial metabolism, we measured oxygen consumption rates (OCRs) in Vector and Drp1 cells as a way of assessing OXPHOS. We also quantified glycolysis in the same samples by measuring the extracellular acidification rate (ECAR). As seen in Figure 5a, the basal OCR of Vector cells, measured during the first 30 min of the experiment, was more than twice that of Drp1 cells (P<0.001), thus closely mirroring the differences in the total mitochondrial mass described in Figures 1b and c and Supplementary Video S1. The subsequent addition of various metabolic inhibitors4, 7 showed responses that were largely attenuated in Drp1 cells. For example, we noted that the level of ATP-linked respiration, that is, the difference in the basal rate of OCR minus that following the addition of oligomycin, was approximately twofold lower in Drp1 cells. Additionally, both the maximal respiratory rate and the spare respiratory capacity of Vector cells, defined as the total area under the curve (AUC) between the addition of cyanide p-trifluoromethoxy-phenylhydrazone (FCCP) and 2-deoxyglucose (2-DG) and the same AUC minus basal respiration, respectively,34, 35, 36 were also significantly attenuated in Drp1 cells. Concurrent measurements showed that basal levels of ECAR were also reduced in Drp1 cells by over twofold, thus mirroring our findings with OCR (Figure 5b). Collectively, these findings were consistent with the subsequent documentation of a state of severe ATP depletion in Drp1 cells (Figure 5c).

Figure 5
figure 5

Metabolic abnormalities in Drp1 cells. (a) Quantification of OXPHOS. Vector and Drp1 cells were analyzed in parallel using a Metabolic Flux Analyzer as previously described.4 Note the higher basal OCR in vector cells as measured between 0 and 30 min (P<0.001). The addition of specific metabolic inhibitors at various times is indicated. (b) Quantification of glycolysis. ECAR was measured on the same samples used to quantify OCR in panel (a). Basal rates were significantly higher in Vector cells (P<0.001). (c) Severe depletion of ATP in Drp1 cells. ATP levels were determined as previously described4 and are indicated as the mean of quadruplicate samples±1S.E. (d) Immunoblotting for total AMPK and phospho-AMPK (α-subunit Thr172) in Vector and Drp1 cells. β-Actin was included as a protein loading control

Reduced ATP levels are generally countered by the phosphorylation-dependent activation of the α-subunit of AMP-dependent protein kinase (AMPK), an energy-sensing enzyme that normalizes ATP levels by tempering anabolic reactions while coordinately increasing glucose uptake and fatty acid β-oxidation.37 As expected, pAMPK levels were markedly elevated in Drp1 cells, indicating that, while this pathway was intact and appropriately responsive, its activation was unable to correct the severe energy deficit in the face of ongoing mitochondrial fission (Figure 5d).

ETC dysfunction can lead to free electron loss, primarily via Complex I and Complex III, and a resultant increase in reactive oxygen species (ROS).30, 31, 38 This would be consistent with the observed Drp1 cell ETC deficits as well as the marked ATP-depleted state, which by itself can increase ROS.38 We therefore measured total cellular ROS using CM-H2DCFDA staining and found a nearly twofold increase in Drp1 cells (Figure 6a) that was confirmed using MitoSOX, a dye that measures superoxide, a specific product of OXPHOS39 (Figure 6b). Indeed, when considered relative to the twofold differences in mitochondrial mass (Figure 1b), the ROS production differential between Vector and Drp1 cells was nearly fourfold. Overall, the increase in ROS by Drp1 cells was comparable with that of myc−/− rat fibroblasts (Ref. 4 and Figure 6c). These findings indicate that Drp1 cell ETC defects are associated with reduced metabolic function, severe ATP depletion, promiscuous electron leak and ROS generation whose origins can largely be traced to the dysfunctional mitochondria.

Figure 6
figure 6

Abnormalities of ROS production by and reduced size of Drp1 cells. (a) Measurement of total cellular ROS. Vector and Drp1 cells were exposed to CM-H2DCFDA as previously described4, 7 and analyzed by flow cytometry. Nearly identical results were observed on three separate occasions (not shown). The ratios of the mean fluorescence are indicated above the panels. (b) Measurement of mitochondrial superoxide production using MitoSOX staining in Vector and Drp1 cells. (c) ROS production by myc+/+ (TGR1) fibroblasts and myc−/− rat fibroblasts. Cells were plated and stained with CM-H2DCFDA and MitoSOX as described in panels (a) and (b). (d) Cell size determinations. Vector and Drp1 cells were plated onto 100-mm tissue culture dishes, and grown for two days to allow the maximal proliferative rates to be achieved. The cells from subconfluent confluent plates were then harvested by trypsinization, stained with trypan blue and the diameters of all viable cells were determined on a Beckman-Coulter Vi Cell analyzer. At least five independent cell size determinations were performed at different times with a minimal of 3000 cells per determination. The mean cell diameter±1 S.E. is shown

The depletion of ATP in Drp1 cells, presumably caused by one or more respiratory chain defects and coupled with an intact but ineffectual pAMPK response, suggested that alternate compensatory mechanisms might be needed to minimize energy consumption while preserving survival and proliferation. One way by which this could be accomplished would be via reduced biomass.40 Indeed, similar to what we previously observed in myc−/− fibroblasts,4 the overall Volume of Drp1 cells was reduced >30% relative to Vector cells (Figure 6d).

Resistance of Drp1 cells to Myc-mediated apoptosis is ATP-dependent

Rat1a-MycER cells were originally selected for their robust pro-apoptotic response following MycER activation by 4-HT. Consistent with this, Vector cell proliferation was significantly impaired following 4-HT addition, even in the presence of high serum concentrations, due to an increasing apoptotic population (Figure 7a and data not shown). By contrast, Drp1 cells, which grew at nearly the same rate as Vector cells in the absence of 4-HT showed marked apoptotic resistance following Myc activation.

Figure 7
figure 7

Drp1 cells are resistant to apoptotic signals. (a) Growth curves. In all, 105 Vector and Drp1 cells were seeded onto 100-mm tissue culture plates in growth medium containing 10% fetal bovine serum and allowed to attach overnight. The following day (day 0), cells were counted and fresh growth medium either lacking or containing 4-HT was then added. Viable cell counts were then performed daily using a Vi-Cell analyzer. The results shown represent the mean of triplicate determinations±1 S.E. (b) Apoptosis profiles. Vector and Drp1 cells were seeded onto 100-mm plates in growth medium and allowed to achieve 80% confluency at which point (day 0) the total number of viable cells was determined from triplicate cultures. The remaining monolayers were then maintained for the duration of the experiment in serum-free medium either lacking or containing 4-HT, and the number of viable cells was determined daily thereafter. (c) Delayed cytochrome c release from Drp1 cells. Cells were initially plated and grown as described in panel (b) and then deprived of serum upon reaching 80% confluency. At the indicated time points, the cells were harvested and fractionated into cytosolic and mitochondrial fractions. Immunoblotting was then performed to allow for detection of cytochrome c release from the mitochondria. Loading controls consisted of β-actin for cytosolic fractions and Ponceau Red staining of the low molecular weight region of the membrane corresponding to the region of cytochrome c migration

We next monitored the survival of sub-confluent cultures of serum-deprived Vector and Drp1 cells in the absence or presence of 4-HT, thus allowing an assessment of both Myc-dependent and Myc-independent apoptosis. As expected,7, 41, 42 Vector cells rapidly succumbed following serum removal, particularly when combined with Myc activation (Figure 7b). By contrast, a greater fraction of Drp1 cells survived under both conditions. These results were supported by studies of cytochrome c release, the inciting event in the intrinsic apoptotic pathway.16, 43 Consistent with their higher sensitivity to Myc-induced apoptosis, mitochondrial cytochrome c release and its cytoplasmic accumulation occurred much more rapidly in Vector cells than in Drp1 cells. Taken together, these results indicate that Drp1 inhibits both Myc-dependent and Myc-independent apoptotic responses.

Because several key steps in both the intrinsic and extrinsic apoptotic pathways are ATP-dependent,44, 45 we next asked whether Drp1 cell survival could be attributable to their ATP-depleted state (Figure 5b). We therefore exposed Drp1 cells to 5-amino-1-β-D-ribofuranosyl-imidazole-4-carboxamide (AICAR), an AMP analog that increases ATP levels by activating AMPK and increasing OXPHOS.37, 46 As seen in Figure 8a, exposure of Drp1 cells to AICAR for as little as 4 h led to over twofold increases in ATP and pAMPK levels (Figure 8b). We then assessed Drp1 cell survival after activating Myc in the absence or presence of AICAR. As seen in Figures 8c, 4-HT alone only modestly reduced survival, thus confirming our previous findings (Figure 7a). Although AICAR alone reduced survival modestly, the combination of Myc activation and AICAR produced a much more pronounced reduction in survival. Moreover, continuous AICAR exposure normalized Drp1 cell size (Figure 8d). Thus, we conclude that AICAR increases ATP levels in Drp1 cells, sensitizes them to Myc-induced apoptosis and increases mean cell volume.

Figure 8
figure 8

AICAR restores ATP levels in Drp1 cells and accelerates Myc-dependent apoptosis. (a) Drp1 cells were seeded in 96-well plates as described for Figure 5b. The following day, fresh medium containing 250 μ M AICAR or vehicle alone was added and ATP levels were determined 4 h later. The histogram shows the mean relative ATP levels±1 S.E. from quadruplicate samples. (b) Parallel cultures of cells, maintained in 100-mm plates, were treated as described in panel (a) and were evaluated by immunoblotting for total and pAMPK levels. In Drp1 cells, the levels of pAMPK relative to total AMPK levels were 2.5-fold higher compared with Vector cells. (c) Drp1 cells were maintained under standard growth conditions (Figure 6a) for 3 days in the presence or absence of 4-HT and 250 μ M AICAR. The total number of viable cells was then determined. The results are expressed as the mean of triplicate sets of cultures±1 S.E. (d) Changes in the mean diameter of Drp1 cells following a 3-day exposure to AICAR. Cells were maintained and counted as described in Figure 6d. The results represent the mean diameters±1 S.E. of at least 3000 cells from triplicate cultures

Discussion

Mitochondria typically participate in fission, fusion and remodeling in response to stress and metabolic change.17, 18 The balance among these activities is maintained by the fission/fusion protein family, including Drp1.17, 18, 23, 24, 25 A role for Drp1 in apoptosis has also been proposed,47, 48, 49 and both its overexpression and inhibition can influence survival in complex ways that are dependent on the cell type, the inciting apoptotic trigger and possibly the nature of the Drp1 isoform that is perturbed.22, 23, 26, 37, 48, 49, 50 Fission is an early apoptotic event18, 48, 51 and is often coordinated with the activation of pro-apoptotic Bcl-2 family member proteins such as Bax, which leads to outer membrane permeabilization, cytochrome c release and the collapse of mitochondrial outer membrane potential.26, 47, 48 However, it has not been determined whether fission itself has a causal role in apoptosis or simply reflects the consequences of downstream processes that initiate cell death. Our results indicate that forced Drp1-mediated mitochondrial fission alone is insufficient to enhance apoptosis and indeed has precisely the opposite effect when cells are serum-deprived and/or when Myc is overexpressed.

It is likely that the common metabolic profiles of Drp1 and myc−/− cells, namely attenuated OXPHOS, ATP depletion and increased ROS generation, reflect the similarities in their ETC defects.4 This suggests that they are a generic response to impaired mitochondrial fusion, irrespective of its cause. Complex I defects are known to promote electron loss,31, 52 thus potentially explaining the higher ROS production by Drp1 cells described here (Figures 4c and d). Moreover, unlike myc−/− cells, where quantitative and qualitative reductions in Complex I occur in parallel,4 the functional defects of Complex I in Drp1 cells are disproportionate to the levels documented by BNGE. This suggests that Drp1 may alter Complex I activity without necessarily affecting its BNGE profile. Together with significantly lowered activity of complex V and reduced overall mitochondrial mass, such abnormalities likely account for the severely ATP-depleted states of Drp1 and myc−/− cells.

One property that does distinguish the mitochondria of Drp1 cells is their excessive rate of abortive fusion (Figure 2b and Supplementary Video S2). Given that Myc-replete Drp1 cells proliferate more rapidly than myc−/− cells,29 they may be under greater pressure to maximize ATP production even though basal levels in both the cells are comparable (Ref. 4 and Figure 5c). Because ectopic Drp1 limits increases in mitochondrial mass and function that would otherwise be needed under such circumstances, the high rates of abortive fusion may represent a compensatory, albeit ultimately futile, attempt to normalize OXPHOS and ATP. Another possibility is that Drp1 overexpression enhances fusion via non-energy-sensing pathways.19 That comparable rates of mitochondrial fusion do not occur in myc−/− cells may reflect the absence of Myc expression itself and its effects on mitochondrial biogenesis6, 24 and/or their greatly reduced proliferative rate and lower ATP requirements.4 Myc may also regulate fission and fusion proteins other than Drp1 that promote the high fusion rate of Drp1 cells.4 At the very least, our findings indicate that, despite their importance, mitochondrial size and integrity are not the only factors that coordinate fission and fusion processes.

Because Myc influences numerous functions,2, 53 it is not surprising that myc−/− cells possess a broader range of phenotypes than Drp1 cells. However, the latter’s behavior suggests that at least some of the myc−/− cell phenotypes are indirect and the result of a state that favors permanent mitochondrial fission and or atrophy. This interpretation may be subject to further refinement given that certain cell types in which fusion is compromised as a result of mitofusion 2 (mfn2) deletion show less impressive changes in mitochondrial ultrastructure than those described here and are actually somewhat larger than their mfn2+/+ counterparts.54

We do not currently understand the precise means by which Drp1 affects mitochondrial phenotype. Possible influences include the identity of the expressed isoform, its subcellular distribution and/or its post-translational regulation by phosphorylation, ubiquitinylation and/or sumoylation.55 Drp1 can affect cristae structure47 and may even have a role in mitochondrial fusion, all of which could be differentially affected by such post-translational modifications.19 These findings are also consistent with the fact that only modest changes in mitochondrial fission and fusion proteins, including Drp1, occur in response to Myc’s absence or overexpression.4

The apoptotic resistance of Drp1 cells in the presence and absence of serum (Figure 7b) indicates that both Myc-dependent and Myc-independent processes require normal mitochondrial function and/or mass. In each case, the common factor that imparts this behavior is likely to be ATP depletion. This is supported by the slower rate of apoptosis following serum deprivation of myc−/− cells,16, 42 whose ATP depletion also likely arises from ETC dysfunction.4 Mutations affecting these complexes are often associated with elevated ROS and ATP depletion and are commonly seen in inherited metabolic diseases and cancer.30, 31, 33, 56 Mutations in specific Complex V subunits such as MTATP6, or general reductions in Complex V, have also been described in cancer, often in association with ATP depletion and apoptotic resistance.57, 58

Previous studies suggest that the initial stages of apoptosis and downstream components of the apoptotic machinery are ATP-dependent, including the activation of caspases 3 and 8, Apaf-1 and apoptosis-associated endonuclease.44, 45 As little as a 50% reduction in ATP levels can significantly delay apoptosis,45 which is consistent with our finding that the apoptotic resistance of Drp1 cells can be reversed by activating AMPK and more than doubling ATP levels (Figure 8). AMPK activation has been reported to impart apoptotic resistance to a variety of cell types, although it is unclear as to whether this is due solely to the normalization of ATP levels.59

That Drp1 cells proliferate and survive offers strong evidence that small mitochondria per se are not themselves a direct trigger of apoptosis but rather require additional factors.26 This is consistent with fission’s central role during normal mitochondrial remodeling in proliferating cells where apoptosis would be an undesirable outcome. Fission is also required for the consumption of mitochondria during periods of starvation-induced autophagy, the purpose of which is to preserve cell viability by providing alternate energy sources.20 Because apoptotic signaling is rapid and irreversible beyond a certain point,16 it is not surprising that it is subject to multiple layers of regulation and that fission would not necessarily trigger an apoptotic response and might, in some cases, even block it.

In summary, our results demonstrate how defects in mitochondrial fission, ETC architecture and ATP production can conspire to promote survival in response to Myc deregulation.31, 52 Although Myc overexpression typically promotes mitochondrial biogenesis and ATP production,4, 15 it does so when essential nutrients and oxygen are abundant. When these become limiting due to poor vascular perfusion and loss of mitochondrial integrity, the attendant depletion of ATP may provide a buffer against the pro-apoptotic tendencies of Myc overexpression, thus favoring tumor cell survival.

Materials and Methods

Generation and maintenance of cell lines

Rat1a fibroblasts were routinely propagated in Dulbecco’s modified Eagle’s minimal essential medium (DMEM) supplemented with 10% fetal bovine serum, glutamine and penicillin/streptomycin as previously described.7, 14 Conditional Myc expression was achieved by stable transduction with the pBABE-puro-MycER retroviral vector (gift of Martin Eilers). Stable expression of the 710 amino-acid long variant 2 isoform of Drp1 (GenBank Accession No: NM_012063, Origene Technologies, Inc. Rockville, MD, USA) as a V5 epitope-tagged fusion protein was obtained in the pLenti6/V5 TOPO vector (Invitrogen, Inc., Carlsbad, CA). The cDNA was amplified using the forward PCR primer 5′-CACCATGGAGGCGCTAATTCCTGTC-3′ and the reverse primer 5′-CCAAAGATGAGTCTCCCGGATTTC-3′, where the underlined bases represent the initiation ATG and the penultimate codons, respectively. Following directional cloning, the orientation of the insert and its in-frame fusion to the V5 epitope coding region were confirmed by automated DNA sequencing. The Drp1-encoding vector, or an empty control vector, were then packaged according to the directions of the supplier and the resultant lentiviral stocks were used to transduce the above-described Rat1a-MycER cell line. Stable clones were selected in 2–5 μg/ml blasticidin, pooled for subsequent characterization and routinely maintained in blasticidin-containing medium. These cell lines are hereafter referred to as Drp1 cells and Vector cells. All retroviral and lentiviral work was approved by the University of Pittsburgh Institutional Biosafety Committee and was performed under BSL2+ conditions.

For confocal microscopic experiments, additional transfections were performed with the pEGFP-Mito and pDsRed2-Mito vectors (Clontech, Inc., Mountain View, CA, USA), which encode green and red fluorescent proteins, respectively, fused to the mitochondrial targeting sequence from subunit VIII of human cytochrome c oxidase. Cells were transfected with 1-2 μg of plasmid DNA using Superfect transfection reagent (Qiagen, Inc., Valencia, CA, USA), selected in 200 μg/ml of G-418 (Sigma-Aldrich, St. Louis, MO, USA) and isolated by fluorescence-activated cell sorting.

Confocal microscopy

Mitochondria were visualized in cells expressing pEGFP-Mito or pDsRed2-Mito vectors. Cells were grown to 50% confluency in six-well, glass-bottom tissue culture plates. Fluorescent images of magnification × 20 were obtained with a Zeiss Axiovert fluorescent microscope (Wake Forest, NC, USA). Images of magnification × 60 were obtained on a Zeiss LSM710 confocal microscope (Wake Forest, NC, USA). Confocal stacks of the mitochondria were analyzed using Imaris software (Bitplane Scientific, South Windsor, CT, USA) and cellular 3D reconstructions were obtained as previously described.4 Mean fluorescence intensities were obtained on at least 50 individual cells and quantified using ImageJ software (http://rsbweb.nih.gov/ij/). Live-cell confocal microscopy was performed on cells stained with MitoTracker Green FM at a concentration of 1 μ M for 30 min before the start of experiment. Cells were counterstained with Hoechst 33342 at a concentration of 2 μg/ml. During the imaging, cells were maintained in phenol red-free DMEM in which the MitoTracker and Hoechst dyes were added at 1/10 the original concentration. Images were captured on the LSM710 microscope approximately every 45 s.

PEG-fusion assays

These were performed as described previously.7 Briefly Vector and Drp1 cells expressing EGFP-Mito were co-plated with equal numbers of the same cells expressing DsRed2-Mito onto glass coverslips in 12-well plates. Following an overnight incubation, fresh medium containing cycloheximide (CHX; 30 μg/ml, Sigma-Aldrich) was added for 30 min to inhibit de novo synthesis of fluorescent proteins. The medium was then removed and 0.5 ml of 100% PEG 1500 (Sigma-Aldrich) was added for 1 min at room temperature. The coverslips were washed three times with DMEM+CHX, and subsequently incubated in the same medium for 2 h. The coverslips were then fixed for 30 min with ice-cold 3.7% para-formaldehyde in PBS, stained with DAPI and fixed onto glass microscope slides. A minimum of 20 heterokaryons were observed for each sample using confocal fluorescence microscopy as described above from which the percentage of co-localized (fused) mitochondria was calculated using the CoLocalizer Pro software (CoLocaliztion Research Software, Kochi, Japan) and expressed as the percentage of mitochondrial fusion.

Determination of mitochondrial mass and ROS

Mitochondrial mass was determined using MitoTracker Green FM and NAO. ROS levels were measured using CM-H2DCFDA and MitoSOX. All dyes were purchased from Life Technologies, Inc. (Carlsbad, CA, USA). Briefly, cells were seeded 1–2 days before staining and allowed to attain 50% confluency unless otherwise stated. Adherent cells were exposed for 45 min to 500 nM MitoTracker Green FM, 2.5 μ M of CM-H2DCFDA or 20 nM NAO in standard growth medium. In the case of MitoSOX, cells were exposed for 10 min to a 5-μ M concentration of the dye. Immediately after exposure, adherent cells were trypsinized, titurated vigorously and then subjected to flow cytometric evaluation using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA, USA) as previously described.7, 14

Measurements of OXPHOS and glycolysis

OXPHOS was quantified using a Seahorse Biosciences (North Billerica, MA, USA) XF24 Extracellular Flux Analyzer as previously described.4, 7 Briefly, 2 × 104 cells were seeded into individual wells of a 24-well plates and cultured overnight before assaying. Fresh medium was added and baseline O2 consumption was quantified as a surrogate measurement of OXPHOS. Several metabolic inhibitors were then added to determine the bioenergetic capacity of the mitochondria. These included: oligomycin, which blocks complex V and causes a build-up of protons across the inner mitochondrial matrix; FCCP, which causes protons exterior to the inner membrane to be carried across to the mitochondrial matrix; 2-DG), which inhibits glucose uptake for the initial step of glycolysis and reduces pyruvate transport into the TCA cycle; and rotenone, which inhibits Complex I and blocks both electron flow and oxygen consumption. Each point was determined from quadruplicate wells, and each experiment was performed at least twice. Glycolysis was quantified in parallel on the same samples by measuring the ECAR as previously described.4, 7

ATP assay

ATP levels were measured using the ATPlite Luminescence Assay System (Perkin Elmer, Waltham, MA, USA) as previously described.7 Briefly, 2 × 104 cells were seeded into individual wells of a 96-well plate and incubated overnight before assay, which was conducted on quadruplicate samples according to the directions of the supplier. The final calculated ATP levels were adjusted for slight variations in cell plating efficiencies (<20%) based on cell counts performed on a parallel series of wells.

Western blotting

Western blotting of total cell lysates was performed as previously described.7, 14 The following primary antibodies were used: Anti-V5: 1:5000 (Invitrogen); anti-Sirt3: 1:1000 (Abcam, Cambridge, MA, USA), anti-Sirt5: 1:1000 (Abcam); anti-MFF: 1:1000 (Abcam); anti-MCAD: 1:2500; anti-cytochrome c: 1:1000 (Abcam); Anti-AMPK 1:2000 (Cell Signaling Technology, Beverley, MA, USA); Anti-phospho-AMPK (α-subunit Thr172) 1:1000 (Cell Signaling Technology); anti β-actin: 1:10,000 (Santa Cruz, Santa Cruz, CA, USA). Horseradish peroxidase-tagged secondary antibodies were used at dilutions recommended by the supplier and included goat anti-mouse IgG:1 : 10 000 (Santa Cruz) and goat anti-rabbit IgG:1 : 5000 (Cell Signaling Technology). Blots were developed using the Pierce ECL Plus Western Blotting Substrate (Thermo Fisher Scientific, Inc., Waltham, MA, USA) as recommended by the supplier.

Cell sizing and apoptosis determinations

Monolayer cultures were trypsinized, washed twice in PBS and re-suspended in 1 ml of PBS. The average diameter of viable cells was then immediately determined with a Beckman-Coulter Cell Viability Analyzer (Beckman-Coulter, Inc., Indianapolis, IN, USA). Only cultures having a starting overall viability of >90% were analyzed. Diameters of at least 3000 viable cells/sample/day were determined on 3–5 individual samples measured on separate days. Apoptosis assays were also performed on the Cell Viability Analyzer by quantifying the fraction of trypan blue-positive cells in the combined floating and attached populations with each assay being performed in triplicate. These assays were in excellent agreement with parallel studies of Annexin V cell surface expression (not shown).4, 7

Mitochondrial purification

Vector and Drp1 cells were harvested by trypsinization, combined with the non-adherent cell population and washed three times with PBS. The final pellets were re-suspended in 1 ml sucrose buffer (0.25 M sucrose; 10 mM Tris-HCl, pH 7.4; 1 mM EDTA and protease inhibitors) and disrupted using a cell homogenizer (Isobiotech, Heidelberg, Germany). The homogenates were then centrifuged at 100 × g for 10 min, and the supernatants were re-centrifuged at 12 000 × g for 15 min. The supernatants containing cytoplasmic proteins were removed and saved. The mitochondria-enriched pellets were washed in the above sucrose buffer and saved. Protein concentrations in each fraction were then determined before SDS-PAGE and immunoblotting for cytochrome c as described above.

BNGE and in situ activity staining

Native ETC complexes were visualized by non-denaturing BNGE as described previously.4 Briefly, isolated mitochondria were lysed by the addition of digitonin to a final ratio of 1 : 8 protein:digitonin and incubated on ice for 20 min. Coomassie blue solution (5% Coomassie blue G250 in 750 mM 6-aminocaproic acid) was added (1/20 v/v) followed by centrifugation at 14 000 × g for 20 min at 4 °C. The supernatant was then electrophoresed on a 3–12% Native PAGE Novex Bis-Tris gel (Invitrogen) in the buffer provided by the supplier, stained with Bio-Safe Coomassie G250 (Bio-Rad, Hercules, CA, USA) and de-stained with water. The gels were scanned and analyzed for band density using an AlphaEaseFC 2200 scanner and software (Alpha Innotech, Inc., San Leandro, CA, USA).

For Complex I activity measurements, the same blue-native gels used to quantify individual complexes were incubated at 37 °C for 1–2 h in 2 mM Tris-HCl, pH 7.4 buffer containing 2.5 mg/ml nitrotertrazolinum blue chloride and 0.1 mg/ml NADH. Densitometric analysis from triplicate gels was used to calculate the mean activity. Quantification of the ATPase activity of Complex V was performed by incubating gels at 37 °C for 3 h in 34 mM Tris-glycine, pH 7.8, 14 mM MgSO4, 0.2% Pb(NO3)2 with freshly added 8 mM ATP. Bands were quantified as described for Complex I.