MISC-1/OGC Links Mitochondrial Metabolism, Apoptosis and Insulin Secretion

We identified MISC-1 (Mitochondrial Solute Carrier) as the C. elegans orthologue of mammalian OGC (2-oxoglutarate carrier). OGC was originally identified for its ability to transfer α-ketoglutarate across the inner mitochondrial membrane. However, we found that MISC-1 and OGC are not solely involved in metabolic control. Our data show that these orthologous proteins participate in phylogenetically conserved cellular processes, like control of mitochondrial morphology and induction of apoptosis. We show that MISC-1/OGC is required for proper mitochondrial fusion and fission events in both C. elegans and human cells. Transmission electron microscopy reveals that loss of MISC-1 results in a decreased number of mitochondrial cristae, which have a blebbed appearance. Furthermore, our pull-down experiments show that MISC-1 and OGC interact with the anti-apoptotic proteins CED-9 and Bcl-xL, respectively, and with the pro-apoptotic protein ANT. Knock-down of misc-1 in C. elegans and OGC in mouse cells induces apoptosis through the caspase cascade. Genetic analysis suggests that MISC-1 controls apoptosis through the physiological pathway mediated by the LIN-35/Rb-like protein. We provide genetic and molecular evidence that absence of MISC-1 increases insulin secretion and enhances germline stem cell proliferation in C. elegans. Our study suggests that the mitochondrial metabolic protein MISC-1/OGC integrates metabolic, apoptotic and insulin secretion functions. We propose a novel mechanism by which mitochondria integrate metabolic and cell survival signals. Our data suggest that MISC-1/OGC functions by sensing the metabolic status of mitochondria and directly activate the apoptotic program when required. Our results suggest that controlling MISC-1/OGC function allows regulation of mitochondrial morphology and cell survival decisions by the metabolic needs of the cell.


Introduction
Mitochondria are the sites of enzymatic reactions fundamental for life, producing ,95% of cellular ATP in eukaryotic cells. Mitochondria also play a dominant role in the control of apoptosis (reviewed in [1]). Several mitochondrial proteins are required for the intrinsic pathway of apoptosis. The succession of events from mitochondrial fragmentation to cell death has been well characterized and shown to be highly conserved from C. elegans to mammals [2,3].
Mitochondria undergo constant rounds of fusion and fission to form or break their tubular structure within the cell. In addition to mitochondrial fragmentation during apoptosis [4], extensive mitochondrial remodelling happens in response to metabolic changes (reviewed in [5]). The exact mechanisms that control the interdependence of metabolic rate and mitochondrial structure are currently unknown. We investigated the possibility that mitochondrial metabolic proteins interact directly with members of the apoptotic machinery to control mitochondrial morphology and cell survival decisions. We focused on the mitochondrial 2oxoglutarate carrier (OGC) to test our hypotheses.
The nuclear gene OGC encodes a monomeric carrier protein that resides in the inner mitochondrial membrane and is responsible for the electroneutral transport of a-ketoglutarate [6,7]. This gene has been studied mostly for its effects on cellular metabolism. However, it has recently been shown that mammalian OGC imports about 40% of mitochondrial glutathione (GSH), a potent anti-oxidant peptide involved in oxidative stress response in mammals and C. elegans. Also, its over-expression in NRK-52E cells prevents chemically induced apoptosis [8][9][10]. These findings suggest that OGC might be involved in stress response as well as control of mitochondrial metabolism. In this report, we charac-terize the C. elegans homologue of OGC, which we called misc-1 (Mitochondrial Solute Carrier), and find a novel role for MISC-1/ OGC in apoptosis.
Mitochondria-induced apoptosis can result from the opening of channels called Mitochondrial Permeability Transition Pores (MPTPs), which are located at the junctions of inner and outer mitochondrial membranes. MPTP contributes to Mitochondrial Outer Membrane Permeabilization (MOMP) during apoptosis and allows exit of cytochrome c [18,19] and other mitochondrial pro-apoptotic factors to the cytoplasm. MPTPs are multi-protein complexes, composed of the Voltage-Dependent Anion Channel (VDAC), the Adenine Nucleotide Translocase (ANT) and Cyclophilin D (Cyp-D; [20][21][22]). Multiple regulators have been identified, including the anti-apoptotic Bcl-2-family proteins Bcl-2 and Bcl-x L [23,18]. The identity of all MPTP components and regulators is not yet known. ANT is an established component of the MPTP in both mammals [24] and C. elegans [25]. We tested the possible interaction of OGC and MISC-1 with components of the MPTP.
Here we show that OGC and its C. elegans orthologue MISC-1 maintain the balance of fusion and fission events in human cells and in C. elegans. Genetically, the absence of MISC-1/OGC results in increased apoptosis rates through the physiological pathway. Furthermore, we provide genetic and molecular evidence of the involvement of MISC-1/OGC in control of insulin secretion and germline stem cell proliferation. Our work points to a novel, conserved and direct interaction between mitochondrial metabolic carriers and the apoptotic machinery to determine cell survival outcomes.

misc-1 expression
We investigated the expression profile of the 2-oxoglutarate carrier in C. elegans. Based on sequence similarity between MISC-1 and human OGC (72% amino acid identity; Supplementary Figure S1), their localization to mitochondria, and the shared roles of the two proteins (see below), we conclude that MISC-1 is the C. elegans orthologue of OGC.
We generated an extrachromosomal transgenic line carrying the full-length misc-1 gene fused to gfp and under control of ,1100 bp of the region 59 of the ATG codon. Although low levels of reporter expression were observed ubiquitously (as expected), a subset of organs showed higher expression of MISC-1::GFP. GFP expression began early in embryogenesis in the intestinal precursor cells (Fig. 1A-B). The protein exhibited intestinal and neuronal localization in newly hatched L1s (Fig. 1C-D), while it was mainly present in the anterior pharynx in adults (Fig. 1E-F). A similar expression pattern was observed in a separate extrachromosomal transgenic line carrying a misc-1 promoter GFP fusion construct (data not shown). Overall, this expression pattern suggests that MISC-1 is required in tissues with high energy demands, i.e. the intestine in developing animals and the pharynx in adults. This is consistent with the increased expression of our MISC-1::GFP reporter in calorierestricted animals ( Figure S2). It is interesting to note that in humans, OGC is most highly expressed in the heart (UCSC Genome Browser, [26]), another organ with high energy demands.
Although MISC-1/OGC is expected to be a mitochondrial carrier, we could find no published in vivo localization data for any organism. C. elegans embryonic and larval cells exhibited a typical mitochondrial pattern [27] for the localization of MISC-1::GFP ( Fig. 1A-D). Co-localization of MISC-1::GFP and the mitochondria-specific marker MitoTracker Red CMXRos in muscle and seam cells indicates that MISC-1 is a mitochondrial protein ( Fig. 1G-H).

MISC-1 and OGC regulate mitochondrial morphology
Under normal cellular conditions, mitochondria are mostly found fused in branched structures. To assess the effects of misc-1 knock-down on mitochondrial morphology, we used a transgenic line with GFP targeted to mitochondria and under control of the muscle-specific myo-3 promoter (myo-3p::mito::gfp; [28]). misc-1 RNAi resulted in robust (84%) down-regulation of MISC-1 protein levels ( Figure S3) and was therefore deemed an appropriate system for the study of misc-1 function. misc-1 RNAi treatment resulted in mitochondrial fragmentation in the mito::gfp strain ( Fig. 2A,B). In order to understand if this phenotype was specific to the mitochondria and not the result of misc-1-mediated defects in muscle fibres, we also tested the effect of misc-1 RNAi on a strain carrying myo-3p::myo-3::gfp [29], a marker for the structure and organization of body wall muscle. Our results indicate that misc-1 RNAi does not affect muscle structure ( Figure S4; see also Fig. 3C-D). MISC-1 seems therefore to specifically control mitochondrial morphology.
To determine whether the effect of the 2-oxoglutarate carrier on mitochondrial morphology is conserved in mammals, we knocked down the misc-1 human orthologue, OGC, in Human Embryonic Kidney 293 (HEK293) cells (Fig. S3). Staining with MitoTracker showed that siRNA against OGC resulted in a mitochondrial fragmentation phenotype reminiscent of misc-1 knock-down worms (Fig. 2C,D). MISC-1 and OGC have the same effect on mitochondrial morphology in C. elegans and human cells, respectively.
Animals homozygous for the misc-1(tm2793) knock-out allele are morphologically normal, have the same developmental time as wild type (measured as time from egg laying to day one of adulthood), and have normal fat staining and lifespan (data not shown). However, when analyzed by transmission electron microscopy, we observed morphological defects in the mitochondria of misc-1 animals. On average, the mitochondria of knockout animals were reduced in size and more spherical in shape compared to N2 controls (Fig. 3A,B). Moreover, they displayed a reduced number of disorganized and abnormal mitochondrial cristae (Fig. 3C,D). These features suggest that absence of MISC-1 shifts the equilibrium of mitochondrial dynamics in the direction of fission and are in agreement with the mitochondrial fragmentation phenotype we observed with the mito::gfp strain. Furthermore, the absence of a key mitochondrial inner membrane protein disrupts normal formation or maintenance of cristae.

Lack of MISC-1 results in increased germline apoptosis
Mitochondrial fragmentation is a hallmark of apoptosis in mammals and in C. elegans [4]. We used SYTO12 to stain apoptotic bodies in the adult germline in N2 and in the misc-1 knock-out. We observed a two-fold increase in the number of apoptotic corpses per gonadal arm in misc-1 knock-outs ( Fig. 4A,B).
Since an increase in the number of apoptotic corpses could be caused either by increased apoptosis or by a defect in corpse removal, we used a transgenic line carrying a CED-1::GFP reporter [30] to monitor corpse engulfment. CED-1 is a receptor localized to the membrane of the sheath cells in the somatic gonad and is responsible for the recognition of cell corpses and the initiation of their engulfment and removal. We observed a two-fold increase in the number of corpses engulfed by CED-1::GFP(+) cells in animals treated with misc-1 RNAi, compared to adults treated with control RNAi ( Figure S5). These results indicate that misc-1 RNAi does not impair the mechanism of cell corpse removal. We conclude that absence or decreased levels of MISC-1 increase the frequency of germline apoptosis.
In order to assess if reduced levels of OGC cause apoptosis in mammalian cells, we knocked-down mouse OGC in the MIN6 mouse insulinoma cell line [31] with an shRNA strategy. OGC knock-down caused robust cleavage and activation of Caspase-3, an established marker of activation of apoptosis (Fig. 4C). We conclude that lower levels of the 2-oxoglutarate carrier induce a phylogenetically conserved apoptosis program that is conserved from nematodes to mammals.

misc-1 acts through the physiological pathway of apoptosis
The increased levels of apoptosis observed upon knocking down misc-1 were controlled at the genetic level, since a ced-3(lf)/Caspase-9 genetic background obliterated apoptosis caused by misc-1 RNAi (Fig. 4D). In C. elegans, apoptosis can be triggered by three main pathways: (a) the DNA damage pathway, the main mediator being the C. elegans p53 orthologue CEP-1 [32]; (b) the cytoplasmic stress pathway, mediated by LET-60/K-Ras [33], and (c) the physiological pathway, mediated by the Retinoblastoma (Rb)-like protein LIN-35 [34]. All three pathways ultimately converge on the CED-9/CED-4/CED-3 core apoptotic machinery.
misc-1 RNAi administered to wild-type N2 (Fig. 4D) recapitulated the germline apoptosis phenotype observed in misc-1 knock-out worms (see Fig. 4B). This result, and the assessed efficiency of misc-1 RNAi in down-regulating MISC-1 protein levels (Fig. S3), further confirms that RNAi is an appropriate tool to study misc-1 function. We treated mutants in each of the three major apoptosis pathways with misc-1 RNAi and assessed apoptosis levels by SYTO12 staining (Fig. 4D). The cep-1(lf)/p53 background did not affect the ability of misc-1 RNAi to increase the number of apoptotic corpses. Using the let-60(n1046)/ K-Ras gain of function allele, which increases the basal rate of apoptosis through the cytoplasmic stress pathway, misc-1 RNAi increased the apoptotic rate in an additive fashion (not synergistic), indicating that misc-1 does not function through this pathway. By contrast, a lin-35(lf) mutant abrogated the apoptosis phenotype induced by misc-1 RNAi. These results suggest that misc-1 acts through the LIN-35/Rb-mediated physiological pathway of apoptosis.

MISC-1 and OGC interact with proteins involved in the activation of apoptosis
Since we found that misc-1 modulates mitochondrial fragmentation, cristae morphology and apoptosis, which are features also   Cell lysates from the three treatments were used for a Western blot employing an antibody that recognizes both total Caspase-3 and cleaved Caspase-3. The latter is the active, proapoptotic form of the caspase and is therefore a marker of apoptosis. The Western blot shows that knock-down of OGC results in Caspase-3 activation and therefore apoptosis in MIN6 cells. Similar results were obtained with MIN6 cells at two different passages. (D) misc-1 mediates apoptosis through the physiological pathway of apoptosis. Synchronized animals with different genetic backgrounds were fed either control or misc-1 RNAi from the L1 stage and were stained with SYTO12 on day 1 of adulthood. SYTO12-positive apoptotic bodies per gonad arm were scored. ced-3 mutants were used as a control, since mutations in this gene encoding an effector caspase obliterates germline apoptosis. cep-1/p53 mutants suppress the DNA-damage pathway of apoptosis. lin-35 mutants lack the physiological pathway of apoptosis. Worms carrying a let-60/K-Ras gain-of-function allele have increased apoptotic rates through the cytoplasmic stress pathway. The data indicate that mutations in lin-35 suppress the apoptotic phenotype of misc-1 RNAi, indicating that misc-1 controls apoptosis through the physiological pathway of apoptosis. *t-test P,0.001. { t-test P,0.05. doi:10.1371/journal.pone.0017827.g004 regulated by the MPTP, we asked if MISC-1 might interact with components or regulators of the MPTP, specifically Bcl-x L and ANT. We performed a co-immunoprecipitation experiment using an antibody against ANT in HEK293 cells. ANT pulled down Bclx L as expected, and it also precipitated OGC (Fig. 5). Likewise, Bcl-x L co-precipitated ANT and OGC (data not shown). As a control, we tested a subunit of complex IV of the electron transport chain, which is also a mitochondrial inner membrane protein. This protein was not detected in the co-immunoprecipitated fraction; it was only present in the input/supernatant.
We observed similar interactions for MISC-1 in C. elegans. We did pull-down experiments in wild type and in our MISC-1::GFP over-expressing transgenic line with several combinations of antibodies and were able to confirm the interaction of MISC-1, CED-9/Bcl-2 and ANT ( Figure S6). These pull-down experiments indicate that MISC-1 and OGC interact with proteins involved in the control of apoptosis.
misc-1 mutations suppress the dauer constitutive phenotype of daf-2 MISC-1 is a metabolic protein, controls mitochondrial function and interacts with CED-9, whose orthologue is Bcl-x L . Interestingly, transgenic mice with strong over-expression of Bcl-x L in their pancreatic beta-cells were previously shown to have altered glucose-sensing at the mitochondria and impaired glucosestimulated insulin release [35]. This observation suggests the possibility that Bcl-x L , like the related pro-apoptotic protein Bad [36], may play a role in insulin release independent of its role in apoptosis. We took advantage of the genetic tools afforded by C. elegans to investigate a possible role for misc-1 in insulin signalling in vivo. The Insulin/Insulin-like Signalling (IIS) pathway is well conserved between C. elegans and mammals (reviewed in [37]). In C. elegans, the Insulin-like Growth Factor 1 Receptor (IGF1R) is encoded by daf-2 (DAuer Formation) [38][39][40]. Mutations in this gene and reduced IIS result in dauer formation, which is a larval diapause state [41][42][43].
To test the possibility that MISC-1 might play a role in insulin secretion in C. elegans, we first tested a possible genetic interaction between misc-1 and daf-2/IGF1R. We created a double mutant carrying the hypomorphic allele daf-2(m41) and the knock-out allele misc-1(tm2793) to ask whether misc-1 modified the daf-2 dauer-constitutive phenotype (Fig. 6A). m41 animals constitutively enter the dauer diapause stage at non-permissive temperatures (.22uC). All daf-2 animals entered the dauer stage at both 22.5uC and 25uC, but no daf-2;misc-1 animals entered the dauer stage at 22.5uC and only 22% formed dauer larvae at 25uC. Therefore misc-1 substantially suppresses the dauer phenotype of daf-2. The suppressive effects of the mutation are stronger at lower temperatures, probably because m41 is a temperature-sensitive allele with a stronger phenotype at higher temperatures.

MISC-1 modulates DAF-28 secretion
The suppression of the daf-2 dauer-constitutive phenotype by misc-1(tm2793) could be interpreted in at least two ways: (a) misc-1 could be genetically downstream of daf-2 or part of a pathway downstream of the insulin-like pathway; (b) mutations in misc-1 may positively affect insulin secretion, which in turn would compensate for the partial loss of DAF-2 function due to the hypomorphic mutation m41. To address this issue, we employed a strain carrying a translational fusion of the C. elegans insulin DAF-28 and GFP (DAF-28::GFP; [44]). We subjected the DAF-28::GFP strain to misc-1 RNAi by feeding. DAF-28::GFP is secreted by the head neurons and the posterior intestine into the pseudo-coelomic fluid, where it is picked up by scavenging cells called coelomocytes. Among a total of six coelomocytes, the number containing DAF-28::GFP (GFP(+)) is proportional to the amount of secreted insulin [44]. When we compared the number of GFP(+) coelomocytes in worms treated with control and misc-1 RNAi, we observed that knock-down of misc-1 induced an increase in GFP(+) coelomocytes (average of two independent experiments; Fig. 6B). In control RNAi animals, we observed no more than two GFP(+) coelomocytes, whereas we observed three GFP(+) coelomocytes in 21% of misc-1 RNAitreated adults, four GFP(+) coelomocytes in 4.6% of the adultss and five GFP(+) coelomocytes in 2.3% of the adults observed. The results show a statistically significant increase in DAF-28::GFP secretion in worms treated with misc-1 RNAi (P,0.05 by two-way ANOVA).
The DAF-28 protein acts as a DAF-2 receptor agonist, but the C. elegans genome encodes over 40 insulin-like proteins, some being agonists and some antagonists of DAF-2 [44][45][46]. To corroborate the DAF-28::GFP reporter results as a general quantitative marker of insulin secretion, we measured the amount of glucose in misc-1 and wild type N2 mixed-stage populations. We reasoned that if indeed insulin secretion is increased in misc-1 mutants, then their glucose utilization should increase and the total glucose level should drop, compared to normal animals. In fact, we observed a 6% decrease (t-test P = 0.0054) in glucose levels in the misc-1 mutant (Fig. 6C), a result consistent with an increase in insulin secretion when MISC-1 levels are reduced. Organismal levels of ATP are not affected by absence of MISC-1 (Fig. 6D), indicating that ATP may not be responsible for the induction of insulin secretion in the case of misc-1 insufficiency.

misc-1 activity negatively modulates germline proliferation
The C. elegans germline is a syncytium of nuclei that mature as they migrate from a distal to a proximal position and eventually become oocytes. Notch signaling initiated from the Distal Tip Cell (DTC) regulates the balance between proliferation (mitosis) and differentiation (meiosis) of these germline nuclei. The DTC produces the Notch ligand LAG-2 [47][48][49], which activates the GLP-1/Notch receptor in nearby germ cells and maintains these cells in the undifferentiated (mitotic) state [50,51]. A recent report showed that increased insulin signalling can increase germ cell Figure 5. OGC interacts with Bcl-x L and ANT. We performed a coimmunoprecipitation experiment in HEK293 cells using an antibody against ANT, which is an integral component of the MPTP. We were able to detect Bcl-X L , OGC and ANT in the co-immuno precipitant. As a control protein, we chose subunit 1 of complex IV of the electron transport chain, which is also localized to the inner mitochondrial membrane. This control protein did not co-immunoprecipitate with Bclx L , OGC and ANT, but was present in the supernatant (input). We obtained similar results by immunoprecipitation with an antibody against Bcl-x L (data not shown). Co-IP: co-immunoprecipitation. WB: western blot. doi:10.1371/journal.pone.0017827.g005 proliferation independently of LAG-2 activation [52]. A similar effect of insulin signalling on female reproduction was observed in mouse [53].
We investigated the possibility that the increased insulin secretion in misc-1 mutants could have an effect on germline proliferation. Proliferative germ cells in late prophase and early mitotic M-phase are positive for histone 3 (H3) phosphorylation (pH3; [54,55]). We performed immunostaining of dissected germlines from wild type and misc-1 knock-out mutants with a pH3 antibody. The results show that absence of MISC-1 greatly increases the number of pH3-positive, mitotic germline stem cells (Fig. 7A,B) and causes an extended mitotic zone (data not shown). pH3 immunostaining strongly suggests that absence of MISC-1 favours mitosis (self-renewal) over meiosis (differentiation) in germline stem cells.
To assess the involvement of LAG-2, we employed a transgenic line expressing a GFP reporter under the control of the lag-2 promoter (lag-2p::gfp; [56]). This reporter is expressed in the DTCs and its processes, called cytonemes (Fig. 7C,D). We found that the cytonemes of animals treated with misc-1 RNAi were longer than the ones in control animals (Fig. 7C-E). Knock-down of misc-1 resulted in a two-fold increase in the length of the cytonemes (mean 6 SEM: control RNAi = 74.9610.80 mm, misc-1 RNAi = 150.9629.14 mm; t-test P = 0.0080). We assessed lag-2p::gfp expression by microscopy in animals treated with control or misc-1 RNAi, but were unable to detect statistically significant differences between the two treatments (Fig. 7F). We concluded that upregulation of lag-2 was not responsible for the extended cytonemes and mitotic zones of misc-1 mutants. Although the length of the cytonemes may be positively correlated with the number of germline cells in a mitotic state, this is independent of lag-2 expression. The germline stem cell proliferation phenotype obtained by lack of MISC-1 is therefore consistent with an increase in insulin signalling. We assessed genetic interactions between misc-1 and the hypomorphic allele daf-2(m41). Mutations in daf-2 result in constitutive dauer formation at 22.5uC and 25uC. However, the double mutant daf-2;misc-1 worms did not arrest at the dauer stage at 22.5uC, and only 22.5% of the double mutants formed dauer larvae at 25uC. These data show that absence of misc-1 expression suppresses the dauer phenotype conferred by daf-2 mutations. T-test P = 0.0021. (B) We tested the possibility that the daf-2 suppression by misc-1 was due to an increase in insulin secretion, which could over-ride the hypomorphic allele m41. We measured insulin secretion by scoring GFP(+) (i.e. DAF-28::GFP-positive) coelomocytes in a transgenic strain upon misc-1 knock-down. misc-1 RNAi animals (N = 44) had on average more GFP(+) coelomocytes than control worms (N = 33) (two-way ANOVA P = 0.04). These data suggest that misc-1 knock-down increases insulin secretion. (C) We quantified glucose concentration in N2 wild type or misc-1 knock-out animals, and observed a 6% reduction in glucose concentration in misc-1 mutants, compared to control (t-test P = 0.0054). The amount of glucose was normalized to protein content. (D) We quantified the amount of cellular ATP in N2 and misc-1 mutants, and observed no significant difference in the concentration of ATP between the two strains (t-test P = 0.09). doi:10.1371/journal.pone.0017827.g006

Discussion
We identified misc-1, the C. elegans orthologue of human OGC. By studying misc-1 in nematodes, we uncovered new functions for this gene that are conserved in higher eukaryotes. We showed that MISC-1/OGC is not merely involved in regulation of mitochondrial metabolism by participating in the malate-aspartate shunt. The protein seems to participate in a higher order of metabolic regulation by interacting with Bcl-2-family members and by regulating insulin secretion. Specifically, our experiments in C. elegans and mammalian cells show that MISC-1 and OGC regulate mitochondrial fusion/fission, cell death and interact with regulators of apoptosis.
First, we showed that reducing or eliminating expression of MISC-1/OGC resulted in mitochondrial fragmentation in C. elegans and in human cells. The mitochondrial fragmentation phenotype and the concomitant reduction in cristae number in misc-1 knock-out animals suggest a reduction in mitochondrial metabolism (reviewed in [5]). Mitochondrial metabolism also decreases and cristae condense in the absence of the fusion machinery [57]. For instance, loss of function of Mitofusin-2 (Mfn2) -a key protein involved in mitochondrial fusion -in cell (C-D) Transgenic animals expressing a lag-2p::gfp reporter were exposed to either control (empty) RNAi vector (C) or misc-1 RNAi (D). misc-1 knock-down results in longer DTC cytonemes than control (arrows delineate the length of the cytonemes). (E) Measurement of cytoneme length in control animals and animals exposed to misc-1 RNAi. The mean cytoneme length was 74.9 mm in control, while it was 150.9 mm in misc-1 knock-down worms. Cytonemes in misc-1 knock-down worms were therefore twice as long as in control (unpaired ttest P = 0.0080). (F) Absence of MISC-1 does not affect lag-2 expression. We quantified reporter expression with ImageJ. We observed no difference in GFP expression between control worms and misc-1 RNAi-treated worms (t-test P = 0.3137). All error bars represent 6SEM. doi:10.1371/journal.pone.0017827.g007 cultures inhibits pyruvate, glucose and fatty acid oxidation and reduces mitochondrial potential [58]. Mfn2 expression is downregulated in the Zucker rat model of obesity [57,58] and in human subjects with obesity or diabetes [59]. Hence, mutations affecting mitochondrial structure also affect metabolism. Here we provide an example that the opposite is also true: a metabolic protein affects mitochondrial structure.
The mitochondrial fragmentation conferred by misc-1 knock-out or knock-down could be mediated by its physical interaction with CED-9/Bcl-2, which is a component of the core apoptotic machinery. The fact that knocking-down misc-1 in C. elegans and OGC in human cells resulted in the same phenotypemitochondrial fragmentation -suggests that MISC-1/OGC might be part of an ancient, phylogenetically conserved mechanism to induce mitochondrial fragmentation. This hypothesis is strengthened by the observation that CED-9 physically interacts with FZO-1/Mfn1,2 to promote mitochondrial fusion [60].
The interaction of MISC-1/OGC with CED-9/Bcl-2 may not be unique, as other mitochondrial proteins localized to the inner mitochondrial membrane might similarly interact with Bcl-2family members in order to regulate the mitochondrial fusion/ fission equilibrium according to metabolic needs. An example of such a protein is ANT, the primary role of which is to mediate the exchange of ADP and ATP across the inner mitochondrial membrane. MISC-1/OGC and ANT are of similar size, share protein domains (three solute carrier or SolCar domains), and are both localized to the inner mitochondrial membrane [61].
Our data suggest that MISC-1 and OGC participate in an evolutionarily conserved apoptosis program. Furthermore, we showed that misc-1 modulates apoptosis through the physiological (LIN-35/Rb) pathway. This pathway is not activated by stress, but functions in the C. elegans germline during oocyte production [33,34], similarly to apoptosis during oocyte production in humans. It has been shown that the physiological pathway activates the core apoptotic machinery through an unknown mechanism, independent of the BH3-only protein EGL-1 [33]. We hypothesize that the physiological pathway of apoptosis is triggered by the efflux of mitochondrial pro-apoptotic factors controlled by the function of MISC-1/OGC and the MPTP.
It has been suggested that the MPTP might function as a lowaffinity Ca 2+ channel [62]. It is interesting to speculate that Ca 2+ may mediate the apoptosis and insulin secretion phenotypes observed in the absence MISC-1/OGC, since we were unable to detect an increase in ATP levels in whole-organism extracts from misc-1 mutants that exhibited higher insulin secretion. It would be interesting to measure ATP levels specifically in the C. elegans insulin-secreting cells. However, because of the lack of techniques to dissect single tissues in the adult, we measured total ATP levels, thereby possibly diluting any effect misc-1 mutations might have on ATP production in insulin-producing cells. Whether through release of mitochondrial Ca 2+ or not, we offer genetic and molecular evidence that absence of MISC-1 results in increased insulin secretion in C. elegans.
Our experiments do not distinguish whether MISC-1 has effects on secretion of only DAF-28 or if it is a more general modulator of insulin secretion. Further studies should be undertaken to clarify this issue. a-ketoglutarate, the main substrate of MISC-1/OGC, is an insulin secretagogue (reviewed in [62]). Furthermore, overexpression of Bcl-x L in mouse b-cells impairs glucose-and KClstimulated insulin secretion and intracellular Ca 2+ levels [35]. Bcl-2-proteins could modulate insulin secretion via release of mitochondrial and ER Ca 2+ [63]. It is therefore possible that MISC-1/OGC affects insulin secretion by modulating the function of Bcl-2 proteins, probably through regulation of Ca 2+ release from mitochondrial stores. Interestingly, an expression microarray study [64] showed that the gene encoding OGC is also down-regulated in muscles of individuals affected by type 2 diabetes or with a family history of the disease, compared to controls.
Absence of MISC-1 favours mitotic proliferation of germline stem cells. This result seems to further corroborate the increase in insulin secretion in misc-1 knock-down and knock-out animals. In fact, increased insulin secretion and insulin signalling induces germline proliferation in C. elegans, independently of the Notch ligand LAG-2 [52]. Here we show that absence of MISC-1 induces germline stem cell proliferation independently of LAG-2. We considered the possibility that increased germline stem cell proliferation may represent a compensatory mechanism for the increased apoptotic rate in the germline of misc-1 animals. However, several reports suggest that germline stem cell proliferation and apoptotic rates are not interdependent. For example, absence of the germline RNA helicase CGH-1 results in increased apoptosis and decreased germline proliferation [65], whereas absence of the RNA binding protein GLD-1 results in reduced apoptosis and increased germline proliferation [66]. We speculate that increased insulin secretion stimulates the growth of DTC cytonemes and that longer cytonemes are responsible for maintaining the extended mitotic zone observed in misc-1 germlines. To the best of our knowledge, such an association between cytoneme length and germline stem cell proliferation has not been reported before.
In conclusion, we identified MISC-1/OGC as a novel regulator of mitochondrial fusion/fission, apoptosis, insulin secretion and germline proliferation. The functions of this protein seem to be phylogenetically conserved from nematodes to mammals. Our data suggest that MISC-1/OGC has a role in assessing the metabolic status of a cell and subsequently in making cell survival decisions. We find it intriguing that a primarily metabolic protein like MISC-1/OGC plays roles in such a wide array of functions. The study of other metabolic mitochondrial carriers in the context of their wider cellular functions might elucidate the complex mechanisms underlying the integration of cell survival signals with cell metabolism.

Transgenic lines
For the transgenic line carrying the MISC-1::GFP reporter, we amplified 1165 bp upstream of the misc-1 ATG codon and the full misc-1 gene from N2 genomic DNA. We then fused misc-1 to gfp using the stitching PCR method [68]. The transgenes were injected with the rol-6(su1006) co-injection marker in the gonads of N2 hermaphrodites according to established protocols [69]. Although the transgene was not integrated in the genome, ,90% of animals in each generation carried the transgene.

RNAi, siRNA and shRNA
RNAi by feeding was performed as previously described [70,71]. The clone used for misc-1 RNAi was B0432.4. misc-1 RNAi and the empty RNAi vector (L4440; [70]) were propagated in Escherichia coli HT115 and administered to worms according to established feeding RNAi protocols [72]. For siRNA in HEK293 cells, we designed oligos targeting OGC with the software OligoEngine Workstation. The oligos were annealed in the siRNA vector pSuper.gfp/neo (OligoEngine) to make pSuperOGC. 1 mg of pSuperOGC or a control vector containing a scrambled mammalian sequence (X-Scramble) were transfected in HEK293 cells using Lipofectamine (Invitrogen). HEK293 transfection with Lipofectamine was done for 6-8 h at 37uC. Cells were grown for 24-48 h before being used for microscopy or RNA extraction. For OGC knock-down in MIN6 b-cells, an shRNA vector targeting mouse OGC was employed (TG504014, OriGene). The cells were grown in 24 well plates and transfected for 6-8 hours using 500 ng of either OGC shRNA vector or a vector containing a scrambled shRNA cassette as control and 1 ml Lipofectamine 2000 (Invitrogen) in 500 mL Opti-MEM, according to the manufacturer's instructions. Following transfection, MIN6 cells were cultured in DMEM with no penicillin/streptomycin for 72 hours before the assessment of apoptosis activation.

Microscopy
Worms were mounted on a 3% agarose pad and immobilized with 10 mM Levamisol (Sigma-Aldrich). Fixed HEK293 cells were mounted on microscope slides using the Vectashield mounting solution (Vector Laboratories). Images were taken using a Zeiss Axioskop with a QImaging Retiga 2000R camera. The imaging software used was Openlab 5.5.0. For some images (Fig. 1G-H,  Fig. 2), we used a Zeiss Axiovert 200 M confocal microscope with the LSM 5 Pascal laser system. Quantification of GFP reporter expression was performed with ImageJ (Rasband WS, ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997-2009.).

Electron microscopy
Adult worms were washed directly into a primary fixative of 2.5% glutaraldehyde in 0.1 M Sorensen phosphate buffer. Worms were transferred to microcentrifuge tubes and fixed for one hour at room temperature. Samples were then centrifuged at 3,000 rpm for one minute, supernatant removed and washed for ten minutes in 0.1 M Sorensen phosphate buffer. The worms were then postfixed in 1% osmium tetroxide in dH2O for one hour at room temperature. Following washing in buffer, specimens were processed for electron microscopy by standard methods; briefly, they were dehydrated in ascending grades of alcohol to 100%, infiltrated with Epon and placed in aluminum planchetes orientated in a longitudinal aspect and polymerized at 60uC for 24 hours. Using a Leica UC6 ultramicrotome, individual worms were sectioned in cross section at 1 mm intervals, starting outside the worm, until the anterior tip of the animal was located as judged by examining the sections stained with toluidine blue by light microscopy. Thereafter, serial ultra-thin sections of 90 nm were taken. Sections were picked onto 200 mesh copper grids, stained with uranyl acetate and lead citrate and examined under a Tecnai Twin (FEI) electron microscope. All mitochondrial images were taken from muscle cells observable in anterior sections of the worm (i.e., typically 5-15 mm from the nose tip). Where required, sections were tilted using the Compustage of the Tecnai to ensure an exact geometrical normalcy to the imaging system. All images were recorded at an accelerating voltage (120 kV) and objective aperture of 10 mm, using a MegaView 3 digital recording system.

qRT-PCR
L1 larvae synchronized by sodium hypochlorite treatment and overnight starvation in M9 buffer were spotted on plates and grown until day 1 of adulthood. RNA was extracted with TRIzol (Invitrogen) and chloroform and purified with RNeasy Mini Kit (Quiagen). Equimolar concentrations of control and experimental RNA were used to make cDNA with SuperScript II Reverse Transcriptase (Invitrogen). Amplification of target genes was done with iTaq SYBR Green Supermix with ROX (Bio-Rad) using an AB 7500 Fast Real-Time PCR System with standard settings. Target gene expression was normalized against endogenous levels of act-2 expression. The same protocol was followed for qRT-PCR experiments with HEK293 cells and normalization was done with respect to Actb expression.

Co-immunoprecipitation and Western blots
HEK293 cells were harvested in co-immunoprecipitation buffer (25 mM TRIS pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.5% NP-40, 1% sodium deoxycholate) with protease inhibitor cocktail (#539134, Calbiochem). Following sonication, the solution was pre-cleared with protein G-sepharose at slow rotation for 1.5 h at 4uC. The antibodies targeting the protein to be co-immunoprecipitated were then added at a concentration of 1:50 and allowed to bind at slow rotation at 4uC overnight. The next morning protein G-sepharose beads were added and the immunoprecipitation was allowed to run at slow rotation for 4 h at 4uC. Beads were washed four times, Laemmli buffer was added to the beads and the protein samples were heated at 65uC for 10 min before being loaded on an SDS-PAGE gel. Western blots were performed using standard techniques.
We used a similar protocol for C. elegans co-immunoprecipitation. Briefly, mixed stage worms were frozen at 280uC in coimmunoprecipitation buffer. The samples were then thawed on ice and sonified four to five times (Sonifier 450 settings: output 3 for 10 s). The lysates were then used for co-immunoprecipitation as described above for mammalian cells. For Western blots using C. elegans, we boiled the samples for 10 min in Laemmli buffer. We generated an a-MISC-1 antibody with GenScript Corporation. Other antibodies used: OGC (developed by FP); Bcl-xL (#2762, Cell Signaling Technology); a-ANT (# MSA02) and a-Complex IV subunit 1 (# MS404, MitoSciences). Western blots for activation of apoptosis employed a Caspase-3 antibody (#9662, Cell Signaling).

Staining C. elegans and HEK293 mitochondria
Mitochondria were stained with MitoTracker CMXRos (Invitrogen). To stain mitochondria in worms, we resuspended MitoTracker in molten (55uC) NG agar to a final concentration of 2 mg/mL. We poured 3-4 mL of the solution in 50 mm plates. The plates were then spotted with 100 mL of OP50. Plates were stored in the dark before and during the experiment. Synchronized L1 larvae were spotted on the plates and allowed to grow until used for microscopy on day 1 of adulthood. To stain mitochondria in HEK293 cells, we added MitoTracker CMXRos to fresh growth medium at a final concentration of 100 nM. The cells were incubated at 37uC for 45 min, fixed and counterstained with DAPI (49,6-diamidino-2-phenylindole, Sigma) according to suppliers' protocols.

Germline apoptosis and DAPI staining and phospho-H3 immunostaining
To stain apoptotic bodies in the germline, we used the dye SYTO12 green fluorescent nucleic acid stain (Invitrogen) according to established protocols [33].
DAPI staining was done following standard techniques. Phospho-H3 immunostaining was performed as previously described [73].

Metabolic assays
Glucose concentration in worms was measured with a Glucose Assay Kit (#K606-100, BioVision) according to manufacturer's instructions. Briefly, synchronized day-one adults were washed 5 times in M9 buffer. The worm pellet was resuspended in 500 mL of the glucose assay buffer provided in the kit, frozen for storage at 280uC, thawed on ice and sonified (Sonifier 450) five times (output level 3, time 10 s) to obtain extracts. After centrifugation at 13,000 rpm at 4uC, the supernatant was used for the glucose assay as per manufacturer's manual. The glucose concentration was normalized to protein content in each sample. The lysates for ATP analysis were generated from synchronized day-one adults, which were frozen, sonicated as described above, and boiled for 15 min. ATP contents were assayed with the Adenosine 59-triphosphate (ATP) Bioluminescent Assay Kit (#FL-AA, Sigma) according to manufacturer's instructions. Measurements of signal intensity were taken with a Tecan M200 micro plate reader.

Statistical analyses
Results are expressed as mean 6 S.E.M. A P-value of less than 0.05 was considered significant. Statistical analyses were performed with the statistical package GraphPad Prism 5.02. Figure S1 Amino acid conservation between C. elegans MISC-1 and human OGC. MISC-1 and OGC share 72% identity and 80% similarity (e-value = 8e 2105 ) at the amino acid level. The two proteins are best reciprocal matches. When doing BLAST similarity searches with the MISC-1 sequence, the second best human match is the dicarboxylate carrier, with an expect value of 2e 251 , amino acid identity of 39% and similarity of 57%. BLAST similarity searches with the human OGC amino acid sequence against the C. elegans proteome showed the second best match as the uncharacterized protein K11G12.5, predicted to be a malate carrier. In this case, the expect value was 8e 250 , amino acid identity was 40% and similarities were 57%. Identical amino acids are shown in black, similarities are shown in grey. (TIF) Figure S2 Starvation increases MISC-1 expression. We observed an increase in MISC-1::GFP reporter expression in animals that were placed for 24 h on plates that contained no food, compared to animals that were fed ad libitum on plates streaked with E. coli OP50. The increase in reporter expression upon starvation was especially noticeable in the posterior bulb of the pharynx (outlined in upper right panel) and in the posterior intestine (outlined in lower right panel). (TIF) Figure S3 Evaluation of misc-1 and OGC knock-down. (A) Synchronized L1 larvae were treated with either empty vector control (L4440) or misc-1 RNAi vector until day 1 of adulthood. We performed a Western blot on the proteins extracted from these strains with our MISC-1 antibody. The results show a ,84% down-regulation of MISC-1 levels upon misc-1 RNAi treatment. (B) Quantitative Real-Time PCR (qRT-PCR) was performed to assess the extent of OGC knock-down in HEK293 cells treated with OGC siRNA. As a control, we transfected cells with a mammalian X-Scramble siRNA vector. OGC expression levels were normalized to Actb. Our results indicate that OGC siRNA reduced target gene expression by ,80%. Error bars: 6S.D. (TIF) Figure S4 misc-1 knock-down does not affect muscle morphology. Control RNAi and misc-1 RNAi on a transgenic strain carrying a myo-3p::myo-3::gfp. This transgene is expressed in muscles and provides a tool to assess muscle morphology. These confocal images show that muscle morphology in misc-1 RNAitreated worms is indistinguishable from that of control RNAitreated worms. The mitochondrial fragmentation phenotype observed in Figure 2B is therefore caused by a specific effect of misc-1 RNAi on mitochondria, not on muscle structure. (TIF) Figure S5 misc-1 RNAi does not compromise the ability of the somatic gonad to remove apoptotic corpses. Control RNAi and misc-1 RNAi treatment on a transgenic strain carrying CED-1::GFP are shown. This fluorescent reporter allows visualization of the apoptotic corpses being engulfed and removed by the somatic sheath cells (arrows). We observed a two-fold increase in the number of engulfing events per gonad arm in worms treated with misc-1 RNAi, compared to control. This result suggests that the increase in apoptotic events observed in Fig. 4A-B was due to an effect of misc-1 on apoptosis, and was not caused by a defect in the mechanism of cell corpse removal.