Abstract

Mitochondria integrate catabolism, anabolism, and signaling

Mitochondria are bioenergetic and biosynthetic organelles that take up substrates from the cytoplasm and use them to drive fatty acid oxidation (FAO), the TCA cycle, the electron transport chain (ETC) and respiration, and to synthesize amino acids, lipids, nucleotides, heme and iron sulfur clusters, as well as NADPH for their own antioxidant defense (Figure 1) (Wallace, 2012). NADH and FADH₂ produced via TCA cycle turning powers the ETC, which in turn generates a proton gradient across the mitochondrial inner membrane that produces ATP through the action of the H⁺-ATP synthase. Dihydroorotate dehydrogenase (DHODH), which is essential for de novo pyrimidine synthesis, requires a functional respiratory chain for activity (Khutornenko et al., 2010). The reactive oxygen species (ROS) generated in mitochondria as a byproduct of the ETC can activate signal transduction pathways such as that of MAP kinase and the HIFs (in normoxia is referred to as pseudohypoxia) (Shadel and Horvath, 2015; Sullivan and Chandel, 2014). Excess ROS production can also lead to cell death. Mitochondria sequester Ca⁺⁺, the selective release of which controls signaling and a broad array of cellular functions. Mitochondria act as signaling platforms for innate immunity (e.g. MAVS and by the release of mtDNA) (Weinberg et al., 2015; West et al., 2015). The BCL-2 family of proteins at the mitochondrial outer membrane control programmed cell death by apoptosis. BCL-2-realated anti-apoptotic proteins block, whereas the BAX and BAK pro-apoptotic proteins promote the release of cytochrome c from the mitochondrial inter membrane space triggers the apoptosome and caspase protease activation in the cytosol, causing cell death (Figure 1) (Czabotar et al., 2014; Moldoveanu et al., 2014). Thus, communication between mitochondria and the cell coordinates a diverse array of functions that are critical for cell metabolism, growth and survival.

Open in a separate window

Figure 1

Mitochondria as biosynthetic and signaling hubs

The TCA cycle uses substrates from glycolysis, fatty acid oxidation, and amino acid catabolism to generate building blocks and high-energy electrons (NADH and FADH₂) to power the ETC. Major biosynthetic products produced by mitochondria and signaling pathways regulated by mitochondrial function are indicated.

In addition to being the bioenergetics power-house and signaling hub, mitochondria also function as a platform for generating biosynthesis building blocks. This requires a continuous inflow of 4 carbon (4C) units to balance the outflow of such units to amino acids and other biosynthetic products (Figure 1). This replenishment of the TCA cycle 4C units is termed anaplerosis and can occur through carboxylation of pyruvate or catabolism of glutamine and other amino acids (Comerford et al., 2014; DeBerardinis et al., 2007; Fan et al., 2009; Mashimo et al., 2014; Yuneva et al., 2007). Such provision of 4C units is in addition to the need for 2C units that are provided by acetyl-CoA, which can be formed from fatty acids, pyruvate, acetate, and many amino acids. The condensation of a 4C and 2C unit yields citrate via citrate synthase, which is solely mitochondrially localized. In addition to being an upstream precursor capable of generating all classical TCA cycle intermediates, citrate can be exported to the cytosol, where it is broken down to oxaloacetate and acetyl-CoA, which is required for lipid synthesis and protein modifications (Figure 1). In addition to 4C and 2C units, mitochondria also play a central role in the metabolism of 1C units required for purine, thymidine, and methionine synthesis. Recent literature has highlighted several advancements in the understanding of mitochondria-mediated biosynthetic pathways.

The mitochondrial genome in cancer

Acquired nuclear somatic and germ line mutations in about one hundred genes are known to provide a selective advantage in cancer (Stratton et al., 2009). Increases in the mutation frequency due to DNA repair defects, replication errors, carcinogen exposure or aging promote cancer by increasing the occurrence of mutations in these genes. These cancer driver mutations and their functional consequence to cancer growth are often well understood and are the subject of targeted anti-cancer drug development. In contrast, the status and role of mutations in the mitochondrial genome in cancer has been less clear. Studies with small sample sizes using the low sensitivity of past sequencing technologies to assess heteroplasmic genomes have rendered association of somatically acquired mitochondrial genome mutations with cancer unclear. Moreover, the now commonplace advanced nuclear sequencing studies usually ignore the mitochondrial genome.

Two recent large-scale efforts to examine the mutational landscape of the mitochondrial genome in over 2000 human cancers across more than 30 tumor types compared with normal tissue from the same patient by massively parallel DNA sequencing has revealed important insights (Ju et al., 2014; Stewart et al., 2015). These studies identified many somatic substitutions with strong replicative strand bias, with C to T and A to G transversions on the mitochondrial heavy strand indicative of mitochondrial polymerase G errors as the major cause of mutations. This is surprising, given the close association of mitochondrial ROS production with mutations, which is apparently dwarfed by the frequency of polymerase errors that occur during the normal process of genome replication. More importantly, missense mutations are selectively neutral and drift toward homoplasmy, whereas there is negative selection for deleterious, pathogenic mutations that remain heteroplamic (Ju et al., 2014; Stewart et al., 2015). These findings clearly indicate that there is generally selective pressure in human cancers for retention of mitochondrial genome function.

Mitochondrial quality control in cancer

For cells to function properly, there needs to be communication back and forth between mitochondria and the nucleus, referred to as antegrade and retrograde signaling (Chandel, 2015). Nuclear genes control mitochondrial biogenesis to synthesize more mitochondria and meet increased metabolic demand. Opposing biogenesis is another layer of nuclear control by the autophagy and mitophagy genes. The lack of accumulation of pathogenic mitochondrial genome mutations in human cancers suggests engagement of active mechanisms for mitochondrial quality control that prevent the accumulation of defective mitochondria. Mitophagy is a specialized from of autophagy that selectively degrades and eliminates superfluous or damaged mitochondria (Randow and Youle, 2014). Biogenesis and mitophagy work in concert to regulate mitochondrial mass, function and quality. In turn, defective mitochondria provide the “eat me” signal directly to the mitophagy machinery by membrane depolarization and a cascade of phosphorylation and ubiquitation of mitochondrial proteins, while reduced cellular energy output (as occurs if the total number of mitochondria is insufficient) triggers mitochondrial biogenesis via the general energy sensor AMP kinase and by other mechanisms.

Depolarization of the outer mitochondrial membrane triggers the activation of the kinase PINK1, which phosphorylates ubiquitin causing the recruitment of the E3 ligase PARKIN to the mitochondrial outer membrane (Lazarou et al., 2015). PARKIN further ubiquitinylates mitochondrial proteins, which recruit the autophagy machinery to capture, engulf and degrade these specific mitochondria in lysosomes. Several other PARKIN-independent mitophagy receptor proteins include BNIP3 (Daido et al., 2004), NIX (Schweers et al., 2007), FUNDC1 (Liu et al., 2012), and BCL2L13 (Murakawa et al., 2015). Although it is reasonable to speculate a role of mitophagy in cancer, direct evidence that mitophagy is involved in tumorigenesis is lacking. The role of mitochondria quality control in cancer is mostly implicated by information obtained using genetic models with manipulation of bona fide autophagy genes.

Ras-driven cancers upregulate basal autophagy by activating the MiTF/TFE family of basic helix–loop–helix leucine zipper transcription factors that promote autophagy and lysosome biogenesis (Guo et al., 2011; Perera et al., 2015; Yang et al., 2011). Defects in autophagy produced by knocking out essential autophagy genes in autochthonous mouse models for cancer causes the accumulation of defective mitochondria and other autophagy substrates, and impairs mitochondrial respiration, cell growth, and survival while increasing cell death and senescence (Guo et al., 2013a; Guo and White, 2013; Guo et al., 2013b; Huo et al., 2013; Rao et al., 2014; Rosenfeldt et al., 2013; Strohecker et al., 2013; Xie et al., 2015; Yang et al., 2014). Systemic, acute autophagy ablation in mice with established Ras-driven lung cancer produces substantial tumor regression prior to significant damage to most normal tissues, suggesting that some tumors are particularly autophagy-dependent (Karsli-Uzunbas et al., 2014). Importantly, in contrast to the carcinomas generated with autophagy intact, autophagy deficient tumors resemble benign oncocytomas, which are tumors characterized by the accumulation of defective mitochondria (Guo et al., 2013a; Karsli-Uzunbas et al., 2014; Strohecker et al., 2013). Thus autophagy is required for tumors to progress from benign to malignant growth. This is also consistent with the requirement for autophagy in the malignant progression of Ras-driven benign pancreatic PANIN lesions (Rosenfeldt et al., 2013; Yang et al., 2014). Thus autophagy is induced by and required for overcoming a barrier to malignant growth in a variety of cancer types.

As a major role of autophagy is mitochondrial quality control and possibly also providing substrates for mitochondrial metabolism, autophagy may promote tumorigenesis and malignancy by these mechanisms. Direct demonstration of this is lacking, as is the identity of specific substrates provided by autophagy-mediated recycling and the metabolic pathways they support is not yet known. Autophagy deficiency in tumor cells causes glutamine dependency suggesting that glutamine may be one autophagy-supplied substrate (Guo et al., 2013a; Strohecker et al., 2013). It is also not known if autophagy eliminates mitochondria with pathogenic mitochondrial genome mutations in tumor cells to preserve the functioning pool of mitochondria or whether the defects in respiration observed without autophagy are the result of protein damage or substrate limitation. In neurons, however, mitophagy is important for the removal of damaged mitochondrial proteins in the setting of mitochondrial genome damage (Pickrell et al., 2015).

Oncocytomas accumulate defective mitochondria

Oncocytomas are rare, benign tumors of most epithelia characterized by the vast accumulation of defective mitochondria due to pathogenic mtDNA mutations (Gasparre et al., 2011). In mouse models, loss of autophagy results in oncocytoma tumors: genetic deletion of an essential autophagy gene causes Ras- and Braf-driven lung carcinomas to become benign oncocytomas (Guo et al., 2013a; Karsli-Uzunbas et al., 2014; Strohecker et al., 2013). Oncocytomas provide an example of tumors with defective mitochondrial respiration caused directly by pathogenic mitochondrial genome mutations. The existence of oncocytomas raises many questions. Why do oncocytomas accumulate defective mitochondria? Why are oncocytomas benign? Is there a role for defective mitochondria in either initiating tumorigenesis or limiting progression to benign disease?

A recent comprehensive analysis of the mutational landscape of human renal oncocytomas revealed that there are two main subtypes (Joshi et al., 2015). Type 1 is characterized by CCND1 rearrangements whereas Type 2 is aneuploid with loss of chromosome 1, and X or Y and/or 14 and 21, and displays male gender bias. There are few and no recurrent somatic mutations in the nuclear genome in either subtype. Type 2 oncocytoma may be a precursor of the more aggressive eosinophilic chromophobe renal cell carcinoma (ChRCC) based on similar transcriptomes and copy number variations (Joshi et al., 2015). What both renal oncocytoma subtypes share in common is recurrent pathogenic mutations in the mitochondrial genome, which may lead to ROS production that could contribute oncogenic signaling or perhaps to accumulation of oncometabolites.

Primary oncocytoma cells are defective for respiration and ROS production, are highly glycolytic, and have a transcriptional signature of p53 and AMP kinase activation, suggesting that loss of respiration may activate a metabolic checkpoint that limits tumor growth to benign disease (Joshi et al., 2015). Moreover, oncocytomas have disruption of the Golgi and vesicle trafficking, particularly loss of processing and activation of major lysosomal cathepsin proteases, accompanied by accumulation of autophagy substrates. Therefore, chronic energy crisis in oncocytomas caused by pathogenic mitochondrial mutations and defects in oxidative phosphorylation, may block trafficking and cause profound cellular dysfunction, limiting these tumors to benign disease (Joshi et al., 2015). Oncocytomas could also have deficits in the synthesis of other important building blocks such as amino acids and they may be expected to require reductive carboxylation of α-ketoglutarate to generate citrate. These hypotheses will be interesting to test in the future.

One long-standing idea is that pathogenic mitochondrial mutations cause oncocytomas, since some tumors are near homoplastic for a mutant mitochondrial genome (Gasparre et al., 2011), a finding that has held up in more recent genomic analyses (Joshi et al., 2015; Lang et al., 2015). Although they share the property of pathogenic mitochondrial mutations, both oncocytoma subtypes also have nuclear genome mutations, specifically CCND1 rearrangements and whole chromosome copy number losses, which can explain why they are tumors. Why they accumulate mtDNA mutations is less clear. One possibility is that pathogenic mitochondrial mutations are selected for because they produce ROS that activates HIFs and perhaps other oncogenic signaling pathways. The mutational signature in the mitochondrial genome in oncocytomas reflects ROS damage in addition to the expected signature of polymerase errors, but as respiration is defective, there is no ROS production or HIF transcriptional signature in tumors (Joshi et al., 2015). Thus, evoking this mechanism would require a temporal sequence of partial mitochondrial dysfunction, ROS production and transient activation of oncogenic signaling that initiates tumor growth followed later by complete loss of respiration and benign disease.

Different subtypes of mitochondria-rich tumors

Striking numerical increases in mitochondria is a characteristic of distinct tumor types with opposing clinical outcomes. Oncocytomas accumulate vast overabundance of mutant mitochondria, and as a result, are well known for their hallmark expansive, eosinophilic cytoplasm. This “oncocytic feature” associated with mitochondrial accumulation in tumor cells is also a characteristic of other tumors, although the significance is unknown.

Oncocytic tumor types include those arising in individuals with Birt-Hogg-Dube (BHD) syndrome that develop spontaneous renal and other tumors (Hasumi et al., 2012; Linehan et al., 2013). Most cases of BHD are caused by mutations in the FLCN tumor suppressor gene that encodes Foliculin, a GTPase activating protein (GAP) for the Rag GTPases (Tsun et al., 2013). Rags promote mTOR activation when amino acids are present, and this activity of the Rags is normally suppressed by FLCN. Loss of FLCN thereby promotes mTOR activation and signaling of cell growth that is a common oncogenic mechanism. Although FLCN may have other functions, its loss removes a repressive signal on mTOR in the absence of amino acids to produce constitutive signaling of cell growth. In sharp contrast to oncocytomas, the vast majority of which are benign, tumors resulting from FLCN loss progress to malignancy (Linehan et al., 2013). So why are tumors that accumulate mitochondria both benign and malignant?

Malignant tumors accumulate respiration functional mitochondria

In contrast to benign oncocytomas, oncocytic renal BHD tumors resulting from FLCN loss accumulate morphological normal, respiration-proficient mitochondria (Lang et al., 2015). Cells from these FLCN-deficient tumors have higher rates of respiration than normal cells, likely explained by the increased number of normal mitochondria (Hasumi et al., 2012). Thus there are two classes of mitochondria-rich oncocytic tumors, those that have mitochondrial defects that are benign, and those where mitochondrial function remains intact that are malignant (Figure 2). This fits with the concept of a tumor-promoting role for mitochondria. The reason why FLCN-deficient tumors accumulate mitochondria is not known, but mTOR promotes mitochondrial biogenesis and inhibits mitochondrial elimination by autophagy. The transcriptional signature for mitochondrial biogenesis is present in FLCN-deficient tumors and it would be interesting to explore the functional role of this and also the potential role of repressed mitophagy.

Open in a separate window

Figure 2

Characteristics of oncocytic (mitochondria-rich) tumors with functional versus defective mitochondria

Mutations in TCA cycle enzymes promote cancer

Mutations in the nuclear encoded TCA cycle enzymes isocitrate dehydrogenase (IDH) 2, succinate dehydrogenase (SDH)× (SDHA-D and SDHAF2), and fumarate hydratase (FH) are found in human cancers (Gaude and Frezza, 2014). IDH2 mutations, like those in the closely related cytosolic protein IDH1, are genetically dominant, with active site mutations in a single allele producing a neomorphic activity where α-ketoglutarate is reduced to the oncometabolite 2-hydroxyglutarate (2-HG) (Dang et al., 2009; Losman and Kaelin, 2013). There are many α-ketoglutarate-dependent enzymes, particularly dioxygenases, and 2-HG is a competitive inhibitor of these enzymes, particularly, hypoxia-inducible factor (HIF) prolyl hydroxylases (PHDs), JmjC domain- containing histone demethylases (part of the JMJD family) and the ten-eleven translocation (TET) family of 5methyl cytosine (5mC) DNA hydroxylases. Oncometabolite-driven changes in the epigenome suppress differentiation and promote proliferation (Dang et al., 2009; Lu and Thompson, 2012; Parker and Metallo, 2015). 2-HG has also been found to inhibit the PHDs that degrade the HIFs and may thereby have a role in cancer. The exact dioxygenase and cancer pathway affected by 2-HG may depend on the cancer type. Nonetheless this provides a clear example of how a mutation in a mitochondrial TCA cycle enzyme can contribute to cancer through production of an oncometabolite.

Loss of function mutations in FH and SDH may act similarly by producing the accumulation of succinate, which also antagonizes the function of these α-ketoglutarate-dependent dioxygenases. Loss of FH also causes the accumulation of fumarate and the inactivation of proteins through aberrant succinylation of proteins such as the KEAP1 tumor suppressor that results in activation of the oncoprotein NRF2 (Adam et al., 2011; Adam et al., 2014). Fumarate also causes succinylation of glutathione, elevated ROS production and NRF2 activation, which can contribute to oncogenesis (Sullivan et al., 2013). Although these findings revealed new cancer-causing paradigms with significant translational implications, they also raise the question of how cancer cells adapt to these significant alterations in an important mitochondrial metabolic activity.

Targeting of mitochondrial metabolism for cancer therapy

Understanding of the role of mitochondria in cancer is revealing novel approaches to targeted therapy. Targeting 1C and glutamine metabolism and aspartate synthesis is being explored, as discussed above. The complex I inhibitor, metformin, a biguanide drug commonly used in the treatment of diabetes, has anticancer activity in diabetics, and it is of great interest to determine if this is more widely applicable (Pollak, 2013). One of the most important questions regarding metformin is whether its anticancer activity arises from reduced systemic glucose and insulin levels, or alternatively through targeting of the tumor ETC. Recently, it is been shown that expression of a metformin-resistant variant of Complex I within a Ras-driven colon cancer xenograft, rendered the tumor insensitive to metformin (Wheaton et al., 2014). Thus, at least within this model system, metformin impairs tumor growth by blocking tumor-intrinsic metabolism. While this could involve redox changes within the tumor, such as elevated NADH and reduced mitochondrial ROS production, the simplest explanation involves a dependence of the tumor on oxidative phosphorylation to make ATP.

Tumors cells with sensitivity to low glucose and reduced oxidative phosphorylation are sensitized to biguanides that inhibit complex I (Birsoy et al., 2014). Certain cancer stem cells, such as Ras-driven pancreatic cancer stem cells, appear to be particularly reliant on oxidative phosphorylation (Lonardo et al., 2013). Such cells are enriched after genetic extinction of Ras signaling, raising the possibility of synergy between oncogene-targeting agents and metformin (Viale et al., 2014).

Promoting mitochondrial ROS production to induce cancer cell death may enhance the activity of chemotherapy (Yun et al., 2015). Inhibiting autophagy and mitophagy can interfere with mitochondrial metabolism by blocking mitochondrial quality control and/or substrate supply (White, 2015; White et al., 2015). This is currently being explored in clinical trials by interfering with lysosome function (White et al., 2015). Targeting the MiTF/TFE family of transcription factors and master regulators of autophagy and lysosomal biogenesis activated in many cancers (Ferguson, 2015) may provide another option and may identify tumors potentially sensitive to autophagy inhibition.

Tumors where TCA cycle enzymes are mutated has also generated novel approaches. Tumors with mutations in TCA cycle enzymes require reductive carboxylation to generate the necessary TCA cycle intermediates, thus their metabolism is distinct from most normal tissues identifying a metabolic vulnerability (Cardaci et al., 2015; Mullen et al., 2012). The most promising is targeting the neomorphic activity of mutant IDH (Rohle et al., 2013; Wang et al., 2013). FH mutations cause a dependency on heme oxygenase (HO) to run a partial TCA cycle and NADH production important for viability (Frezza et al., 2011). Similarly, pyruvate carboxylase (PC) enables aspartate synthesis in SDH-deficient tumor cells creating a metabolic vulnerability (Cardaci et al., 2015). More generally, tumors with defective mitochondrial function, including benign oncocytomas and SDH and FH-deficient malignant tumors, require up-regulated glycolysis to meet their energy demands. Accordingly, such tumors may be particularly sensitive to inhibition of glucose uptake or glycolysis. Such inhibition may be close to what Warburg envisioned 80 years ago. Yet greater opportunities, however, lie in the bulk of tumors that do not conform to Warburg’s initial hypothesis of mitochondrial dysfunction, in which case is the secret may be to target effectively the mitochondria themselves.

Acknowledgements

Footnotes

References