Pyruvate dehydrogenase complex (PDC) export from the mitochondrial matrix

Abstract

Studies on mitochondria protein import had revealed in detail molecular mechanisms of how peptides and proteins could be selectively targeted and translocated across membrane bound organelles. The opposite process of mitochondrial export, while known to occur in various aspects of cellular physiology and pathology, is less well understood. Two very recent reports have indicated that a large mitochondrial matrix protein complex, the pyruvate dehydrogenase complex (PDC) (or its component subunits), could be exported to the lysosomes and the nucleus, respectively. In the case of the latter, evidence was presented to suggest that the entire complex of 8–10 MDa could translocate in its entirety from the mitochondrial matrix to the nucleus upon mitogenic or stress stimuli. We discuss these findings in perspective to what is currently known about the processes of transport in and out of the mitochondrion.

• Lysosome
• mitochondria
• mitochondria-derived vesicles (MDVs)
• nucleus
• pyruvate dehydrogenase complex (PDC)

Introduction

The evolutionary transfer of protein encoding capacity to the nucleus by the endosymbiont ancestor of the eukaryotic cell mitochondrion necessitated specific mitochondrial targeting mechanisms for mitochondrial proteins that have a biogenic origin at the cytosolic ribosome. For those that function in the matrix, the polypeptides would need to be translocated across both the outer and inner mitochondrial membranes. These translocations are accomplished by a series of mitochondrial translocase complexes (Dudek et al., 2013; Höhr et al., 2014; Qiu et al., 2013; Rehling et al., 2004) that are functionally (as well as physically) connected. Precursor polypeptides first enter into the intermembrane space (IMS) through the Translocase of the Outer Membrane (TOM) complex. Proteins that contain the mitochondrial targeting presequence are then inserted into the Inner Mitochondrial Membrane (IMM) by the Translocase of the Inner Membrane 23 (TIM23) complex. Translocation of soluble proteins into the matrix requires the activity of the presequence translocase-associated import motor (PAM) and an established membrane potential across the IMM. Other translocase complexes and protein networks such as the Translocase of the Inner Membrane 22 (TIM22) complex, the mitochondrial Sorting and Assembly Machinery (SAM) (Höhr et al., 2014), and the Mitochondrial Intermembrane space import and Assembly (MIA) pathway (Stojanovski et al., 2008) also facilitates import and IMS residence of a variety of polypeptides.

In spite of the fact that the mitochondria could import large proteins, the opposite process of mitochondrial export is much less well understood. Mitochondrial exit of apoptosis-inducing components, such as cytochrome c and apoptosis inducing factor (AIF), is indeed well known (Kroemer et al., 2007; Vaux, 2011). These proteins are however largely found in the IMS. The IMM could be compromised on occasions of mitochondrial membrane permeabilization, when opening of pores (Bernardi, 2013) at the IMM allows a two-way flow of solutes and small molecules that are less than 1.5 kDa in size. On the whole, however, specific or targeted export of larger mitochondrial matrix proteins to another cellular destination is not known to occur.

The pyruvate dehydrogenase complex (PDC) is responsible for the generation of acetyl-CoA from pyruvate, which feeds into the tricarboxylic acid (TCA or Krebs) cycle. The PDC harbors 3 enzyme activities (E1, E2 and E3), each comprises of multiple polypeptide subunits. It is one of the largest multiprotein complexes known, with the smaller bacterial homologue at about 4.5 MDa and the mammalian form about 9 MDa in size, which are significantly larger than the largest complex in the respiratory chain, as well as ribosomes (Zhou et al., 2001; Patel & Korotchkina, 2006). Intriguingly, recent reports have indicated that PDC or its components could exit the mitochondria, and are then targeted to the lysosome and nucleus, respectively. The former is associated with a class of oxidative stress-induced mitochondria-derived vesicles (MDVs) (McLelland et al., 2014; Soubannier et al., 2012a), while the latter has no clearly delineated mechanism (Sutendra et al., 2014). In the following paragraphs, we discuss these new findings in the light of what is currently known about protein import and export from the mitochondria.

Evidence for mitochondrial exit of the matrix localized pyruvate dehydrogenase complex (PDC)

MDVs are first described by the McBride laboratory as membranous vesicles of 70–100 nm in diameter that contains a specific membrane protein, the outer-membrane mitochondria-anchored protein ligase (MAPL) (Neuspiel et al., 2008). Over-expression of MAPL induces mitochondria fragmentation that is suppressed by a dominant negative mutant of the mitochondrial fission regulator dynamin-related protein 1 (Drp1), and the McBride laboratory have subsequently showed that MAPL acts as a SUMO E3 ligase that regulates mitochondrial fission (Braschi et al., 2009). Cells expressing both MAPL and dominant-negative DRP1 exhibited extensively fused mititochondrial reticulum, but in spite of this, a small pool of small vesicles that are either positive for the outer membrane TOM complex component TOM20, or MAPL, persisted. Interestingly, the authors showed that these MAPL-containing MDVs are targeted to the peroxisomes, and later speculated that these MDVs could contribute to de novo peroxisome biogenesis (Sugiura et al., 2014). This mitochondria-peroxisome transport process requires the retromer complex (Braschi et al., 2010), which was better known for its role in endosome-Golgi transport (Seaman, 2012). The retromer component Vps35 associates with MAPL, is recruited to MDVs, and it’s silencing significantly attenuated MAPL transport to peroxisomes (Braschi et al., 2010).

Subsequent findings from the McBride laboratory noted that MDVs are not phenotypically homogenous, and may indeed have different modes of generation under different physiological or pathological conditions. Mitochondrial stress induces the formation of MDVs that are enriched in oxidized or damaged cargo (Soubannier et al., 2012b). These MDVs containing damaged proteins do not go to the peroxisomes, but are instead targeted to lysosomes (Soubannier et al., 2012a), where they presumably degraded. Although reminiscent of mitochondrial autophagy or mitophagy, the process is apparently independent of classical autophagy components such as ATG5 and LC3, and was recently proposed to be classified as type 3 mitophagy, or micromitophagy (Lemasters, 2014). Importantly, a recent paper from the McBride and Fon laboratories showed that Parkin and PTEN-induced putative kinase 1 (PINK1), both mutated in early onset Parkinsonism and known to be important in multiple aspects of mitochondrial quality control (Scarffe et al., 2014), are involved in the generation of the stress-induced, lysosome targeting MDVs. These findings are of tremendous interest as they provide novel mechanistic insight into the mitochondrial associated function of these key disease genes (Sugiura et al., 2014; Winklhofer, 2014).

Of particular interest to our discussion here though are data in these reports that showed possible incorporation of PDC subunits into MDVs (McLelland et al., 2014; Soubannier et al., 2012a, b). Morphologically, MDVs could be either single membrane or double membrane bound (Soubannier et al., 2012a), and the latter type could contain matrix proteins such as subunits of the PDC. The authors detected the presence of PDC subunits E2/E3 binding protein (E3BP, found only in eukaryotes) in stress-induced Parkin/PINK1-dependent MDVs, but not the presence of E1α. The authors surmised that their findings “suggest that the megadalton PDH complex is at least partially disassembled before incorporation into vesicles” (McLelland et al., 2014). These findings provide the first evidence of a rather unexpected transport of large polypeptides or proteins (albeit damaged and destined for lysosomal degradation) from the mitochondrial matrix to another organelle.

In another, arguably even more unexpected finding, the Michelakis laboratory demonstrated that the PDC is not exclusively found in the mitochondrial matrix, but could also be present in the nucleus (Sutendra et al., 2014). The authors imaged the presence of PDC components E1, E2 and E3BP in the nucleus, and showed that nuclear PDC is functional in generating acetyl-CoA from pyruvate. Nuclear PDC is also not subjected to regulation by pyruvate dehydrogenase kinase (PDK), which phosphorylates and inhibits mitochondrial PDC. This in situ generation of acetyl-CoA in the nucleus is apparently important for histone acetylation. In connection to this, nuclear PDC is also important for the proliferative phenotype by facilitating S phase entry and cell cycle progression (Sutendra et al., 2014).

A particularly intriguing finding made by the authors pertains to the origin of nuclear PDC. There is not any clear evidence for these to be abundant in the cytosol. However, the authors showed that nuclear PDC levels are significantly elevated in the nucleus upon the addition of serum after a period of serum starvation. This elevation in nuclear PDC levels is concomitant with an obvious reduction in PDC levels of the mitochondria. The authors thus surmised that the mitochondrial PDC could have been translocated en bloc to the nucleus. In fact, this apparent translocation is also observed when cells were treated with a mitogenic factor like epidermal growth factor (EGF), and an ER oxidative stressor such as rotenone. The mechanism of mitochondrial-nuclear translocation was not explored, but presumably it would involve mechanisms that will allow the very large PDC to negotiate both the IMM and the outer membrane, followed by subsequent nuclear translocation through the nuclear pore complex. An interesting, and potentially important point noted by the authors is that the translocation is dependent on the ubiquitously present chaperone heat shock protein of 70 kDa (Hsp70). Previously known to facilitate nuclear transport, it was also shown to be associated with both PDC E1 and E2.

Known mechanisms for mitochondrial export

Are there known processes that might be able to explain the findings of mitochondrial export or exit of PDC? Processes of export from the mitochondrial are not particularly well defined or understood. Unlike the endoplasmic reticulum (ER), the mitochondrion is not known to possess the ability to retro-translocate proteins through the entry translocon, as is known to occur in the process of ER-associated degradation (ERAD) (Olzmann et al., 2013). The IMM localized Oxa-1 insertase serves import functions (Hildenbeutel et al., 2012), and is also classically known to facilitate the insertion of mitochondrial genome encoded respiratory chain complexes that are synthesized within the matrix into the IMM (Stuart, 2002). This mechanism is about the only well-defined mechanism whereby a sizable matrix polypeptide could be incorporated into, and therefore potentially translocate across, the IMM.

Other than the Oxa-1 insertase complex mediated process, there are other known processes of ER export. Firstly, during programmed cell death, the mitochondria do export a variety of cell death effector which largely resides in its IMS (Kroemer et al., 2007; Vaux, 2011). These include the well-known cytochrome c, apoptosis inducing factor (AIF) and second mitochondria-derived activator of caspases (SMAC/Diablo). Secondly, pathological stresses could also cause mitochondrial membrane permeabilization, largely as a result of the opening of the mitochondrial permeability transition pore (MPTP) (Bernardi, 2013) at the IMM. This pore could potentially allow solutes and small molecules of < 1.5 kDa to leaked out from the matrix. One possible peptide that could exit the matrix through such pores is the mitochondrial DNA encoded, 21 aa neuroprotective peptide humanin (Nishimoto et al., 2004). Thirdly, the process of mitochondrial unfolded protein response (UPR^mt) could in some contexts involve the pumping of peptides (generated by proteolytic degradation of matrix proteins) by the IMM localized HAF-1/Mdl-1 ABC transporter into the cytosol to trigger retrograde nuclear signaling (Haynes & Ron, 2010). Mitochondrial-based UPR^mt and retrograde signaling could well be mediated, at least in Caenorhabditis elegans, in a non-cell autonomous manner (Durieux et al., 2011) by mitochondrial signals of unknown molecular nature known as mitokines (Long et al., 2014). All these processes of mitochondrial export are not well-defined in terms of mechanisms and the molecular nature of the translocases involved. Importantly, none of these appear able to translocate large proteins, never mind the megadalton-size PDC.

How could PDC exit the mitochondria intact?

All things considered, incorporation of PDC subunits and components into oxidative stress and damage-induced, parkin/PINK1-dependent formation of MDVs appears to be completely plausible. However, it is difficult to imagine how an intact PDC, which has a diameter of ∼ 45 nm (Sumegi et al., 1987), could be effectively enclosed in a 70–100 nm vesicle. The observation of E2/E3BP but not E1α in these MDVs attested to the above notion. It should be noted that the authors used the monoclonal antibody 13G2AE2BH5, which epitope is usually simply defined as E2/E3BP by the vendors, and it is unclear exactly which epitope of the subcomplex it detects. There are functional PDC complexes that are smaller (∼1 MDa in size) (Sumegi et al., 1987; Zhou et al., 2001), but even these would appear too large for MDVs.

The finding by Sutendra et al. (2014) that the PDC could potentially be transferred intact from the mitochondrial matrix into the nucleus is therefore puzzling and begs an explanation. Unless there is a novel translocation mechanism that the entire field has somehow completely missed over the years (such as the translocases working in reverse), the membrane barriers involved in the translocation of such a large complex appears unsurmountable, and importantly could not be logically justified in terms of energetics. PDC is not the only known enzyme that could generate acetyl-CoA in the nucleus. Nuclear localization of ATP-citrate lyase and acetyl-CoA synthetase has been documented, and the former appears to be particularly important for histone acetylation (Wellen et al., 2009). In view of this, there is hardly a need to translocate PDC from the mitochondrial matrix to satisfy requirements for acetyl-CoA in the nucleus, unless in conditions of severely restricted availability of citrate. A cytosolic source of PDC would have been logistically easier for nuclear import, but this was not particularly prominent. It should also be noted that although the nuclear pore complex is known to allow large protein and ribonucleoprotein complexes to go through, the sheer size of PDCs would plausibly still present a problem.

An earlier paper, which is also cited by Sutendra et al. (2014), appeared to suggest that PDC could be assembled outside the mitochondrial matrix, at least in some cancer cells (Hitosugi et al., 2011). The evidence presented consists of Western blot analyses which found E1, E2 and E3BP in the outer mitochondrial membrane fraction in comparable amounts to that in the matrix. The problems are that relative concentrations between these fractions are unclear, and whether the prevalence of extra-matrix PDC in some cancers cells could be extended to the fibroblast used by Sutendra et al. (2014) is also unclear. However, if there is indeed significant amount of functional PDC that is assembled and associated with the outer membrane, any nuclear translocation of this pool of PDC would be considerably easier. This is an important point that warrants a quick verification by ultrastructural analysis.

Concluding remarks

The discussion above highlighted recent findings that the large PDC complex that resides in the mitochondrial matrix could be transported or translocated into other organelles in the cell (summarized in Figure 1). Stress-induced MDVs containing disassembled PDC subunits being targeted to the lysosome appears to be a novel pathway for mitochondrial quality control that is distinct from the more classical modes of mitophagy. On the other hand, evidence suggesting that mitochondrial matrix PDC could be translocated as an intact complex to the nucleus, while fascinating, is also puzzling in terms of the apparent difficulty and energetics concerns. We await further work that would either unveil a completely novel mechanism of intracellular transport, or provide novel insights into how known mitochondrial translocases could function in export.

Figure 1. A schematic diagram depicting the recently demonstrate transport/translocation of PDC to the lysosome (via mitochondria-derived vesicles, MDVs) and the nucleus. The former likely involved damaged and disassembled PDC subunits, while the latter was postulated to involve intact, functional PDCs. The mechanism of translocation for the latter is not known (marked as “?”), but should need to go through the nuclear pore complex (NPC). Nuclear translocated PDCs are functional in producing acetyl CoA from pyruvate diffused in from the cytosol. These transport/translocation processes are not connected to known mitochondria membrane translocases that import polypeptides from the outer mitochondrial membrane (OMM) to the intermembrane space (IMS). Translocases at the inner mitochondrial membrane (IMM) could translocate polypeptides from IMS to the matrix, or vice versa.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

This work was supported by the NUS Graduate School for Integrative Sciences and Engineering.