Prohibitins and the functional compartmentalization of mitochondrial membranes

Two decades have passed since the discovery of the first member of a remarkably conserved family of membrane proteins now known as prohibitins (McClung et al., 1989). Since then, members of this family have been identified in virtually all species, with two closely related proteins being expressed in each case (Nijtmans et al., 2002; Merkwirth and Langer, 2009). The absence of prohibitins leads to severe phenotypes and deficiencies in higher eukaryotes, indicating that these proteins have fundamentally important functions (Artal-Sanz et al., 2003; He et al., 2008; Merkwirth et al., 2008; Park et al., 2005). However, how prohibitins act at the molecular level has remained poorly understood, and even the subcellular localization of prohibitins has been controversial. Recent studies have now significantly advanced our understanding of the cellular role of prohibitins. Mitochondria have been identified as the predominant site of action of these proteins. In addition, high-molecular-weight complexes composed of multiple prohibitin subunits have been defined as the physiologically active structure in mitochondria, and the main downstream targets of prohibitin function have been characterized.

In this Commentary, we review the versatile cellular processes that have been linked to prohibitins, with a focus on their role in mitochondria. We also discuss a model in which prohibitins ensure mitochondrial membrane organization and functionality by serving as scaffolds for both proteins and lipids.

Prohibitins comprise two evolutionarily conserved proteins, prohibitin-1 (PHB1) and prohibitin-2 (PHB2), which share more than 50% identical amino acid residues. Both proteins are present in organisms of all phylogenetic kingdoms. PHB1 was the first of the prohibitins to be discovered, and was identified on the basis that the expression of the gene was higher in normal cells compared with regenerating liver cells. Microinjection of Phb1 mRNA into mouse embryonic fibroblasts (MEFs) prohibited cell-cycle progression, which gave rise to the name prohibitin (McClung et al., 1989). Although the antiproliferative activity of PHB1 was later attributed to the 3′ untranslated region of its mRNA, the name remained (Jupe et al., 1996a; Jupe et al., 1996b). PHB2 was isolated a few years later – together with PHB1 – as an interaction partner of the IgM receptor in mouse B cells, leading to the alternative names B-cell receptor-associated protein of 32 kDa (BAP32) and BAP37 for PHB1 and PHB2, respectively (Terashima et al., 1994).

Since these discoveries were made, a set of diverse functions has been attributed to both PHB1 and PHB2, opening up the possibility that prohibitins are multifunctional proteins. Prohibitins have been implicated in transcriptional regulation (Kurtev et al., 2004; Wang et al., 2002a) – giving rise to yet another name for PHB2, repressor of estrogen receptor activity (REA) (Montano et al., 1999) – as well as in the regulation of sister-chromatid cohesion (Takata et al., 2007), cellular signaling (Rajalingam et al., 2005), apoptosis (Fusaro et al., 2003), mitochondrial biogenesis and maintenance of mitochondrial DNA (Kasashima et al., 2008). It should be noted that these processes occur in different cellular compartments – the nucleus, the plasma membrane and mitochondria (Box 1). However, increasing evidence suggests that pleiotropic phenotypes that are observed in the absence of prohibitins reflect a dysfunction of mitochondria.

Prohibitins have been found to localize to mitochondria in all cell types examined to date. In vivo immunofluorescence studies, immunogold labeling and biochemical subcellular-fractionation experiments in various cell types and in different organisms have been used to identify prohibitins as integral membrane proteins of the mitochondrial inner membrane (Artal-Sanz et al., 2003; Berger and Yaffe, 1998; Coates et al., 1997; Ikonen et al., 1995; Snedden and Fromm, 1997).

PHB1 and PHB2 are essential for cell proliferation and embryonic development in mice. Mouse embryos lacking either the Phb1 or Phb2 gene fail to develop beyond embryonic day 8.5 (He et al., 2008; Merkwirth et al., 2008; Park et al., 2005), and depletion of PHB1 or PHB2 impairs the proliferation of endothelial cells and MEFs (Merkwirth et al., 2008; Schleicher et al., 2008). These findings are in striking contrast to the previously proposed anti-proliferative role of prohibitins and their predicted function as a negative regulator of E2F-mediated transcription (as described in Box 1) (Wang et al., 1999a). The expression of PHB2 with mutations in its putative nuclear-localization sequences did not interfere with the proliferation of prohibitin-deficient MEFs (Merkwirth et al., 2008). By contrast, cells expressing PHB2 variants with mutations in residues that are essential for mitochondrial targeting were unable to maintain cell growth, indicating that the mitochondrial localization of PHB2 is essential for cell proliferation (Kasashima et al., 2006; Merkwirth et al., 2008). Whether the same holds true for PHB1 has yet to be determined.

Although compelling evidence indicates that prohibitins have a predominantly mitochondrial function, they might serve functions outside mitochondria under certain conditions or in certain cell types (Box 1). For example, PHB2 was shown to translocate from mitochondria to the nucleus following the binding of estradiol by ERα and following exposure of HeLa cells to capsaicin, the active component of chilli peppers, which binds to PHB2 but not to PHB1 (Kasashima et al., 2006; Kuramori et al., 2009). These studies suggest that prohibitins have non-mitochondrial functions, depending on regulatory cues. However, further evidence to support the relocalization of prohibitins to other cellular compartments, and for the underlying molecular mechanism of this relocalization, is still needed, because protein export from mitochondria has not been observed previously under non-apoptotic conditions.

The function of prohibitins in mitochondria has been linked to respiration and the stability of mitochondrial DNA (mtDNA) and, additionally, to the morphology of mitochondria. Crosslinking studies identified PHB1 and PHB2 as peripheral components of mitochondrial nucleoids, which are nucleoprotein complexes that contain mtDNA (Bogenhagen et al., 2008; Bogenhagen et al., 2003; Wang and Bogenhagen, 2006). Proteins incorporated in these nucleoid complexes, such as mitochondrial transcription factor A (TFAM) and mitochondrial single-stranded DNA-binding protein (mtSSB), regulate the stability, packaging, replication, transcription and maintenance of mtDNA (Chen and Butow, 2005). Remarkably, downregulation of PHB1 expression in HeLa cells affects the organization of mitochondrial nucleoids and the steady-state level of TFAM (Kasashima et al., 2008). These findings suggest that PHB1 maintains the organization and copy number of mtDNA by regulating TFAM stability (Kasashima et al., 2008).

However, depletion of prohibitins in various cell lines does not significantly affect (or at least does not completely abolish) cellular respiration, indicating that prohibitins are not essential for the maintenance of the mitochondrial genome. Knockdown of Phb1 expression in epithelial cells induces cellular senescence (Schleicher et al., 2008). This was shown to be due to an inhibition of complex I of the mitochondrial electron transport chain by a mechanism that is not currently understood at the molecular level but that leads to an increased production of reactive oxygen species (ROS) and the depolarization of mitochondrial membrane potential (Schleicher et al., 2008). Similarly, prohibitin-deficient yeast cells display a reduced replicative lifespan that is associated with age-dependent mitochondrial degeneration (Coates et al., 1997; Wang et al., 2008). This phenotype is suppressed if cytosolic protein expression is reduced, which might prevent the accumulation of non-assembled proteins in the mitochondrial inner membrane that, in turn, triggers proton leakage (Wang et al., 2008). Notably, respiration and the assembly of respiratory supercomplexes was not impaired in Phb2^–/– MEFs (Merkwirth et al., 2008). It remains to be determined whether variable effects of the loss of prohibitins on respiratory activities reflect differences in experimental conditions, because different approaches have been used to deplete cells of prohibitins. Alternatively, cell-type-specific differences in how prohibitins affect the stability of mtDNA and the biogenesis of the respiratory chain might exist.

Independent of their role in respiration, prohibitins regulate the morphology of mitochondria. Abnormal mitochondria accumulate in the nematode Caenorhabditis elegans after RNA-interference (RNAi)-mediated depletion of prohibitins or in prohibitin-deficient yeast cells (Artal-Sanz et al., 2003; Berger and Yaffe, 1998; Osman et al., 2009). These findings were corroborated by studies in mammalian cells: RNAi-mediated knockdown of either Phb1 or Phb2 expression in HeLa cells caused fragmentation of the mitochondrial network (Kasashima et al., 2006), which is composed of highly interconnected tubules formed by balanced fusion and fission events (Hoppins et al., 2007). Similarly, fragmented mitochondria were observed in prohibitin-deficient MEFs, which suggests that the fusion of mitochondrial membranes is impaired in the absence of prohibitins (Merkwirth et al., 2008). In addition, prohibitins are required for cristae morphogenesis, as revealed by an ultrastructural analysis of mitochondria in prohibitin-deficient MEFs (Merkwirth et al., 2008). A disturbed cristae morphology might facilitate the release of cytochrome c from the intracristal space and thereby explain the increased sensitivity of prohibitin-deficient MEFs to apoptotic stimuli.

The aberrant mitochondrial morphology observed in the absence of prohibitins can be explained by an altered processing of OPA1 (Merkwirth et al., 2008), a large dynamin-like GTPase that is found in the mitochondrial intermembrane space and that regulates both mitochondrial fusion and cristae morphogenesis (Hoppins et al., 2007). Mutations in OPA1 cause degeneration of retinal ganglion cells in autosomal dominant optic atrophy (Alexander et al., 2000; Delettre et al., 2000). Proteolytic processing of OPA1 splice variants, which are expressed in a tissue-specific manner (Akepati et al., 2008; Delettre et al., 2001), results in the accumulation of long and short OPA1 isoforms (Delettre et al., 2001; Ishihara et al., 2006; Satoh et al., 2003). Strikingly, long OPA1 isoforms are destabilized in prohibitin-deficient MEFs, indicating accelerated processing and degradation of OPA1 (Merkwirth et al., 2008). The importance of prohibitins for normal OPA1 processing explains the mitochondrial morphology defects observed in Phb2^–/– cells, because mitochondrial fusion depends on both long and short OPA1 isoforms (Ishihara et al., 2006; Song et al., 2007). Indeed, the ectopic expression of a non-cleavable long OPA1 isoform restored a tubular mitochondrial network, cristae morphogenesis and apoptotic resistance of Phb2^–/– MEFs (Merkwirth et al., 2008). These experiments thus identify OPA1 processing as the key process that is regulated by prohibitins in MEFs.

The mechanism by which prohibitins affect OPA1 processing remains to be determined. Prohibitins assemble with the ATP-dependent m-AAA protease in large supercomplexes in the inner membrane of yeast mitochondria and modulate the proteolysis of non-assembled membrane proteins by this protease (Steglich et al., 1999). Mammalian m-AAA proteases have been linked to OPA1 processing in cellular systems (Ishihara et al., 2006) and are indeed able to cleave OPA1 variants when they are expressed in yeast (Duvezin-Caubet et al., 2007). Accordingly, prohibitins might regulate proteolytic processing of OPA1 by the m-AAA protease.

Interestingly, a protein that is distantly related to prohibitins, known as stomatin-like protein 2 (SLP-2), has been identified in the mitochondrial inner membrane of mammalian cells (Da Cruz et al., 2008; Da Cruz et al., 2003). SLP-2 directly interacts with and stabilizes prohibitins (Da Cruz et al., 2008). Similarly to prohibitins, the function of SLP-2 has recently been linked to mitochondrial fusion and the processing of OPA1. Crosslinking studies led to the identification of the GTPase mitofusin-2 (MFN2), a central component of the fusion machinery in the mitochondrial outer membrane (Hoppins et al., 2007), as an interaction partner of SLP-2 (Hajek et al., 2007). SLP-2, but not PHB2, is required for mitochondrial hyperfusion, which is induced in response to stress conditions including UV irradiation or low concentrations of cycloheximide (Tondera et al., 2009). Depletion of SLP-2 under these stress conditions results in the destabilization of long OPA1 isoforms (Tondera et al., 2009), which resembles the result obtained when prohibitins are depleted under normal conditions. Therefore, this suggests that a complex network of related proteins regulates OPA1 processing and mitochondrial fusion.

The finding that a plethora of functions are associated with prohibitins raises the intriguing issue of how prohibitins mediate these diverse cellular processes at the molecular level. The analysis of the native structure of PHB1 and PHB2 in yeast, nematodes and mammals has revealed that both proteins are present in a high-molecular-weight complex (∼1.2 MDa) in the inner membrane of mitochondria (Artal-Sanz et al., 2003; Nijtmans et al., 2000; Steglich et al., 1999; Tatsuta et al., 2005). Co-immunoprecipitation experiments in both yeast and mammalian cells provide support that PHB1 and PHB2 physically interact (He et al., 2008; Ross et al., 2008; Steglich et al., 1999). The function of both prohibitin subunits is interdependent in various organisms (Artal-Sanz et al., 2003; Berger and Yaffe, 1998; Kasashima et al., 2006; Merkwirth et al., 2008). The absence of either prohibitin does not affect the expression of the other, but results in its degradation. These observations substantiate immunodepletion experiments that show that all PHB1 and PHB2 subunits are present in a complex with the respective other subunit, never as free proteins (Coates et al., 2001). Thus, the mitochondrial prohibitin complex (PHB complex) represents the physiologically active structure and, because the stabilities of PHB1 and PHB2 depend on one another, similar phenotypes are expected in the absence of either protein. In the light of these findings, previous reports that have proposed distinct functions of PHB1 or PHB2 in the nucleus or at the plasma membrane (Box 1) should be revisited.

The assembly of PHB complexes in the mitochondrial inner membrane has been investigated in detail in the yeast Saccharomyces cerevisiae. The analysis of purified yeast PHB complexes by single-particle electron microscopy revealed a ring-like structure with an outer diameter of about 200-250 Å (Tatsuta et al., 2005) (Fig. 1A). Crosslinking experiments detected interactions exclusively between Phb1 and Phb2 subunits and did not detect any homodimeric crosslink adducts, suggesting that the ring complexes are built up of multiple Phb1 and Phb2 subunits that alternate with each other (Back et al., 2002) (Fig. 1B). Phb1 and Phb2 are targeted to mitochondria by means of unconventional N-terminal targeting sequences, which are not cleaved upon mitochondrial import (Fig. 1C). Insertion of N-terminal hydrophobic sequences into the inner membrane is mediated by the Tim23 translocase and facilitated by the Tim8-Tim13 complex, a chaperone complex in the mitochondrial intermembrane space. The C-terminal coiled-coil domains of Phb1 and Phb2 (Fig. 1C), both of which expose large domains to the intermembrane space, are required for the subsequent oligomerization into large complexes. Initially, intermediate complexes of ∼120 kDa that probably represent hetero-oligomeric tetramers of Phb1 and Phb2 subunits are formed. These intermediates act as building blocks for the formation of large ring-like assemblies (Back et al., 2002).

The fact that PHB complexes have large dimensions and form ring-like structures stimulated different proposals on the molecular function of prohibitins. The formation of ring complexes is reminiscent of chaperonins, which are molecular-chaperone proteins that mediate protein folding. On the basis of the finding that prohibitins interact with unassembled respiratory-chain subunits, and that there is limited sequence similarity between prohibitins and chaperonins (Nijtmans et al., 2002; Nijtmans et al., 2000), a chaperone or holdase function for prohibitins has been proposed. Although the absence of prohibitins has not been found to cause general defects in the assembly of respiratory-chain complexes in various organisms, it has been observed that non-assembled subunits of the respiratory-chain complex are more rapidly degraded by the m-AAA protease in the absence of prohibitins (Nijtmans et al., 2002; Steglich et al., 1999). It should be noted, however, that this might also be due to the fact that m-AAA proteases have altered proteolytic activity in the absence of prohibitins. Moreover, the dimensions of PHB ring complexes substantially exceed those of chaperonins, and the sequence similarity between prohibitins and chaperonins is limited and is confined to a hinge region located between the ATPase and substrate-binding domains – that is, to a region that is not directly involved in substrate interaction. Therefore, a function of prohibitins as chaperones or holdases remains speculative.

An alternative but not mutually exclusive model suggests that prohibitins have a scaffolding function in the mitochondrial inner membrane. Accordingly, prohibitins might be required for inner-membrane organization and, for instance, for the recruitment of m-AAA proteases to specific functional sites. Such a view is supported by the similarities between prohibitins and a large family of distantly related membrane proteins, the SPFH family (Tavernarakis et al., 1999), which also includes mitochondrial SLP-2 (Box 2). A number of SPFH proteins have been found in defined membrane domains, suggesting that they act as scaffold proteins. The finding that SPFH-family members form high-molecular-weight complexes with a similar structural organization to those formed by prohibitins suggests that these proteins serve similar functions in distinct cellular membranes (Browman et al., 2007; Morrow and Parton, 2005). A role of prohibitins as scaffold proteins is also consistent with recent genetic evidence that links their function to mitochondrial phospholipid metabolism in yeast.

Although prohibitins are essential in higher eukaryotes (He et al., 2008; Park et al., 2005), yeast cells deficient for prohibitins are viable but show a decreased replicative lifespan (Coates et al., 1997). However, a number of genes have been identified as essential for the survival of prohibitin-deficient yeast cells in various studies (Berger and Yaffe, 1998; Birner et al., 2003; Osman et al., 2009; Osman et al., 2007; Steglich et al., 1999). A systematic genetic survey of ∼4,600 yeast strains lacking non-essential genes identified 35 genes that were essential in the absence of prohibitins and were therefore termed genetic interactors of prohibitins (GEP) genes (Osman et al., 2009). The majority of these genes encode mitochondrial proteins, highlighting the major role of prohibitins in mitochondria. Strikingly, GEP genes fall into distinct functional groups, including genes that control respiratory-chain assembly and those required for mitochondrial morphology and the assembly of β-barrel proteins. ATP10, ATP23 and OXA1 – three genes that are specifically required for the assembly of the membrane-embedded F_o-part of the mitochondrial ATP synthase (Jia et al., 2007; Tzagoloff et al., 2004; Zeng et al., 2007) – are among the synthetic lethal interactors of prohibitins, indicating that an impaired assembly of the F_o-part is deleterious in the absence of prohibitins (Osman et al., 2007). The enzymes that mediate the terminal steps in phosphatidylethanolamine (PE) and cardiolipin (CL) biosynthesis, Psd1 and Crd1, are essential for the survival of yeast cells in the absence of prohibitins (Birner et al., 2003; Osman et al., 2009).

The analysis of two GEP genes, GEP1 and UPS1, provided further insight into the molecular function of prohibitins. Gep1 and Ups1 are members of the conserved but largely uncharacterized MSF1′/PRELI protein family (Dee and Moffat, 2005; Sesaki et al., 2006). Whereas Ups1 has been linked to processing of the OPA1 homolog Mgm1 in yeast (Sesaki et al., 2006), the function of Gep1 is unknown. Strikingly, overexpression of the phosphatidylserine (PS) synthase Cho1 restores the growth of Δgep1Δphb1 cells, suggesting that increased levels of PS, the precursor for PE, were beneficial in these cells (Osman et al., 2009). Further experiments unraveled an essential role for Gep1 and Ups1 in promoting normal PE and CL levels within mitochondria, respectively, corroborating the functional link of prohibitins to the metabolism of mitochondrial phospholipids. Indeed, the levels of mitochondrial PE and/or CL are reduced in the majority of yeast strains lacking GEP genes (Osman et al., 2009) (Fig. 2). Downregulation of prohibitins in cells with low PE levels causes severe disturbances of the mitochondrial inner membrane and ultimately results in the dissipation of the mitochondrial membrane potential and cell death (Osman et al., 2009). It can therefore be concluded that prohibitins are essential for the integrity of the mitochondrial inner membrane if the levels of PE and CL are reduced.

The characterization of the genetic interactome of prohibitins in yeast suggests that prohibitins and the membrane lipids CL and PE contribute to similar processes within the mitochondrial inner membrane that are essential for cell survival. Alterations in CL levels or in the composition of the fatty-acid chain of CL are associated with many pathophysiological states (Chicco and Sparagna, 2007; Joshi et al., 2008). CL ensures the optimal activity of various inner-membrane proteins, the stability of respiratory-chain complexes (Schlame, 2008; Wenz et al., 2009) and modulates apoptosis (Choi et al., 2007; Gonzalvez et al., 2008). Interestingly, PE and CL are related phospholipids and share the tendency to form non-bilayer, hexagonal phases; these play roles in membrane contact zones during fusion and fission processes and in the transmembrane movement of proteins (Cullis and de Kruijff, 1979; Dowhan, 1997; Schlame et al., 2000; van den Brink-van der Laan et al., 2004). The physiological relevance of these biophysical similarities is highlighted by the observation that yeast or Escherichia coli cells with dramatically decreased levels of PE and CL are inviable (Gohil et al., 2005; Rietveld et al., 1993). Both PE and CL tend to aggregate and form defined lipid clusters in the plasma membrane of bacteria (Kawai et al., 2004; Matsumoto et al., 2006; Nishibori et al., 2005). These clusters might play important roles in bacterial cell division and sporulation, and might serve as platforms for the recruitment of soluble proteins that bind to lipid head groups or to proteins with transmembrane segments that preferentially localize in these clusters (Matsumoto et al., 2006). It seems likely that this holds true also for the mitochondrial inner membrane, and that a lateral segregation of PE and CL in the inner membrane is crucial for mitochondrial processes.

We propose that the scaffold function of PHB complexes supports the formation of such functional microdomains within the mitochondrial inner membrane (Fig. 3A). When levels of PE and CL are not limiting, the function of prohibitins is dispensable in yeast, and an asymmetric lipid distribution can be maintained in the membrane owing to the distinct biophysical properties of different phospholipids. However, prohibitin function becomes essential in yeast if PE or CL levels decrease. PHB complexes might support the formation of PE or CL patches under these conditions and locally enrich these phospholipids within the inner membrane. Clearly, further experimental support for a function of prohibitins as lipid scaffolds is needed, including evidence for a direct interaction between phospholipids and PHB complexes, or for an asymmetric lipid distribution in the inner membrane. It should be noted, however, that members of the SPFH family of proteins (Box 2) associate with defined membrane domains and interact directly with specific phospholipids (Browman et al., 2007; Huber et al., 2006). For example, flotillins, stomatins and podocin in other membranes cofractionate with detergent-resistant lipid rafts composed of cholesterol and sphingolipids (Bickel et al., 1997; Schwarz et al., 2001; Snyers et al., 1999). However, mitochondrial membranes contain only low amounts of cholesterol and sphingolipids, meaning that a similar lipid composition of prohibitin-associated membrane is unlikely. Moreover, a recent study did not provide any evidence for the existence of detergent-resistant lipid rafts in mitochondrial membranes (Zheng et al., 2009). It is therefore an attractive possibility that different SPFH proteins stabilize membrane domains that differ in their phospholipid compositions. The functional link between prohibitins and CL and PE suggests that prohibitins assist in the formation of membrane domains that are specifically enriched in CL and PE.

The interaction between PHB complexes and m-AAA proteases indicates that prohibitins can also act as protein scaffolds (Steglich et al., 1999) (Fig. 3B). m-AAA proteases might be recruited to distinct membrane domains whose lipid composition facilitates the degradation or processing of specific substrates or activates the proteolytic activity. The sequestration of m-AAA proteases by PHB complexes could modulate the access of proteolytic substrates to these membrane domains. In agreement with this scenario, loss of prohibitins leads to an increased turnover of non-assembled membrane proteins by the m-AAA protease in yeast (Steglich et al., 1999). Conversely, a function for PHB complexes as protein scaffolds could exclude proteins from specific areas of the membrane and thereby allow the formation of protein-free areas in the mitochondrial inner membrane, which is considered to be the most protein-rich membrane in the cell (Fig. 3C). This fence-like function of prohibitins might be of particular importance for membrane fusion events, which are known to depend on non-bilayer lipids (Dowhan, 1997).

A role of prohibitins as membrane scaffolds suggests that a defined spatial membrane organization is required to maintain various mitochondrial activities. The assembly of m-AAA proteases with PHB complexes suggests that distinct membrane domains have a role in proteolytic processes. Indeed, the processing of Mgm1 and the morphogenesis of cristae were identified as processes that require CL and PE within mitochondria (Osman et al., 2009; Sesaki et al., 2006). Consistently, Mgm1 binds anionic phospholipids including CL in vitro (Meglei and McQuibban, 2009). PHB complexes acting as membrane scaffolds might recruit proteases, or ensure a specific lipid environment that facilitates Mgm1 processing, and thereby prevent mitochondrial fragmentation under conditions of limited CL or PE. The accelerated processing of OPA1 in Phb2^–/– MEFs might result from a similar lipid dependence and impaired inner-membrane organization (Merkwirth et al., 2008). Similarly, the importance of CL for OPA1 cleavage could explain the disturbed mitochondrial morphology in cells lacking CL synthase or tafazzin, an enzyme involved in cardiolipin remodeling that is mutated in Barth syndrome (Choi et al., 2007; Claypool et al., 2008; Schlame, 2008). Differences in PE and CL levels in the mitochondrial membranes of different species might also provide an explanation for the puzzling observation that prohibitins are dispensable in yeast but essential in vertebrates. Alternatively, prohibitins might have acquired additional functions in higher eukaryotes. It is of interest in this context that the processing of OPA1, which has been identified as the key activity regulated by prohibitins in MEFs, is probably mediated by the m-AAA protease in mammalian cells (Ishihara et al., 2006). By contrast, the OPA1 homolog Mgm1 in yeast is cleaved by the rhomboid protease Pcp1 (Freeman, 2008; McQuibban et al., 2003).

Mitochondria-dependent apoptosis is another process that is influenced by the phospholipid composition of mitochondrial membranes and might depend on the scaffolding function of PHB complexes. CL provides an anchor and activating platform for caspase-8 at the mitochondrial surface (Gonzalvez et al., 2008), and affects Bax insertion and activation in liposomes in vitro (Lucken-Ardjomande et al., 2008). Depletion of CL releases cytochrome c from the mitochondrial inner membrane and accelerates stimuli-elicited apoptosis (Choi et al., 2007). It is therefore possible that a disturbed inner-membrane organization in the absence of PHB complexes facilitates cytochrome-c release and increases the sensitivity towards apoptotic stimuli in Phb2^–/– MEFs (Merkwirth et al., 2008).

The assembly of the respiratory chain or of the F₁F_o-ATP synthase might be other processes that rely at least partially on the formation of prohibitin-assisted membrane domains. This is supported by the synthetic lethal interaction of yeast prohibitins with a number of known assembly factors of the respiratory chain (Osman et al., 2009). The assembly of the F_o-part of ATP synthase seems to be particularly sensitive to the loss of prohibitins (Osman et al., 2007). Assembly intermediates that accumulate in prohibitin-deficient yeast cells might disrupt the integrity of the inner membrane, resulting in proton leakage and cell death. Similarly, the increased generation of ROS and the aging phenotype in endothelial cells or in yeast (Coates et al., 1997; Schleicher et al., 2008; Wang et al., 2008) could be explained by a compromised assembly of respiratory complexes in the absence of both prohibitins and a defined membrane organization.

The emerging role of prohibitins as protein and lipid scaffolds highlights the importance of spatial membrane organization for normal mitochondrial function. The inner membrane, which is probably the most protein-rich membrane of the cell, has a complex ultrastructure, and increasing evidence suggests a functional compartmentalization by macromolecular complexes. Interactions between components of protein translocases in the outer and inner membrane (TOM and TIM complexes, respectively) allow the coordinated transport of nuclear-encoded proteins across both mitochondrial membranes (Bolender et al., 2008; Reichert and Neupert, 2002). Supercomplexes between respiratory-chain complexes are thought to increase the efficiency of oxidative phosphorylation by promoting substrate channeling (Acin-Perez et al., 2008; Schägger and Pfeiffer, 2000; Strauss et al., 2008). Prohibitin ring complexes might provide yet another means for inner-membrane organization and define functional membrane domains by acting as a scaffold for both proteins and lipids. The finding that distantly related membrane proteins of the SPFH family are present in other cellular membranes indicates that protein-assisted formation of functional membrane domains might be a novel organizing principle of general relevance. However, a challenging task for future studies will be to provide direct evidence for the existence of prohibitin-dependent membrane domains. It has been difficult thus far to unambiguously prove the existence of lipid microdomains in any membrane, because microdomains are believed to be extremely small and short-lived, and therefore be near the limits of detection using currently available methods. New, more powerful microscopic methods might allow visualization of an asymmetric lipid distribution in mitochondrial membranes in the future and open up new possibilities for further defining the role of prohibitins in organizing the mitochondrial inner membrane.