JXB Advance Access originally published online on March 23, 2006
Journal of Experimental Botany 2006 57(6):1225-1243; doi:10.1093/jxb/erj151
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
REVIEW ARTICLE |
The mitochondrial compartment
School of Biology, Sir Harold Mitchell Building, University of St Andrews, St Andrews, Fife KY16 9TH, Scotland, UK
* To whom correspondence should be addressed. E-mail: david.logan{at}st-andrews.ac.uk
Received 21 November 2005; Accepted 6 February 2006
 |
Abstract |
|---|
Top
Abstract
IntroductionMitochondria are vital organelles that perform a variety of fundamental functions ranging from the synthesis of ATP through to being intimately involved in programmed cell death. Comprised of at least six compartments: outer membrane, inner boundary membrane, intermembrane space, cristal membranes, intracristal space, and matrix, mitochondria have a complex, dynamic internal structure. This internal dynamism is reflected in the pleomorphy and motility of mitochondria. Mitochondria contain their own DNA (mtDNA), encoding a small number of vital genes, but this role as a genetic vault is not compatible with the role of mitochondria in bioenergetics since electron transport results in the generation of reactive oxygen species (ROS) that induce lesions in the mtDNA. It is hypothesized that ROS shape the morphological organization of the higher plant cell mitochondrial population into a discontinuous whole, and that ROS are a selective pressure affecting the organization of the mitochondrial genome. This review describes how inter- and intra-mitochondrial compartmentalization underpins the biology of this complex organelle.
Key words: Cytoskeleton, discontinuous whole, division, dynamics, fusion, mitochondria, mitochondrial genome, morphology, mutants, ultrastructure
|
Introduction |
|---|
Mitochondria are highly dynamic, pleomorphic organelles composed of a smooth outer membrane surrounding an inner membrane of significantly larger surface area that, in turn, surrounds a protein-rich core, the matrix. Although mitochondria contain their own genome and protein-synthesizing machinery (Leaver et al., 1983
; Unseld et al., 1997; Gray et al., 1999) they are only semi-autonomous: the majority of mitochondrial polypeptides are encoded in the nuclear genome, synthesized in the cytosol and imported into the mitochondria post-transcriptionally (Unseld et al., 1997; Whelan and Glaser, 1997; Duby and Boutry, 2002). The role of the mitochondrion in the synthesis of ATP formed by oxidative phosphorylation is well established (Saraste, 1999) and, in addition, mitochondria are involved in numerous other metabolic processes including the biosynthesis of amino acids, vitamin cofactors, fatty acids, and iron-sulphur clusters (Mackenzie and McIntosh, 1999; Bowsher and Tobin, 2001). Apart from the role of the mitochondrion in ATP synthesis and various biosynthetic pathways the mitochondrion is one of three cell compartments involved in photorespiration (Douce and Neuburger, 1999), is implicated in cell signalling (Vandecasteele et al., 2001; Logan and Knight, 2003), and has been shown recently to be involved in programmed cell death (Jones, 2000; Youle and Karbowski, 2005). This review deals with the complex biology of the mitochondrion and describes how various levels of compartmentalization within the mitochondrion and cellular mitochondrial population as a whole (the chondriome) underpin the multiple functions of this vital organelle. Although focused on the higher plant mitochondrial compartment, frequent reference will be made to studies using non-plant model organisms. In some cases, this is simply due to a paucity of information about specific aspects of plant mitochondrial biology; in all cases it is because I believe the information is of fundamental relevance. A short article such as this can only provide a brief overview of the importance of compartmentalization to the life of the mitochondrion. A great deal has been left out (e.g. co-ordination of the mitochondrial and nuclear genomes, control of protein import, the mitochondrial proteome, biochemical defence against ROS, amongst other topics) and it is possible, even likely, that my choices of topics to include might not be of interest to all with an interest in mitochondria but, in the end, this is a personal view of the mitochondrial compartment.
|
Compartmentalization and the chemiosmotic theory |
|---|
The vast majority of biological energy (ATP) production is associated with energy-transducing membranes: the prokaryotic plasma membrane of bacteria and blue-green algae, the thylakoid membranes of chloroplasts, and the inner mitochondrial membrane. The energy-transducing membrane is central to the chemiosmotic theory that explains the basic mechanism of biological energy production, whereby ATP production is coupled to the controlled dissipation of a proton electrochemical gradient (proton motive force). The membrane allows compartmentalization of protons, via their vectorial transport across the membrane, by the action of a primary proton pump(s). In mitochondria the primary proton pumps comprise complexes I, III, and IV. These primary pumps generate a high gradient of protons that forces a secondary pump (the ATP synthase complex) to reverse, energized by the flow of protons ‘downhill’, thereby synthesizing ATP from ADP and Pi. Any proton leak across the membrane would cause a short-circuit, destroy the compartmentalization of protons and uncouple the proton motive force from the ATP synthase. The energy-transducing membrane must, therefore, be essentially closed and have a high resistance to proton flux.
The energy-transducing membrane of mitochondria, the inner mitochondrial membrane, is a highly pleomorphic structure. Although there are an almost endless variety of inner mitochondrial membrane morphologies in mitochondria from different species, from different cell types within the same species or from the same cell types but in different metabolic states (Munn, 1974), some generalizations can be made. Transmission electron microscopy led to the development of models of the internal structure of mitochondria. Palade's model (Palade, 1952), also called the baffle model, depicted the invaginations of the inner mitochondrial membrane, the cristae, as random, wide in-folds of the membrane (the typical text book image, Fig. 1) while Sjostrand suggested the cristae were composed of a stack of independent membranous lamellae (Sjostrand, 1953). It is clear from two ground-breaking research papers published in 1994 (Lea et al., 1994; Mannella et al., 1994), describing results obtained using high-resolution scanning electron microscopy or electron tomography, respectively, together with subsequent investigations, that neither model was entirely correct (Mannella, 2006).
|
The results obtained using advanced tomographic imaging techniques demonstrate that, at least in animal tissue, tubular rather than lamellar cristae predominate and that the morphology of cristae infers that they are structurally distinct from the rest of the inner mitochondrial membrane. An additional finding was confirmation that the cristae were connected to the inner boundary membrane (cortical inner mitochondrial membrane, parallel to the outer membrane) by membranous tubules, instead of the cristae being simple in-folds of the membrane as suggested by Palade (1952)
. Daems and Wisse (1966) first reported that cristae attach to the inner boundary membrane via narrow tubules termed pediculi, but this finding was not consistent with the baffle paradigm. Subsequently, it has been shown that the connections between the cristae and the inner boundary membrane, the term crista junction has superseded pediculi, have a preferred size and morphology and are independent of the source of the mitochondrion and the means of fixation (Mannella et al., 1994, 1997; Perkins et al., 1997a, b, 1998). Indeed, it has been proposed that crista junctions are a uniform structural component of all mitochondria (Perkins and Frey, 2000). For example, in rat liver mitochondria, crista junctions are 30–50 nm long although tubules three times that length have been measured, and in Neurospora crassa the slot-like crista junctions have been measured at up to 200 nm, although the average length is 30–40 nm (Frey et al., 2002; Perkins et al., 1997a).
The number of crista junctions and the morphology of the intercristal space have been shown to change with the metabolic state of the mitochondria (Hackenbrock, 1968; Mannella et al., 1994, 1997). In the orthodox state, corresponding to partial matrix expansion, the intercristal space is compressed and tubular with few cristae interconnections and one or two crista junctions with the inner boundary membrane. In the condensed state, corresponding to partial matrix contraction, the intercristal spaces are dilated and there are more numerous intercristal membrane connections and crista junctions. Hackenbrock (1968) demonstrated, by rapid fixation of purified mouse liver mitochondria in different respiratory steady-states, that mitochondria in state 3 (maximum respiratory rate in the presence of excess ADP and respiratory substrate) were in the condensed conformation, but reverted to an orthodox morphology after entering state 4 respiration (characterized by a reduction in respiration due to the depletion of ADP). Addition of ADP to these mitochondria caused a reversion to the condensed form within 35 s, followed by a gradual return to the orthodox conformation as all the ADP is phosphorylated.
Dry, quiescent maize embryos contain mitochondria with little internal membrane structure and an electron-light matrix (Logan et al., 2001). Upon imbibition, mitochondrial biogenesis is stimulated and within 24 h (protrusion of the radicle typically took place after 36–48 h imbibition) mitochondria in the embryo have a normal, orthodox, conformation (Fig. 2; Logan et al., 2001). By contrast, mitochondria isolated from germinated embryos (after 48 h imbibition) had a condensed conformation (Fig. 2; Logan et al., 2001). It is tempting to speculate that the switch from an orthodox to a condensed conformation during mitochondrial biogenesis is indicative of the changing biochemistry of the organelle as it switches from being reliant on the provision of electrons from external NADH dehydrogenases to the newly assembled TCA cycle (Logan et al., 2001).
|
A condensed morphology, large intercristal spaces with narrow crista junctions to the intermembrane space, has been shown by computer simulation to lead to a reduction in diffusion of ADP into the cristae, reduction in the transport of ADP across the inner mitochondrial membrane and, therefore, ATP production (Mannella, 2006
). Adoption of an orthodox confirmation when the bulk ADP concentration is low might therefore act to minimize the negative effect on ATP production of limited diffusion of ADP through the crista junctions by concentrating the ADP within a smaller intercristal volume. The results of Hackenbrock (1968) and those from the computer simulation suggest that inner mitochondrial membrane remodelling, which affects the degree of compartmentalization, is a mechanism enabling the control of ATP production by mediating ADP availability (Mannella, 2006). Whether this control mechanism operates in vivo remains to be determined. What is clear from the above discussion is that at least six discrete mitochondrial compartments can be recognized on a structural basis: outer membrane, intermembrane space, inner boundary membrane, cristal membrane, intercristal space, and matrix. |
Biogenesis of the cristal membranes is dependent on ETC biogenesis |
|---|
The extent to which the structural organization and compartmentalization of the energy-transducing inner mitochondrial membrane to form three components (inner boundary membrane, cristal membranes and intercristal space) are reflected in, or indeed due to, a different protein complement of each compartment is not fully understood. It has been demonstrated recently, with bovine heart mitochondria, that approximately 94% of both Complex III and ATP synthase protein, as detected by immuno-gold labelling, resides in the cristal membrane, the remaining 6% is located in the inner boundary membrane (Gilkerson et al., 2003
). The authors concluded that there is restricted diffusion of respiratory complexes through the crista junctions and that the cristae comprise a regulated functionally distinct subcompartment of the inner mitochondrial membrane (Gilkerson et al., 2003). A similar compartmentalization of cytochrome c oxidase in the cristae has been recorded in Jerusalem artichoke (Kay et al., 1985; Moller et al., 1987) and rat cardiac muscle and pancreas (Perotti et al., 1983) mitochondria, and in the cristae and inner boundary membrane of mouse liver mitochondria (Hiraoka and Hirai, 1992). In addition, indirect evidence to support the hypothesis that the cristal membrane is the site of oxidative phosphorylation comes from examination of Rho0 cells that lack mitochondrial DNA (Gilkerson et al., 2000). Human mitochondrial DNA encodes 13 polypeptide components of the respiratory chain and, therefore, in Rho0 cells, the oxidative phosphorylation machinery is incompletely assembled. This selective loss of only a small proportion of respiratory complex subunits has a dramatic effect on the internal structure of the mitochondria: the cristal membranes are greatly reduced and disorganized, yet the inner boundary membrane remains visibly unaltered (Gilkerson et al., 2000). This specific effect on the cristal membranes can be explained if the cristal membranes are functionally distinct from the inner boundary membrane and are dependent on the correct biogenesis of the respiratory chain for their own biogenesis.
Two supernumerary F0-ATPase-associated subunits, g and Tim11p (also called e), that are not essential for growth in yeast and are restricted to mitochondria (Walker et al., 1991; Higuti et al., 1993; Collinson et al., 1994; Boyle et al., 1999), are involved in the dimerization of the F1F0-ATPase and cristae biogenesis and morphology (Paumard et al., 2002). However, although these subunits are conserved between yeast and mammals there are no significant homologues in Arabidopsis. In S. cerevisiae, absence of either subunit, g or Tim11p, results in the absence of cristae, although the inner boundary membrane is present (Paumard et al., 2002). A similar aberrant mitochondrial phenotype has been described in mutants of a large GTPase called Mgm1p (Wong et al., 2000), and it was proposed that Mgm1p is involved in inner membrane remodelling events in yeast (Wong et al., 2000). Subsequently, Mgm1p was identified independently by two groups (Herlan et al., 2003; McQuibban et al., 2003) as a substrate of a yeast rhomboid-type protease named Rbd1p (rhomboid) or Pcp1p (processing of cytochrome c peroxidase (Esser et al., 2002) and that cleavage of Rbd1p/Pcp1p regulates inner membrane remodelling (Herlan et al., 2003; McQuibban et al., 2003). Rbd1p/Pcp1p contains six transmembrane domains and is embedded in the inner mitochondrial membrane (McQuibban et al., 2003). Upon import of an Mgm1p precursor, the N-terminal hydrophobic region becomes tethered in the inner membrane at the site of the first transmembrane domain, by what is assumed to be a translocation-arrest mechanism, leaving the N-terminal mitochondrial targeting presequence exposed to the matrix (Herlan et al., 2003). Cleavage by the matrix-processing peptidase generates what is called the large isoform of Mgm1p (l-Mgm1p) (Herlan et al., 2003). Next, l-Mgm1p is further translocated into the matrix and the second transmembrane domain becomes inserted into the inner membrane, whereupon it undergoes further proteolytic cleavage by Rbd1p/Pcp1p producing a smaller isoform, s-Mgm1p, which is released into the intermembrane space and becomes associated with either the outer or inner mitochondrial membrane (Herlan et al., 2003). Both isoforms function in the maintenance of mitochondrial morphology and respiratory competence, but the mechanism controlling the ratio of l-Mgm1p to s-Mgm1p is unknown (Herlan et al., 2003). Recently, Amutha et al. (2004) integrated the Tim1p, Mgm1p, and Rbd1p/Pcp1p data by demonstrating that Mgm1p is an upstream regulator of Tim1p subunit stability, of the assembly of the F1F0-ATPase, and of cristae biogenesis. Homologues of Mgm1p and Rbd1p/Pcp1p genes are present in Arabidopsis: Mgm1p=members of the Arabidopsis dynamin-like gene family (Hong et al., 2003), the closest being DRP3B, At2g14120; Rbd1p/Pcp1p=At1g18600. At the time of writing, only DRP3B has been shown to be required for normal mitochondrial morphology (Arimura and Tsutsumi, 2002), but no information is available on the internal morphology of mitochondria in DRP3B mutants.
|
Contact sites |
|---|
Contact sites were first described by Hackenbrock (1968)
as specific regions where the outer membrane and inner boundary membrane are closely apposed, with no discernible space between them. It is now known that at least two types of contact site exist. One is as described by Hackenbrock, while in the second, the outer and inner boundary membranes are connected by bridge-like structures that maintain a constant separation between the membranes (Senda and Yoshinaga-Hirabayashi, 1998; Perkins et al., 2001). Senda and Yoshinaga-Hirabayashi (1998) suggested that the bridges might keep the outer and inner membranes apart thus maintaining the intermembrane space as a physically distinct compartment. The close apposition of the outer and inner boundary membranes as reported by Hackenbrock led him to suggest that these contact sites could facilitate the passage of solutes and small molecules between the cytosol and the matrix (Hackenbrock, 1968). Subsequently, it was demonstrated that translationally-arrested polysomes were selectively bound to the outer membrane surface at contact sites (Kellems et al., 1975) and that precursor proteins, trapped during translocation, were stuck within both outer and inner boundary membranes (Schleyer and Neupert, 1985; Schwaiger et al., 1987). Using chimeras composed of the N-terminal portion of a mitochondria-targeted precursor protein fused to a cytosolic protein which become trapped during translocation, Pon and colleagues were able to show that the partly translocated precursors are enriched at contact sites and that contact sites contain import activity (Pon et al., 1989). A similar approach, using arrested translocation intermediates, enabled the co-isolation of the translocase of the outer membrane (TOM) and the preprotein translocase of the inner membrane (TIM23 complex) (Dekker et al., 1997; Schulke et al., 1999).
A component of contact sites in Arabidopsis was identified recently. The translocase of the inner membrane 17 (AtTIM17-2) was shown to link the inner and outer membranes by means of its C-terminal region that is also essential for protein import (Murcha et al., 2005). Interestingly, the Arabidopsis protein can complement a yeast TIM17 mutant, but only when the C-terminal region of 85 amino acids, not present in the yeast protein, is removed (Murcha et al., 2003). A new component of the S. cerevisiae TIM23 complex, Tim21, has been identified (Mokranjac et al., 2005). Tim21 is anchored in the inner boundary membrane and, via its C-terminal domain, specifically interacts with the TOM complex, possibly stabilizing the contact site (Mokranjac et al., 2005). It is possible that the C-terminal regions of AtTIM17-2 and S. cerevisiae Tim21 perform a similar role in the respective organisms. The exact relationship between morphological contact sites and translocation contact sites is not known, i.e. whether all contact sites are also import sites or whether there is a subset of the contact sites, for example, the closely-apposed type, that function as sites of protein import while the bridge-type contact sites are structural only.
|
Compartmentalization within the matrix |
|---|
The matrix contains the enzymes of the pyruvate dehydrogenase complex (PDC), TCA cycle, and glycine oxidative decarboxylation during photorespiration, and contains pools of metabolites including NAD, NADH, ATP, and ADP. However, little is known about how the different proteins and metabolites are distributed in the matrix. GFP targeted to the matrix of mitochondria in various types of animal cell is fully dispersed throughout the available space and FRAP (fluorescence recovery after photobleaching) studies have shown diffusion rates of GFP to be close to that of a protein in a dilute aqueous solution (Partikian et al., 1998
). The fact that the measured diffusion rate of GFP in the matrix is only 3–4-fold less than in water led Partikian and colleagues to question the widely-held view that metabolite channelling, where the product of one enzyme is transferred, as substrate, directly to the next enzyme in the pathway, circumventing free aqueous-phase diffusion, is necessary. Instead, Partikian et al. (1998) suggested that the arrangement of metabolic pathways into metabolons, particles containing the enzymes of a part or the whole of a metabolic pathway (Robinson and Srere, 1985; Velot et al., 1997), enabled the establishment of an uncrowded, enzyme-free, aqueous space through which solutes could easily diffuse. PDC is a multienzyme complex considered to be a prototypical metabolon. Analysis of the distribution of protein fusions between GFP and PDC subunits in human fibroblasts revealed a discrete, heterogeneous distribution of PDC in the matrix (Margineantu et al., 2002a). Since human fibroblast mitochondria typically form a reticulum of tubules, the heterogeneous distribution of GFP fluorescence indicates hotspots of PDC along the mitochondrial tubules (Margineantu et al., 2002a). It will be very interesting to discover whether this heterogeneity is maintained under conditions that cause a fragmentation of the reticulum, i.e. will there be discrete mitochondria lacking PDC? Unfortunately, to my knowledge, nothing is known about the inter-mitochondrial distribution of PDC or the TCA-cycle complexes in the physically discrete mitochondria of higher plants. |
Glycolysis |
|---|
Recently, the application of proteomics has demonstrated that seven of the ten glycolytic enzymes are present in a mitochondrial fraction from Arabidopsis suspension culture cells, four of the seven (glyceraldehyde-3-P dehydrogenase, aldolase, phosphoglycerate mutase, and enolase) were also identified in the intermembrane space/outer membrane fraction (Giege et al., 2003
). The purified mitochondrial fraction was capable of metabolizing 13C-glucose to 13C-labelled TCA cycle intermediates, demonstrating that the full glycolytic pathway was present and active, and fusions of enolase or aldolase to yellow fluorescent protein demonstrated co-localization with Mitotracker Red stained mitochondria (Giege et al., 2003). Giege et al. (2003) concluded that the complete glycolytic pathway is associated with mitochondria (possibly as a structurally linked glycolytic metabolon) enabling pyruvate to be provided directly to the mitochondrion where it is a substrate for the matrix-localized PDC. The discoveries of a heterogeneous distribution of PDC along human mitochondrial tubules and the association of glycolysis with mitochondria in Arabidopsis raises the intriguing possibility of the two types of compartmentalization existing in the same organism. The glycolytic pathway, partly associated with the outer mitochondrial membrane, would then be adjacent to the matrix-located PDC thereby enabling the direct channelling of pyruvate from glycolyis to the TCA cycle. It is conceivable that this putative juxtaposition of glycolysis and PDC would occur at contact sites (or induce the formation of contact sites) thereby increasing the efficiency of pyruvate channelling. |
Intrinsic control of mitochondrial morphology and motility |
|---|
The conformation of the inner membrane, believed to be continuously variable between the two extremes detailed above (orthodox and condensed) and dependent on the energy state of the mitochondrion, has been shown to affect the external morphology and motility of mitochondria (Bereiter-Hahn and Voth, 1983
). Change in the external morphology of mitochondria, the bending, branching, formation, and retraction of localized protrusions (Logan et al., 2004) that are typical of mitochondria in living cells have all been ascribed to the rearrangement of cristae (Bereiter-Hahn and Voth, 1994). However, the extent to which these shape changes are truly intrinsic, or involve the activity of molecular motors on the cytoskeleton, is not known. Bereiter-Hahn and Voth (1983) analysed shape changes and motility of mitochondria in endothelial cells from Xenopus laevis tadpole hearts. In the condensed state, mitochondria are immobile, while in the orthodox state they are motile (Bereiter-Hahn and Voth, 1983). Inhibition of electron transport or oxidative phosphorylation causes a decrease in mitochondrial motility and a concomitant transition to the condensed conformation (Bereiter-Hahn and Voth, 1983). Injection of ADP, which induces extreme condensation, also immobilizes mitochondria. In addition to their affect on mitochondrial motility, inhibitors of electron transport induce the formation of large disc-shaped mitochondria, an identical morphology is seen in tissues under anoxic conditions (Bereiter-Hahn and Voth, 1983). Low oxygen pressure, achieved by mounting cells at high density under a coverslip on a microscope slide, also induces the formation of disc-like mitochondria in tobacco suspension cultured cells (Van Gestel and Verbelen, 2002). Over a time period of 4 h (shorter at higher cell densities) the normal discrete mitochondria (0.5–5 µm in length) have fused to form a reticulum composed of linear and ring-shaped tubular sections interspersed with large plate-like structures (Van Gestel and Verbelen, 2002). Mitochondria in Arabidopsis leaf epidermal cells have been observed undergoing similar morphological transitions during prolonged (40 min) incubation of sections of leaf between slide and coverslip (DC Logan, unpublished observations). Interestingly, unlike the Xenopus mitochondria, tobacco suspension cell mitochondria did not change morphology in response to respiration inhibitors or uncouplers (KCN, dinitrophenol or carbonyl cyanide m-chlorophenylhydrazone) nor did oxidative stress induced by paraquat, menadion, hydrogen peroxide, or CuSO4 induce changes in the normal mitochondrial morphology (Van Gestel and Verbelen, 2002). Van Gestel and Verblen suggest that this may be due to up-regulation of the alternative respiratory pathway which has been suggested to mitigate against ROS damage in plant cells (Van Gestel and Verbelen, 2002). However, paraquat and hydrogen peroxide do induce a change in the mitochondrial morphology in Arabidopsis leaf epidermal cells and mesophyll protoplasts (I Scott, AK Tobin, DC Logan, unpublished data, see below).
The effect of the metabolic status of the mitochondrion on mitochondrial morphology and motility has been suggested to help ensure the mitochondria are located where they are needed. Association of mitochondria with energy-requiring structures or organelles has been well described in a variety of systems (Munn, 1974; Tyler, 1992; Bereiter-Hahn and Voth, 1994). One classic example is the formation of the Nebenkern, a collar around the sperm axoneme formed during spermatogenesis and comprising two giant mitochondria formed by repeated fusion events (Hales and Fuller, 1996, 1997). In plant tissues containing chloroplasts, visualization of mitochondria stained with DiOC6 or expressing GFP has shown the frequent close proximity of these two organelles (Stickens and Verbelen, 1996; Logan and Leaver, 2000). It is assumed that this facilitates exchange of respiratory gases and possibly metabolites, although direct evidence for this is lacking. In characean internode cells, it has been suggested that the spatiotemporal distribution of mitochondria within the cell promotes their association with chloroplasts (Foissner, 2004). A further example of mitochondrial association with energy-consuming structures is the association of mitochondria with the endoplasmic reticulum. One explanation for this association has recently been gaining acceptance. It has been demonstrated in HeLa cells that there are micro-domains of the mitochondrial reticulum where it is in very close contact (<60 nm) with the ER (Rizzuto et al., 1998). The functional significance of these micro-domains has been explained on the basis of Ca2+ dynamics (Rutter and Rizzuto, 2000). For example, localized agonist-induced release of Ca2+ from the ER may stimulate uptake into the closely associated mitochondria where the transient increase in Ca2+ may modulate mitochondrial function.
|
Mitochondrial inheritance |
|---|
Mitochondria cannot be created de novo and so mitochondrial propagation can only occur through the process of division of the parental organelle. It follows that individuals must inherit their mitochondria. For a recent review of mtDNA inheritance in plants, readers are directed to Barr et al. (2005)
. Mitochondrial inheritance is maternal in the majority of angiosperms (Mogensen, 1988) and therefore represents a form of extreme compartmentalization whereby paternal mtDNA is excluded or removed from the zygote. In contrast to the mechanism in mammals, wherein sperm mitochondria are ubiquitinated during spermatogenesis leading to their specific degradation in the zygote (Sutovsky et al., 1996, 1999a, 2003; Thompson et al., 2003), maternal inheritance in angiosperms appears to occur primarily as a result of the exclusion of paternal mitochondrial genomes from the male reproductive cells before fertilization (Nagata et al., 1999a, b; Sodmergen et al., 2002), although a mechanism similar to that active in mammals may also operate in angiosperms (Liu et al., 2004). In species that show biparental inheritance of mitochondria, the abundance of mtDNA increases during pollen development (Nagata et al., 1999a). The question of why the majority of eukaryotes display uniparental inheritance of mitochondria is addressed in the section below on the mitochondrial theory of ageing.
Mitochondrial inheritance in yeast is a highly ordered process and many proteins have been implicated in its control (for a review see Okamoto and Shaw, 2005). By contrast, little is known about the mechanisms controlling the inheritance and proliferation of mitochondria in multicellular organisms. During the human cell cycle, mitochondria switch between two predominant morphological states (Barni et al., 1996; Karbowski et al., 2001; Margineantu et al., 2002b). During the G1 phase of the cell cycle, mitochondria fuse to form reticula, bringing the number of individual organelles to half the number prior to M phase (Karbowski et al., 2001). As cells proceed from G1 to S phase mitochondrial numbers increase due to fragmentation (division) of the mitochondrial reticula (Barni et al., 1996; Karbowski et al., 2001; Margineantu et al., 2002b). A similar series of events occurs in plants (Sheahan et al., 2004, 2005). Using protoplast culture, Sheahan and co-workers showed that, prior to protoplast division, mitochondria undergo a period of extensive elongation as a result of what is termed massive mitochondrial fusion (MFF; see below) followed by a phase of mitochondrial fission which, in turn, is followed by the active and uniform dispersal of the mitochondrial population within the cell (Sheahan et al., 2005). The active and uniform dispersal of mitochondria throughout the cytoplasm ensures the unbiased distribution of mitochondria into each daughter cell following cytokinesis, thus the cellular inheritance of mitochondria is not a strictly random process.
Little is known concerning the mechanisms regulating the cellular distribution of mitochondria in higher plants, but disruption of the actin cytoskeleton affected the dispersal of mitochondria and resulted in a biased distribution of mitochondria in the daughter cells (Sheahan et al., 2004). Mutations of the Arabidopsis FRIENDLY MITOCHONDRIA (FMT) gene lead to a grossly altered cellular distribution of mitochondria (Logan et al., 2003). Disruption of FMT causes the mitochondria to form large clusters of ten or hundreds of organelles, although some mitochondria remain apparently normally distributed as singletons throughout the cytoplasm (Logan et al., 2003). FMT is a conserved eukaryotic gene but, apart from a short tetratricopeptide repeat (TPR) domain that is thought to function in protein–protein interactions, the FMT protein has no homology to proteins of known function. Disruption of FMT homologues in Dictyostelium discoideum (cluA) or S. cerevisiae (CLU1) causes aberrant mitochondrial phenotypes (Zhu et al., 1997; Fields et al., 1998). In the cluA– mutant of D. discoideum the mitochondria cluster near the cell centre (Zhu et al., 1997) while in the S. cerevisiae clu1 mutant the mitochondrial tubules collapse to one side of the cell (Fields et al., 1998).
Although at the time of writing the role of the FMT protein is unknown, it is proposed that FMT is involved in the interaction of mitochondria with the microtubule cytoskeleton (Logan et al., 2003). Plant mitochondrial movement has been shown to rely on actin microfilaments (Olyslaegers and Verbelen, 1998; Van Gestel et al., 2002). However, microtubules have been shown to have a role in the positioning and/or tethering of mitochondria (Van Gestel et al., 2002). Mitochondria have been shown to associate with the microtubule-specific motor protein, kinesin (Khodjakov et al., 1998) that binds cargo at the tetratricopeptide repeat (TPR) domains in the kinesin light chains (Stenoien and Brady, 1997; Verhey et al., 2001). When mitochondria are moving on actin filaments they are presumably prevented from binding to microtubules until this is required to immobilize the mitochondria or effect small-scale adjustments to their position. Based on the phenotype of the fmt mutant and the presence of a conserved TPR domain in FMT it is hypothesized that when mitochondria are moving on actin filaments FMT binds to the kinesin-like receptor on the mitochondrion via the TPR domains thereby preventing any unwanted association with microtubules. This putative role of FMT/CluA as a cap for the kinesin-like receptor on the mitochondrion leaves the microtubule-associated kinesin motor free to associate with other mitochondria or different types of cargo. Applying this hypothesis to the fmt mutant, in which receptor-capping would not occur, leaves the mitochondria free to bind to microtubules which prevents their movement on actin. A cluster of mitochondria then develops as mitochondria divide but are unable to move apart. This hypothesis is consistent with the lack of a clustered mitochondrial phenotype in Clu-mutants of Caenorhabditis elegans (M Clarke, personal communication) since, in this organism, large-scale mitochondrial movement is microtubule-based. Application of this receptor-capping hypothesis for FMT function to D. discoideum suggests an explanation for the interconnections between clustered mitochondria in the D. discoideum cluA– mutant reported by Fields et al. (2002). In D. discoideum, final separation of mitochondria following their division may require movement on the cytoskeleton and, therefore, when this is prevented in the case of the cluA– mutant, because of inexorable binding to microtubules, the mitochondria remain connected. The absence of a similar phenotype in Arabidopsis may simply reflect differences in the architecture of the actin cytoskeleton or the contribution of movement to the mitochondrial division process. This hypothesis for FMT function is supported by the observation that the pattern of mitochondrial distribution in fmt mutants mimics the effect of latrunculin-B, which promotes the rapid depolymerization of the actin cytoskeleton (DC Logan, unpublished observations). The Arabidopsis genome contains 61 identified kinesin genes, the highest number in any sequenced eukaryotic genome (Reddy and Day, 2001). However, no heavy chains have been identified in any plant, although some light chains have been predicted in the Arabidopsis genome (Reddy and Day, 2001). By contrast, there are many microtubule-associated processes unique to plants that are likely to require additional plant-specific microtubule-associated proteins including motors (Reddy and Day, 2001; Hussey, 2004). Yeast-two-hybrid screens are currently being used to identify proteins interacting with FMT in vivo.
|
Control of mitochondrial number, size, and gross morphology |
|---|
Mitochondria are pleomorphic organelles and, at any point in time, the mitochondrial population of a given cell consists of a heterogeneous mix of mitochondrial morphologies. In higher plants, mitochondria are typically discrete, spherical to sausage-shaped organelles although more extreme morphologies are frequently seen (see www.st-andrews.ac.uk/
dclogan for movies of Arabidopsis leaf mitochondria and www.plantcellbiologyoncd.com for a range of movies of mitochondria in living plant tissue). The number of mitochondria per cell is variable, depending on cell type and physiological state, however, a typical Arabidopsis mesophyll cell contains approximately 200–300 discrete mitochondria while tobacco mesophyll protoplasts typically contain 500–600 (Sheahan et al., 2004). This basic organization of the higher plant mitochondrial population into discrete organelles sets plants apart from many other eukaryotes. For example, in S. cerevisiae there are typically 5–10 tubular mitochondria forming an extended reticular network beneath the cell cortex (Stevens, 1977). In various mammalian cell types the mitochondrial morphology is complex, consisting of a mixture of small discrete spherical mitochondria and elongate tubular regions that often form a reticulum. Mitochondrial pleomorphism in mammalian cells is known to be affected by culture conditions since different research groups working on the same cell lines have described different morphologies (Karbowski and Youle, 2003). |
Mitochondrial fission |
|---|
Mitochondrial shape, size, and number are controlled by the dynamically opposing processes of fission and fusion (for review see Logan, 2003
; Okamoto and Shaw, 2005). Hermann et al. (1997) screened a population of UV-mutagenized S. cerevisiae for individuals that displayed aberrant mitochondrial morphologies as revealed by staining with DiOC6. In one of the identified mutants, mdm29, the mitochondrial reticulum collapses into a bundle of tubules along one side of the cell. The mdm29 mutation was mapped to DNM1 (Otsuga et al., 1998), which encodes a protein structurally related to dynamin, a large GTPase required for membrane scission during exocytosis (Herskovits et al., 1993; van der Bliek et al., 1993; Damke et al., 1994, 1995; Hinshaw and Schmid, 1995). Independently, Smirnova et al. (1998) demonstrated that Drp1, another member of the dynamin superfamily of large GTP-binding proteins, is involved in the maintenance of normal mitochondrial morphology and distribution in humans. At around the same time as it was suggested that dynamin-like proteins function in mitochondrial fission, a second large GTP-binding protein, called fuzzy onions (fzo), was shown to be required for normal development of the Nebenkern (see above), during spermatogenesis in Drosophila (Hales and Fuller, 1997). Analyses of the fzo1 mutation in yeast confirmed the role of this protein in mitochondrial fusion (Hermann et al., 1998; Rapaport et al., 1998); see below) and led to confirmation of the role of the dynamin-like protein, Dnm1p, in yeast mitochondrial fission (Bleazard et al., 1999; Sesaki and Jensen, 1999). The Fzo1p GTPase regulates mitochondrial fusion, and loss of Fzo1p function causes the mitochondrial network to fragment rapidly (Hermann et al., 1998). By contrast, mutation of DNM1 causes mitochondria to lose their normal structure and, instead, form a large network of interconnected tubules (Sesaki and Jensen, 1999). In fzo1p dnm1p double mutants the wild-type yeast mitochondrial phenotype is restored. This analysis elegantly demonstrated that a delicate balance of fission and fusion is required to maintain yeast mitochondrial shape, size, and number.
Dnm1p is a soluble protein but assembles on the outer mitochondrial membrane at sites where subsequent division occurs (Okamoto and Shaw, 2005). Three further proteins are required for yeast mitochondrial fission and have been shown to regulate the assembly of Dnm1p. Fis1p (Mozdy et al., 2000; Tieu and Nunnari, 2000; not to be confused with the Arabidopsis Fertilization Independent Seed 1 protein) is an evolutionarily conserved integral outer membrane protein tethered by a C-terminal hydrophobic domain and, unlike Dnm1p, is evenly distributed on the mitochondrial outer surface. The mitochondrial phenotype of yeast FIS1 deletion mutants, in which the mitochondria have a net-like morphology, is identical to the DNM1 deletion mutants and double FIS1/DNM1 deletion mutants suggesting that Fis1p and Dnm1p act in the same pathway (Mozdy et al., 2000). Fis1p is required for targeting of cytosolic Dnm1p to the mitochondria (Mozdy et al., 2000; Tieu and Nunnari, 2000). The second protein that interacts with Dnm1p, independently identified by four research groups, is Mdv1p, a mitochondria-associated cytosolic protein that, like Dmn1p, is localized to the site of mitochondrial division (Cerveny et al., 2001; Fekkes et al., 2000; Mozdy et al., 2000; Tieu and Nunnari, 2000). Cells disrupted in MDV1 showed aberrant mitochondrial morphology characterized by a net-like sheet of interconnected tubules, similar to the phenotype of DNM1 deletion mutants. The third Dnm1p-interacting protein, Caf4p, was identified in an affinity screen aimed at identifying proteins interacting with Fis1p (Griffin et al., 2005). Caf4p interacts with each of the other members of the mitochondrial division apparatus: the N-terminal half of Caf4p interacts with Fis1p and Mdv1p; the C-terminal WD40 domain interacts with Dnm1p (Griffin et al., 2005).
Dynamin-like proteins are also required for mitochondrial division in Arabidopsis. Disruption of either of two Arabidopsis dynamin-like genes, DRP3A or DRP3B, results in an aberrant mitochondrial morphology characterized by an increase in the size of individual mitochondria and a concomitant decrease in the number of mitochondria per cell (Arimura and Tsutsumi, 2002; Arimura et al., 2004; Logan et al., 2004; I Scott, AK Tobin, DC Logan, unpublished results). The mutant mitochondria also have frequent constrictions along their length suggesting that they are arrested at the stage of membrane scission (Logan et al., 2004; I Scott, AK Tobin, DC Logan, unpublished results). In addition to DRP3A and DRP3B, a further two members of the Arabidopsis dynamin-like superfamily have been implicated in the control of mitochondrial morphology (Jin et al., 2003). DRP1C and DRP1E, members of the DRP1 subfamily with closest homology to soybean phragmoplastin (Gu and Verma, 1996), were reported to co-localize partially with mitochondria and disruption of either gene was reported to increase the proportion of mitochondria with an elongated morphology (Jin et al., 2003). However, DRP1C and DRP1E have also been localized to the developing cell plate and so these dynamin-like proteins may have a complex role in plant development.
Apart from dynamin-like proteins, only one other member of the plant mitochondrial division apparatus has been identified. An Arabidopsis functional homologue of hFis1/Fis1p was recently identified, and named BIGYIN (At3g57090) to reflect the mitochondrial phenotype in homozygous T-DNA insertion mutants (Scott et al., 2006). Disruption of BIGYIN leads to a reduced number of mitochondria per cell, coupled with a