Proteomic studies have demonstrated that yeast mitochondria contain roughly 1000 different proteins. Only eight of these proteins are encoded by the mitochondrial genome and are synthesized on mitochondrial ribosomes. The remaining 99% of mitochondrial precursors are encoded within the nuclear genome and after their synthesis on cytosolic ribosomes must be imported into the organelle. Targeting of these proteins to mitochondria and their import into one of the four mitochondrial subcompartments – outer membrane, intermembrane space (IMS), inner membrane and matrix – requires various membrane-embedded protein translocases, as well as numerous chaperones and cochaperones in the aqueous compartments. During the last years, several novel protein components involved in the import and assembly of mitochondrial proteins have been identified. The picture that emerges from these exciting new findings is that of highly dynamic import machineries, rather than of regulated, but static protein complexes. In this review, we will give an overview on the recent progress in our understanding of mitochondrial protein import. We will focus on the presequence translocase of the inner mitochondrial membrane, the TIM23 complex and the presequence translocase-associated motor, the PAM complex. These two molecular machineries mediate the multistep import of preproteins with cleavable N-terminal signal sequences into the matrix or inner membrane of mitochondria.
Mitochondria are double-membrane bounded organelles that originated from a bacterial endosymbiotic event approximately two billion years ago. Through the course of evolution the ancestral endosymbiont displaced a large proportion of its genes to the host's nucleus and, in parallel, novel proteins encoded by the host's genome were acquired. Consequently, the majority of mitochondrial precursors are synthesized on cytosolic ribosomes and must subsequently be imported into the organelle. Since mitochondria contain four subcompartments, the outer membrane, the intermembrane space (IMS), the inner membrane and the innermost aqueous compartment, the matrix, proteins need to be additionally sorted into these different locations. This protein sorting process is mediated by import signals within the preproteins. In a simplified view, two main classes of mitochondrial targeting signals can be distinguished: amino-terminal, cleavable presequences and internal signals, which are usually distributed throughout the mature part of the protein. The targeting information is specifically recognized by receptors and often also by the channel-forming proteins, which allow passage of preproteins across the membranes.
Initially, three distinct mitochondrial protein import and assembly complexes were identified: the Translocase of the Outer Membrane (TOM complex) (Herrmann & Neupert et al., 2000; Endo et al., 2003; Pfanner et al., 2004; Wiedemann et al., 2004a), and two Translocases of the Inner Membrane dedicated to the import of proteins into the matrix or the inner membrane (TIM complexes) (Jensen & Dunn et al., 2002; Neupert & Brunner et al., 2002; Koehler et al., 2004a; Rehling et al., 2004) (Fig. 1).
The TOM complex represents the central entry pore for essentially all nuclear-encoded mitochondrial precursors. However, after passage through the TOM complex precursor import pathways diverge dictated by the targeting information they possess. β-Barrel proteins of the outer membrane are selectively directed to the Sorting and Assembly Machinery (SAM) complex, which then mediates their insertion and assembly into the outer membrane (Pfanner et al., 2004; Paschen et al., 2005) (Fig. 1a).
A group of IMS proteins such as the small Tim proteins (Tim9, Tim10, Tim8, Tim13) depend on the recently established IMS-specific import machinery consisting of the IMS residents Mia40 and Erv1 for their transport and subsequent assembly (Chacinska et al., 2004; Naoe et al., 2004; Allen et al., 2005; Mesecke et al., 2005; Rissler et al., 2005; Terziyska et al., 2005) (Fig. 1b). In contrast, inner membrane proteins with internal targeting signals, such as the metabolite carriers of the inner mitochondrial membrane, are bound by specialized chaperone complexes in the IMS, the Tim9–Tim10 complex or Tim8–Tim13 complex, which guide these largely hydrophobic precursors across the aqueous IMS to the inner membrane carrier translocase (TIM22 complex) that mediates their insertion (Jensen & Dunn et al., 2002; Koehler et al., 2004a; Rehling et al., 2004) (Fig. 1c).
Preproteins with N-terminal signal sequences (presequences) are destined for the mitochondrial matrix or in some cases the inner membrane and are transported across the inner membrane by the presequence translocase (TIM23 complex) (Jensen & Dunn et al., 2002; Neupert & Brunner et al., 2002; Endo et al., 2003; Rehling et al., 2004) (Fig. 1c and d). In the mitochondrial matrix the presequence is proteolytically removed from the precursor by the matrix processing peptidase (MPP) generating the mature form of the imported protein. While both the TIM22 and TIM23 complex use the membrane potential (Δψ) across the inner membrane as a driving force to translocate the precursors into or across the inner membrane, complete translocation of presequence-containing precursors into the matrix by the TIM23 complex requires the ATP-powered mitochondrial import motor as a second driving force (Jensen & Dunn et al., 2002; Neupert & Brunner et al., 2002; Koehler et al., 2004a; Rehling et al., 2004) (Fig. 1d). The core part of this so-called Presequence translocase Associated Motor complex (PAM) is a mitochondrial member of the Hsp70 chaperone family (mtHsp70), which is tethered to the TIM23 complex by the adaptor protein Tim44. In addition, members of a specific class of presequence-containing preproteins typically destined for the mitochondrial inner membrane or IMS contain bipartite signal sequences (Jensen & Dunn et al., 2002; Koehler et al., 2004a; Herrmann & Hell et al., 2005). In the primary structure of this class of precursors such as cytochrome b2 or cytochrome c1, the N-terminal matrix-targeting signal is followed by a hydrophobic sorting signal. The sorting signal stalls the translocation of these precursors into the matrix by arresting them within the TIM23 translocon and induces the lateral release of the precursor into the inner membrane (Fig. 1c). Such precursors may remain anchored in the inner membrane or undergo a second processing event on the IMS side of the inner membrane (Jensen & Dunn et al., 2002; Koehler et al., 2004a; Herrmann & Hell et al., 2005), which results in the release of a soluble IMS protein. However, recent analyses have shown that the underlying molecular mechanisms for this sorting process is more complicated than expected, and it is discussed in detail below.
Proteins pass the outer membrane through the TOM complex
Preproteins destined for the outer membrane, the IMS, the inner mitochondrial membrane or the matrix pass the outer membrane via the TOM complex (Herrmann & Neupert et al., 2000; Jensen & Dunn et al., 2002; Koehler et al., 2004b; Pfanner et al., 2004; Rehling et al., 2004; Wiedemann et al., 2004a) (Fig. 1). The TOM complex contains two primary receptor proteins, Tom20 and Tom70, which associate with the core of the TOM complex, also termed the General Import Pore (GIP). Five proteins, the receptor Tom22, the small Tom proteins Tom5, Tom6, and Tom7 and the essential, pore-forming subunit Tom40, form the GIP complex (Fig. 2).
The receptor Tom70 recognizes mainly internal targeting signals within precursor proteins, such as the carrier proteins of the inner membrane, which contain multiple internal targeting signals distributed throughout the protein (Herrmann & Neupert et al., 2000; Jensen & Dunn et al., 2002; Koehler et al., 2004a; Rehling et al., 2004). Several Tom70 dimers are recruited to a single substrate protein. Cytosolic chaperones, such as members of the Hsp70 and Hsp90 families, deliver the carrier precursors to Tom70 in an ATP-dependent manner (Young et al., 2003). The second peripheral receptor, Tom20, recognizes mainly matrix precursor proteins with N-terminal, cleavable presequences. Tom20 and Tom70 deliver precursor proteins to the secondary receptor Tom22. Tom22 is more tightly associated with the channel-forming core of the TOM complex, Tom40, and mediates the transfer of preproteins to the central pore (Herrmann & Neupert et al., 2000; Endo et al., 2003; Koehler et al., 2004a; Wiedemann et al., 2004a) (Fig. 2). Translocation of preproteins from Tom22 into the import channel requires Tom5, the smallest subunit of the TOM complex (Wiedemann et al., 2004a). Tom5 consists of a single transmembrane helix and exposes a small, negatively charged domain to the cytosol. Another small TOM complex protein, Tom6, is needed to establish a stable interaction of Tom22 with Tom40 and seems to be of general importance for the stability of the TOM complex. Tom7 appears to be an antagonist of Tom6, which destabilizes the TOM complex to allow some conformational flexibility for the dynamic assembly of the core complex (Wiedemann et al., 2004a).
The β-barrel protein Tom40 forms the protein-conducting unit for preprotein transfer across the outer membrane (Endo et al., 2003; Pfanner et al., 2004; Wiedemann et al., 2004a; Paschen et al., 2005) (Fig. 2). In the TOM complex, several Tom40 molecules form two to three pores. These import channels are cation-selective and have an estimated diameter of 20–25 Å, as judged by electrophysiological analyses and electron microscopy. Recombinantly expressed, reconstituted Tom40 was shown to form a single channel of very similar dimensions and properties (Becker et al., 2005). When a presequence-containing preprotein emerges from the Tom40 channel at the IMS side of the outer membrane, it initially interacts with the IMS domain of Tom22 via its presequence (see below). The presence of sequential interaction sites within the TOM complex has led to the so-called ‘binding chain hypothesis’, which postulates that preproteins pass through the TOM complex via a chain of low-affinity binding sites from Tom20/Tom70 via the cytosolic domain of Tom22 and Tom5 into the Tom40 channel and thereafter to the IMS domain of Tom22. Finally, the IMS domain of Tom22 presents the presequence-containing proteins to the TIM23 complex (Wiedemann et al., 2004a).
The SAM complex mediates the insertion of outer membrane β-barrel proteins
How are proteins transported into the outer membrane? It appears that outer membrane proteins utilize internal targeting signals for transport across the outer membrane. While it is generally believed that outer membrane proteins with simple topology, such as proteins with single transmembrane spans, are laterally released from the TOM complex into the lipid bilayer, recent analyses of the transport of outer membrane β-barrel proteins has led to the identification of a novel membrane protein assembly complex of the outer membrane, which was termed the SAM (Sorting and Assembly Machinery) complex (Wiedemann et al., 2003; Pfanner et al., 2004; Paschen et al., 2005) (Fig. 2). Interestingly, the biogenesis of outer membrane β-barrel precursors involves an initial translocation of precursors into the IMS via the TOM complex where, with the aid of the small Tim proteins, they are guided to the SAM complex for their subsequent integration and assembly into the outer membrane (Kozjak et al., 2003; Paschen et al., 2003; Wiedemann et al., 2003; Gentle et al., 2004; Hoppins & Nargang et al., 2004; Ishikawa et al., 2004; Milenkovic et al., 2004; Waizenegger et al., 2004; Wiedemann et al., 2004b) (Fig. 2). To date, four subunits of the SAM complex are known. Sam37 (or Mas37) (Wiedemann et al., 2003) is an outer-membrane protein exposed to the cytosol. Sam37-deficient mitochondria are selectively impaired in the assembly of β-barrel proteins of the outer membrane such as Tom40, porin and Mdm10 (Wiedemann et al., 2003). Sam50 (Tob55) (Kozjak et al., 2003; Paschen et al., 2003; Gentle et al., 2004) is the essential core component of the SAM complex (Fig. 2) and is likely to form a β-barrel itself. Another essential constituent of the SAM complex, Sam35 (Tob38) (Ishikawa et al., 2004; Milenkovic et al., 2004; Waizenegger et al., 2004), is a peripheral outer membrane protein. Temperature-sensitive mutants of sam35 show a similar β-barrel protein assembly defect as mutants in sam37 or sam50 (Milenkovic et al., 2004). Recently, Meisinger. (2004) demonstrated that Mdm10, a protein that was previously shown to be involved in mitochondrial distribution and morphology, is a component of the SAM complex (Fig. 2). However, Mdm10-deficient mitochondria are selectively impaired in the assembly of Tom40, while the biogenesis of other β-barrel proteins such as porin is not affected. These unexpected findings show an intimate link between mitochondrial morphogenesis and protein import (Meisinger et al., 2004). Even more factors that act downstream of the SAM complex appear to be required for the biogenesis of the TOM complex. Depletion of the outer-membrane protein Mim1 was found to affect the assembly of the TOM complex (Ishikawa et al., 2004; Waizenegger et al., 2005). The biogenesis of the mature 450-kDa TOM complex proceeds via two distinct intermediates of 250-kDa and 100-kDa, respectively. The 250-kDa intermediate represents a SAM complex-associated form of newly imported Tom40 (Milenkovic et al., 2004). While in mutants with impaired SAM core-complex functions none of these intermediates are formed efficiently, the early 250-kDa intermediate can still be formed in Mim1-depleted mitochondria while further steps of Tom40 assembly are affected (Ishikawa et al., 2004; Waizenegger et al., 2005). In this respect mim1 mutant mitochondria resemble mdm10Δ mitochondria (Meisinger et al., 2004). Thus, Mim1 seems to be involved in the biogenesis of Tom40; however, an association of Mim1 with the SAM or TOM complex has not yet been observed.
Taken together, the biogenesis of the TOM complex appears to be a highly complicated process that we are just beginning to understand, and additional components of this pathway may indeed be realised. Moreover, it is still elusive, where, when and how the newly imported Tom40 is folded into a β-barrel and what the molecular roles of the SAM complex and Mim1 are in the biogenesis of the TOM complex.
Mia40 and Erv1 are required for the import of small intermembrane space proteins
Import into mitochondria and assembly of several small IMS proteins was recently shown to depend on the essential protein, Mia40 (Chacinska et al., 2004; Naoe et al., 2004; Terziyska et al., 2005) (Fig. 1b). Mia40 exposes a large soluble domain containing its active site to the IMS. The interaction of small Tim proteins and other IMS proteins such as Cox17 and Cox19 with Mia40 is an essential step in their import and a prerequisite for the assembly of Tim9 and Tim10 into their native heterohexameric complex (Chacinska et al., 2004). Mia40 forms transient disulfide bridges with its substrates and, as a consequence, strong reducing conditions inhibit Mia40-dependent protein import in vitro. Completion of the Mia40 reaction cycle requires the action of the sulfhydryl oxidase Erv1 (Allen et al., 2005; Mesecke et al., 2005; Rissler et al., 2005) (Fig. 1b). However, it is still unclear at which stages of the Mia40-dependent IMS import Erv1 is needed. It has been suggested that the function of Erv1 is to reoxidize Mia40 after substrate release (Mesecke et al., 2005), while another study indicates that Erv1 acts directly on the substrates or on Mia40-substrate complexes (Rissler et al., 2005) further downstream in the import pathway. Mia40, as well as its so far identified substrates, contains characteristic cysteine-rich motifs, which are able to bind divalent metal ions, such as zinc (Zn2+) and, in the case of Mia40, also copper (Cu2+) (Koehler et al., 2004b; Terziyska et al., 2005; Wiedemann et al., 2006). However, the physiological relevance of this affinity for metals is not clear. For the small Tim complexes, it is becoming more evident that folding and assembly strongly depend on disulfide formation rather than on metal binding (Lu et al., 2004). In addition, the recently solved X-ray structure of the Tim9–Tim10 complex revealed that each subunit contains two intramolecular disulfide bridges (Webb et al., 2006; Wiedemann et al., 2006).
Carrier transport across the inner membrane
A large group of multispanning inner membrane proteins possess internal targeting information. Among these proteins are the metabolite carriers of the inner mitochondrial membrane, e.g. the ADP/ATP translocator (AAC). Internal signals are initially recognized by Tom70 (see above), upon which carrier precursor proteins are translocated across the outer membrane through the TOM complex in a hairpin-loop conformation (Endo et al., 2003; Koehler et al., 2004a; Rehling et al., 2004). Carrier release from the receptors and transport across the GIP requires that the Tim9–Tim10 complex of the IMS associates with the carrier precursor (Fig. 1c). The Tim9–Tim10 complex forms a six-blade α-helical propeller structure (Webb et al., 2006), which acts in a chaperone-like manner in that it protects the hydrophobic carrier molecules during transport across the aqueous IMS. A specialized transport machinery in the inner membrane, termed the carrier translocase or TIM22 complex, accepts the carrier precursor from the small Tim proteins and mediates its subsequent insertion into the inner mitochondrial membrane (Endo et al., 2003; Rehling et al., 2004) (Fig. 1c). The 300-kDa TIM22 complex uses the membrane potential across the inner membrane as the only external energy source to drive the insertion process. Tim22 represents the pore-forming core subunit of the translocase to which the proteins Tim12, Tim18, and Tim54, together with the small Tim proteins, are bound. The analysis of the isolated TIM22 complex revealed that it contains two Tim22 pores, which cooperate during protein transport (Rehling et al., 2003). The insertion process of the carrier precursor occurs in multiple steps that depend on the presence of a membrane potential across the inner membrane and the recognition of internal signals within the preprotein by the TIM22 complex (Jensen & Dunn et al., 2002; Koehler et al., 2004a; Rehling et al., 2004; Brandner et al., 2005).
Evolving views on the presequence translocase of the inner membrane
The presequence translocase of the inner membrane (TIM23 complex) and the associated import motor are required for the translocation of precursors with cleavable N-terminal signal sequences into the matrix (Fig. 3). A number of recent studies have demonstrated that the dynamic behaviour and signal sequence-dependent regulation of the TIM23 complex and its associated motor is much more complicated than initially anticipated. The discovery of several novel subunits of the mitochondrial import motor has made clear that the long-known components Tim44, mtHsp70 and the nucleotide exchange factor Mge1 are in fact part of a sophisticated molecular machine now known as the presequence translocase-associated motor (Saccharomyces genome database; SGD) or PAM complex (D'Silva et al., 2003; Truscott et al., 2003; Frazier et al., 2004). The PAM complex interacts and cooperates with the membrane-embedded TIM23 complex in a versatile and highly regulated manner in order to mediate preprotein import into the matrix.
Interestingly, it recently has been demonstrated that the TIM23 complex directly interacts with the TOM complex via a physical contact between the IMS domain of Tom22 and the newly identified Tim21 protein (Chacinska et al., 2005; Mokranjac et al., 2005a) (Fig. 4; see below). This interaction, however, is not stable, but regulated by the import signals of incoming preproteins. Tim21 also affects the interaction of PAM with the TIM23 complex, suggesting an intricate reaction cycle coupled to a signal transduction cascade from the mitochondrial surface across two membranes into the matrix (Chacinska et al., 2005; Oka & Mihara et al., 2005). In the following sections, we will discuss in detail our current view on the composition, the energetic requirements and the molecular mechanism of the TIM23/PAM complex.
Where TOM and TIM meet: direct transfer of preproteins from the outer to the inner membrane
Even before identification of the TOM and TIM translocases, it was shown that preproteins in transit to the matrix span both the outer and inner mitochondrial membranes simultaneously (Fig. 4). Thus, when the TOM and TIM23 complexes were identified, it was speculated that there was a physical interaction and concomitant preprotein transfer from the TOM complex to the TIM23 complex. Many groups arrested preproteins in transit across both membranes using fusion proteins of a mitochondrial precursor protein to the stably folded C-terminal domain of dihydrofolate reductase (DHFR). In the presence of its inhibitor methotrexate (MTX), DHFR becomes tightly folded and its translocation across the outer membrane is inhibited. These analyses revealed that the N-terminal ends of accumulated preprotein translocation-intermediates were inserted into the TIM23 complex and partially translocated into the matrix, where processing by MPP could take place, while their C-terminal region was still exposed to the cytosolic side of the outer membrane. Such TOM–TIM–preprotein supercomplexes could be visualized by Blue-Native polyacrylamide gel electrophoresis (PAGE) analysis and were subsequently isolated by affinity purification (Geissler et al., 2002; Chacinska et al., 2003; Endo et al., 2003). This two-membrane-spanning intermediate represents a productive import intermediate, since the trapped preprotein can be chased into the matrix after release of the import block (Chacinska et al., 2003). However, for a long time it remained elusive if the TOM and TIM23 complexes were in direct physical contact via transient interactions, or just connected via the preprotein in transit. In addition, the molecular mechanisms underlying the formation and stabilization of such intermediates were not understood. In this regard, the observation that the extreme N-terminus of inner membrane channel protein Tim23 protrudes through the outer membrane was an interesting observation. However, deletion of the Tim23 IMS domain portion that was suggested to span the outer membrane did not affect the formation of the TOM–TIM preprotein supercomplex nor protein transport (Chacinska et al., 2003, 2005). The recent identification of two novel subunits of the TIM23 complex, Tim50 and Tim21, has provided many new insights into the protein–protein interactions that govern supercomplex formation (Fig. 4).
Tim50 is an essential protein in yeast located in the inner mitochondrial membrane (Geissler et al., 2002; Yamamoto et al., 2002; Mokranjac et al., 2003a). Tim50 spans the inner membrane via a single transmembrane segment with its N-terminal end protruding into the matrix and its C-terminus exposed to the IMS (Figs 3 and 4). Depletion of Tim50 strongly inhibits import into the matrix, while the sorting of inner membrane proteins is only mildly affected. Tim50 could be copurified with tagged Tim23 (Geissler et al., 2002; Mokranjac et al., 2003a) and with tagged Tom22 after formation of a TOM–TIM–preprotein supercomplex (Geissler et al., 2002). These analyses demonstrate that Tim50 is a subunit of the functional TIM23 translocase during import. In addition, Tim50 can be crosslinked to a preprotein that is accumulated as an early transport intermediate in the TOM complex (Yamamoto et al., 2002; Mokranjac et al., 2003a). This observation suggests that an interaction between preproteins and Tim50 occurs at an early stage of protein transport, namely before the preprotein is transferred from the TOM complex to the TIM23 complex (Fig. 4). Indeed, Tim50 promotes the formation of a TOM complex-arrested precursor intermediate (Chacinska et al., 2005). At this stage, the presequence is associated with the IMS domain of Tom22 (Tom22IMS), since this domain is essential for the formation of a TOM-arrested precursor intermediate (Frazier et al., 2003) (Fig. 4). Moreover, at the inner membrane the IMS domain of Tim50 interacts directly with the IMS domain of Tim23 (Geissler et al., 2002; Yamamoto et al., 2002). Thus, Tim50 appears to have multiple functions in protein transport: (i) in the recognition of preproteins as soon as they emerge from the TOM complex; (ii) in the formation of a precursor-TOM transport intermediate in which the presequence is bound to Tom22IMS, and (iii) in the transfer of precursors to the import channel of the inner membrane.
Another novel subunit of the TIM23 complex that was recently identified by affinity purification of tagged Tim23 is Tim21 (Chacinska et al., 2005; Mokranjac et al., 2005a; Oka & Mihara et al., 2005). Tim21 is a protein of the inner mitochondrial membrane that exposes a domain to the IMS (Tim21IMS), which interestingly has been shown to bind to Tom22IMS (Chacinska et al., 2005; Mokranjac et al., 2005a) (Fig. 4). This is the first described direct physical contact between the TOM and TIM23 complex. Strikingly, the interaction between Tom22IMS and Tim21IMS is lost in the presence of presequences, suggesting that Tim21 and presequences compete for the same binding site on Tom22IMS (Chacinska et al., 2005). Consequently, increasing the amounts of Tim21 in mitochondria by overexpression strongly reduces the amount of a TOM-arrested intermediate by promoting the release of the presequence from Tom22IMS.
Thus, Tim21 and Tim50 have antagonistic functions in the transfer of preproteins from the TOM to the TIM23 core complex. After passage through the Tom40 channel, the preprotein first binds to Tim50. This interaction is required to generate a stable interaction of the presequence with Tom22. At this stage, the incoming preprotein initially tethers the TOM and TIM23 complexes (Fig. 4). Completion of preprotein transfer to the TIM23 core complex requires release of the presequence from Tom22. Binding of Tim21IMS to Tom22IMS initiates this process, which is the prerequisite for the subsequent interaction of the presequence with the channel-forming Tim23 protein (Fig. 4).
The protein-conducting channel of the TIM23 complex: more than just a hole in the membrane
The essential proteins Tim23 and Tim17, which were first identified as central import components by genetic screens in the early 1990s, form the membrane-integrated core of the TIM23 complex (Jensen & Dunn et al., 2002; Neupert & Brunner et al., 2002; Endo et al., 2003; Koehler et al., 2004a; Rehling et al., 2004; Wiedemann et al., 2004a) (Fig. 3). Both proteins exhibit sequence similarity to each other and share the same topology. They contain four transmembrane segments and their N- and C-termini are located in the IMS. Tim23 has an N-terminal domain, which has been implicated in preprotein recognition (see above). Both Tim23 and Tim17 have been crosslinked to preproteins in transit. Tim17 and Tim23 together form the preprotein import site in the inner membrane of mitochondria that cooperates via the adaptor protein Tim44 with mtHsp70 (Jensen & Dunn et al., 2002; Neupert & Brunner et al., 2002; Rehling et al., 2004; Wiedemann et al., 2004a) (Fig. 3). The so-called core complex of Tim23 and Tim17 migrates as a 90-kDa species on Blue-Native PAGE (90K complex). Electrophysiological experiments with purified and reconstituted Tim23 show that this protein forms cation-selective, voltage-gated channels. The Tim23 channel was estimated to have a diameter of 13–24 Å. Based on this observation, it is believed that the protein-conducting pore through the inner membrane is mainly formed by the transmembrane helices of Tim23. The N-terminal IMS domain of Tim23, which interacts with presequences, is required for the selectivity of the channel. Indeed, presequence peptides have been shown to activate the Tim23 channel specifically (Jensen & Dunn et al., 2002; Endo et al., 2003; Rehling et al., 2004; Wiedemann et al., 2004a).
In contrast to Tim23, very little is known about the molecular function of Tim17. A temperature-sensitive mutation in TIM23 (tim23–2) has been described, which selectively destabilizes the 90K complex. Mitochondria from tim23–2 mutant cells are affected in preprotein transport, indicating that the association of Tim17 with Tim23 in the 90K complex is required for preprotein import. Tim17 and Tim23 interact via their hydrophobic, membrane-integral domains and appear to be present in the TIM23 complex in equimolar amounts (Jensen & Dunn et al., 2002; Endo et al., 2003; Rehling et al., 2004; Wiedemann et al., 2004a). Two recent studies proposed mechanistic roles for Tim17 in preprotein import. Meier. (2005) showed that the expression of an N-terminally truncated variant of Tim17 (Tim17ΔN) leads to a significant reduction of the electrical potential across the inner membrane. However, the import defect seen in Tim17ΔN mutant mitochondria is probably not due to the diminished membrane potential but rather to the fact that the passage of preproteins through the TIM23 complex was affected. Thus, it was concluded that the N-terminus of Tim17 plays a role for the initial insertion of preproteins into the Tim23 channel by affecting its gating properties. Chacinska. (2005) reported on another function of Tim17. Generating temperature-sensitive alleles of TIM17 they obtained one mutant (tim17–5) that was specifically affected in matrix import, while another mutant (tim17–4) displayed defects in the sorting of inner membrane proteins. Thus, Tim17 plays an important role in both matrix import and inner membrane sorting, and these two functions can be genetically separated. Remarkably, purification of the TIM23 complex from a tim17–5 mutant background demonstrated that the association of the PAM complex with the TIM23 complex was affected. This observation as well as other data suggests that Tim17 plays a key role in recruiting the motor complex to the TIM23 complex (Figs 3 and 4).
What drives matrix import? PAM as an intricate molecular machine
The membrane potential (Δψ) provides the energy for the initial translocation of the presequence across the inner membrane through the Tim23 channel. However, further import of the preprotein requires ATP hydrolysis by mtHsp70 (Neupert & Brunner et al., 2002; Endo et al., 2003; Koehler et al., 2004a; Rehling et al., 2004) (Figs 3 and 4). The mitochondrial Hsp70 (mtHsp70), which in yeast is encoded by the SSC1 gene, is essential for preprotein import into mitochondria. mtHsp70 binds directly to an incoming preprotein as soon as the N-terminus has entered into the matrix. Like other members of this large chaperone family, mtHsp70 undergoes a regulated reaction cycle. Hsp70 systems generally consist of an ATP-consuming chaperone, a nucleotide exchange factor and a so-called J-domain protein (J-protein) that stimulates the ATPase activity of Hsp70s (for review see Mayer et al., 2004). The best-studied example of such a triad is the Escherichia coli Hsp70 DnaK with its nucleotide exchange factor GrpE and J-protein DnaJ. Hsp70 systems play a key role in the correct folding of newly synthesized proteins in the cytoplasm of both prokaryotic and eukaryotic cells. The yeast mtHsp70 is not only essential for protein folding inside mitochondria but also for the import of preproteins into the mitochondrial matrix. The reaction cycle of mtHsp70 at the translocation site involves initial low-affinity preprotein binding by the ATP-bound form of the protein. Subsequent ATP hydrolysis allows tight binding to the preprotein. Finally, release of the preprotein requires exchange of ADP against ATP. The nucleotide exchange factor of mtHsp70, Mge1, has been shown to cooperate with mtHsp70 in both protein folding and preprotein import (Neupert & Brunner et al., 2002). The J-domain protein that assists mtHsp70 in protein folding was found to be Mdj1. However, while deletion of Mdj1 does interfere with proper protein folding, it has no effect on preprotein import. Thus, folding and import mediated by mtHsp70 appeared to require different cochaperones. This view was supported by the identification of two distinct Hsp70 complexes, which sequentially interact with an imported preprotein. The ‘import complex’, consists of mtHsp70, Mge1 and the peripheral inner membrane protein Tim44, while the ‘folding complex’ that interacts with newly imported proteins during the later stage of their maturation, contains Mdj1. Two other known mitochondrial J-domain-containing proteins, Jac1 and Mdj2, are dispensable for preprotein import. Therefore, a DnaJ-like function of Tim44 was assumed, especially as Tim44 contains an 18-residue segment that shows limited sequence similarity to J-proteins. This stretch of amino acids was shown to be essential for productive interaction of Tim44 with mtHsp70. However, later analyses demonstrated that Tim44 did not stimulate the ATPase activity of mtHsp70 (D'Silva et al., 2003; Truscott et al., 2003).
Since mtHsp70 binds to the incoming preprotein, it was suggested that this event traps the preprotein in transit through the Tim23 channel on the way to the mitochondrial matrix. In this model the ability of the preprotein to diffuse forward and backwards in the Tim23 channel by Brownian motion is limited by binding of mtHsp70 to the precursor and thus a directionality of the movement is provided. In this model, complete import of the polypeptide chain into the matrix requires a successive series of interactions between mtHsp70 molecules and the precursor. An alternative model suggests that mtHsp70 is bound to the translocase and performs a conformational change during the ATPase cycle that translates into a power stroke. This movement of mtHsp70 at the matrix side of the inner membrane could actively pull preproteins across both mitochondrial membranes. Tim44 is a central component of the translocase and critical in both models for the function of mtHsp70 in protein import. Tim44 can be crosslinked to a preprotein in transit through the inner membrane and is required for preprotein import. In addition, Tim44 interacts with mtHsp70 and appears to recruit it to the TIM23 complex (Neupert & Brunner et al., 2002; Endo et al., 2003; Rehling et al., 2004).
For the last 10 years, the ‘Brownian ratchet model’ and the ‘power stroke model’ have been controversially discussed and a wealth of literature has been published with data supporting both models. In the meantime, it has become more and more evident that probably both mechanisms contribute to preprotein import, but to what extent seems to depend on the nature of the preprotein (Neupert & Brunner et al., 2002; Koehler et al., 2004a; Mayer et al., 2004; Rehling et al., 2004). Moreover, both models were based on the known proteins that regulate mtHsp70. However, we now know that there are three additional components of the import motor (now termed the PAM complex) and each is clearly involved in the regulation of mtHsp70. It will be interesting to see how their function can be incorporated into these models.
Unusual complexity of the mitochondrial Hsp70 import motor (PAM)
The apparent lack of a DnaJ homolog protein in the import motor that acts on mtHsp70 during import was recently solved by three independent reports on the identification of a novel mitochondrial J-domain containing protein, Pam18 (Tim14) (Fig. 3). Indeed, this protein was shown to play a crucial role in mtHsp70-dependent matrix protein import (D'Silva et al., 2003; Mokranjac et al., 2003b; Truscott et al., 2003). Pam18 is an essential protein of the inner mitochondrial membrane in yeast. It contains a single transmembrane segment and a classical J-domain including the conserved ‘HPD’ signature motif that is exposed to the matrix. As expected, the J-domain of Pam18 stimulates the ATPase activity of mtHsp70 (D'Silva et al., 2003; Truscott et al., 2003). Pam18 copurifies with the TIM23/PAM complex and can be crosslinked to Tim44. Interestingly, Pam18 promotes the formation of a stable complex between Tim44 and mtHsp70. Mutation or depletion of Pam18 leads to a selective impairment of preprotein import into the matrix, while motor-independent inner membrane sorting of preproteins is not affected, indicating that Pam18 is required for the proper function of the PAM complex (D'Silva et al., 2003; Mokranjac et al., 2003b; Truscott et al., 2003). Taken together, Pam18 represents the long-searched-for J-protein of mtHsp70 at the preprotein import site and thus plays a central role in the PAM complex reaction cycle.
Besides Tim44, Pam18 and the nucleotide exchange factor Mge1, the regulation of the PAM complex requires further components. The mtHsp70-stimulating activity of Pam18 is regulated by Pam16 (Tim16), another recently identified component of the PAM complex that is also essential for cell viability (Frazier et al., 2004; Kozany et al., 2004; D'Silva et al., 2005) (Fig. 3). Similar to Pam18, Pam16 can be copurified with the TIM23/PAM complex. Pam16 interacts with Pam18 and is required for the association of Pam18 with the TIM23 complex. Pam18 and Pam16 migrate together as an 80-kDa complex on Blue-Native PAGE, suggesting that they represent a distinct module of the PAM complex. Depletion of Pam16 from mitochondria as well as temperature-sensitive alleles of PAM16 lead to a specific defect in preprotein import into the matrix, but do not affect inner membrane sorting. Pam16 displays sequence similarity to J-proteins, but lacks the essential and conserved ‘HPD’ signature motif. Consequently, Pam16 is not able to stimulate the ATPase activity of mtHsp70 (Frazier et al., 2004; Kozany et al., 2004; D'Silva et al., 2005). In contrast, the addition of soluble Pam16 to an in vitro system containing mtHsp70, the J-domain of Pam18 and Mge1 leads to a reduction of mtHsp70 ATPase activity. Thus, it appears that Pam16 affects preprotein import through its antagonistic activity on Pam18-induced ATPase stimulation of mtHsp70 (Li et al., 2004; D'Silva et al., 2005).
Recently, van der Laan. (2005) reported that the PAM complex contains yet another subunit, Pam17 (Fig. 3). Deletion of PAM17 selectively inhibits the import of preproteins into the matrix, while preprotein sorting into the inner membrane is unaffected. But what is the molecular role of Pam17? Blue-Native PAGE experiments showed that the interaction between Pam18 and Pam16 is significantly weakened upon deletion of PAM17, while the overall architecture of the TIM23/PAM complex remains intact. Moreover, purification of the TIM23/PAM complex from mitochondria lacking Pam17 revealed that the efficient association of the Pam18/Pam16 module with the translocase requires Pam17. Thus, Pam17 stabilizes the interaction of Pam18 with Pam16 in addition to the binding of the Pam18/Pam16 module to the TIM23 core complex, and thereby promotes the PAM reaction cycle.
In summary, the PAM complex consists of at least six subunits: mtHsp70, the ATP-driven core of the PAM complex, the docking protein Tim44, the transiently associated nucleotide exchange factor Mge1, the J-protein Pam18, the Pam18-adaptor and regulator protein Pam16 and Pam17, the organizer of the Pam18/Pam16 module (Fig. 3). Mokranjac. (2005b) reported that the role of Pam18 can be partially taken over by the mitochondrial J-protein Mdj2, a conclusion that is based on the observation that overexpression of Mdj2 partially restores growth of Pam18-deficient yeast cells. However, since deletion of Mdj2 has no effect on preprotein import and wild-type levels of Mdj2 do not compensate for a defect in Pam18, a physiological role of Mdj2 in preprotein import remains unlikely. Recently, the mitochondrial matrix protein, Zim17 (Tim15 or Hep1) was suggested to cooperate with mtHsp70 in preprotein import (Burri et al., 2004; Yamamoto et al., 2005). However, two later studies demonstrated that Zim17 is rather a matrix heat-shock protein that prevents mtHsp70 from aggregation (Sanjuan Szklarz et al., 2005; Sichting et al., 2005). After prolonged inactivation of Zim17 aggregation of mtHsp70 leads to pleiotropic effects on preprotein import and mitochondrial morphology (Sanjuan Szklarz et al., 2005).
Dynamic association of the PAM complex with the TIM23 complex is regulated by Tim21
Preproteins and even different domains of a particular preprotein can differ substantially in their motor requirement during import into mitochondria. This suggests that the association of the PAM complex to the TIM23 complex may be a dynamic process. Indeed, Chacinska. (2005) showed that Tim21 is involved in regulating the TIM23/PAM complex dynamics. When the TIM23 complex is isolated via a tagged Tim21 rather than Tim23, none of the known motor components are copurified. In contrast, when the TIM23 complex is isolated via its core component Tim23, both PAM subunits and Tim21 copurify. Thus, the TIM23 complex exists in two distinct states: a Tim21-bound but PAM-free complex, which is competent for inner membrane sorting but not for import into the matrix and, a Tim21-free but PAM-bound complex that is competent for import into the matrix (Chacinska et al., 2005) (Fig. 4). Recruitment of Tim21 into the complex promotes the dissociation of the PAM complex from the TIM23 complex. In contrast, deletion of TIM21 increases the amount of PAM complex associated to TIM23. Therefore, it appears that at steady state the two different forms of the TIM23 complex exist in equilibrium with each other. But how is the switch between these two forms of the TIM23 complex initiated and how is the dissociation of Tim21 and binding of PAM mediated? Insight into this issue came from the observation that Tim17 is involved in protein sorting into the inner membrane, but also interacts specifically with the IMS domain of Pam18 (Chacinska et al., 2005). It is therefore conceivable that Tim17 recognizes the preprotein to be translocated and then initiates the switch between the PAM-bound and PAM-free form of the TIM23 complex. In this model, Tim17 is actively involved in the sorting of certain preproteins destined for the inner membrane and their lateral release via the PAM-free form of the TIM23 complex (Fig. 4). However, the PAM-free form cannot mediate full translocation of a precursor protein across the inner membrane. Therefore, for the transport of matrix-targeted preproteins, Tim21 is released from the TIM23 core complex and Tim17 interacts with Pam18 to allow the association of the PAM complex to the TIM23 core complex (Fig. 4). The PAM-free form of the TIM23 complex solely uses the electrical potential across the inner membrane as external energy source, which is sufficient for inner membrane sorting.
With the identification of Tim21 a complete chain of physical interactions between import components from the cytosolic surface of mitochondria into the matrix can be assigned. It appears that the TOM–TIM23 and TIM23–PAM interactions are not independent events but represent steps of a reaction cascade that couples the specific roles of TOM, TIM23 and PAM in preprotein translocation across both mitochondrial membranes (Fig. 4). This intricate interaction and signaling cascade directly involves Tom22, Tim21, Tim17, Pam18 and probably further subunits of the TIM23/PAM complex. Remarkably, this reaction chain appears to be regulated by the import signals of the preproteins in transit. Such import signals are, for example, hydrophobic stop-transfer sequences or tightly folded domains within preproteins. Sensing of these signals may be transferred via a signaling cascade of protein interactions to the components downstream in the reaction sequence and thereby priming the TIM23/PAM complex for the tasks to come.
Conclusions and perspectives
The last few years have brought about many new and unexpected insights into the complexity and flexibility of the mitochondrial import machinery. Many novel players in the processes of protein transport have been discovered, some of which play a well-defined and essential role, while others seem to be involved in regulation and optimization of protein import under various conditions. Many intriguing questions about the molecular mechanisms underlying protein sorting into the different mitochondrial subcompartments remain to be addressed. The discovery of the SAM complex and the Mia40/Erv1 system have provided new concepts for outer-membrane and IMS protein biogenesis, respectively. Now that the proteins involved are known, their specific properties and functions have to be defined.
Moreover, it is important that we begin to address how import components of the outer membrane, IMS, inner membrane and matrix communicate and cooperate in order to assure protein import fidelity and accuracy. It appears that differential import signals within preproteins determine the transient assembly of distinct import supercomplexes. How and where these signals are recognized and how such information is transduced to downstream components is not understood. A major aspect will be to get more insight into the molecular mechanisms by which the PAM complex exerts its function. Given the complexity of this unique Hsp70 system and its strikingly dynamic behaviour, one may ask what governs the assembly and exchange of PAM, and how its activity can be so tightly regulated in space and time.
In summary, after several years of searching for new proteins, it is now time to put all the pieces of this puzzle together and try to achieve a deeper understanding of signal transduction and integration processes that control mitochondrial protein import.
Owing to space limitation we have only been able to quote the most recent literature in this review, other important but older publications can be found in the excellent reviews of colleagues we refer to. However, we apologize to those colleagues whose work could not be cited here. We would like to thank Drs C. Meisinger, A. Chacinska, and D. Stojanovski for critically reading the manuscript and helpful discussions. The work of the author's laboratory was supported by the Deutsche Forschungsgemeinschaft, the Sonderforschungsbereich 388. M.v.d.L. is a recipient of an EMBO postdoctoral fellowship.