Plastids are a diverse group of essential organelles in plants that include chloroplasts. The biogenesis and maintenance of these organelles relies on the import of thousands of nucleus-encoded proteins. The complexity of plastid structure has resulted in the evolution of at least four general import pathways that target proteins into and across the double membrane of the plastid envelope. Several of these pathways can be further divided into specialty pathways that mediate and regulate the import of specific classes of proteins. The co-ordination of import by these specialized pathways with changes in gene expression is critical for plastid and plant development. Moreover, protein import is acutely regulated in response to physiological and metabolic changes within the cell. In the present review we summarize the current knowledge of the mechanism of import via these pathways and highlight the regulatory mechanisms that integrate the plastid protein-trafficking pathways with the developmental and metabolic state of the plant.
Plastids are a highly divergent group of organelles that provide essential metabolic and signalling functions within all plant cells . The archetypical plastid is the chloroplast – the organelle that provides the capacity for all plants to perform photosynthesis. Plastids, and consequently the advent of the Plant Kingdom, arose by the endosymbiosis of a photosynthetic cyanobacterium-like prokaryote by a nucleated host cell [1–3]. As the evolution of complex multicellular plants progressed to give rise to specialized cell types and tissues, the endosymbiont also evolved into distinct plastid types to provide specialized metabolic functions for specific cells and tissues. As a result, modern land plants contain at least a half-dozen plastid types with distinct morphologies and functions . Despite this diversity, two unifying principles apply to all plastids. First, plastid differentiation is reversible, allowing plastids to interconvert and differentiate in concert with their cellular host . Secondly, gene transfer to the nucleus has reduced the plastid genome to about 120 genes in land plants . As a result, plastid biogenesis is reliant on the import of thousands of nucleus-encoded proteins (~3500 proteins in Arabidopsis thaliana) after completion of their synthesis on cytoplasmic ribosomes [5,6].
The remarkable diversity of plastids, and the reliance of plastid biogenesis on nucleus-encoded genes, emphasizes the central importance of the protein import and sorting processes in the development and maintenance of these key organelles. Research over the past several years has revealed a complex set of protein-trafficking pathways that have evolved in response to the tremendous diversity in protein substrates and the architectural complexity of plastids themselves [7–10]. At least four general targeting systems have evolved for importing nucleus-encoded proteins into the organelle after synthesis on cytoplasmic ribosomes (Figure 1). This includes one recently discovered pathway in which passage through the secretory pathway serves as an intermediate step in targeting of proteins to plastids [11–13]. In addition, the major import pathway, the TOC–TIC (translocon at the outer envelope membrane of chloroplasts–translocon at the inner envelope membrane of chloroplasts) pathway, actually consists of several distinct pathways that mediate the import of different preprotein subclasses.
Overview of the pathways for targeting nucleus-encoded proteins to plastids
In many cases, import of proteins into the organelle represents only the first step in the trafficking process, because targeting also requires sorting to the proper suborganellar compartment. Chloroplasts, the most structurally complex plastids, contain six compartments that are delineated by the outer and inner envelope membranes and the internal thylakoid membrane. The protein import and sorting systems that operate at the double-membrane envelope contain some core components that appear to have been adapted from proteins present in the original endosymbiont . However, these proteins have clearly evolved to perform new functions in the context of the protein-trafficking pathways. Moreover, numerous unique proteins found only in plants have been added to the translocation systems, giving rise to machineries found only in plastids. By contrast, the protein-targeting systems of the thylakoid membrane are structurally and functionally conserved from those found in prokaryotes . The major difference is the relocation of the targeting systems from their location at the cytoplasmic membrane in bacteria to the thylakoid membrane in chloroplasts.
The present review will focus primarily on the pathways for the import of nucleus-encoded proteins into the organelle and the sorting pathways to the envelope membranes that are associated with the import process. In addition to covering our current knowledge of the mechanism of the import pathways, we will also highlight recent developments in understanding the diversity and regulation of the import pathways as they relate to the metabolic and developmental state of the plant. The conserved pathways for sorting proteins to the thylakoid membrane will not be discussed in detail. There are many excellent reviews that cover thylakoid targeting (see, for example, ), and space limitations prevent us from providing adequate coverage of all aspects of these systems.
THE TOC–TIC TRANSLOCONS
The first discovered and best characterized protein-import system is the TOC–TIC translocon machinery of the plastid envelope. Analysis of the vascular plant genomes suggest that between 2100 and 4000 nucleus-encoded proteins utilize this pathway for targeting to plastids [5,16]. As such, the pathway is estimated to mediate about 90% of the trafficking into the organelle. The TOC–TIC pathway initiates the targeting of proteins destined for all of the plastid subcompartments (Figure 1), including nucleus-encoded thylakoid proteins. Proteins utilizing the TOC–TIC import system are synthesized as preproteins containing a cleavable N-terminal transit peptide of from 30 to more than 100 amino acids in length . The transit peptide is necessary and sufficient for targeting of proteins from the cytoplasm into the stroma of the organelle. In contrast with the amphipathic α-helical presequences of nucleus-encoded mitochondrial proteins, universal structural features that define plastid transit peptides have not been identified. Bioinformatic, structural and mutagenesis studies all have identified key residues within individual transit peptides that are required for function. However, translating the results from these studies to the broader array of transit peptides from other preproteins has proved difficult. Previous studies have demonstrated that the transit peptide receptors of the TOC–TIC translocons bind to distinct, but overlapping, regions of transit peptides . This observation, in addition to the existence of numerous TOC–TIC receptors with distinct transit peptide selectivities in land plants [19–21], suggests a complex molecular-recognition system that might account for the difficulties in defining universal recognition motifs or structural elements that are common to all transit peptides. The analysis of several recently sequenced genomes of green and red algae suggest that these organisms possess much less receptor diversity and transit peptides that are of shorter length on average [22–24]. A re-examination of recognition determinants within transit peptides using examples from these apparently simpler systems might prove to be more fruitful.
The TOC–TIC system consists of two multimeric membrane complexes that mediate the recognition and direct translocation of preproteins from the cytoplasm across the double-membrane envelope into the stroma. This system shares fundamental elements that are characteristic of protein translocation systems that operate at other cellular compartments (e.g. the endoplasmic reticulum and mitochondria) . TOC, the translocon at the outer envelope membrane of chloroplasts, contains a central membrane channel that forms a stable complex with two transit peptide receptors  (Figure 2). The receptors mediate recognition of preproteins by specifically binding to transit peptides and control the initiation of transfer of preproteins into the channel via their intrinsic GTPase activities [27,28]. The TOC translocon directly associates with TIC, the translocon at the inner envelope of chloroplasts, to form an uninterrupted passageway for preproteins from the cytoplasm into the stroma. The TIC translocon also consists of a membrane channel and a set of components that mediate the interaction of preproteins with molecular chaperones as they emerge into the stroma from the TIC channel [7–10]. Molecular chaperones in the cytoplasm, intermembrane space and stroma bind to the preprotein to maintain it in an unfolded state during membrane translocation, provide the driving force for the translocation reaction and ensure proper folding, assembly or suborganellar targeting of proteins once they reach the stroma.
Known components that constitute TOC
The analysis of the energetics of import and the identification of several intermediates in the import reaction have revealed a complex set of molecular interactions that control both preprotein recognition and membrane translocation. The binding and insertion of preproteins at the TOC translocon involves at least four discernible steps (Figure 3). Preproteins initially interact at the TOC receptors in a reversible reaction that does not require nucleoside triphosphate hydrolysis [29,30]. This step has been designated ‘energy-independent binding’ (Figure 3, step 1). Delivery of preproteins to the receptors does not require cytoplasmic factors in vitro; however, a number of molecular chaperones and associated proteins have been proposed to facilitate the targeting of several preproteins to the TOC complex [31–33]. The second stage of binding is promoted by GTP. This step has been designated ‘energy-dependent association’ (Figure 3, step 2) . At this stage, the preprotein remains exposed at the chloroplast surface and appears to be more tightly bound to the TOC GTPase receptors. The non-hydrolysable analogues of GTP and GDP, guanosine 5′-[γ-thio]triphosphate and guanosine 5′-[β-thio]diphosphate respectively, inhibit energy-dependent association and all subsequent steps in the import reaction . These observations suggest that GTP hydrolysis and/or GTP–GDP exchange are required for this step and are indicative of the GTP-regulated molecular switch [18,35–37].
A consolidated model for the function of the TOC GTPase receptors in mediating the recognition and regulating translocation of preproteins at the chloroplast surface
Transfer of the preprotein from the TOC GTPase receptors to the TOC channel (insertion; Figure 3, step 3) requires the hydrolysis of low levels of ATP. At this point, the preprotein has partially penetrated the outer envelope membrane and the transit peptide appears to have completely traversed the TOC translocon. GTP analogues have no effect on subsequent stages in import once the insertion intermediate is formed, indicating that the role of GTP is limited to the initial stages of preprotein interactions with the TOC translocon . Insertion represents the high-affinity committed step in the import reaction, as this intermediate is irreversibly bound to chloroplasts. A subsequent ATP-dependent step (penetration; Figure 3, step 4) brings the preprotein into contact with the TIC translocon [34,38]. This step also has been referred to as the ‘early import intermediate’ , and protease-sensitivity experiments indicate that the preprotein has largely traversed the outer membrane [34,39]. During penetration, the transit peptide binds to a recognition site on the TIC translocon at the stromal face of the inner membrane . Although insertion and penetration both require ATP hydrolysis within the intermembrane space, they are defined by distinct temperature requirements. Insertion occurs at 4 °C, whereas penetration is observed at 25 °C . The molecular basis for the two distinct ATP-dependent steps is unclear. A likely, although untested, possibility is that the two steps represent distinct interactions with an Hsp70 (70 kDa heat-shock protein)-type chaperone that resides within the intermembrane space. The Hsp70-type chaperone is directly associated with early import intermediates and therefore could provide the ATP-dependent driving force for translocation through the TOC channel.
In addition to contributing to the unidirectional movement of the protein across the membrane, the penetration of the preprotein and its interaction with TIC components stimulates the recruitment of an inner-membrane-bound Hsp93 chaperone–co-chaperone system to the exit site . This triggers the transfer of the emerging chain from the TIC translocon to the chaperone, and membrane translocation across the coupled TOC–TIC systems commences with the assistance of Hsp93 at the expense of ATP hydrolysis in the stroma [30,42,43] (Figure 4). The stromal processing peptidase recognizes transit peptides as they emerge from the translocon and mediates cleavage and assists in degradation of the released peptide [44,45].
Known components of TIC
TOC receptor GTPases and preprotein recognition
The TOC receptor system at the surface of plastids is formed by two related outer-membrane GTPases (Figure 2). The receptors are encoded by small gene families in vascular plants, and we will refer to them as the Toc159 and Toc34 families respectively. The Toc159 and Toc34 families are found in all plant and algal genomes from across the phylogenetic tree  and have been shown to be essential proteins in thale cress (Arabidopsis thaliana) [20,21,47–49]. As such, they constitute core components of the TOC translocon. Direct binding and covalent cross-linking studies have demonstrated that the TOC GTPase receptors are responsible for transit peptide recognition at the TOC translocon [19,29,38,50,51].
The TOC receptors define a novel family of GTPases that are unique to plants [28,52,53]. Members of both receptor families contain structurally related G-domains (GTPase domains) and have been shown to hydrolyse GTP in vitro [54,55]. This activity directly implicates them in regulating the GTP-dependent initiation of preprotein translocation (i.e. binding and energy-dependent association; Figure 3). Toc34 consists of an ~30 kDa cytoplasmic G-domain that is anchored to the outer membrane by a short α-helical transmembrane segment [56,57]. Toc159 has a more complex structure. Its G-domain is positioned between an N-terminal acidic region (A-domain) and a C-terminal ~54 kDa membrane-anchor domain (M-domain) [56,58].
Although the X-ray crystal structure of pea (Pisum sativum) Toc34 indicates a conserved Ras-like GTPase fold, there is a major rearrangement of two of the four canonical GTP-binding motifs that are found in other small GTPases (e.g. Ras, signal recognition particle and small G-proteins) . The receptors bind nucleotide with relatively high (submicromolar) affinity, but their intrinsic GTPase activities are very low (kcat<0.5 min−1) [59,60]. These properties are characteristic of regulatory GTPases and suggest that specific exchange factors or GTPase-activating factors might regulate the receptors, thereby interconverting the receptors between active and inactive states.
In the case of the TOC receptors, one of the factors regulating GTP binding and hydrolysis is transit-peptide binding. In vitro binding studies have shown that the G-domains of the receptors interact directly with transit peptides [18,19,51]. Binding stimulates GTP hydrolysis [18,61], providing an important clue to the switch that activates the molecular gate to the translocon. Covalent cross-linking studies demonstrate that the Toc159 and Toc34 receptors interact with preproteins simultaneously during the reversible docking stage . The ability to concurrently interact with preproteins at the chloroplast surface is facilitated by the fact that they bind to distinct regions of transit peptides . Toc159 appears to preferentially bind the N-terminal region, whereas Toc34 interacts with the C-terminal region. Docking is inhibited by non-hydrolysable GTP analogues and is promoted by GTP or GTP+GDP . This suggests that nucleotide exchange or different nucleotide-bound states of the two receptors is required for preprotein binding, a scenario consistent with the energetics of energy-dependent association (Figure 3). This co-ordinated recognition system is likely to be facilitated by the interaction of the two receptors with each other. Biochemical and structural studies demonstrate that the receptors can form homo- and hetero-oligomeric complexes [52,55]. Several investigators have hypothesized that reorganization of receptor–receptor interactions at the translocon by transit-peptide binding and associated GTP exchange/hydrolysis represents the gating mechanism that initiates the membrane translocation reaction [28,53,62].
Receptor dimerization originally was hypothesized also to stimulate GTP hydrolysis [52,53]. However, structural and biochemical studies [60,62] suggest that dimerization has a very modest effect on GTP hydrolysis (~1.5-fold). This is in contrast with the affects of typical GAPs (GTPase-activating proteins) that increase hydrolysis by at least one order of magnitude. Furthermore, the crystal structure of the Toc34 dimer, in both the GDP- and GTP (guanosine 5′-[β,γ-imido]triphosphate)-bound states, indicate that dimerization is not strictly nucleotide-dependent . Although additional GAPs for the TOC GTPases might exist, the current data point to the transit peptide as a key regulator of TOC GTPase function. Interestingly, the structural studies also identified a potential peptide-binding site on the Toc34 GTPase that lies near the nucleotide-binding pocket . The transit peptide is known to bind to the G-domains of both TOC receptors, and the peptide-binding pocket is an excellent candidate for the transit-peptide binding site. This potential peptide-binding site partially overlaps with several residues at the dimer interface in models of the Toc34–Toc159 heterodimer . This leads to the possibility that transit-peptide binding influences both the interaction of receptors with each other and regulates their GTPase activities. As such, transit-peptide binding would control opening of the translocon gate and transfer of the preprotein to the membrane channel.
Although there is a general consensus for the role of TOC GTPase activity in regulating the early stages of import, there has been considerable discussion as to the relative roles of Toc159 and Toc34 in binding and penetration. This has led to two main models for Toc GTPase function [27,37,63]. The first model (targeting or gating hypothesis) proposes that Toc159 and Toc34 function as co-receptors for transit peptides at the chloroplast surface [28,53]. This proposal is consistent with the observation that both receptors cross-link to preproteins at the early docking step in import [29,38]. Transit-peptide binding is proposed to trigger two key events: dissociation of receptor–receptor interactions, which would reveal access to the TOC channel, and stimulation of GTP hydrolysis, which would trigger preprotein dissociation from the receptors and transfer to the channel. Hsp70 binding to the preprotein in the intermembrane space as it inserts across the channel is proposed to trap the preprotein in the translocon, and cycles of binding and release by the Hsp70 ATPase cycle would pull the preprotein across the channel. In this model, the role of Toc159 is primarily in transit-peptide recognition, whereas Toc34 is proposed to provide the GTP-driven push for preprotein transfer to the channel. Support for this model relies on the observation that protein import into isolated chloroplasts or in transgenic plants can occur in the absence of Toc159 GTPase activity, albeit at reduced levels [64,65]. Import into isolated chloroplasts remains sensitive to non-hydrolysable analogues in the absence of the Toc159 GTPase, implicating Toc34 in the initiation of translocation . Furthermore, Toc159 is the principle cross-linking target for preproteins docked at the TOC translocon , and members of the Toc159 receptor family appear to constitute the primary structural and functional difference in TOC complexes that exhibit distinct selectivities for different classes of transit peptides [20,21] (see below).
A second model (motor hypothesis)  proposes that Toc34 functions as the initial transit-peptide receptor. The bound preprotein is proposed to facilitate Toc34–Toc159 interactions and the formation of a ternary complex containing the two receptors and preprotein. This step is supported by the observation that synthetic transit peptides promote the stable association of Toc159 and Toc34 . The ternary complex would serve to pass the preprotein to Toc159 and initiate membrane translocation. Toc159 is envisioned to function as a GTP-driven motor that threads the preprotein into the TOC channel via repeated rounds of GTP hydrolysis. Support for this step comes from reconstitution studies in which a fragment of Toc159 (corresponding to its G- and M-domains) and Toc75 were shown to catalyse partial translocation of a preprotein into proteoliposomes in the absence of Toc34 . However, the demonstration of GTP-dependent protein import into chloroplasts in which the Toc159 GTPase domain has been deleted [64,65] argues against the role of the receptor as a repetitive translocation motor in the intact translocon.
Although these models present very different pictures for the relative roles of the two GTPase receptors in preprotein recognition, several unifying observations are worth emphasizing. First, both models envision the TOC GTPase receptors to function in concert during the initial docking of preproteins at the chloroplast surface. Cross-linking studies using intact chloroplasts demonstrate that both receptors are in close contact with preproteins during their initial docking at the translocon. Secondly, GTP hydrolysis is required to initiate outer-membrane translocation, and the GTPase activities of both receptors are stimulated by transit-peptide binding. Thirdly, once transferred to the TOC channel, complete translocation of the preprotein requires ATP hydrolysis at the Hsp70 chaperone in the intermembrane space. The major difference between the two models revolves around which of the two receptors initiates translocation via a GTPase push. Figure 3 attempts to provide a unifying model that takes into account the major elements of each of the models and the supporting data. We have attempted to reconcile the differences proposed for the GTPase activities of the two receptors by proposing that they act in concert during preprotein recognition. In this model, the combination of the two GTPases would define the specificity of the translocon. Transit-peptide binding would control receptor–receptor interactions and receptor GTPase activity. In turn, GTP binding and hydrolysis at both receptors would regulate translocon selectivity and the initiation of translocation. The model proposes that the GTPase activities of the two receptors are to some extent redundant or overlapping. This would be consistent with the concerted binding observed in intact TOC complexes and explain why translocation across the membrane can occur when one or the other GTPase is absent or inactivated. The resolution of this issue and further insight into the relative roles of Toc159 and Toc34 GTPase activities are likely to come from the analysis of mutants that selectively alter the activities or specificities of one or both of the two receptors in transgenic plants.
Cytoplasmic factors that aid in targeting to the TOC–TIC import system
Several cytoplasmic factors have been implicated in assisting in the targeting of preproteins to the TOC receptors (Figure 2). However, the precise role of these proteins in targeting remains to be clarified, because of contradictory results from in vitro and in vivo studies. A guidance complex containing an Hsp70 chaperone and a 14-3-3 protein has been proposed to facilitate the targeting of preSSU [precursor of the small subunit of rubisco (ribulose-1,5-bisphosphate carboxylase)], to the chloroplast surface . The interaction of the guidance complex with preSSu was shown to be dependent on in vitro phosphorylation of the preSSu transit peptide. In vitro phosphorylation was shown to regulate binding of preSSu to the TOC GTPase receptors [18,61], leading to the proposal that a cycle of phosphorylation/dephosphorylation of the transit peptide was involved in controlling docking and transfer of the preprotein to the translocon. However, it is unclear whether this is a general mechanism to facilitate targeting, because analyses of the phosphorylation model have raised questions about its in vivo relevance [66,67].
Two additional chaperone-associated targeting factors have been implicated in preprotein targeting. A cytoplasmic Hsp70 family member, termed Com70 (chloroplast outer-membrane Hsp70), was shown to associate with the outer-envelope membrane and to bind preproteins early in the import reaction . Unlike the endoplasmic-reticulum or mitochondrial systems, a stimulatory effect of Com70 on preprotein targeting has not been demonstrated. More recently, a role for cytoplasmic Hsp90 in preSSu targeting has been proposed . In vitro binding studies suggest that an Hsp90–preprotein complex docks at the TPR (tetratricopeptide repeat) domains of the outer-membrane protein Toc64. Toc64 is proposed to mediate the transfer of the preprotein from Hsp90 to the Toc34 receptor. However, deletion mutants of the genes encoding Toc64 in the moss Physcomitrella patens  and Arabidopsis  give no apparent phenotypes, and isolated chloroplasts from these mutants have no detectable import defects. Furthermore, clear orthologues of Toc64 are not found in many algal species [22–24]. These observations suggest that the Toc64–Hsp90 targeting function is not essential for protein import, but might represent an accessory function. It is also possible that the lack of effects from deleting the Toc64–Hsp90 or guidance complex targeting pathways in vivo could be due to overlapping or redundant functions.
The channel and outer-membrane translocation
A major component of the TOC membrane channel is formed by the β-barrel protein Toc75, an essential component of the import apparatus  (Figure 2). Toc75 is predicted to contain 16 membrane-spanning strands , and is related to the Omp85 (85 kDa outer-membrane protein) family of β-barrel membrane proteins that are found exclusively in bacterial and mitochondrial outer membranes . Omp85 is required to facilitate the integration of β-barrel proteins into the Escherichia coli outer membrane . Other members of this family include the Tom40 (translocase of outer-membrane 40 kDa subunit homologue) channel protein of the mitochondrial preprotein translocase and the Sam50 (sorting and assembly machinery 50)/Tob50 protein that facilitates β-barrel protein integration into the mitochondrial outer membrane [73,74]. It has been postulated that Toc75 and Tom40 arose from the bacterial Omp85 progenitor by gene duplication and functional diversification at the very earliest stages of endosymbiosis . Toc75 and Tom40 were probably adapted to facilitate protein import in response to gene transfer from the endosymbiont to the host nucleus.
Reconstitution studies demonstrate that Toc75 can form a transit-peptide-activated membrane channel, and conductivity measurements suggest that it can potentially form a channel with a diameter of ~14 Å (1 Å=0.1 nm) . This size is sufficient to allow the passage of a polypeptide chain with minimal secondary structure. Toc75 also is a major cross-linking target of preproteins during membrane translocation [38,76], consistent with its proposed role as a membrane channel. Toc159 and Toc34 form a stable complex with Toc75 in stoichiometric ratios estimated at 1:4:4 or 1:3:3 respectively . The C-terminal M-domain of Toc159 has also been shown to interact with preproteins during membrane translocation . This suggests that the M-domain of Toc159 might interact with Toc75 within the membrane bilayer to form the TOC channel and/or associate with the preprotein in the intermembrane space as the protein exits the TOC channel. Like the mitochondrial and endoplasmic reticulum protein translocation channels, the TOC channel does not appear to be selective with regards to the sequence of the polypeptide substrate. However, folded substrates are not efficiently transported across the outer membrane .
Preprotein insertion into the TOC channel is initiated by GTP hydrolysis at the TOC receptors. However, subsequent translocation across the membrane is insensitive to non-hydrolysable GTP analogues and relies exclusively on the hydrolysis of ATP . An Hsp70 chaperone has been shown to associate with import intermediates in the intermembrane space , and the ATP-dependence of outer-membrane translocation has been attributed to the activity of this chaperone. A small J-domain (DnaJ-like domain) protein, Toc12, has been shown to associate with the core TOC components and Hsp70. Toc12 is an integral membrane protein with its J-domain extending into the intermembrane space of the envelope . It has been proposed to function in localizing the Hsp70 to the exit site of the TOC translocon, thereby ensuring that preproteins are prevented from folding or mislocalizing to the intermembrane space. Toc12 could also serve as an anchor for Hsp70, allowing it to function as the molecular driving force for translocation.
THE TIC COMPLEX ASSOCIATES DIRECTLY WITH TOC TO PROVIDE A DIRECT PATHWAY TO THE STROMA
TOC and TIC complexes physically associate with one another at membrane contact sites of the chloroplast envelope to facilitate translocation into the organelle  (Figure 4). The precise mechanism of this interaction has not been defined, but the so-called TOC–TIC supercomplexes are detected in the absence of preprotein translocation. This indicates that components of the import machinery directly mediate TOC–TIC association. Several factors have been proposed to co-ordinate the interaction between TOC and TIC complexes. Tic22, a peripheral protein located in the intermembrane space , and Toc64 have been shown to interact , leading to the proposal that they mediate interactions between the outer and inner membranes during import. The M-domain of Toc159 extends into the intermembrane space and also has been implicated in the interaction between TOC and TIC complexes. Although the precise mechanism of contact site formation is unknown, the formation of TOC–TIC supercomplexes appears to be regulated because their abundance increases in relation to the flux of protein import .
Analysis of TOC and TIC complexes by blue native gel electrophoresis and sucrose-gradient centrifugation demonstrated that the components associate into very large multimeric assemblies [83–85]. TOC complexes, containing Toc34, Toc159 and Toc75, were shown to be 800–900 kDa in size, and TOC–TIC supercomplexes were ~1400 kDa . Kinetic analysis of protein import demonstrated that preproteins initially dock at the 800 kDa TOC complex before engaging the larger TOC–TIC supercomplexes , consistent with preprotein triggered association of the two translocons. A smaller, ~500 kDa, TOC complex has been isolated and shown to mediate minimal membrane translocation when reconstituted into proteoliposomes . Negative-staining electron microscopy of this complex suggested the existence of up to four membrane pores per complex. These data suggest that each TOC complex can potentially mediate the simultaneous transport of several preprotein substrates.
Composition of the TIC channel remains elusive
Unlike the TOC channel, the assembly of functional TIC complexes appears to be much more dynamic. As such, defining the precise roles of various TIC components has been more challenging. At least three components, namely Tic20, Tic21 and Tic110, have been proposed to function in forming the channel of the TIC translocon (Figure 4). Tic20  and Tic21  are similar hydrophobic inner-membrane proteins consisting of four α-helical transmembrane segments. Tic20 covalently cross-links to preproteins during translocation across the inner membrane. Genetic studies have revealed that a null mutant of the TIC20 gene exhibits a severe albino phenotype  and that the reduction in Tic20 levels by small interfering RNA  in Arabidopsis specifically disrupts preprotein translocation across the inner membrane.
Unlike other components, Tic21 was identified by a forward genetic screen in Arabidopsis for mutants that affect protein import . Subsequent genetic studies suggest that Tic21 and Tic20 have functional similarity. Tic21 appears to be more important for later stages of leaf development, whereas Tic20 functions earlier in development, suggesting that they might represent differentially expressed, but functionally overlapping, TIC components . Although no significant sequence similarity was found between Tic20 and Tic21, they are similar in size and topology. One of the most compelling, albeit indirect, arguments for Tic20 and Tic21 as channels is their distant relationship to the Tim17/23 channels of the mitochondrial inner-membrane import machinery . Tim17 and Tim23 also are small hydrophobic proteins and are known to form the major constituents of the TIM (translocase of the mitochondrial inner membrane) channels. Both the TIC and TIM proteins appear to have evolved from metabolite (i.e. amino acid) transporters of bacterial cytoplasmic membranes. Tic21 has been also proposed to act as iron transporter and might have a dual function within plastids .
Tic110 also has been implicated as a component of the TIC channel. This hypothesis was based on the observation that a denatured fragment of Tic110 formed a β-barrel-type channel when reconstituted into proteoliposomes . However, subsequent in vitro and in vivo studies indicate that the fragment corresponds to the soluble stromal domain of Tic110 . Therefore it appears unlikely that the stromal domain forms a β-barrel-type channel, as proposed in the original study. Both Tic20 and Tic21 are detected in TOC–TIC supercomplexes in association with Tic110 [81,86]. Therefore it is possible that all three components participate in the formation of the TIC channel.
TIC translocation and preprotein maturation
Although a direct role for Tic110 in forming the translocation channel has not been demonstrated, it has been shown to play a key role in co-ordinating the stromal events during protein import. Among TIC components, Tic110 is the most abundant TIC protein and is encoded by a single gene in all species examined . The exception is the moss P. patens, which has two copies, apparently the result of a genome duplication event. Tic110 is essential for Arabidopsis viability [91,92], highlighting its essential role in plastid biogenesis and protein import. It consists of two short α-helical transmembrane segments at its N-terminus and a large (approx. 95 kDa) stromal domain . It covalently cross-links to preproteins during import and directly binds to the transit peptide via a site on its ~95 kDa stromal domain that lies adjacent to the inner face of the inner membrane . This transit-peptide-binding site provides the trans-docking site for transit peptides at the inner membrane as they emerge from the TIC channel. This ensures the unidirectionality of the translocation reaction and triggers the recruitment of molecular chaperones that will assist in TIC translocation, folding or suborganellar targeting.
Translocation across the inner envelope requires stromal ATP in isolated chloroplasts or can be driven by light, consistent with the involvement of stromal chaperones. Although several chaperones have been shown to associate with the translocon machinery, a stromal AAA+ (ATPases associated with a variety of cellular activities)-type ATPase, Hsp93, is found to associate with Tic110 stoichiometrically . Hsp93 is highly homologous with the chaperone components of the bacterial-type Clp (caseinolytic proteinase) proteases that bind, unfold and translocate target substrates into protease subunits for degradation . Hsp93 has been shown to interact with preproteins during translocation into chloroplasts, confirming a role in the import process . These observations have led to the proposal that ATP hydrolysis by Hsp93 is most likely to provide the driving force for translocation across the inner envelope membrane. Consistent with this hypothesis, knockout mutants of the major form of plastid Hsp93 have defects in protein import and chloroplast biogenesis.
Another inner-membrane protein, Tic40, has been shown to participate with Tic110 and Hsp93 during TIC translocation . Tic40 consists of a single, N-terminal transmembrane segment, followed by stromally exposed TPR and Hip/Hop (Hsc70-interacting protein/Hsc70/Hsp90-organizing protein) domains. The presence of the TPR and Hip/Hop domains suggest that Tic40 functions as a co-chaperone at the TIC translocon. The Hip/Hop domain of Tic40 is functionally interchangeable with that of human Hip proteins . Most importantly, it stimulates ATP hydrolysis by Hsp93, indicating that it indeed functions as a co-chaperone. Null mutants of Tic40 in Arabidopsis are viable, but have significant defects in chloroplast biogenesis , demonstrating that it functions as a critical regulator of the TIC translocon.
Biochemical studies suggest that Tic110, Tic40 and Hsp93 form a ternary complex in vivo [92,96]. Genetic studies also suggest that they may function in close association with each other and the proper stiochiometric expression of these three proteins is important for proper chloroplast biogenesis . Further investigation of these interactions has provided striking evidence that Tic110 and Tic40 co-ordinate the association of preprotein with Hsp93 chaperone to drive import and maturation events, leading to a model for the events that drive TIC translocation . As summarized in Figure 5, the transit peptide first binds to the N-terminal part of Tic110 as it emerges from the TIC channel. This binding induces a conformational change in Tic110 and allows the Tic40 TPR domain to associate with Tic110. The interaction between Tic110 and Tic40 results in the release of the transit peptide from Tic110. This interaction also unshields the Hip/Hop domain of Tic40, allowing it to stimulate ATP hydrolysis of Hsp93. ATP-driven cycles of Hsp93 binding to the preprotein would drive membrane translocation and account for the stromal ATP requirement for import.
A model for the co-ordination of chaperone-driven translocation of preproteins at the TIC complex
The involvement of Hsp93 in chloroplast protein translocation is unique to chloroplasts. This mechanism is somewhat more analogous to the role of the p97 AAA+-family ATPase in the retrotranslocation of misfolded proteins from the endoplasmic recticulum during endoplasmic-recticulum-associated degradation  than the Hsp70-mediated translocation events in mitochondria. In this comparison, Tic110 and Tic40 in chloroplasts and VIMP [VCP (valosin-containing protein)-interacting membrane protein] in the endoplasmic reticulum could act as analogous platforms for the AAA+ ATPases to pull the proteins across their respective membranes. In contrast with mitochondria and the endoplasmic reticulum , a role for Hsp70 family members in TIC translocation has not been conclusively demonstrated.
REGULATION OF TOC–TIC TRANSLOCON FUNCTION IN RESPONSE TO DEVELOPMENTAL AND PHYSIOLOGICAL STATUS
Recent studies have demonstrated the involvement of the plastid protein-import apparatus in regulating and responding to long- and short-term changes in plant developmental and physiological status. Plastids have evolved structurally and functionally distinct TOC–TIC translocons to generate multiple pathways for import of different classes of preprotein substrates . These pathways appear to be important in maintaining organelle homoeostasis when the gene-expression profiles of plastid proteins change during plant and plastid development. The co-ordination of protein import with changes in gene expression appears to be critical for differentiation and maintenance of distinct plastid types in different tissues. In addition to the presence of diverse import pathways, TOC–TIC translocons also have the ability to adapt to acute changes in metabolic and physiological states, such as those that accompany changes in photosynthetic activity (e.g. redox state) .
Functionally distinct TOC complexes are required for plastid biogenesis
The analysis of available plant genomes indicates that the Toc34 and Toc159 receptors are encoded by small gene families in all land plants. In Arabidopsis, the Toc34 family is encoded by two genes, atTOC33 and atTOC34, and the Toc159 family is encoded by four genes, atTOC159, atTOC132, atTOC120 and atTOC90 . AtToc33 and atToc159 are considered to be the orthologues of pea Toc34 and Toc159 respectively. Each of these genes appears to be differentially expressed. For example, atToc159 and atToc33 are the predominant receptor forms in photosynthetic and developing tissues that contain chloroplasts, whereas atToc34, atToc132 and atToc120 display relatively uniform expression throughout development and in different plastid types [20,21,102]. These observations led to the proposal that the distinct receptor types might play a role in differential protein import during plastid development [47,48], and the accumulating evidence supports this hypothesis.
The Toc159 family appears to encode functionally distinct receptors. An atTOC159-null mutant, ppi2, lacks developed chloroplasts in leaf tissue but has normal plastids in roots, consistent with its expression predominantly in green tissues . Furthermore, ppi2 plants fail to accumulate many of the normally abundant photosynthetic proteins, whereas many other plastid proteins accumulate normally in ppi2 plastids. In vivo targeting and in vitro binding studies indicate that atToc159 has a binding specificity for the transit peptides of photosynthetic preproteins . On the basis of these data, it has been proposed that the atToc159 receptor defines translocons that mediate the import of a subset of highly expressed preproteins that are required for chloroplast development.
Additional evidence supports the hypothesis that the two other members of the atToc159 family, namely atToc132 and atToc-120, mediate the import of constitutively expressed housekeeping proteins. These genes are expressed at a uniformly low level such that they are relatively prominent in non-photosynthetic tissues [20,21]. Indeed, atToc132 preferentially binds to non-photosynthetic proteins in vitro . Genetic studies suggest that the functions of atToc132 and atToc120 are largely redundant, and double-knockout mutants result in either embryo or seedling lethal phenotypes [20,21]. This is consistent with the proposal that these proteins are involved in the import of housekeeping proteins. The role of atToc90 is less clear, but null mutants augment the albino phenotype of the atToc159 ppi2 mutant, suggesting a complementary role for the two receptors in the targeting of photosynthetic proteins .
AtToc159, atToc132 and atToc120 also form distinct TOC complexes that include a common channel component, atToc75 . Immunoprecitation studies suggest that atToc33 and atToc34 differentially assemble with the atToc159 family members. Although genetic studies demonstrate that the two Toc34 isoforms are functionally redundant, they do appear to have distinct preprotein binding characteristics [48,49]. Furthermore, some species, such as rice (Oryza sativa), contain multiple Toc159 genes, but only one Toc34 gene. Therefore the precise role of the Toc34 family members in defining functionally distinct translocons remains to be clarified. Nonetheless, it is clear that they differentially associate with the Toc159 family members and are likely to contribute to the selectivity of these complexes in the context of the translocon. Figure 3 incorporates the function of the distinct TOC GTPases by proposing that the combination of different Toc159 and Toc34 family members define the selectivity of the TOC translocon for specific subclasses of preproteins (Figure 3, step 1). This model envisions that the GTPase activities of the receptors regulate this selectivity in addition to promoting the initiation of translocation. It is intriguing to speculate that the multiple possible arrangements of TOC receptors could generate a variety of highly selective translocons in a mechanism similar to the combinatorial model of promoter recognition that uses a limited number of transcription factors to generate transcriptional complexes of defined selectivity.
Analyses of mutants lacking the Toc159 family have also suggested that the protein-import pathway is tightly linked to the transcription of plastid genes in the nucleus. The expression of nuclear genes encoding photosynthetic proteins as well as the accumulation of photosynthetic proteins is significantly impaired in the ppi2 mutant . By contrast, the majority of down-regulated genes in a mutant lacking Toc132 alone encodes non-photosynthetic proteins, suggesting that each mutant has a unique transcriptional response . Several studies have also suggested that the expression of genes encoding translocon components is up-regulated in these mutants [21,104]. This response might represent a compensatory mechanism during defective protein import.
Intriguingly, the proliferation of multiple Toc159 genes appears to correlate with the advent of morphologically and functionally distinct plastid types in land plants. For example, higher plants have at least four copies of TOC159 genes, whereas the motile unicellular green alga Chlamydomonas has only one Toc159-type receptor . Although the validity of this correlation is limited by the small number of complete plant genomes, it is tempting to speculate that expansion of the Toc159 gene family accompanied the evolution of distinct plastid types by providing a mechanism to regulate the import of subclasses of proteins encoded by developmentally regulated genes. The existence of the multiple pathways would be essential to facilitate organelle development by maintaining the proper balance of housekeeping (e.g. amino-acid- and lipid-synthetic enzymes) as against specialized proteins (e.g. photosynthetic proteins). The necessity to regulate and co-ordinate import via multiple pathways might also account for the evolution of the complex GTP-dependent molecular switch at the TOC translocon (Figure 3). In addition to discriminating between plastid as against non-plastid proteins, this switch would also be important in regulating the selectivity of the distinct translocons for different classes of plastid preproteins. It remains to be determined whether the composition of TIC complexes also varies in conjunction with TOC complexes. Tic110 appears to be a common component of both pathways, but the differential expression of Tic20 and Tic21 suggests that they might also play roles in the differential import of plastid protein subclasses.
Regulation in response to metabolic state
In addition to the regulation of import via the generation of distinct translocon pathways, import also appears to be regulated directly in response to metabolic and diurnal changes. A set of redox proteins (Tic62, Tic55 and Tic32) has been shown to associate with TIC complexes [105–108] (Figure 2). Indeed, the import of ferredoxin–NAD(P)+ oxidoreductase  and ferredoxin-III  is strictly regulated by light via the redox state of chloroplasts. Therefore a redox mechanism might serve to co-ordinate import activity in response to metabolic activities of the organelle. Redox regulation might also involve Ca2+ and calmodulin. Evidence has been provided that Tic32 is both an NADPH-dependent dehydrogenase and a calmodulin-binding protein . In vivo analysis of these components suggests that Tic55 may not play essential roles in import , whereas atTIC32 is essential for plant viability . Further investigation should provide insight into the function of these components in vivo and the mechanism by which they associate and regulate TIC function.
Another unique regulatory mechanism is illustrated by the import of the photo-activated enzyme PORA (protochlorophyllide oxidoreductase A). PORA catalyses a critical step in chlorophyll biosynthesis, and its import is regulated by its substrate availability within plastids . The substrate-dependence of pre-PORA import appears to be encoded within its transit peptide  and also is regulated in an organ-specific manner . Accumulation of POR protein is significantly reduced in mutants carrying defective TOC–TIC components, and Toc34 has been identified as a component of the prePORA import pathway . However, an OEP (outer-envelope protein), OEP16, has been proposed to function as a pore for PORA import rather than the Toc75 channel . Intriguingly, OEP16 has been also shown to function as an amino acid transporter , suggesting that OEP16 might perform dual functions. It remains to be seen whether or not the pathway for prePORA import is unique to this one enzyme.
NOVEL IMPORT PATHWAYS HIGHLIGHT THE COMPLEXITY OF PLASTID BIOGENESIS AND FUNCTION
Protein targeting to the outer-envelope membrane
The majority of proteins destined for the outer-envelope membrane of chloroplasts lack cleavable transit peptides (Figure 1). This probably corresponds to approx. 24 proteins. For outer-membrane proteins that contain typical α-helical transmembrane segments, the targeting signals for membrane integration appear to be localized within, and adjacent to, their transmembrane regions. The targeting of Toc34 , Toc64  and OEP7/14 [119,120] have been studied in most detail. None of these model proteins utilize the standard TOC–TIC system, and their integration appears to occur without an input of external energy. Analysis of the targeting determinants of Toc64 indicate that sequences adjacent to the transmembrane domain acts as a Sec translocon avoidance signal, thereby preventing the Toc64 transmembrane segment from functioning as a signal sequence . Very recently, a cytoplasmic AKR (ankyrin repeat) protein, AKR2A, has been shown to function as a cytosolic mediator for sorting and targeting of outer-envelope membrane proteins to the chloroplast . The elegant combination of genetic and cell-biological approaches in Arabidopsis demonstrated that AKR2A interacts with the OEP7/14 targeting signal and mediates delivery to the chloroplast surface. An akr2a mutant exhibited defects in chloroplast protein accumulation and chloroplast biogenesis, confirming a key role for this targeting factor in outer-membrane biogenesis. Separate work showed that OEP7/14 cross-links to Toc75 during its integration into isolated chloroplasts . Furthermore, the protein has been shown to integrate into chemically pure proteoliposomes containing only reconstituted Toc75, but not liposomes alone . These data suggest that Toc75 serves as the integration channel for outer-membrane proteins with α-helical transmembrane spans. The involvement of Toc75 in multiple distinct outer-membrane insertion/translocation events is consistent with the observation that Toc75 exists in significant molar excess over the TOC receptors and TOC complexes.
Much less information is available on the integration of β-barrel outer-membrane proteins. Plastids contain a homologue of the Omp85 protein, termed OEP80, which is involved in β-barrel protein biogenesis in bacteria, and it has been shown to be essential in Arabidopsis . However, direct evidence for its role in protein targeting/integration has not been demonstrated. Plastid β-barrel proteins lack cleavable transit peptides and they do not appear to utilize the standard TOC–TIC system for targeting. Therefore it appears that the integration pathway may not be analogous to that in mitochondria, in which β-barrel proteins initially engage the TOM (translocase of outermembrane) and are passed to an Omp85 orthologue, Sam50/Tob50, for integration .
Toc75 is the notable exception among β-barrel proteins. It is synthesized with an N-terminal transit peptide, and the initial events in targeting involve the TOC–TIC system . The transit peptide is followed in tandem by a glycine-rich sequence that inhibits complete translocation of Toc75 across the envelope . This polyglycine signal triggers the integration of the protein into the outer membrane via an unknown mechanism that might involve the TOC channel. A signal peptidase similar to those operating at bacterial and thylakoid membranes has been demonstrated to localize at the inner-envelope membrane and be essential for removing the glycine-rich signal, yielding mature Toc75 .
The Toc159 receptor also appears to utilize a unique targeting pathway. It is synthesized without a transit peptide, but its membrane-anchor domain appears to integrate into the membrane via the assistance of the TOC translocon. In vitro and in vivo analysis of Toc159 targeting in isolated chloroplasts and reconstituted proteoliposomes indicates that it requires both Toc34 and Toc75 for targeting [55,126,127]. Toc34 appears to be the docking site for cytoplasmic Toc159 at the membrane, and the Toc75 channel appears to be required for membrane integration. This unusual mechanism of membrane insertion is probably due to the unusual nature of the Toc159 membrane anchor. The M-domain alone can assemble into TOC complexes with low efficiency when expressed in transgenic plants , and it competes with targeting of the full-length receptor in vitro . No discernable transmembrane segments or β-barrel-type structure is predicted in the M-domains of Toc159 from any species. Therefore its association with the membrane is likely to depend upon its assembly into the TOC translocon. A short hydrophobic region is present at the C-terminal region of the protein. Although it is insufficient to serve as a transmembrane segment, it could interface with the core of the lipid bilayer in the context of the TOC complex, thereby accounting for the inability to extract Toc159 from the membrane with aqueous perturbants.
Protein targeting to the intermembrane space
In contrast with the abundance of key electron-transport components that are localized to the intermembrane space in mitochondria, very few proteins have been localized to the compartment between the outer- and inner-envelope membranes in chloroplasts. In fact, the targeting of only two proteins to the intermembrane chloroplast space, MGD1 [MGDG (monogalactosyldiacyl glycerol) synthase] and Tic22, has been studied in detail [128,129]. Both proteins are synthesized with N-terminal cleavable targeting signals that resemble transit peptides. Original studies on pre-Tic22 targeting suggested that it utilized a TOC–TIC-independent pathway for targeting across the outer membrane . However, more recent studies provided evidence that pre-Tic22 and pre-MGD1 interact with the TOC machinery to initiate translocation across the outer membrane  (Figure 1). Interestingly, the targeting of the two proteins appears to diverge within the intermembrane space. Pre-Tic22 appears not to engage the TIC machinery and is targeted directly to the intermembrane space, where it is processed to its mature form. In contrast, pre-MGD1 does appear to partially access the stroma and be processed by the stromal processing peptidase before release to the intermembrane space.
Protein targeting and integration at the inner-envelope membrane
The targeting of proteins to the chloroplast inner-envelope membrane appears to involve several pathways. The majority of inner-membrane proteins appear to be synthesized with transit peptides, and they utilize the TOC–TIC system for the initial steps in targeting to the organelle (Figure 1). However, at least two pathways for the insertion of inner-membrane protein with transit peptides appear to exist. The first mechanism appears to be a stop-transfer mechanism that triggers lateral insertion of transmembrane helices into the membrane as they pass through the TIC channel. The data from in vitro import studies of several polytopic membrane proteins are consistent with a stop-transfer mechanism [130–132]. The stop-transfer signals have not been well defined, but probably involve one or more transmembrane helices and adjacent residues.
In addition to the stop-transfer mechanism, a number of inner-membrane proteins, including Tic40 and Tic110, have been shown to utilize a ‘post-import’ targeting mechanism for inner-membrane insertion [133–135]. Tic40 and Tic110 have been shown to utilize soluble stromal intermediates en route to the inner membrane. Insertion of these intermediates is not obligatorily coupled to the import process and appears to require proteinaceous factors at the envelope . The components of the post-import pathway have not been identified, but probably represent a unique protein-targeting system. An analogous pathway, termed the Oxa1p pathway, has been identified in mitochondria for the insertion of mitochondrially and several nucleus-encoded inner-membrane proteins [136–139]. This so-called conservative pathway is also found in chloroplasts and prokaryotes. The chloroplast Oxa1p homologue Alb3 functions in the insertion of a subset of membrane proteins at the thylakoid membrane . An Oxa1p homologue has not been detected in proteomic studies of the chloroplast envelope [5,140–143]. Therefore it is not clear whether a conservative sorting machinery similar to the Oxa1p protein functions at the inner envelope, and the nature of the post-import pathway machinery remains to be addressed. The mechanism by which inner-membrane and thylakoid-membrane proteins are distinguished, and why the stop-transfer and post-import mechanisms are both required for inner-membrane targeting, remain fascinating questions. The topology and/or size of domains residing within the intermembrane space or stroma are probable factors that will dictate the targeting pathway.
Transit-peptide-independent targeting to plastids
A survey of the plastid proteome identifies 100 or more plastid proteins that appear to lack cleavable transit peptides (Figure 1). This observation reveals the existence of novel pathways for protein import and targeting in plastids that were not apparent previously. The import of two proteins of this class has been studied in some detail. Tic32 is an inner-membrane component proposed to be involved in the regulation of the TOC–TIC pathway and is synthesized on cytoplasmic ribosomes as a mature-sized polypeptide . The protein can be imported into isolated chloroplasts and correctly targeted to the inner membrane, indicating that the mature polypeptide encodes all the information necessary for proper localization. Deletion analysis suggests that regions within the N-terminus of the protein are required for targeting, but a precise targeting signal has not been mapped in detail. Tic32 import is not competed by standard TOC–TIC substrates, suggesting that it uses a distinct translocon for entry into the organelle. Tic32 import does require ATP, but the components of this novel translocon have not been identified.
A second protein, ceQORH (chloroplast envelope quinone oxidoreductase), has also been identified as a substrate of the transit-peptide-independent import pathway [145,146]. ceQORH is a peripheral membrane protein that is bound to the stromal face of the inner-envelope membrane. Deletion analysis of the polypeptide reveals a relatively complex set of targeting determinants that are not localized to a single contiguous region within the polypeptide. As with Tic32, ceQORH appears not to utilize the standard TOC–TIC translocon for import. It is notable that the TOC–TIC substrates used in the competition assays in the studies of Tic32 and ceQORH targeting are all postulated to use the Toc159 TOC pathway. It is therefore possible that the transit-peptide-independent substrates could utilize one of the alternative TOC translocons (e.g. Toc132/120 or Toc90).
Protein import involving the secretory pathway
The diversity of plastid protein-trafficking pathways was recently expanded by the remarkable discovery of protein delivery to plastids via the secretory pathway [11–13] (Figure 1). Investigations into the function of an α-amylase and a nucleotide pyrophosphatase/phosphodiesterase in rice and CAH1, an α-type CAH (carbonic anhydrase) in Arabidopsis, led to the surprising discovery that the precursors of these glycoproteins contained endoplasmic-reticulum signal sequences, but were localized to the chloroplast stroma [11–13]. These three studies utilized an elegant combination of cell-biological techniques, including the analysis of trafficking of green-fluorescent-protein fusion proteins in the presence of pharmacological agents that disrupt Golgi trafficking, to provide conclusive evidence that these proteins pass through the secretory pathway en route to plastids. Additional studies identified several other stromal proteins that carried the hallmark glycosylation associated with passage through the Golgi apparatus [12,13].
It has long been known that protein targeting to complex plastids in certain algae and apicomplexans (unicellular spore-forming animal parasites) involves routing through the secretory pathway . The complex plastids in these organisms arose from a secondary endosymbiotic event in which a photosynthetic eukaryote was engulfed by another nucleated cell, resulting in a plastid that is surrounded by a remnant of the endosymbiont's plasma membrane. However, green algae and flowering plants arose from primary endosymbiosis of a cyanobacterial ancestor and therefore do not contain this fourth membrane system. In plants, the double membrane envelope corresponds to the outer membrane and cytoplasmic membranes of Gram-negative bacteria. Although it is clear that the secretory pathway for protein targeting exists in green plants, the mechanism of trafficking via this system has not been defined . It remains to be seen whether protein targeting involves vesicle trafficking, a process that would require fusion of Golgi-derived vesicles with the outer-envelope membrane. Subsequent targeting of cargo across the inner membrane could involve the TIC complex, an unknown translocon or a second vesicle-trafficking process . In this case, specific trafficking factors [e.g. V-SNARES (vesicle soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptors), T (target organelle)-SNARES, coat proteins and Rab proteins] would be expected to mediate this pathway. Alternatively, it is possible that the glycoproteins are directly transported into plastids after being retrieved from the Golgi and retrotranslocated out of the endoplasmic reticulum into the cytoplasm. CAH1 was shown not to import into isolated chloroplasts in vitro, favouring the model of vesicular trafficking . Regardless of the specific mechanism, the discovery of the pathway adds another notable example of the diverse array of pathways that have evolved to mediate protein trafficking to plastids.
The diversity of protein targeting pathways in plastids and their regulation in response to developmental and metabolic status reflects the central role of plastids in cellular function. Considerable work remains to define the molecular mechanisms of each component of this complex network of trafficking pathways. Although we have a general outline of the molecular activities that constitute the TOC–TIC system, many key questions remain to be answered. In particular, the role of the TOC GTPase receptors in regulating protein translocation into the organelle and the molecular basis of the specificity of distinct TOC translocons remain to be resolved. Another major issue is the molecular nature of the TIC translocon. Although it is commonly referred to as the ‘TIC complex’, a stable membrane complex comparable with the TOC complex or the TOM–TIM complexes in mitochondria has yet to be observed. This raises the possibility that multiple distinct TIC translocons containing some shared components (e.g. Tic110) might arise in response to the import of specific substrates, such as stromal or inner-membrane proteins. This situation would be analogous to the distinct TIM complexes that mediate stromal and inner-membrane carrier-protein import in mitochondria.
Even less is known about the molecular details of the other trafficking pathways that operate at the plastid envelope. It remains to be determined whether or not these distinct pathways might share some common components, as is suggested by studies on targeting to the outer membrane. Regardless, genomic and proteomics analyses clearly demonstrate the potential for these pathways to mediate the targeting of hundreds of proteins to plastids, and it is clear that the definition of these pathways is essential to understanding the biogenesis of these essential organelles.
Perhaps one of the most exciting new areas in the plastid biogenesis is the discovery of trafficking pathways that are required for the proper development and differentiation of plastids in different tissues. Defining the precise roles of the multiple TOC translocons and their potential roles in co-ordinating protein import with changes in gene expression opens new avenues to understanding the role of protein import in plant development. Connecting these developmentally regulated import pathways with the signalling pathways that mediate plastid–nuclear communication will also be interesting. Similarly, the regulated import of specific proteins in response to the metabolic or physiological state of the cell further dispels the concept of protein import as a basic housekeeping function. It is now clear that the multiple pathways for protein trafficking to this diverse array of organelles are as tightly regulated as any metabolic pathway and as responsive to cellular cues as many signalling pathways.
This work was supported by NIH (National Institutes of Health) grants GM061893 and GM063879 to D. S., the 21st Century Centers of Excellence Program and Grant-in-Aid for Young Scientists 18780247 to T. I., and a joint Invitation Fellowship for Research from the Japan Society for the Promotion of Science S07133 to both authors.
ATPases associated with a variety of cellular activities
chloroplast envelope quinone oxidoreductase
chloroplast outer-membrane Hsp70
MGDD (monogalactosyldiacylglycerol) synthase
85 kDa outer-membrane protein
protochlorophyllide oxidoreductase A
precursor of the small subunit of rubisco (ribulose-1,5-bisphosphate carboxylase)
soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor
translocon at the inner envelope membrane of chloroplasts
translocase of the mitochondrial inner membrane
translocon at the outer envelope membrane of chloroplasts
translocase of mitochondrial outer membrane
translocase of mitochondrial outer membrane 40 kDa subunit homologue