PLASTIDS — such as chloroplasts, which carry out photosynthesis in the green parts of plants — are present in every plant cell. Chloroplasts originated from an endosymbiotic event, in which an ancestral photosynthetic cyanobacterium was taken up by a heterotrophic host cell that already contained mitochondria1, 2. This endosymbiotic process led to a massive transfer of genetic information from the endosymbiont to the emerging host nucleus. A prerequisite for the successful completion of this process was the development and establishment of protein-import machinery for the import of chloroplast-localized, but now cytosolically synthesized, polypeptides. As chloroplasts are the most recent organelle to be added to the eukaryotic cell, several post-translational protein-targeting systems probably already existed in the host cell, for example, those belonging to mitochondria3, peroxisomes4 and the plasma membrane5. Therefore, the arising chloroplast protein-import system had to develop unique features to ensure organelle specificity, as the correct sorting of proteins in a eukaryotic cell is essential for its functionality.
The chloroplast proteome is estimated to consist of 3,500–4,000 polypeptides, whereas the coding capacity of the chloroplast genome rarely exceeds 200 genes6, 7. Newly imported polypeptides are frequently integrated into protein complexes that also contain proteins that are synthesized inside the organelle. The most prominent examples are the two photosystems, the ATP synthase and the CO2-fixing enzyme ribulose-1,5-biphosphate carboxylase oxygenase (Rubisco). Because of this dual genetic origin, there must be a tight coordination of transcription, translation and protein import between the parent cell and the organelle. Only the tight coordination of these processes in time, space and quantity ensures the successful biogenesis of the organelle.
Chloroplasts are highly structured and contain three distinct membrane systems — the outer- and inner-envelope membranes, which surround the organelle, and the thylakoid membrane network, which contains the photosynthetically active protein complexes (Box 1). In addition, three soluble subcompartments — the space between the envelope membranes, the stroma and the thylakoid lumen — can be clearly distinguished. So, in addition to the targeting and import systems that are the main focus of this article, several intra-organelle sorting and transport systems must also be present in chloroplasts8, 9, 10.
Although chloroplasts are thought to originate from a unique primary endosymbiotic event, many algae contain photosynthetically active chloroplasts that are surrounded by three or four membranes. These chloroplasts are called complex plastids and they originated from a secondary endosymbiotic event in which a photosynthetic eukaryotic cell was taken up by a non-photosynthetic eukaryotic host cell. Relic non-photosynthetic complex plastids called APICOPLASTS are also present in several intracellular parasites such as Plasmodium falciparum and Toxoplasma gondii, which are the causative agents of malaria and toxoplasmosis, respectively. Protein targeting and translocation into plastids is therefore of broad interest11.
A general outline of chloroplast protein import
The vast majority of chloroplast proteins are synthesized as precursor proteins (preproteins) in the cytosol and are imported post-translationally into the organelle. Most proteins that are destined for the thylakoid membrane, the stroma and the inner envelope are synthesized with an amino-terminal extension called a presequence, or transit sequence, which is proteolytically removed after import (Fig. 1). The transit sequence is both necessary and sufficient for organelle recognition and translocation initiation. Preproteins that contain a cleavable transit peptide are recognized in a GTP-regulated manner12 by receptors of the outer-envelope translocon, which is called the TOC complex13. The preproteins cross the outer envelope through an aqueous pore and are then transferred to the translocon in the inner envelope, which is called the TIC complex. The TOC and TIC translocons function together during the translocation process (Fig. 2). Completion of import requires energy, which probably comes from the ATP-dependent functioning of molecular chaperones in the stroma14. The stromal processing peptidase then cleaves the transit sequence to produce the mature form of the protein, which can fold into its native form.
Figure 1 | Protein-import pathways into chloroplasts.
In the cytosol, preproteins with an amino-terminal presequence or transit sequence (yellow) can form a cytosolic guidance complex on phosphorylation in the cytosol. This guidance complex consists of an HSP70 (heat shock protein-70) chaperone and a 14-3-3 dimer. Altenatively, non-phosphorylated preproteins can associate with TOC159 (translocon of the outer chloroplast envelope-159) or with HSP70. All these complexes bind to TOC receptors in a GTP-dependent manner. A soluble TOC159 receptor might shuttle between the cytosol and the TOC complex bringing preproteins to the organelle surface. In a joint effort between TOC and TIC, preproteins are imported across both membranes in an NTP-dependent manner. The transit sequences are cleaved off by the stromal processing peptidase (SPP). Mature proteins either fold and assemble in the stroma or are directed to the thylakoids by various pathways. Preproteins without a cleavable transit sequence are mostly bound by HSP70 and targeted to the outer envelope. They can insert spontaneously into the membrane in vitro, although the process might be facilitated by proteinaceous factors in vivo, which still need to be identified (highlighted by a question mark). ALB3, albino3; IE, inner envelope; IES, inter-envelope space; OE, outer envelope; OEP, outer-envelope protein; SEC, secretory pathway; SRP; signal-recognition particle; TAT, twin-arginine translocase; TIC, translocon of the inner chloroplast envelope.
Figure 2 | The chloroplast translocon complexes.
The import machines in the outer (TOC) and inner (TIC) envelope from Pisum sativum have distinct subunit compositions. The TOC core complex is formed by the GTP-dependent receptor TOC34, the import channel TOC75 and the GTP-driven motor protein TOC159. TOC34 and TOC159 can be regulated by protein phosphorylation. The insert shows a three-dimensional reconstruction of a negatively stained electron-microscopy image that was obtained of the purified TOC core complex from P. sativum. TOC64 might function as a docking protein for HSP70 (heat shock protein-70)-guided preproteins — it is only loosely attached to the core complex. TIC110 might form the translocation channel of the TIC complex, and TIC20 might also be part of this channel (please refer to the text for further details). TIC40, which is closely associated with TIC110, might function as a chaperone-recruiting site. TIC22, which is localized in the inter-envelope space, might coordinate TIC and TOC function. The redox components TIC55 and TIC62 are proposed to regulate translocation through the TIC complex, and the binding of ferredoxin NAD(P) reductase (FNR) to TIC62 allows the TIC complex to sense the redox state of the chloroplast. TIC62 also contains a conserved NAD(P)-binding site. TIC55 contains an iron–sulphur centre and a mononuclear iron-binding site. The figure reflects only the subunit composition of the TIC and TOC complex as published to date, and does not reflect stoichiometric ratios, definitive functions or correct protein topologies. The electron-microscopy insert in the figure was reproduced with permission from Ref. 39 © (2003) The Rockefeller University Press.
From translation to the organelle surface
When the nascent polypeptide chain emerges from the ribosome it encounters a highly concentrated proteinaceous environment. Non-productive interactions or the exposure of hydrophobic amino-acid stretches result in aggregation and the loss of newly synthesized proteins. Molecular chaperones of the heat shock protein (HSP)70 FAMILY interact with non-native preproteins and keep them in a soluble, conformation that is not fully folded15, 16, 17, 18. This is also necessary because polypeptides have to cross the envelope translocons in a largely unstructured, extended conformation.
A common motif for targeting has not yet been identified in chloroplast transit sequences, but, in general, the transit sequences vary in length from 20–150 amino acids. Proteins that are localized in the thylakoid lumen contain bipartite targeting signals. The amino-proximal portion functions as a chloroplast-targeting, envelope-transfer domain, whereas the carboxy-proximal portion functions as a thylakoid-transfer domain19, 20, 21, 22. Other preproteins are synthesized without a cleavable transit sequence and contain their targeting information in the mature part of the protein (Fig. 1). This group of preproteins includes almost all of the outer-envelope proteins23, as well as one protein from the inner-envelope membrane so far24. Cleavable transit peptides carry an overall positive charge and are enriched in the hydroxylated amino acids serine and threonine. In aqueous solution, the transit peptide forms a random-coil structure25. Several, but not all, chloroplast preproteins can be phosphorylated in the transit sequence by a cytosolic ATP-dependent protein kinase26. Phosphorylation leads to the binding of 14-3-3 PROTEINS, which, together with HSP70, can form a cytosolic guidance complex27 (Fig. 1). Preproteins that are bound to the guidance complex are imported into chloroplasts more quickly than monomeric preproteins. Phosphorylation in the cytosol might therefore select a subclass of preproteins for preferential import26, 27. Phosphorylation is not directly involved in targeting, because mutating the phosphorylation site does not result in mistargeting in vitro. Furthermore, the nature of the kinase that performs this phosphorylation is still unknown, so in vivo evidence for the importance of precursor phosphorylation is still missing. Transit-sequence recognition is achieved by the receptor polypeptide of the TOC complex. Less is known about the mechanism of presequence-independent insertion into the outer envelope.
The TOC translocon
The subunits of the TOC translocon were initially identified by chemically crosslinking precursor proteins to neighbouring polypeptides or by membrane-complex isolation in the presence of detergent28, 29, 30, 31, 32, 33. Due to the different energy requirements for preprotein binding (<50 
M NTP) and translocation (>100 M NTP), different translocon subunits could be identified. TOC75 and TOC159 were typically found crosslinked to preproteins at low GTP and ATP concentrations31, 34, whereas TOC34 was mainly found to be crosslinked in the absence of GTP or ATP (Ref. 35). From these studies, it was proposed that TOC159 functions as an initial receptor, whereas TOC34 has a later regulatory function. However, it should be noted that the experimental set up that was used was, biochemically, extremely complex. Chloroplasts always contain residual NTPs and NDPs, which, owing to the presence of myokinase and nucleoside diphosphate kinase36, can result in the production of every type of NTP or interconversion between NTPs during the experiment. Nucleotide 'free' experiments might therefore actually not be completely nucleotide free. Furthermore, crosslinking experiments require that labelled preproteins accumulate at a certain site, so that they become detectable. Rapid kinetic intermediates are therefore difficult to identify in this type of experiment, so it is hard to deduce a clear series of steps in the import process from in organello studies.
Nevertheless combining the techniques that are described above resulted in the identification of four TOC translocon subunits — the two GTP-binding proteins TOC159 and TOC34 (Refs 28–30), the protein-import channel TOC75 (Refs 31,32,37), and TOC64 (Ref. 38 Fig. 2). The former three subunits form a stable core complex of 
550 kDa, which has a stoichiometric ratio of 4:4:1 for TOC34:TOC75:TOC159 (Ref. 39). In the model plant Arabidopsis thaliana, several genes exist for every TOC subunit40 (see below). Transcriptional control and organ-specific expression might result in the formation of TOC complexes that have different selectivities and translocation properties. Most of the in vitro biochemical data summarized below were obtained using TOC proteins from Pisum sativum, whereas the in vivo data were generally obtained from studies of A. thaliana. To distinguish between P. sativium and A. thaliana proteins, the prefixes Ps and At will be used.
TOC33/34. TOC34 is anchored by its carboxy-terminal tail, and most of the protein, including the GTP-binding domain, is exposed to the cytosol30. Reconstituted TOC34 can directly interact with preprotein12, 41. Preproteins are recognized with a high affinity when TOC34 is in its GTP-bound form12, 42. The low endogenous GTPase activity of TOC34 (Ref. 30) is stimulated up to 50-fold by the preprotein43. Therefore, the preprotein seems to function as a GTPase-activating protein, and such proteins are key for the regulation of GTPases in various systems44. TOC34–GDP has a much lower affinity for the preprotein42, which is subsequently handed to the next TOC subunit.
The next step in the TOC34 cycle is the release of GDP, which would normally be promoted by a GDP-exchange factor45. Such a factor has not yet been identified, but a domain of TOC159 could possibly fulfil this role (see below). The nucleotide-free form of TOC34 can then take one of two routes. First, it can be recharged with GTP and enter a new round of precursor recognition and binding or, second, it can be phosphorylated by an outer-envelope-localized, ATP-dependent protein kinase46. Phosphorylation inhibits GTP binding and therefore results in a complete switching off of the receptor. In A. thaliana, two homologues of Ps TOC34 are present40 — At TOC33, which seems to be a functional analogue of Ps TOC34, can also be switched off by phosphorylation, whereas At TOC34 seems not to be phosphorylated47. At TOC34 might therefore function as a constitutive receptor for protein import into different plastid types (see below)48, 49 (Fig. 3).
Figure 3 | Model of the Arabidopsis thaliana TOC33 and TOC34 receptor cycles.
The Arabidopsis thaliana (At)TOC33 (translocon of the outer chloroplast envelope-33) and At TOC34 receptors are activated by GTP binding. At TOC33/34–GTP binds the phosphorylated preprotein with high affinity. The preprotein activates the endogenous GTPase activity of At TOC33/34, which results in the hydrolysis of GTP. At TOC33/TOC34–GDP has a lower affinity for the preprotein, which is released to the next translocon subunit. After the release of GDP from At TOC33/34, the receptor can be recharged with GTP and enter a new receptor cycle. Alternatively, the At TOC33 receptor — which seems to be the functional analogue of the Pisum sativum (Ps) TOC34 receptor — can be turned off by phosphorylation, which inhibits GTP binding. Dephosphorylation of At TOC33 and Ps TOC34 is required for re-activation through GTP binding. The reaction cycles for a single At TOC 33 and a single At TOC34 are shown. Pi, inorganic phosphate. Modified with permission from Ref. 21 © (2002) Elsevier.
A. thaliana has become a model organism, and it can be used for forward and reverse genetic approaches. The roles of several TOC and TIC subunits in chloroplast biogenesis have therefore been studied in plants containing either antisense-RNA constructs — which lead to a reduction, but not a complete absence, of the target protein — or T-DNA insertions, which lead to a complete knockout of the target gene product. The reduction or knockout of translocon subunits that are essential for translocon function should result in severe defects in chloroplast/plastid biogenesis (for example, a loss of the ability to grow PHOTOAUTOTROPHICALLY), or should lead to lethality, which is normally manifested as an embryo lethality that results in early SEED ABORTION.
The plastid-protein-import (ppi) mutants have been numbered consecutively and the first to be isolated was a knockout of At toc33 (Ref. 50; At TOC33 is a functional analogue of Ps TOC34 (see above)). This ppi1 mutant has a pale-green phenotype and retarded chloroplast development. However, later in development, plants partially recover and are able to grow photoautotrophically on soil. This phenotype is probably due to the presence of a homologue of At TOC33 in this organism, that is, At TOC34 (Refs 48,50). Expression studies indicate that At TOC34 is constitutively expressed at low levels in all organs, whereas At TOC33 is upregulated in MERISTEMATIC and rapidly expanding leaf tissue48, 50. In line with these data is the observation that At TOC33 is dispensable for root development49. Together, the in vivo and in vitro results indicate that At TOC33 and At TOC34 have partially redundant functions. However, both receptors show clear preferences for certain classes of preproteins47, 48, 50. Finally, the different post-translational regulation of At TOC33 and At TOC34 must be kept in mind. As the At TOC33 receptor can be switched off by phosphorylation, the amount of At TOC33 protein does not represent the amount of active protein47.
How do plants respond to the inactivation of genes that encode translocon subunits? Only a limited amount of data are available, which makes it difficult to draw a comprehensive picture. However, the most details are available for TOC33/TOC34 (Ref. 51). In ppi1 knockout plants, the nuclear genes that encode thylakoid-localized photosynthetic proteins tend to be downregulated, whereas genes that encode the proteins involved in carbon metabolism are upregulated. Genes that encode stress proteins — such as HSP70 and the chaperonin-60 system — are also upregulated, which indicates that the chloroplasts experience an overall stress. The transcription of the TOC and TIC subunits is slightly upregulated. Surprisingly, though, the gene that encodes At TOC34 is downregulated in ppi1 plants51, although artificial overexpression of At TOC34 in a ppi1 background can completely reverse the ppi1 phenotype50.
TOC159. TOC159 is a second GTP-binding subunit of the translocon28, 29. The full-length protein can be divided into three regions: an amino-terminal A-domain, which is rich in acidic amino acids; a central G-domain, which contains the GTP-binding site and is highly similar to the corresponding region in TOC34; and finally the carboxy-terminal membrane or M-domain52, 53, 54. A small family of TOC159-like proteins is present in A. thaliana53, and it comprises TOC159, TOC132, TOC120 and TOC90. The A-domain becomes progressively smaller as the molecular size of the family members decreases and it is finally absent from At TOC90. The G-domain and M-domain are conserved between members of the family53. It has therefore been speculated that the A-domain defines certain substrate selectivities, whereas the G- and M-domains represent the more conserved or catalytic properties of this translocon subunit52, 53. From in vitro experiments, we have evidence for two functions of TOC159 (Ref. 41). First, it recognizes preproteins in a GTP-dependent manner, which is consistent with its proposed role as a receptor protein. Second, it functions in translocation. The 52-kDa M-domain of TOC159 is close to the preprotein35, which indicates that TOC159 is part of the translocation pore or that it might also contact the preprotein on the inter-envelope-space side of the outer envelope. In addition, TOC159 seems to provide the driving force for membrane translocation41 (see below). TOC159 is the most prominent phosphoprotein in the outer envelope of P. sativum chloroplasts46, so it might be subject to a similar type of regulation as TOC34.
A soluble form of TOC159 has been detected in the supernatant of mechanically ruptured P. sativum leaves54. Furthermore, overexpression of a TOC159–GFP (green fluorescent protein) fusion protein in A. thaliana, which was under the control of the strong heterologous CaMV 35S-promotor, led to the presence of the fusion protein in the cytosol. TOC159 has therefore been proposed to function as a soluble receptor that shuttles between the cytosol, where it picks up its cargo (preproteins), and the chloroplast surface, where it delivers its cargo to the TOC translocon54. Direct experimental support for this idea is lacking. However, the insertion of soluble TOC159 into the outer envelope is facilitated by interaction with TOC34, which could provide a mechanism for TOC-complex assembly and insertion of a receptor–preprotein shuttle55, 56, 57. Furthermore, Hwang and colleagues made stable complementation lines of the ppi2 mutant, which is defective in TOC159, using either full-length At TOC159 fused to GFP or the M-domain of At TOC159 fused to a T7 reporter tag (both were under the control of the CaMV 35S-promoter)58. Under the conditions used, they did not observe a soluble form of the At TOC159 M-domain — the fusion protein was localized exclusively to the outer envelope. However, the full-length fusion protein had a dual location — it partitioned between the chloroplast surface and the cytosol58. As already noted for mitochondria, even mild overexpression can result in protein-import failure and can therefore lead to mislocalization to other cellular compartments59. Our own data indicate that the soluble form of TOC159 is associated with chloroplast-specific lipids, which could indicate that TOC159 is present in low-density membrane fragments97.
The ppi2-mutant phenotype52 is much more severe than that of ppi1, the At toc33 knockout. The mutant lines have an albino phenotype — leaf plastids do not differentiate into chloroplasts (that is, no thylakoid formation occurs), and the plants are unable to grow on soil and die at the COTYLEDON STAGE52. The typical nuclear-encoded gene products that are required for photosynthesis — such as the light-harvesting chlorophyll a/b-binding protein and the small subunit of Rubisco — are dramatically downregulated in ppi2 plastids52. This indicates that other members of the TOC159 family, such as TOC132 or TOC120, cannot fully replace TOC159 and have specialized functions in organelle biogenesis, for example, in the import of non-photosynthetic gene products52. The levels of At TOC75 and At TIC110 in ppi2 mutants are unaffected, while the transcription of At toc34 is enhanced52. How the regulation of At TOC33, which is the predominant isoform in developing chloroplasts, is affected is unknown52. Chloroplasts from ppi2 plants contain very low levels of typically imported photosynthetic proteins — the residual import capacity is insufficient to support proper chloroplast development and results in the death of the seedling52. Whether chloroplasts from ppi2 plants can actually import non-photosynthetic gene-product precursors or contain levels of these polypeptides that are comparable to those in wild-type plants is not known52.
Complementation of ppi2 plants with different TOC159 constructs showed that both the G- and M-domain are required to recover a wild-type-like phenotype, whereas the A-domain is dispensable58. Plants complemented with the M-domain alone had a greenish phenotype, which indicates that the G-domain is required for full restoration of function. In the absence of the G-domain, plants were yellowish, which again indicates that the M- and A-domains are not sufficient for complementation and that the highly charged A-domain exerts an inhibitory effect58. Similar results were obtained by transiently expressing different deletion constructs of At TOC159 and simultaneously studying the import of the presequence of the Rubisco small subunit fused to GFP. The deletion constructs containing the G- and M-domain together or the M-domain alone resulted in complete complementation, whereas in the absence of the G-domain — that is, when the A- and M-domains were fused — preprotein import was slow and some preproteins accumulated in the cytoplasm. This again indicated that the A-domain exerts an inhibitory effect on import in this mutant58. Together, these data indicate that the M-domain is probably essential, not only for TOC-complex assembly, but also as a functional part of the translocon. The G-domain is required for full complementation, as it constitutes part of the receptor and motor function. The role of the G-domain should become more obvious in kinetic studies, in which it will be possible to compare import rates rather than just a single endpoint after a long incubation time. From in organello studies, it is clear that removal of the G-domain still allows residual (by-pass) import under the appropriate conditions (for example, a high ATP concentration and a long incubation time)60, 61, 62. In these situations, the import yield can be up to 20% of the normal yield.
TOC75. TOC75 is the most abundant outer-envelope protein. Structural analysis of TOC75 predicts that it forms a 
-barrel-type channel that, according to computer predictions, is lined by 16 transmembrane -sheets63. Heterologously expressed TOC75 forms a cation-selective, high-conductance ion channel when it is reconstituted into planar lipid bilayers64. Electrophysiological measurements indicated that the channel is 25 Å wide at the entrance, and constricts to 15–17 Å wide inside the channel. A width of 15 Å could accommodate a polypeptide stretch that still retains some secondary structure65. The electrophysiological data also showed that TOC75 has a cytosolic preprotein-binding site, which is able to distinguish between the precursor protein and the mature form34, 64, 65. The binding affinity of TOC75 for preproteins is, however, lower than that of TOC34 or TOC159.
A number of TOC75 isoforms are present in A. thaliana, and they are named according to the chromosomal location of the genes40. At TOC75III is the most abundant outer-envelope protein and is present in the isolated TOC core complex39. A viable At toc75III-knockout mutant has not been reported so far, which could indicate an essential role for this protein in chloroplast biogenesis. At TOC75V is present in the outer envelope with an abundance of less than 10% that of At TOC75III (Ref. 66). It is not present in the TOC core complex, which could indicate that At TOC75V forms a specialized subset of import translocons. At TOC75V is homologous to proteins in the outer membrane of Gram-negative bacteria (Box 2), and it probably represents the most ancestral form of an organelle protein-import channel66, 67, 68. It is not clear whether At TOC75I and At TOC75IV are expressed at all in plants, because no expressed-sequence-tag clones have been found so far40.
TOC64. The role of TOC64 is less well-defined. It has three exposed TETRATRICOPEPTIDE MOTIFS on the cytosolic face of the organelle38, which are involved in protein–protein interactions. The peroxisomal import receptor PEX5 and the mitochondrial import receptor TOM70 have similar motifs3, 4. TOM70 functions as a receptor for hydrophobic preproteins (such as carrier proteins that have several internal transport signals), which arrive at the organelle surface as a preprotein–HSP70 complex. The HSP70 chaperone docks onto a specialized tetratricopeptide-repeat domain of the TOM70 import receptor and catalyses the productive transfer of the preprotein to the translocon69, 70. A similar role could be envisioned for TOC64.
Reconstitution of TOC translocation. In vitro reconstitution experiments represent a useful tool to define more precisely the sequence of events that lead from recognition to translocation, and to characterize single subunits biochemically. The TOC core complex can be isolated from purified envelope membranes as a functionally active unit33, 71. Analysis of the complex by transmission electron microscopy shows a particle with a diameter of 130 Å, and a three-dimensional reconstruction map indicates a central finger-like region that separates four curved translocation channels39 (see the inset in Fig. 2). Reconstitution of this TOC core complex into liposomes showed that it is fully import competent — that is, it could recognize and translocate a preprotein across the liposomal membrane in a GTP-dependent manner41. No other nucleoside triphosphate could replace the GTP requirement. Further dissection of this system using single purified TOC subunits showed that TOC34 and TOC159, but not TOC75, can function as GTP-dependent preprotein-binding proteins, that is, receptors. In addition, the reconstitution of TOC75 together with TOC159, but not with TOC34, resulted in preprotein import into a lipid micelle in a GTP-dependent manner41. The hydrolysis of GTP by TOC159 is therefore the sole driving force for preprotein translocation in vitro. GTP hydrolysis by TOC159 might provoke a conformational change that, in a 'sewing machine'-type mechanism, pushes the preprotein through the TOC75 channel (Fig. 4). TOC75 and TOC159 form the minimal translocon unit in vitro that is able to specifically recognize and translocate chloroplast preproteins across a membrane. TOC34 could therefore represent the initial receptor for incoming preproteins. The transfer of the preprotein from TOC34 to TOC159 might be facilitated by the formation of a heterodimer. The recently obtained crystal structure of TOC34 (Ref. 72) indicates that it can form homo-oligomers. Due to the extensive sequence similarity between TOC34 and TOC159 (Ref. 28), hetero-oligomerization might also be possible56, 57, which would clearly facilitate a smooth transfer of preproteins from one subunit to another. Homotypic interactions between TOC34 and TOC159 occur preferentially when both proteins are in their GDP-bound form73. One could speculate that the interaction between TOC34 and TOC159 triggers a GDP-to-GTP exchange in TOC159, which would result in the transfer of the preprotein from TOC34–GDP to TOC159–GTP and subsequently in the dissociation of the TOC34–TOC159 heterodimer (not shown in Fig. 4). The GTP-regulated interaction of translocon subunits is also verified by co-translational protein-translocation systems in bacteria and eukaryotes74.
Figure 4 | Working hypothesis for the action of TOC159.
A preprotein binds to translocon of the outer chloroplast envelope-34 (TOC34)–GTP and, on GTP hydrolysis, this preprotein is transferred to TOC159–GTP. TOC34 then releases GDP and rebinds GTP. In an action that also requires GTP hydrolysis, TOC159 then pushes the preprotein through the TOC75 import channel. A conformational change in TOC159 that is caused by GTP hydrolysis might move the preprotein forward into the channel in a 'sewing machine'-type mechanism. TOC159 then releases GDP and rebinds GTP. Although the TOC core complex is shown in the figure, the minimal complex that is needed for preprotein translocation across a membrane is composed of just TOC75 and TOC159. The gating mechanism of TOC75 is unknown at present. IES, inter-envelope space; OE, outer envelope; Pi, inorganic phosphate.
The TIC translocon
In most cases, proteins probably translocate through the TOC and TIC complexes simultaneously, although the TIC complex has the ability to retrieve and translocate preproteins that have been released into the inter-envelope space by the TOC translocon75. The translocation of preproteins across the inner-envelope membrane requires ATP hydrolysis76, 77, 78, and this ATP is probably needed for the action of molecular chaperones in the stroma, which provide the driving force to complete import into the organelle79, 80. A membrane potential is not required for any step of the import process76, 77, 78. Several TIC subunits have been identified, and these are TIC110, TIC62, TIC55, TIC40, TIC22 and TIC20 (Fig. 2).
TIC110, TIC62 and TIC55. TIC110 has one or two hydrophobic transmembrane 
-helices in its amino-terminal region79, 81, which are important to target and anchor the protein in the inner-envelope membrane81. The topology of the large 80-kDa carboxy-terminal region is not completely resolved (see below). TIC110 is the most abundant and best studied TIC subunit79, 80, 81, 82. On the inter-envelope-space side, it seems to be in the close vicinity of, or even in contact with, the TOC translocon, because TIC110 often co-fractionates with the TOC complex32. On the stromal side of the inner envelope, TIC110 could interact with molecular chaperones such as HSP93 and chaperonin-60 (Refs 79,80). Whether such interactions occur directly or through other TIC subunits remains to be established. Reconstitution experiments using heterologously expressed protein showed that TIC110 can form a cation-selective channel in lipid bilayers83. The electrophysiological properties of this channel indicate a pore size of 15–20 Å, which is similar to the size estimated for TOC75 and is sufficient to allow the passage of a polypeptide chain. Whether further proteins, such as TIC20, participate in channel formation in vivo is unknown at present84.
The exact function of TIC110 is, however, not fully understood at present. Topological studies that assessed the protease accessibility of TIC110 in either intact chloroplasts or isolated inner-envelope-membrane vesicles gave conflicting results. Studies in our laboratory found that TIC110 was accessible to proteases from the inter-envelope space, which is indicative of an exposed domain or multiple transmembrane spans that are connected by protease-accessible loops81. Keegstra and colleagues, however, observed that TIC110 was inaccessible to proteases from the inter-envelope space, and therefore concluded that TIC110 exposes a large domain to the chloroplast stroma85. Circular dichroism (CD) spectra of heterologously expressed soluble TIC110 (residues 93–966) indicated a largely -helical conformation86, whereas CD spectra recorded after the refolding of a similar TIC110 domain following treatment with 8 M urea indicated a high propensity for -strands83. Clearly more work is needed to settle this important issue.
The TIC complex that was purified by blue-native polyacrylamide gel electrophoresis has an apparent molecular weight of 250 kDa (Ref. 87). Abundant subunits in this complex are TIC110, TIC62 and TIC55. The latter two have the potential to catalyse electron-transfer reactions87, 88. TIC55 contains a IRON–SULPHUR CENTRE and a mononuclear iron-binding site87. TIC62 has a conserved NAD(P)-binding site. The carboxyl terminus of TIC62 is exposed to the stroma and interacts with ferredoxin NAD(P) reductase (FNR; Fig. 2), as deduced from yeast two-hydrid studies and affinity-chromatography experiments88. FNR couples photosynthetic electron flux to the reduction of NAD(P)H. NAD(P)H is the primary reductant in CO2 fixation, nitrogen and sulphur reduction, as well as in fatty-acid and isoprenoid biosynthesis. The interaction between FNR and TIC62 might link the metabolic status of the chloroplast — that is, the NAD(P)H:NAD(P) ratio — to the import capacity of the TIC complex. Support for this idea comes from import experiments that used different ferredoxin isoforms89. Ferredoxin I, which is involved in photosynthetic electron flux, is imported both in light and dark conditions. Ferredoxin III, which is more involved in metabolic processes, is mistargeted in the light to the inter-envelope space. Only in the dark is ferredoxin III correctly imported into the stroma89. Together, the available data indicate a clear potential for the redox regulation of preprotein import through the TIC complex. It is probable that not all preproteins are sensitive to this control, or that there are distinct TIC subtranslocons that have different regulatory properties.
TIC40. TIC40 is an integral membrane component of the TIC complex and it seems to be tightly associated with TIC110 (Refs 90,91). The carboxy-terminal region is exposed on the stromal site of the inner envelope, and this region seems to have two functional domains. First, it is homologous to HSP70-interacting proteins as well as to HSP70/HSP90-organizing proteins90 and, second, it has a tetratricopeptide domain for protein–protein interactions91. Immunoprecipitation experiments indicate not only a close association with TIC110, but also with stromal HSP93 (Ref. 91).
TIC40 is encoded by a single gene in A. thaliana40. A. thaliana plants that contain a T-DNA insertion in the At tic40 gene have a pale-green phenotype91. Their flowering is retarded and their thylakoid grana stacks are not as pronounced as they are in wild-type plants. Chloroplasts isolated from At tic40-knockout plants bind preproteins with an efficiency similar to that of wild-type plants; however, translocation into the stroma is reduced by 50% (Ref. 91). These results support the proposed role for TIC40 in recruiting molecular chaperones to the TIC translocon, which concentrates these chaperones at, and coordinates chaperone action with, the sites of import90.
TIC22. TIC22 is localized to the inter-envelope space and is only loosely bound to the inner-envelope membrane84. It might have a role in the coordination of TOC and TIC functions, or in the guidance of preproteins across the inter-envelope space.
TIC20. TIC20 is an integral protein of the inner-envelope membrane and it has four putative transmembrane -helices84. Due to its similarities to bacterial amino-acid transporters and to the mitochondrial import component TIM17 (Ref. 92), it was proposed to participate in the formation of the TIC import channel84. However, no in vitro data support this idea so far.
The importance of TIC20 for chloroplast biogenesis was studied using an antisense approach93. Two genes — At tic20I and At tic20V — encode the TIC20 isoforms (Ref. 40). A. thaliana plants that were treated with antisense RNA against At tic20I showed reduced levels (between 20–50%) of At TIC20I protein compared to wild-type plants. Plants were viable on soil, but they were pale and the chloroplast ultrastructure showed impaired thylakoid formation93. Preprotein import into chloroplasts that were isolated from these antisense plants was inhibited by 50%, although preprotein binding to TOC receptors remained normal, which indicates a selective defect in translocation across the inner membrane93. The expression of At tic20IV and the role of this isoform in antisense-At-tic20I-treated plants were not studied, but At TIC20IV might help to maintain a viable phenotype in these plants93. Although there were changes in the expression and protein levels of At TIC20I in plants treated with antisense At tic20I, no significant changes in the amount of the other TOC and TIC subunits could be detected93.
Further insights into the transcriptional control of translocon subunits has come from an independent genetic screen that searched for protein-import mutants. In this screen, a leaf-specific transcription factor was isolated that specifically upregulates At toc33 and At toc75III gene expression94. Mutant plants showed only basal transcription rates of the two genes, which were not sufficient to produce enough translocon subunits for normal chloroplast biogenesis. Consequently, these plants had a pale-green phenotype, but could still grow photoautotrophically. Understanding the developmental and organ-specific control of the expression of translocon subunits should, in the future, enable us to predict translocon composition, selectivity and regulation within different plastid types.
Conclusion and perspectives
Protein import into chloroplasts is more complex than was previously anticipated. Regulatory circuits seem to operate at three different levels — in the cytosol and at the TOC and TIC translocons. GTP-dependent regulation at the TOC complex superficially resembles the signal-recognition-particle–Sec61 system. However, the actual import mechanism is a hybrid machine that uses a GTP-dependent motor to move proteins across the outer envelope, and it might be similar to the bacterial SecA-dependent system, which translocates proteins across the inner bacterial membrane through the SecYEG translocon. The TIC translocon probably uses molecular chaperones as the driving force for translocation and, in this respect, resembles the mitochondrial import system. Although most preproteins seem to use a general import pathway, the unexpected number of genes for isoforms of the TOC and TIC subunits seems to tell a different story. Clearly the protein composition of these translocons in different organs needs to be determined to verify the expression profiles. Biochemical and biophysical methods must be used in reconstituted systems to pinpoint differences in the specificity and regulation of translocon subunits, as well as to understand the cooperation between them. Owing to its inborn complexity, the in vivo system is unable to tell us everything. Finally, our efforts to determine the structures of single subunits and of entire complexes should help us to define the mechanistic basis of protein import into chloroplasts.
pH-dependent pathways