Protein trafficking to the apicoplast

With the identification of the apicoplast genome, it became apparent that the transcription and translation of its resident genes (including many ribosomal proteins), as well as any apicoplast-specific functions, would require the collaboration of additional proteins. These proteins are encoded in the nucleus, and hence are called nucleus-encoded apicoplast-targeted (NEAT) proteins. A search of the T. gondii EST databases for proteins homologous to those found in chloroplasts led to the identification of several candidates for apicoplast ribosomal proteins, as well as some enzymes involved in fatty acid biosynthesis. Initially two such proteins, acyl carrier protein (ACP) and ribosomal protein S9 (S9), were confirmed to be localized to the apicoplast by microscopic analysis using specific antisera (Waller et al., 1998). These genes, in turn, provided tools for dissecting the manner in which proteins are targeted to the apicoplast.

Sequence analysis clearly showed that these proteins, all of which were predicted to reside within the lumen of the apicoplast, possessed N-terminal extensions as compared to their bacterial orthologs. In eukaryotic organisms, such extensions often contain topogenic information. For example, proteins localized to primary chloro-plasts are usually targeted via an N-terminal transit peptide directly from the cytosol. Localization to the secretory system usually involves an N-termi-nal signal sequence. These presequences are rapidly removed by specific processing enzymes upon import (Bruce, 2001).

Interestingly, in those organisms with secondary chloroplasts, N-terminal extensions appear to contain a signal sequence followed by a transit sequence (Nassoury et al., 2005). This organization is exactly what is observed for both T. gondii and P. falciparum NEAT proteins predicted to reside in the apicoplast stroma (Waller et al., 1998). Using gene fusions, Waller et al. (1998) showed that the N-terminal extension of T. gondii ACP was able to target green fluorescent protein (GFP) to the apicoplast. Furthermore, presequences of P. falciparum predicted NEAT proteins were able to target GFP to the T. gondii apicoplast. These data suggest that at least some mechanisms of targeting are conserved across the Apicomplexa (Waller et al., 1998; Jomaa et al., 1999).

Both the signal and transit regions of the N-terminal extension of NEAT proteins are required to target a reporter to the apicoplast (Figure 9.6); deletion of either region results in mis-targeting. Without a transit sequence, the S9 signal sequence targets GFP for secretion (DeRocher et al., 2000; Yung et al., 2001), indicating that the reporter protein had entered the secretory system. Similar work in P. falciparum showed that the N-terminal region of ACP is able to target GFP for secretion (Waller et al., 2000). The signal sequences for these proteins do not appear any different from those of proteins targeted to other destinations in the secretory system. Indeed, in unpublished studies mentioned by Roos et al. (1999), replacing the endogenous signal sequence with one from a heterologous secretory protein did not alter targeting to the apicoplast. Without a signal sequence, the S9 transit peptide directs GFP to the mitochondrion (Figure 9.6), while a cytosolic localization is seen with GFP fusions to the transit peptide of ferredoxin-NADP+ reductase (FNR) in T. gondii (Harb et al., 2004). Similar findings were seen for ACP in malaria parasites (Waller et al., 2000). Taken together, these studies indicate that the first step in protein targeting to the apicoplast lumen is entry into the secretory system.

FIGURE 9.6 Both domains of the N-terminal extension are required for targeting of NEAT proteins. GFP fusions containing the entire N-terminal extension of ribosomal protein S9 (aa 1-159, S+T-GFP), its signal sequence (aa 1-42, S-GFP) or its transit sequence (aa 33-159-159, T-GFP) were expressed in T. gondii. The left-hand panels show GFP fluorescence, while the right-hand panels show DIC images of the same cells, which are residing within a parasitophorous vacuole in host fibroblasts. Co-localization with apicoplast markers (DNA or acetyl CoA carboxy-lase) demonstrated that the single dot observed upon expression of S+T-GFP corresponds to the apicoplast. Co-localization with mitochondrial markers HSP-60 and mitotracker showed that the T-GFP protein is found in the mitochondrion. S-GFP is found primarily in the parasitophorous vacuole, although some material can be seen within the endomembrane system of the parasite. Image courtesy of Dr Amy DeRocher.

The transit peptides of NEAT proteins vary in length, from about 50 to 200 amino acids (aa), and are very diverse in sequence. Like chloroplast transit peptides, these peptides have few acidic or hydrophobic residues (Foth et al., 2003). In fact, the transit peptide of T. gondii S9, when fused to

GFP allows GFP to be imported into isolated pea chloroplasts (DeRocher et al., 2004). Apicoplast transit peptides have a net positive charge. The T. gondii transit sequences are enriched for serine and threonine, amino acids shown to be important in plant transit peptides (Bruce, 2001). P falciparum transit peptides are enriched for asparagine and lysine residues ((Foth et al., 2003), and mutational analysis showed that hydroxylated residues are not crucial for targeting (Waller et al., 2000). Site-directed mutagenesis of the P. falciparum ACP transit peptide indicated that while basic residues at positions 2 and 6 were not essential, an acidic residue at position 2 prevented apicoplast targeting (Foth et al., 2003). A predicted HSP70 binding site in the transit peptide was also found to be important (Foth et al., 2003), suggesting that maintenance of the unfolded structure could be important in apicoplast targeting. Several studies have shown that T. gondii apicoplast transit peptides contain redundant information, since non-overlapping segments can still mediate targeting (DeRocher et al., 2000; Yung et al., 2001, 2003; Harb et al., 2004). Detailed mapping of transit peptide functions of T. gondii FNR suggests that release from the ER, localization to the apicoplast, binding to chaper-ones, and processing are specified by discrete domains (Harb et al., 2004).

The identification of NEAT proteins and characterization of the bipartite apicoplast targeting sequence has allowed development of bioinfor-matic models that predict which proteins may be localized to the apicoplast of malaria parasites. The genome of P. falciparum is completely assembled, and the essential features of the transit peptide are well-described, allowing the generation of neural network (PATS; Zuegge et al., 2001) and rule-based algorithms (PlasmoAP; Foth and McFadden, 2003) to identify candidate Plasmodium NEAT proteins. Both of these programs are available through a web-based interface at PlasmoDB.org, and PATS can also be found at the Modlab website at the Universität Frankfurt am Main (http://gecco.org.chemie.uni-frankfurt.de/pats/pats-index.php). Algorithms for the identification of T. gondii NEAT proteins were not available when this review was prepared; however, with completion of annotation for the

T. gondii genome (ToxoDB 4.0, released July 2006), development of such algorithms will be a high priority.

Processing of chloroplast transit peptides is rapid, such that only the mature protein is seen under steady-state conditions (see, for example, Shanklin et al., 1995). In contrast, both the mature form and the precursor protein containing the transit peptide are observed for NEAT proteins, whether native proteins or artificial gene fusions (Waller et al., 1998; DeRocher et al., 2005). Pulse-chase studies indicate that little processing is seen until 45-120 minutes after synthesis for T. gondii and P. falciparum ACP-GFP (van Dooren et al., 2002; DeRocher et al., 2005). Whether this delay (as compared to chloroplast protein processing) reflects the time required for complete import or the relative inefficiency of processing is not yet clear. A P. falciparum putative ortholog of the chloroplast stromal processing peptidase has been identified (van Dooren et al., 2002). The predicted protein includes a bipartite targeting sequence, suggesting that it lies within the apicoplast. A putative T. gondii ortholog can be detected in the gene models of ToxoDB (E values of ~10-50). A functional apicoplast is required for processing of the transit peptide in T. gondii (He et al., 2001b). However, analysis of the S9 transit peptide indicates that multiple processing steps can occur, the first of which may take place before import is complete (Yung et al., 2001) - a possibility supported by studies of the transit peptide of FNR (Harb et al., 2004).

The path that NEAT proteins follow to go from the ER to the apicoplast stroma is not yet clear, although several models have been proposed (Figure 9.7). One hypothesis is that the apicoplast lies literally within the secretory system, with its outer membrane contiguous with the ER. All proteins would pass it on their way to other destinations and NEAT proteins would be grabbed by the apicoplast by virtue of their transit peptide. They would then be imported through the next three membranes. Another possibility is that proteins move by vesicular trafficking from the ER to apicoplast, either directly or indirectly. In this case,

FIGURE 9.7 Three models for protein targeting to the apicoplast. Model 1 shows direct import from the ER. Model 2 postulates vesicular trafficking through the Golgi. Model 3 proposes vesicular trafficking from the ER, bypassing the Golgi. Organelles indicated are: A, apicoplast; ER endo-plasmic reticulum; G, Golgi; and N, nucleus.

FIGURE 9.7 Three models for protein targeting to the apicoplast. Model 1 shows direct import from the ER. Model 2 postulates vesicular trafficking through the Golgi. Model 3 proposes vesicular trafficking from the ER, bypassing the Golgi. Organelles indicated are: A, apicoplast; ER endo-plasmic reticulum; G, Golgi; and N, nucleus.

the transit peptide would be responsible for packaging the proteins into appropriate vesicles, and specificity of vesicle trafficking would likely be conveyed by an additional component - possibly proteins such as SNAREs at the vesicle and apicoplast surfaces, as occurs for targeting to other destinations in the secretory system (Hong, 2005).

Recent studies have indicated that it is unlikely that T. gondii NEAT proteins that reside in the apicoplast lumen traffic through the Golgi. Exposure of T. gondii to brefeldin A (BFA) or low temperature block (15°C) both inhibit Golgi trafficking but are without effect on protein localization to the apicoplast as judged by deconvolution microscopy (DeRocher et al., 2005). These studies are complicated by the presence of pre-existing marker proteins in the apicoplast. To circumvent this difficulty, DeRocher et al. (2005) used a system that allows regulated exit of proteins from the ER (Figure 9.8). It is based on a tandem repeat of a conditional aggregation domain (CAD) fused to a reporter that bears the necessary topogenic information (Rollins et al., 2000). In the absence of a synthetic ligand, aggregation of the CAD domains blocks trafficking of the fusion protein. The addition of the ligand solubilizes the aggregated protein,

FIGURE 9.8 Apicoplast protein targeting studied by conditional aggregation. A GFP fusion protein bearing the bipartite extension of ribosomal protein S9, plus a tandem array of four CAD domains, was expressed in T. gondii. Removal of ligand causes aggregation of the CAD domains, while addition of ligand yields monomerization. When ligand is removed from the stable transfectants, the GFP is detected in the ER. When ligand is added for 4 hours the protein trafficks to the apicoplast. This tracking occurs even in the presence of BFA. GFP was detected using anti-GFP antibodies. DAPI staining reveals the DNA in the parasite nucleus and apicoplast, and in the upper and lower images, a portion of the host cell nucleus is visible (asterisk). Image courtesy of Dr Amy DeRocher. This figure is reproduced in color in the color plate section.

FIGURE 9.8 Apicoplast protein targeting studied by conditional aggregation. A GFP fusion protein bearing the bipartite extension of ribosomal protein S9, plus a tandem array of four CAD domains, was expressed in T. gondii. Removal of ligand causes aggregation of the CAD domains, while addition of ligand yields monomerization. When ligand is removed from the stable transfectants, the GFP is detected in the ER. When ligand is added for 4 hours the protein trafficks to the apicoplast. This tracking occurs even in the presence of BFA. GFP was detected using anti-GFP antibodies. DAPI staining reveals the DNA in the parasite nucleus and apicoplast, and in the upper and lower images, a portion of the host cell nucleus is visible (asterisk). Image courtesy of Dr Amy DeRocher. This figure is reproduced in color in the color plate section.

releasing it for trafficking (Rivera et al., 2000). To study protein targeting to the T. gondii apicoplast, ligand was withdrawn for 2 days from parasites that were stably transfected with an apicoplast-targeted CAD-GFP This blocked trafficking of the fusion protein, depleting it from the apicoplast during parasite division. Instead, the GFP reporter accumulated in the ER (Figure 9.8). Subsequent addition of ligand released the fusion protein, which rapidly localized to the plastid region (DeRocher et al., 2005). This localization was not blocked by BFA, or by 15°C block of ER to Golgi trafficking. However, a BFA-sensitive step is involved in protein targeting or maturation, since no transit peptide cleavage is observed in pulse-chase experiments in the presence of BFA (DeRocher et al., 2005).

In P. falciparum, expression of mRNAs for known NEAT proteins is temporally regulated during the erythrocytic cycle, peaking in the late trophozoite/early schizont stages (Bozdech et al.,

2003; Le Roch et al., 2003). When a P falciparum NEAT GFP fusion protein was expressed earlier than normal (by driving transcription using a promoter active earlier in the erythrocytic cycle), it trafficked to the parasitophorous vacuole in a BFA-dependent manner and also appeared in the apicoplast at the usual time in the cycle (Cheresh et al., 2002). However, the general significance of this finding is not yet clear.

Little is known about the import apparatus of the apicoplast. The assumption, based on phyloge-netic origins and similarity of targeting sequences, is that the apicoplast translocon resembles that of chloroplasts. This translocon is composed of multiple components in the inner (Tic) and outer (Toc) membranes (Jarvis and Soll, 2002; Nassoury et al., 2005). Among the functions provided in the complex are specific binding to the transit sequence, channel formation, energy generation, and chaperone activity. However, other than a putative Tic22 and Toc34 (unpublished work cited in Waller and McFadden, 2005), database searches have not revealed promising candidates for orthologs of the chloroplast Tic and Toc complexes. Presumably, orthologous functions are present but the machinery is substantially diverged with respect to protein sequence. Isolated secondary plastids from the cryptomonad Guillardia theta, described by the authors as having two membranes, were capable of importing proteins bearing transit peptides from nucleomorph-encoded plastid proteins but not those bearing transit peptides from nuclearly encoded proteins (Wastl and Maier, 2000). These data raise the possibility of multiple mechanisms of protein import. Interestingly, the cryptomonad nucleomorph encodes at least two proteins recognizable as functioning in the import pathway for the inner membrane: Tic22 and Iap100 (Douglas et al., 2001).

The above studies concern the localization of soluble matrix proteins to the apicoplast. To date, proteins residing in the apicoplast membrane have not been conclusively identified, much less their route to the plastid. It can be presumed that a substantial membrane proteome is required for the import of proteins and substrates, and the export of products. As noted above, a candidate Tic22 has been identified, and it bears a bipartite targeting sequence. In chloroplasts, Tic22 lies in the intermembrane space (Kouranov et al., 1999). Thus it seems likely that at least some non-lume-nal proteins reach the apicoplast via protein targeting sequences related to those already defined for soluble proteins. This mechanism is reminiscent of targeting of proteins to the inner membrane of the chloroplast, which is often mediated by a transit peptide (Silva-Filho et al., 1997; Roth et al., 2004). Other chloroplast proteins lack recognizable transit sequences (Funes et al., 2004), as do some potential apicoplast membrane proteins, such as the T. gondii sugar phosphate transporter (Karnataki, DeRocher, Feagin and Parsons, unpublished results). Certain proteins targeted to the outer membrane of the chloroplast, such as Toc75, have a transit sequence, while others have an N-terminal region resembling a signal sequence (Hofmann and Theg, 2005). It appears likely that a variety of mechanisms are employed for targeting to apicoplast membranes, possibly related to the final destination.

With four bounding membranes, the apicoplast offers multiple options for localization. Membrane composition may differ between them, and may convey some specificity to the targeting of membrane proteins. The lipid composition of chloroplast membranes is very distinct from that of other cellular membranes, with high levels of galactoglyc-erolipids (Block et al., 1983). Although little is known about apicoplast membranes directly, galac-tolipids have been identified in P falciparum and T. gondii (Marechal et al., 2002; Bisanz et al., 2006).

0 0

Post a comment