Plasmodium falciparum-encoded exported hsp70/hsp40 chaperone/co-chaperone complexes within the host erythrocyte

Authors


For correspondence. E-mail przybors@staff.uni-marburg.de; Tel. (+49) 6421 2826596; Fax (+49) 6421 2826531.

Summary

Malaria parasites modify their host cell, the mature human erythrocyte. We are interested in the molecules mediating these processes, and have recently described a family of parasite-encoded heat shock proteins (PfHsp40s) that are targeted to the host cell, and implicated in host cell modification. Hsp40s generally function as co-chaperones of members of the Hsp70 family, and until now it was thought that human Hsp70 acts as the PfHsp40 interaction partner within the host cell. Here we revise this hypothesis, and identify and characterize an exported parasite-encoded Hsp70, referred to as PfHsp70-x. PfHsp70-x is exported to the host erythrocyte where it forms a complex with PfHsp40s in structures known as J-dots, and is closely associated with PfEMP1. Interestingly, Hsp70-x is encoded only by parasite species that export the major virulence factor EMP1, implying a possible role for Hsp70-x in EMP1 presentation at the surface of the infected erythrocyte. Our data strongly support the presence of parasite-encoded chaperone/co-chaperone complexes within the host erythrocyte, which are involved in protein traffic through the host cell. The host–pathogen interaction within the infected erythrocyte is more complex than previously thought, and is driven notonly by parasite co-chaperones, but also by the parasite-encoded chaperone Hsp70-x itself.

Introduction

Malaria tropica, caused by Plasmodium falciparum is responsible for over 0.6 million deaths annually (World Health Organization, 2011). Human erythrocytes infected by this parasite undergo radical changes in their biochemical and structural properties, induced by approximately 250 parasite-encoded proteins that are exported into the erythrocyte, collectively termed the Exportome (Hiller et al., 2004; Marti et al., 2004; Maier et al., 2009). Among these is the major virulence factor PfEMP1, which inserts into the erythrocyte plasma membrane, and is largely responsible for the novel adherent properties possessed by infected erythrocytes, referred to as cytoadherence (Baruch et al., 1995; Smith et al., 1995; Su et al., 1995). The adhesion of infected erythrocytes to various tissues in the body leads directly to the severe disease outcomes related with malaria tropica, including cerebral and pregnancy-associated malaria (Miller et al., 2002).

In the host cell, the parasite develops within a membrane-bound compartment known as the parasitophorous vacuole (PV) (Lingelbach and Joiner, 1998). Thus, proteins synthesized within the parasite that are destined for the erythrocyte cytosol must pass not only the parasite plasma membrane, but also the membrane of the PV (parasitophorous vacuolar membrane, PVM). Transport appears to take place initially via the default secretory pathway to the PV, then through a translocon complex within the PVM (de Koning-Ward et al., 2009; Crabb et al., 2010). Passage across the PVM requires protein unfolding, hinting at an important role for molecular chaperones within the PV that can carry out this function (Gehde et al., 2009). Having emerged from the PVM, exported proteins must be refolded and trafficked to various destinations within the host cell. This process is still not well understood, but represents an interesting paradigm in eukaryotic cell biology: how to transport proteins across the cytosol of a denucleated host cell, lacking the endomembrane transport system generally required for protein trafficking. Several analyses have revealed that a surprisingly large percentage of the P. falciparum-specific Exportome is involved in the correct presentation of PfEMP1 on the host cell surface (Cooke et al., 2006; Maier et al., 2007; Glenister et al., 2009). Apparently, the parasite invests a huge amount of resources in ensuring that PfEMP1 (and likely other variant antigens) reach the cell surface, probably driven by the high selective advantage of cytoadherence and thus avoidance of splenic clearance. Many of the proteins involved in these transport processes are P. falciparum-specific, suggesting that the large size and complicated domain structure of PfEMP1 and other variant antigens requires a specialized transport system not present in other Plasmodium species.

Among the proteins involved in presentation of PfEMP1 are members of the Hsp40/DnaJ family (Maier et al., 2008). Several exported PfHsp40s are known to localize to J-dots, highly mobile parasite-induced structures within the host cell (Kulzer et al., 2010). In other systems, Hsp40s act as co-chaperones, regulating the efficient binding and functioning of Hsp70s, and are thus intimately involved in protein folding and trafficking mechanisms, and have also been implicated in post-translational membrane integration (Kampinga and Craig, 2010; Hegde and Keenan, 2011). To date, however, no exported PfHsp70s have been reported, leading to the current hypothesis that the parasite can only utilize human Hsp70 to manipulate exported protein substrates, likely in concert with exported PfHsp40s (Banumathy et al., 2002; Charpian and Przyborski, 2008; Kulzer et al., 2010). Here we present data supporting an alternative hypothesis by identifying an exported parasite-encoded Hsp70, known as PfHsp70-x, which forms a complex with exported PfHsp40s, and possibly PfEMP1, in the infected cell. We suggest that the PfHsp70/40 complex here reported is likely to represent an essential requirement for efficient protein traffic within the infected cell, and thus can be viewed as an important pathogenicity factor and potential drug target.

Results

Identification of a potentially secreted PfHsp70, PfHsp70-x

Plasmodium falciparum encodes multiple members of the Hsp70 family: a homologue of the yeast canonical cytosolic Hsp70 ScSsa1p (Hsp70-1, PF08_0054), a ScKar2p homologue (Hsp70-2, PFI0875w, also called BiP, GRP78), a mitochondrial ScEcm10p homologue (PfHsp70-3, PF11_0351), and both cytosolic and endoplasmic reticulum (ER) nucleotide exchange factors, homologous to ScSse1p/ScLhs1p (PfHsp70-z, also known as CG4, PF07_0033; PfHsp70-y, MAL13P1.540) (Shonhai et al., 2007). Additionally, a sixth, two-exon gene exists (PfHsp70-x, MAL7P1.228), which is annotated to contain an in-frame stop codon in the 5′ end of the gene, and is thus referred to as a pseudogene. Our own cDNA sequencing reveals that the hsp70x coding sequence actually begins within the predicted intron sequence, and encodes a slightly shorter, but complete open reading frame, lacking introns (Fig. S1). The discrepancy between our gene annotation and the original database sequence is due to a single base pair mistake in the original sequencing data, which removed the correct AUG start codon (Fig. S1). This analysis is supported by examination of data at PlasmoDB, which reveals a single nucleotide polymorphism (SNP) at this position differing between 3D7 and all other strains (CombinedSNP.MAL7.1058). The correct 3D7 cDNA sequence has been deposited at GenBank (JN900252), is available from GeneDB v3.0 (Logan-Klumpler et al., 2012) (PF3D7_0831700) and is found as an incomplete (missing start codon) expressed sequence tag at PlasmoDB (Aurrecoechea et al., 2009) (PF07TR002). The experimentally validated cDNA sequence encodes a full-length Hsp70 protein, including an N-terminal hydrophobic signal sequence, followed by eight amino acids preceding the highly conserved ATPase domain (Figs S1 and S2).

Hsp70-x is a Laverania-specific chaperone

To understand the evolutionary origin of PfHsp70-x, we carried out fine-scale phylogenetic analysis using sequences from available apicomplexan genome databases. Our analyses reveal that genes encoding Hsp70-x can only be identified in malaria parasites belonging to the Laverania subgenus, represented by P. falciparum and the chimpanzee parasite Plasmodium reichenowi, and are absent in all other Plasmodium species and further apicomplexans (Figs 1 and S3). PfHsp70-x is most closely related to the canonical cytosolic PfHsp70-1, and is likely to have evolved from gene duplication on chromosome 8, followed by acquisition of targeting information and functional specialization (Fig. 1; full tree is shown in Fig. S4). Unusually, in our analysis Hsp70-x roots basally to the speciation of Plasmodium, maybe suggesting that this protein was present in the last common ancestor of the species studied, but has subsequently been lost from all species barring P. falciparum and P. reichenowi (Fig. 1).

Figure 1.

Phylogenetic tree of Hsp70s from Plasmodium, Cryptosporidium, Toxoplasma and Babesia spp. Three α-proteobacterial DnaK orthologues (Octadecabacter antarcticus, Paracoccus denitrificans, Bartonella henselae) were used as an out-group. Yeast Ssa1p is included to represent cytosolic Hsp70. Only Hsp70-x is shown as an expanded branch. Only P. falciparum (Pfal) and P. reichenowi (Prei) possess Hsp70-x (red shading). Bootstrap values < 95% are shown. Accession numbers of protein sequences used to generate the tree, and explanation of species abbreviation can be found in Fig. S3. A full tree including expanded branches of all Hsp70s used is shown in Fig. S4. Assembly of Prei-x coding sequence from individual clones is shown in Fig. S5.

PfHsp70-x localizes to the erythrocyte cytosol and PV

As the presence of a predicted hydrophobic signal sequence was a strong indication for a secretory protein, we explored the localization of PfHsp70-x by generating parasites expressing a GFP-tagged chimeric protein. We episomally expressed a construct encoding the first 32 N-terminal amino acids of the protein, fused to GFP and followed by the remainder of the PfHsp70-x coding sequence, under control of the PfCRT promoter (Fig. 2A). This N-terminal tagging arrangement is necessary to avoid blockage of the C-terminal ‘lid’ domain of the protein, which is essential for Hsp70 function. In erythrocytes infected with this cell line (3D7G:70x), a diffuse fluorescent signal can be detected in the erythrocyte cytosol and in a ring structure surrounding the parasite, likely the PV (Fig. 2B). Additionally, strongly fluorescent foci could be visualized within the host cytosol (Fig. 2B) which were highly mobile, as evidenced by live cell imaging (Movie S1). The distribution of fluorescence did not change considerably during the parasite life cycle (Fig. 2B).

Figure 2.

PfHsp70-x localizes to the host cell and PV.

A. Structure of the expressed chimeric protein. Red, signal peptide; blue, PfHsp70-x; green, GFP; PB, peptide binding domain. The region between putative signal peptide cleavage site and start of GFP is shown expanded. At the C-terminus of the protein is an –EEVN motif likely related to the –EEVD motif commonly found in cytosolic chaperones.

B. Live cell imaging of 3D7G:70x-infected erythrocytes. In the merge and overlay images: green, GFP; blue, Hoechst. In all stages punctate GFP fluorescence can be visualized in the host cell. Additionally, a ring of fluorescence indicative of a PV localization can also be seen.

C. Western blot analysis of 3D7G:70x-infected erythrocytes. Arrow indicates PfHsp70-x:GFP chimera; star represents endogenous PfHsp70-x. Size markers in kDa.

D. Indirect immunofluorescence localization of PfHsp70-x. In the merge and overlay images: green, α-70x/MRax; blue, Hoechst. R, ring stage; T, trophozoite stage; S, schizont stage.

E. Fractionation of infected cells with SLO. Schematic of experiment and Western blot analysis of samples prepared as presented in the scheme. 1 × 107 cell equivalents/lane using MRaX antisera. The fractionation was verified by analysing the distribution of marker proteins for the PV (PfSERP/SERA5), erythrocyte cytosol (HsHsc70), parasite cytosol (PfAldolase) and PVM (PfExp1). The distribution of PfHsp70-x is consistent with a soluble pool of protein in both the erythrocyte cytosol, and PV lumen. Size markers in kDa.

The partial export of PfHsp70-x was unexpected, as this protein does not contain the PEXEL/HT signal usually required to direct parasite proteins to the host cell (Hiller et al., 2004; Marti et al., 2004). To verify the localization of PfHsp70-x by secondary means we generated two independent antisera raised against PfHsp70-x (referred to as MRaX and BIaX). Both antisera recognized a band of the predicted molecular mass (c. 72 kDa with cleaved signal peptide) in Western blot analysis, with both pre-immune sera and host cell only controls being clear (Fig. S6). Additionally, Western blot analysis of protein extracts derived from 3D7G:70x-infected erythrocytes reveals two bands, corresponding to the endogenous PfHsp70-x, as well as the GFP-tagged chimera (predicted molecular mass 99 kDa, Fig. 2C). Having verified the specificity of the antisera, we carried out immunofluorescence analyses. The MRaX antisera labels ‘dotted’ structures within the infected erythrocyte, similar to live cell images of the 3D7G:70x cell line (Fig. 2D). We followed the distribution of the endogenous protein throughout the blood-stage life cycle. This analysis revealed that PfHsp70-x is exported already in early stages of parasite development, and that the fluorescent pattern does not recognizably change during progression through the life cycle (Fig. 2D). Similar results were obtained when using the BIaX antisera (Fig. S6). The lower level of diffuse cytosolic and PV staining seen when using the MRaX antisera is likely due to the conditions required to retain epitopes, but which do not allow good fixation of soluble proteins. As a further verification of the specificity of our antisera, we carried out immunofluorescent analysis on cells infected with the 3D7G:70x cell line, using the MRaX antisera. This analysis revealed a strong colocalization between the GFP signal, and that visualized using the specific antisera (Fig. S6). Control experiments on non-infected erythrocytes, and secondary antibody only controls revealed no significant fluorescence. In contrast to the above result, PfHsp70-1 is found only in association with parasite cytosol, as previously described (Fig. S6) (Gitau et al., 2011). For all following experiments we used the MRaX antisera.

As the immunofluorescence conditions required for detection of PfHsp70-x did not allow adequate fixation and thus immunofluorescent detection of soluble PV-resident proteins, we examined the distribution of the endogenous protein by cell fractionation. This analysis verified the distribution of PfHsp70-x, with partial export to the host cell, but also a substantial fraction being found within the PV (Fig. 2E). This data suggests that endogenous PfHsp70-x is present both in the erythrocyte cytosol and PV, and is thus consistent with the data obtained by imaging of the 3D7G:70x transgenic parasite line.

These data verify, by three independent means, and using two independently generated antibodies, that PfHsp70-x is partially exported to the host cytosol, but is also retained within the lumen of the PV.

PfHsp70-x partially localizes to the J-dots

The fluorescent pattern seen both with the GFP fusion and by immunofluorescence was indicative of localization to the J-dots, and was supported by the high mobility of these structures in intact cells (Movie S1) (Kulzer et al., 2010). To further investigate this, we carried out immunocolocalization analysis to localize PfHsp70-x in cells infected with the previously described PFE55CRT and PFA660CRT cell lines, which export two different GFP-tagged PfHsp40s to the J-dots (Kulzer et al., 2010). These analyses revealed, in both cases, a strong signal overlap (Fig. 3A), suggesting that PfHsp70-x is closely associated with at least two exported PfHsp40s at the J-dots.

Figure 3.

PfHsp70-x colocalizes with exported Hsp40s and PfEMP1.

A. Colocalization with exported Hsp40s. In merge and overlay: green, GFP; red, α-70x; blue, Hoechst. A high level of colocalization can be observed in both cases.

B. Colocalization with the ATS domain of PfEMP1. In merge and overlay: green, α-70x; red, α-ATS; blue, Hoechst. LR, late ring stage; ET, early trophozoite stage; LT, late trophozoite stage; S, schizont stage. Signal colocalization can be observed already in ring-infected cells, and gradually decreases during the progression to trophozoite and schizont stages.

PfHsp70-x shows stage-dependent colocalization with PfEMP1

We, and others, have proposed a potential role of exported Hsp40s in enabling trafficking of PfEMP1 to the surface of the infected cell (Maier et al., 2008; Kulzer et al., 2010). We were therefore interested in studying a potential interaction of PfHsp70-x with PfEMP1. To this end we carried out co-immunofluorescence assays on erythrocytes infected with parasites at various stages of development. These analyses revealed that, at early stages of parasite development, there is a high level of signal colocalization between PfHsp70-x and PfEMP1 (Fig. 3B). The level of colocalization drops during progression through the life cycle (Fig. 3B). Occasionally, both a late stage and early stage parasite could be captured in one frame. This internal control verifies our observations, with differing levels of colocalization being seen in the same frame (Fig. S7). As a further control, we analysed the same slides using confocal fluorescence microscopy. Again we observe colocalization between PfHsp70-x and PfEMP1 (Fig. S7). To exclude the possibility that PfHsp70-x partially localizes to further intra-erythrocytic structures such as Maurer's clefts (MC) and tethers, we carried out co-labelling analysis on cells infected with 3D7G:70x using the established MC markers SBP1 and STEVOR, and tethers marker MAHRP2 (Blisnick et al., 2000; Blythe et al., 2008; Pachlatko et al., 2010). These analyses revealed a lack of signal colocalization between the GFP-tagged chimera and the marker proteins (Fig. S7).

PfHsp70-x is found in a common complex with both exported PfHsp40s and PfEMP1

To investigate if PfHsp70-x, exported PfHsp40s and PfEMP1 are contained within a common complex, we applied a proximity ligation assay (PLA). This method is based on the ability of two complementary DNA strands to anneal if they are within 40 nm of each other. Following annealing, an in situ amplification step followed by addition of a fluorescent probe allows highly sensitive detection of protein–protein interactions (Fredriksson et al., 2002). In all three cases tested (PfHsp70-x/PfEMP1, PfHsp70-x/A660-GFP, PfHsp70-x/E55-GFP) we were able to detect a strong punctate fluorescent signal, strongly supporting a direct interaction between PfHsp70-x and PfEMP1, A660-GFP and E55-GFP respectively (Fig. 4A). As a negative control, we carried out the same experiment, but with the addition of only the PfHsp70-x antisera, revealing no non-specific signal in infected cells (Fig. 4A). To further substantiate these interactions, we performed co-immunoprecipitation experiments on protein extracts from erythrocytes infected with the PFE55CRT transfectant line. PFE55-GFP co-immunoprecipitates with PfHsp70-x only in the presence of anti-PfHsp70-x antibodies, but not in the negative control (agarose beads only, Fig. 4B). As a further negative control, we probed for human Hsc70, which is present in large amounts in the erythrocyte. HsHsc70 was not detected in any fraction, verifying the specificity of the interaction (Fig. 4B). We were not able to consistently detect PfEMP1 co-immunoprecipitating with PfHsp70-x, possibly due to the large molecular weight of this protein. To further understand the nature of complexes containing PfHsp70-x and exported PfHsp40s, we carried out gel filtration experiments on erythrocyte cytosolic fractions derived from the PFE55CRT transfectant line. This analysis reveals that PfHsp70-x is incorporated into macromolecular complexes with a vast array of molecular masses, ranging from approximately 100 kDa up to over 700 kDa (Fig. S8). Similarly, PFE55-GFP appears to be involved in many different complexes (Fig. S8). The protein composition of these complexes remains to be characterized in detail, but the size distribution of PfHsp70-x in gel filtration analysis differs from that of HsHsc70, suggesting that these two proteins may be incorporated into distinct complexes (Fig. S8), consistent with the co-immunoprecipitation experiments above.

Figure 4.

PfHsp70-x interacts with exported Hsp40s and PfEMP1.

A. PLA of (upper panel) PfHsp70-x/PfEMP1, (second panel) PfHsp70-x/A660-GFP, (third panel) PfHsp70-x/E55-GFP. A red signal indicates a positive interaction. Lower panel shows negative control. PfHsp70-x appears to closely interact with the ATS of PfEMP1, and both exported Hsp40s studied.

B. Co-immunoprecipitation using α-70x. Size markers in kDa. E55-GFP specifically co-precipitates with PfHsp70-x, whereas HsHsc70 is not detected.

Taken together, these experiments show that PfHsp70-x and exported PfHsp40s are present in high molecular weight complexes within the infected erythrocyte, and that PfHsp70-x directly and specifically interacts with exported PfHsp40s and also potentially with PfEMP1.

PfHsp70-x is expressed in parasite lines lacking PfHRPII

The gene encoding PfHsp70-x (hsp70x) is located directly adjacent to the hrpii gene, in a region often subject to chromosomal truncation events (Pologe and Ravetch, 1988; Gamboa et al., 2010) which lead to loss of hrpii. We wished to study if such truncation events could also cause loss of hsp70x. We thus carried out Western blot analysis on both 3D7 (which retain hrpii) and D10 (lacking hrpii) parasite strains. This analysis reveals that, despite loss of the hrpii gene (and thus protein) in D10, both D10 and 3D7 parasite still express PfHsp70-x (Fig. S9). This result adds support to the hypothesis that PfHsp70-x fulfils an important role in parasite survival, as loss of hsp70x would appear to be selected against, even in long-term cultured parasites.

PfHsp70-x targeting to the host cell requires an unusual export sequence

Although clearly transported to the host cell, PfHsp70-x does not contain a PEXEL/HT signal, which is usually required for protein traffic to the host cell (Hiller et al., 2004; Marti et al., 2004). To dissect the cryptic export signal of PfHsp70-x, we generated transfectant parasites expressing GFP-tagged nested deletions (Fig. 5A). Transgenic lines expressing GFP fused to up to the first 25 amino acids of PfHsp70-x exhibited a ring of fluorescence surrounding the parasite, characteristic of a localization to the lumen of the PV (Fig. 5B). Unexpectedly, a construct containing only the first 20 amino acids fused to GFP is also trafficked to the PV, despite lack of the predicted endogenous signal peptidase cleavage site. Upon further examination we noted that the cloning strategy reintroduced a potential signal peptidase recognition site, and it is thus likely that the signal sequence is cleaved at this new site. These data indicate that the first 25 amino acids of PfHsp70-x function as an ER-type signal sequence, which allows secretion into the PV. Upon addition of a further eight amino acids, the fluorescence signal changed dramatically, with the chimera now visualized in the lumen of the PV but also the host cell cytosol (Fig. 5B). Addition of further amino acids had minimal effect on the signal distribution, with GFP fluorescence being detected in both erythrocyte cytosol and PV, even upon inclusion of up to 45 N-terminal amino acids (Fig. 5B). These observations were verified by cell fractionation using streptolysin O (SLO) (Ansorge et al., 1996) (Fig. 5C). Western blot analysis did not reveal any evidence of export sequence processing suggesting that, contrary to PEXEL/HT containing proteins, the export signal remains intact (Chang et al., 2008; Boddey et al., 2010; Russo et al., 2010) (Fig. 5C). We do, however, note that due to the small size (eight amino acids) of the putative export motif, we may not be able to detect signal cleavage using this method. Targeting to the J-dots appears to be directed by sequences further downstream in the mature protein, as chimera containing only the first 45 amino acids are seen diffusely in the erythrocyte cytosol and do not concentrate into fluorescent foci (Fig. 5B).

Figure 5.

Targeting of PfHsp70-x to the host cell is dependent on an eight-amino-acid sequence.

A. Structure of N-terminal region of PfHsp70-x. SP, predicted signal peptide cleavage site. Numbering refers to amino acids from N-terminus. The linker region between putative signal sequence cleavage site and the start of the ATPase domain is shown in dark blue.

B. Live cell imaging of parasites expressing various N-terminal GFP fusion chimera. Numbers refer to amino acids from N-terminus. In merge and overlay: green, GFP; blue, Hoechst. Chimera containing up to 25AA show fluorescence exclusively in the PV, whereas chimera containing more than 32AA also show fluorescence in the host cell cytosol.

C. Streptolysin O treatment of erythrocytes infected with the transgenic lines shown in B. SNT, supernatant following SLO lysis; P, pellet following SLO lysis. The fractionation was verified by analysing the distribution of marker proteins for the PV (PfSERP/SERA5), erythrocyte cytosol (HsHsc70) and parasite cytosol (PfAldolase). Size markers in kDa. In parasite lines exporting GFP to the erythrocyte, a commonly reported GFP degradation product could also be detected, likely the result of non-specific uptake of GFP during the parasite's feeding process.

These experiments demonstrate that PfHsp70-x contains a novel octameric export signal between amino acids 25 and 32. Taken together, these data define a new minimal host cell-targeting signal, distinct in both sequence and position from previously reported export sequences.

Discussion

It has long been recognized that P. falciparum exports proteins to and through the host cell, the mature human erythrocyte. How exactly these proteins are able to be transported across the cytosol of a denucleated cell, lacking the necessary equipment for protein traffic remains until this point only poorly understood. We, and others have reported the export of parasite-encoded Hsp40/DnaJ proteins to the host cell, and linked export of these proteins to the transport of parasite proteins (including PfEMP1) across the host erythrocyte (Hiller et al., 2004; Maier et al., 2008; Kulzer et al., 2010). As Hsp40s usually fulfil their function in concert with Hsp70s (Kampinga and Craig, 2010), and in the absence of an exported parasite Hsp70, it has been generally assumed that infected erythrocytes must contain a ‘patchwork’ Hsp70/40 complex, composed of parasite Hsp40s and residual human Hsp70s (Charpian and Przyborski, 2008).

Here, using both specific antisera and GFP chimera, we provide solid evidence for trafficking of a parasite-encoded Hsp70, PfHsp70-x, to both the PV and host cell. PfHsp70-x may thus play an important role on both cis- and trans-sides of the PVM. PfHsp70-x is incorporated into high molecular weight complexes within the infected cell, and these complexes contain both exported PfHsp40s and possibly PfEMP1. We suggest that the exported chaperone complex here reported may be involved in chaperoning parasite-encoded proteins, including PfEMP1, through the human host cell, and in aiding insertion of these proteins into the erythrocyte plasma membrane. Previous studies have noted that PfEMP1, although predicted to be a membrane protein, is actually synthesized as a soluble protein, and remains in a largely soluble/membrane peripheral state during its transport to the host cell surface (Knuepfer et al., 2005; Papakrivos et al., 2005). Fluorescence recovery after photobleaching experiments has suggested that PfEMP1 is trafficked through the host cell as part of a multi-protein complex, possibly containing chaperones (Knuepfer et al., 2005). Our data may well go some way towards mechanistically explaining this observation, as a role for a Hsp40/70 complex in both cytosolic chaperoning and post-translational membrane insertion of proteins is not without precedent (Pfanner and Geissler, 2001; Soll and Schleiff, 2004; Hegde and Keenan, 2011).

Although definite evidence for export of Hsp40 proteins only so far exists in P. falciparum-infected cells, many other Plasmodium spp. encode potentially exported Hsp40s, but do not encode Hsp70-x. In these cases other molecules must provide this functionality, possibly residual host Hsp70. In P. falciparum-infected cells, we cannot formally exclude a role for HsHsp70 in protein transport processes, but rather suggest that PfHsp70-x is required to handle substrates over and above the core Exportome of all Plasmodium species. In this light, it is interesting to note that, as well as possessing PfHsp70-x, P. falciparum is predicted to export a much larger number of Hsp40 proteins than other species (Sargeant et al., 2006). This is possibly driven by the expansion of exported proteins in this species, and the need for additional chaperones and co-chaperones that can manipulate P. falciparum-specific substrates such as PfEMP1.

PfHsp70-x transport to the host cytosol requires an eight-amino-acid sequence located directly following the predicted signal sequence. This sequence lacks features generally conserved in the well-characterized PEXEL/HT motif (Hiller et al., 2004; Marti et al., 2004). So-called PEXEL negative exported proteins (PNEPS) have been described, which are exported despite lacking a recognizable PEXEL (Spielmann and Gilberger, 2010). Most of these proteins share the presence of a predicted membrane-spanning region, which is required for transport to the host cell (Spielmann and Gilberger, 2010); however, two soluble exported PNEPS have also been reported, MAHRP2 and REX1 (Dixon et al., 2008; Pachlatko et al., 2010). Of these, only REX1 contains a similar N-terminal organization to PfHsp70-x, with a predicted ER-type signal sequence, and a short (10AA) downstream export sequence, NSIKENANSK (Dixon et al., 2008). Comparing this sequence to the minimal export sequence we have identified in PfHsp70-x (ASNNAEES) reveals that, although six amino acids are shared between the sequences, there is no obvious conservation of amino acid order. Directed single amino acid mutagenesis experiments on these minimal export sequences will be required to decipher any possible functional relatedness of these signals.

In P. falciparum, hsp70x is located in a sub-telomeric region on chromosome 8, in a region lacking synteny with the chromosomes of other Plasmodium species (Kooij et al., 2005). This region is vulnerable to chromosomal breakage in both long-term cultured, as well as patient isolates (Pologe and Ravetch, 1988; Gamboa et al., 2010). Breakage has been shown to commonly occur in and around the hrpii gene (Pologe and Ravetch, 1988) (MAL7P1.231), which lies directly adjacent to hsp70x. Despite this, in all parasite lines so far studied in detail, hsp70x has been retained (Scherf and Mattei, 1992; Gamboa et al., 2010). Although circumstantial, these data imply a highly important, possibly essential role for PfHsp70-x in parasite survival.

A recent publication suggests that Plasmodium Hsp70s are sufficiently different to their human counterparts to allow identification of small molecules selectively inhibiting Hsp70 function (Cockburn et al., 2011). This tactic has the potential to target parasite proteins that are, by necessity, highly conserved and subject to strong structural and functional constraints limiting their genetic diversity and thus ability to mutate.

To conclude: phylogenetic linkage, localization, interaction analyses and chromosomal positioning of Hsp70-x are strong indications for a role of this protein in transport of parasite proteins, including possibly EMP1, through the erythrocyte cytosol. Indeed, our data could be interpreted to mean that the appearance of Hsp70-x during evolution may have been a prerequisite for the expansion of surface-expressed parasite antigens, and thus the pathology associated with P. falciparum infection.

Experimental procedures

Phylogenetic analysis

Hsp70 sequences were retrieved by blast (Akaike, 1974) of 14 yeast Hsp70 sequences on available RefSeq proteins from selected apicomplexans and by manual text and motif-based searches at EupathDB (Aurrecoechea et al., 2010). Best hits from all species were aligned with MAFFT (Katoh and Toh, 2008) and a phylogenetic tree was reconstructed with PhyML3 (Guindon et al., 2010) using the best-fit model as inferred with ProtTest3 (Darriba et al., 2011) by the Akaike information criterion measure (Akaike, 1974).

Parasite culture and transfection

Plasmodium falciparum (clone 3D7) was cultured as previously described (Trager and Jensen, 1976). Ring stage parasites were transfected as previously described (Fidock and Wellems, 1997) and selected with 2.5 nM WR99210 (a kind gift of Jacobus Pharmaceuticals). Generation of transfection constructs is described in Appendix S1.

Antisera

MRaX

Rabbit antiserum (raised against peptide CQEPQKAEATNLRGRNS) was generated commercially (Eurogentec, Belgium) (peptide position shown in Fig. S2).

BIaX

Rabbit anti-protein antiserum was raised to the recombinantly expressed C-terminal region of PfHsp70-x (amino acids 579-stop). Anti-HRPII was obtained commercially (Abcam, ab9206).

Live cell imaging and immunofluorescence

Live imaging and immunofluorescence experiments were carried out as previously described (Kulzer et al., 2010). Details of the experimental protocol and antibodies used can be found in Appendix S1.

Proximity ligation assay

Proximity ligation assay was performed according to the manufacturer's guidelines using the Duolink In Situ PLA probes and Duolink In Situ detection reagent orange (Olink Bioscience). Rabbit anti-70x/MRaX (1/500), mouse anti-ATS [1/50 (Waterkeyn et al., 2000)] and goat anti-GFP (1/1500; Rockland) were used in combination with Duolink PLA probes anti-rabbit-MINUS, anti-mouse-PLUS and anti-goat-PLUS.

Cell fractionation

Trophozoite-infected erythrocytes were washed in PBS pH 7.4 and subjected to either SLO permeabilization as described previously (Ansorge et al., 1996) or hypotonic lysis using 10 mM Tris pH 7.4. Details of the experimental protocol can be found in Appendix S1.

Acknowledgements

We particularly wish to thank Ulrike Boehme and the Pathogen Sequencing Unit at the Sanger Centre for their help with sequence analysis. Kerstin Maas-Enriquez provided excellent technical assistance and Addmore Shonhai assisted in choice of PfHsp70x sequences for immunization. We furthermore thank Rowan Hatherley for assistance with preparation of r70x and Alan Cowman for the gift of anti-ATS monoclonal antibodies. This work was supported by DFG project PR1099/3-1 (SPP1580, J. M. P.), and by the Australian NHMRC (B. S. C.). The authors gratefully acknowledge the contribution to this work of the Victorian Operational Infrastructure Support Program. We thank the Australian Red Cross Blood Bank for the provision of human blood and serum. S. C. is supported by a Monash Graduate Scholarship, and gratefully acknowledges funding from OzEMalaR Researcher Exchange and the Australian Society of Parasitology.

Ancillary