A High-Resolution Map of SBP1 Interactomes in Plasmodium falciparum-infected Erythrocytes.

The pathogenesis of malaria parasites depends on host erythrocyte modifications that are facilitated by parasite proteins exported to the host cytoplasm. These exported proteins form a trafficking complex in the host cytoplasm that transports virulence determinants to the erythrocyte surface; this complex is thus essential for malaria virulence. Here, we report a comprehensive interaction network map of this complex. We developed authentic, unbiased, highly sensitive proteomic approaches to determine the proteins that interact with a core component of the complex, SBP1 (skeleton-binding protein 1). SBP1 interactomes revealed numerous exported proteins and potential interactors associated with SBP1 intracellular trafficking. We identified several host-parasite protein interactions and linked the exported protein MAL8P1.4 to Plasmodium falciparum virulence in infected erythrocytes. Our study highlights the complicated interplay between parasite and host proteins in the host cytoplasm and provides an interaction dataset connecting dozens of exported proteins required for P. falciparum virulence.

Introduction

Malaria remains one of the world's most important infectious diseases, affecting ∼200 million people worldwide annually (WHO, 2018). When Plasmodium falciparum, one of the most virulent forms of the human malaria parasite, establishes infection in host erythrocytes, the parasites export numerous proteins to the host cell's cytoplasm and plasma membrane to drastically remodel the host cell (Hiller et al., 2004, Maier et al., 2009). These exported proteins are collectively referred to as the exportome, and some of these proteins form a large protein complex that leads to profound structural and morphological changes in the host erythrocytes (LaCount et al., 2005); for example, Maurer's clefts (Lanzer et al., 2006) and knobs (Wickham et al., 2001) are established in the cytoplasm and on the surface of erythrocytes, respectively. Consequently, the infected erythrocytes become more rigid and adhere to the vascular endothelium, which prevents clearance of the infected erythrocytes by the spleen and subsequently disrupts normal blood flow, resulting in severe malaria in humans (De Niz et al., 2016, Maier et al., 2008).

The adherence of infected erythrocytes to the vascular endothelium is mediated by interactions between parasite adhesins on the erythrocyte surface and host endothelial receptors (Fairhurst et al., 2005, Janes et al., 2011, Waller et al., 2003). Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) is an antigenically variant adhesin that is transported to knobs on the erythrocyte surface (Waller et al., 1999). Knobs are macromolecular complexes of knob-associated histidine-rich protein (KAHRP) that anchor PfEMP1 to the membrane skeleton (Oh et al., 2000, Waller et al., 1999). In contrast, Maurer's clefts are involved in the trafficking of PfEMP1 to the erythrocyte surface (Lanzer et al., 2006, Maier et al., 2008, Wickham et al., 2001). Many exported proteins, represented by skeleton-binding protein 1 (SBP1) (Cooke et al., 2006, Maier et al., 2007), membrane-associated histidine-rich protein 1 (MAHRP1) (Spycher et al., 2008), ring-exported protein 1 (REX1) (Hanssen et al., 2008), subtelomeric variant open reading frame (STEVOR) (Przyborski et al., 2005), and PfEMP1-trafficking protein 1 and 5 (PTP1 and PTP5) (Maier et al., 2008, Rug et al., 2014), have been shown to reside in Maurer's clefts. Some of these exported proteins are essential for the intracellular transport of PfEMP1 to the erythrocyte surface, suggesting that they form a large protein complex in the Maurer's clefts that serves as protein-trafficking machinery to transport exported proteins to their final destinations (Rug et al., 2014). However, essential information regarding the interactions between these exported proteins is lacking, because of the technical difficulties of studying protein-protein interactions in the cytoplasm of erythrocytes infected with Plasmodium falciparum (Batinovic et al., 2017, Rug et al., 2014).

The P. falciparum exportome had been predicted to comprise approximately 400 proteins since the discovery of a motif sequence called PEXEL (P. falciparum exported elements) or HT (host-targeting sequence), which is conserved at the N-terminal end of many exported proteins (Hiller et al., 2004, Marti et al., 2004). However, a previous study also identified the presence of many proteins that lack the canonical PEXEL/HT motif but could be efficiently exported to host cytoplasm (Heiber et al., 2013), thereby complicating the identification of the exported proteins that comprise the P. falciparum exportome. These PEXEL-negative exported proteins (PNEPs) include SBP1, MAHRP1, and REX1, all of which are related to and are indispensable for malaria virulence (Cooke et al., 2006, Hanssen et al., 2008, Maier et al., 2007, Spycher et al., 2008). Given the difficulty to predict and identify PNEPs based on protein sequences, an alternative approach is needed to directly identify PNEPs based on protein-protein interactions in the host cytoplasm.

In this study, we developed an alternative proteomic approach to identify exported proteins involved in the trafficking complex and to clarify the P. falciparum exportome. Highly sensitive mass spectrometry identified multiple proteins that interact with SBP1 through its intraerythrocytic-trafficking process. Further interaction, localization, and functional assays demonstrated that the SBP1 interactomes established in our study represent a powerful and invaluable platform to identify exported proteins related to severe malaria caused by P. falciparum.

Results

A High-Resolution Map of SBP1 Interactomes

To identify proteins that interact with SBP1, we preformed FLAG-tag-based immunoprecipitations with lysates of erythrocytes infected with parasites overexpressing C-terminally FLAG-tagged SBP1 (SBP1-FLAG), and analyzed the co-immunoprecipitated proteins by mass spectrometry (Gorai et al., 2012, Watanabe et al., 2014). We first generated parasites episomally expressing SBP1-FLAG and confirmed that the FLAG-tagged SBP1 retained the localization of wild-type SBP1 in Maurer's clefts (Figures 1A and 1B).

Figure 1

An Unbiased Comparative Proteomic Approach to Identifying Proteins That Interact with SBP1

(A) Immunofluorescence assay on 3D7/SBP1-FLAG cells: the FLAG signal (green) co-localizes with that of endogenous protein (SBP1; red). Erythrocyte and parasitophorous vacuole membranes are circled by solid and dotted lines, respectively. Scale bar, 3 μm.

(B) Western blot analysis of 3D7/SBP1-FLAG cells: parasite-infected RBCs were purified by magnetic-activated cell sorting (MACS), lysed with SDS/Triton lysis buffer, and equivalent amounts of samples were analyzed by western blot.

(C and D) (C) Western blot and (D) silver stain analyses of proteins co-immunoprecipitated with SBP1-FLAG. The arrow indicates a bait protein (SBP1-FLAG).

(E) Venn diagram comparing proteins identified in 3D7 and SBP1-FLAG proteomes. A total of 256 proteins were specifically detected in the SBP1-FLAG proteome.

(F and G) (F) Average and (G) distribution of Mascot Protein Scores of proteins in two proteomic datasets. P values were calculated by using Mann-Whitney's U and Kolmogorov-Smirnov tests described in Transparent Methods supplemental file.

(H) Unbiased comparative analysis of proteins in two proteomes based on Mascot protein score; 106 proteins were selected by fold change with a cutoff of 5.0.

(I) Selected representatives of top-ranked proteins identified in the SBP1 interactomes.

Human erythrocytes were infected with 3D7 wild-type parasites, or parasites expressing SBP1-FLAG, and late trophozoite or schizont-enriched red blood cells (RBCs) were purified via Percoll gradient centrifugation and magnetic isolation. Whole-cell lysates were prepared, immunoprecipitated with an anti-FLAG antibody, and subjected to SDS-PAGE followed by western blot and silver staining. Although the bait protein was detected among the precipitants from the lysates of RBCs infected with SBP1-FLAG-expressing parasites (Figure 1C), multiple proteins were non-specifically bound to the immunoprecipitation beads and specific bands for immunoprecipitates from the SBP1-FLAG samples were not detected (Figure 1D). We, therefore, employed a highly sensitive shotgun proteomic approach, and compared the proteomes of the two precipitates. The SBP1-FLAG and 3D7 proteomes contained 1,381 and 1,367 proteins, respectively, with 1,125 (∼81%) overlapping proteins (Figure 1E and Table S1). No statistically significant differences were observed in the power of protein detection or protein coverage between the two proteomes, indicating that the proteomes were comparable (Figures 1F and 1G). A total of 256 proteins, 51 host and 205 parasite proteins, was specifically detected in only the SBP1-FLAG proteome (Figure 1E); however, these included many proteins with a relatively low Mascot protein score (Table S2).

We then refined the list of proteins based on a comparison of their Mascot protein scores and identified 106 proteins (88 parasite and 18 host proteins) that specifically co-immunoprecipitated with SBP1 (Figure 1H and Table S3; details are provided in the Transparent Methods). In addition to spectrin (SPTA1 and SPTB), a host factor known to interact with SBP1 (Blisnick et al., 2000, Kats et al., 2015), we identified several parasite proteins, including PfEMP1, PIESP2, and PTP1, all of which form a large protein complex with SBP1 inside host cells (Batinovic et al., 2017, Cooke et al., 2006, Maier et al., 2007, Mbengue et al., 2015, Rug et al., 2014), as top-ranked hit proteins, thereby demonstrating the reliability of our approach relative to previous studies (Figure 1I).

We next performed a functional enrichment analysis (Wang et al., 2015) and categorized the potential SBP1 interactors according to subcellular compartments. Gene Ontology (GO) analysis identified Maurer's clefts (Benjamini-adjusted p = 1.6 × 10−4), and the cytoskeleton (Benjamini-adjusted p = 0.032) as the most enriched by parasite and host proteins, respectively (Figure 2A). In addition, most of the high-ranked GO terms for the parasite proteins were predominantly related to proteins exported to host cytoplasm (p = 1.6 – 3.5 × 10−4) (Table S4). The GO analysis further identified enrichment in multiple parasite cellular compartments, including the Sec61 translocon complex (Benjamini-adjusted p = 0.0089), rough endoplasmic reticulum (Benjamini-adjusted p = 0.014), and nuclear pore (Benjamini-adjusted p = 0.024) (Figures 2A and 2B). These results suggest that the identified proteins represent parasite proteins exported to the host cytoplasm, where SBP1 functions to anchor the Maurer's clefts to the host cytoskeleton (Blisnick et al., 2000, Kats et al., 2015) as well as proteins associated with the intracellular trafficking of SBP1 within the parasites.

Figure 2

Verification and Validation of SBP1 Interactomes

(A) Functional enrichment analyses of likely SBP1-interacting proteins. Gene Ontology analysis was conducted for the parasite (upper panel) and host proteins (lower panel), respectively.

(B) The relationships of the GO terms enriched for the parasite proteins were visualized using Cytoscape with the ClueGO plugin.

(C and D) Biological binding and functional validation of 11 SBP1 interactors by use of co-immunoprecipitation (C) and immunofluorescence (D) assays on endogenous SBP1-FLAG cells. The asterisk indicates a non-specific band that reacted with the light chain of IgG. Scale bar, 3 μm. The ratio of co-localization between two factors was analyzed by using Python 3.7.3 and is shown as the mean ± SD (n = 10).

To verify and validate our findings, we focused on the following 11 proteins: five parasite proteins that interacted with the Maurer's clefts (PIESP2, REX1, REX2, MAHRP1, and MAHRP2), three parasite proteins that have not previously been reported to interact with SBP1 (PF10_0018, STEVOR, and TryThrA), and three host proteins (STOM, KPNB1, and MPP1). To test whether these proteins interact with endogenous SBP1, we generated parasites expressing FLAG-tagged SBP1 under the control of endogenous promoters by introducing a FLAG tag after the full-length C terminus of SBP1 using single crossover homologous recombination (Figure S1). Whole-cell lysates were prepared, and co-immunoprecipitation was carried out as described above. The precipitated proteins were subjected to SDS-PAGE followed by western blot using polyclonal serum from mice immunized with recombinant parasite proteins (details are provided in the Transparent Methods) and commercially available antibodies against the host proteins. We confirmed that all the 11 proteins tested co-immunoprecipitated with endogenous SBP1 by western blot analyses (Figure 2C). Of note, six of the parasite proteins (PIESP2, REX1, REX2, MAHRP1, STEVOR, and TryThrA) co-localized with SBP1 in the Maurer's clefts, whereas the remaining five proteins (two parasite proteins, PF10_0018 and MAHRP2, and three host proteins, STOM, KPNB1, and MPP1) showed partial co-localization with SBP1 (Figure 2D), presumably reflecting a different role for these proteins in the host cytoplasm; that is, the “full” co-localization proteins may be involved in the trafficking complex in the Maurer's clefts, whereas the “partial” co-localization proteins may be involved in anchoring the Maurer's clefts. Taken together, our unbiased comparative proteomic approach allowed us to systematically identify SBP1-interacting candidates and establish a comprehensive map of SBP1 interactomes in human erythrocytes infected with P. falciparum.

Host-Parasite Interactions Based on SBP1 Interactomes

Our identification of SBP1-interacting candidates sheds light on the complex interplay between multiple parasite proteins and SBP1 throughout the trafficking pathway and highlights plausible parasite-host interactions in the host cytoplasm. Of the host proteins that interacted with SBP1, we focused on STOM and MPP1, which are involved in the membrane organization of erythrocytes (Biernatowska et al., 2017, Lach et al., 2012, Salzer and Prohaska, 2001). To determine whether these proteins are associated with the erythrocyte cytoskeleton, we examined their solubility (Figure 3A). Solubility assays revealed that STOM remained exclusively in the pellet corresponding to membrane ghost fractions (similar to spectrin), whereas a portion of MPP1 was released into the detergent-sensitive supernatant fractions, suggesting the presence of soluble MPP1 in the host cytoplasm, consistent with previous studies (Biernatowska et al., 2017, Lach et al., 2012).

Figure 3

MPP1 Is Associated with SBP1 in Maurer's Clefts

(A) Solubility of the host STOM and MPP1 proteins, which interacted with SBP1; uninfected and MACS-purified P. falciparum 3D7-infected RBCs (iRBCs) were lysed with 0.09% saponin lysis buffer and equivalent amounts of supernatant and pellet samples were analyzed by western blot.

(B) Immunofluorescence assay of host factors in 3D7/ΔSBP1 cells. Scale bar, 2 μm.

(C) Double-stained immunoelectron microscopy of tetanolysin-permeabilized iRBCs to confirm MPP1 association with SBP1 in Maurer's clefts. SBP1 and MPP1 were labeled with 5- and 15-nm gold particles, respectively. MPP1 associated with SBP1 in Maurer's clefts apposed to the RBC membrane. MC, Maurer's clefts; P, parasites; RBCM, red blood cell membrane. Scale bars, 1 μm and 200 nm, for left and right panels, respectively.

To examine the effects of SBP1 on the localization of these two host proteins, we next generated SBP1-deficient transgenic parasites (Cooke et al., 2006, Maier et al., 2007) and analyzed the protein localization in erythrocytes infected with wild-type or ΔSBP1 parasites (Figure 3B). The absence of SBP1 did not influence the localization of these two proteins; the MPP1 signal was detected in the Maurer's clefts and infected RBC (iRBC) membrane regardless of whether SBP1 was present. To further examine whether MPP1 associates with SBP1 in Maurer's clefts, we performed immunoelectron microscopy using gold particle co-labeling of MPP1 and SBP1 (Figure 3C). We observed that MPP1 was closely associated with SBP1 in the Maurer's clefts, but that a portion of MPP1 was detectable just beneath the erythrocyte membrane where the SBP1 signal was not detected (Figure 3C). Thus, we identified the host protein MPP1 as being recruited into Maurer's clefts, suggesting that it may play a role in the membrane organization of these clefts.

Using SBP1 Interactomes to Reveal the Exportome

Previous proteomic studies of the protein components of the trafficking complex (Batinovic et al., 2017, Rug et al., 2014) and of secreted microvesicles (Mantel et al., 2013) identified a limited number of exported proteins (Figure S2). Our SBP1 interactomes highly enrich our knowledge of the known exported proteins identified in previous proteomic studies (Figure S2) and reveal a previously unknown association between a host protein, MPP1, and the Maurer's clefts.

Among the 88 parasite proteins identified in our SBP1 interactomes, we found 63 proteins with the canonical PEXEL motif [(R/K)xLx(D/E/Q)], which is present in many exported proteins (Figure 4A). We focused on the following 16 proteins: 5 proteins with the PEXEL motif (PF10_0018, PF10_0020, PCNA2, PF08_0069, and PF10_0208) and 11 proteins lacking the PEXEL motif (ACS3, HGPRT, MAL7P1.170, PF07_0065, PFF0290w, PF14_0057, PFB0826c, ACS7, VP1, VPS29, and PF13_0012a), and re-examined their intracellular localization.

Figure 4

GFP-Based Protein Export Assay Identifies Four Exported Proteins

(A) Abundance of proteins with a PEXEL sequence [(R/K)xLx(D/E/Q)] among 88 parasite factors that interacted with SBP1.

(B and C) Live-cell images of parasites expressing the N terminus or full-length of 5 PEXEL (B) and 11 non-PEXEL (C) proteins fused with GFP. GFP alone and GFP-fused SBP1 and KAHRP are shown as controls. Nuclei are stained with DAPI. Scale bar, 2 μm. The expression of GFP fusion proteins for the cell lines tested is shown in Figure S3A. The results of immunofluorescence assays using anti-GFP antibodies on three mutants showing weak GFP fluorescence (indicated by the asterisk) are shown in Figure S3B.

(D) Western blot analyses using anti-GFP with extracts from selective permeabilization assays show bands with full-length or truncated versions of four exported GFP fusion proteins. SN, supernatants; Tet, tetanolysin (host cytosol); Sap, saponin (Maurer's clefts and parasitophorous vacuoles); P, pellet (final pellet); REX3, soluble parasite proteins in host cells; EXP2, parasitophorous vacuole marker. Asterisk, degradation product.

To test whether these proteins are exported into the host cytoplasm, we conducted a GFP-based protein export assay as established previously (Dixon et al., 2008, Heiber et al., 2013, Spielmann et al., 2006). The full length or N terminus of each candidate was C-terminally tagged with GFP and expressed episomally in P. falciparum. In parallel, GFP alone and GFP-tagged SBP1 and KAHRP were expressed and used as controls. The expression of the GFP-fused proteins of each mutant was confirmed by western blot (Figure S3A). Of the mutant cell lines tested, three showed weak GFP fluorescence (PF10_0208, ACS3, and ACS7), nine showed no evidence of export (PCNA2, PF08_0069, HGPRT, PF07_0065, PFF0290w, PF14_0057, PFB0826c, VP1, and VPS29), and four showed export (PF10_0018, PF10_0020, MAL7P1.170, and PF13_0012a) (Figures 4B and 4C). For the three mutants showing weak GFP fluorescence, we performed immunofluorescence staining using an α-GFP antibody and showed no evidence of export (Figure S3B). Of the four exported proteins, PF10_0018-GFP, PF10_0020-GFP, and MAL7P1.170-GFP were distributed throughout the host cytoplasm, whereas PF13_0012a-GFP produced characteristic fluorescence in the Maurer's clefts.

To define the subcellular localization of the four exported proteins, we conducted a selective permeabilization assay on erythrocytes infected with each mutant parasite (Boddey et al., 2016, Gruring et al., 2012, Mantel et al., 2016) (Figure 4D). Sequential treatment of iRBCs with tetanolysin and saponin separates the cells into RBC cytosol, a parasitophorous vacuole and Maurer's clefts, a parasite membrane fraction, and a host membrane fraction (Mantel et al., 2016). Western blot analyses demonstrated that all four exported proteins were present in the soluble fractions representing RBC cytosol, thereby confirming the observed localization of these proteins in the live-cell imaging analyses (Figure 4D). Thus the SBP1 interactomes established by our alternative proteomic approaches led us to discover four exported proteins regardless of whether they contained the PEXEL motif.

Necessity and Significance of SBP1-Interacting Proteins

In light of the functional importance of SBP1 for the host cell remodeling that is required for malaria virulence, we asked whether disruption of genes encoding SBP1 interactors could affect malaria survival and pathogenesis. To concentrate on genes that encoded exported proteins for knockout experiments, we first curated a list of genes by selecting those specifically expressed during the ring stage from Table S2, as well as the top-ranked 25 hit genes from Table S3, resulting in a set of 37 genes. This set contained seven genes that encode well-known exported proteins required for malaria virulence (SBP1, MAHRP1, MAHRP2, REX1, REX2, PTP1, and PTP5) (Figure S4). We therefore focused on 13 SBP1-interacting genes (PF10_0018, ACS3, PF13_0309, PF08_0069, vapA, FIKK10.2, TryThrA, ACS7, VP1, STARP, PFL2530w, PF14_0045, and MAL8P1.4) whose functions remain to be fully elucidated, plus SBP1, and performed targeted gene disruption experiments by using the double homologous recombination system described previously (Maier et al., 2008) (Figures S5 and S6).

Of 14 genes tested, seven (SBP1, PF10_0018, FIKK10.2, TryThrA, STARP, PFL2530w, and MAL8P1.4) could be genetically disrupted, suggesting that the other seven genes are essential to the parasite during the asexual blood stage (Figures 5A and S4–S6). Of the seven genes that were disrupted, six encode exported proteins (SBP1, PF10_0018, FIKK10.2, TryThrA, STARP, and MAL8P1.4), which is consistent previous findings that genes encoding exported proteins are less essential for parasite asexual development than those encoding non-exported proteins (Bushell et al., 2017, Maier et al., 2008). We did not see any substantial differences in parasite growth or invasion between the wild-type parasite and each knockout mutant (Figures S7A and S7B). These results suggest that the exported proteins encoded by these six genes may play roles in host cell remodeling by interacting with SBP1.

Figure 5

Necessity and Significance of SBP1 Interactors for Parasites in the Asexual Blood Stage

(A) Genes selected for knockout studies. Shown in the first to fourth columns are the accession number, previous ID, gene symbol, and gene description from PlasmoDB (http://www.plasmo.org). The fifth column shows the subcellular localization of the protein in the infected erythrocyte if known and whether or not the protein is exported based on the results of the protein export assay shown in Figure 4. The sixth column shows a transcriptional profile where yellow denotes an increased period of transcription and blue denotes a decreased period or no transcription. Gene expression data were obtained from the DeRisi's microarray data available in PlasmoDB. R, ring; T, trophozoite; S, schizont asexual life cycle stages. The seventh column shows the Protein Mascot Score of each protein in the SBP1-FLAG/3D7 proteomes; - denotes not detected. The eighth column refers to whether the gene can be genetically disrupted (Yes) or not (No). Asterisk, undetermined because of delayed growth of parasites transfected with gene-targeting plasmid.

(B) Adherence of mutant parasite-infected erythrocytes to CD36 receptors under static conditions. Two clones of each mutant parasite were tested for binding to CD36 receptors; adherent cells were counted and are shown as a ratio relative to the level of 3D7-infected erythrocytes. The values presented are the average of three independent experiments ± SD. P values were calculated by using one-way ANOVA with Tukey's post-hoc test (*P < 0.05).

(C) Overview of the protein-protein interactions uncovered by our SBP1 interactomes. Three host proteins (STOM, KPNB1, and MPP1), two parasite proteins (PF10_0018, and PF10_0020), and six parasite exported proteins (STEVOR, Pf332, TryThrA, STARP, PF13_0012a, and MAL8P1.4) identified in our study are shown in boldface. Proteins involved in cytoadherence are underlined.

We then focused on three knockout cell lines (ΔTryThrA, ΔSTARP, and ΔMAL8P1.4), in which genes that encoded proteins located in Maurer's clefts were disrupted, and tested their cytoadherence by using a CD36 static binding assay. We found that ΔTryThrA and ΔSTARP showed comparable levels of cytoadherence relative to those of the wild-type, whereas disruption of the MAL8P1.4 gene resulted in increased cytoadherence of iRBCs to CD36 (Figure 5B). Intriguingly, we did not observe any changes in the localization of known exported proteins, represented by SBP1, PfEMP1, MAHRP1, and REX1, or any alternations in knob formation or Maurer's cleft morphology in the erythrocytes infected with ΔMAL8P1.4 parasites compared with those in the erythrocytes infected with wild-type parasites (Figures S7C–S7G). Thus, by performing gene knockout experiments followed by cytoadherence assays on 14 genes that encode potential SBP1 interactors, we identified MAL8P1.4 as being associated with the cytoadherence of iRBCs and thereby affecting malaria virulence.

Discussion

Here, we developed an authentic, unbiased, and highly sensitive comparative proteomic approach to characterize proteins that interact with SBP1. This approach allowed us to identify more than 100 putative proteins, which were enriched with known exported proteins that play central roles in host cell remodeling. On the basis of SBP1 interactomes, we further conducted systematic analyses to validate the factors that were associated with SBP1 in the cytoplasm of erythrocytes (Figure 5C) and identified a parasite protein, MAL8P1.4, which regulates the cytoadherence of P. falciparum-infected erythrocytes to vascular endothelial receptors.

Regarding the quality of the interactome we have presented, it must be noted that we cannot exclude the possibility that our dataset contains false-positives. However, by performing validation studies on previously known and unknown SBP1-interacting proteins, we have been able to show that our SBP1 interactomes are reliable and invaluable for the clarification of the exportome. This significantly increases the value of our dataset as a resource because the entire dataset used for our analyses is presented here and can be analyzed independently. It is also important to note that our established system to identify SBP1-interacting proteins is completely authentic and unbiased regarding protein identification. Because of technical difficulties using co-immunoprecipitation to study protein-protein interactions in the cytoplasm of erythrocytes infected with parasites, most previous interactome studies (Batinovic et al., 2017, Dietz et al., 2014, Oberli et al., 2016) have been done on parasite lysates after saponin lysis to remove the host cell cytosol, Maurer's clefts, and soluble parasitophorous vacuole contents. This is because erythrocytes contain abundant cytoskeleton proteins, which easily and non-specifically bind to beads, leading to loss of important information regarding protein interactions in the host cytoplasm. To overcome these difficulties, we carried out co-immunoprecipitation experiments on whole-cell lysates of purified iRBCs, analyzed the precipitated proteins by using shotgun proteomics, and performed quantitative analyses of the two proteomes obtained. The dataset derived by taking this approach includes many of the exported proteins previously characterized. In fact, our proteomic data covered 16 of the known exported proteins (PfEMP1, FIKK10.1, PTP1, PFF0090w, PFA0210c, GEXP10, SBP1, PIESP2, REX1, MAHRP1, MAHRP2, Pf332, REX2, PFE0050w, GEXP07, and MC-2TM) of the 21 overlapping proteins identified in three previous proteomic studies (Batinovic et al., 2017, Mantel et al., 2013, Rug et al., 2014). Our study further confirmed that these proteins made authentic interactions in the host cytoplasm (Figure S2). With these advantages, our SBP1 interactomes could serve as intermediaries to connect previous studies (Batinovic et al., 2017, Dietz et al., 2014, Mantel et al., 2013, Oberli et al., 2016, Rug et al., 2014) of exported proteins. The application of our proteomic approaches to other exported proteins resident in Maurer's clefts would greatly increase our understanding of the interaction network of the P. falciparum exportome.

Recently, the P. falciparum orthologs of SBP1 and MAHRP1 were discovered in the rodent malaria Plasmodium berghei by De Niz et al. (De Niz et al., 2016), and disruption of these genes resulted in the decreased cytoadherence of iRBCs to CD36 in a mouse model, suggesting that these genes are likely involved in the transport of an unidentified parasite ligand that allows binding of iRBCs to vascular endothelium, similar to P. falciparum SBP1 and MAHRP1 (Cooke et al., 2006, Maier et al., 2007, Spycher et al., 2008). Given that our proteomic studies identified a solid interaction between SBP1 and PfEMP1, our approach could be used to identify such unidentified parasite ligands of rodent malaria.

Our analyses led us to identify three parasite proteins (PF10_0018, STEVOR, and TryThrA) associated with the trafficking complex, as well as several host-parasite protein interactions (STOM, KPNB1, and MPP1). PF10_0018 was identified as a highly ranked SBP1 interactor, and its homolog PF10_0020 was also identified with a high probability-based score in our proteomic analyses (see Tables S2 and S3). These two genes encode proteins that belong to the α/β hydrolase family, have the highest sequence similarity in our study (Figure S8), and were recently shown to be efficiently exported (Spillman et al., 2016). However, their roles in the host cytoplasm remain completely unknown. Given the predicted lysophospholipase activity of PF10_0018 and PF10_0020, and that a variety of pathogens exploit host lipids to modify their membrane compositions (van der Meer-Janssen et al., 2010), these proteins might play a role in the lipid metabolism required for the generation of Maurer's clefts.

A tryptophan-threonine-rich antigen (termed by TryThrA) was also identified as a parasite protein associated with SBP1 in the trafficking complex. TryThrA was previously characterized as a protein expressed on merozoite surface and involved in parasite invasion, which is supported by the inhibitory effect of synthetic peptides of TryThrA antigen on merozoite invasion of erythrocytes (Curtidor et al., 2006). In the current study, we found that TryThrA was expressed across the asexual cycle and localized in Maurer's clefts (See Figure S9). Moreover, we demonstrated that its gene could be genetically disrupted without affecting parasite invasion (Figure S7B), which is inconsistent with previous studies (Alam et al., 2015, Curtidor et al., 2006). This discrepancy could be explained by off-target effects of the synthetic peptides used in the previous studies, or by alternative expression of molecules that could compensate for the loss of TryThrA in our knockout parasites. Further studies are warranted to elucidate the precise function of TryThrA in infected erythrocytes.

Host-parasite protein interactions play an essential role in malaria progression and pathogenesis (Egan et al., 2015, Miller et al., 2002, Olszewski et al., 2009). Here, we identified several host-parasite protein interactions in the host cytoplasm. Of three host factors identified (STOM, KPNB1, and MPP1), we found that MPP1, membrane palmitoylated protein 1 (also termed p55, 55 kDa erythrocyte membrane protein), was recruited into the Maurer's clefts (Figure 3C). MPP1 is a member of the membrane-associated guanylate kinase (MAGUK) family and plays essential roles in membrane organization of erythroid cells, composition of lipid rafts on erythrocyte membranes, and erythrocytopoiesis (Biernatowska et al., 2017, Egan et al., 2015, Lach et al., 2012, Quinn et al., 2009). Moreover, a previous proteomic study of microvesicles, which are secreted from the surface of P. falciparum-infected erythrocytes and likely bud from Maurer's clefts (Mantel et al., 2013), identified the presence of this protein. These results suggest that MPP1 might contribute to the organization of the membranous structures of Maurer's clefts.

Our series of knockout experiments followed by cytoadherence assays on potential SBP1 interactors identified MAL8P1.4 as being involved in the cytoadherence of iRBCs to vascular endothelial receptors (Figure 5B). MAL8P1.4 is a member of the Plasmodium helical interspersed subtelomeric (PHIST) family of exported proteins, which play diverse roles in parasite-infected erythrocytes (Kumar et al., 2018, Oberli et al., 2014, Oberli et al., 2016, Proellocks et al., 2014). Although the function of PHIST genes remains to be fully elucidated, previous studies have revealed that specific PHIST proteins can bind to the acidic C-terminal (ATS) domain of PfEMP1 and that depletion of genes that encode PHIST proteins results in decreased cytoadherence (Oberli et al., 2016, Proellocks et al., 2014). Moreover, the binding capacity of a PHIST protein differs for each PfEMP1 depending on the sequence of its ATS domain (Kumar et al., 2018, Oberli et al., 2016), suggesting that PHIST genes might have coevolved with specific ATS domains to create interaction pairs with maximum binding strength to transport a specific PfEMP1 to the erythrocyte membrane (Kumar et al., 2018, Oberli et al., 2014, Proellocks et al., 2014). In contrast, it has also been reported that MAL8P1.4 does not localize to the surface of erythrocytes or a knob and has low binding affinity for an ATS domain of a specific PfEMP1 (Oberli et al., 2014). Although there are many unknowns regarding the function of MAL8P1.4, given that multiple PHIST proteins are cooperatively and selectively involved in the transport of a specific PfEMP1 to the erythrocyte surface (Oberli et al., 2016), the altered cytoadherence by ΔMAL8P1.4 parasites may arise from the alternation of PfEMP1 being transported and presented on the erythrocyte surface. Further investigations of the relationships between cognate genes, in addition to a genetic complementation study of the MAL8P1.4 gene, would reveal the mechanistic details of the cytoadherence required for malaria virulence.

In summary, we present a comprehensive map of SBP1 interactomes to better understand the intraerythrocytic trafficking complex and interactions between host and parasite proteins in the host cytoplasm. We used this information to identify exported proteins required for cytoadherence of P. falciparum-infected erythrocytes and found that disruption of MAL8P1.4 resulted in altered cytoadherence of iRBCs to endothelial receptors with no appreciable effects on parasite viability or host cell remodeling. These findings demonstrate that our original approach using unbiased comparative shotgun proteomics provides significant insights into the complicated interplay between host and parasite proteins in the trafficking complex and creates pathways to study the molecular basis of the virulence of P. falciparum-infected erythrocytes.

Limitation of the Study

The present study led us to identify multiple host- and parasite-derived proteins that interact with SBP1, but it did not lead to the detailed function elucidation of the identified factors. Detailed analyses of how MPP1 or MAL8P1.4 is involved in the formation and configuration of Maurer's clefts, as well as cell adhesion, are needed. Moreover, to determine whether our results are universally applicable, further studies using different P. falciparum strains should be done.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.