A member of the tryptophan-rich protein family is required for efficient sequestration of Plasmodium berghei schizonts

Protein export and host membrane remodeling are crucial for multiple Plasmodium species to establish a niche in infected hosts. To better understand the contribution of these processes to successful parasite infection in vivo, we sought to find and characterize protein components of the intraerythrocytic Plasmodium berghei-induced membrane structures (IBIS) that form in the cytoplasm of infected erythrocytes. We identified proteins that immunoprecipitate with IBIS1, a signature member of the IBIS in P. berghei-infected erythrocytes. In parallel, we also report our data describing proteins that co-precipitate with the PTEX (Plasmodium translocon of exported proteins) component EXP2. To validate our findings, we examined the location of three candidate IBIS1-interactors that are conserved across multiple Plasmodium species, and we found they localized to IBIS in infected red blood cells and two further colocalized with IBIS1 in the liver-stage parasitophorous vacuole membrane. Successful gene deletion revealed that these two tryptophan-rich domain-containing proteins, termed here IPIS2 and IPIS3 (for intraerythrocytic Plasmodium-induced membrane structures), are required for efficient blood-stage growth. Erythrocytes infected with IPIS2-deficient schizonts in particular fail to bind CD36 as efficiently as wild-type P. berghei-infected cells and therefore fail to effectively sequester out of the circulating blood. Our findings support the idea that intra-erythrocytic membrane compartments are required across species for alterations of the host erythrocyte that facilitate interactions of infected cells with host tissues.


Introduction
Well-tuned host-parasite interactions allow Plasmodium parasites to be effective pathogens, which results in more than half of the world's population to be at risk of contracting malaria caused by one of the Plasmodium species able to infect humans.
Plasmodium parasites successfully infect two cell types in their vertebrate hosts: the initial liver stage of infection is followed by the infection of blood cells, the stage causing disease. Malaria parasites strategically remodel the red blood cells in which they reside. These adaptations create an intracellular environment suitable for parasite development and allow the infected cells to deftly navigate the host organism. Host alterations are carried out by proteins delivered from the parasite into the host cell. These exported Plasmodium proteins are delivered across the parasitophorous vacuole membrane into the host cell via the parasite-encoded PTEX protein export apparatus [1][2][3]. The effects of protein export have been primarily described in Plasmodium falciparum where delivery and presentation of the P. falciparum-specific adhesion protein PfEMP1 on the surface of infected red blood cells is a fundamental consequence of protein export. Protein export and host remodeling are not, however, exclusive to this Plasmodium species. The PTEX components are conserved across species [1], and infection of red blood cells with any described Plasmodium species leads to alterations of the host cell [4][5][6][7][8]. These alterations can include the transport of parasite proteins to the erythrocyte surface in addition to the morphologically evident membrane alterations [9][10][11][12][13].
The establishment of membrane compartments in the red blood cell cytoplasm of infected red blood cells are found across Plasmodium species. Although these structures have various morphologies and different names including Maurer's clefts in P. falciparum-infected cells [7], Sinton Mulligan stipplings/clefts in P. knowlesi-infected cells [14], Schüffner's dots or caveolae-vesicle complexes in P. vivax-infected cells [6], and intra-erythrocytic P. berghei-induced structures (IBIS) in P. berghei-infected cells [15,16], the conservation of a few "cleft"-localized proteins suggests that at least some of the functions of these membrane compartments are conserved across species [17][18][19]. When genetic studies have targeted P. falciparum proteins that localize to Maurer's clefts, the gene deletions affect presentation of PfEMP1 on the surface of the red blood cells, knob morphology and/or Maurer's cleft morphology [20][21][22][23][24][25][26][27], which are all unique to P. falciparum-infected cells.
To learn more about the host cell-localized membrane modifications in other Plasmodium species we used the rodent malaria parasite, P. berghei as a model. We aimed to investigate the contribution of these conserved membrane features in a system in which we can investigate the consequences of targeting exported proteins on parasite success in vivo. To identify additional protein components that localize to these membrane compartments in Plasmodiuminfected cells, we immunoprecipitated an IBIS signature protein, IBIS1, from P. bergheiinfected red blood cells. We focused on two proteins that co-precipitated with IBIS1, PBANKA_0524300 and PBANKA_0623100, which we found localize to IBIS1-positive membranes in the liver and blood infection stages. We termed these proteins intra-erythrocytic Plasmodium-induced structure proteins 2 and 3 (IPIS2 and IPIS3, respectively) and found that while they are both required for robust parasite propagation in the blood, loss of IPIS2 specifically prevents efficient sequestration of P. berghei schizonts from the blood circulation.

Identification of Plasmodium proteins co-precipitating with IBIS1-mCherry and EXP2-mCherry
To identify additional protein components of the IBIS membrane structures, we immunoprecipitated IBIS1-mCherry from infected erythrocytes. As controls, we also immunoprecipitated mCherry from erythrocytes infected with other P. berghei lines expressing mCherry or mCherry fusion proteins ( Fig 1A). In these control samples, mCherry was localized in the cytoplasm of the parasite (labeled in Fig 1A as mCherry) [28], the cytoplasm of the host erythrocyte (PEXEL-mCherry) [29], or in the parasitophorous vacuole membrane (EXP2-mCherry) [30]. In the EXP2-mCherry line, full length EXP2 is tagged with mCherry and expressed from the endogenous promoter. Because EXP2 is a well-characterized Plasmodium protein that is a component of the Plasmodium translocon of exported proteins (PTEX), immunoprecipitation of EXP2-mCherry further served as a positive control for the pulldown assay since it is known to interact with other PTEX components [31,32].
The mCherry-tagged proteins were immunoprecipitated with anti-RFP nanobody-coupled resin and precipitating proteins were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The results were analyzed based on the label-free quantitation (LFQ) intensities of the detected peptides (full results in S1 Table). Differences in starting material quantities affected the overlap of peptides detected in all three independent experimental replicates. We therefore focused on Plasmodium proteins detected in the IBIS1-mCherry or EXP2-mCherry pulldown in at least two of the three independent experimental replicates, and analyzed enrichment using the mean LFQ intensities across all three replicates ( Fig 1B).
Several of the established protein components of the PTEX complex [1], including HSP101, PTEX150 and PTEX88 co-precipitated specifically with EXP2-mCherry ( Fig 1C). Several additional proteins that have previously co-purified with PTEX proteins [33,34] were also identified in our analysis as being specifically enriched in the EXP2-mCherry sample. Furthermore, a protein previously identified in the intra-erythrocytic IBIS of P. berghei-infected cells, PbMAHRP1 [17], co-precipitated specifically with IBIS1-mCherry. Together, these results confirm the specificity of the pulldowns and validity of the analyses.
The tryptophan-rich proteins IPIS2 and IPIS3 colocalize with IBIS1 in both liver and blood stages of infection PBANKA_0623100 and PBANKA_1400700, were selected for further analysis (Fig 2A). Two of the three proteins do not have predicted secretion or export signals, but do have a tryptophan-threonine rich domain (PF12319) that is conserved in several proteins across Plasmodium species. To visualize the expression and localization of these proteins, they were each Cterminally tagged with the fluorescent protein mCherry via integration of the tag at the endogenous gene locus (S1A and S1B Fig).
All three proteins were found to be exported to the erythrocyte cytoplasm (S1C Fig). Consistent with previous reports [11], PBANKA_0623100-mCherry localizes to punctate structures in the cytoplasm of infected red blood cells (S1C Fig). We found PBAN-KA_0524300-mCherry exhibits a similar punctate localization in the erythrocyte cytoplasm. A previous study reported exported GFP-tagged PBANKA_1400700 evenly distributed throughout the erythrocyte cytoplasm [35]. In contrast, we detected the exported fusion protein in a punctate pattern (S1C Fig). In both this study and the previous study, much of the fluorescently tagged PBANKA_1400700 remained associated with the parasites. While all three of the tagged candidate proteins could be seen localizing to the parasite in some infected cells, this localization was most pronounced for PBANKA_1400700-mCherry.
All three proteins were also detected in the liver infection stage (S2A Fig). PBAN-KA_0524300-mCherry and PBANKA_0623100-mCherry were first detected 24 hours postinfection in the developing liver-stage forms and surround the parasite in a pattern indicative of the liver-stage parasitophorous vacuole and associated tubovesicular network. PBAN-KA_1400700-mCherry was detected throughout the liver stage, but it was not secreted from the parasite despite the presence of a signal peptide.
To determine if the localization of these three proteins overlaps with IBIS1, the parasite lines in which these proteins are tagged with mCherry were crossed with a line in which IBIS1 is tagged with GFP via co-infections of mosquitoes. A proportion of the parasites resulting from the cross express both fluorescent proteins. Signals from PBANKA_0524300-mCherry and PBANKA_0623100-mCherry overlapped considerably with the IBIS1-GFP signal surrounding the liver-stage parasites (Fig 2B). This colocalization was confirmed with confocal microscopy (S2B Fig) and indicates that IBIS1, PBANKA_0524300 and PBANKA_0623100 are present within the same domains of the liver-stage parasitophorous vacuole and tubovesicular network. In infected red blood cells, PBANKA_0524300-mCherry and PBAN-KA_0623100-mCherry colocalize with IBIS1-GFP. Although the majority of PBANKA_1400700-mCherry remains parasite-associated, the exported mCherry signal colocalizes with IBIS1-GFP. These results indicate that all three studied proteins localize to the same membrane compartments formed in the infected erythrocyte cytoplasm (Fig 3A).

Fig 1. Immunoprecipitation of P. berghei proteins from infected red blood cells identifies proteins that coprecipitate specifically with IBIS1-mCherry and EXP2-mCherry.
A) The indicated mCherry-tagged proteins were immunoprecipitated from lysates of late stage P. berghei using RFP-specific nanobodies and co-precipitating proteins were identified by LC-MS/MS. B) P. berghei proteins that were detected in the IBIS1-mCherry sample in at least two of the experimental replicates are depicted in the heatmap, which indicate the mean normalized LFQ intensities of detected proteins of three independent experiments. Each row represents a protein detected in the IBIS1-mCherry eluates. The columns represent the immunoprecipitates prepared from the mCherry-expressing parasite lines indicated in (A) (1: IBIS1-mCherry, 2: mCherry, 3: PEXEL-mCherry, 4:EXP2-mCherry). Proteins that were enriched in the IBIS1-mCherry pulldown relative to at least two of the three controls (top 10% of the normal distribution) are annotated. Those in bold were enriched relative to all three controls. Proteins marked with a pink dot were followed up further in this study. C) P. berghei proteins that were detected in the EXP2-mCherry sample in at least two of the experimental replicates are depicted in the heatmap, which indicate the mean normalized LFQ intensities of detected proteins of three independent experiments. Each row represents a protein detected in the EXP2-mCherry eluates. The columns represent the immunoprecipitates prepared from the different mCherry-expressing parasite lines indicated in (A). Proteins marked with dark blue dots are confirmed to play roles in PTEX-mediated export, and those marked with light blue dots were previously found biochemically associated with components of the PTEX complex [33,34]. https://doi.org/10.1371/journal.ppat.1010846.g001

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts Because PBANKA_0524300 and PBANKA_0623100 clearly localize at the interface with the host cell during both the liver and blood stages of the infection and their expression and localization patterns so closely mimic those of IBIS1, we chose to further investigate the two tryptophan-rich domain-containing proteins. Because the current naming systems for genes containing this tryptophan-rich domain are inconsistent across species and none are ideal for indicating orthologs across species, we propose naming these two proteins based on their localization to intra-erythrocytic Plasmodium induced structures (IPIS). We therefore refer to PBANKA_0524300 as IPIS2 and PBANKA_0623100 as IPIS3.

Localization of putative P. vivax orthologs in P. berghei
Unlike IBIS1, which appears to be exclusive to the rodent Plasmodium species, tryptophanrich domain-containing proteins are found across the Plasmodium genus (S3 Fig). Syntenic orthologs of IPIS2 are conserved across the Plasmodium clade and are expressed in P. vivax, P. knowlesi, P. malariae and P. cynomolgi. Notably, no syntenic IPIS2 ortholog exists in the Laverania Plasmodium species P. reichenowi or P. falciparum. IPIS3 has syntenic orthologs only in the rodent Plasmodium species; however, PVP01_0202200/PvTRAg2 was previously annotated

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts as an ortholog on PlasmoDB because of sequence similarity. In order to determine if the putative orthologs from P. vivax also targeted the membrane structures in the cytoplasm of Plasmodium-infected red blood cells, we generated P. berghei lines that expressed HA-tagged PvTRAg8, the syntenic P. vivax ortholog of IPIS2, or HA-tagged PvTRAg2, which shares similarity to IPIS3. These genes were integrated into the P. berghei genome such that they are expressed under the control of the IPIS2 or IPIS3 promoters, respectively. Both PvTRAg8-HA and PvTRAg2-HA are exported efficiently by P. berghei and localize to discrete structures in the cytoplasm of infected erythrocytes ( Fig 3B).

Absence of IPIS2 or IPIS3 confers a fitness disadvantage during Plasmodium growth in the blood
In order to determine if IPIS2 or IPIS3 contribute to successful growth of the parasite, each gene was deleted and replaced by selection markers via double homologous recombination (S4 Fig). The successful gene deletion and selection of isogenic knockout lines indicated that neither IPIS2 nor IPIS3 are essential for blood-stage growth. Infection of Anopheles stephensi with the knockout lines and resulting development of salivary gland sporozoites revealed that these genes are also dispensable for Plasmodium development in mosquitoes (S5A and S5B Fig).
To assess the ability of the knockout lines to establish infection in mammals, C57BL/6 mice were infected with IPIS2-or IPIS3-deficient sporozoites. The wild-type and ipis2-P. berghei lines were detected in the blood between day three and four after infection with 1,000 sporozoites. In contrast, the ipis3-parasite line became visible in the bloodstream as late as five days after infection, and in two of the ten infected mice, IPIS3-deficient parasites were never detected in the blood (Fig 4A). This delay in patency prompted us to examine growth of the parasite lines in the livers of infected animals. Quantification of the parasite burden in the liver 42 hours after infection with sporozoites revealed no significant differences in growth of the ipis2-or ipis3-lines in the liver in comparison to wild-type P. berghei (Fig 4B). In parallel we assessed the liver-stage development of the knockout lines in cultured hepatoma cells in vitro. The size of the exo-erythrocytic forms in HepG2 cells were determined 48 hours after infection with sporozoites from each line, and no restriction of parasite growth was detected in the absence of IPIS2 or IPIS3 in vitro (S5C Fig), which is consistent with the in vivo data.
Although the time to patency was statistically similar in mice infected with either wild-type or ipis2-sporozoites, the parasitemia of IPIS2-deficient parasites in the blood was on average lower than wild-type P. berghei following sporozoite infection (S5D Fig). The delay in patency observed upon infection with ipis3-sporozoites was accompanied by a corresponding decrease in parasitemia during the blood stages (S5D Fig). That P. berghei suffers growth consequences in the absence of IPIS2 or IPIS3 is underscored by the observation that fewer mice infected with ipis2-or ipis3-sporozoites succumb to infection within the first 12 days of infection A) The appearance of ipis3-parasites in the blood following sporozoite injection is delayed in comparison to wild-type. C57BL/6 mice were infected with 1,000 sporozoites of either ipis2-, ipis3-or wild-type parasites, and blood-stage infections were monitored starting from day 3 post-infection with Giemsa-stained blood smears. Statistics: Log-rank (Mantel-cox) test; n = 5 (ipis2-), n = 10 (ipis3-). B) The growth of ipis2-and ipis3-parasites in the liver does not differ significantly from wild-type P. berghei. The relative parasite burden in the liver was measured by qPCR of RNA extracted from livers forty-two hours after intravenous injection of either ipis2-, ipis3-, or wild-type sporozoites. Levels of Pb18s rRNA were normalized to levels of mouse GAPDH RNA in each infected liver. Statistics: unpaired two-tailed t-test; n = 5; ns: not significant. C) The ipis2-and ipis3-lines do not grow as efficiently as wild type in the blood of co-infected animals. Mice were infected with equal amounts of wild-type P. bergheiinfected erythrocytes and ipis2-[mCherry; PyrS] or ipis3-[mCherry; PyrS] infected erythrocytes. Parasitemia was measured by flow cytometry between days 4-7 after infection. � p <0.05, �� p < 0.01, ��� p < 0.005 determined by unpaired two-tailed t-test; n = 5. D) The absence of one IPIS protein does not dramatically influence the expression level of the other. Erythrocyte-associated mCherry signal was quantified in fixed unstained infected cells. The mCherry signal intensity was plotted as a ratio to parasite size to account for differences in expression in different stages. In the ipis2-line, a statistically significant increase in mCherry signal was detected; however the extent was negligible. � p <0.05, unpaired two-tailed t-test. https://doi.org/10.1371/journal.ppat.1010846.g004

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts (S5E Fig) in comparison to mice infected with wild-type P. berghei, and this reduction in pathogenicity is significant for the ipis3-line.
Competitive growth analyses, in which mice are infected with both wild-type and knockout lines, have successfully identified growth phenotypes in the blood of infected mice [36, 37]. We selected for knockout parasites expressing single fluorescent markers as described [38], since the expression of distinct genetically-encoded fluorescent proteins in the wild-type and knockout lines allows their differentiation by flow cytometry. To assess growth of the knockout lines specifically in the blood stage of infection, mice were infected with equal numbers of red blood cells infected with wild-type and either ipis2-or ipis3-parasites. (Fig 4C). Although the same number of parasites were injected intravenously, fewer IPIS2-deficient parasites were detected in the blood of infected animals compared to wild-type parasites at each time point from four to seven days post-infection. While similar numbers of ipis3-and wild-type parasites were detected initially at day four, the wild-type line out-competed IPIS3-deficient parasites as the infection progressed. These results suggest that both knockout lines grow less efficiently in the blood than wild-type parasites.
The localization of IPIS2 and IPIS3 to the intraerythrocytic membrane structures and their requirement for efficient blood-stage growth prompted us to investigate whether the membrane structures are still formed in the cytoplasm of erythrocytes infected with the ipis2-and ipis3knockout lines. In order to visualize these structures in the knockout lines, we tagged IPIS2 with the fluorescent protein mCherry in the IPIS3-deficient line, and likewise, tagged IPIS3 with mCherry in the line lacking IPIS2 (S6 Fig). The IPIS proteins still appear in punctate structures distributed across the erythrocyte cytoplasm in the absence of either IPIS2 or IPIS3 (S7A Fig). Because IPIS2 and IPIS3 share expression and localization patterns in addition to the tryptophan-rich domain, we hypothesized that upregulation of one may compensate for the absence of the other. Because the fluorescent tags are on the natively expressed proteins, we were also able to use these lines to compare expression of IPIS2 and IPIS3 in the knockout lines based on the fluorescence intensity of mCherry. To account for the increase in expression of both IPIS2 and IPIS3 over the course of normal parasite development in red blood cells (S7B Fig), we normalized the fluorescence intensity of the IPIS2-mCherry or IPIS3-mCherry signal to parasite size in mixed blood stages. We detected no notable difference in the expression of IPIS2-m-Cherry or IPIS3-mCherry in the ipis3-or ipis2-lines, respectively, in comparison to the expression of the mCherry-tagged proteins in a wild-type genetic background ( Fig 4D). These results indicate that IPIS2 and IPIS3 are not upregulated in the knockout lines.

Schizonts are detected in the circulation in the absence of IPIS2 or IPIS3, but only IPIS2 is required for adherence of schizonts to CD36
Schizonts are only rarely detected in peripheral blood of P. berghei-infected mice due to their sequestration in lung and adipose tissue [39]. In animals infected with the IPIS2-or IPIS3-deficient lines, we detected schizonts in the peripheral blood during blood-stage growth more often than in animals infected with wild-type P. berghei lines (S5F Fig). Circulating schizonts can be recognized and cleared by the spleen, and infection of mice with P. berghei lines that fail to sequester out of circulation has been correlated with an increase in the weight of spleens of infected animals [17,[39][40][41]. In comparison to mice infected with wild-type P. berghei, the spleen-to-body weight ratio was on average 29% higher in mice infected with the ipis2-line and 19% higher in mice infected with the ipis3-line (Fig 5A). This data further indicated that sequestration may be compromised in red blood cells infected with the knockout parasites.
In order to compare circulating schizonts directly and quantitatively across multiple lines, synchronized infections in mice were required. To this end, purified schizonts were injected

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts intravenously into naïve animals, and the relative proportions of each life cycle stage present in blood smears taken between 18-24 hours after infection were quantified (Fig 5B). The injection of purified late-stage parasites resulted in a maximum of fifteen-fold more schizonts in the peripheral blood compared to wild type in the absence of IPIS2, and five-fold more schizonts in the absence of IPIS3. The parasitemia in animals 18-20 hours after injection of schizonts was similar across parasite lines (S5G Fig). The parasites used in this assay express GFP-Luciferase. Twenty-three hours after infection, the time when we saw the maximal number of schizonts, animals were sacrificed and the luciferase activity in the lung and adipose tissue was assessed. The luciferase activity in these tissues was lower in animals infected with the ipis2-line in comparison with wild-type-infected animals ( Fig 5C). This result is consistent with the hypothesis that the wild-type schizonts sequester in these tissues more efficiently than the IPIS2-deficient line. In contrast, no significant difference in tissue-associated luciferase activity was seen with the ipis3-infected animals compared to wild-type parasites. Organs from these animals infected with IPIS3-deficient parasites had a wide range of luciferase activity, but the data suggest that there is no reduction in tissue-associated parasites caused by the absence of IPIS3. Values are normalized to organs from wild-type-infected mice. �� p < 0.01, ���� p < 0.0001, ns: not significant, unpaired two-tailed t-test. D) Red blood cells infected with IPIS2-deficient P. berghei do not bind efficiently to CD36-expressing CHO cells. Schizonts were incubated with CHO cells and the number of infected-red blood cells bound per cell was quantified. The background binding to control CHO cells was subtracted, and the values were normalized to wild-type-infected red blood cell binding. The depicted data comprises three experimental replicates with three technical replicates for each parasite line and CHO cell line. �� p < 0.01, ns: not significant, unpaired two-tailed t-test. https://doi.org/10.1371/journal.ppat.1010846.g005

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts P. berghei schizont sequestration in mouse tissues has been shown to be mediated by binding to CD36 in the host endothelium [39]. To examine the ability of erythrocytes infected with P. berghei schizonts lacking IPIS2 or IPIS3 to bind CD36, blood infected with late blood stage parasites was incubated with CHO cells modified to express CD36 on the cell surface [42] ( Fig  5D). Erythrocytes infected with wild-type P. berghei bound to the CD36-expressing cells on average five times better than to control CHO cells that do not express CD36. The adherence of erythrocytes containing IPIS2-deficient parasites to CD36 was reduced by about 60% compared to wild-type-infected cells. In contrast, there was no difference in binding of cells infected with the ipis3-line in comparison to wild-type P. berghei, indicating that erythrocytes infected with IPIS3-deficient parasites remain capable of adhering to CD36. Taken together, these data demonstrate that the absence of these exported IPIS proteins influences the blood infection stage. The absence of IPIS2 in particular influences the interaction between P. berghei schizonts and the host tissue endothelium, making IPIS2 required for efficient sequestration of schizonts from the blood circulation.

Discussion
The formation of intracellular membrane structures containing parasite proteins is a conserved feature within the Plasmodium genus and occurs in all intracellular stages. These membranes demarcate regions where the parasite can interact with the environment of its host, and the functional activities occurring at these membranes will depend on their protein and lipid compositions.
The collection of proteins known to localize to the Maurer's cleft-like structures in cells infected by Plasmodium species other than P. falciparum is limited, and this study has expanded this protein repertoire. By immunoprecipitating a P. berghei protein (IBIS1) present in host-localized membranes across intracellular stages, we identified additional proteins present in the IBIS in the cytoplasm of P. berghei-infected erythrocytes. Previous studies have localized P. berghei proteins to punctate structures in the infected-red blood cell cytoplasm [11, 15-17, 35, 43], and many of these were identified based on similarity to known exported P. falciparum proteins or on the presence of a PEXEL/HT (Plasmodium export element / host targeting) motif. Although neither IPIS2 nor IPIS3 contain a predicted PEXEL/HT motif [44], they are exported from the parasite into the host erythrocyte where they localize to IBIS. We cannot determine from this data whether PBANKA_1400700, IPIS2 and IPIS3 interact with each other and/or IBIS1 directly or indirectly as part of one or more protein complexes. However, colocalization of the candidate proteins with IBIS1 suggests the IBIS1 pulldown successfully identified proteins present in the membrane structures formed in the host erythrocyte cytoplasm.
Previous knock-outs of IPIS2 showed no severe effect on blood stage growth [40], however large scale screening suggested there indeed was a slower growth phenotype associated with the knockout of IPIS2 and of IPIS3 [45]. We found the absence of IPIS2 influences the ability of P. berghei schizonts to sequester from the blood circulation. In support of previous findings [17,40,41,46], efficient CD36-dependent cytoadherence supports robust in vivo growth, and we found that the failure of ipis2-schizonts to sequester from the circulation correlates with a fitness disadvantage during the blood infection stage.
The conclusion that IPIS2 is required for efficient sequestration of late-stage P. berghei from the circulation is supported by the detection of ipis2-schizonts in the peripheral blood and their relative absence from the tissues. At the end of one developmental cycle, wild-type schizonts are mostly absent from circulation and can instead be found in the lung and adipose tissue. Although ipis3-schizonts are found in the peripheral blood at this time, ipis3-parasite

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts material can also be detected in the tissues of perfused animals. In contrast, the tissue-associated ipis2-parasite material is notably reduced, and this correlates with higher numbers of circulating ipis2-schizonts. Adherence assays using CD36-expressing cells demonstrated that erythrocytes infected specifically with IPIS2-deficient parasites fail to bind CD36 as efficiently as cells infected with wild-type P. berghei. The fitness cost of not sequestering is due to increased splenic clearance of infected cells [40], and we see a corresponding increase in spleen weight in animals infected with the ipis2-knockout line.
Deletion of the IPIS3 gene also affected parasite fitness, but the phenotype appears to be distinct from that caused by the loss of IPIS2. The ipis3-line grows poorly in the blood in comparison to wild-type P. berghei; however, IPIS3-deficient schizonts are competent in CD36 binding and are tissue-associated in synchronized infections. The minor growth phenotype could therefore be due to a number of possible causes. Although ipis3-schizonts can still associate with tissues, a proportion of them can still be detected in the circulation. Synchronized growth suggested no major defect in the transition of ipis3-parasites from rings to schizonts, but protracted development in the late stages or defects in parasite exit from the red blood cell would increase the time ipis3-parasites spend as schizonts and perhaps consequently the likelihood that schizonts can be detected in circulation. These circulating schizonts may be sufficient to cause the increase in spleen weight that is observed in ipis3--infected mice in comparison to wild-type infection. Alterations in parasite deformability or differences in how the ipis3-parasites trigger immune responses and inflammation may also be responsible for these observations.
The loss of IPIS3 also affects the transition between the liver and blood growth stages. Although ipis3-growth in the liver appeared to be robust when parasite burden was measured following sporozoite infection, we saw a delay in the time to patent blood-stage infection, and ipis3-parasites failed to establish blood-stage infection in a small proportion of infected mice. Comparisons to the ipis2-line indicate that the defect in ipis3-blood stage growth is not sufficient to cause this effect on patency. An intriguing possibility is that a defect in merozoite development or parasite egress in the liver and blood could explain the phenotypes of the ipis3-line, but these two phenotypes (time to patency and blood-stage growth) may also have alternative and independent causes.
Infected erythrocytes circulating freely in the blood represent only a proportion of the parasites in Plasmodium-infected vertebrates [47][48][49][50][51]. Although cytoadherence to the host vasculature and sequestration out of the blood circulation are most prominent in P. falciparum infections, these traits are not exclusive to this species. While some tissue-associated parasites are extravascular [51,52], there is also evidence of intravascular sequestration of red blood cells infected with various Plasmodium species: Erythrocytes infected with P. berghei schizonts sequester in the tissue vasculature in a manner dependent on host CD36 [39]. Although latestage P. knowlesi or P. vivax can be detected in the blood, erythrocytes infected with P. knowlesi can accumulate in the brain vasculature [53] and can bind to ICAM-1 or VCAM [54]. A proportion of late-stage P. vivax are sequestered out of circulation [47, 48,51], and P. vivaxinfected erythrocytes have been shown to bind to endothelial cells [48,[55][56][57]. P. vivax members of the PIR protein family appear to be involved in this binding [10,55,58]; however evidence that PIR proteins play a role in cytoadherence of other Plasmodium species where they are expressed is lacking [59,60].
The molecules mediating the binding of P. berghei-infected erythrocytes to host CD36 are thus far unknown. Although the IPIS2 knockout line fails to bind to CD36 and sequester efficiently, the intracellular localization of IPIS2 precludes its direct involvement in cytoadherence. We therefore infer that there is an alteration in parasite-induced modification of the host cell surface in cells infected with the ipis2-line. The IBIS-localized proteins PbSBP1 and PbMAHRP1 have also been shown to be essential for efficient sequestration of P. berghei schizonts from the blood circulation [17]. Expression of IPIS3-mCherry in the ipis2-line confirmed that absence of IPIS2 causes no broad defect in parasite protein export or protein trafficking to IBIS, but failure of PbMAHRP1 or PbSBP1 to localize to IBIS in the ipis2-line would be expected to result in loss of schizont sequestration.
The requirement of PbSBP1 and PbMAHRP1 for P. berghei sequestration indicated that the IBIS formed in P. berghei-infected cells are functionally related to the Maurer's clefts of P. falciparum-infected cells [17]. Although it is not yet known if the P. knowlesi ortholog of SBP1 (PkSBP1) contributes to sequestration, PkSBP1 has been found to localize to the Sinton and Mulligans Stippling-an encouraging suggestion for conserved protein roles in these membrane compartments [18]. While SBP1 and MAHRP1 were first described in P. falciparum [61,62], IPIS2 and IPIS3 do not have definitive orthologs in P. falciparum. The gene encoding IPIS2 is, however, conserved and syntenic across the primate-and human-infecting Plasmodium clade, such as P. vivax and P. knowlesi. Our data consequently support the idea that membrane structures formed in the infected erythrocyte cytoplasm play conserved roles in the host cell remodeling that facilitate the interactions between the infected cell and the host tissues.
Both IPIS2 and IPIS3 possess a tryptophan-threonine-rich domain (PFAM: PF12319) that is found in multiple Plasmodium proteins. Proteins with this domain are sometimes referred to as TRAg, PyAg or Pv-fam-a. Importantly however, in P. berghei Fam-A refers to an unrelated family of proteins, which contain a predicted START domain. The tryptophan-threonine-rich domain is characterized by positionally conserved tryptophan residues, and proteins containing this domain can be immunogenic and are implicated in a range of processes including adherence to and invasion of erythrocytes [63][64][65][66]. Our study shows for the first time that a protein with this tryptophan-rich domain is involved in remodeling of the host erythrocyte and supports the idea that this domain does not necessarily define the function of these proteins. The family of proteins containing this domain is expanded in the Plasmodium clade of Plasmodium species, which contains all human-infecting species apart from P. falciparum. For example, P. vivax encodes forty Trp-Thr-rich domain-containing proteins and P. knowlesi has twenty-nine. In contrast, P. falciparum encodes only three proteins with this domain. Various expression and localization patterns have been described for Plasmodium proteins possessing this domain [65,67]. A P. falciparum member of this family, TryThrA, was recently shown to immunoprecipitate with PfSBP1 and localize to Maurer's clefts [68]. However, genetic targeting of TryThrA did not influence the ability of P. falciparum-infected erythrocytes to cytoadhere to CD36. In this study, we demonstrate that putative P. vivax orthologs of IPIS2 and IPIS3 appear to engage the protein export machinery when expressed in P. berghei and, similar to PbIPIS2 and PbIPIS3, localize to punctate structures in the cytoplasm of P. berghei-infected erythrocytes. When the P. berghei protein IBIS1 was expressed in P. falciparum, it was found to localize to the Maurer's clefts [69], which further supports the cross-species export recognition we also demonstrate here with the expression of P. vivax proteins in P. berghei. While the results of trans-species expression should be interpreted with caution, these data may suggest that these orthologs may have a similar target in the cytoplasm of P. vivaxinfected red blood cells.
In addition to colocalizing with IBIS1 in infected erythrocytes, IPIS2 and IPIS3 are also expressed in the liver stage, and like IBIS1, they surround the liver-stage parasites. Such dual expression and differing localizations in these two stages have also been demonstrated for multiple other exported Plasmodium proteins [59,70]. P. berghei members of the PIR, Fam-A and Fam-B protein families have been shown to be expressed in both stages, exported into the host erythrocyte and retained in the liver-stage parasitophorous vacuole [59]. It should be noted

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts that in these examples, proteins were tagged with the fluorescent protein mCherry, as was done here with IPIS2 and IPIS3. While fluorescent protein tags do not hinder protein export across the parasitophorous vacuole in blood stages, we cannot exclude the possibility that this tag may interfere with protein translocation in the liver stage of infection. Although the localization differs in the two infection stages, IPIS2 and IPIS3 are at the boundary with the host cell in both, which is noteworthy considering the physiological differences between erythrocytes and hepatocytes.
In parallel to our IBIS1-mCherry pulldown, we immunoprecipitated mCherry-tagged EXP2 from P. berghei-infected erythrocytes. We not only pulled out previously characterized components of the PTEX apparatus, which validated our experimental approach and analysis, but we also identified and highlighted some additional candidate proteins that have been previously immunoprecipitated with HA-tagged PTEX components from P. falciparum-infected cells [33,34]. These proteins, marked in light blue in Fig 1, were pulled down previously with either PTEX150 or PTEX88 [33, 34] and have not yet been investigated, but their association with different PTEX components in multiple Plasmodium species makes them strong candidates for having functional association with the PTEX complex in infected cells. Nevertheless, these and the other novel proteins we found associating specifically with EXP2 need further characterization before any role in protein export can be established. Likewise, our validation of IPIS2 and IPIS3 as resident proteins of the IBIS in infected erythrocytes suggests that the additional proteins we detected specifically in the IBIS1-mCherry immunoprecipitates are potential protein components of the cytoplasmic membrane structures of Plasmodiuminfected erythrocytes.
Multiple Plasmodium species have cytoadhesive potential, and parasite sequestration has broad consequences for disease progression and parasite persistence [71][72][73][74]. Studying both the conserved and the species-specific features that mediate host cell remodeling can improve our understanding of how Plasmodium species promote favorable interactions within their hosts' tissues. Our studies show IPIS2 and IPIS3 localize to erythrocyte-localized membrane structures in P. berghei-infected cells and suggest that orthologs of these proteins may localize to such membranes across several of the Plasmodium species in which protein export and host remodeling are less well characterized. Our results support the idea that the membrane structures in the cytoplasm of Plasmodium-infected cells have conserved functions across species. While variable in morphology, they are important for the remodeling that allow Plasmodiuminfected cells to navigate the tissues of their host.

Experimental animals, cells and parasite lines
Female NMRI, SWISS and C57BL/6 mice were either purchased from Charles River Laboratories (Sulzfeld, Germany) or bred in-house. NMRI or SWISS mice were used for blood-stage growth assays. Sporozoite inoculations, spleen weight measurements, and sequestration experiments were performed using C57BL/6 mice. The P. berghei reference lines used were P.

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts berghei ANKA, P. berghei GFP-Luc (RMgm-29) [75], and Plasmodium berghei Bergreen [76]. The HepG2 and Huh7 hepatoma cell lines were cultured in DMEM supplemented with fetal calf serum (FCS). Chinese hamster ovary (CHO) cells expressing GFP or CD36-GFP were a gift from the laboratory of Iris Bruchhaus (BNITM, Hamburg) [42] and were maintained in Ham's F12 supplemented with 10% FCS and 1 mg/ml G418. Anopheles stephensi mosquitoes were raised under a 14 h light/10 h dark cycle at 28˚C and 80% humidity and were fed daily on 10% sucrose.

Construction of plasmids
For the construction of the plasmids used for endogenous C-terminal mCherry tagging, regions containing the IPIS2, IPIS3 or PBANKA_1400700 genes were amplified from P. berghei genomic DNA using the respective primers (S2 Table) and integrated into the SacII and XbaI sites of the B3D+mCherry plasmid [77]. The sequences of PvTRAg8 and PvTRAg2 were codon optimized (sequences provided in S8 Fig), synthesized (GenScript) and cloned into the NotI and SpeI sites of b3D.DT^H.^D (provided by Dr. Andrew Waters, Glasgow University). The 5' and 3' untranslated regions of the IPIS2 and IPIS3 genes were cloned into the GOMO-GFP luciferase vector using the respective primers (S2 Table) containing SacII and NotI sites and the XhoI and KpnI sites, respectively, in order delete IPIS2 and IPIS3 via double homologous recombination.

Generation of transgenic P. berghei parasites
Transfection P. berghei schizonts and selection of transgenic parasites was carried out as previously described [75]. Briefly, the plasmid constructs used for genetic alteration were either linearized, or in the case of the knockouts, the vectors were digested with restriction enzymes whose cut sites flanked the disruption element. Schizonts were isolated from overnight cultures, and transfection of the schizonts was done using the Nucleofactor T-cell transfection kit in combination with an Amaxa gene pulser using the program U-33. Transfected parasites were injected into recipient mice, and on the second day, transgenic parasites were selected for using 70ng/μl, pH 3.5-5.5 pyrimethamine drinking water.
To remove the DHFR-resistance from the knockout lines and concurrently select for parasites encoding single fluorescent reporters [38], the knockout lines were put under selective pressure for five days with 1.5mg/ml 5-fluorocytosine (5-FC) to the mouse drinking water, which was administered fresh daily and protected from light. Isogenic clonal lines were generated by FACS sorting parasites that expressed only mCherry or GFP-luciferase. This process yielded the lines ipis2-[GFP-Luc; PyrS] and ipis3-[GFP-Luc; PyrS]. Genotypes were confirmed with PCR (S2 Table)

Cross fertilization of transgenic P. berghei lines
To generate parasites expressing both IBIS1-GFP and the mCherry-tagged candidate proteins, the lines were genetically crossed. Mice were co-infected with equal numbers of the mCherryexpressing P. berghei lines and the IBIS1-GFP P. berghei line and fed to Anopheles stephensi mosquitoes. Sporozoites comprising a mixed population including double-fluorescent

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts parasites were isolated from the salivary glands of mosquitoes infected with cross-fertilized lines on day 21 after transmission to the mosquitos.
Overnight ex vivo culture and purification of P. berghei schizonts P. berghei schizonts were cultured ex vivo as described [75] in DMEM culture medium supplemented with FCS and heparin for up to 22 hours at 37˚C on a shaking platform rotating at 15 rpm with a gas composition of 5% CO 2, 10% O 2 , and 85% N 2 . Schizonts were isolated from the culture through a one-step Nycodenz gradient separation using 55% Nycodenz in PBS.

Immunoprecipitation of IBIS1-mCherry
Parasites were cultured as described above for 16-20 hours and late-stage parasites were purified with a 60% Nycodenz gradient. Cells were lysed on ice for thirty minutes with periodic agitation in a buffer containing 10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5 mM EDTA and 0.5% NP40 supplemented with 1mM PMSF and a protease inhibitor cocktail (Sigma-Aldrich). Immunoprecipitations from cell lysates were performed with the RFP-Trap-A kit (Cromotek) according to manufacturer's instructions using equal numbers of parasites per line, which varied from 7x10 8 to 7x10 9 across experimental replicates. The P. berghei lines used included a line expressing IBIS1-mCherry [15], a line expressing mCherry from the HSP70 promoter [28], a line in which mCherry is fused to the predicted PEXEL motif of the circumsporozoite protein and expressed from the IBIS1 promoter [29], and a line expressing EXP2-mCherry [30].
Proteins eluted from the IP beads were digested using an automated sample-preparation workflow (Axel-Semrau Proteome Digest-O-r [78]. Briefly, the samples were reduced by 1 mM tris(2-carboxyethyl)phosphine (TCEP, Merck) and free sulfhydryl groups carbamidomethylated using 5.5 mM chloroacetamide (Sigma-Aldrich). Proteins were digested with 0.5 μg sequencing grade endopeptidase LysC (Wako) overnight at room temperature. The reaction was terminated by adding trifluoroacetic acid (TFA, Merck) to a final concentration of 1% resulting in a final pH of 2. The peptides were purified using C18 stage-tips (Empore SPE disks, 3M) [79] and measured on a Q-Exactive plus mass spectrometer (Thermo-Fisher, Germany) coupled to a nano-LC system (easy-nLC, Thermo-Fisher, Germany). An amount of 2 μg of the peptide sample were injected and separated using a 3h gradient (4 to 76% acetonitrile and 0.1% formic acid in water) at a flow rate of 0.25 μl/min on an in-house prepared nano-LC column (0.075 mm x 250 mm, 3 μm Reprosil C18, Dr. Maisch GmbH). The separated peptides were ionized on a proxeon ion source and directly sprayed into the mass spectrometer (Q-Exactive Plus, Thermo). The MS1 acquisition was performed at a resolution of 70,000 in the scan range from 300 to 1700 m/z. The top 10 intense masses were selected for MS2 analysis. MS2 scans were carried out at a resolution of 15,500 with the isolation window of 2.0 m/z. Dynamic exclusion was set to 30 s and the normalized collision energy to 26 eV. For the automatic interpretation of the recorded spectral data, the MaxQuant software package version 1.6.0.16 was used [80]. Carbamidomethylation was set as a fixed modification while oxidized methionine and acetylated lysine were set as variable modifications. An FDR of 1% was applied to peptide and protein level, and an Andromeda-based search was performed using a P. berghei Uniprot database.
The mean LFQ intensities across all three replicates were used for analyses. Enrichment in the experimental samples was determined by calculating the ratios of the intensities of all protein groups detected in the experimental samples relative to the control samples, and the binary logarithm of these ratios was formed. The standard normal distribution was calculated using the means and standard deviation for the obtained log2 ratios (S1 Table).

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts

Blood stage immunofluorescence assays
For infected erythrocytes, 1μl of blood from an infected mouse was washed, diluted in PBS, and added to concanavalin A-coated coverslips (0.3mg/ml concanavalin A). The sample was then fixed with 4% PFA + 0.0075% glutaraldehyde in PBS for 15 minutes at room temperature.

Calculation of mCherry signal in infected blood
Infected erythrocytes were fixed and mounted in DAPI-containing Fluoromount-G (South-ernBiotech). Fluorescence microscopy was done using an Axio Observer Z2 epifluorescence microscope (Zeiss) and 63x/1.4 objective. Images were taken with AxioCam MR3 camera at identical exposures to quantify the mCherry signal in the presence and absence of IPIS2 and IPIS3. Images were imported into FIJI for quantification of both the mCherry signal in the red channel and parasite size in the DIC channel [81].
The outline of the red blood cell visible in the DIC channel was used to define the area of mCherry calculation within the red channel. To account for the asynchronous infections, parasite size was used as a proxy for parasite development, and the mCherry signal intensity calculated was expressed as a ratio to parasite size, which was also determined in the DIC channel.

Infections with P. berghei sporozoites
Sporozoites were extracted from salivary glands of infected mosquitoes, and 1,000 sporozoites in RPMI were injected into the tail vein of naïve mice for assessing growth, and 10,000 sporozoites were injected for qPCR to determine the parasite burden in the infected liver. Thin Giemsa-stained blood smears were made every day starting from day 3 after injection in order to determine parasitemia by light microscopy. Animals were monitored daily and sacrificed immediately if they showed at least three behavioral or functional abnormalities or presented with ataxia, paralysis or convulsions. To assess the weight of spleens in infected animals, eight days after the injection of 1,000 sporozoites, the infected mice were weighed and sacrificed. The spleens were then dissected and weighed. To account for differences in body weight, a ratio of spleen weight to body size was calculated.
For in vitro infections, HepG2 cells were seeded and infected with isolated sporozoites. After addition of sporozoites, cells were centrifuged at 1800 x g for 5 minutes. The cells were fixed with 4% PFA for 15 minutes. The fixed cells were permeabilized with methanol and then treated with a block + quenching solution of 3% BSA + 4% FCS serum in ammonium chloride in PBS for 1 hour at RT.
Antibodies against RFP (ab62341; Abcam), GFP (ab13970; Abcam) and PbHSP70 [82] were used for immunofluorescence analyses. All prepared samples were visualized with fluorescence microscopy. Exoerythrocytic form size of the developing parasites was calculated using FIJI imaging software [81].

qPCR determination of parasite burden in infected mouse livers
The parasite burden in infected mouse livers was determined with qPCR quantification of parasite 18s rRNA. Mice were infected i.v. with 10,000 sporozoites. 42 hours post infection, the mice were sacrificed, and the livers harvested. They were washed briefly in PBS before homogenization in TRIzol. Liver homogenates were diluted 1:1 in TRIzol and stored at -80˚C. mRNA was extracted using the Direct-zol RNA MiniPrep Plus kit according to the manufacturer's

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts protocol. The extracted mRNA was diluted to 2ng/μl. cDNA was generated using RETROscript (Invitrogen) according to the manufacturer's protocol. Relative parasite burden was determined via quantitative PCR by comparing the mean Ct value of the P. berghei 18s ribosomal subunit (Gene ID: 160641) to the mean Ct value of the Mus musculus gapdh (Gene ID: 281199965) in the generated cDNA.

Flow cytometry analysis of P. berghei blood stage infections
Flow cytometric detection of blood stage parasites was performed as described [36]. Blood stage co-infections were initiated with parasites transferred from donor mice in which the parasitemia was below 1% and quantified by flow cytometry on a BD FACS Canto II. The gating strategy was as described previously [36]. The blood from the donor mice was serially diluted such that 500 parasites of both wild type and either ipis2-[mCherry; PyrS] or ipis3-[mCherry; PyrS] in 100ul of RPMI were injected per animal into naïve recipient mice. On days 4-7 following injection, parasitemia of both lines was measured by flow cytometry from a drop of tail blood diluted in Alsever's solution. Blood cells infected with parasites of both lines (the mCherry+GFP+ population) were excluded from the parasitemia calculation.

Synchronous blood-stage infections
In order to assess schizont sequestration in a synchronous infection, schizonts were isolated from an overnight ex vivo culture as described above. Each mouse was injected with approximately 5x10 7 infected red blood cells isolated from the overnight cultures. Giemsa-stained thin blood smears using tail blood were made at 2 hours post infection to ensure successful invasion of merozoites. After this, giemsa-stained smears were made at the indicated time points after infection. The stages were identified by light microscopy as rings, early trophozoites, late trophozoites, and schizonts. Early trophozoites were classified when trophozoites were 1/3 the size of the red blood cell or smaller, and late trophozoites were classified when parasites were larger than 1/3 of the infected cell. 300 total parasites were counted per time point per mouse, and the population percentages were calculated as a portion of the total infected erythrocytes. For quantification of tissue-associated luciferase signal, mice were sacrificed and perfused 23 hours after injection of schizonts. The lung and adipose tissue were weighed and homogenized through a 70 μm cell strainer in RPMI medium and spun down at 400 x g for 3 minutes. The pellet was resuspended in 1 ml RPMI. The cell suspension was mixed 1:1 with ONE-Glo Luciferse Assay Buffer (Promega), and the luminescence was measured with the Synergy HTX Multi-Mode Reader. To account for differences in organ weight, the ratio of tissue mass to luminescence was calculated. To include data from three experimental replicates, values were normalized to those obtained for the organs of the wild-type-infected mice.

CD36 adherence assay
Experiments assessing the adherence of infected red blood cells to CHO cells expressing either GFP or CD36 were performed as previously described [42,83]. 1x10 5 CHO cells were seeded on coverslips in 24 well dishes. To obtain late-stage P. berghei, infected blood was incubated in ex vivo cultures as described above. At 16-18 hours blood was resuspended in binding media (RPMI, 2% glucose) and added to the CHO cells at 1.5x10 6 infected cells per well. Cells were incubated at 37˚C for 1 hour with periodic shaking, washed in binding media, fixed in 1% glutaraldehyde and stained with a 1:10 Giemsa-Weiser buffer solution. The number of infected red blood cells bound per CHO cell was assessed by microscopy. At least 250 CHO cells per coverslip were quantified. For each parasite line, the binding to the control GFP-expressing CHO cell line was assessed for reference. The experiment was performed three times with

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts triplicates of all samples. To include data from all experiments, the binding was normalized to the binding of the wild-type P. berghei-infected cells Supporting information S1

PLOS PATHOGENS
A tryptophan-rich protein is required for efficient sequestration of Plasmodium berghei schizonts S5 Fig. Phenotypes of ipis2-and ipis3-in mosquito and blood infection stages. (A) Midguts of mosquitoes were harvested 14 days after an infectious blood meal and microscopy was used to determine the presence or absence of oocysts. (B) The number of sporozoites per mosquito were calculated at least 17 days following infected blood meal. (C) Liver stage ipis2-and ipis3parasites grow similarly to wild-type parasites in vitro. HepG2 cells infected with either wildtype, ibis2-[GFP-Luc;mCherry], or ibis3-[GFP-Luc;mCherry] sporozoites were fixed 48 hours after infection, and the parasites were stained with antibodies against PbHSP70. Exo-erythrocytic forms were imaged with the fluorescence microscope, and their size was determined using FIJI. ns, not significant, unpaired t-test. (D) Parasitemia in the blood of mice following infection with 1000 sporozoites was quantified by microscopy from Giemsa-stained blood smears. � , P < 0.05, �� , P < 0.01, unpaired t-test. (E) ipis2-and ipis3-are less pathogenic in mice than wild-type P. berghei. Survival of mice after infection with 1000 sporozoite. n = 5 (ipis2-), n = 10 (ipis3-). ns, not significant �� , P < 0.01, log-rank test. (F) Mice were infected with 5000-10000 infected red blood cells. Parasites in the infected mice were quantified and classified by stage from Giemsa-stained smears of peripheral blood at the indicated days following infection. � , P < 0.05, unpaired t-test. (G) Mice were infected with equal numbers of schizonts isolated from overnight cultures. Parasites in the infected mice were quantified from Giemsa-stained smears of peripheral blood 18-20 hours after infection. ns, not significant, unpaired t-test. Erythrocytes infected with P. berghei expressing mCherry-tagged proteins in either a wild-type or knock-out genetic background were fixed, labeled with anti-mCherry antibodies and fluorescence microscopy was used to detect the presence of the IBIS via detection of either IPIS2-mCherry or IPIS3-mCherry. DAPI depicts parasite nuclear labelling. GFP is expressed in the parasite cytoplasm exclusively in the knock-out lines. The mCherry signal was amplified by immunofluorescence using antibodies recognizing mCherry. Scale bar = 10 μm. (B) Erythrocyte-associated mCherry signal was quantified in fixed, unstained infected cells. The intensity of the mCherry signal from IPIS2-mCherry and IPIS3-mCherry in both wild-type and knock-out backgrounds was plotted against parasite size. (TIF) Rauch, Berit Söhl-Kielczynski and Werner Stenzel for technical assistance and Mariana de Niz for experimental advice. The authors acknowledge the instruments and expertise of Microscopy Australia at the Centre for Advanced Microscopy, Australian National University, a facility that is funded by the University and the Federal Government through NCRIS.