Plasmodium falciparum Plasmodium helical interspersed subtelomeric proteins contribute to cytoadherence and anchor P. falciparum erythrocyte membrane protein 1 to the host cell cytoskeleton

Summary Adherence of Plasmodium falciparum‐infected erythrocytes to host endothelium is conferred through the parasite‐derived virulence factor P. falciparum erythrocyte membrane protein 1 (PfEMP1), the major contributor to malaria severity. PfEMP1 located at knob structures on the erythrocyte surface is anchored to the cytoskeleton, and the Plasmodium helical interspersed subtelomeric (PHIST) gene family plays a role in many host cell modifications including binding the intracellular domain of PfEMP1. Here, we show that conditional reduction of the PHIST protein PFE1605w strongly reduces adhesion of infected erythrocytes to the endothelial receptor CD36. Adhesion to other endothelial receptors was less affected or even unaltered by PFE1605w depletion, suggesting that PHIST proteins might be optimized for subsets of PfEMP1 variants. PFE1605w does not play a role in PfEMP1 transport, but it directly interacts with both the intracellular segment of PfEMP1 and with cytoskeletal components. This is the first report of a PHIST protein interacting with key molecules of the cytoadherence complex and the host cytoskeleton, and this functional role seems to play an essential role in the pathology of P. falciparum.


Introduction
After invading the human erythrocyte, the malaria parasite Plasmodium falciparum refurbishes its host cell dramatically. The most important changes lead to sequestration of infected cells to the microvasculature of human organsthe sole cause of morbidity and mortality in malaria tropica. These changes also allow the malaria parasite to grow in a parasitophorous vacuole inside the erythrocyte and enable nutrient uptake. The parasite invests approximately 10% of its proteome to refurbish the host cell in this way. Hundreds of exported parasite proteins fall into one of two groups (Spillman et al., 2015). The first group is well defined and consists of proteins containing a pentameric motif, termed PEXEL/HT (Plasmodium export element/host targeting signal) (Hiller et al., 2004;Marti et al., 2004), which allows the establishment of a P. falciparum exportome with approximately 400 proteins (Sargeant et al., 2006). A second group of exported proteins, which do not contain a PEXEL/HT motif or any other identifiable export motif, has also been observed (PEXEL-negative exported proteins). It is difficult to predict the true number of PEXEL-negative exported proteins and hence the total number of exported proteins (Heiber et al., 2013). The export of both groups of proteins results in profound structural and morphological changes in the erythrocyte. For example it causes the formation of electron-dense protrusions on the erythrocyte surface, called knobs (Watermeyer et al., 2016), alters red blood cell (RBC) rigidity (Maier et al., 2008) and increases membrane permeability (Nguitragool et al., 2011).
A key molecule and ligand for binding infected red blood cells (iRBCs) to host cell receptors on the vascular endothelium is the P. falciparum erythrocyte membrane protein 1 (PfEMP1). This major parasite virulence factor is embedded in the knobs through a transmembrane helix and comprises a highly variable ectodomain and a semiconserved intracellular segment, the acidic terminal segment (ATS) (Lavstsen et al., 2003;Mayer et al., 2012). The extracellular part of PfEMP1 consists of multiple adhesion domains, enabling the infected cell to bind to host adhesins including CD36, intercellular adhesion molecule-1 (ICAM-1) and chondroitin sulfate A (CSA). This binding leads to iRBC sequestration within the microvasculature (Kraemer and Smith, 2006). In contrast, the cytoplasmic domain is relatively conserved and was previously thought to interact with the knob-associated histidine-rich protein (KAHRP) (Crabb et al., 1997) and, potentially, with the erythrocyte cytoskeleton components actin and spectrin (Kilejian et al., 1991;Waller et al., 1999Waller et al., , 2002Oh et al., 2000). Recent data, however, do not support a direct ATS-KAHRP interaction but rather an ATS interaction with PHIST proteins PFI1780w and PFE1605w (Mayer et al., 2012;Oberli et al., 2014).
The proteins encoded by the phist multigene family are defined by the presence of a 150-amino acid domain consisting of four consecutive α-helices. Almost all members include a signal sequence and a PEXEL motif (Sargeant et al., 2006). The phist family underwent dramatic lineagespecific proliferation in P. falciparum and is suspected of playing a major role in host cell modifications in cytoplasmic protein associations (Sargeant et al., 2006;Oakley et al., 2007;Frech and Chen, 2013). To date, only a few PHIST proteins have been partially characterized and almost no molecular functions have been assigned, despite their wide distribution within the iRBC. So far, members of the PHIST protein family have been implicated in knob formation (Maier et al., 2008), in altered host cell rigidity (Mills et al., 2007;Maier et al., 2008), in trafficking of and interaction with PfEMP1 (Maier et al., 2008;Mayer et al., 2012;Oberli et al., 2014) and in iRBC adhesion to the brain microvasculature (Daily et al., 2005;Claessens et al., 2012). Moreover, PHIST proteins have been shown to localize to the iRBC periphery (Tarr et al., 2014), possibly binding erythrocyte cytoskeletal components (Kilili and LaCount, 2011;Parish et al., 2013;Proellocks et al., 2014). They have also been found in detergent-resistant membrane fractions (Sanders, 2005) and in exosomes mediating cell-cell communication (Regev-Rudzki et al., 2013).
Previously, we showed that PFE1605w, another member of the PHIST protein family, is exported to knobs and binds directly to the PfEMP1 ATS domain, displaying similar temporal and spatial export as PfEMP1 (Oberli et al., 2014). This finding differs somewhat from those of Proellocks et al. (2014), who suggested an alternative localization. Fluorescence polarization experiments using PFE1605w and a set of ATS domains from different PfEMP1 molecules showed substantial variation in binding affinity, suggesting that different PHIST proteins might have been optimized for different PfEMP1 members (Oberli et al., 2014). Here we present the first functional analysis of a PHIST protein by using inducible downregulation of PFE1605w and also a unique controlled system that blocks PFE1605w at Maurer's clefts. Both approaches showed that reduced levels of PFE1605w within the knobs lead to strongly reduced adhesion of iRBC to endothelial receptors, but that PFE1605w plays no role in transporting PfEMP1 or its surface exposure. PFE1605w directly binds the C-terminus of different ATS domains in vitro and in iRBC and interacts with components of band 3 and junctional complexes at the erythrocyte membrane. This is the first report of a functional role for PFE1605w, which anchors a variety of PfEMP1 variants to the cytoskeleton of the iRBC.

Inducible regulation of PFE1605w
To investigate the function of PFE1605w, we generated a parasite cell line that allowed a conditional expression of the endogenous protein by using the human FK506 binding protein (FKBP) destabilization domain (DD) technique (Banaszynski et al., 2006;Armstrong and Goldberg, 2007). For this, parasites were generated that expressed endogenous PFE1605w as a C-terminally tagged green fluorescent protein-DD fusion protein and whose PFE1605w-DD was rapidly degraded if not stabilized by Shield-1. The integration of the plasmid containing the coding sequence for the PFE1605w-DD construct at the correct locus was confirmed by Southern blot (Fig. S1A). Immunofluorescence assays (IFAs) showed the expected localization of tagged PFE1605w at Maurer's clefts and at the iRBC membrane ( Fig. 1A) as previously shown with non-modified protein (Oberli et al., 2014). In parasites grown for 96 h without Shield-1 (PFE1605w OFF ), PFE1605w levels were highly reduced and the residual protein was visible only within the parasite. Parasites grown for 96 h under the presence of Shield-1 (PFE1605w ON ) displayed normal levels and distribution of PFE1605w. Western blot analysis of synchronized PFE1605w ON /PFE1605w OFF parasites with polyclonal antibodies against PFE1605w showed significantly reduced levels of PFE1605w in PFE1605w OFF parasites compared with PFE1605w ON and 3D7 wild-type parasites (Fig. 1B). To test the PFE1605w ON / PFE1605w OFF parasites' adherence to recombinant CD36, a semistatic adhesion assay was performed. PFE1605w OFF parasites showed a 50% (±9%, n = 3) decrease in adhesion to CD36 (Fig. 1C) compared with PFE1605w ON parasites, a result similar to that observed with a PFE1605w gene disruption (Proellocks et al., 2014). Subsequent addition of Shield-1 to the cultures (PFE1605w RES ) restored adhesion to CD36 to the same level as that for PFE1605w ON parasites (Fig. 1C). This indicates that reduced levels of exported PFE1605w results in a significant reduction of adhesion of iRBCs.

Inducible tethering of PFE1605w at Maurer's clefts
To confirm the importance of PFE1605w presence in knobs for cytoadherence, an alternative approach was used. By conditionally tethering PFE1605w to the cytoplasmic domain of a Maurer's cleft protein, we prevented its transport to the knobs, thereby blocking the presence of PFE1605w at the knob structure. The technique is based on the heterodimerization of the FKBP12 to the FKBP-rapamycin binding (FRB) domain of human mechanistic target of rapamycin in the presence of rapamycin (Haruki et al., 2008;Busch et al., 2009;Robinson et al., 2010;Xu et al., 2010). First, we generated parasites that expressed PFE1605w C-terminally fused to FKBP under the control of the endogenous promoter (Fig. S1A). These parasites were subsequently transfected with a plasmid that episomally expressed membrane-associated histidine-rich protein 1 (MAHRP1) fused to an mCherry tag and an FRB domain under the mal7 promoter (Figs 2A and S1C). In ring-stage parasites, the MAHRP1-FRB fusion protein was exported to the Maurer's clefts, whereas the FKBP-tagged PFE1605w still resided within the parasite (Fig. 2B). In trophozoite and schizont parasites, PFE1605w-FKBP was correctly exported to Maurer's clefts and to knobs as previously described (Oberli et al., 2014). Upon adding 100 nM rapalog (a rapamycin analogue) to ring-stage parasites, a ternary complex at the Maurer's clefts composed of MAHRP1-FRB, rapalog and PFE1605w-FKBP was formed as soon as PFE1605w-FKBP was exported to Maurer's clefts (Fig. 2B) and PFE1605w-FKBP was blocked from localizing in the knobs. Next, we tested the cytoadhesive properties of parasites grown in the presence or absence of rapalog (PFE1605w +RAP / PFE1605w ÀRAP ) to recombinant CD36. Parasites cultured in the presence of rapalog (PFE1605w +RAP ) showed a 62% (±9%, n = 3) reduction in binding to CD36 compared with parasites grown without rapalog (PFE1605w -RAP ) (Fig. 2C). To demonstrate that endogenous untagged PFE1605w does not bind MAHRP1-FRP upon addition of the rapalogue, 3D7 wild-type parasites were transfected with the MAHRP1-FRB plasmid. In both cases (control +RAP / control ÀRAP ), the parasites showed comparable levels of binding to CD36, indicating that no heterodimerization occurred (Fig. 2C).
PFE1605w has no significant role in P. falciparum erythrocyte membrane protein 1 transport To test whether PFE1605w reduction or tethering impairs transport of other well-characterized exported proteins, we analysed the PFE1605w O N /PFE1605w O F F / PFE1605w +RAP /PFE1605w ÀRAP parasites by IFA by using antibodies against PfEMP1, PfEMP3, KAHRP, . The graph displays mean values across triplicate samples normalized to 3D7 wild-type parasite binding. The error bars represent SDs of three independent experiments. An arbitrary threshold (dashed line) for unspecific binding was calculated as the mean level of iRBC binding to 1% w/v BSA plus 2 SDs. P values were calculated by using a two-tailed Student's t-test, asterisk indicates P ≤ 0.05. mature parasite-infected erythrocyte surface antigen (MESA), ring-infected erythrocyte surface antigen (RESA), MAHRP1, MAHRP2 and HSP70x. All tested proteins revealed correct subcellular localization in parasites (data shown for PFE1605w ON /PFE1605w OFF parasites; Fig. S2).
To test whether the observed reduction in CD36 binding was due to a reduction of PfEMP1 surface exposure, we treated iRBCs with trypsin. In all parasite cultures, PfEMP1 was correctly displayed on the iRBC surface ( Fig. 3A and B) as evident by PfEMP1 proteolysis that yields intact ATS domains. The trypsin cleavage assay also revealed that the size of PfEMP1 in all parasites were identical, suggesting that the same PfEMP1 variant was expressed in the parasite lines compared. Scanning electron microscopy (SEM) showed the presence of knobs in all parasite cell lines (Fig. S3); thus, the reduction in cytoadherence observed in PFE1605w OFF and PFE1605w +RAP parasites was not due to decreased knob formation.
Cytoadhesive properties of infected red blood cells expressing different P. falciparum erythrocyte membrane protein 1 in the absence of PFE1605w Previously, we showed that the recombinant PHIST domain of PFE1605w interacted with six different ATS variants with up to 25-fold differences in affinity (Oberli et al., 2014). This suggested that PFE1605w might be optimized for binding to a subset of PfEMP1 variants; hence, it might be relevant only for iRBC binding to a subset of endothelial receptors. Therefore, we selected parasites expressing PFE1605w-DD on different host receptors, including CD36, ICAM-1 and CSA, in order to isolate parasites expressing different PfEMP1 molecules. After four rounds of pre-selection, we obtained parasites binding to CD36, ICAM-1, or CSA (Fig. 4). Expression of var genes in all the pre-selected parasite lines was tested by quantitative PCR (qPCR) and showed a clear differential expression of var genes, suggesting the The nuclei were stained with DAPI. Scale bar = 2 μm. C. Semistatic adhesion assay of PFE1605w ÀRAP /PFE1605w +RAP parasites to immobilized recombinant CD36 protein at 50 μg ml À1 concentration.
The graph displays mean values across triplicate samples, and the error bars represent the SDs of three independent experiments. An arbitrary threshold (dashed line) for unspecific binding was calculated as the mean level of iRBC binding to 1% w/v BSA plus two SDs. P values were calculated by using a two-tailed Student's t-test; the asterisks indicate P ≤ 0.0001.
display of a distinct PfEMP1 variant on the iRBC surface (Fig. 4B). Pre-selected parasites were grown with and without Shield-1 and after 96 h they were allowed to bind to their respective receptor in a semistatic adhesion assay. Parasites grown in the absence of Shield-1 showed an approximately 64% reduction in binding to ) of 625 nM Shield-1. The graphs display overall mean values across triplicate experiments using linear regression with a random effect for experiment. The error bars represent the SDs of the triplicate experiments. An arbitrary threshold (dashed line) for unspecific binding was calculated as the mean level of iRBC binding to 1% w/v BSA plus two SDs. P values were calculated by using a two-tailed Student's t-test. The asterisks indicate P ≤ 0.0001. 'ns' indicates P ≥ 0.05. B. Pie charts show the var transcript distribution in the selected lines. qPCR was performed with specific primers for each var gene as previously reported. CD36 (Fig. 4A). Binding to ICAM-1 was reduced by 30% and binding to CSA showed no reduction at all (Fig. 4A), indicating that PFE1605w plays no role in CSA-mediated cytoadherence.
PFE1605w binds the C-terminal part of the acidic terminal segment Previously, we showed that the recombinant PHIST domain of PFE1605w binds with low-micromolar affinity to the C-terminal part of the ATS domain (ATS-C) of PfEMP1 variant PF08_0141 (Oberli et al., 2014). Sequence conservation among ATS domains suggested that ATS-C provides the PFE1605w binding epitope in most PfEMP1 variants. To test this, we performed in vitro fluorescence polarization binding experiments by using the PFE1605w PHIST domain and fluorescein-labelled recombinant ATS-C fragments from PfEMP1 variants dominantly expressed in preselected parasites (Figs 4B and S4A). In almost all cases, we observed direct PFE1605w-ATS-C binding with dissociation constants (K d ) in the 4-90 μM range (Fig. 5A).
To test whether PFE1605w binds to ATS-C in P. falciparum iRBCs, we designed two mini-PfEMP1 constructs consisting of an N-terminal part of a PEXEL protein (PF13_0275) including a signal sequence and a PEXEL motif, the ATS-C of two PfEMP1 variants and a 3xHA tag to allow detection (Figs 5B and S4B). The PfEMP1 variants selected, PF08_0141 and PFF0010w, display approximately 13-fold difference in in vitro affinity (5 and 65 μM K d respectively) for the PHIST domain of PFE1605w (Oberli et al., 2014). Because the fusion proteins were expressed under the crt promoter, they were found early in the life cycle. Due to the lack of a TM domain, the mini-PfEMP1 was soluble and exported to the erythrocyte cytosol with the predicted size (Fig. S1D). Potential ATS-C interaction partners were detected by co-immunoprecipitation (Co-IP) followed by mass spectrometry (MS) for protein identification. Trophozoite extracts from parasites expressing a mini-PfEMP1 fusion protein were used to isolate potential interacting proteins through an hemagglutinin (HA)-affinity matrix. As a negative control, parasite extract with an excess of soluble HA peptide was added during the affinity-matrix binding of the mini-PfEMP1 fusion proteins. Western blot analysis confirmed that the mini-PfEMP1 fusion proteins were successfully purified and that PFE1605w was coeluted with both mini-PfEMP1 constructs ( Fig. 5C and D). In addition, from duplicate Co-IP experiments, the liquid chromatography-mass spectrometry (LC-MS)/MS analysis detected from both mini-PfEMP1 constructs more than 10 peptide hits for PFE1605w ( Fig. 5C and D). These results demonstrate a direct protein-protein interaction of PFE1605w with the C-terminal part of the ATS domain of different PfEMP1 variants.

Potential PFE1605w interaction partners
To detect other potential PFE1605w interaction partners, we performed Co-IP experiments with parasites express-ing the PFE1605w-3xHA fusion protein, followed by MSbased protein identification. Different components of the human erythrocyte cytoskeleton were detected, in addition to two Plasmodium proteins of unknown function (Fig. 6A).
To confirm these potential PFE1605w interaction partners from the human cytoskeleton, we performed reverse Co-IP experiments with 3D7 wild-type parasite lysate and specific antibodies against human bands 3 and 4.2, both of which locate equally at band 3 and junctional complexes (Mankelow et al., 2012), and ankyrin 1. From the elution of reverse Co-IP with antibodies against human band 4.2, Western blots detected a dominant band, identified as PFE1605w (Fig. 6C). MS identified other components of band 3 and junctional complexes, including band 3, band 4.2, band 4.1, the αand β-chains of spectrin and ankyrin, but no other Plasmodium protein except PFE1605w (Fig.  6B), again suggesting that PFE1605w interacts with one or several cytoskeletal components. To further probe the cytoskeletal interactions of PFE1605w, we produced a fluorescein-labelled recombinant PFE1605w fragment (PFE1605w-C) comprising the C-terminal tail of this protein that follows the PHIST domain. PFE1605w-C was previously shown to bind to inside-out vesicles prepared from uninfected erythrocytes (Proellocks et al., 2014). In the fluorescence-polarization binding experiments, PFE1605w-C interacted with recombinant band 3 with approximately 1.5 μM K d (Fig. 6D). This result demonstrates the direct interaction of PFE1605w with a specific cytoskeletal protein, although we do not exclude the possibility that PFE1605w partakes in a larger multiprotein complex.

Discussion
The remarkable number of exported PHIST proteins predicted and the dramatic lineage-specific proliferation of this multigene family in P. falciparum only (Sargeant et al., 2006) suggest an important role for PHIST proteins in host cell modifications. These modifications lead to the dramatic morbidity and mortality observed with this parasite. This observation is reflected in the number of recent publications showing that PHIST proteins are involved in altering host cell rigidity (Mills et al., 2007;Maier et al., 2008), binding erythrocyte components (Silva et al., 2005;Mills et al., 2007;Pei et al., 2007a;Parish et al., 2013;Proellocks et al., 2014), reducing cytoadherence under flow (Maier et al., 2008;Proellocks et al., 2014), mediating cell-cell communication (Regev-Rudzki et al., 2013), cytoskeletal association (Tarr et al., 2014) and elevated transcript levels of some phist genes in patients (Daily et al., 2005;Mok et al., 2007;Claessens et al., 2012). Although it has been assumed that most PHIST proteins contain one or more interaction epitopes (Sargeant et al., 2006), to the best of our knowledge, no detailed characterization of protein interactions directly linked to the functional role of a PHIST protein has been reported.
Here, we have functionally characterized PFE1605w, which has been shown to bind to the ATS domain of PfEMP1, comigrates with PfEMP1 in space and time and localizes to Maurer's clefts and knobs (Oberli et al., 2014), although an alternative localization has been suggested (Proellocks et al., 2014). In addition to the well-known and frequently used conditional post-translational regulation using an FKBP DD, we applied a 'knock-sideways' or 'anchor-away' system (Haruki et al., 2008;Busch et al., 2009;Robinson et al., 2010;Xu et al., 2010). With this method, we took advantage of the rapalog-induced heterodimerization of the FKBP12 and FRB domains to tether PFE1605w at Maurer's clefts, the transient location for a variety of parasite proteins destined to the iRBC membrane and surface. The tethering technique is a powerful way of revealing the function of an exported protein in host cell refurbishment and helps to dissect the role of these proteins within the export pathway. Both methods, tethering of PFE1605w at Maurer's clefts and protein destabilization, confirmed that mislocalization or depletion of PFE1605w did not result in reduced surface exposed PfEMP1, suggesting no obvious role for PFE1605w in PfEMP1 transport. At the same time, however, both methods of PFE1605w depletion from knobs resulted in large reduction of cytoadherence to CD36.
The different levels of reduction in parasite cytoadherence to specific endothelial receptors upon PFE1605w depletion suggest a highly specialized role for this protein. Previously, we tested six PfEMP1 ATS variants for binding the PHIST domain of PFE1605w (Oberli et al., 2014) and revealed up to 25-fold differences in binding affinities. This suggests that sequence variation in ATS has optimized PFE1605w for binding to a PfEMP1 subset and that perhaps other PHIST proteins might have coevolved with specific ATS domains to create interaction pairs with maximum binding strength. In this simplified model, differences in the PFE1605w-ATS binding affinity might be expected to account for differences in the cytoadherence phenotype upon PFE1605w depletion.
Our assays partly support this model of PFE1605w function, as evidenced by the lack of an effect of PFE1605w depletion on cytoadherence observed for CSA-binding parasites, where the ATS-C fragment of the dominantly expressed PfEMP1 variant (PFL0030c, VAR2CSA) had a very weak binding affinity to the PFE1605w PHIST domain (K d~9 0 μM). In contrast, the ATS-C fragments of PfEMP1 variants most often found in ICAM-1-binding parasites, PF07_0050 and PFL0020w, have up to sixfold higher PFE1605w affinity, and ICAM-1 parasite cytoadherence is reduced by 30% upon PFE1605w depletion.
The complete picture, however, is more nuanced. Co-IP assays coupled with MS robustly detected the in vivo association of PFE1605w with two mini-PfEMP1 constructs encompassing the ATS-C fragment of two PfEMP1 variants. These variants represent the two main subtypes of PfEMP1 ATS, groups A (PF08_0141) and B (PFF0010w). The in vitro affinity of these ATS domains for the PFE1605w PHIST, however, varies by more than 10-fold. We also observed that the ATS-C fragment of the PfEMP1 variant dominantly expressed in CD36-binding parasites, PFD0615c, displays essentially no direct affinity for PFE1605w, despite the large decrease in CD36 cytoadherence upon PFE1605w depletion. The result seems to contradict the simple PFE1605w functional model presented above.
To reconcile these results, we must consider the following: PFE1605w directly binds the majority of ATS domains tested here and in previous studies (Oberli et al., 2014). This interaction is present in vivo for both strongly and weakly associated PFE1605w-ATS pairs. PFE1605w does not affect PfEMP1 transport and it colocalizes to knobs with PfEMP1 (Oberli et al., 2014). Indeed, the significant reduction in receptor binding upon tethering PFE1605w to Maurer's clefts strongly indicates that this protein exercises its functional role in knobs. There, PFE1605w is likely to be joined by (and might act together with) other ATS-binding PHIST proteins such as PFI1780w (Mayer et al., 2012;Oberli et al., 2014), thereby accounting for the partial disruption of cytoadherence upon PFE1605w depletion. Further, in certain PfEMP1 variants, such as the CD36-binding PFD0615c, the PFE1605w-ATS interaction might be mediated or strongly reinforced by other PHIST proteins. Because most of the PHIST proteins are expressed simultaneously, it seems that partnering must occur in the cytosol of the host or directly at the periphery (Fig. 7). Transcriptome analyses of all selected cell lines grown with or without shield excluded a possible upregulation of certain PHIST (data not shown).
To date, only a few direct interactions of exported proteins with cytoskeletal components of erythrocytes have been described and confirmed, e.g. KAHRP (Pei et al., 2005;Weng et al., 2014), PfEMP3 (Pei et al., 2007b), RESA (Pei et al., 2007a) and MESA (Magowan et al., 2000). Both KAHRP and PfEMP3 are required for correct trafficking and functional PfEMP1 display on the erythrocyte surface (Crabb et al., 1997;Waterkeyn et al., 2000), while PFE1605w is not. Co-IP experiments with the full-length PFE1605w-HA fusion protein identified a number of host integral membrane proteins and components of the erythrocyte cytoskeleton as putative binders, and fluorescence polarization experiments confirmed the direct interaction of the PFE1605w C-terminus with the cytosolic domain of band 3. Thus, we have now shown the presence of two interaction epitopes within PFE1605w, at its PHIST domain and the C-terminus, making it an anchoring molecule between PfEMP1 and the host cytoskeleton. The findings are consistent with previous assays suggesting an association of PFE1605w Cterminal fragments with erythrocyte-derived inside-out vesicles (Proellocks et al., 2014). Both co-IP and in vitro experiments suggest that PFE1605w, and thus PfEMP1, targets the host's band 3 and junctional complexes, while, interestingly, PFE1605w was the only parasite protein detected. The next step would be to map the exact interaction epitopes of PFE1605w with band 3 and possibly other cytoskeletal proteins. In addition to eluci- Fig. 7. Schematic representation of the proposed functional role of PFE1605w within the iRBC knob structure. A. Both Co-IP experiments and in vivo experiments showed that the PHIST domain of PFE1605w binds the C-terminal part of different ATS domains and the C-terminal part of PFE1605w targets the host band 3 and junctional complexes, thus making it an anchoring molecule between PfEMP1 and the host cytoskeleton. B. As various PHIST domains interacted with the same PfEMP1 epitope but with different affinities, it is conceivable that another PHIST protein might take over the function of PFE1605w depending on the surface exposed PfEMP1 molecule. dating the complex that anchors PfEMP1 to the cytoskeleton, it would be valuable to study whether other PHIST proteins might bind PfEMP1 variants, in particular VAR2CSA, where CSA cytoadherence was not reduced upon depletion of PFE1605w. The PHIST interactome invites further studies to fully understand the remodelling of the host cell leading to pathology.
In summary, we show that the PHIST protein PFE1605w binds not only to PfEMP1 but also to members of band 3 and junctional complexes of the host cell. PFE1605w, however, plays no role in the transport of PfEMP1. We also show that various PfEMP1 molecules interact differently with PFE1605w and binding to endothelial receptors is partially disrupted upon conditional knock-down or misplacement of PFE1605w. A profound analysis of other exported PHIST proteins and their interaction partners should help to reveal key components of the cytoadherence complex.

Experimental procedures
Parasite culture and transfection Plasmodium falciparum 3D7 cell culture and transfection were performed according to standard procedures (Ljungström et al., 2008). Transfected parasites were grown in the presence of the indicated combinations of 10 nM WR99210 (Jacobs Pharmaceuticals, Cologne, Germany), 2.5 mg ml À1 blasticidin (Life Technologies, Zug, Switzerland), 625 nM Shield-1 and 100 nM A/C Heterodimerizer (Clontech).

Southern blot analysis
Genomic DNA of saponin-lysed parasites was isolated as previously described (Beck, 2002). DNA was digested with AflII and XhoI restriction enzymes (New England Biolabs), separated on a 0.8% w/v agarose gel and transferred to a Amersham Hybond-N + membrane (GE Healthcare). The blot was probed with [ 32 P]-dATP-labelled hdhfr PCR fragments.

Fluorescence microscopy
Immunofluorescence assays were performed on acetone-fixed blood smears of infected parasite cultures (Spielmann et al., 2003) and blocked with 3% v/w BSA. Primary antibodies included

Scanning electron microscopy
After knob selection and Percoll purification, the erythrocytes/iRBCs were fixed in 2% v/v glutaraldehyde in phosphate buffer for 1 h at room temperature. After three washes in PBS, the samples were transferred to coverslips preliminary coated with poly-L-lysine (Sigma), dehydrated in increasing 90% v/v and 2× 100% v/v, 10 min each) and dried at the critical point. Finally, coverslips were mounted onto stubs, sputtered with 5 nm platinum (LEICA EM ACE600) and imaged at 5 kV with a SEM Versa 3D (FEI). The micrographs were coloured in PHOTOSHOP CS6.

Trypsin cleavage assay
For trypsin cleavage, Percoll-purified trophozoite stage parasites were incubated either in L-(tosylamido-2-phenyl) ethyl chloromethyl ketone-treated trypsin (Sigma, 100 μg ml À1 in PBS) or in trypsin and 1 mg ml À1 soybean trypsin inhibitor (Sigma, 1 mg ml À1 in PBS) for 15 min at 37°C. The digest was stopped by the addition of soybean trypsin inhibitor to a final concentration of 1 mg ml À1 . PfEMP1 extraction and subsequent analysis was done as previously described (Van Schravendijk et al., 1993;Waterkeyn et al., 2000). Quantitative PCR for P. falciparum erythrocyte membrane protein 1 expression Synchronized cultures of PFE1605w-DD expressing parasites preselected to bind CD36, ICAM-1 or CSA were split and cultured 96 h in the presence (+) or absence (À) of Shield-1, and ringstage parasites were used for var transcript profiling, as previously described (Dahlbäck et al., 2007). Transcript abundance of each 3D7 var gene was determined relative to internal control transcripts by qPCR by using gene-specific primers and complementary DNA synthesized from total RNA extracted from pelleted infected erythrocytes dissolved in TRIzol.

Recombinant protein expression
Codon-optimized genes encoding the ATS-C fragments of PfEMP1 variants (Fig. S4A) were cloned in a modified pET-16 vector (Merc Millipore). Gene fragments coding for amino acids 300-528 of PFE1605w (PFE1605w-C) or amino acids 1-379 of human erythrocytic band 3 were cloned in a pFloat2 vector (Rogala et al., 2015), which provides an N-terminal His 6 tag.
Purification and fluorescent labelling of ATS-C and PFE1605w-C was performed as previously described (Mayer et al., 2012); briefly, clones were transformed in Escherichia coli strain BL21 (DE3), grown in Luria-Bertani medium and protein expression was induced with 0.1 mM isopropyl β-D-1-thiogalactopyranoside.
Cells were lysed by sonication, and proteins were purified from lysate supernatants by using metal-affinity, ion-exchange and size-exclusion chromatography. Fluorescent labelling was performed by N-(5-fluoresceinyl)maleimide (5-FAM; Invitrogen) conjugating to a single cysteine residue at the protein Nterminus that was added during cloning. Labelled ATS-C and unreacted dye were separated by size-exclusion chromatography. Protein identity and 5-FAM labelling was confirmed by electrospray ionisation (ESI) MS.
Purification of the PFE1605w PHIST domain and the cytosolic band 3 domain was performed as previously described (Zhang et al., 2000;Oberli et al., 2014).
Changes in fluorescence polarization were fitted by using a single binding model in the program ORIGIN (OriginLab).

Coimmunoprecipitation experiments
Three hundred millilitres of cell culture (5% haematocrit, 5-8% parasitaemia) of 3D7 parasites or 3D7 parasites episomally expressing PFE1605w-3xHA/mini-PfEMP1 was cross-linked in 1% v/v formaldehyde. The reaction was stopped after 10 min by adding 2.5 M glycine. Immunoprecipitation was performed as previously described (Dietz et al., 2014). For the Co-IP experiments with the mini-PfEMP1 fusion protein, Pierce α-HA magnetic beads (Thermo Scientific) were used. For the reverse Co-IP with α-band 4.2 antibodies, Dynabeads Protein G were used together with the cross-linking reagent BS3 to avoid coelution of antibodies, according to the manufacturer's protocol (Life Technologies). The eluted fraction was analysed on a 4-12% w/v polyacrylamide bis-Tris gel (Invitrogen) and fractions of it or TCA precipitated pellets were sent to the central core facility for LC-MS/MS analysis. Fig. S1. Plasmid maps, Southern blot and Western blots of extracts from cell lines used in this study. Fig. S2. Localization of well-characterized exported proteins upon PFE1605w reduction. Fig. S3. Reduced levels of PFE1605w do not alter knob formation. Fig. S4. ATS-C and mini-PfEMP1 constructs.