elifeeLifeeLife2050-084XeLife Sciences Publications, Ltd6528210.7554/eLife.65282Research ArticleMicrobiology and Infectious DiseaseStructural Biology and Molecular BiophysicsMalaria parasites use a soluble RhopH complex for erythrocyte invasion and an integral form for nutrient uptakeSchureckMarc A1DarlingJoseph E2MerkAlan2ShaoJinfeng1DaggupatiGeervani3SrinivasanPrakash3OlinaresPaul Dominic Bhttps://orcid.org/0000-0002-3429-66184RoutMichael P5ChaitBrian T4WollenbergKurt6SubramaniamSriramhttps://orcid.org/0000-0003-4231-4115Sriram.Subramaniam@ubc.ca7DesaiSanjay Ahttps://orcid.org/0000-0003-2150-2483sdesai@niaid.nih.gov1Laboratory of Malaria and Vector Research, NIAID, National Institutes of HealthRockvilleUnited StatesLaboratory of Cell Biology, Center for Cancer Research, National Cancer Institute, National Institutes of HealthBethesdaUnited StatesDepartment of Molecular Microbiology and Immunology, and Johns Hopkins Malaria Institute, Johns Hopkins Bloomberg School of Public HealthBaltimoreUnited StatesLaboratory of Mass Spectrometry and Gaseous Ion Chemistry, The Rockefeller UniversityNew YorkUnited StatesLaboratory of Cellular and Structural Biology, The Rockefeller UniversityNew YorkUnited StatesOffice of Cyber Infrastructure & Computational Biology, National Institute of Allergy and Infectious Diseases, National Institutes of HealthBethesdaUnited StatesDepartment of Biochemistry and Molecular Biology, University of British ColumbiaVancouverCanadaSilvieOlivierReviewing EditorSorbonne Université, UPMC Univ Paris 06, INSERM, CNRSFranceSoldati-FavreDominiqueSenior EditorUniversity of GenevaSwitzerland04012021202110e652822911202004012021http://creativecommons.org/publicdomain/zero/1.0/This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Malaria parasites use the RhopH complex for erythrocyte invasion and channel-mediated nutrient uptake. As the member proteins are unique to Plasmodium spp., how they interact and traffic through subcellular sites to serve these essential functions is unknown. We show that RhopH is synthesized as a soluble complex of CLAG3, RhopH2, and RhopH3 with 1:1:1 stoichiometry. After transfer to a new host cell, the complex crosses a vacuolar membrane surrounding the intracellular parasite and becomes integral to the erythrocyte membrane through a PTEX translocon-dependent process. We present a 2.9 Å single-particle cryo-electron microscopy structure of the trafficking complex, revealing that CLAG3 interacts with the other subunits over large surface areas. This soluble complex is tightly assembled with extensive disulfide bonding and predicted transmembrane helices shielded. We propose a large protein complex stabilized for trafficking but poised for host membrane insertion through large-scale rearrangements, paralleling smaller two-state pore-forming proteins in other organisms.

eLife digest

Malaria is an infectious disease caused by the family of Plasmodium parasites, which pass between mosquitoes and animals to complete their life cycle. With one bite, mosquitoes can deposit up to one hundred malaria parasites into the human skin, from where they enter the bloodstream. After increasing their numbers in liver cells, the parasites hijack, invade and remodel red blood cells to create a safe space to grow and mature. This includes inserting holes in the membrane of red blood cells to take up nutrients from the bloodstream.

A complex of three tightly bound RhopH proteins plays an important role in these processes. These proteins are unique to malaria parasites, and so far, it has been unclear how they collaborate to perform these specialist roles.

Here, Schureck et al. have purified the RhopH complex from Plasmodium-infected human blood to determine its structure and reveal how it moves within an infected red blood cell. Using cryo-electron microscopy to visualise the assembly in fine detail, Schureck et al. showed that the three proteins bind tightly to each other over large areas using multiple anchor points. As the three proteins are produced, they assemble into a complex that remains dissolved and free of parasite membranes until the proteins have been delivered to their target red blood cells. Some hours after delivery, specific sections of the RhopH complex are inserted into the red blood cell membrane to produce pores that allow them to take up nutrients and to grow.

The study of Schureck et al. provides important new insights into how the RhopH complex serves multiple roles during Plasmodium infection of human red blood cells. The findings provide a framework for the development of effective antimalarial treatments that target RhopH proteins to block red blood cell invasion and nutrient uptake.

malariacryo-EMnutrient uptakeerythrocyte invasionprotein structurepore-forming proteinsResearch organismP. falciparumhttp://dx.doi.org/10.13039/100000060National Institute of Allergy and Infectious DiseasesDesaiSanjay Ahttp://dx.doi.org/10.13039/100000054National Cancer InstituteSubramaniamSriramhttp://dx.doi.org/10.13039/100000002National Institutes of HealthP41 GM103314ChaitBrian Thttp://dx.doi.org/10.13039/100000002National Institutes of HealthP41 GM109824RoutMichael PChaitBrian Thttp://dx.doi.org/10.13039/501100002784Canada Excellence Research Chairs, Government of CanadaSubramaniamSriramThe funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.Author impact statementA high-resolution structure reveals how the RhopH complex traffics within Plasmodium-infected human erythrocytes to serve divergent, essential functions.Introduction

Malaria parasites evade host immunity by replicating within vertebrate erythrocytes. In humans, the virulent Plasmodium falciparum pathogen uses multiple ligands for erythrocyte invasion (Cowman et al., 2012) and then remodels its host cell to achieve tissue adherence and nutrient acquisition (Goldberg and Cowman, 2010; Wahlgren et al., 2017; Desai, 2014). Remarkably, a single protein complex, termed RhopH, contributes to each of these activities despite their separate timings and cellular locations (Gupta et al., 2015; Goel et al., 2010). The three subunits of the RhopH complex, known as CLAG, RhopH2, and RhopH3, are conserved and restricted to Plasmodium spp.; none have significant homology to proteins in other genera (Kaneko, 2007), suggesting that these proteins and the complex they form evolved to meet the specific demands of bloodstream parasite survival.

While RhopH2 and RhopH3 are single-copy genes in all Plasmodium spp., CLAG proteins are encoded by a multigene family with variable expansion in malaria parasite species infecting humans and other vertebrates including birds, rodents, and primates (Kaneko et al., 2001; Cortés et al., 2007; Rovira-Graells et al., 2015). Each of these subunits is transcribed in mature schizont-infected erythrocytes (Figure 1A; Ling et al., 2004); during translation, these proteins assemble with unknown stoichiometries into a complex that is packaged into rhoptry organelles (Ito et al., 2017). Upon host cell rupture, RhopH3, but not CLAG or RhopH2 subunits, facilitates invasion of the next erythrocyte. Some 18 hr later, CLAG3, a paralog encoded by the parasite chromosome 3, inserts in the host erythrocyte membrane to form the plasmodial surface anion channel (PSAC) for nutrient uptake (Desai et al., 2000; Nguitragool et al., 2011; Pillai et al., 2012); other paralogs may also contribute to PSAC (Gupta et al., 2020) or, in the case of CLAG9, to cytoadherence (Trenholme et al., 2000; Goel et al., 2010; Nacer et al., 2011). RhopH2 and RhopH3 also traffic to the host membrane and are required for PSAC activity (Ito et al., 2017; Counihan et al., 2017). Because these proteins have no homologs in other genera, how they traffic within infected cells and serve these multiple roles is unknown.

A stable, soluble RhopH complex with a 1:1:1 subunit stoichiometry in schizont-infected erythrocytes.

(A) Schematic showing RhopH complex synthesis in schizonts (t = 0–6 hr), role of RhopH3 in erythrocyte invasion (t = 6 hr), and contribution to plasmodial surface anion channel (PSAC)-mediated nutrient uptake at the host membrane (t = 24–44 hr). (B) Immunoblot showing that hypotonic lysis does not release CLAG3, but that alkaline carbonate treatment (CO3=) and freeze–thaw release distinct pools from membranes (membr). (C) Coomassie-stained gel of three RhopH proteins recovered by coimmunoprecipitation after freeze–thaw release. Ribbon at bottom, C-terminal multi-tag strategy used for purification of CLAG3-tv2. (D) Negative staining electron microscope image of purified RhopH complexes; scale bar, 100 nm. (E) Deconvolved native mass spectrometry (MS) spectrum for endogenous RhopH complexes. (F) Negative stain 2D class averages without (top) or with C-terminal tagging with green fluorescent protein (GFP) variants (CLAG3-GFP or RhopH2-mVenus, respectively). GFP-variant density is denoted by black arrows. (G) Negative stain 3D reconstructions using freeze–thaw preparation or harvest from spent media without freeze–thaw. Note similar architectures.

Native mass spectrometry (MS) analysis of the endogenous RhopH complex immunoprecipitated via CLAG3-tv2.

(A) Raw and deconvolved MS spectra of RhopH complex purified from schizont-infected cells. Peak series assignments for intact CLAG3-tv2/RhopH2/RhopH3 complex and a CLAG3-tv2/RhopH2 subcomplex are indicated. The highest charge state for each peak series is noted. Additional peak series at 430.0 kDa and 429.1 kDa were observed at lower intensities (10–15% relative peak signals); these smaller heterotrimeric forms may reflect RhopH3 proteolytic processing (Doury et al., 1994) or low-level degradation. (B) Table of experimental masses obtained from native MS analysis. *Estimated from mean ± standard deviation of measured masses across the charge-state distribution (n ≥ 4). **The expected masses for each subunit calculated from subunit primary sequences with predicted N-terminal signal peptides removed, as indicated in parenthesis: CLAG3-tv2 (C3-tv2, 1–24), 173,995 Da; RhopH2 (R2, 1–19), 160,453 Da; RhopH3 (R3, 1–24), 102,046 Da. ***Lower than expected measured masses may reflect incomplete mapping of subunit N-termini after signal-peptide processing or C-terminal proteolytic processing. (C) Derivation of expected masses for various stoichiometries of the three component proteins, with best matches to the measured masses highlighted.

Subunit modifications and thermostability of the RhopH complex.

(A) Coomassie-stained gels showing recovery of three RhopH subunits after coimmunoprecipitation from schizont-infected cells using a tagged CLAG3 as the bait. CLAG3 and RhopH2 exhibit expected mobility shifts when tagged. Modified constructs are described in Materials and methods. (B) Thermal unfolding of purified RhopH complex. Top, relative fluorescence intensity associated with protein denaturation during indicated temperature ramp. Bottom, first-derivative plot, showing denaturation at >55°C. (C) Scattering intensities at 266 nm and 473 nm, indicative of small and large aggregate formation, during thermal ramp application. Red and blue traces reflect independent samples.

Our data reveal essential features of the RhopH complex. We combine mass spectrometry, single-particle cryo-electron microscopy (cryo-EM) and biochemical studies using conditional knockdown of protein export to determine that the RhopH is initially produced as a soluble complex that functions in erythrocyte invasion. The complex remains soluble in extracellular merozoites and, upon completed invasion, is deposited into the parasitophorous vacuole surrounding the intracellular parasite. A protein translocon on the parasitophorous vacuolar membrane, PTEX (de Koning-Ward et al., 2009), contributes to RhopH export via an unknown mechanism (Ito et al., 2017). Our high-resolution de novo RhopH complex structure and biochemical studies suggest large-scale conformational changes for eventual conversion to an integral form at the host erythrocyte membrane. This conversion is PTEX dependent and enables channel-mediated uptake of host plasma nutrients.

ResultsFreeze–thaw releases a soluble RhopH complex with 1:1:1 subunit stoichiometry

To address these questions, we sought to recover well-behaved RhopH complexes. Alkaline Na2CO3 extraction but not hypotonic treatment partially released CLAG3 from infected cell membranes (Figure 1B, top row), implicating both integral and peripheral membrane pools. We found that simple freeze-thaw also releases some CLAG3 from the peripheral pool (bottom row); although Na2CO3 extraction releases a larger amount, freeze–thaw is gentler and does not denature many proteins. Neither of these treatments is expected to release integral membrane proteins. Using multiple C-terminal tags engineered into the single clag3h gene of the KC5 line (Gupta et al., 2018; CLAG3-tv2; Figure 1C, bottom), we effectively harvested this minor fraction from human blood cultures. This CLAG3 remained associated with RhopH2 and RhopH3 (Figure 1C) and yielded monodisperse protein complexes in negative stain imaging (Figure 1D). Native mass spectrometry (MS) yielded a molecular weight of 433,790 ± 10 Da (Figure 1E, Figure 1—figure supplement 1), matching the expected mass for a heterotrimeric complex with a 1:1:1 stoichiometry; a 0.6% mass error may reflect post-translational modification and/or proteolytic processing, as reported for RhopH3 (Ito et al., 2017). A smaller 333,232 ± 3 Da fraction corresponded to a minor CLAG3–RhopH2 heterodimer. Thus, freeze–thaw permits gentle, detergent-free harvest of this essential complex.

RhopH complexes segregated into 2D classes with two primary views (two-lobe and side views, Figure 1F, top row). We next used a green fluorescent protein (GFP)-derivative-tagging approach (Ciferri et al., 2012), confirmed integrity of each variant (Figure 1—figure supplement 2A), and detected densities reflecting addition of this globular epitope tag. This independently confirmed single copies of each subunit and established an orthogonal arrangement for CLAG3 and RhopH2 (arrows, Figure 1F). Three-dimensional reconstruction provided a low-resolution image of the entire complex and established a two-lobed structure (Figure 1G). A similar two-lobed structure was obtained for RhopH complexes recovered from spent media without protease inhibitors, detergents, or freeze–thaw, implicating a highly stable complex. Finally, the purified RhopH complex resisted aggregation and unfolding at temperatures above those seen in malaria fevers (Figure 1—figure supplement 2B,C). We submit that a thermostable RhopH complex is well-equipped for transit through diverse subcellular environments.

Structure of the RhopH complex

We next determined the complex’s de novo structure using cryo-EM and concentrated protein from sequential coimmunoprecipitation (0.8–2 mg/mL, FLAG and His10 tags). Initial analyses with 2D and 3D classifications yielded a two-lobed structure with a 3.3 Å resolution (Figure 2, Figure 2—figure supplement 1; Table 1); per-particle contrast transfer function (CTF) estimation and motion correction improved overall resolution to 2.9 Å.

High-resolution structure of the soluble RhopH complex and stabilizing interactions.

(A) Side view of the cryo-electron microscopy (cryo-EM) reconstruction with CLAG3, RhopH2, and RhopH3 color scheme maintained in all figures that show the structure. (B) Side and 90o-turned bowl ribbon diagrams of the RhopH complex. Buildable N- and C-terminal residues of each subunit are labeled. In (A) and (B), the CLAG3 HVR and single validated transmembrane α-helix are colored yellow and green, respectively. (C) CLAG3 domain architecture, with residues numbered from N-terminus. (D) Ribbon schematic illustrating pairwise interactions between subunits. The visualized N- and C-termini of each subunit are indicated by ribbon color change. (E) CLAG3–RhopH3 binding interface, as determined by the CLAG3 1300 loop and 300 regions, shown from separate angles. Enlarged views at bottom show critical CLAG3 residues involved in hydrophobic and charge–charge interactions. (F) The CLAG3–RhopH2 binding interface from different views. Enlarged image at bottom left shows CLAG3 α-helices that define the RhopH2 bridge, with helix numbering from one at the protein N-terminus. Right, Space-filling view of the CLAG3–RhopH2 surfaces at their binding interface. The proteins are separated from one another and rotated to expose the binding surfaces; blue and red shading reflect positive and negative electrostatic potential, respectively. Complementary surface potentials on these surfaces form salt bridges and contribute to tight interactions. (G) Ribbon schematic showing positions of cysteines that form intramolecular disulfides (black), unbonded cysteines (red), and cysteines that were not visualized (dashed black lines). Intermolecular disulfides were not observed. (H) Tabulated list of disulfides and unbonded cysteines.

Cryo-electron microscopy (cryo-EM) data processing scheme.

(A) A raw representative cryo-EM micrograph showing RhopH complexes; scale bar, 100 nm. (B) Example 2D classes generated from 214,233 particles. (C) Flow chart of data processing procedures. An initial model was used for 3D classification, resulting in one well-resolved class (class 3). Initial refinement yielded a map with a poorly resolved small lobe; particle polishing and CTF refinement improved resolution to 2.92 Å. Multibody refinement with indicated masks permitted more complete reconstruction of the small lobe.

Observed and predicted movements of the RhopH complex.

(A) First (inward) component of motion derived from multibody analysis and corresponding first normal mode elNémo prediction; 90o rotated views are shown below each model. Arrows indicate direction of movement. (B) Second (torsional) component of motion. (C) Third, fourth, and fifth normal modes of motion predicted by elNémo, left to right, respectively. The third mode predicts tilting of the two lobes, while the others predict a slight expansion and contraction of the RhopH complex about its center.

Multiple residues involved in CLAG3–RhopH2 binding.

Ribbon diagrams of the exposed surfaces of CLAG3 and RhopH2 at their binding interface (A and B, respectively). Conserved residues are indicated using Consurf color coding; scale bar at top.

Conservation of amino acids at subunit interfaces.

Surface representations of RhopH subunit binding sites, showing sequence conservation at the RhopH2 and RhopH3 binding sites on CLAG3 (A and B, respectively) and at the CLAG3 binding sites on RhopH2 and RhopH3 (C and D). Consurf color scale bar at top for all panels.

Neighbor-joining tree and structural similarities.

(A) Major CLAG paralog clades are indicated by labeled red arcs. The ‘Clade F’ label represents the Plasmodium taxonomy proposed by Galen et al., 2018 and does not reflect the precise structure of our phylogeny tree. Bootstrap support is indicated by branch width, with thin branches have less than 70% support. (B) Three-dimensional alignment of Bcl-xL to RhopH2, as identified through an exhaustive Protein Data Bank (PDB) search using the Dali Server. Expanded view shows the aligned region. Z score, 4.2; root-mean-square deviation (RMSD), 3.4 Å; number of residues aligned, 95. (C) Three-dimensional alignment of SepL to RhopH3, based on a similar PDB search. Z score, 6.0; RMSD, 3.6 Å; number of residues aligned, 115.

Cryo-electron microscopy (cryo-EM) data collection, refinement, and validation statistics.
Freeze–thaw non-inserted (EMDB-22890) (PDB 7KIY)
Data collection and processing
Microscope Camera Calibrated magnificationTitan Krios K2 Summit 59,500
Voltage (kV) Exposure time Frame/total (s) Number of frames per image300 23.2 2.5 frames/s 58
Electron exposure (e−/Å2)69.6
Defocus range (μm)0.5–3.5
Pixel size (Å) Box size (pixels)0.82 400
Symmetry imposedC1
Initial particle images (no.)311,390
Final particle images (no.)68,216
Map resolution (Å) FSC threshold2.92 0.143
Map resolution range (Å)2.89–12.10
Refinement
Initial model used (PDB code)na
Model resolution (Å)3.06
Model resolution range (Å)2.9–12.10
Map sharpening B factor (Å2)−34
Model composition Nonhydrogen atoms Protein residues Ligands19,943 2388 0
B factors (Å2) Protein Ligand43.95 Na
R.m.s. deviations Bond lengths (Å) Bond angles (°)0.012 1.33
Validation MolProbity score Clashscore Poor rotamers (%)2.51 13.52 1.42
Ramachandran plot Favored (%) Allowed (%) Disallowed (%)78.25 19.38 2.36

Soluble RhopH is a heterotrimeric complex consisting of single CLAG3, RhopH2, and RhopH3 subunits (Figure 2A), as predicted above. CLAG3 mediates subunit associations through independent contacts with RhopH2 and RhopH3, which do not directly interact with one another. The visualized complex assumes a ‘shallow bowl with a short base’ appearance due to an out-of-plane orientation of RhopH2 relative to CLAG3 (Figure 2B). On the opposite face, a CLAG3 mid-section protrudes to create a short ‘base’ that includes a critical amphipathic α-helix proposed to line the PSAC pore at the host membrane, as described below. The bowl’s opposite rim is formed by globular α-helices from CLAG3 and RhopH3.

From other angles, an asymmetric two-lobed architecture is apparent, with a well-resolved large lobe that enabled confident de novo model building for CLAG3 and RhopH3 (Figure 2C, Figure 2—figure supplement 1C). In contrast, the small lobe was initially not well-resolved.

We hypothesized that both lobes have defined structures that undergo relative movement and therefore used multi-body refinement (Nakane et al., 2018) to identify rigid but mobile substructures. Assuming two bodies joined by a CLAG3 stem, we refined each lobe separately and improved the small lobe’s resolution (Figure 2—figure supplement 1C). The small lobe’s hammer-shaped ends were now clearly visualized, improving model building from 225 to 513 residues for RhopH2. Excluding their flexible N- and C-terminal tails, ≥90% of CLAG3 and RhopH3 residues were also confidently localized. Multibody refinement also defined the directions and extent of motion between the two lobes (Figure 2—figure supplement 2; Videos 1 and 2). Interestingly, consideration of protein energy landscapes using normal mode analysis (Suhre and Sanejouand, 2004) predicted remarkably similar motions (Videos 3–7). Although the biological significance of this mobility is uncertain, conservation of the stem sequence and length in P. falciparum CLAG paralogs and among other Plasmodium spp. supports an important role (Figure 2—figure supplements 3 and 4; 48% bridge region identity between divergent human P. falciparum and P. vivax CLAGs).

First (inward) component of motion derived from multibody analysis.Second (torsional) component of motion derived from multi-body analysis.First structural movement predicted by elNémo normal mode analysis.Second structural movement predicted by elNémo normal mode analysis.Third structural movement predicted by elNémo normal mode analysis.Fourth structural movement predicted by elNémo normal mode analysis.Fifth structural movement predicted by elNémo normal mode analysis.Subunit interactions and roles

CLAG3 contains three visually distinct domains (Figure 2C): an N-terminal sphere, an elongated central bridge for binding RhopH2, and a C-terminal bundle encasing an amphipathic helix that later integrates in the host erythrocyte membrane (Sharma et al., 2015). The N- and C-terminal domains hold RhopH3 tightly through bidentate interactions via a ‘1300 loop’ and a ‘300 region’ that form orthogonal pincer-grasp interactions. We illustrate these high-confidence interactions between CLAG3 and the other subunits in Figure 2D.

The 1300 loop packs against RhopH3 with discrete foci of hydrophobic and salt-bridge interactions (formed by CLAG3 residues Y1331, F1338, F1347, L1348 and D1334, D1339, D1340, E1345, respectively, Figure 2E, left panels). These CLAG3 residues and the cognate RhopH3 residues are highly conserved (Figure 2—figure supplement 4B and D), implicating essential roles in stabilizing the complex. The less strictly conserved 300 region consists of three α-helices, with two helices (10 and 11) interacting with RhopH3 residues 397–412 and 575–588 to create a hydrophobic core with a convergence of aromatic side chains (core formed by F314, F322, V330, M333, M394, Y399, Figure 2E, bottom right). The third CLAG3 helix (helix 14) and an upstream loop are closely apposed to RhopH3 through complex interactions. Together, the 300 region and 1300 loop produce an extensive 3700 Å2 CLAG3 interface with RhopH3.

The 2005 Å2 CLAG3–RhopH2 interface is much more fragmented (residues 706–715, 787–805, and 920–939 of CLAG3 and 414–435, 580–594, 682–708, and 760 from RhopH2). Interestingly, the CLAG3 backbone threads back and forth through the bridge domain (Figure 2F, bottom left) to form a wall-like interface; both surfaces are enriched with hydrophobic, conserved residues that form stable interactions (Figure 2F, bottom right; Figure 2—figure supplements 3 and 4A,C).

Each subunit has numerous conserved cysteines that contribute to tight assembly of this large complex through the formation of observed and possible disulfide bonds (Figure 2G,H). Although we did not detect bonding between subunits, several cysteines were not visualized and could form such interactions. Conserved cysteines are a common feature of rhoptry proteins (Kaneko, 2007); they presumably contribute stability during egress and erythrocyte invasion and may also be critical for RhopH enzymatic activity at its final erythrocyte membrane destination (Carter, 1973).

CLAG3’s central position in the structure, together with its surface exposure on erythrocytes and immune selection (Iriko et al., 2008), likely accounts for expansion of the clag gene family in all Plasmodium spp. We examined CLAG phylogeny and found that P. falciparum paralogs cluster into well-supported groups containing species infecting other mammals (Figure 2—figure supplement 5A). CLAG9 clustered independently and represented an older lineage. Sequences from Plasmodium spp.-infecting birds formed a separate group (labeled ‘Clade F’ in Figure 2—figure supplement 5A, based on taxonomy proposed by Galen et al., 2018). These sauropsid CLAG sequences are split into two well-supported orthologous groups, one that is basal to the CLAG2/CLAG3/CLAG8 orthologs and one that is basal to the CLAG9 orthologs. This pattern suggests an ancient split into two paralogs in the common ancestor of sauropsid and mammalian Plasmodium spp., with subsequent diversification of mammalian paralogs. This diversification and ongoing gene family expansion (Otto et al., 2018) may yield distinct RhopH complexes capable of divergent functions including erythrocyte invasion, cytoadherence, and nutrient uptake. Expansion may also permit fine-tuning of PSAC permeabilities to allow nutrient uptake in both malnourished and well-fed hosts (Mira-Martínez et al., 2017).

Structural similarity searches of the Protein Data Bank (PDB) revealed weakly significant hits for each subunit that may guide structure–function studies of this Plasmodium-restricted complex (Figure 2—figure supplement 5B and C). RhopH3 exhibited the greatest structural similarity with alignment to domains from SepL, a regulator of type III translocon-based secretion in bacteria (Burkinshaw et al., 2015). RhopH2 partially aligned with Bcl-xL, an anti-apoptotic protein that also regulates membrane permeabilization (Finucane et al., 1999). Both hits from our structural similarity searches raise the tantalizing possibility that RhopH2 and RhopH3 function to regulate PSAC. Such regulation could produce the unprecedented selectivity of this channel, which imports diverse nutrients including purines, amino acids, sugars, and some vitamins while maintaining very low Na+ permeability to prevent host cell osmotic lysis (Cohn et al., 2003).

Transmembrane domains are shielded in the soluble complex

Biochemical studies point to a direct contribution of the RhopH complex in PSAC-mediated nutrient uptake (Gupta et al., 2018; Gupta et al., 2020), with a single confidently predicted CLAG3 transmembrane domain distal to a 10–30 residue hypervariable region (HVR, Figure 3A). Site-directed mutagenesis of a conserved A1215 residue in this transmembrane domain (α-helix 44 in our structure) alters channel gating, selectivity, and conductance, supporting a pore-lining helix (Sharma et al., 2015). Notably, a PDB structure search identified this and several neighboring helices with a significant alignment to APH-1, an integral membrane component of human γ-secretase (Figure 3—figure supplement 1A). The corresponding APH-1 α-helix makes stable interactions with phospholipid in that structure (Bai et al., 2015), further supporting membrane insertion of CLAG3 α-helix 44.

A CLAG3 transmembrane helix is buried in the soluble complex.

(A) Posterior probability plots for transmembrane (TM) domain prediction, determined for residues 940–1240 of CLAG3 using indicated algorithms. Green circle on the plots’ single confidently predicted TM (α-helix 44) represents A1215; HVR, hypervariable region. (B) Cylinder view of RhopH complex map showing the buried α-helix 44 (green). (C) Enlarged and turned view from (B). Additional helices that may interact with membranes are labeled. (D) Corresponding cryo-electron microscopy (cryo-EM) density and an adjacent poorly ordered HVR. (E) Top and side views of CLAG3 α-44. Note that hydrophobic side chains cluster on the upper helix surface in these views; polar residues are at the opposite surface and may line the eventual pore. (F) Slice-through view showing a thin interior section of the RhopH complex. CLAG3 is shown as sticks and colored by Consurf conservation score for each residue. Note that α-44 exhibits higher sequence variation than neighboring domains. (G) Cylinder view with known and predicted TM helices in green; these helices are buried and physically separated from one another.

Predicted transmembrane (TM) domains are buried in the soluble RhopH complex.

(A) Structural alignment of CLAG3 to APH-1, an integral membrane component of human γ-secretase, as identified through an exhaustive PDB search. The aligned region includes α-helices 39, 40, 43, and 44 of CLAG3; a phospholipid molecule is seen in stable association with a corresponding α-helix in APH-1. Dali Z score, 4.5; RMSD, 4.3 Å; number of residues aligned, 116. (B) Cylinder view diagrams of CLAG3 and two unrelated ‘two-state’ proteins—colicin Ia and Bax. In their soluble states, each protein buries its hydrophobic TM helices within the structure for stability (green cylinders); these helices later become exposed to allow membrane insertion. (C and D) Posterior probability plots for TM domain prediction for RhopH2 and RhopH3 based on the Phobius algorithm (Käll et al., 2004). Each protein has a single confidently predicted TM that corresponds to an α-helix in our structure. (E and F) Cylinder view representations of RhopH2 and RhopH3, showing that predicted TM helices are shielded.

This important helix is buried within a CLAG3 C-terminal bundle (Figure 3B–D), paralleling buried hydrophobic helices in some much smaller pore-forming proteins (Dal Peraro and van der Goot, 2016; Figure 3—figure supplement 1B). Transverse and longitudinal views establish that multiple Phe side chains segregate to one surface of helix 44 and that polar side chains line up at the opposite face (Figure 3E), as expected for a helix that lines an aqueous pore (Sharma et al., 2015). Although its physicochemical properties are conserved in CLAG orthologs, helix 44 exhibits little primary sequence conservation (Figure 3F). In contrast to this helix, the nearby HVR was poorly ordered, consistent with an unstructured extracellular loop that functions as an immune decoy (Figure 3D). The single predicted transmembrane domains on RhopH2 and RhopH3 are also buried in the soluble structure (helices defined by V740-D757 and G595-Y622 of these subunits, respectively; Figure 3—figure supplement 1C–F). Thus, known and predicted transmembrane domains are shielded in the trafficking RhopH complex (Figure 3G), implicating large-scale protein rearrangements for their membrane insertion.

RhopH is synthesized as a non-integral complex

The peripheral and integral membrane pools (Ito et al., 2017) of the RhopH complex may both be formed during protein synthesis. Alternatively, the complex may be produced exclusively as a soluble form for trafficking and membrane insertion at a later point in the cell cycle. To distinguish between these models, we performed fractionation studies with synchronous cultures at defined developmental stages. During stage-specific synthesis in schizont-infected cells (Ling et al., 2004), both peripheral and integral membrane pools were reproducibly detected (Figure 4A, top row). This finding’s interpretation is complicated by preexisting CLAG3 derived from the preceding cycle and trafficked to the infected cell surface (Figure 1A). To address this uncertainty, we treated early schizont-stage cultures with protease to identify prior-cycle CLAG3 inserted at the erythrocyte membrane. As the integral pool was quantitatively proteolyzed (Figure 4A,B), we conclude that the integral pool in these cells reflects protein made in the previous cycle; the larger carbonate-extractable pool represents newly synthesized protein.

RhopH is produced as a soluble complex and requires interaction with the PTEX translocon for membrane insertion.

(A) Immunoblot showing that pretreatment of mature schizont-infected cells with pronase E, a broad specificity protease, reduces CLAG3 in the membrane fraction (membr) without affecting freeze–thaw released or Na2CO3-extractable (CO3=) pools. (B) Mean ± S.E.M. fractional reduction of indicated CLAG3 pools upon pronase E treatment, determined from changes in band intensities from matched immunoblots as in (A). *p = 0.01, n = 3. (C) Immunoblots showing membrane fractionation of CLAG3 and Band3, a host membrane marker, in purified merozoites and their schizont-infected progenitor cells. Representative of two independent trials. (D) Similar fractionation studies at indicated stages throughout the P. falciparum bloodstream cycle. While schizont- and trophozoite-infected cells were enriched by the percoll–sorbitol method, ring-infected cells cannot be similarly enriched, presumably accounting for CLAG3 detection in the soluble lane and non-additive fractionation in rings. (E) Mean ± S.E.M. band intensities from three independent trials as in (D). *p < 0.005. (F) Schematic shows PTEX-mediated protein translocation and refolding in host erythrocyte cytosol. Middle, Anti-CLAG3 immunoblots from 13F10 cellular fractions with and without trimethoprim (TMP) (top and bottom blots, respectively). Bar graph shows mean ± S.E.M. fraction of integral membrane CLAG3 (Fint), determined from band intensities. *p < 0.015; n = 3. (G) CLAG3-tv2 fractionation studies using enriched mature infected cells. Top, Anti-HA blot showing that soluble CLAG3 (freeze–thaw and CO3 = lanes) is not susceptible to extracellular protease, but the integral pool (membr) is. The ~40 kDa cleavage product remains membrane embedded. EXP2, an intracellular parasite membrane protein, is primarily integral and is protease insensitive. Aldolase, a parasite cytosolic protein, is quantitatively released by freeze–thaw and carbonate treatment. Representative of more than three trials.

Fractionation studies using purified merozoites revealed carbonate-extractable CLAG3 and undetectable levels of integral protein (Figure 4C), consistent with packaging of newly synthesized RhopH complex into rhoptries and jettisoning of the prior-cycle integral host membrane pool upon schizont rupture; the host membrane marker, Band3, is also discarded at egress. Thus, CLAG3 is synthesized as a soluble protein that associates with other RhopH subunits to interact peripherally with membranes in rhoptries; whorls seen in rhoptries may provide a membranous surface for transfer of these proteins to the next erythrocyte (Bannister et al., 1986).

We then tracked this newly synthesized pool through the parasite bloodstream cycle and found that merozoites transfer their peripheral CLAG3 pool to immature ring-stage parasites, which also carry negligible amounts of the integral form (Figure 4D,E, rings). With parasite maturation, CLAG3 transitions from a primarily extractable form upon synthesis in schizonts into a growing integral pool after transfer into new erythrocytes (Figure 4D,E, trophozoites). During this conversion, CLAG3 remains associated with other RhopH subunits and eventually localizes to the infected host cell membrane (Vincensini et al., 2008; Nguitragool et al., 2011; Ahmad et al., 2020).

How does this 440 kDa soluble RhopH complex convert into an integral form? Upon erythrocyte invasion, these and other rhoptry proteins are deposited into the parasitophorous vacuole. The PTEX protein translocon exports proteins secreted by the intracellular parasite into host cytosol (de Koning-Ward et al., 2009; Beck et al., 2014; Ho et al., 2018). It may therefore also export RhopH proteins into host cytosol; such transfer would be novel as it has not been established for other merozoite proteins deposited in the vacuole. While two studies have obtained conflicting results about whether RhopH proteins are exported via this translocon, both reported that PTEX knockdown abolishes activation of PSAC-mediated nutrient uptake at the host membrane (Beck et al., 2014; Ito et al., 2017). To examine membrane insertion, we performed CLAG3 fractionation using 13F10, a conditional PTEX knockdown parasite (Beck et al., 2014) whose protein export requires trimethoprim (TMP, Figure 4F). We found that CLAG3 transitions to an integral form in this parasite normally in the presence of TMP, but that PTEX knockdown produces a loss of integral CLAG3 (-TMP, p = 0.01, n = 3). CLAG3 that failed to insert into the membrane was more readily solubilized (-TMP, soluble lane), possibly due to protein crowding as a result of blocked export from the parasitophorous vacuole. Thus, RhopH membrane insertion is dependent on PTEX activity.

Stage-dependent membrane insertion was further evaluated in CLAG3-tv2 parasites with protease susceptibility studies. Both the freeze–thaw released and carbonate-extractable pools of CLAG3 were unaffected by extracellular protease, but the integral pool at the host membrane yielded a C-terminal cleavage product that remained membrane embedded (Figure 4G). α-Helix 44 is within this cleavage fragment and likely provides the responsible transmembrane anchor. Collectively, these findings indicate that CLAG3 is synthesized and trafficked in a soluble RhopH complex that undergoes marked rearrangements during its export to enable insertion at the host membrane.

Discussion

We propose that RhopH evolved as a modular three-protein complex suited for essential and divergent functions at separate points in the bloodstream parasite cycle (Figure 5). A soluble form, packaged into rhoptry secretory organelles, facilitates RhopH3 contribution to erythrocyte invasion through still unknown mechanisms that presumably involve surface interactions. A large exposed surface area of ~32,000 Å2 and globular architecture of RhopH3 provide candidates for inquiry. Our structure similarity searches found that RhopH3 residues 434–665 align with domains 2 and 3 of SepL; because domain three mediates interaction with the Tir receptor (Burkinshaw et al., 2015), one possibility is that RhopH3 interacts with an unidentified host cell receptor at this site. The RhopH3 C-terminus provides another surface for the presumed interactions, as suggested by site-directed mutagenesis of serine 804 and by studies with a monoclonal antibody against a 134 aa recombinant fragment (Doury et al., 1994; Ekka et al., 2020). This entire region (residues 716–897) is not resolved in our structure and appears to be flexible. Invasion-inhibiting antibodies that bind here may directly or indirectly prevent essential interactions with a cognate receptor. These findings and recent structural studies of the Rh5-CyRA-Ripr (Wright et al., 2014; Wong et al., 2019) should enable structure-guided therapies targeting erythrocyte invasion, an Achilles heel in the parasite’s bloodstream cycle.

Model of RhopH synthesis and trafficking.

The complex is produced in a soluble form and packaged into rhoptries (1) before transfer via extracellular merozoites to the nascent parasitophorous vacuole of a new host erythrocyte (2). The rhoptry may also contribute lipids to the nascent parasitophorous vacuole (Dluzewski et al., 1995). The soluble RhopH complex then crosses the PVM and undergoes membrane insertion via a PTEX-dependent mechanism (3). Finally, it is deposited on the host membrane with a small variant region on CLAG3 exposed to plasma, enabling channel-mediated nutrient uptake (4).

A soluble RhopH complex may also facilitate transfer to new erythrocytes for a second role in PSAC-mediated nutrient uptake (Nguitragool et al., 2011). We determined that the complex is transferred to the new host cell and deposited in the parasitophorous vacuole in a soluble form. The member subunits may then be exported into host cell cytosol via PTEX, as suggested by confocal immunofluorescence assays showing blocked export of each RhopH subunit in PTEX knockdown parasites (Ito et al., 2017). Forward and reverse coimmunoprecipitation experiments also suggest that the RhopH complex directly interacts with PTEX to enter host cell cytosol (de Koning-Ward et al., 2009; Counihan et al., 2017). We next show that CLAG3 membrane insertion occurs via a PTEX-dependent mechanism (Figure 4F). Insertion may occur either concurrently with or after export. Because exported chaperones are thought to facilitate refolding of exported proteins and subsequent transit to specific host cell sites, failed CLAG3 membrane insertion may result from blocked export of multiple effector proteins.

Our structure reveals several intriguing and unique problems faced by the RhopH complex during its export and host membrane insertion. How this large complex crosses the parasitophorous vacuolar membrane remains unclear. If it transits directly through PTEX, this tightly assembled ternary complex with numerous disulfide bonds would require carefully coordinated unfolding and disassembly by HSP101 and possibly other vacuolar activities before translocation (Beck et al., 2014; Ho et al., 2018; Matthews et al., 2019). Subsequent reassembly in host cytosol may be even more complicated, with largely uncharacterized machinery needed to reform a stable complex without denaturation.

Another dilemma exposed by these studies is the precise mechanism by which one or more RhopH subunits become integral to the host erythrocyte membrane while remaining strictly associated with each other (Ito et al., 2017; Ahmad et al., 2020). Although membrane insertion during transit through PTEX would follow the precedent of Sec translocon-mediated membrane insertion in bacteria and other eukaryotes (Denks et al., 2014), PTEX appears to lack a lateral gate, as used by other translocons to transfer cargo proteins into the adjacent lipid bilayer (Egea and Stroud, 2010; Corey et al., 2019; Ho et al., 2018). We tend to favor membrane insertion after transfer into host cytosol. In this scenario, the energetically demanding process of conformational rearrangement to expose and insert specific α-helical domains into the host membrane may be facilitated by interactions with parasite-derived chaperones and Maurer’s cleft organelles (Proellocks et al., 2016).

Although various studies support a role of the RhopH complex in PSAC formation and nutrient uptake (Nguitragool et al., 2011; Mira-Martínez et al., 2019; Sharma et al., 2013; Ito et al., 2017; Counihan et al., 2017), whether this ternary complex directly forms the aqueous pore in the host erythrocyte membrane remains debated. In vitro selection previously implicated a short, but critical amphipathic CLAG3 motif in solute selectivity and PSAC single-channel gating (Lisk et al., 2008; Sharma et al., 2015). Our de novo structure establishes that this motif indeed forms an α-helix with hydrophobic and polar side chains segregated to opposite faces (helix 44, Figure 3), supporting a pore-lining helix in the host membrane. While studies suggest that CLAG3 oligomerizes at the host membrane and has an surface-exposed variant region (Figure 5; Gupta et al., 2018; Nguitragool et al., 2014), RhopH2 and RhopH3 are not exposed based on protease susceptibility studies (Ito et al., 2017).

The CLAG3 helix 44 and the individual predicted transmembrane domains on RhopH2 and RhopH3 are separated from one another by 46–101 Å in the soluble structure (Figure 3G). If all three helices come together to form the eventual nutrient pore, a remarkable rearrangement of the complex will be required during its conversion from a soluble to a membrane-inserted form. While our findings suggest interactions with PTEX or exported chaperone proteins, these rearrangements may also be facilitated by post-translational modifications such as site-specific phosphorylation and lysine acetylation (Cobbold et al., 2016; Pease et al., 2013).

Our findings provide a framework for understanding two unique and essential functions in bloodstream malaria parasites. Structure-guided development of therapies can now be pursued against a strictly conserved target exposed to plasma at two key points in the parasite cycle.

Materials and methods
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Plasmodium falciparum)KC5• 10.1128/mBio.02293–17Wt control
Cell line (Plasmodium falciparum)CLAG3-tv2This paperC-terminal His10-FLAG-thrombin-TEV-HA-twinstrept-BC2 tag
Cell line (Plasmodium falciparum)CLAG3-GFPThis paperC-terminal His8-mGFP-FLAG-Twin Strep tag
Cell line (Plasmodium falciparum)CLAG3-tv1This paperC-terminal 3xFLAG-3xHA-His8-Strept II tag
Cell line (Plasmodium falciparum)CLAG3-tv1+RhopH2-mVThis paperCLAG3-tv1 with tandem RhopH2 C-terminal mVenus tag
Cell line (Plasmodium falciparum)13F10• 10.1038/nature13574TMP-dependent HSP101 conditional knockdown
AntibodyAnti-CLAG3 (mouse polyclonal)• 10.7554/eLife.23485(1:1000)
Sequence-based reagentCLAG3 sgRNAThis paperFor CRISPR editing5′-TAAAAACACTAATAAGACCA-3′
Recombinant DNA reagentpUF1-Cas9• 10.1038/nbt.2925Cas9 expression
Recombinant DNA reagentpL6• 10.1038/nbt.2925sgRNA expression and homology cassette
Recombinant DNA reagentpL7- CLAG3-tv2This studyModification of pL6 for parasite transfection
Recombinant DNA reagentpL7-CLAG3-GFPThis studyModification of pL6 for parasite transfection
Recombinant DNA reagentpL7-CLAG3-tv1This studyModification of pL6 for parasite transfection
Chemical compound, drugDSM1BEI Resources RepositoryCat# MRA-1161
Chemical compound, drugWR99210David Jacobus
Commercial assay or kitAnti-FLAG M2 affinity agarose resinSigma–AldrichCat# A2220
Commercial assay or kit3xFLAG peptideSigma–AldrichCat# F4799
Commercial assay or kitNi-NTA Agarose resinQiagenCat# 30210
Commercial assay or kitZeba microspin desalting columns, 40 kDa MWCOThermo ScientificCat# 87764
OtherGold-coated quartz emitterThis studyNative mass MS study
Commercial assay or kitSYPRO OrangeThermo ScientificCat# S6650Protein stability assay (1:5000)
OtherCarbon film gridsElectron Microscopy SciencesCat# CF200-Cu
OtherQuantifoil Cu 300 mesh gridsElectron Microscopy SciencesCat# Q3310CR1.3
Other4–15% Mini-PROTEAN TGX gelBio-RADCat# 4561086
Software, algorithmThermo Xcalibur Qual BrowserThermo Scientificversions 3.0.63 and 4.2.47
Software, algorithmUniDec• 10.1021/acs.analchem.5b00140; • 10.1007/s13361-018-1951-9versions 3.2 and 4.1http://unidec.chem.ox.ac.uk/
Software, algorithmm/zProteometrics LLC
Software, algorithmEPUThermoFischer
Software, algorithmLatitudeGatan Inc
Software, algorithmRELION 2.0; RELION 3.0• 10.1016/j.jsb.2012.09.006https://www3.mrc-lmb.cam.ac.uk/relion
Software, algorithmMotionCor2• 10.1038/nmeth.4193https://emcore.ucsf.edu/ucsf-software
Software, algorithmGctf• 10.1016/j.jsb.2015.11.003https://www2.mrc-lmb.cam.ac.uk/research/locally-developed-software/zhang-software/
Software, algorithmUCSF Chimera• 10.1002/jcc.20084https://www.cgl.ucsf.edu/chimera/
Software, algorithmCoot• 10.1107/S0907444910007493https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/
Software, algorithmPHENIX assign_sequence• 10.1107/S2059798319011471https://www.phenix-online.org/
Software, algorithmPHENIX real space refine• 10.1107/S2059798318006551https://www.phenix-online.org/documentation/reference/real_space_refine.html
Software, algorithmJPred• 10.1093/nar/gkv332http://www.compbio.dundee.ac.uk/jpred/
Software, algorithmelNémo server• 10.1093/nar/gkh368http://www.sciences.univ-nantes.fr/elnemo/
Software, databasePlasmoDB• 10.1093/nar/gkn814https://plasmodb.org/plasmo/
Software, algorithmMAFFT server• 10.1093/bib/bbx108https://mafft.cbrc.jp/alignment/server/
Software, algorithmMEGA X• 10.1093/molbev/msy096;10.1093/molbev/msz312https://www.megasoftware.net/
Software, algorithmConSurf server• 10.1093/nar/gkw408https://consurf.tau.ac.il/
Software, algorithmNCBI Protein BLAST• 10.1093/nar/25.17.3389https://blast.ncbi.nlm.nih.gov/Blast.cgi
Software, algorithmClustal Omega• 10.1093/nar/gkz268https://www.ebi.ac.uk/Tools/msa/clustalo/
Software, algorithmPymolSchrödinger, LLChttps://pymol.org/2/
Software, algorithmDali server• 10.1093/bioinformatics/btz536http://ekhidna2.biocenter.helsinki.fi/dali/
Software, algorithmImageJ• 10.1186/s12859-017-1934-zhttps://imagej.nih.gov/ij/index.html
Software, algorithmSigmaPlot 10.0Systat
Software, algorithmPrism 8.2GraphPad
Parasite culture

P. falciparum laboratory strains were grown in O+ human erythrocytes (Interstate Blood Bank) using standard methods and maintained at 5% hematocrit under 5% O2, 5% CO2, 90% N2 at 37°C.

Endogenous tagging

CRISPR-Cas9 gene editing was used to produce engineered P. falciparum lines using the KC5 laboratory clone carrying a single clag3h gene to avoid epigenetic switching (Gupta et al., 2018). Transfections were performed by electroporation of pUF1-Cas9 and modified pL6 plasmids for homologous replacement of the genomic site as described (Ghorbal et al., 2014); 1.5 µM DSM1 and 2 nM WR99210 were used to select for integrants, which were detected by PCR. All experiments were performed with limiting dilution clones that were confirmed with DNA sequencing.

Primary protein purifications used the edited CLAG3-tv2 clone, in which a C-terminal multiple affinity tag consisting of His10-FLAG-thrombin-TEV-HA-twinstrept-BC2 nanobody binding site was appended to an otherwise unmodified CLAG3h. The CLAG3-GFP incorporates a C-terminal His8-monomeric GFP-FLAG-Twin strept tag on CLAG3h. The CLAG3-tv1+RhopH2-mV strain contains a C-terminal 3xFLAG-3xHA-His8-Strept II tag on CLAG3 and a monomeric Venus tag at the RhopH2 C-terminus; this parasite was produced by sequential CRISPR-Cas9 editing of the two genomic loci and used for negative stain imaging of mVenus-tagged RhopH2.

Protein purification

Up to 1 mL of enriched schizont-stage parasites were harvested by the percoll–sorbitol method and frozen in liquid nitrogen at 20% v/v in 200 mM NaCl, 10 mM Tris, pH 7.5 with 1 mM phenylmethylsulfonyl fluoride (PMSF). Frozen parasites were thawed at room temperature, and insoluble debris was pelleted at 20,000 × g for 10 min at 4°C. NaCl was added to 500 mM before overnight incubation of the clarified lysate with anti-FLAG M2 affinity agarose resin (Sigma–Aldrich) at 4°C with gentle agitation. The resin was subsequently washed with 1–5 mL of 10 mM Tris, pH 7.5 and 500 mM NaCl before elution in 10 mM Tris, pH 7.5, 200 mM NaCl and 0.15 mg/mL 3xFLAG peptide. The eluate was concentrated for native mass spectrometry and cryo-EM studies via a second affinity purification on Ni-NTA agarose resin (Qiagen) and small volume elution in 200 mM NaCl, 300 mM imidazole, 10 mM Tris, pH 7.5. After overnight dialysis to remove imidazole, purified RhopH complex was further concentrated by ultracentrifugation at 150,000 × g for 1 hr, yielding 0.8–2 mg/mL protein in 30 µL.

Native mass spectrometry analysis

Purified RhopH complex was buffer-exchanged into native mass spectrometry (MS) solution (200 mM ammonium acetate, pH 7.5, 0.01% Tween-20) using Zeba microspin desalting columns with a 40 kDa cut-off (ThermoScientific; Olinares et al., 2016; Olinares and Chait, 2020). Buffer-exchanged sample (3 µL) was loaded into a locally prepared gold-coated quartz emitter and electrosprayed into an Exactive Plus EMR instrument (ThermoFisher Scientific) with a modified static nanospray source (Olinares and Chait, 2020). The MS parameters used include spray voltage, 1.2–1.3 kV; capillary temperature, 150–250°C; in-source dissociation, 10 V; S-lens RF level, 200; resolving power, 17,500 at m/z of 200; AGC target, 1 × 106; maximum injection time, 200 ms; number of microscans, 5; injection flatapole, 8 V; interflatapole, 4 V; bent flatapole, 4 V; high-energy collision dissociation, 200 V; ultrahigh vacuum pressure, 7–8 × 10−10 mbar; total number of scans, ≥100. Mass calibration in positive extended mass range (EMR) mode was performed using cesium iodide.

The acquired MS spectra were visualized using Thermo Xcalibur Qual Browser (versions 3.0.63 and 4.2.47). Spectra deconvolution was performed either manually or using the software UniDec versions 3.2 and 4.1 (Marty et al., 2015; Reid et al., 2019). The resulting deconvolved spectrum from UniDec was plotted using the m/z software (Proteometrics LLC). Experimental masses were reported as the mean ± SD across all calculated mass values within the observed charge state series. Mass accuracies were calculated as the % difference between the measured and expected masses relative to the expected mass.

Protein thermostability

Thermal denaturation of the RhopH complex was evaluated with two methods. ThermoFluor assays were performed with 20 µL of 0.4 mg/mL freeze–thaw extracted RhopH complex and a 1× dilution of SYPRO Orange. Fluorescence intensity was continuously monitored during a thermal ramp from 25°C to 95°C in 0.5°C/10 s increments. Raw fluorescence and first-derivative plots were used to assess unfolding. RhopH complex aggregation was also evaluated using sizing with thermal ramp application on Uncle (Unchained Labs) and duplicate samples of 8.9 µL of 0.1 mg/mL RhopH complex. Aggregation was measured by monitoring static light scattering at 266 and 473 nm with a ramp from 20°C to 80°C at a constant rate of 1.0°C/min for 1 hr with measurements at 0.5°C increments.

Negative stain data acquisition

Purified RhopH protein (4.8 μL of a 0.05 mg/mL solution) was applied to carbon film grids (CF200-Cu, Electron Microscopy) and stained with 4.8 μL of 0.75% uranyl formate for 30 s. After drying, grids were loaded onto a ThermoFischer Tecnai 12 electron microscope with a Gatan Ultra Scan camera operating at 120 kV. Images were collected using EPU software (ThermoFischer) at 67,000× magnification for a pixel size of 1.77 Å. The datasets consisted of between 69 and 142 micrographs (culture-media RhopH, 69 micrographs; complexes containing RhopH2-mV, 109; CLAG3-tv1, 124; CLAG3-GFP, 142).

Negative stain image processing

All negative stain image processing was performed using RELION 2.0 (Scheres, 2012). Micrographs were processed without CTF correction. Initial auto-picking was performed using a Gaussian blob. Well-behaved classes from 2D classification of Gaussian blob-picked particles were used for template-based auto-picking. Further 2D classification was performed to clean the particle set. For datasets with GFP derivative tagging, additional density for the bulky epitope was visible is several 2D classes. For freeze–thawed solubilized and spend-media RhopH, an initial model was generated and used for 3D auto-refinement in RELION. Three-dimensional models represent views in Chimera (Pettersen et al., 2004).

Cryo-EM data acquisition

2.5 μL of 0.8 mg/mL RhopH was applied to glow-discharged Quantifoil Cu 300 mesh grids (1.2/1.3), blotted for 3 s, and plunge frozen in liquid ethane cooled by liquid nitrogen using a Vitrobot plunge freezing instrument (FEI/ThermoFisher). The blotting chamber was maintained at 20°C and 100% humidity. One thousand three hundred and ten micrographs were collected on a Titan Krios (ThermoFisher) transmission electron microscope operated at 300 kV. Images were recorded on a K2 Summit camera (Gatan Inc) operated in super-resolution counting mode and a physical pixel size of 0.84 Å. The detector was placed at the end of a GIF Quantum energy filter (Gatan Inc), operated in zero-energy-loss mode with a slit width of 20 eV. Each image was fractionated into 58 frames with a frame exposure of 0.4 s and a dose rate of 3 e–/Å2/s, giving a total accumulated dose of 70 e–/Å2 over the 23.2 s exposure. All data was collected using the Latitude S software (Gatan Inc).

Cryo-EM image processing

All cryo-EM image processing was performed in RELION 3.0. Movies were motion corrected and dose-weighted using MotionCor2 (Zheng et al., 2017). Contrast transfer function (CTF) parameters were determined using the Gctf (Zhang, 2016) wrapper in RELION. Initial particle picking was performed with the Laplacian-of-Gaussian (LoG) picker in RELION. Subsequent 2D classes from the LoG-picked particles were used for template-based auto-picking performed in RELION resulting in 311,390 particles. After two rounds of 2D classification, the initial collection was cleaned to 214,233 particles and used to generate an initial 3D model. Three-dimensional classification using five classes with regularization parameter T = 4 resulted in one well-resolved class of 68,216 particles. Three-dimensional auto-refinement of these particles resulted in a 3.26 Å map. Two rounds of particle polishing and one round of CTF refinement further improved the resolution to 2.92 Å. Although the large lobe was well-resolved and permitted de novo model building, the small lobe and C-terminal bundle of CLAG3 were resolved to lower resolution inhibiting interpretation. Further 3D classification did not improve small subunit interpretability. To better resolve RhopH2 and CLAG3 C-terminal domain, multibody refinement was performed (Nakane et al., 2018). Multibody refinement using masked region 1 of the large subunit and masked region two as the small subunit and the bridge between the large and small subunit resulted in better EM density for mobile elements of the small subunit although a lower overall resolution for the second masked region. Multibody analysis also yielded the top components of motion.

Model building and refinement

Model building was performed in Coot (Emsley et al., 2010). EM density maps were generated in RELION by post-processing with a constant B factor or locally sharpened regions of the maps in Local Resolution. Initially, a poly-alanine model was built for well-ordered regions of the RhopH complex in Coot. The sequence registry was determined by a combination of manual examination of side-chain density and the PHENIX assign_sequence program (Liebschner et al., 2019), which predicts sequence registry based on side-chain density. Regions of the map with low resolution were built through a combination EM density interpretation and secondary structure prediction performed in JPred (Drozdetskiy et al., 2015). Real space refinement with secondary structure restraints was performed in PHENIX real space refine (Afonine et al., 2018). Structural figures were generated in PyMOL 2.1.0 (Schrödinger) or Chimera. Prediction of motion in the final model was performed using the elNémo server (Suhre and Sanejouand, 2004).

Phylogenetic analysis

CLAG DNA sequences were downloaded from PlasmoDB (http://PlasmoDB.org) and aligned using the MAFFT server (Katoh et al., 2019) with default parameters. Sequences shorter than 2000 nucleotides in length were removed to maximize sequence overlap. The multiple-sequence alignment was corrected manually to preserve the reading frame. Phylogenetic analysis of the remaining 147 sequences was performed using the MEGA X software (Kumar et al., 2018; Stecher et al., 2020). A phylogenetic tree was inferred using the neighbor-joining method (Saitou and Nei, 1987) based on pairwise distances computed using the maximum composite likelihood method (Tamura et al., 2004), with the rate variation among sites modeled with a gamma distribution (shape parameter = 1). To assess how well the data supported the groups in the tree, 250 bootstrap replicates were performed (Felsenstein, 1985).

Conservation analysis

The ConSurf server (https://consurf.tau.ac.il/; Ashkenazy et al., 2016) was used to generate per-residue conservation scores and map conservation values on the 3D RhopH complex structure. Non-redundant sequences of RhopH subunits from Plasmodium spp. were identified through the use of PlasmoDB and NCBI Protein BLAST. The sequences were aligned using Clustal Omega. ConSurf was then used to evaluate evolutionary conservation of amino acid residues; the resulting conservation scores were used for color-coding residues in PyMOL.

Structural similarity searches

The Dali server (http://ekhidna2.biocenter.helsinki.fi/dali/; Holm, 2019) was used to search for proteins with 3D structures like that of the RhopH complex. Exhaustive PDB database searches revealed significant matches to specific domains from individual RhopH subunits, as defined by Dali Z-scores ≥ 3.0. PyMOL alignments of RhopH domains and PDB structures of corresponding hits were used to evaluate biological significance.

Membrane fractionation

Synchronization for stage-dependent membrane fractionation assays utilized two 5% sorbitol treatments ~6 hr apart. Ring-stage infected cells were harvested immediately without enrichment. Trophozoite- and schizont-stage-infected cells were then harvested 18 hr and 40 hr after sorbitol treatment, respectively, and enriched through the percoll–sorbitol method. Cells infected with 13F10 growth with or without TMP were harvested without enrichment as these cells lack PSAC activity (Beck et al., 2014; Ito et al., 2017).

Freed merozoite studies were performed with 3D7 parasites using synchronous schizonts enriched using the percoll–sorbitol method. Purified schizonts were cultured with 25 µM E64D at 7.5 × 107 cells/mL and closely monitored for 4–5 hr for the development of segmenters containing fully formed merozoites. Cells were then washed, adjusted to 2.5 × 107 cells/mL in complete media, and allowed to recover at 37°C for 15 min. Freed merozoites (2.5 × 108 cells/mL) were obtained by sequential passage through two 1.2 µm syringe filters to rupture the mature segmenters. A hemocytometer was used to confirm that merozoites were free of contaminating intact erythrocytes before pelleting (4500 × g, 5 min) and freezing along with matched intact schizonts.

Fractionation studies were performed using matched cell pellets resuspended in lysis buffer (7.5 mM Na2HPO4, 1 mM EDTA, pH 7.5) at 3.5% hematocrit; this cell lysate corresponded to the ‘total’ input. Cellular debris and membranes were pelleted by ultracentrifugation at 100,000 × g for 1 hr at 4°C. The supernatant was kept as the ‘soluble’ fraction. Membranes were resuspended and incubated in 200 µL of 100 mM Na2CO3, pH 11 at 4°C for 30 min before ultracentrifugation (100,000 × g, 1 hr, 4°C) to separate peripheral from integral membrane proteins. Samples were neutralized with 1 M HCl and solubilized in a modified Laemmli buffer with a final 6% sodium dodecyl sulfate (SDS) concentration.

Protease susceptibility experiments used percoll–sorbitol-enriched cells. Infected cells were treated with Pronase E in phosphate-buffered saline (PBS) supplemented with 0.6 mM CaCl2 and 1 mM MgCl2 for up to 1 hr at 37°C. They were then extensively washed in PBS with 1 mM PMSF prior to membrane fractionation.

Immunoblotting

Samples were prepared in a modified Laemmli buffer with a final 6% SDS concentration. Proteins were separated on a 4–15% Mini-PROTEAN TGX gel (Bio-RAD) and transferred to nitrocellulose. After blocking, antibodies against CLAG3 (Nguitragool et al., 2011), Band3 (Santa Cruz), HA epitope tag (Sigma–Aldrich), EXP2 (European Malaria Reagent Repository), or aldolase (Abcam) were applied and visualized as described (Ito et al., 2017). Band intensities were quantified using ImageJ and analyzed in Prism (GraphPad).

Statistical analysis

Statistical significance for numerical data was calculated by unpaired Student’s t-test or one-way ANOVA. Significance was accepted at p < 0.05 or indicated values.

Acknowledgements

We thank Arasu Balasubramaniam for help with ThermoFluor assays, Anthony Armstrong for guidance on elNémo normal mode analysis, Ryan Kissinger and Anita Mora for artwork, Daniel Goldberg for the 13F10 clone, and David Jacobus for WR99210. DSM1 (MRA-1161) was obtained through MR4 as part of the BEI Resources Repository, NIAID, NIH. This work utilized the computational resources of the NIH HPC Biowulf cluster (http://hpc.nih.gov).

Additional informationCompeting interests

Reviewing editor, eLife

No competing interests declared

Author contributions

Conceptualization, Formal analysis, Investigation, Writing - original draft, Writing - review and editing

Formal analysis, Investigation, Writing - review and editing

Formal analysis, Investigation, Writing - review and editing

Formal analysis, Investigation, Writing - review and editing

Investigation, Writing - review and editing

Supervision, Methodology, Writing - review and editing

Formal analysis, Investigation, Writing - review and editing

Formal analysis, Supervision, Writing - review and editing

Formal analysis, Supervision, Writing - review and editing

Formal analysis, Investigation, Writing - review and editing

Conceptualization, Resources, Formal analysis, Supervision, Writing - review and editing

Conceptualization, Formal analysis, Supervision, Writing - original draft, Project administration, Writing - review and editing

Additional filesData availability

All data generated or analysed during this study are included in the manuscript and supporting files. Cryo-EM maps have been deposited in EMDB and PDB.

The following datasets were generated:

SchureckMADarlingJEMerkASubramaniamSDesaiSA2021Plasmodium falciparum RhopH complex in soluble formRCSB Protein Data Bank7KIY

SchureckMADarlingJEMerkASubramaniamSDesaiSA2021Plasmodium falciparum RhopH complex in soluble formEMDataResourceEMD-22890

ReferencesAfonine PVPoon BKRead RJSobolev OVTerwilliger TCUrzhumtsev AAdams PD2018Real-space refinement in PHENIX for cryo-EM and crystallographyActa Crystallographica Section D Structural Biology7453154410.1107/S2059798318006551Ahmad MManzella-Lapeira JSaggu GIto DBrzostowski JADesai SA2020Live-Cell FRET reveals that malaria nutrient channel proteins CLAG3 and RhopH2 remain associated throughout their tortuous traffickingmBio11e0135410.1128/mBio.01354-2032900800Ashkenazy HAbadi SMartz EChay OMayrose IPupko TBen-Tal N2016ConSurf 2016: an improved methodology to estimate and visualize evolutionary conservation in macromoleculesNucleic Acids Research44W344W35010.1093/nar/gkw40827166375Bai XCYan CYang GLu PMa DSun LZhou RScheres SHWShi Y2015An atomic structure of human γ-secretaseNature52521221710.1038/nature1489226280335Bannister LHMitchell GHButcher GADennis ED1986Lamellar membranes associated with rhoptries in Erythrocytic merozoites of Plasmodium knowlesi: a clue to the mechanism of invasionParasitology9229130310.1017/S00311820000640642423944Beck JRMuralidharan VOksman AGoldberg DE2014PTEX component HSP101 mediates export of diverse malaria effectors into host erythrocytesNature51159259510.1038/nature1357425043010Burkinshaw BJSouza SAStrynadka NCJ2015Structural analysis of SepL, an enteropathogenic Escherichia coli type III secretion-system gatekeeper proteinActa Crystallographica Section F Structural Biology Communications711300130810.1107/S2053230X15016064Carter JR1973Role of sulfhydryl groups in erythrocyte membrane structureBiochemistry1217117610.1021/bi00725a0284683480Ciferri CLander GCMaiolica AHerzog FAebersold RNogales E2012Molecular architecture of human polycomb repressive complex 2eLife1e0000510.7554/eLife.0000523110252Cobbold SASantos JMOchoa APerlman DHLlinás M2016Proteome-wide analysis reveals widespread lysine acetylation of major protein complexes in the malaria parasiteScientific Reports6srep1972210.1038/srep19722Cohn JVAlkhalil AWagner MARajapandi TDesai SA2003Extracellular lysines on the plasmodial surface anion channel involved in na+ exclusionMolecular and Biochemical Parasitology132273410.1016/j.molbiopara.2003.08.00114563534Corey RAAhdash ZShah APyle EAllen WJFessl TLovett JEPolitis ACollinson I2019ATP-induced asymmetric pre-protein folding as a driver of protein translocation through the sec machineryeLife8e4180310.7554/eLife.4180330601115Cortés ACarret CKaneko OYim Lim BYIvens AHolder AA2007Epigenetic silencing of Plasmodium falciparum genes linked to erythrocyte invasionPLOS Pathogens3e10710.1371/journal.ppat.003010717676953Counihan NAChisholm SABullen HESrivastava ASanders PRJonsdottir TKWeiss GEGhosh SCrabb BSCreek DJGilson PRde Koning-Ward TF2017Plasmodium falciparum parasites deploy RhopH2 into the host erythrocyte to obtain nutrients, grow and replicateeLife6e2321710.7554/eLife.2321728252383Cowman AFBerry DBaum J2012The cellular and molecular basis for malaria parasite invasion of the human red blood cellJournal of Cell Biology19896197110.1083/jcb.201206112Dal Peraro Mvan der Goot FG2016Pore-forming toxins: ancient, but never really out of fashionNature Reviews. Microbiology14779210.1038/nrmicro.2015.326639780de Koning-Ward TFGilson PRBoddey JARug MSmith BJPapenfuss ATSanders PRLundie RJMaier AGCowman AFCrabb BS2009A newly discovered protein export machine in malaria parasitesNature45994594910.1038/nature0810419536257Denks KVogt ASachelaru IPetriman NAKudva RKoch HG2014The sec translocon mediated protein transport in prokaryotes and eukaryotesMolecular Membrane Biology31588410.3109/09687688.2014.90745524762201Desai SABezrukov SMZimmerberg J2000A voltage-dependent channel involved in nutrient uptake by red blood cells infected with the malaria parasiteNature4061001100510.1038/3502300010984055Desai SA2014Why do malaria parasites increase host erythrocyte permeability?Trends in Parasitology3015115910.1016/j.pt.2014.01.00324507014Dluzewski ARZicha DDunn GAGratzer WB1995Origins of the parasitophorous vacuole membrane of the malaria parasite: surface area of the parasitized red cellEuropean Journal of Cell Biology684464498690024Doury JCBonnefoy SRoger NDubremetz JFMercereau-Puijalon O1994Analysis of the high molecular weight rhoptry complex of Plasmodium falciparum using monoclonal antibodiesParasitology10826928010.1017/S00311820000761137517517Drozdetskiy ACole CProcter JBarton GJ2015JPred4: a protein secondary structure prediction serverNucleic Acids Research43W389W39410.1093/nar/gkv33225883141Egea PFStroud RM2010Lateral opening of a translocon upon entry of protein suggests the mechanism of insertion into membranesPNAS107171821718710.1073/pnas.101255610720855604Ekka RGupta ABhatnagar SMalhotra PSharma P2020Phosphorylation of rhoptry protein RhopH3 is critical for host cell invasion by the malaria parasitemBio11e00166-2010.1128/mBio.00166-2033024030Emsley PLohkamp BScott WGCowtan K2010Features and development of cootActa Crystallographica. Section D, Biological Crystallography6648650110.1107/S090744491000749320383002Felsenstein J1985Confidence limits on phylogenies: an approach using the bootstrapEvolution3978379110.1111/j.1558-5646.1985.tb00420.xFinucane DMBossy-Wetzel EWaterhouse NJCotter TGGreen DR1999Bax-induced Caspase Activation and Apoptosis via Cytochrome c Release from Mitochondria Is Inhibitable by Bcl-xLJournal of Biological Chemistry2742225223310.1074/jbc.274.4.2225Galen SCBorner JMartinsen ESSchaer JAustin CCWest CJPerkins SL2018The polyphyly of plasmodium: comprehensive phylogenetic analyses of the malaria parasites (order Haemosporida) reveal widespread taxonomic conflictRoyal Society Open Science517178010.1098/rsos.17178029892372Ghorbal MGorman MMacpherson CRMartins RMScherf ALopez-Rubio JJ2014Genome editing in the human malaria parasite Plasmodium falciparum using the CRISPR-Cas9 systemNature Biotechnology3281982110.1038/nbt.292524880488Goel SValiyaveettil MAchur RNGoyal AMattei DSalanti ATrenholme KRGardiner DLGowda DC2010Dual stage synthesis and crucial role of cytoadherence-linked asexual gene 9 in the surface expression of malaria parasite var proteinsPNAS107166431664810.1073/pnas.100256810720823248Goldberg DECowman AF2010Moving in and renovating: exporting proteins from plasmodium into host erythrocytesNature Reviews Microbiology861762110.1038/nrmicro242020706280Gupta AThiruvengadam GDesai SA2015The conserved clag multigene family of malaria parasites: essential roles in host-pathogen interactionDrug Resistance Updates18475410.1016/j.drup.2014.10.00425467627Gupta ABalabaskaran-Nina PNguitragool WSaggu GSSchureck MADesai SA2018CLAG3 Self-Associates in malaria parasites and quantitatively determines nutrient uptake channels at the host membranemBio9e02293-1710.1128/mBio.02293-1729739907Gupta ABokhari AABPillai ADCrater AKGezelle JSaggu GNasamu ASGanesan SMNiles JCDesai SA2020Complex nutrient channel phenotypes despite mendelian inheritance in a Plasmodium falciparum genetic crossPLOS Pathogens16e100836310.1371/journal.ppat.100836332069335Ho CMBeck JRLai MCui YGoldberg DEEgea PFZhou ZH2018Malaria parasite translocon structure and mechanism of effector exportNature561707510.1038/s41586-018-0469-430150771Holm L2019Benchmarking fold detection by DaliLite v.5Bioinformatics355326532710.1093/bioinformatics/btz53631263867Iriko HKaneko OOtsuki HTsuboi TSu XZTanabe KTorii M2008Diversity and evolution of the rhoph1/clag multigene family of Plasmodium falciparumMolecular and Biochemical Parasitology158112110.1016/j.molbiopara.2007.11.00418155305Ito DSchureck MADesai SA2017An essential dual-function complex mediates erythrocyte invasion and channel-mediated nutrient uptake in malaria parasiteseLife6e2348510.7554/eLife.2348528221136Käll LKrogh ASonnhammer EL2004A combined transmembrane topology and signal peptide prediction methodJournal of Molecular Biology3381027103610.1016/j.jmb.2004.03.01615111065Kaneko OTsuboi TLing ITHowell SShirano MTachibana MCao YMHolder AATorii M2001The high molecular mass rhoptry protein, RhopH1, is encoded by members of the clag multigene family in Plasmodium falciparum and Plasmodium yoeliiMolecular and Biochemical Parasitology11822323110.1016/S0166-6851(01)00391-711738712Kaneko O2007Erythrocyte invasion: vocabulary and grammar of the plasmodium rhoptryParasitology International5625526210.1016/j.parint.2007.05.00317596999Katoh KRozewicki JYamada KD2019MAFFT online service: multiple sequence alignment, interactive sequence choice and visualizationBriefings in Bioinformatics201160116610.1093/bib/bbx10828968734Kumar SStecher GLi MKnyaz CTamura K2018MEGA X: molecular evolutionary genetics analysis across computing platformsMolecular Biology and Evolution351547154910.1093/molbev/msy09629722887Liebschner DAfonine PVBaker MLBunkóczi GChen VBCroll TIHintze BHung L-WJain SMcCoy AJMoriarty NWOeffner RDPoon BKPrisant MGRead RJRichardson JSRichardson DCSammito MDSobolev OVStockwell DHTerwilliger TCUrzhumtsev AGVideau LLWilliams CJAdams PD2019Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in PhenixActa Crystallographica Section D Structural Biology7586187710.1107/S2059798319011471Ling ITFlorens LDluzewski ARKaneko OGrainger MYim Lim BYTsuboi THopkins JMJohnson JRTorii MBannister LHYates JRHolder AAMattei D2004The Plasmodium falciparum clag9 gene encodes a rhoptry protein that is transferred to the host erythrocyte upon invasionMolecular Microbiology5210711810.1111/j.1365-2958.2003.03969.x15049814Lisk GPain MGluzman IYKambhampati SFuruya TSu XFay MPGoldberg DEDesai SA2008Changes in the plasmodial surface anion channel reduce leupeptin uptake and can confer drug resistance in Plasmodium falciparum-Infected ErythrocytesAntimicrobial Agents and Chemotherapy522346235410.1128/AAC.00057-08Marty MTBaldwin AJMarklund EGHochberg GKBenesch JLRobinson CV2015Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensemblesAnalytical Chemistry874370437610.1021/acs.analchem.5b0014025799115Matthews KMKalanon Mde Koning-Ward TF2019Uncoupling the threading and unfoldase actions of Plasmodium HSP101 Reveals Differences in Export between Soluble and Insoluble ProteinsmBio10e01106-1910.1128/mBio.01106-1931164473Mira-Martínez Svan Schuppen EAmambua-Ngwa ABottieau EAffara MVan Esbroeck MVlieghe EGuetens PRovira-Graells NGómez-Pérez GPAlonso PLD'Alessandro URosanas-Urgell ACortés A2017Expression of the Plasmodium falciparum Clonally Variant clag3 Genes in Human InfectionsThe Journal of Infectious Diseases21593894510.1093/infdis/jix05328419281Mira-Martínez SPickford AKRovira-Graells NGuetens PTintó-Font ECortés ARosanas-Urgell A2019Identification of Antimalarial Compounds That Require CLAG3 for Their Uptake by Plasmodium falciparum -Infected ErythrocytesAntimicrobial Agents and Chemotherapy63e00052-1910.1128/AAC.00052-19Nacer ARoux EPomel SScheidig-Benatar CSakamoto HLafont FScherf AMattei D2011Clag9 is not essential for PfEMP1 surface expression in non-cytoadherent Plasmodium falciparum parasites with a chromosome 9 deletionPLOS ONE6e2903910.1371/journal.pone.002903922205992Nakane TKimanius DLindahl EScheres SH2018Characterisation of molecular motions in cryo-EM single-particle data by multi-body refinement in RELIONeLife7e3686110.7554/eLife.3686129856314Nguitragool WBokhari AAPillai ADRayavara KSharma PTurpin BAravind LDesai SA2011Malaria parasite clag3 genes determine channel-mediated nutrient uptake by infected red blood cellsCell14566567710.1016/j.cell.2011.05.00221620134Nguitragool WRayavara KDesai SA2014Proteolysis at a specific extracellular residue implicates integral membrane CLAG3 in malaria parasite nutrient channelsPLOS ONE9e9375910.1371/journal.pone.009375924699906Olinares PDDunn ADPadovan JCFernandez-Martinez JRout MPChait BT2016A robust workflow for native mass spectrometric analysis of Affinity-Isolated endogenous protein assembliesAnalytical Chemistry882799280710.1021/acs.analchem.5b0447726849307Olinares PDBChait BT2020Native mass spectrometry analysis of Affinity-Captured endogenous yeast RNA exosome complexesMethods in Molecular Biology206235738210.1007/978-1-4939-9822-7_1731768985Otto TDBöhme USanders MReid ABruske EIDuffy CWBull PCPearson RDAbdi ADimonte SStewart LBCampino SKekre MHamilton WLClaessens AVolkman SKNdiaye DAmambua-Ngwa ADiakite MFairhurst RMConway DJFranck MNewbold CIBerriman M2018Long read assemblies of geographically dispersed Plasmodium falciparum isolates reveal highly structured subtelomeresWellcome Open Research35210.12688/wellcomeopenres.14571.129862326Pease BNHuttlin ELJedrychowski MPTalevich EHarmon JDillman TKannan NDoerig CChakrabarti RGygi SPChakrabarti D2013Global analysis of protein expression and phosphorylation of three stages of Plasmodium falciparum intraerythrocytic developmentJournal of Proteome Research124028404510.1021/pr400394g23914800Pettersen EFGoddard TDHuang CCCouch GSGreenblatt DMMeng ECFerrin TE2004UCSF chimera--a visualization system for exploratory research and analysisJournal of Computational Chemistry251605161210.1002/jcc.2008415264254Pillai ADNguitragool WLyko BDolinta KButler MMNguyen STPeet NPBowlin TLDesai SA2012Solute restriction reveals an essential role for clag3-associated channels in malaria parasite nutrient acquisitionMolecular Pharmacology821104111410.1124/mol.112.08122422949525Proellocks NICoppel RLMohandas NCooke BM2016Malaria parasite proteins and their role in alteration of the structure and function of red blood cellsAdvances in Parasitology9118610.1016/bs.apar.2015.09.00227015947Reid DJDiesing JMMiller MAPerry SMWales JAMontfort WRMarty MT2019MetaUniDec: high-throughput deconvolution of native mass spectraJournal of the American Society for Mass Spectrometry3011812710.1007/s13361-018-1951-929667162Rovira-Graells NCrowley VMBancells CMira-Martínez SRibas de Pouplana LCortés A2015Deciphering the principles that govern mutually exclusive expression of Plasmodium falciparum clag3 genesNucleic Acids Research438243825710.1093/nar/gkv730Saitou NNei M1987The neighbor-joining method: a new method for reconstructing phylogenetic treesMolecular Biology and Evolution440642510.1093/oxfordjournals.molbev.a0404543447015Scheres SH2012RELION: implementation of a bayesian approach to cryo-EM structure determinationJournal of Structural Biology18051953010.1016/j.jsb.2012.09.00623000701Sharma PWollenberg KSellers MZainabadi KGalinsky KMoss ENguitragool WNeafsey DDesai SA2013An epigenetic antimalarial resistance mechanism involving parasite genes linked to nutrient uptakeJournal of Biological Chemistry288194291944010.1074/jbc.M113.468371Sharma PRayavara KIto DBasore KDesai SA2015A CLAG3 mutation in an amphipathic transmembrane domain alters malaria parasite nutrient channels and confers leupeptin resistanceInfection and Immunity832566257410.1128/IAI.02966-1425870226Stecher GTamura KKumar S2020Molecular evolutionary genetics analysis (MEGA) for macOSMolecular Biology and Evolution371237123910.1093/molbev/msz31231904846Suhre KSanejouand YH2004ElNemo: a normal mode web server for protein movement analysis and the generation of templates for molecular replacementNucleic Acids Research32W610W61410.1093/nar/gkh36815215461Tamura KNei MKumar S2004Prospects for inferring very large phylogenies by using the neighbor-joining methodPNAS101110301103510.1073/pnas.040420610115258291Trenholme KRGardiner DLHolt DCThomas EACowman AFKemp DJ2000clag9: a cytoadherence gene in Plasmodium falciparum essential for binding of parasitized erythrocytes to CD36PNAS974029403310.1073/pnas.04056119710737759Vincensini LFall GBerry LBlisnick TBraun Breton C2008The RhopH complex is transferred to the host cell cytoplasm following red blood cell invasion by Plasmodium falciparumMolecular and Biochemical Parasitology160818910.1016/j.molbiopara.2008.04.00218508137Wahlgren MGoel SAkhouri RR2017Variant surface antigens of Plasmodium falciparum and their roles in severe malariaNature Reviews Microbiology1547949110.1038/nrmicro.2017.4728603279Wong WHuang RMenant SHong CSandow JJBirkinshaw RWHealer JHodder ANKanjee UTonkin CJHeckmann DSoroka VSøgaard TMMJørgensen TDuraisingh MTCzabotar PEde Jongh WATham WHWebb AIYu ZCowman AF2019Structure of Plasmodium falciparum Rh5-CyRPA-Ripr invasion complexNature56511812110.1038/s41586-018-0779-630542156Wright KEHjerrild KABartlett JDouglas ADJin JBrown REIllingworth JJAshfield RClemmensen SBde Jongh WADraper SJHiggins MK2014Structure of malaria invasion protein RH5 with erythrocyte basigin and blocking antibodiesNature51542743010.1038/nature1371525132548Zhang K2016Gctf: real-time CTF determination and correctionJournal of Structural Biology19311210.1016/j.jsb.2015.11.00326592709Zheng SQPalovcak EArmache JPVerba KACheng YAgard DA2017MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopyNature Methods1433133210.1038/nmeth.41932825046610.7554/eLife.65282.sa1Decision letterSilvieOlivierReviewing EditorSorbonne Université, UPMC Univ Paris 06, INSERM, CNRSFrance

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study provides novel insights into the structure and trafficking of an essential protein complex of the malaria parasite. Plasmodium falciparum relies on the RhopH protein complex for invasion of erythrocytes and for nutrient uptake. Here, Schureck et al. use cryo-electron microscopy to determine the first 3D structure of the RhopH complex purified from the parasite. They provide compelling evidence that the complex is synthesized as a soluble form that is exported to the infected host cell and subsequently inserted into the erythrocyte membrane.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Malaria parasites use a soluble RhopH complex for erythrocyte invasion and an integral form for nutrient uptake" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that in its current form, your work will not be considered further for publication in eLife.

As you will see, the four reviewers are broadly supportive of this work, which addresses an important topic and provides the first atomic resolution cryo-EM structure of endogenous P. falciparum RhopH complex. However, they were especially concerned that the conclusions drawn from the structural and biochemical analysis are not sufficiently supported by the data. Importantly, they raised three main issues that will take some time to address (probably more than two months) and in consequence, we have together opted for a rejection with a strong encouragement for re-submission. If you are able to address these issues we will consider a newly submitted form of this paper that we will treat as a revised manuscript. If you choose this course of action please submit a separate cover letter detailing the changes you have made.

Main points to be addressed:

1) Issues regarding the structure quality should be addressed (Reviewer #3), and some clarification and precision in the structure description is required (Reviewers #2 and #3).

2) The four reviewers had concerns about the role of PTEX translocon in RhopH export. You should either provide direct evidence that RhopH interacts with PTEX, or revise your model.

3) A more thorough biochemical characterization of RhopH complex is required to strengthen your model, including experiments with pronase (Reviewers #2 and #4) or free merozoites (Reviewer #1).

Reviewer #1:

This is an interesting study of Plasmodium falciparum RhopH, a complex of three proteins (CLAG, RhopH2 and RhopH3) that plays a dual role in the malaria parasite, first in merozoites for invasion of erythrocytes, and then at the erythrocyte membrane for nutrient uptake by the parasite. The authors developed a freeze-thaw procedure for gentle detergent-free harvest of a soluble form of the RhopH complex, and used cryo-EM to determine de novo the structure of the complex at high resolution. Interestingly, predicted transmembrane domains are shielded in the soluble complex, suggesting a model where the complex is transferred to the erythrocyte as a soluble form, and then undergoes large-scale conformational changes for insertion in the host erythrocyte membrane. The authors then conducted a biochemical characterization of RhopH and report that the complex is synthesized as a non-integral complex, and that insertion in the erythrocyte membrane requires export through the parasite PTEX protein translocon.

While the structure data do not provide clues on the function of RhopH3 during invasion or on the structure of the integral membrane complex, this study provides a framework for future functional studies, and gives novel and important insights into how the malaria parasite exports a complex of rhoptry proteins into the host cell.

I do not have sufficient expertise to evaluate the quality of the cryo-EM work.

1) Cryo-EM as used to determine the structure of a soluble form of RhopH, which apparently corresponds to a minor fraction of the complex "not associated with membranes". It is not clear what exactly this fraction corresponds to. In a previous report (Ito et al., 2017) the same group indicated that members of the complex are membrane-associated, but extractable in part by carbonate. Why is this fraction not recovered after hypotonic lysis?

It is not clear whether the soluble form corresponds to peripheral membrane proteins.

Have the authors attempted recovering the complex from purified merozoites or spent media for cryoEM? The authors propose that a change in conformation explains the differential behavior of the complex, but they should discuss alternative explanations, such as post-translational modifications or protein interactions, which could alter the association of the complex with the membranes.

2) The biochemical characterization of the different pools of RhopH complex in schizonts is complicated by the coexistence of previously synthesized RhopH (integral form) and newly synthesized protein (peripheral, in the rhoptries). In Figure 4, the authors should include purified merozoites, which should contain only the peripheral/soluble complex, and erythrocyte ghost membranes, which should only contain the integral form. This would provide clearer evidence to support their model.

3) In the PTEX mutant experiment, it is crucial to analyze parasites at the same stage. Beck et al., 2014, showed that in the absence of TMP, 13F10 mutant parasites are blocked at the ring stage, so it is not surprising that the protein extraction profile in Figure 4E is similar to the ring profile in Figure 4C, and does not prove that export through PTEX is required for RhopH membrane insertion. This is a major weakness in the model.

Reviewer #2:

Schureck et al. present some interesting work on the structure of the soluble RhopH complex in Plasmodium falciparum. They use cryo-electron microscopy to show that clag3, RhopH2 and RhopH3 are present in a 1:1:1 stoichiometry, and that known and predicted transmembrane helices are shielded in the soluble complex. This work presents a significant advance in the field and in my opinion warrants publication in eLife.

Reviewer #3:

I see real merit in this study as it describes and reports the biochemical characterization and first atomic resolution cryo-EM structure of the ternary RhopH complex from the malaria parasite purified directly from endogenous source, using a strategy similar to what has been described by Ho and Beck for the PTEX vacuolar translocon. The manuscript however needs clarification and editing to make it more easily readable to a broader public in eLife.

Authors show that a soluble and remarkably stable 1:1:1 RhopH endogenous complex can be isolated directly from infected red blood cells using CRISPR-Cas9 engineered methods. Stoichiometry is determined by native MS combined with the NS-EM analysis of complexes where one subunit is selectively tagged with a “bulky” (relatively speaking compared to each subunit) GFP tag. The NS-EM and biochemical/MS-proteomics data support their claims and are clearly described and illustrated (Figure 1 and its supplementary figures).

An atomic resolution (~2.9Å) cryo-EM structure is described with a particular emphasis on the protein-protein interfaces and their conservation. Based on the nature, conservation and intricacy levels of these CLAG3-RhopH1 and CLAG3-RhopH2 surfaces the authors conclude that the soluble RhopH complex is robust and built to promote invasion and undergo trafficking throughout the cell to reach its final destination the red blood cell membrane.

3 TMDs are predicted, one in each subunit, all buried in this soluble form of the RhopH complex, a pro-PSAC form so to speak. They propose a model where this soluble RhopH converts into a membrane-integrated form in the RBCM with TMDs becoming inserted in the RBCM to form the functional PSAC transporting diverse solutes/nutrients.

Description and illustration of the structure is lacking clarity and precision in some key aspects (TMDs). The main weakness of the present work might reside in the proposed interaction between PTEX and RhopH for the trafficking of the latter throughout the PVM into the infected erythrocyte.

Abstract. Although the Abstract is clear it should introduce the names of the two other rhoptry proteins RhopH2 and RhopH3 proteins simultaneously within that third sentence for the sake of clarity and consistency, as it sounds odd to bring all the attention on CLAG3 (or RhopH1) but not mention the other two subunits so intimately associated with CLAG3.

Second paragraph in Introduction. I would clarify “expansion in humans and vertebrates” as in “Plasmodium spp. infecting humans and other vertebrates such as birds, rodents and primates” as they explain more clearly later in the manuscript.

At the end of the Introduction (and maybe also in the Abstract) I would suggest being a bit more explicit in introducing PTEX (I am aware that referencing the initial study from DeKoning-Ward et al. is done much later in the text) and its function as a vacuolar translocon. In general the Introduction assumes that most readers would be familiar with the exquisitely complicated life cycle of apicomplexan parasites such as Plasmodium (or Toxoplasma): Introducing the existence and formation of a host-derived parasitophorous vacuole (membrane) following invasion of the host red blood cell by the parasite sounds necessary to clarify the general picture. Again this is done quite late in the body of the manuscript.

1) Structure determination

End of Introduction. Results section and in general. Concerning the structure itself presented at 2.9Å resolution as per cryo-EM standards. 2.9 Å is not high resolution: atomic resolution should suffice.

Table 1. Statistics for cryo-EM structural analysis.

In Data collection and processing. Any reason why defocus, FSC threshold and map resolution values are not indicated?

In Refinement. Any reason why model resolution range is not indicated?

Going through the PDB preliminary validation report indicates the presence of a fair amount of clashes in this large structure. RhopH2 in particular seems less well defined.

Upon inspection of the structure using the provided PDB file, I will comment about disulfide bonds and "free" cysteines (reduced). The three proteins contain a fair amount of cysteines and including a total of 7 clearly identified disulfide bonds (overall the 3 subunits) as one could expect in "secreted" proteins facing harsher conditions and as the Authors say "suited for trafficking and transfer to a new host erythrocyte". These are fairly complex patterns of disulfide linkages that did not seem to catch the attention of the Authors.

In CLAG3 (chain A in the PDB file) they are 14 cysteines.

with lone cysteines C779 C1065 C1217 C1431

and three listed disulfide bonds (C409-C415) (C523-C544) and (C1352-C1355)

but pairs C335-?-C363 and C519-?-C547 seem also close enough to form disulfide bonds.

In RhopH2 (chain B in the PDB file) they are 8 cysteines.

with two listed disulfide bonds (C233-C240) and (C791-C851) and lone cysteines C268 and C531; C531 seems to have a problem in that loop? Here again C871 is close enough to C909 to form a disulfide bond?

In RhopH3 (Chain C in the PDB file) they are 14 cysteines.

with lone cysteines C336 and C446 and two listed disulfide bonds: (C157-C231); (C244-C253). That said, pairs (C42//C99) (C262-?-C276) (C421-?-C620) and (C475//C536) are also close enough to form disulfide bonds?

Is this a definitive assignment or are further adjustments needed/expected?

2) TMDs in RhopH complex.

Authors describe precisely the TMD in CLAG3 (α44) as potentially forming the trans-membrane pore in PSAC (where they previously mapped the mutation A1215) but they don't define explicitly the predicted TMDs in RhopH2 and RhopH3 in the manuscript. They are drawn on Figure 3G and Figure 3—figure supplement 1 panels E and F and one can roughly guess their position from the Phobius plots in panels C and D. It took me a while to figure out where to look in the structure.

As pointed earlier, it is a very interesting structure but it is not always well described. Since this is a new structure describing 3 novel proteins from Plasmodium (with no real structural homologues available), a potential membrane protein (complex) it is not like that they are legions. I count two so far, it is worth the effort to give a complete classical diagram for each protein with secondary structure elements on top of the sequence and highlighting some of their most salient features on it (at least the TMDs, the HVR, the interactions surfaces). I understand that we cryo-EM and X-ray people are solving structures of increasing complexity, faster than we can write and maybe think but this should not prevent us from describing them with a minimum of precision so it makes the work of the reviewer and potential readers less painful and more interesting.

Using DALI, similarities with BCl-xL, SepL and γ-secretase APH-1 for TMD portions of RhopH2, H3 and CLAG3, respectively to support their claims.

Figure 3—figure supplement 1B. Despite its legend and knowing what Colicin and Bxl are and look like, I have absolutely no clue what that panel actually shows? It is not clearly illustrated.

3) “Interaction with PTEX” for trafficking across the PVM.

In Figure 4, the Authors demonstrate the existence of two pools but I am not convinced that the Authors provide strong evidence that PTEX is interacting with RhopH for its translocation across the PVM. One could argue that PTEX knockdown results in the lack of export of another protein that is directly or indirectly required for the subsequent proper maturation of RhopH from its soluble to integral form (PSAC) at the RBCM while RhopH follows another route for translocation across the PVM (not through the PTEX, not HSP101-dependent)? That point of the manuscript is rather weak and remains contentious as the Authors mention contradicting reports from Beck et al., 2014, and their own work from Ito et al., 2017. Should RhopH proteins really transit through PTEX then they are PEXEL-negative proteins.

Although the model is attractive, it is hard for me to understand how a ternary complex so tightly assembled with so many disulfide bonds would be threaded through the membrane via PTEX to be refolded and reassembled on the other side (whether in it is in the same soluble form or directly into a membrane-inserted integral form). All three TMDs predicted in each subunit are in the middle of each protein. Integration/insertion in the host cell membrane will then not only require considerable conformational rearrangement but also the translocation of protein sub-domains on opposite sides of the bilayer. While this is discussed where authors mentioned the absence of a lateral gate in the PTEX PVM pore subunit EXP2 for lateral insertion of TM segments, they do not rule out the intervention of other chaperones or other trafficking pathways: like vesicular trafficking maybe via tubular network extensions/Maurer's clefts. That latter route seems more likely.

Nevertheless, the Authors certainly draw 1) an interesting model for the “life cycle” of the RhopH complex and 2) a parallel between their system and the smaller pore forming toxins that also transition from a soluble monomeric form to an oligomeric membrane-inserted pore. However, their system is heterotrimeric thus suggesting a singular increase in complexity compared to the majority of single- or bi-component PFTs. What would be the receptor (usually a lipid in PFTs) triggering insertion inside the host cell membrane?

It is unfortunate that the Authors cannot visualize insertion (if it is spontaneous like PFTs and does not require extra energy) on membranes using their purified endogenous soluble complex. I realize they are limited by the amount of material available.

The Figure 5 conceptualizing their model is nice but somehow confusing.I suggest labeling the PV and/or PVM, the parasite, the RBC for clarification. Although the numbers give a false impression of ordered steps. While 1 and 2 are connected, I would think that here 3 does not follow 2. 3 is de novo endogenous synthesis of RhopH from a parasite dividing inside the infected red blood cell. And while PSAC in 4 might be the result of translocation through PTEX as the Authors propose in this manuscript, it could also be the result of another insertion/secretion path through tubular network maybe. Is there a reason why the inserted form of CLAG3 (PSAC) has two TM spanning segments drawn? Is it confirmed that both Rhop2 and Rhop3 are on the cytoplasmic side of the erythrocyte in the inserted form?

Reviewer #4:

Please note that I can only judge the cell biological and biochemical aspects of the manuscript. The authors investigate the dual soluble/membrane integral nature of the RhopH complex described previously, showing that it is soluble at late stages (in schizonts) and transforms into a membrane-integral form after invasion and that PTEX activity is required for this transition. The work is well done and the data are clear and convincing. The structures show that the TM are folded into interior of the protein, providing a convincing reason why the complex is initially soluble. However, there is no indication how the transition from soluble to membrane-bound may occur may occur and the data implicating the PTEX in transport of the complex are very indirect and open to other interpretations.

1) The authors claim that the PTEX is involved in transport of the RhopH complex to the host cell, but do not provide evidence for this. The experiment in Figure 4E shows that PTEX is required for insertion of the complex into the membrane, but does not prove any direct interaction of the PTEX with the complex. There are several other explanations for the finding that PTEX activity is required for membrane insertion of the complex, such as the requirement of an exported accessory factor.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Malaria parasites use a soluble RhopH complex for erythrocyte invasion and an integral form for nutrient uptake" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Dominique Soldati-Favre as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary:

This is an interesting study of Plasmodium falciparum RhopH, a complex of three proteins (CLAG, RhopH2 and RhopH3) that plays a dual role in the malaria parasite, first in merozoites for invasion of erythrocytes, then at the erythrocyte membrane for nutrient uptake by the parasite. The authors developed a freeze-thaw procedure for gentle detergent-free harvest of a soluble form of the RhopH complex, and used cryo-EM to determine de novo the structure of the complex at high resolution. Interestingly, predicted transmembrane domains are shielded in the soluble complex, suggesting a model where the complex is transferred to the erythrocyte as a soluble form, and then undergoes large-scale conformational changes for insertion in the host erythrocyte membrane. The authors then conducted a thorough biochemical characterization of RhopH and show that the complex is synthesized as a non-integral complex, which is transferred into the infected erythrocyte during merozoite invasion. The complex is then exported to the host cell membrane, a process that possibly depends on the PTEX protein translocon.

Revisions:

The four reviewers agree that this work warrants publication in eLife and that the revised manuscript would be suitable without additional experimental data. However, they also pointed at remaining issues of clarity and presentation that need to be addressed. The individual comments are pasted below.

Reviewer #1:

The authors resubmit an improved version of their manuscript. In particular, they describe the structure of the RhopH complex with more details, and added new data concerning the biochemical characterization of the complex. In particular, analysis of purified merozoites confirms that only the peripheral form of CLAG is found at this stage, corroborating the proposed model for RhopH synthesis and export. The authors partially clarified in the revised text the nature of the peripheral (carbonate-extractable) and integral (carbonate-resistant) protein pools. However, some aspects in the fractionation studies still need to be clarified, and the role of the PTEX translocon should be considered with more caution.

– There is still a lack of clarity regarding the soluble form in the fractionation studies. In the Materials and methods, the authors indicate that after parasite resuspension in lysis buffer (7.5 mM Na2HPO4, 1 mM EDTA, pH 7.5), followed by ultracentrifugation, the supernatant was kept as the "soluble" fraction. CLAG3 is detected in this soluble fraction, as shown in Figure 4A, 4D, 4F. It is unclear what this soluble form corresponds to. Is it the same fraction as indicated as "freeze-thaw" in Figure 4G? If not, why is CLAG3 released in the supernatant upon parasite treatment with the buffer alone? This contradicts the conclusions drawn from Figure 1B, stating that hypotonic lysis is not sufficient to release the peripheral complex.

– Subsection “RhopH is synthesized as a non-integral complex”: the pronase experiment (Figure 4A-B) shows that the integral pool is exposed on the cell surface, but is not informative on the timing of synthesis.

– In Figure 4C, an additional loading control should be included to probe a merozoite integral membrane protein, similarly to band3 used for schizonts.

– In Figure 4F, why is there an increase in the soluble form upon removal of TMP? In Figure 4G, why is there a reduction in the carbonate-extracted pool after protease treatment?

– The authors acknowledge that there is no evidence for a direct role of the PTEX translocon in export, and now refer to a "PTEX translocon-dependent" export mechanism. However, PTEX conditional knockdown parasites are unable to progress through development and arrest at the ring stage. The authors cannot exclude that the observed defect in the switch from soluble to integral protein complex is due to the developmental arrest of the 13F10 mutant, possibly before the expression (or activation) of a translocon-independent mechanism for export, which normally takes place later during parasite development. This limitation should be mentioned in the Discussion, and an alternative pathway (arrow) with a question mark could be added in the model in Figure 5.

Reviewer #2:

The manuscript has been clearly improved and the Discussion is also strong to describe and propose this exciting structure and model. I support publication.

Concerning the structure.

Thank you for Figures 2D, 2G and H. It helps (which is nice) and I sense it is more accurate now than it was before (which is desirable).

The disulfide issue has been addressed. I am not particularly fond of disulfide bonds but since these are secreted proteins, with no obvious structural homologues, I think it was worth checking them as it could be of use to others in the future. I am happy to see that now 4 observed disulfide bonds have appeared (in addition to the first 7 described) in a decent model, a sign that the model builder(s) took the time and effort to build a structure as good as possible, that reflects the data and that was thoroughly looked at by its “owners” and not only by this reviewer. Thank you for providing us with a complete cryo-EM table including PDB and EMDB codes.

It seems now to pass quality control for an overall 2.9 A resolution structure. Thank you!

Thank you for clarifying Figure 5 to me and strengthening and improving the discussion on the model for RhopH with Figures 4C and 4G. I got it now.

I really look forward to see the membrane-inserted form. This will also be a beautiful story.

Reviewer #3:

The authors do a very good job addressing the reviewers' comments on the manuscript. Especially the addition of the investigation of the merozoites and the fractionation experiment in Figures 4C and 4G, respectively, provide clear insight into the localization of the soluble and membrane-bound fractions and the timing of the transition between the two states. Scientifically, I have no more comments. However, there are a few minor issues that may lead to confusion on the part of the reader and although the authors did an admirable job adjusting the language describing the model, some of the new data do not seem to be reflected in the model.

Subsection “Subunit interactions and roles”: The description of RhopH2 and BCl-xL and RhopH3 and SepL, respectively, as orthologues overstates the amount of similarity; orthologues are defined having identical function and potentially the ability to complement each other's function. The authors correctly state that they performed a similarity search, so it would be more correct to state that these proteins contain regions of structural similarity, rather than refer to them as homologues.

Discussion paragraph one: the first half of the sentence "Alternatively, monoclonal antibody…" is unclear.

Discussion paragraph two: the phrase "failed transit and membrane insertion in the PTEX knockdown (Figure 4F) may, nonetheless, be an indirect result of blocked export of multiple effector proteins" is only partially supported by the results. The authors very convincingly show that the complex can be present in a soluble (non-membrane-inserted) state and that the complex is found in this state in parasites in which PTEX synthesis has been knocked down. However, no evidence for the localization of RhopH in these parasites is provided and hence no conclusion about failed transit of the complex (assuming that the authors refer to the passage of the PVM) in these parasites can be drawn.

The sentence "Upon erythrocyte invasion, these and other rhoptry proteins are deposited into the parasitophorous vacuole, where the PTEX protein translocon mediates export into host cytosol (de Koning-Ward et al., 2009; Beck et al., 2014; Ho et al., 2018)." This sentence suggests that the PTEX transports rhoptry proteins across the PVM. To my knowledge, there are no data showing that rhoptry protein is transported by the PTEX; RhopH would be the first, if the role of PTEX in the transit of the complex past the PV is confirmed. Perhaps this can be rephrased to remove the suggested (although not explicitly stated) link between deposition of rhoptry proteins into the PV and protein export by the PTEX?

Figure 4D. The experiment showing the membrane association of the complex at different stages in the erythrocytic lifecycle is a valuable addition to the manuscript. It is unclear how the results fit into the model that the authors present, however. RhopH is deposited into the PV upon invasion and PTEX-dependent protein export starts at most minutes later. It is thus expected that RhopH is transported to the cytosol of the host cell almost immediately after invasion. As the complex remains in a soluble state for an extended time after this, it seems unlikely that PTEX is directly responsible for the membrane insertion of CLAG. The results rather seem to support a model in which an accessory factor, produced during the trophozoite stage and exported through the PTEX, is responsible for the transition from a soluble to a membrane-bound state.

Reviewer #4:

Further comments relating to the revised text are as follows:

– In Figure 1, I find the labelling of the timing of each step confusing, as the timing is relative to Rhoptry complex synthesis instead of from invasion

– In Figure 4D, the soluble, CO3 and membrane bands to do not seem to "add up" to what is observed in the total for Ring material, has some material been lost?

– In the same figure, the most amount of CO3 material is observed in schizont material, does this represent the newly synthesised material? Can this be made more clear in the text

– In Figure 4G, what stage are the parasites at?

– In Figure 4G, what is the purpose of adding the adolase panel?

If these comments are addressed I would recommend this paper be published in eLife.

10.7554/eLife.65282.sa2Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

[…]

1) Cryo-EM as used to determine the structure of a soluble form of RhopH, which apparently corresponds to a minor fraction of the complex "not associated with membranes". It is not clear what exactly this fraction corresponds to. In a previous report (Ito et al., 2017) the same group indicated that members of the complex are membrane-associated, but extractable in part by carbonate. Why is this fraction not recovered after hypotonic lysis?

It is not clear whether the soluble form corresponds to peripheral membrane proteins.

Have the authors attempted recovering the complex from purified merozoites or spent media for cryoEM? The authors propose that a change in conformation explains the differential behavior of the complex, but they should discuss alternative explanations, such as post-translational modifications or protein interactions, which could alter the association of the complex with the membranes.

We apologize for the confusion created by the phrase “not associated with membranes”. We intended to indicate that the CLAG3 released by freeze-thaw is no longer associated with membranes. Our present findings remain consistent with quantitative association of CLAG3 with membranes, first as a peripheral protein after synthesis and much later as an integral protein at the host membrane. We have revised this phrase to clarify and to emphasize that the freeze-thaw released protein is peripheral (“some CLAG3 from the peripheral pool”).

It is correct that hypotonic lysis liberates negligible amounts of RhopH proteins at any parasite stage, establishing that the RhopH complex is constitutively associated with membranes in either peripheral or integral pools. Partial release with simple freeze-thaw, as we now report for the first time, implicates loose membrane association for the peripheral pool, at least at some parasite stages. Alkaline carbonate treatment is harsher (pH 11) and is the most broadly accepted test for distinguishing between peripheral and integral proteins. Consistent with this rank order of stringency, we found that freeze-thaw releases smaller amounts of peripheral RhopH than alkaline carbonate treatment (Figure 1B). Because carbonate extraction is harsh and denatures proteins, identification of gentle, non-denaturing release with freeze-thaw proved enabling for structure determination. We have revised the manuscript to clarify this (“although Na2CO3 […] many proteins”).

Our manuscript includes a 3D low resolution structure from spent medium, using negative stain imaging (original Figure 1G). Because it is similar to the freeze-thaw reconstruction in Figure 1G, we could not commit resources and manpower to perform cryo-EM imaging of the essentially identical complex from spent medium.

We agree that post-translational modifications may facilitate the conformational changes needed for CLAG3 membrane insertion and have revised the Discussion accordingly (“these rearrangements […] acetylation”). Please note, however, that post-translational modifications should not be viewed as an “alternate explanation” because the CLAG3 α-helix 44 is buried in the soluble complex and must undergo reorientation for membrane insertion regardless of subsequent modifications.

2) The biochemical characterization of the different pools of RhopH complex in schizonts is complicated by the coexistence of previously synthesized RhopH (integral form) and newly synthesized protein (peripheral, in the rhoptries). In Figure 4, the authors should include purified merozoites, which should contain only the peripheral/soluble complex, and erythrocyte ghost membranes, which should only contain the integral form. This would provide clearer evidence to support their model.

We agree with these comments and have now performed fractionation studies with freed merozoites, as requested. The new Figure 4C shows that freed merozoites contain CLAG3 in only a peripheral form, in contrast to their corresponding schizont progenitor. We include Band3 as an endogenous erythrocyte membrane protein to confirm that the freed merozoites are not contaminated with RBC membranes and to further establish that carbonate extraction reveals both peripheral and integral CLAG3 pools in schizonts but only an integral pool for Band3, a more conventional membrane protein. We agree with this reviewer that this result provides independent evidence for the proposed model. Please see the revised Results, “Fractionation studies using purified merozoites …”, and corresponding figure legend.

Unfortunately, there is not a good procedure for harvesting erythrocyte ghost membranes from infected cells. A commonly discussed protocol (PMID: 2001227) uses host membrane lysis with saponin, mechanical shearing, and differential centrifugation; their paper used membrane-specific markers to assess harvest and contamination. While their data are useful, our unpublished studies with this procedure (previously and during the present work) have produced low recovery and likely contamination with Maurer’s clefts (a problem not evaluated in their study).

To provide analogous evidence for the model, we instead performed protease susceptibility studies with the epitope-tagged parasite used in our cryo-EM studies. The revised Figure 4G shows that while freeze-thaw released and carbonate-extractable CLAG3 are not cleaved by extracellular protease, the integral pool at the host membrane is susceptible. Notably, the observed c-terminal cleavage product is also carbonate-resistant, consistent with α-helix 44 serving as a transmembrane domain. Importantly, this finding indicates that host membrane CLAG3 is indeed integral. Please see Figure 4G, the revised Results “Stage-dependent membrane […] at the host membrane.” and revised Figure 4 legend.

These findings compellingly show that CLAG3 is synthesized and transferred to new erythrocytes as a soluble protein and that it subsequently becomes integral at the host membrane to become exposed to host plasma.

3) In the PTEX mutant experiment, it is crucial to analyze parasites at the same stage. Beck et al., 2014, showed that in the absence of TMP, 13F10 mutant parasites are blocked at the ring stage, so it is not surprising that the protein extraction profile in Figure 4E is similar to the ring profile in Figure 4C, and does not prove that export through PTEX is required for RhopH membrane insertion. This is a major weakness in the model.

PTEX conditional knockdown parasites are indeed unable to progress through development and resemble rings in their morphology. This concern, an indirect effect of PTEX block, applies equally well to other exported parasite proteins (e.g. PfEMP1 in the original 13F10 mutant studies, PMID: 25043010, where the authors stated “although […] delivery of PfEMP1 to the RBC surface is HSP101-dependent, we cannot exclude the possibility that this block is an indirect result of a failure to export other proteins required for PfEMP1…”).

The evidence for RhopH is better than for PfEMP1 and some other proteins because co-IP experiments reveal that PTEX components immunoprecipitate RhopH2 (Table S4 of PMID: 19536257) and reverse co-IP using RhopH2 as the bait pulls down PTEX components (Figure 3B of PMID: 28252383).

The new data we added to the revised manuscript (question #2 above), establish that CLAG3 is synthesized as soluble protein (new Figure 4C), remains soluble upon deposit into the ring-stage parasitophorous vacuole upon invasion (Figure 4D-E), and later becomes integral at the host membrane (new Figure 4G). This timeline for membrane insertion is also suggestive of a PTEX role.

We agree that the exact mechanism of PVM translocation remains unclear; indeed, the RhopH complex may face a unique dilemma if it requires unfolding for translocation and subsequent refolding in the RBC compartment. We also agree that CLAG3 membrane insertion may occur either through direct PTEX interaction or indirectly through the action of other exported proteins.

We have significantly revised the Discussion section to present our PTEX mutant findings more cautiously, to include citations for the forward and reverse pull-down experiments that support RhopH-PTEX interaction, and to more clearly describe the intriguing questions raised by our structural and biochemical studies. We have also revised the Abstract and Introduction sections to clarify this uncertainty. We have also changed Figure 5 to represent PTEX-dependent membrane insertion more cautiously.

Reviewer #3:

[…]

Abstract. Although the Abstract is clear it should introduce the names of the two other rhoptry proteins RhopH2 and RhopH3 proteins simultaneously within that third sentence for the sake of clarity and consistency, as it sounds odd to bring all the attention on CLAG3 (or RhopH1) but not mention the other two subunits so intimately associated with CLAG3.

Agreed and corrected.

Second paragraph in Introduction. I would clarify “expansion in humans and vertebrates” as in “Plasmodium spp. infecting humans and other vertebrates such as birds, rodents and primates” as they explain more clearly later in the manuscript.

Agreed and corrected.

At the end of the Introduction (and maybe also in the Abstract) I would suggest being a bit more explicit in introducing PTEX (I am aware that referencing the initial study from DeKoning-Ward et al. is done much later in the text) and its function as a vacuolar translocon. In general the Introduction assumes that most readers would be familiar with the exquisitely complicated life cycle of apicomplexan parasites such as Plasmodium (or Toxoplasma): Introducing the existence and formation of a host-derived parasitophorous vacuole (membrane) following invasion of the host red blood cell by the parasite sounds necessary to clarify the general picture. Again this is done quite late in the body of the manuscript.

This is an important suggestion, which we have accepted with revisions to both the Abstract (“After transfer…”) and last paragraph of the Introduction “The complex remains […] nutrients.”

1) Structure determination

End of Introduction. Results section and in general. Concerning the structure itself presented at 2.9Å resolution as per cryo-EM standards. 2.9 Å is not high resolution: atomic resolution should suffice.

Actually, there is an ongoing debate about “high”, “atomic”, and “near atomic” resolution in protein structures, as discussed in “Atomic resolution: a badly abused term in structural biology” (PMID: 28375149) and the provocative responses of 5 experts and the IUCr Commission on Biological Macromolecules (PMID: 28375150). The “Sheldrick criterion” of resolution dmin of 1.2 Å for “atomic” resolution may be the most conservative definition, with some respondents taking a more generous view. We agree with the consensus of these respondents that 2.9 Å, as we achieved, does not meet the standard of “atomic” or “near-atomic”. We have removed two instances of “high-resolution” that do not contribute meaning to our manuscript, but have chosen to keep two instances as they recognize that our resolution is higher than those achieved in all other cryo-EM studies of malaria parasite non-ribosomal proteins published to date.

Table 1. Statistics for cryo-EM structural analysis.

In Data collection and processing. Any reason why defocus, FSC threshold and map resolution values are not indicated?

In Refinement. Any reason why model resolution range is not indicated?

We have added the missing parameters to Table 1 and further improved refinement in the process.

Going through the PDB preliminary validation report indicates the presence of a fair amount of clashes in this large structure. RhopH2 in particular seems less well defined.

RhopH2 is indeed less well-defined than the other subunits. Our initial refinements permitted building of the region of RhopH2 that binds CLAG3, but multi-body refinement further improved RhopH2 density for model building, as we highlighted in the Figure 2—figure supplements 1 and 2. Nevertheless, RhopH2 remains incompletely built, possibly because of mobile domains that may serve yet-unknown functions.

Upon inspection of the structure using the provided PDB file, I will comment about disulfide bonds and "free" cysteines (reduced). The three proteins contain a fair amount of cysteines and including a total of 7 clearly identified disulfide bonds (overall the 3 subunits) as one could expect in "secreted" proteins facing harsher conditions and as the Authors say "suited for trafficking and transfer to a new host erythrocyte". These are fairly complex patterns of disulfide linkages that did not seem to catch the attention of the Authors.

In CLAG3 (chain A in the PDB file) they are 14 cysteines.

with lone cysteines C779 C1065 C1217 C1431

and three listed disulfide bonds (C409-C415) (C523-C544) and (C1352-C1355)

but pairs C335-?-C363 and C519-?-C547 seem also close enough to form disulfide bonds.

In RhopH2 (chain B in the PDB file) they are 8 cysteines.

with two listed disulfide bonds (C233-C240) and (C791-C851) and lone cysteines C268 and C531; C531 seems to have a problem in that loop? Here again C871 is close enough to C909 to form a disulfide bond?

In RhopH3 (Chain C in the PDB file) they are 14 cysteines.

with lone cysteines C336 and C446 and two listed disulfide bonds: (C157-C231); (C244-C253). That said, pairs (C42//C99) (C262-?-C276) (C421-?-C620) and (C475//C536) are also close enough to form disulfide bonds?

Is this a definitive assignment or are further adjustments needed/expected?

Yes, our original submission failed to mention the numerous cysteines in the three subunits, as recognized by early workers (PMID: 15953647). We have checked the maps (confirming nearly all this reviewer’s careful tallies above) and added both an orienting ribbon schematic and a tabulated list (new Figure 2G and H). We have also revised the Results (“Each subunit has numerous…”), Discussion (“If it transits directly….”) and Abstract (“tightly assembled with extensive disulfide bonding”) to highlight this important feature.

2) TMDs in RhopH complex.

Authors describe precisely the TMD in CLAG3 (α44) as potentially forming the trans-membrane pore in PSAC (where they previously mapped the mutation A1215) but they don't define explicitly the predicted TMDs in RhopH2 and RhopH3 in the manuscript. They are drawn on Figure 3G and Figure 3—figure supplement 1 panels E and F and one can roughly guess their position from the Phobius plots in panels C and D. It took me a while to figure out where to look in the structure.

We have explicitly defined the positions of these helices in the text “helices defined by V740-D757 and G595-Y622 of these subunits, respectively”.

As pointed earlier, it is a very interesting structure but it is not always well described. Since this is a new structure describing 3 novel proteins from Plasmodium (with no real structural homologues available), a potential membrane protein (complex) it is not like that they are legions. I count two so far, .it is worth the effort to give a complete classical diagram for each protein with secondary structure elements on top of the sequence and highlighting some of their most salient features on it (at least the TMDs, the HVR, the interactions surfaces). I understand that we cryo-EM and X-ray people are solving structures of increasing complexity, faster than we can write and maybe think but this should not prevent us from describing them with a minimum of precision so it makes the work of the reviewer and potential readers less painful and more interesting.

A full length sequence diagram with features highlighted is not practical as such a representation for the ~3700 total residues in the complex would take many pages to represent a relatively small number of features. We hope the reviewer agrees that the new ribbon diagrams (Figure 2D and 2G) serve this purpose well and concisely.

Using DALI, similarities with BCl-xL, SepL and γ-secretase APH-1 for TMD portions of RhopH2, H3 and CLAG3, respectively to support their claims.

Figure 3—figure supplement 1B. Despite its legend and knowing what Colicin and Bxl are and look like, I have absolutely no clue what that panel actually shows? It is not clearly illustrated.

The goal of panel B in Figure 3—figure supplement 1 is to show that some other pore-forming proteins also have soluble forms with their TMs buried. We show zoom-in structures of the relevant domains of colicin Ia and Bax with their pre-membrane inserted hydrophobic helices in green. To clarify this point for readers, we have revised this graphic to include a similar zoom-in view of CLAG3 and revised the figure legend, “(B) Cylinder view diagrams of CLAG3 […] insertion.”

3) “Interaction with PTEX” for trafficking across the PVM.

In Figure 4, the Authors demonstrate the existence of two pools but I am not convinced that the Authors provide strong evidence that PTEX is interacting with RhopH for its translocation across the PVM. One could argue that PTEX knockdown results in the lack of export of another protein that is directly or indirectly required for the subsequent proper maturation of RhopH from its soluble to integral form (PSAC) at the RBCM while RhopH follows another route for translocation across the PVM (not through the PTEX, not HSP101-dependent)? That point of the manuscript is rather weak and remains contentious as the Authors mention contradicting reports from Beck et al., 2014, and their own work from Ito et al., 2017. Should RhopH proteins really transit through PTEX then they are PEXEL-negative proteins.

We agree that PTEX-dependent membrane insertion of CLAG3 may reflect either direct interaction and translocation through PTEX or an indirect mechanism such CLAG3 membrane insertion though the action of chaperone(s) that are exported via PTEX. Previous publications have used forward and reverse pull-down experiments to suggest direct interaction between RhopH2 and PTEX components (Table S4 of PMID: 19536257 and Figure 3B of PMID: 28252383).

Most importantly, we have significantly revised the Discussion section to present our PTEX mutant findings more cautiously and to more clearly describe the intriguing questions raised by our structural and biochemical studies. We have also revised the Abstract and Introduction sections to clarify this uncertainty. Finally, we added new experimental data (Figure 4C and 4G) to more strongly support the model of a soluble complex delivered to new host erythrocytes and eventually inserted at the host membrane.

Although the Beck et al., 2014 and Ito et al., 2017 papers have different IFA results for CLAG3 export into host erythrocyte cytosol, they both report that PSAC activity is abolished in the PTEX knockdown (see Figure 3A-B of the Beck paper PMID: 25043010). We understand the Beck group also confirmed that CLAG3 fails to insert in the host membrane in the knockdown, based on communications with their authors. Thus, these studies are not contradictory for the point being made in Figure 4, namely CLAG3 membrane insertion through a PTEX-dependent process. We have revised the Results to clarify this, “While two studies have obtained conflicting results about whether RhopH proteins are exported via this translocon, both reported that PTEX knockdown abolishes activation of PSAC-mediated nutrient uptake at the host membrane (Beck et al., 2014; Ito et al., 2017)”.

Although the model is attractive, it is hard for me to understand how a ternary complex so tightly assembled with so many disulfide bonds would be threaded through the membrane via PTEX to be refolded and reassembled on the other side (whether in it is in the same soluble form or directly into a membrane-inserted integral form). All three TMDs predicted in each subunit are in the middle of each protein. Integration/insertion in the host cell membrane will then not only require considerable conformational rearrangement but also the translocation of protein sub-domains on opposite sides of the bilayer. While this is discussed where authors mentioned the absence of a lateral gate in the PTEX PVM pore subunit EXP2 for lateral insertion of TM segments, they do not rule out the intervention of other chaperones or other trafficking pathways: like vesicular trafficking maybe via tubular network extensions/Maurer's clefts. That latter route seems more likely.

We agree with this excellent assessment. We have revised the Discussion to highlight this intriguing problem.

Nevertheless, the Authors certainly draw 1) an interesting model for the “life cycle” of the RhopH complex and 2) a parallel between their system and the smaller pore forming toxins that also transition from a soluble monomeric form to an oligomeric membrane-inserted pore. However, their system is heterotrimeric thus suggesting a singular increase in complexity compared to the majority of single- or bi-component PFTs. What would be the receptor (usually a lipid in PFTs) triggering insertion inside the host cell membrane?

It is unfortunate that the Authors cannot visualize insertion (if it is spontaneous like PFTs and does not require extra energy) on membranes using their purified endogenous soluble complex. I realize they are limited by the amount of material available.

Another excellent point. The amount of purified RhopH complex is actually not limiting for functional reconstitution experiments, which were pursued using several distinct but unsuccessful avenues during the course of this work (esp. reconstitution of the soluble complex into planar lipid bilayers for electrical recordings of PSAC activity). The failure of these experiments may reflect involvement of chaperones in the membrane insertion process, but there are other explanations also.

The Figure 5 conceptualizing their model is nice but somehow confusing.I suggest labeling the PV and/or PVM, the parasite, the RBC for clarification. Although the numbers give a false impression of ordered steps. While 1 and 2 are connected, I would think that here 3 does not follow 2. 3 is de novo endogenous synthesis of RhopH from a parasite dividing inside the infected red blood cell. And while PSAC in 4 might be the result of translocation through PTEX as the Authors propose in this manuscript, it could also be the result of another insertion/secretion path through tubular network maybe. Is there a reason why the inserted form of CLAG3 (PSAC) has two TM spanning segments drawn? Is it confirmed that both Rhop2 and Rhop3 are on the cytoplasmic side of the erythrocyte in the inserted form?

We have revised the graphic to add key labels.

In fact, #3 does follow #2 because it is the rhoptry protein pool delivered into the PV during invasion that must then be exported into host cytosol; there is no detectable transcription/translation of RhopH proteins at ring or early trophozoite stages as would be required for de novo endogenous synthesis by the intracellular parasite.

We revised the PTEX interaction step because we agree that the original graphic too strongly suggests direct threading of RhopH proteins through PTEX. While the complex may indeed diffuse along the tubular network, the topological problem of membrane translocation would remain. Our evidence supports translocation and membrane insertion in a PTEX-dependent manner but cannot distinguish between direct or indirect mechanisms.

Prior studies have established that only a small variable region of CLAG3 (HVR in the manuscript) is surface exposed, so a second TM upstream of the HVR and helix 44 is necessary. RhopH2 and RhopH3 are not susceptible to extracellular protease, but remain associated with CLAG3 at the host membrane based on FRET studies, so are drawn as cytoplasmic. We have revised the figure legend and the corresponding Discussion section to clarify these points.

Reviewer #4:

Please note that I can only judge the cell biological and biochemical aspects of the manuscript. The authors investigate the dual soluble/membrane integral nature of the RhopH complex described previously, showing that it is soluble at late stages (in schizonts) and transforms into a membrane-integral form after invasion and that PTEX activity is required for this transition. The work is well done and the data are clear and convincing. The structures show that the TM are folded into interior of the protein, providing a convincing reason why the complex is initially soluble. However, there is no indication how the transition from soluble to membrane-bound may occur may occur and the data implicating the PTEX in transport of the complex are very indirect and open to other interpretations.

1) The authors claim that the PTEX is involved in transport of the RhopH complex to the host cell, but do not provide evidence for this. The experiment in Figure 4E shows that PTEX is required for insertion of the complex into the membrane, but does not prove any direct interaction of the PTEX with the complex. There are several other explanations for the finding that PTEX activity is required for membrane insertion of the complex, such as the requirement of an exported accessory factor.

We agree that PTEX-dependent membrane insertion of CLAG3 may reflect either direct interaction and translocation through PTEX or an indirect mechanism such CLAG3 membrane insertion though the action of chaperone(s) that are exported via PTEX. Previous publications have used forward and reverse pull-down experiments to suggest direct interaction between RhopH2 and PTEX components (Table S4 of PMID: 19536257 and Figure 3B of PMID: 28252383).

We have significantly revised the Discussion section to present our PTEX mutant findings more cautiously, to include citations for forward and reverse pull-down experiments that support RhopH-PTEX interaction, and to more clearly describe the intriguing new questions raised by our structural and biochemical studies. We have also revised the Abstract and Introduction sections to clarify this uncertainty. We have also added new experiments (Figure 4C and 4G) that delimit the timing of RhopH membrane insertion, supporting PTEX-dependence. Finally, we changed Figure 5 to represent PTEX-dependent membrane insertion more cautiously.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Reviewer #1:

The authors resubmit an improved version of their manuscript. In particular, they describe the structure of the RhopH complex with more details, and added new data concerning the biochemical characterization of the complex. In particular, analysis of purified merozoites confirms that only the peripheral form of CLAG is found at this stage, corroborating the proposed model for RhopH synthesis and export. The authors partially clarified in the revised text the nature of the peripheral (carbonate-extractable) and integral (carbonate-resistant) protein pools. However, some aspects in the fractionation studies still need to be clarified, and the role of the PTEX translocon should be considered with more caution.

– There is still a lack of clarity regarding the soluble form in the fractionation studies. In the Materials and methods, the authors indicate that after parasite resuspension in lysis buffer (7.5 mM Na2HPO4, 1 mM EDTA, pH 7.5), followed by ultracentrifugation, the supernatant was kept as the "soluble" fraction. CLAG3 is detected in this soluble fraction, as shown in Figure 4A, 4D, 4F. It is unclear what this soluble form corresponds to. Is it the same fraction as indicated as "freeze-thaw" in Figure 4G? If not, why is CLAG3 released in the supernatant upon parasite treatment with the buffer alone? This contradicts the conclusions drawn from Figure 1B, stating that hypotonic lysis is not sufficient to release the peripheral complex.

Figure 4A and 4B have been corrected. The “soluble” label in these panels have been changed to “freeze-thaw” as these samples were frozen before thawing for membrane fractionation.

The experiments shown Figure 4D did not use frozen samples; a “soluble” band is detected, primarily in ring-infected cells. We suspect this relates to differences in harvest of these stages. While schizont- and trophozoite-infected cells were enriched by the percoll-sorbitol method, ring-infected cells cannot be similarly enriched. We therefore harvested and performed fractionation on rings without enrichment. Thus, a somewhat more complicated fractionation procedure with unenriched rings may account for detectable amounts of CLAG3 in panel 4D, ring stage parasites. We clarify this point in the figure legend. “While schizont- […] non-additive fractionation.”.

Figure 4F, where we used a conditional PTEX knockdown parasite for CLAG3 fractionation studies, is the most interesting. It is notable that PTEX knockdown (-TMP) yields a band in the soluble lane but +TMP does not. We do not understand this well, but a conservative hypothesis is that some of the non-integral CLAG3 pool is released from its peripheral association with membranes because of protein crowding after blocked export in the PTEX knockdown. We clarify this possibility in the Results: “CLAG3 that failed to insert […] export from the parasitophorous vacuole.”.

– Subsection “RhopH is synthesized as a non-integral complex”: the pronase experiment (Figure 4A-B) shows that the integral pool is exposed on the cell surface, but is not informative on the timing of synthesis.

Yes, the timing of synthesis was previously established based on stage-specific transcription and translation in schizonts. This was stated in the manuscript Introduction with citation to Ling et al., 2005.

– In Figure 4C, an additional loading control should be included to probe a merozoite integral membrane protein, similarly to band3 used for schizonts.

Figure 4C shows Band3, not as a loading control, but to demonstrate that the freed merozoites are not contaminated by erythrocyte membranes. Band3 is detected in schizont-infected erythrocytes but not in freed merozoites, indicating that these merozoites are not contaminated with erythrocyte membrane fragments. As this experiment is not designed to quantitate protein abundance in merozoites vs schizonts, we do not see value in a merozoite “loading control”. Indeed, the procedure for freed merozoite harvest sacrifices yield to obtain high purity (PNAS 108:13275-80).

– In Figure 4F, why is there an increase in the soluble form upon removal of TMP? In Figure 4G, why is there a reduction in the carbonate-extracted pool after protease treatment?

As described above, we clarify the increase in the soluble form upon PTEX knockdown (Figure 4F) in the Results: “CLAG3 that failed to insert […] export from the parasitophorous vacuole.”

The difference the reviewer observes between – and + protease for the “CO3=” lanes is not statistically significant (P = 0.42, paired Student t test from 4 matched trials). We do not think it is necessary to explicitly state this in the paper as it is tangential to this experiment’s purpose.

– The authors acknowledge that there is no evidence for a direct role of the PTEX translocon in export, and now refer to a "PTEX translocon-dependent" export mechanism. However, PTEX conditional knockdown parasites are unable to progress through development and arrest at the ring stage. The authors cannot exclude that the observed defect in the switch from soluble to integral protein complex is due to the developmental arrest of the 13F10 mutant, possibly before the expression (or activation) of a translocon-independent mechanism for export, which normally takes place later during parasite development. This limitation should be mentioned in the Discussion, and an alternative pathway (arrow) with a question mark could be added in the model in Figure 5.

Apologies for an intentional double negative, but we do not acknowledge that there is no evidence for a direct role in of the PTEX translocon in export. To the contrary, previous IFA experiments provide strong evidence. Please see Figure 7B of eLife 6:e23485 (2017), where immunofluorescence experiments show that all three RhopH components are trapped in the parasitophorous vacuole upon TMP removal in the PTEX conditional knockdown. Forward and reverse coimmunoprecipitation experiments from other groups (cited in the first paragraph) also provide evidence for direct interaction between PTEX and RhopH proteins to mediate export.

We also do not refer to a “PTEX translocon-dependent export mechanism”. The 5 instances of “translocon-dependent”, “PTEX-dependent” or “dependent on PTEX activity” are all careful to state that RhopH membrane insertion is PTEX-dependent.

These reviewer statements incorrectly conflate PTEX-dependent “export” and “membrane insertion”. These processes should be considered separately. Export has been studied previously (eLife 6:e23485 and Nature 511:592, where only some findings are contradictory), requires further study, but is not a question we have pursued here because it will require new technologies.

The membrane insertion event is the important question our manuscript examines for the first time, as it is enabled by our high-resolution structure. We agree with the reviewer that insertion may occur either directly during PVM translocation or later through the action of exported chaperones, and hence refer to membrane insertion as “PTEX-dependent”. This word choice is important and appropriately cautious.

We have revised the Discussion to clarify these points, “We determined that the complex […] may result from blocked export of multiple effector proteins”.

We do not agree with the reviewer suggestion to add an arrow with question mark to Figure 5 because it would incorrectly imply that there is a route for RhopH membrane insertion that is completely independent of PTEX. The data clearly indicate that there is no such route, so a graphic such as this would be misleading.

Reviewer #3:

The authors do a very good job addressing the reviewers' comments on the manuscript. Especially the addition of the investigation of the merozoites and the fractionation experiment in Figures 4C and 4G, respectively, provide clear insight into the localization of the soluble and membrane-bound fractions and the timing of the transition between the two states. Scientifically, I have no more comments. However, there are a few minor issues that may lead to confusion on the part of the reader and although the authors did an admirable job adjusting the language describing the model, some of the new data do not seem to be reflected in the model.

Subsection “Subunit interactions and roles”: The description of RhopH2 and BCl-xL and RhopH3 and SepL, respectively, as orthologues overstates the amount of similarity; orthologues are defined having identical function and potentially the ability to complement each other's function. The authors correctly state that they performed a similarity search, so it would be more correct to state that these proteins contain regions of structural similarity, rather than refer to them as homologues.

Although it is conventional to refer to protein 3D structure comparisons, as used to find BCl-xL and SepL, as “structural homology searches” and to call the hits “structural homologs”, we want to be abundantly cautious and address this reviewer’s point. We have therefore changed “structural homologs” to “hits from our structural similarity searches”.

Discussion paragraph one: the first half of the sentence "Alternatively, monoclonal antibody…" is unclear.

Yes, this was a bit awkward. We have revised to this sentence to clarify. “The RhopH3 C-terminus …”.

Discussion paragraph two: the phrase "failed transit and membrane insertion in the PTEX knockdown (Figure 4F) may, nonetheless, be an indirect result of blocked export of multiple effector proteins" is only partially supported by the results. The authors very convincingly show that the complex can be present in a soluble (non-membrane-inserted) state and that the complex is found in this state in parasites in which PTEX synthesis has been knocked down. However, no evidence for the localization of RhopH in these parasites is provided and hence no conclusion about failed transit of the complex (assuming that the authors refer to the passage of the PVM) in these parasites can be drawn.

There is already published data for this point. Please see Figure 7B of eLife 6:e23485 (2017), where immunofluorescence experiments revealed that all three RhopH components are trapped in the parasitophorous vacuole upon TMP removal in the PTEX conditional knockdown.

We have revised the Discussion to clarify that PTEX knockdown traps RhopH proteins in the parasitophorous vacuole, based on the prior IFA experiments. “We determined that the complex […] may result from blocked export of multiple effector proteins”.

The sentence "Upon erythrocyte invasion, these and other rhoptry proteins are deposited into the parasitophorous vacuole, where the PTEX protein translocon mediates export into host cytosol (de Koning-Ward et al., 2009; Beck et al., 2014; Ho et al., 2018)." This sentence suggests that the PTEX transports rhoptry proteins across the PVM. To my knowledge, there are no data showing that rhoptry protein is transported by the PTEX; RhopH would be the first, if the role of PTEX in the transit of the complex past the PV is confirmed. Perhaps this can be rephrased to remove the suggested (although not explicitly stated) link between deposition of rhoptry proteins into the PV and protein export by the PTEX?

We have revised this sentence to avoid such inference and to highlight that PTEX-mediated export of RhopH would be a first. “The PTEX protein translocon […] deposited in the vacuole”.

Figure 4D. The experiment showing the membrane association of the complex at different stages in the erythrocytic lifecycle is a valuable addition to the manuscript. It is unclear how the results fit into the model that the authors present, however. RhopH is deposited into the PV upon invasion and PTEX-dependent protein export starts at most minutes later. It is thus expected that RhopH is transported to the cytosol of the host cell almost immediately after invasion. As the complex remains in a soluble state for an extended time after this, it seems unlikely that PTEX is directly responsible for the membrane insertion of CLAG. The results rather seem to support a model in which an accessory factor, produced during the trophozoite stage and exported through the PTEX, is responsible for the transition from a soluble to a membrane-bound state.

We respectfully point out that this comment assumes that PTEX-dependent export is constitutively applied to all cargo proteins present in the vacuole. This assumption has not been tested experimentally. Because resident soluble proteins are not exported, it seems likely that there are unidentified mechanisms that control timing and targeting of proteins in the PV for export, in contradiction to this implicit assumption. We also point out that IFA experiments with PTEX knockdown parasites reveal RhopH proteins trapped in the PV. The revised the paragraph addresses this point, “We determined that the complex […] may result from blocked export of multiple effector proteins”.

Reviewer #4:

Further comments relating to the revised text are as follows:

– In Figure 1, I find the labelling of the timing of each step confusing, as the timing is relative to Rhoptry complex synthesis instead of from invasion

Because this manuscript tracks RhopH from its point of synthesis, we think it is best to have the graphic labelled according to this complex’s synthesis. As the biology and trafficking of this complex are complicated, we believe this will help readers. A timing based on invasion would force labeling of RhopH synthesis at 38-44 h and confuse more readers than not.

– In Figure 4D, the soluble, CO3 and membrane bands to do not seem to "add up" to what is observed in the total for Ring material, has some material been lost?

We have added a statement about “non-additive fractionation” to the Figure 4, panel D legend. This likely resulted because ring-infected cells cannot be enriched by the percoll-sorbitol method, requiring fractionation of rings at lower parasitemias than possible for schizont- and trophozoite-infected cells.

– In the same figure, the most amount of CO3 material is observed in schizont material, does this represent the newly synthesised material? Can this be made more clear in the text

Yes, this is almost certainly the newly synthesized protein, as supported by freed merozoite and protease-susceptibility studies (Figure 4C and 4A-B). We have made this more explicit: “a primarily extractable form upon synthesis in schizonts …”

– In Figure 4G, what stage are the parasites at?

– In Figure 4G, what is the purpose of adding the adolase panel?

This experiment used mixed trophozoite- and schizont-infected cells, as now indicated in the figure legend, “enriched mature infected cells”. Aldolase is shown as conventional soluble protein that is quantitatively extracted by freeze-thaw and carbonate treatment. Similarly, EXP2 is shown as a more conventional integral membrane protein that is minimally extracted by these treatments. Comparison to these proteins highlights that CLAG3 is unusual, present as both soluble and integral forms within infected cells. We hope this comparison is clear from the Figure 4G legend.