Rab5b-Associated Arf1 GTPase Regulates Export of N-Myristoylated Adenylate Kinase 2 From the Endoplasmic Reticulum in Plasmodium falciparum

Plasmodium falciparum extensively remodels human erythrocytes by exporting hundreds of parasite proteins. This remodeling is closely linked to the Plasmodium virulence-related functions and immune evasion. The N-terminal export signal named PEXEL (Plasmodium export element) was identified to be important for the export of proteins beyond the PVM, however, the issue of how these PEXEL-positive proteins are transported and regulated by Rab GTPases from the endoplasmic reticulum (ER) to the cell surface has remained poorly understood. Previously, we identified new aspects of the trafficking of N-myristoylated adenylate kinase 2 (PfAK2), which lacks the PEXEL motif and is regulated by the PfRab5b GTPase. Overexpression of PfRab5b suppressed the transport of PfAK2 to the parasitophorous vacuole membrane and PfAK2 was accumulated in the punctate compartment within the parasite. Here, we report the identification of PfRab5b associated proteins and dissect the pathway regulated by PfRab5b. We isolated two membrane trafficking GTPases PfArf1 and PfRab1b by coimmunoprecipitation with PfRab5b and via mass analysis. PfArf1 and PfRab1b are both colocalized with PfRab5b adjacent to the ER in the early erythrocytic stage. A super-resolution microgram of the indirect immunofluorescence assay using PfArf1 or PfRab1b- expressing parasites revealed that PfArf1 and PfRab1b are localized to different ER subdomains. We used a genetic approach to expresses an active or inactive mutant of PfArf1 that specifically inhibited the trafficking of PfAK2 to the parasitophorous vacuole membrane. While expression of PfRab1b mutants did not affect in the PfAK2 transport. In contrast, the export of the PEXEL-positive protein Rifin was decreased by the expression of the inactive mutant of PfRab1b or PfArf1. These data indicate that the transport of PfAK2 and Rifin were recognized at the different ER subdomain by the two independent GTPases: PfAK2 is sorted by PfArf1 into the pathway for the PV, and the export of Rifin might be sequentially regulated by PfArf1 and PfRab1b.

Plasmodium falciparum extensively remodels human erythrocytes by exporting hundreds of parasite proteins. This remodeling is closely linked to the Plasmodium virulence-related functions and immune evasion. The N-terminal export signal named PEXEL (Plasmodium export element) was identified to be important for the export of proteins beyond the PVM, however, the issue of how these PEXEL-positive proteins are transported and regulated by Rab GTPases from the endoplasmic reticulum (ER) to the cell surface has remained poorly understood. Previously, we identified new aspects of the trafficking of Nmyristoylated adenylate kinase 2 (PfAK2), which lacks the PEXEL motif and is regulated by the PfRab5b GTPase. Overexpression of PfRab5b suppressed the transport of PfAK2 to the parasitophorous vacuole membrane and PfAK2 was accumulated in the punctate compartment within the parasite. Here, we report the identification of PfRab5b associated proteins and dissect the pathway regulated by PfRab5b. We isolated two membrane trafficking GTPases PfArf1 and PfRab1b by coimmunoprecipitation with PfRab5b and via mass analysis. PfArf1 and PfRab1b are both colocalized with PfRab5b adjacent to the ER in the early erythrocytic stage. A super-resolution microgram of the indirect immunofluorescence assay using PfArf1 or PfRab1b-expressing parasites revealed that PfArf1 and PfRab1b are localized to different ER subdomains. We used a genetic approach to expresses an active or inactive mutant of PfArf1 that specifically inhibited the trafficking of PfAK2 to the parasitophorous vacuole membrane. While expression of PfRab1b mutants did not affect in the PfAK2 transport. In contrast, the export of the INTRODUCTION Secretory proteins are synthesized in the endoplasmic reticulum (ER) and are delivered to their destinations via the Golgi apparatus. ER-to-Golgi trafficking is highly conserved among eukaryotes, and two types of Ras super families of GTPases, the Sar/Arf and Rab families, mediate protein sorting and transport as molecular switches between these organelles (Suda et al., 2018;Kurokawa and Nakano, 2019). A guanine nucleotide exchange factor (GEF) acts on GDP-bound form of GTPase to convert it to a GTP-bound active state, and a GTPase accelerating protein (GAP) binds to the GTPase to catalyze hydrolysis of the bound GTP to GDP and thereby convert the GTPase back to its inactive state (Hutagalung and Novick, 2011). The GTPase cycle between GDP-and GTP-bound forms causes conformational changes and specific effector molecules are recruited to the GTP-bound active form to complete various membrane trafficking events (Stenmark, 2009;Donaldson and Jackson, 2011). Many regulatory functions performed by the proteins in the Sar/Arf and Rab families were identified by their interaction with diverse effector proteins that select cargo, promote vesicle movement, and verify the correct site for fusion (Hutagalung and Novick, 2011).
The Sar/Arf and Rab families of GTPases regulate the bidirectional transport between ER and the Golgi in mammalian cells and land plants (Brandizzi and Barlowe, 2013). Newly synthesized secretory proteins translocate into the ER lumen using the N-terminal signal peptide (Nickel and Rabouille, 2009). The ER luminal chaperone BiP assists in folding newly synthesized proteins to maintain protein homeostasis (Craig et al., 1993). Activation of the ER-localized GTPase Sar1 and its GTP hydrolysis triggers the recruitment of the COPII coat component and causes the enrichment of cargo proteins into the ER exit sites (ERES) where COPII-coated vesicles form (Kung et al., 2012;Yorimitsu and Sato, 2012). The COPII machinery performs two critical functions: first, Sar1 and the inner layer Sec23-Sec24 COPII subunits bind to and select specific cargo for packaging; second, polymerization of the outer layer of Sec13-Sec31 COPII subunits occurs into a cage structure to drive vesicle formation (Bi et al., 2007). After uncoating of the COPII coat, diffusive vesicles are anchored to cis-Golgi membranes by the extended coiled-coil domain that tethers factor protein p115, the yeast homolog of Uso1, as well as the multi-subunit TRAPPI (transport protein particle I) complex that activates Rab1 GTPase (Cai et al., 2007). The fusion of vesicles depends on the assembly of integral membrane SNARE complexes between the donor and acceptor membranes (Allan et al., 2000). In contrast, a heptameric complex COPIcoated vesicle facilitates the retrieval of escaped ER luminal proteins that contain KDEL signals, which are recognized by the resident KDEL receptor ERD2 in the cis-Golgi, and by SNARE proteins Semenza et al., 1990). Once Arf1 is activated by the GEF which contains a conserved Sec7 domain, the membrane-localized Arf1 recruits the COPI coat to the cis-Golgi membranes (Zhao et al., 1999). The COPI subunits recognize the sorting motifs of transmembrane cargo proteins and are incorporated into nascent COPI vesicles and subsequently retrieve cargo proteins to the ER (Eugster et al., 2000). Most Rab and Arf GTPases carry lipid modifications that are necessary for membrane recruitment. In the case of Rab, a carboxy-terminal geranylgeranylation (20-carbon unsaturated fatty acid) group at the C-terminal cysteine residue is responsible for the membrane attachment of the Rab-GTP form (Leung et al., 2006). In contrast, the N-terminal glycine residue of Arf is myristoylated (14-carbon saturated fatty acids) (Pasqualato et al., 2002).
Plasmodium falciparum is the causative agent of malaria, and it extensively remodels the human erythrocytes in which it resides (de Koning-Ward et al., 2016). The remodeling process is conducted by hundreds of proteins exported from the parasites into the host cell compartment, and these enable virulencerelated functions including cytoadherence to the vascular endothelium, immune evasion, and nutrient uptake (Miller et al., 2013). Many of the exported proteins contain a five amino acid motif termed the Plasmodium export element (PEXEL) which is found~20 amino acids downstream of the signal peptide (Hiller et al., 2004;Marti et al., 2004). The PEXEL motif is removed in the ER lumen of the parasite by the ER resident aspartic protease plasmepsin V before exit of the ER (Boddey et al., 2010;Russo et al., 2010). All repertoires of COPII and COPI components are conserved in the Plasmodium spp. (Kibria et al., 2019), suggesting that the early secretory system of Plasmodium may resemble that in higher eukaryotes. Several studies have reported that the fungal metabolite brefeldin A (BFA), which inhibits the Golgi-localized GEF Arf1, disrupts the accurate export of PEXEL-containing proteins (Akompong et al., 2002;Chang et al., 2008), indicating that these proteins are trafficked through the classical ER-Golgi pathway. Subsequently, the exported proteins are transported across the parasite plasma membrane and the parasitophorous vacuole membrane (PVM) via the multimeric Plasmodium translocon of exported proteins (PTEX) into the erythrocyte cytosol (de Koning-Ward et al., 2009). In these Plasmodium trafficking systems, the roles of small GTPases are poorly understood such that the subcellular localization and function of only one Sar1 and six Rab GTPases has been being characterized functionally; i.e., PfSar1, PfRab1a, PfRab5a, PfRab5b, PfRab6, PfRab7, and PfRab11A (Adisa et al., 2001;Struck et al., 2005;Elliott et al., 2008;Agop-Nersesian et al., 2009;Rached et al., 2012;Krai et al., 2014;Ebine et al., 2016;Morse et al., 2016). However, effectors and binding proteins have only been identified for PfRab11A (Agop-Nersesian et al., 2009). Further, PfRab11A is associated with the myosin-tail-interacting-protein and is crucial for the parasite invasion of host cells (Agop-Nersesian et al., 2009).
Several exported proteins lacking an N-terminal PEXEL motif and are called PEXEL-negative proteins (PNEPs). These are also translocate through PTEX for export into the erythrocyte cytosol (Spielmann and Gilberger, 2010;Elsworth et al., 2014). For example, the skeleton binding protein 1 (SBP1), which is a PEXEL-negative transmembrane protein, is supposed to export via the classical secretory pathway, because the truncated SBP1 internalized in the perinuclear staining corresponding to the ER (Saridaki et al., 2009). However, recent report that SBP1 was included within electron-dense materials in the parasite cytoplasm (Iriko et al., 2020), strongly suggests that the presence of alternate export pathways except for the classical ER-to-Golgi transport pathway in the parasite trafficking system. Plasmodium falciparum adenylate kinase 2 (PfAK2), which lacks a signal peptide and the PEXEL motif, but is modified via myristoylation (14-carbon) and palmitoylation (16-carbon saturated fatty acid) at the N-terminal glycine and cysteine residues respectively, is transported to the parasite PVM (Rahlfs et al., 2009;Thavayogarajah et al., 2015;Ebine et al., 2016). Previously, we have reported that trafficking of PfAK2 to the PVM was repressed and the PfAK2 accumulated in the punctate structure within the parasite cytoplasm due to the overexpression of PfRab5b, indicating that PfRab5b is involved in the transport of PfAK2 to the PVM (Ebine et al., 2016). PfRab5b possesses a structurally unique motif that is similar to PfAK2, which lacks a C-terminal geranylgeranylation modification, while retaining the N-terminal myristoylation and palmitoylation motifs (Ezougou et al., 2014). In rodent malaria P. berghei, gene-deletion of PbRab5b was unsuccessful, indicating Plasmodium Rab5b is essential for the blood stage growth (Ezougou et al., 2014). Subcellular localization of PfRab5b is proximal to the region of the BiP-positive ER, and segregated from COPII subunit Sec13-positive ERES (Ebine et al., 2016). Interestingly, the inhibition of transport was observed in PfAK2, however the PEXEL-positive erythrocyte vesicle protein 1 (PfEVP1) and SBP1 were correctly exported upon the overexpression of PfRab5b (Ebine et al., 2016). These findings suggested that PfRab5b may regulate the transport of PfAK2 by the COPII independent non-classical pathway (Ebine et al., 2016).
Currently, none of GTPase, which regulates the trafficking of PNEPs including PfAK2, is elucidated, A bioinformatic technique previously identified casein kinase1 (CK1) as a PfRab5b binding protein (Rached et al., 2012); however, it remains unclear how PfRab5b is involved in the selective transport of PfAK2. To understand the mechanisms underlying the PfRab5b-dependent specific trafficking pathway, we isolated PfRab5b associated proteins using coimmunoprecipitation and mass analysis approaches. Among the candidate proteins, we found that the PfArf1 and PfRab1b GTPases were colocalized with PfRab5b in the compartment close to the ER. Indirect immunofluorescence assay for PfArf1or PfRab1b-expressing parasites using super-resolution microscopy revealed that PfArf1 and PfRab1b were localized to different ER subdomains. Further, using a genetic approach to express an active or inactive mutant indicated that PfArf1, but not PfRab1b, was involved in the transport of PfAK2 to the PVM. Unexpectedly, PfRab1b participates in the trafficking of the PEXEL-positive export protein Rifin. Our data suggest that the trafficking pathways of PfAK2 and Rifin are separated in the PfArf1-positive ER subdomain, and this is the first report for the identification of GTPase which regulates transport of PfAK2 in Plasmodium.

Ethics Statement
Human RBCs and plasma were obtained as donations from anonymized individuals at the Japanese Red Cross Society (no. 28J0023).

Strain Culture and Transfection
Plasmodium falciparum line MS822 (Nakazawa et al., 2011) was cultured as described previously (Alexandre et al., 2011), and transfection was performed as described (Deitsch et al., 2001;Alexandre et al., 2011). Transfected parasites were selected with 5 nM of WR99210-HCl (a kind gift from D Jacobus) or 2.5 µg/ml of blasticidin S-HCl (BSD, Funakoshi). Transfectants were selected 2 weeks after the addition of drugs.

Coimmunoprecipitation Assay and Mass Spectrometry Analysis
Transgenic parasites carrying the PfRab5b-YFP or PfRab5b-YFP-FLAG with pCHD43(II) episomal plasmids were cultured in 250 ml medium (5% hematocrit, 7% parasitemia), and infected red blood cells (iRBCs) were collected by centrifugation at 560 × g for 5 min and permeabilized using 0.075% saponin in phosphate buffered saline (PBS) for 30 min on ice. Samples were crosslinked using 0.5 mM 3,3'-dithiodipropionic acid di (Nhydroxysuccinimide ester) (DSP) (Sigma-Aldrich, St. Louis) for 30 min at room temperature, and were subsequently quenched by the addition of 50 mM Tris-HCl (pH 7.5) according to the manufacturer's protocol. Next, the samples were solubilized with 250 µl of 1.0% Triton X-100 in PBS and kept on ice for 20 min. The insoluble fraction was removed by centrifugation at 9,100 × g for 5 min, and the supernatant fractions were incubated with Protein G Sepharose (GE Healthcare) at 4°C for 60 min to reduce non-specific binding during coimmunoprecipitation. The PfRab5b protein complex was immunoprecipitated using anti-FLAG antibody conjugated agarose (EZview Red Anti-FLAG M2 Affinity Gel, Sigma-Aldrich) at 4°C for 3.5 h, and the immunoprecipitated was washed thrice with 1.0% Triton X-100 in PBS. The PfRab5b protein complexes were eluted with 50 µl of 0.2 mg/ml FLAG peptide (Sigma-Aldrich) in PBS containing 50 mM DTT and 10 mM EDTA at 4°C for 16 h. The eluted proteins were separated via 12% SDS-PAGE and visualized using silver staining. The in-gel trypsin digestion of proteins, liquid chromatography, and time-of-flight tandem mass spectrometry (LC-ToF MS/MS) (Orbitrap, Thermo Fisher Scientific, Waltham, MA, USA) were performed as described previously (Makiuchi et al., 2013). The quantitative value, normalized with unweighted spectrum counts, was used to estimate relative quantities of proteins in the samples. Specific binding proteins were selected by the following criteria: 1) peptide fold enrichment that showed >3.5 higher in PfRab5b-YFP-FLAG samples compared with those in the PfRab5b-YFP control were selected. 2) Protein with >3.5 higher in PfRab5b-YFP-FLAG samples with no value in the PfRab5b-YFP control were selected. 3) Proteins those involved in cytosolic proteins such as proteasome and ribosomal proteins were removed from the list.

Indirect Immunofluorescence Assay
Transgenic parasites carrying the PfRab5b, PfArf1, or PfRab1b mutant proteins fused with a DD system with artificial centromere plasmid PfCenV-ef1a-double were stabilized with 0.5 µM Shld1 for 24 h (Clontech) (Armstrong and Goldberg, 2007). Next, cultures were synchronized by the 5% sorbitol for 10 min at room temperature. After washing, ring-rich parasites were cultured until indicated erythrocytic stages under the 0.5 µM Shld1. Infected erythrocytes were collected and fixed with 4% paraformaldehyde (Thermo) in PBS at 4°C for more than 12 h. Fixed samples were permeabilized with 0.5% Triton X-100 in PBS for 30 min and blocked with 3% skim milk prepared with PBS for 10 min. Primary antibodies were used at the following dilutions: anti-BiP (1:100, kindly gift from Prof. Kita) and anti-ERD2 (1:100) (Elmendorf and Haldar, 1993). Alexa 488conjugated anti-mouse IgG (1:10,000, Molecular Probes) was used as the secondary antibody. The number of parasites that showed colocalization of the RFP and PfERD2 or BiP signals was counted in 20-30 trophozoites from three independent experiments. A test for statistical significance was performed using the Student t-test.

Microscopy Techniques
Images were acquired using an LSM700 or LSM780 confocal laser-scanning microscope (Zeiss, Germany). For distance measurements between PfBiP and PfArf1 or PfRab1b, superresolution imaging was performed using the Zeiss LSM880 with Airyscan confocal laser-scanning microscope, that is equipped with an oil-immersion 100× objective lens (alpha Plan-Apochromat 100×/1.46 oil DIC M27 Elyra) (Zeiss, Germany). The background fluorescence from the non-transfected parasites was set as baseline. All images were acquired in the same laser voltage and gain (exposure time). Raw data were processed using the Zeiss Zen2 software to measure fluorescent intensities. The images were analyzed using the Zeiss Zen2 software or Fiji-ImageJ software (Schindelin et al., 2012). Scoring of images was judged by the blinded two experienced microscopists.

Coimmunoprecipitation of PfRab5b-YFP-FLAG to Isolate PfRab5b Associated Candidate Proteins
Previous studies have shown that PfRab5b is localized adjacent to the ER and is involved in the trafficking of the N-myristoylated protein AK2 to the PVM (Ebine et al., 2016). In general, transport of newly synthesized transmembrane or luminal proteins from the ER is regulated by the Sar1 family of GTPases, but not by the Rab family (Nakano and Muramatsu, 1989). Proteins are packed into COPII-coated vesicles at the ERES and trafficked to the cis-Golgi cisternae under the control of the Rab1 and Arf1 GTPases Martinez et al., 2016). To investigate the mechanism controlled by PfRab5b from the ER, we attempted to identify PfRab5b interacting proteins by coimmunoprecipitation. Lysates prepared from PfRab5b expressing parasites, which were C-terminally fused with the YFP and FLAG-tags (PfRab5b-YFP-FLAG) or a negative control PfRab5b-YFP (Supplementary Figure 1A), were immunoprecipitated with an anti-FLAG antibody and eluted using the FLAG peptide. The eluted samples were analyzed via LC-ToF MS/MS, and 677 peptides were identified from the PfRab5b-YFP-FLAG expressing lysate (Supplementary Table  2). Six candidate proteins showing a 3.5-fold change in the expression of PfRab5b-YFP-FLAG relative to PfRab5b-YFP were obtained. These included the ADP-ribosylation factor (PfArf GTPase) (PF3D7_1020900), erythrocyte binding antigen-181 (EBA181) (PF3D7_0102500), protein transport protein SEC7 (PfSec7) (PF3D7_1442900), early transcribed membrane protein 10.2 (ETRAMP10.2) (PF3D7_1033200), PfRab1b GTPase (PF3D7_0512600), and the PVM protein S16 (Pfs16) (PF3D7_0406200) ( Table 1). In other organisms, the Arf GTPase, Sec7, and Rab1b are involved in the transport from the cis-Golgi to the ER (Monetta et al., 2007). Another Plasmodium-specific candidate protein EBA181 is a ligand localized on the surface of merozoites and plays an important role in the entry of parasites into erythrocytes by binding to surface receptors on erythrocyte cell membranes (Gilberger et al., 2003). ETRAMP10.2 is expressed at an early point of the intraerythrocytic stage and is localized at the parasite periphery, which is assumed to be the PVM . The Pfs16 is expressed in gametocytes and is localized at the parasite periphery, similar to ETRAMP10.2 (Bruce et al., 1994). In this study, we focused on the functions of the PfArf and PfRab1b GTPases to elucidate the mechanism of intracellular transport mediated by PfRab5b. Studies on the other candidate proteins will be described elsewhere.

PfArf1 and PfRab1b Colocalize With PfRab5b
In the P. falciparum genome, 11 Rab GTPase-encoding genes have been identified (Quevillon et al., 2003). Among them, PfRab1a and PfRab1b have been reported as two human Rab1 homologs (Quevillon et al., 2003). PfRab1a is localized to the ER and regulates trafficking from the ER to the apical organelles known as rhoptries (Morse et al., 2016). However, there are no reports for PfRab1b function and its subcellular localization. For the Sar/Arf family, the Sar1 homolog PfSar1 alone has been shown to localize the ER and define network membranes surrounding the parasite nuclei (Adisa et al., 2007). A BLASTP search revealed the presence of six Sar/Arf proteins in the P. falciparum 3D7 genome database (Supplementary Figure 2A). Amino acid sequencing showed that five of these proteins are Arf family GTPases (PF3D7_1020900, PF3D7-1034700, PF3D7_1442000, Pf3D7_0920500, Pf3D7_1316200), and one is a Sar1 homolog (PF3D7_0416800) based on the conserved effector sequence and the overall amino acid identities (Supplementary Figure 2C). PF3D7_1020900, which was obtained as PfRab5b interacting protein (Table 1), showed the highest identity (75%) to human Arf1 among the Plasmodium and human Arf families (Supplementary Figure 2B). Thus, the PfRab5b-associated protein candidate PF3D7_1020900 is a Plasmodium Arf1 homolog and was annotated as PfArf1 (GenBank Accession number, BR001667). The intracellular colocalization of PfRab5b with PfArf1 or PfRab1b was demonstrated using double-expressing parasites. First, the constitutively active mutant PfRab5b Q94L was fused with YFP and a DD (PfRab5b Q94L -YFP-DD) and expressed in a Shld1 ligand-dependent manner (Armstrong and Goldberg, 2007;Ebine et al., 2016). The Q-to-L substitution in the conserved GTP binding consensus domain impairs intrinsic GTPase activity which favor formation of the active GTP-bound form to small GTPases (Der et al., 1986: Stenmark et al., 1994. As conventional Rab GTPases are modified with Cterminal geranyl-geranylation (Joberty et al., 1993), PfRab1b was fused with the N-terminal RFP (RFP-PfRab1b). In contrast, Arf1 is modified with N-terminal myristoylation (Sewell and Kahn, 1988), and PfArf1 was fused with a Cterminus RFP fusion (PfArf1-RFP). PfRab5b Q94L -YFP-DD and RFP-PfRab1b, or PfArf1-RFP were placed under the control of the Pbef1a dual promoter and the constructs were transformed into parasites (Supplementary Figure 1B). Immunoblots using an anti-RFP antibody showed the 48 and 50 kDa bands of PfArf1-RFP and RFP-PfRab1b, respectively (Supplementary Figure 3). The anti-GFP antibody detected a 62 kDa band corresponding to PfRab5b Q94L -YFP-DD, indicating that the full length of fusion constructs were expressed (Supplementary Figure 3). In the mononuclear early trophozoite stage, 2 or 3 nuclear late trophozoite, and multinucleated early schizont stages, the RFP signals of PfArf1-RFP and RFP-PfRab1b were observed as juxtanuclear punctate structures, and these colocalized with the YFP fluorescence from PfRab5b Q94L -YFP-DD ( Figures 1A, B). Previous report showed that PfRab5b localized near the parasite plasma membrane at the schizont stage (Ezougou et al., 2014). We additionally showed the localization of PfRab5b to the ER and PVM between ring and late schizont stages (Ebine et al., 2016). Thus we focus on the localization of PfArf1 and PfRab1b between early trophozoite and early schizont stages. PfRab5b Q94L -YFP-DD showed good colocalization with PfArf1-RFP (average Pearson's correlation coefficient: R = 0.51 ± 0.088. n = 10 parasites). PfRab1b-RFP showed mild colocalization with PfRab5b Q94L -YFP-DD, as analyzed by YFP and RFP signals (average Pearson's correlation coefficient: R = 0.34 ± 0.05, n = 10 parasites). This result indicated that PfArf1-RFP rather than PfRab1b-RFP closely associated to PfRab5b Q94L -YFP-DD (Supplementary Figure 4). The interaction of PfRab5b Q94L -YFP and RFP-PfArf1 was confirmed by reciprocal coimmunoprecipitation of RFP-PfArf1 and PfRab5b Q94L -YFP double-expressing parasites with anti-RFP antibody ( Figure 1C). The immunoprecipitated PfRab5b Q94L -YFP was recognized as 62 kDa band using mouse anti-GFP antibody together with RFP-PfArf1 visualized with mouse anti-RFP antibody, while the negative control marker cytosolic protein Hsp90 was not detected in the sample. The interaction between PfRab5b Q94L -YFP and PfRab1b-RFP was not confirmed in this reciprocal coimmunoprecipitation (data not shown).

PfArf1 and PfRab1b Were Localized in Different Subdomains of the ER and the cis-Golgi
In an analysis conducted previously, we have shown that PfRab5b was localized adjacent to the ER (Ebine et al., 2016), which was labeled with the ER luminal chaperone PfBiP (Kumar et al., 1991;Kumar and Zheng, 1992). To examine the subcellular localization of PfArf1 and PfRab1b, which were colocalized with PfRab5b ( Figure 1B), PfArf1-RFP, or RFP-PfRab1b expressing parasites (Supplementary Figure 1C) were stained with the anti-PfBiP antibody. Punctate structure signals for PfArf1-RFP and RFP-PfRab1b were closely localized with the PfBiP signals ( Figure 2A). More than 70% of PfArf1-RFP expressing parasites showed colocalization of the PfArf1-RFP and PfBiP signals (71 ± 11%). This proportion is higher than that of RFP-PfRab1b and PfBiP in RFP-PfRab1b expressing parasites (46 ± 7%, p < 0.05) ( Figure 2B). This result was unexpected because the Arf1 and Rab1 GTPases were previously reported to be localized and targeted to the cis-Golgi in most other organisms (Stearns et al., 1990;Moyer et al., 2001). Next, PfArf1-RFP and RFP-PfRab1b expressing parasites were stained with an anti-PfERD2 antibody, which stained the Plasmodium homolog of the cis-Golgi membrane protein ERD2 Elmendorf and Haldar, 1993). Most of the PfArf1-RFP expressing parasites did not show colocalization of the PfArf1-RFP and PfERD2 signals (32 ± 9%) ( Figures 2C, D). The ratio of PfRab1b colocalization with PfERD2 was increased to 59 ± 3% in PfRab1b-RFP expressing parasites (p < 0.05) ( Figures 2C, D). These results indicate that both PfArf1 and PfRab1b simultaneously localize to the ER and cis-Golgi in this organism; however, the subcellular localization of fraction differed between PfArf1 and PfRab1b, as most of the PfArf1 was localized to the proximal region of the ER, and half of PfRab1b was individually localized to the ER and the cis-Golgi.
Detailed analysis using super-resolution microscopy enabled the identification of the distinct subcellular localization of PfArf1 and PfRab1b on the ER, and whether both GTPases localize to the same subdomain or reside in distinct regions. The immunostained slides were processed with a super-resolution microscope LSM880 with Airyscan and processed with Zeiss Zen2 software, which provides a lateral resolution of 140 nm, to analyze the precise cellular locations of the proteins. Peak signal intensities of the most proximal staining between PfBiP and PfArf1-RFP ( Figure 3A) or PfRab1b ( Figure 3B) were calculated. The average distance from PfBiP was closer to PfArf1-RFP than to RFP-PfRab1 (PfArf1-RFP: 0.33 ± 0.08 µm vs. RFP-PfRab1 0.45 ± 0.11 µm, p < 0.001) ( Figure 3C). These data indicate the presence of compartmentalization in the ER or ER adjacent novel membrane structures, and PfArf1 showed significant localization close to the PfBiP-positive ER rather than PfRab1b.

PfArf1 and PfRab1b Are Involved in the Transport of the PEXEL-Positive Transmembrane Protein Rifin
Blood stage parasites export several proteins into the host erythrocyte cytosol and the PV (Sargeant et al., 2006;van Ooij et al., 2008). The PEXEL sequence is a five-residue motif in the Immunoprecipitated PfArf-RFP (a white arrowhead) and PfRab5b Q94L -YFP-DD (a black arrowhead) was visualized with mouse anti-RFP or anti-GFP antibodies, respectively. In the absence of rabbit anti-RFP antibody during immnoprecipitation (IP: −), neither PfArf-RFP nor PfRab5b Q94L -YFP-DD was detected. Anti-Hsp90 antibody was used as a negative control. Two 50 kDa bands in pull down fraction (an asterisk) were non specific recognition of secondary antibody against antirabbit IgG. downstream N-terminal signal peptide, and it has been detected in many exported proteins (Hiller et al., 2004;Marti et al., 2004). PEXEL-positive proteins have been suggested to pass through the classical ER/Golgi pathway (Akompong et al., 2002). However, the presence of several PNEPs indicates the existence of an alternative export pathway (Möskes et al., 2004;Thavayogarajah et al., 2015). We have previously shown that overexpression of PfRab5b did not disrupt the export of the PEXEL-positive transmembrane protein EVP1 to the iRBC cytosol (Ebine et al., 2016). To examine whether PfArf1 and PfRab1b are involved in trafficking of PEXEL-positive export proteins, we chose a Rifin variant PFA0745w, whose fusion construct with YFP was secreted into the erythrocyte cytosol (Marti et al., 2004). The N-terminal PEXEL domain and the Cterminal transmembrane region were fused with RFP (Rifin-RFP) and co-expressed with the PfArf1 and PfRab1b mutant constructs, whose expression was driven by the Shld1 ligand (Supplementary Figure 1D). In parasites expressing the PfArf1 WT -YFP-DD and active mutant PfArf1 Q71L -YFP-DD constructs, Rifin-RFP signals were detected in the iRBC plasma membrane and at the parasite periphery (PfArf1 WT : 91 ± 4%, PfArf1 Q71L : 88 ± 6%) ( Figures 4A, B). In contrast, the export of Rifin-RFP was reduced in the inactive PfArf1 T31N -YFP-DD expressing parasites (53 ± 12%) ( Figures 4A, B), suggesting that PfArf1 is involved in PEXEL-positive Rifin transport. Similarly, the expression of wild-type DD-YFP-PfRab1b and the active mutant DD-YFP-PfRab1b Q67L did not show differences for the export of Rifin-RFP (PfRab1b WT : 89 ± 2%, PfRab1b Q67L : 90 ± 10%), whereas the expression of the inactive mutant DD-YFP-PfRab1b S22N reduced the export activity (39 ± 9%) ( Figures 4C, D).  transport of PfAK2 (Ebine et al., 2016). Therefore, we examined whether the overexpression of PfArf1 perturbs the transport of PfAK2-RFP to the PVM. Parasites that double-expressed PfAK2-RFP and PfArf1 WT , or the constitutively active PfArf1 Q71L or inactive PfArf1 T31N mutants ( Figure 1A), whose expression is driven by the Shld1 ligand were established (Supplementary Figure 1D). The PfArf1 Q71L and PfArf1 T31N mutants are corresponding to human Arf1 Q71L and Arf1 T31N , respectively (Dascher andBalch, 1994: Teal et al., 1994). Expression of PfArf1 Q71L or PfArf1 T31N mutants for 48 h did not show growth defect (Supplementary Figure 5). In the PfArf1 WT -YFP-DD expressing parasite, all PfAK2-RFP signals were localized at the parasite periphery, indicating a typical PVM staining pattern ( Figure 5A). In contrast, the active mutant PfArf1 Q71L -YFP-DD expressing parasites reduced PfAK2-RFP targeting to the PVM (38 ± 6.7%). Several parasites showed a faint RFP signal and a punctate RFP signal within the parasite cytoplasm (faint: 35 ± 7.1%, punctate: 27 ± 7.7%, respectively) ( Figures 5A, B). In the inactive PfArf1 T31N -YFP-DD expressing parasite, the transport of PfAK2-RFP was reduced to 58 ± 6.8%, and further, 13 ± 7.2% and 29 ± 11% of the parasites showed a faint RFP signal and a punctate pattern in the cytoplasm, respectively ( Figures 5A, B). The faint signal of PfAK2-RFP was more abundant in PfArf1 Q71L -YFP-DD than in PfArf1 T31N-YFP-DD (p < 0.05). These results indicate that PfArf1 is directly involved in the trafficking of PfAK2 to the PVM. The specific role of PfArf1 in the transport of PfAK2 was highlighted by the co-expression of PfRab1b mutants ( Figure 5C). In the co-  expression with the wild-type DD-YFP-PfRab1b, active DD-YFP-PfRab1b Q67L , or the inactive DD-YFP-PfRab1b S22N constructs, which corresponds human Rab1b Q67L and Rab1b S22N , respectively (Tisdale et al., 1992), the transport of PfAK2-RFP was not inhibited and all parasites showed a peripheral pattern for their expression (PfRab1b WT : 99 ± 2%, PfRab1b Q67L : 96 ± 4%, PfRab1b S22N : 94 ± 3%) ( Figures 5C, D). Expression of Rab1b Q67L and Rab1b S22N mutants for 48 h did not show growth defect (Supplementary Figure 5). These results suggest that PfArf1 is extensively involved in the transport of the N-acylated protein PfAK2 to the PVM.

Isolation of PfArf1 and PfRab1b as PfRab5b Associated Proteins
In this study, we demonstrated the isolation of PfRab5b proteins using a coimmunoprecipitation approach. We identified two GTPases, PfRab1b and the human Arf1 homolog PfArf1, that colocalized with PfRab5b and were associated with the ER marker PfBiP in erythrocytic stage parasites. Further, we demonstrated that PfArf1 and PfRab1b are closely located near the ER, but the precise localization differed as shown by superresolution microscopy, suggesting that the ER-Golgi interface in Plasmodium is highly compartmentalized. Additionally, we provide direct cell biological evidence that PfArf1 and PfRab1b are involved in different types of cargo selection. PfArf1 regulates the transport of N-acylated PfAK2 to the PVM, whereas PfRab1b controls the trafficking of the PEXEL-positive exported protein Rifin in the erythrocyte cytosol. Thus, PfRab5b and its associated GTPases are involved in the sorting of several families of exported proteins in the different ER subdomains. Rab GTPase recycles GTP-bound active and GDP-bound inactive forms (Hutagalung and Novick, 2011). To immunoprecipitate efficiently, we used DSP cross-linker, whose spacer arm length 12.0 Å, before the coimmunoprecipitation to fix the PfRab5b-nucleotide bound state. Such approach, the use of cross-linker before coimmunoprecipitation, identifies many (> 100) proteins in LC-ToF MS/MS (Watanabe et al., 2020). Our approach also identified 677 proteins and PfArf1 was listed as highly enriched proteins among 677 proteins (Supplementary The rate of parasites that showed a Rifin-RFP signal was detected in the erythrocyte cytoplasm in PfArf1-RFP (B) and RFP-PfRab1b (D) expressing cells are shown in graphs. Thirty individual early trophozoites and early schizonts were counted from three independent experiments. Infected RBCs, recognized by the DAPI and YFP signals under the microscope, were imaged by the laser microcopy, and then analyzed for the localization of RFP and whether Rifin-RFP was exported to the iRBC. The statistical significance was determined using the Student's t-test.  Table 2). Contrary, a casein kinase CK1, which is identified as PfRab5b binding protein by bioinformatic technique (Rached et al., 2012), was not enriched in our approach (Accession Number, C6S3F7-1, Supplementary Table 2). Colocalization of PfRab5b and CK1 at parasite periphery was reported in schizont stage (Ezougou et al., 2014). These observations suggested that PfRab5b and CK1 may localized to different membrane subdomain on PVM in schizont stage. Further biological and biochemical studies are needed to prove direct and indirect interaction between PfRab5b and effector proteins.

PfArf1 Exports the N-Myristoylated Protein PfAK2 to the PVM
The regulatory mechanisms underlying the trafficking of acylated proteins are not clearly understood in Plasmodium and other organisms. Dual acylated proteins are first myristoylated at the ER membrane by N-myristoyl transferase after palmitoylation by a palmitoyltransferase, and the acylated proteins are then trafficked to the apical organelle and parasite plasma membrane surface (Cabrera et al., 2012). For other proteins, such as the Plasmodium falciparum calcium-dependent protein kinase 1 (PfCDPK1) and the Drosophila transgultaminase A (TG-A), the dual acylated proteins are packed into multivesicular bodies and are subsequently exported to the apical organelle or extracellular space via the unconventional ER-Golgiindependent pathway (Möskes et al., 2004;Shibata et al., 2017). We have previously shown that the transport of Nmyristoyl PfAK2 to the PVM was inhibited by the overexpression of PfRab5b, and that PfAK2 accumulated in the punctate structure within the parasite cytoplasm together with PfRab5b, indicating that PfRab5b and PfAK2 were included in the internal vesicle of the multivesicular body and were then transported to the PVM (Ebine et al., 2016). In this study, we have shown that PfArf1, but not PfRab1b, is involved in the regulation of PfAK2 (Figures 4 and 5). Our results indicated that the expression of active or inactive PfArf1 mutants inhibited (C, D) were examined using the immunofluorescence assay. Fluorescence signals from RFP (red), YFP (green), and DAPI (blue) are shown. Wild-type PfArf1 or PfRab1b (upper panels), the active mutant PfArf1 Q71L or PfRab1b Q67L (middle panels), and the inactive mutant PfArf1 T31N or PfRab1b S22N (lower panels) are shown. Representative images for PfAK2-RFP are shown and are divided into three patterns: peripheral PVM staining (PVM), faint signal (faint), and dot-like punctate signal within the parasites (punctate). The bars indicate 2 µm. The parasites that showed a signal for PfAK2-RFP were classified into three patterns based on the PfArf1-RFP expressing cells (B) and then shown in bar graph. The rate of parasites showing peripheral staining for PfAK2-RFP in RFP-PfRab1b expressing cells (D). Thirty individual early trophozoites and early schizonts were counted from three independent experiments. Infected RBCs, recognized by the DAPI and YFP signals under the microscope, were imaged by the laser microcopy, and then analyzed for the localization of RFP signal as PV, faint, and punctate. The statistical significance was determined using the Student's t-test.
PfAK2 transport to PVM ( Figure 5A), whereas the expression of active or inactive PfRab1b mutants had no effect ( Figure 5C). These results suggest that GTP hydrolysis by PfArf1 is required for proper PfAK2 recognition and further transport. For mammalian Arf1, it has been biochemically demonstrated that GTP hydrolysis by human Arf1 promotes the selective cargo selection from the Golgi membrane, and the concentration of cargo proteins into the COPI-coated vesicles (Lanoix et al., 1999). This observation indicated that the role of GTP hydrolysis in cargo selection is similar between PfArf1 and the human Arf1.

Presence of Subdomains Near the ER and the Sequential Roles of PfArf1 and PfRab1b in Cargo Selection
Both PfArf1 and PfRab1b are localized close to the ER as punctate structures in the early erythrocytic stage ( Figure 1B). The punctate structure near the ER was observed in other proteins, such as the COPII coat components PfSec13 and PfSec24 (Lee et al., 2008;Struck et al., 2008) and in another PfRab1 isotype PfRab1a (Morse et al., 2016). Previously, we have shown that PfRab5b did not colocalize with the COPII component PfSec13 (Struck et al., 2008;Ebine et al., 2016), suggesting that PfArf1 and PfRab1b might be localized in different domains from the COPII vesicle budding site, the ERES. In this study, the results indicate that the PfArf1 signal was not completely consistent with that for PfRab1b and was localized to different membrane subdomains around the ER (Figure 3). These findings suggest that the previously reported COPII component, PfArf1, and PfRab1b might be localized to an independent subdomain and may have different roles in the transport of cargo proteins and cargo selection from the ER. PfRab1b was involved in the transport of the PEXEL-positive export protein, Rifin, and the expression of the inactive PfRab1b mutant showed decreased activity of Rifin export to erythrocyte cytosol ( Figures 4C, D). In contrast, the expression of the active PfRab1b mutant did not inhibit Rifin export, indicating that the GTP-bound state of PfRab1b may be necessary for the proper functioning of PfRab1b. A similar observation was reported in mammalian Rab5 during endosome fusion, where it was demonstrated that the GTP-bound mutant form showed the same effect as the wild-type Rab (Barbieri et al., 1996).
Interestingly, the export of Rifin was also decreased by the expression of inactive mutants of PfArf1, but not the active GTPfixed mutant and the wild-type PfArf1, which cycles GTP-and GDP-bound states ( Figures 4A, B). This result suggests that the GTP-bound state of PfArf1, but not GDP-fixed mutant, is necessary for the export of Rifin, or for the correct activity of PfRab1b. Although not included in this study, it is notable that a Sec7 domain containing protein was listed as a candidate for the FIGURE 6 | A proposed model for the two pathways regulated by PfArf1 and PfRab1b. PfRab5b (blue circle) recruited PfArf1 (red circle) probably through the binding to PfArf1 activating protein (light grey ellipse) on the ER membrane ( Figure 1). Membrane targeted PfArf1 selected the N-acylated AK2 and sorted into the pathway for the PV ( Figure 5). According to the partial involvement of PfArf1-positive membrane for the selection of PEXEL-positive Rifin (Figure 4), and PfRab1b (yellow circle) localized to distant membrane from the ER (Figure 3), PfArf1-positive membrane might be matured into PfRab1b-positive compartment. PfRab1bpositive compartment selected and further transport the PEXEL-positive Rifin to the RBC surface. ER, endoplasmic reticulum; ERES, endoplasmic reticulum exit site; PV, parasitophorous vacuole; RBC, red blood cell. PfRab5b binding protein using mass analysis ( Table 1). The 200 amino acid residue Sec7 domain is conserved among eukaryote and has the guanine nucleotide exchange activity toward Arf (Arf GEF) (Cherifils et al., 1998). Sec7 binds GDP-bound wild-type Arf GTPase and promotes the exchange of GDP for GTP. In other organisms, Sec7 is involved in the "Rab cascade": activated GTPase triggers the recruitment of GEF for the downstream GTPase, and thus a series of GTPase activations is feasible (Barr, 2013). For example, in Saccharomyces cerevisiae, Ypt1 (in yeast Rab1) and the Arf-like GTPase Arl1 recruit Sec7 to the Golgi membrane, and subsequently Arf1 is activated by the GEF activity of membrane-localized Sec7, following the stimulation of Ypt31 (yeast Rab11) on the trans-Golgi membrane for cargo sorting to secretory vesicles (McDonold and Fromme, 2014). Thus, we generated a sequential hypothetical model of PfArf1 and PfRab1b adjacent to the ER that is presented in Figure 6. Membrane-localized PfRab5b may recruit PfSec7 to exchange GDP for GTP of PfArf1 in the adjacent ER subdomain, where the N-myristoylated protein PfAK2 is selected and packed into the pathway destined for the PVM. This is based on the finding that colocalization of the active mutant of PfRab5b and PfArf1 (Figure 1), and wild-type PfArf1 is necessary for the transport of PfAK2 ( Figure 5A). It has been shown that Plasmodium PfArf1 possesses GTPase activity in vitro (Stafford et al., 1996) and that PfSec7 accelerated the nucleotide exchange activity against PfArf1 (Baumgartner et al., 2001). Thus, the PfArf1 may colocalize with PfSec7 adjacent to the ER. The role of PfSec7 in intracellular traffic remains elusive, and further studies are needed to unravel the regulation of PfAK2 transport together with PfArf1. Subsequently, the PfArf1 positive membrane matures and recruits PfRab1b at the membrane. This model is based on our results that indicate that PfRab1b localizes to the different subdomains of PfArf1 that are adjacent to the ER ( Figures 1B and 3), and that the sequential recruitment of different GTPases to the membrane occurs via Sec7 (McDonold and Fromme, 2014). In this hypothesis, PfSec7 could not recruit on the membrane in GDP-fixed PfArf1 mutant and may showed the defect in the function of downstream PfRab1b, suggesting expression of inactive mutant of PfArf1 showed the reduction of the export of Rifin ( Figures 4A, B). The Rab1 positive membrane subdomain near the ER is found in mammalian tubulovesicular membrane clusters of the ER-Golgi intermediate compartment (ERGIC) (Appenzeller-Herzog and Hauri, 2006). ERGIC clusters lie close to the COPII-positive ERES. Transport from the ER to the ERGIC is controlled by COPII coat vesicles, and Rab1 is involved in membrane tethering at the ERGIC in anterograde transport (Allan et al., 2000). Sorting in the ERGIC involves another protein coat of COPI and the Arf family (Goldberg, 2000). Thus, ERGIC is a sorting and recycling platform for the transport between the ER and Golgi in mammalian cells. In Plasmodium, the PfRab1b-positive compartment was involved in the export of the PEXEL-positive protein Rifin (Figure 4) and this suggests the existence of a novel sorting compartment for PEXEL-positive cargo proteins. The model shown in Figure 6 indicates the segregation of the subdomain for sorting of different cargo proteins. Conversely, the expression of active and inactive PfRab1b mutants did not affect the export of PfAK2 ( Figure 5).

Transport to the Golgi and the Presence of a Few Parts of PfArf1 on the PfERD2-Positive Golgi Membrane
In most mammalian secretory cells, electron microscopy has revealed the presence of ribosome-coated rough ER and partly smooth surfaced structures in the vicinity of the Golgi complex (Saraste and Kuismanen, 1992). Golgi is displayed as stacks of flattened cisternae, which are often laterally linked into a ribbonlike structure (Zhang and Wang, 2016). In contrast, the ER and Golgi of malarial parasites are not well characterized and are reported to be loosely associated vesicles (Aikawa, 1971). To overcome the difficulty in the visualization of ER and the Golgi, we used super-resolution fluorescence microscopy ( Figure 3) and quantified the number of colocalized parasites (Figures 2-5). In this study, about 70% of parasites showed colocalization of PfArf1 and the ER marker PfBiP, and 30% showed colocalization with the cis-Golgi marker PfERD2 ( Figure 2). These results may explain the following two possibilities. One; PfArf1 is a Golgi resident which is involved in retrograde trafficking from the Golgi. This model is based on reports from mammalian and yeast models, where Arf1 is involved in different steps of cargo sorting together with specific effector or Sec7 containing proteins such as retrograde traffic from the cis-Golgi to the ER, or the secretory pathway from the trans-Golgi (Donaldson and Jackson, 2011). As it is not yet characterized, the Golgi resident PfArf1 may be involved in retrograde trafficking to the ER because the homolog of ERD2 is present and localized to the Golgi in Plasmodium (Elmendorf and Haldar, 1993). ERD2 retrieves the conserved Cterminal tetrapeptide sequence HDEL-containing ER luminal proteins from the cis-Golgi in yeast and mammalian cells (Hsu et al., 1992;Townsley et al., 1994). A second possibility is the presence of the Golgi subdomain in Plasmodium. The Plasmodium ERD2 was previously reported to colocalize with the Golgi re-assembly stacking protein (GRASP) (Struck et al., 2005), and PfArf1 was shown to be colocalized with GRASP in the early trophozoite stage (Thavayogarajah et al., 2015). Therefore, it appears that PfArf1, GRASP, and PfERD2 are colocalized in the same Golgi membrane. However, mammalian GRASP55 is present in the medial/trans-cisternae of Golgi stacks as shown by cryo-immunoelectron microscopy (Langreth et al., 1978). Thus, this may be the reason for the Golgi-stacked protein GRASP and HDEL-receptor ERD2 to colocalize in the same Golgi subdomain in Plasmodium.

CONCLUSION AND FUTURE PERSPECTIVES
In conclusion, our results show that PfArf1 mediates the transport of N-myristoylated PfAK2 from the adjacent ER, and PfRab1b, which localizes differently than PfArf1, is involved in the export of PEXEL-positive Rifin to the erythrocyte cytosol. Currently, the mechanism how PfArf1 recognize cargo protein at the ER subdomain and sort to the pathway for the PVM, remains elusive. In other cases, there are very few reports on trafficking of dual acylated protein via a multivesicular body (Möskes et al., 2004;Shibata et al., 2017). In mammalian cells, myristoylated and palmitoylated GFP localized to the membrane subdomain enriched with cholesterol and ganglioside at the plasma membrane (McCabe and Berthiaume, 2001). It may plausible that PfArf1 together with PfRab5b and unidentified effector proteins organize the acylated cargo-recognition subdomain at the ER lipid subdomain. Consideration of the two facts that PfRab5b is essential for the growth (Ezougou et al., 2014;Ebine et al., 2016) and Nmyristoyl transferase is a promising drug target for malaria (Schlott et al., 2018), suggests that elucidation of further molecular mechanism on PfArf1 and its regulatory proteins may help the plasmodium biology as well as pathogenesis.

DATA AVAILABILITY STATEMENT
The data sets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
IT, TH, and YS-N conceived the study. IT, TH, TM, NS, SI, and YS-N designed and performed experiments and data analysis. IT, TM, TA, and YS-N drafted the paper. TA, KN, and TN participated in data analysis and edited the paper. YS-N acquired grants. All authors contributed to the article and approved the submitted version.

FUNDING
This research was funded by Grants-in-Aid for Scientific Research (C) (JP19K07531 to YS-N) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), and a grant from The Naito Foundation to YS-N.

ACKNOWLEDGMENTS
We thank Kiyoshi Kita (Nagasaki University) for providing the anti-PfBiP antibody, Osamu Kaneko (Nagasaki University) for the pCHD43(II) episomal plasmid, Daniel E Goldberg (Washington University School of Medicine) for the YFP-FKBP episomal plasmid, and Shusuke Nakazawa (Nagasaki University) for the P. falciparum MS822 line. The P. falciparum 3D7 line was obtained from MR4 (contributed by D. J. Carucci, MRA-102). We thank the Japanese Red Cross Society for providing human RBCs and plasma, and Jacobus Pharmaceuticals for WR99210. We thank all members of the Department of Parasitology, NIID, for various discussions related to the study. Supplementary Table 2 | List of proteins detected from immunoprecipitated samples prepared from PfRab5b-YFP-FLAG and PfRab5b-YFP expressing parasites using anti-FLAG antibody. The most enriched proteins were shown in red and listed in Table 1.

SUPPLEMENTARY MATERIAL
Supplementary Figure 1 | Schematic structure of the constructs used in this study.
Supplementary Figure 2 | Pf3D7_1020900 is a mammalian homolog of Arf1.
(A) Amino acid percentage identity matrix of the six Plasmodium Sar/Arf familiy of GTPases and human Arf1 and Sar1. Plasmodium protein sequences for the Sar/Arf family of GTPases were retrieved from the PlasmoDB genome database (release 46, Nov 2019) using the Pathema Bioinformatics Resource Center (https:// plasmodb.org/plasmo/) (Collaborative, 2001). Plasmodium Sar/Arf homologs were retrieved via the BLASTP algorithm using the human Arf1 and Sar1 GTPases as queries in the P. falciparum 3D7 database. (B) Sequence alignment of Homo sapiens Arf1 and Sar1 and six Plasmodium Sar1/Arf families of proteins. Amino acid sequences were aligned using Clustal Omega (EMBL-EBI, Welcome Genome Campus). The N-terminus myristoylation site with a glycine residue specific for the Arf family and the GTP-binding consensus boxes are indicated in blue and red boxes, respectively (Pereira-Leal and Seabra, 2000). The effector regions of the Arf/Arf-like (Arl) and Sar1 families are indicated in light blue and magenta, respectively. Note that PF3D7_1442000 and PF3D7_1316200 lack an N-terminal glycine residue. PF3D7_0920500 and PF3D7_1316200 are missing the second GTP-binding box. Two conserved residues, that are substituted to create constitutively active and inactive mutants, are indicated by black and white arrowheads, respectively. (C) Amino acid percentage identity matrix of Pf3D7_1020900 and the five human Arf1.
Supplementary Figure 3 | Immunoblots showing the expression of full length of PfArf1-RFP, and RFP-PfRab1b. Parasite lysates expressing PfRab5b Q94L -YFP-DD and Arf1-RFP or RFP-PfRab1b were subjected to immunoblot analysis using anti-RFP and anti-GFP antibodies. Anti-Hsp90 antibody was used as a loading control. Mock is the non-transformant parasite.
Supplementary Figure 4 | Pearson's correlation coefficient between PfRab5b Q94L -YFP-DD and PfArf1-RFP or RFP-PfRab1b. The fluorescence intensities of the merged YFP and RFP signals were analyzed using Fiji-Image J software to generate Pearson correlation coefficients. The scatter plot represents Pearson correlation coefficients values from ten independent parasites, which are Taku et al. Rab5b-Associated Proteins in Plasmodium obtained from three independent transfectants. The statistical significance was determined using Student's t-test.
Supplementary Figure 5 | Expressions of PfArf1 or PfRab1b mutants showed no effect on parasites growth. The population doubling time of non-transformant MS822 line, PfArf1 and PfRab1b mutants were measured for 2 days cultivation from three independent assays. The average (bar graph) and the standard deviation (thin bar) are indicated. Mock was cultured without drugs and PfArf1, PfRab1b mutants were cultured with BSD. Except from MS822 line, transformed parasite were cultured in the presence of 2.5 µg/ml of BSD and the expressions of PfArf1 and PfRab1b mutants were stabilized by 0.5 µM Shld1. A statistical significance was evaluated using Student's t-test.