Rhoptry kinase protein 39 (ROP39) is a novel factor that recruits host mitochondria to the parasitophorous vacuole of Toxoplasma gondii

ABSTRACT Most intracellular pathogens replicate in a vacuole to avoid the defense system of the host. A few pathogens recruit host mitochondria around those vacuoles, but the molecules responsible for mitochondrial recruitment remain unidentified. It is only in the apicomplexan parasite Toxoplasma gondii, that mitochondrial association factor 1b (MAF1b) has been identified as an association factor for host mitochondria. Here, we show that rhoptry kinase family protein 39 (ROP39) induces host mitochondrial recruitment in T. gondii. We found that the abundance of ROP39 was increased on host mitochondria extracted from human foreskin fibroblasts (HFFs) infected with T. gondii. ROP39 expressed exogenously in HFFs localized on host mitochondria, indicating that it has the potential to bind to host mitochondria without assistance from other parasite factors. Confocal microscopy revealed that ROP39 colocalized with host mitochondria on the membrane of parasitophorous vacuoles, in which the parasites reside. Moreover, we observed about a 10% reduction in the level of mitochondrial association in rop39-knockout parasites compared with a parental strain.


INTRODUCTION
Toxoplasma gondii is an obligate intracellular protozoan parasite that can infect almost all warm-blooded animals, and nearly onethird of the world's population is infected with this parasite (Grigg and Sundar, 2009). Toxoplasma gondii causes a latent infection in most humans, but leads to lethal diseases, including encephalitis, in immunosuppressed people due to acquired immunodeficiency syndrome (AIDS) or organ transplantation. In pregnant women, initial infection with T. gondii may cause fetuses to encounter the parasites through vertical transmission, and this can result in serious symptoms such as retinochoroiditis, hydrocephalus and psychomotor retardation (Montoya and Liesenfeld, 2004).
Toxoplasma gondii is sequestered in host cells by the parasitophorous vacuole (PV), which permits intracellular replication of parasites. It is well known that T. gondii associates host mitochondria and endoplasmic reticulum (ER) around PVs (Sinai et al., 1997). Such mitochondrial association is also observed in other intracellular pathogens, such as Legionella pneumophila, Chlamydia psittaci and Encephalitozoon cuniculi (Horwitz et al., 1983;Matsumoto et al., 1991;Scanlon et al., 2004), whose molecular mechanism and function remains largely unclear. In a recent quantitative trait locus (QTL) analysis of T. gondii, mitochondrial association factor 1b (MAF1b) was identified in T. gondii as a parasite factor that associates host mitochondria around its PV membranes (PVMs) (Pernas et al., 2014). However, which factor(s) recruits host mitochondrial to PVMs remains unknow.
Rhoptry kinase family proteins (ROPs) are secretory proteins that are discharged into host cells from an apicomplexan parasite organelle called a rhoptry. Following the formation of PVMs, these molecules localize on PVMs or in PVs or host nuclei, and modify host function to optimize the milieu around parasites. ROP2 and ROP8 were suggested to induce mitochondrial association in an antisense-RNA based study (Sinai and Joiner, 2001), but this was not supported by use of a rop2a/rop2b/rop8 triple knockout mutant (Pernas and Boothroyd, 2010). However, ROPs are still fascinating molecules with regard to their relevance in mitochondrial recruitment. ROPs are secreted into host cells immediately after invasion by T. gondii (Boothroyd and Dubremetz, 2008), and this corresponds to the observation that host mitochondrial association occurs within a few minutes after penetration of host cells (Sinai and Joiner, 2001). Thus, secretion of ROPs is coincident with onset of mitochondrial association. Moreover, T. gondii discharges many evacuoles containing ROPs into glycosylphosphatidylinositoldeficient cells, and the mitochondrial association is enhanced in these cells (Tahara et al., 2016). These observations may indicate a link between ROPs and mitochondrial recruitment. Considering MAF1b is a dense granule protein which is secreted after invasion, we speculated certain ROPs secreted into host cell prior to invasion may work to recruit host mitochondria to PVMs.
In this study, we search for ROPs whose abundance are upregulated on host mitochondria in T. gondii infection to find new factor(s) to induce recruitment of host mitochondria using a quantitative proteomics method and ROP39 was identified. As a result of the further analysis, we found ROP39 as a causative factor in host mitochondrial recruitment.

Increased abundance of ROP39 on host mitochondria during infection
To explore ROPs involved in host mitochondrial recruitment, we employed a quantitative proteomics method: isobaric tag for relative and absolute quantitation (iTRAQ) (Wiese et al., 2007) to identify ROPs with increased levels on host mitochondria. Host mitochondria were extracted from parasite-infected cells, and proteins were purified from them. The proteins were purified with the same way from a mixture of HFFs and parasites as a control to exclude non-specific binding. A total of 320 proteins of T. gondii were detected with an amino acid sequence of at least one peptide fragment that matched the reference sequence with >95% confidence. Among these proteins, eleven ROPs were found and only ROP39 (TGGT1_262050) was more abundantly detected from mitochondria isolated from T. gondii-infected cells (Table 1). We performed evacuole assays to check that ROP39 actually was injected into host cytosol prior to PVMs formation. We confirmed that cytochalasin treatment blocked the invasion of parasites but not the secretion of vacuole (evacuole), which indicates that ROP39 secretion occurs followed by the building up of PVMs (Fig. S1).

ROP39 localizes on host mitochondria in mammalian cells
To assess whether localization of ROP39 is due to its inherent properties or to the effects of other parasite-derived factors, we investigated the molecular structure of ROP39 to use a domain prediction tool. ROP39 has been predicted to have a signal peptide at amino acid positions 1-49 by SP-HMM/SP-NN (https://toxodb. org/toxo/). Moreover, the presence of a mitochondrial targeting sequence (MTS) in ROP39 was estimated on Mitoplot, a prediction tool to identify mitochondrial pre-sequence and cleavage sites (Fukasawa et al., 2015). A cleavage site in mitochondria was predicted for the peptide bond between amino acids 63 and 64 (Fig. 1A). We expressed ROP39 lacking the signal peptide (ROP39δSP) exogenously in human foreskin fibroblasts (HFFs) using a mammalian expression vector. FLAG-tagged ROP39δSP was found to localize partially on host mitochondria (Fig. 1B). The pattern of partial colocalization of ROP39 and host mitochondria was similar to that of counter staining of ROP2 and mitochondria (Sinai and Joiner, 2001). Furthermore, we generated a vector to express ROP39 lacking the N-terminal 63 amino acids (ROP39δSPδMTS) based on the prediction of Mitoprot. The expressed protein ROP39δSPδMTS was scattered around the cytosol in HFFs (Fig. 1B). These findings suggest the possibility that ROP39 bind to host mitochondria in the cytosol through its MTS prior to the formation of PVMs.

ROP39 colocalizes with host mitochondria on PVMs in T. gondii
To investigate whether ROP39 is actually targeted to PVMs, we confirmed the localization of ROP39 in infected cells with T. gondii. ROP39 was found to be localized on PVMs and to partially colocalize with host mitochondria in cells ( Fig. 2A). To further investigate the localization of ROP39, we labeled ROP39 with MAF1b (PVMs marker) by conducting an immunostaining (Fig. 2B). We observed that ROP39 colocalized with MAF1b on PVMs (Fig. 2B). These findings support the localization of ROP39 on PVMs and are consistent with the localization of ROP39 on host mitochondria indicated by the data from iTRAQ and Fig. 1.
A rop39-knockout mutant exhibits decreased association of host mitochondria To show the relevance to ROP39 in the mitochondrial recruitment, we generated a rop39 deficient mutant RHΔrop39 in the RH strain  background using a CRISPR/Cas9 system (Fig. 3A) and deletion of ROP39 was checked by PCR (Fig. 3B). We measured ratio of the perimeter of PVs in contact with host mitochondria and defined the ratio as the level of host mitochondrial association, which indicates the ability of ROP39 to recruit host mitochondria. Host mitochondrial association decreased by 10% in RHΔrop39 (51.6 ±18.2%) compared with the parental RH strain (61.9±16.7%) in analysis of immunofluorescence microscopy images (P<0.05, oneway ANOVA followed by Tukey-Kramer post hoc test) (Fig. 3A). A RHΔrop39 complemented strain, RHΔrop39:ROP39, showed complete rescue of mitochondrial association (62.2±14.5%) (Fig. 4A). The kinase activity of ROPs are related to virulence of parasites and modification of gene expression in host cells (Taylor et al., 2006;Saeij et al., 2007). This led us to explore whether the kinase activity of ROP39 was involved in host mitochondrial recruitment. To assess this aspect of ROP39, we generated a strain to express ROP39 without kinase activity in a rop39-deleted background (RHΔrop39:ROP39δKAS) by introducing ROP39 with a deleted kinase active site amino acids 402 to 414 (VHSDLKPENVLV) predicted in PROSITE (https://prosite. expasy.org/), into RHΔrop39. RHΔrop39:ROP39δKAS was not rescued in mitochondrial association (50.7±17.8%) (Fig. 4A), which indicates ROP39 recruits host mitochondria through its kinase activity. We confirmed that host mitochondria association declined in RHΔrop39 (14.6±9.0%) in comparison with the RH strain (26.1±15.7%) in analysis of electron microscopy images (P<0.01, one-way ANOVA followed by Tukey-Kramer post hoc test) (Fig. 4C). On the other hand, host ER association was accelerated in RHΔrop39 and the maf1b-deleted strain RHΔmaf1b, in order of RHΔmaf1b (39.6±16.3%) >RHΔrop39 (27.1±11.4%) >RH (19.0±10.6%) (P<0.01, one-way ANOVA followed by Tukey-Kramer post hoc test) ( Fig. 4C; Fig. S2). We wondered that the gene deletion lowered the fitness of parasites, resulting in decreased association of host mitochondria. To investigate whether rop39-deletion has any impact on the fitness of T. gondii, we performed a plaque assay. There were no significant differences in the mean plaque area and the number of plaques between RH parental strain and RHΔrop39, indicating that decreased association in RHΔrop39 was not attributed to the gene disruption (Fig. S3). The percentage of PVs covered by host mitochondria or ER was almost constant (about 50%) in these three strains (Fig. 4D). This phenomenon happened possibly because ER associated the PVMs from which mitochondria were detached due to the disruption of ROP39 or MAF1b. Collectively, these findings show that ROP39 is a novel factor involved in recruitment of host mitochondria and the perimeter of PVs that can associate with host organelles may be fixed (Fig. 4E).

DISCUSSION
Although MAF1b was identified as a mitochondrial association factor, the molecule(s) responsible for host mitochondrial recruitment was unraveled. Here, we showed that ROP39 seemed to be a causative molecule for the recruitment. We observed that mitochondrial association in rop39-knockout parasites was decreased compared with a parental strain, which supports that ROP39 works on host mitochondrial association. Considering the secretion timing of ROPs and localization of ROP39 in host cell, there is a high possibility that ROP39 binds to host mitochondria prior to PVMs formation. The unknown domain(s) of ROP39 or other parasite protein(s) may work on anchoring ROP39 to host mitochondria because proteins possessing an MTS are internalized when it is inserted into mitochondria. Moreover, substrate(s) of ROP39 may be related to the anchoring process because kinasedead ROP39 could not rescue the level of host mitochondrial association in rop39-deleted mutants. Once the formation of PVMs occurs, ROP39 is recruited to PVMs possibly together with host mitochondria. Then, MAF1b may anchor host mitochondria on PVMs because it binds to host's mitochondrial interspace bridging (MIB) complex on its C-terminal region (Kelly et al., 2017), which can accelerate host mitochondrial association (Fig. 5).
The recruitment of host mitochondria was thought to be a means by which T. gondii acquires metabolites from host cells to help with its replication (Sinai et al., 1997;Crawford et al., 2006). However, it has been reported that host mitochondria associated with PV restrict the growth of T. gondii by intercepting the uptake of fatty acids (Pernas et al., 2018). However, it is unknown whether T. gondii exploit metabolites different from fatty acids through host mitochondrial recruitment. Further studies are needed to elucidate the reason that host mitochondrial association occurs in infected cells with T. gondii.
Collectively, our results identify T. gondii ROP39 as a new recruitment factor for host mitochondria. Exploiting components involved with recruitment of host mitochondria advances the understanding of why host mitochondrial recruitment diverges over a wide variety of intracellular parasites and how parasites employ mitochondrial recruitment to survive in host cells.

Mitochondria isolation and iTRAQ analysis
HFFs were cultivated to confluency in 145-mm dishes and 1.0×10 7 parasites were inoculated and cultured at 37°C for 24 h. Host mitochondria were extracted from the HFF infected with T. gondii and the mixture of 1.0×10 7 HFFs and 4.0×10 7 T. gondii (as control) by using a Mitochondrial Isolation Kit for Cultured Cells (Thermo Fisher Scientific, Waltham, MA, USA), which was performed following the manufacturer's protocol. These samples were maintained at 4°C and shipped to Filgen, Inc. (Aichi, Japan). The proteins extracted from the samples were reduced, alkylated and digested, resulting in generation of peptides. Control peptides were labeled with 117 isobaric tags, and test peptides derived from host mitochondria in parasiteinfected cells were labeled with 118 isobaric tags. The labeled peptides were separated using two-dimensional liquid chromatography followed by tandem mass spectrometry. Identification and quantification of proteins were performed using the Paragon™ algorithm in ProteinPilot™ software (AB Sciex, Framingham, MA, USA). The data from one technical replicate was processed in the analysis.

Mammalian cell transfection
HFFs were cultivated to confluency on a cover slip in 12-well plates. The cells were transfected with mammalian transient vectors ( pCMV-puro-P2A-BAFF, a kind gift from Dr Taishi Onodera, Department of Immunology, National Institute of Infectious Diseases, Japan) containing each rop39 version using Lipofectamine 3000 (ThermoFisher Scientific).

Generation of rop39 and maf1b deletion mutants
To knockout the gene of interest (GOI), CRISPR/Cas9 system was employed. Briefly, single guide RNAs bind to the homologous region of GOI and CRISPR/Cas9 breaks the double-strand DNA on GOI. The wound DNA strands were repaired with a homologous recombination and GOI is replaced with a drug resistant gene flanked by the homologous regions (Jinek et al., 2012;Cong et al., 2013).

Generation of the stably transfected parasites
RHΔrop39:ROP39 complemented strain and RHΔrop39:ROP39δKAS kinase-dead strain were generated from RHΔrop39 strain to be transfected with pROP39-ROP39-FLAG or pROP39-ROP39δKAS-FLAG vectors, respectively. A chloramphenicol acetyltransferase (chloramphenicol resistance gene) contained plasmid was co-transfected together with each vector. The transfectants were cultured under 20 mM chloramphenicol for 2 weeks and cloned by limiting dilution.

Transmission electron microscopy
HFFs were cultivated to confluency on a gold grid in 12-well plates and 1.0×10 6 parasites were inoculated and cultured at 37°C for 24 h. The culture was freeze-dried in liquid nitrogen and chemical fixation was performed. The sample was scanned on a transmission electron microscope (Tokai-Denshi Inc., Aichi, Japan). Host mitochondria and ER were distinguished by its membranous structure and electron density.

Quantification of the strength of host organelle recruitment
The perimeter of PVs and the length of host mitochondria and ER in contact with PVs were measured using ImageJ software (http://imagej.nih.gov/ij/) in immunofluorescence and electron microscopy images. A total length of mitochondria and/or ER in contact with PVs was divided by the perimeter of PVs and the ratio was defined as the strength of host organelle recruitment.

Evacuole assay
Parasites were collected by scraping the cell monolayer and released from host cells by passage through a 21-gauge needle. Extracellular parasites were filtered onto a polycarbonate membrane filter (3.0-μm pore size; Millipore, Bedford, MA, USA) and washed with Hanks' balanced salt solution (Wako) containing 0.1 mM EGTA and 10 mM HEPES. HFFs were cultivated on a cover slip (Matsunami Glass, Osaka, Japan) in 12-well plates for 48 h before the experiment. Parasites were pretreated with 1 μM cytochalasin D (Sigma-Aldrich) at room temperature for 10 min. 1.0×10 8 parasites were inoculated on host cells and cultured at 37°C for 60 min in the presence of 1 μM cytochalasin D. After washing with 1 mM D-PBS, the culture was fixed with 4% paraformaldehyde at 4°C for 20 min.

Plaque assays
HFFs were cultivated to confluency in six-well plates and 250 parasites were inoculated and cultured at 37°C for 11 days without disturbance. After washing with 1 mM D-PBS, the culture was fixed with methanol (Sigma-Aldrich) at 4°C for 20 min. The fixed culture was treated with 0.25% CBB R-250 (Wako) for 3 h. The images of plaques were procced on ImageJ software (http://imagej.nih.gov/ij/) by calculating the area size of all plaques. The plaque area was defined manually and marked with lines to major the area size by using Tools in Polygon Section.

Statistical analysis
All data are shown as means or mean±s.d. The normality was checked by Shapiro-Wilk test. All statistical tests were performed using R software (https://www.r-project.org).