Host MOSPD2 enrichment at the parasitophorous vacuole membrane varies between Toxoplasma strains and involves complex interactions

ABSTRACT Toxoplasma gondii is an obligate, intracellular parasite. Infection of a cell produces a unique niche for the parasite named the parasitophorous vacuole (PV) initially composed of host plasma membrane invaginated during invasion. The PV and its membrane (parasitophorous vacuole membrane [PVM]) are subsequently decorated with a variety of parasite proteins allowing the parasite to optimally grow in addition to manipulate host processes. Recently, we reported a proximity-labeling screen at the PVM–host interface and identified host endoplasmic reticulum (ER)-resident motile sperm domain-containing protein 2 (MOSPD2) as being enriched at this location. Here we extend these findings in several important respects. First, we show that the extent and pattern of host MOSPD2 association with the PVM differ dramatically in cells infected with different strains of Toxoplasma. Second, in cells infected with Type I RH strain, the MOSPD2 staining is mutually exclusive with regions of the PVM that associate with mitochondria. Third, immunoprecipitation and liquid chromatography tandem mass spectrometry (LC-MS/MS) with epitope-tagged MOSPD2-expressing host cells reveal strong enrichment of several PVM-localized parasite proteins, although none appear to play an essential role in MOSPD2 association. Fourth, most MOSPD2 associating with the PVM is newly translated after infection of the cell and requires the major functional domains of MOSPD2, identified as the CRAL/TRIO domain and tail anchor, although these domains were not sufficient for PVM association. Lastly, ablation of MOSPD2 results in, at most, a modest impact on Toxoplasma growth in vitro. Collectively, these studies provide new insight into the molecular interactions involving MOSPD2 at the dynamic interface between the PVM and the host cytosol. IMPORTANCE Toxoplasma gondii is an intracellular pathogen that lives within a membranous vacuole inside of its host cell. This vacuole is decorated by a variety of parasite proteins that allow it to defend against host attack, acquire nutrients, and interact with the host cell. Recent work identified and validated host proteins enriched at this host–pathogen interface. Here, we follow up on one candidate named MOSPD2 shown to be enriched at the vacuolar membrane and describe it as having a dynamic interaction at this location depending on a variety of factors. Some of these include the presence of host mitochondria, intrinsic domains of the host protein, and whether translation is active. Importantly, we show that MOSPD2 enrichment at the vacuole membrane differs between strains indicating active involvement of the parasite with this phenotype. Altogether, these results shed light on the mechanism and role of protein associations in the host–pathogen interaction.

a transiently expressed, epitope-tagged construct (34). MOSPD2 has been described in prior studies of uninfected cells as mediating inter-organellar associations through protein-protein interactions via its major sperm protein (MSP) and cellular retinalde hyde-binding protein (CRAL)/triple functional domain protein (TRIO) domain (35,36). It was tantalizing to hypothesize, therefore, that MOSPD2 might be mediating host ER-PVM interactions in infected cells, and in this study, we utilize microscopy, genetic, and biochemical techniques to investigate the mechanism and function of MOSPD2 localization at the PVM. We take advantage of three common strains of Toxoplasma to narrow down possible functions. In addition, biochemical methods allowed us to immunoprecipitate MOSPD2 and identify interacting Toxoplasma proteins at the PVM which were then tested for a role in and/or dependence on MOSPD2 association. We also assessed MOSPD2 association in the context of other parasite/host interactions at the PVM. The results reveal a unique interaction between Toxoplasma tachyzoites and the host cells they infect.

RESULTS
Association of MOSPD2 with the PVM was previously described by Cygan et al. using Type I RH. To determine whether MOSPD2 association differs between strains of Toxoplasma, human foreskin fibroblasts (HFFs) were infected with either Type I RH, Type II ME49, Type III CTG, or the close relative apicomplexan parasite, Neospora caninum (Nc) for 21 h and then stained using anti-MOSPD2 antibodies. The results showed association of MOSPD2 with the PVMs of cells infected with RH and ME49 at substantially higher levels than those infected with CTG and Nc (Fig. 1A). To quantify these apparent differences, we used Fiji to measure the florescence intensity at the PVM 21 h post-infection (hpi) while excluding the host cytoplasm and lumen of the PV. The results showed that while there was considerable variability between the mean fluorescence intensity of each PVM in a given monolayer, overall, the PVMs in cells infected with RH and ME49 averaged almost twice the fluorescence intensity of cells infected with CTG or Nc (Fig. 1B). To more precisely characterize this association, we performed immune-electron microscopy (IEM) with MOSPD2 antibodies on RH-infected HFFs. IEM images showed MOSPD2 located at the ER-PVM interface, although the extremely close apposition of these two membranes does not allow determination of precisely which membrane MOSPD2 is anchored within (Fig. 1C).
Although the major difference in MOSPD2 association seen during infection is between RH/ME49 and CTG/Nc, RH and ME49 also appear to differ in the pattern of MOSPD2 association at the PVM (Fig. 1A). To quantify the fluctuation in MOSPD2 signal in infected cells, the PVM-localized fluorescent signal was measured using Fiji (see Materials and Methods). MOSPD2 fluorescence at the PVM appeared considerably more patchy (i.e., had more variation in fluorescence intensity around any given PVM) in cells infected with RH than ME49 ( Fig. 2A and B). A major difference known for the PVM of cells infected with RH vs ME49 is the phenomenon of host mitochondrial association or HMA (14). MOSPD2 is known to have a role in membrane contact sites between organelles and is present at ER-mitochondria interface in HeLa cells (35). Therefore, it was possible that MOSPD2 might be playing a role in HMA at the PVM. To first determine whether MOSPD2 co-localizes with mitochondria at the PVM, RH-infected HFFs were stained with Mito Tracker and antibodies to endogenous MOSPD2. The results showed that localization of host mitochondria and MOSPD2 at the PVM were, in fact, anti-correlated (Fig. 2C). This was confirmed by immuno-electron microscopy with the results showing MOSPD2 at the PVM only in regions that did not have HMA ( Fig. 2D and E). These results suggest that MOSPD2 can only associate with regions of the PVM that are not occupied by host mitochondria.
The parasite protein responsible for HMA is known to be MAF1b (14). It co-localizes with host mitochondria at the PVM and is not expressed in ME49. To determine if MAF1b influences the loci where MOSPD2 localizes, whether directly or indirectly, HFFs were infected with ME49 parasites that ectopically express MAF1b. Immunostaining showed wild-type ME49 had low signal fluctuations around the PVM while ME49 that express MAF1b had MOSPD2 signal mirroring the pattern seen in RH; i.e., MOSPD2 appeared absent wherever MAF1b was present ( Fig. 3A through C). While we cannot exclude the possibility that MAF1b itself directly prevents MOSPD2 association, these results seem most likely to be due to active exclusion by host mitochondria recruited to the PVM.
The so-called Myc regulation 1 (MYR) complex is known to translocate soluble GRA effector proteins across the PVM and into the host cell (37)(38)(39). The first component characterized of the MYR complex was MYR1 and knocking out this gene leads to the range. Data are from one experiment that pooled three technical replicates. Significance was tested using a one-way ANOVA and Tukey post-hoc test thereafter (** indicates P < 0.01, *** indicates P < 0.001). One of three fully independent experiments is shown; all three had similar results (i.e., RH and ME49 significantly higher than CTG). (C) Left panel, electron micrograph of 24 h RH-infected HFFs stained with MOSPD2 antibody (diluted 1/50) and visualized using protein A-gold particles. Scale bar = 500 nm. Right panel, zoomed image of boxed region in left image with an arrow pointing at the PVM. ANOVA, analysis of variance; HFFs, human foreskin fibroblasts; MFI, mean fluorescence intensity; MOSPD2, motile sperm domain-containing protein 2; PVM, parasitophorous vacuole membrane. Datapoints are from one experiment that pooled the results from three replicate cover-slips for each strain, thereby yielding 15 vacuoles total per condition. Data are representative of three independent experiments producing similar results (i.e., RH range substantially higher than ME49). (B) The standard deviation (y-axis) of the ratios from (A) for each vacuole were plotted.
Significance was tested using Student's t test (*** indicates P < 0.001). The other two biological replicates showed a similar difference between the two strains with P < 0.008. (C) HFFs were infected with RH parasites for 21  To further explore which Toxoplasma proteins at the PVM might interact with MOSPD2, HFFs overexpressing V5-tagged MOSPD2 were infected with ME49 and immunoprecipitated using anti-V5 nanobodies for LC-MS/MS identification. Western blot stains for V5 showed strong enrichment of the epitope-tagged protein in the eluted material (Fig. 5A). VAP-A, a host protein known to interact with MOSPD2 at the ER (42), was also enriched in the elution while calreticulin which is also an ER protein but is  for LC-MS/MS protein identification. A value of 1 was added to all spectral counts for control and V5-MOSPD2 conditions then averaged and the results for each parasite protein were listed in descending rank for enrichment in the MOSPD2 samples relative to controls (Data Set S1). Intensity of red shading reflects relative number of spectral counts detected in each sample. Hyperplexed localization of organelle proteins by isotope tagging (LOPIT) data describe subcellular localization within intact parasites for identified proteins (49). Such data do not relate to a given protein's final location within the infected host cell. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MOSPD2, motile sperm domain-containing protein 2; PVM, parasitophorous vacuole membrane.
Research Article mSphere known to not associate with MOSPD2 was not co-precipitated and acted as a negative control (Fig. 5A). To determine the proteins most enriched in our immunoprecipitation, ratios between our V5-MOSPD2 and untagged control were generated (Fig. 5B). The most enriched Toxoplasma protein was ROP17, a serine/threonine kinase known to be at the PVM and having at least two functions (25,43); and among the top 15 most enriched proteins, 7 (ROP1/5/17/18, GRA12/14, and MAF1a) have previously been reported to localize to the PVM (20,25,30,(44)(45)(46)(47)(48) indicating that our protocol was indeed enriching for proteins associating, directly or indirectly, with MOSPD2 at the PVM.

Research Article mSphere
To determine whether the top Toxoplasma candidate proteins from the V5-MOSPD2 immunoprecipitation play a direct role in MOSPD2 association at the PVM, HFFs were infected with wild-type parasites or ones carrying mutations in the candidate genes. The results showed no difference in MOSPD2 association for RH∆rop17 relative to wild-type RH (Fig. 6A), indicating that ROP17 is not needed for MOSPD2 association with the PVM. Similarly, mutant parasites in ROP1, ROP5, ROP18, TGGT1_247350, and GRA12 showed no change in host MOSPD2 association at the PVM relative to infection with RH-wild type ( Fig. 6B to D). Together, these results suggest that while there are parasite proteins at the PVM that MOSPD2 is interacting with, at least the candidates tested so far are not individually responsible for the presence of MOSPD2 at this location.
The MOSPD2 association could either be pre-existing MOSPD2 recruited to the PVM-ER interface from other cellular compartments or newly synthesized material that is drawn to this location as its first (and only) destination. To distinguish between these competing hypotheses, HFFs were pre-treated with cycloheximide (CHX) for 1 h then infected with ME49 for 6 additional hours (Fig. 7A). CHX blocks host cell translation and it was previously shown that Toxoplasma tachyzoites can grow for at least 16 h in host cells blocked for protein translation (50). Quantitation of the MOSPD2 fluorescent signal at the PVM in HFFs pre-treated with CHX was greatly reduced relative to untreated controls ( Fig. 7B and C). To determine if CHX was somehow disrupting the trafficking and insertion of parasite-derived PVM proteins, we infected and treated HFFs as described above and stained for GRA7 and MAF1b. Imaging these monolayers showed GRA7 and MAF1b localized to the PVM, as expected ( Fig. 7D). To determine whether global levels of MOSPD2 change with CHX treatment, a Western blot was performed under the same conditions as the immunofluorescence assay described above. Levels of MOSPD2 did not perceptibly change up to 7 h post-CHX treatment, indicating MOSPD2 was not being degraded to any significant extent during this period of treatment (Fig. 7E). To confirm translation was successfully inhibited in HFFs, ubiquitin levels were assessed under the assumption that the proteasome would still be rapidly degrading ubiquitinated proteins during CHX treatment, a result reported in other cell types (51). Western blot analysis of ubiquitin levels showed a marked reduction of ubiquitinated proteins after CHX treatment relative to untreated controls (Fig. 7E), indicating that the CHX treatment was working, as expected. Together, these results indicate that association of MOSPD2 with the PVM depends on active translation, suggesting that newly synthesized material is what associates with the PVM, rather than previously synthesized MOSPD2 being "stolen" from other membranes.
Three main domains have been previously described for MOSPD2: the CRAL/TRIO, MSP, and a c-terminal tail anchor (Fig. 8A). To determine which of these domains plays a role in association with the PVM, V5-tagged mutant constructs of each domain were generated and overexpressed in wild-type HFFs using lentivirus transduction methods. For the CRAL/TRIO and tail anchor regions, the mutants were deletion constructs (∆CT and ∆TA, respectively), whereas for the MSP domain, point mutations at Arg404 and Leu406 were generated (R404D/L406D) since these mutations were previously reported to destroy the ability of MSP to associate with the FFAT motifs they normally recognize (35). Western blot analysis shows protein bands at the expected size for each mutant and in the wild type (Fig. 8C), albeit with different expression levels for each, possibly because of toxicity associated with some of these constructs. To determine if association at the PVM was still achieved with the mutant constructs, these HFF cell lines were infected with ME49 and stained by immunofluorescence. Association at the PVM was seen in infected cells expressing the wild type, ∆CT, and R404D/L406D constructs, although the association was reduced for the ∆CT version (Fig. 8B). For ∆TA, however, which was expressed at reduced but still detectable levels, no PVM association was seen (Fig. 8B). Quantifying the ratio of PVM-localized V5 signal (MOSPD2) relative to signal within the host cytosol showed a significant drop in the ∆CT line and no significant PVM-enrich ment in the ∆TA line compared to the wild-type control. Interestingly, the double point mutant appeared to have the opposite effect; i.e., a greater tendency to localize to the PVM than the wild-type control (Fig. 8D). Together, these results indicate that the CRAL/TRIO and tail anchor of MOSPD2 are necessary for its efficient association at the Toxoplasma PVM whereas the MSP domain might even work against such association.
Arginine-rich amphipathic helices (AHs) present in many ROPs are known to be sufficient to enable association with the PVM (26), and MOSPD2 was recently described to harbor one positively charged AH within the CRAL/TRIO (35). Using AlphaFold (52, 53), MOSPD2 is also predicted to have a second c-terminal AH, proximal to the tail anchor.
To determine if these structures are sufficient for PVM-association, the CRAL/TRIO AH Research Article mSphere (AH-CT) and the AH just preceding the tail anchor along with the tail anchor itself (AH-TA) were conjugated to the c-terminal end of enhanced green fluorescent protein (eGFP) and transiently transfected into HFFs (Fig. 9A). Live-cell imaging and quantitation of eGFP signal showed prominent association of eGFP-MOSPD2 with the PVM while the negative control had no apparent association ( Fig. 9B and C). AH-CT and AH-TA constructs, like the negative control, also had no apparent association with the PVM (Fig.  9B and C). Together, these results indicate that neither the AH-CT nor the tail anchor and its associated AH are sufficient for association with the PVM. MOSPD2 has been reported to not be essential for human cell growth (54) and so to generate a stable knockout in HFFs, a Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) guide targeting MOSPD2 was transduced into host cells while control cells received a non-targeting (Control) guide. After puromycin-selection for uptake of the gRNA-containing constructs, which also carry a puromycin-resistance cassette, knockout (∆MOSPD2) efficiency was assessed in the population by Western blot. Immunostaining showed a band for endogenous MOSPD2 in the HFFs receiving no virus or the Control gRNA while no detectable signal was seen in the ∆MOSPD2 knockout population (Fig. 10A), indicating the knockout efficiency was extremely high. To further confirm the ∆MOSPD2 HFFs were knockouts, they were infected for 21-24 h and stained for endogenous MOSPD2. Toxoplasma vacuoles in "No-virus" and Control HFFs had association of MOSPD2 while all observed ∆MOSPD2 HFFs had no detectable MOSPD2 at the PVM confirming efficient knockout (Fig. 10B). To determine whether loss of MOSPD2 has an impact on in vitro growth of Toxoplasma tachyzoites, plaque assays were performed in triplicate for each of Type I/II/III parasites in Control vs ∆MOSPD2 HFFs. The results at 9-day post-infection (Fig. 10C) showed a small but significant difference in plaque sizes across all strains between control and ∆MOSPD2 HFFs with RH, ME49, and CTG plaques being about 82%, 64%, and 64% the size of the plaques in Control HFFs, respectively. Interestingly, there was no significant difference in plaque size after growth on the two host cell lines for Type I (RH) and Type II (ME49) at 10 days and 13 days, respectively. For Type III (CTG, 13 days), however, there was a small difference that showed statistical significance with plaques in the ∆MOSPD2 cultures being about 88% the size of those in the Control HFFs. Overall, these data indicate that loss of host MOSPD2 has, at most, a modest impact on growth in HFFs in vitro.
The fact that MOSPD2 normally localizes to the ER and has a prominent role in mediating contact sites between host organelles suggested the possibility that it might have a role in ER association with the PVM in Toxoplasma-infected cells. To test this hypothesis, wild type and ∆MOSPD2 HFFs were infected with RH for 6 and 24 h and then imaged using electron microscopy. Control HFFs showed strong HMA and host ER association with the PVM, as expected (Fig. 11A); interestingly, however, infected ∆MOSPD2 HFFs also exhibited clear association of the PVM with these two classes of host organelle (Fig. 11B). To determine if there were any changes in the fraction of PVM associated with ER, this was quantified in Control and ∆MOSPD2 HFFs. The results for both 6-and 24 h post-infection showed no significant difference in ∆MOSPD2 relative to Control HFFs (Fig. 11C). These results argue against MOSPD2 playing an important role in mediating association of host ER with the Toxoplasma PVM.
It has been hypothesized that the arginine-rich AH domain of ROPs drives associa tion with the PVM through the attraction of the AH to the negative curvature of the PVM (26), but this possibility has not been directly tested. The MSP domain is known to be a protein-protein interactor (55) and so we decided to test whether the MSP domain might be interacting with the ROP-AH domain and whether this, in fact, is what draws ROP proteins to the PVM. To test this possibility, Control and ∆MOSPD2 HFFs were infected with ME49, and then fixed and stained with a monoclonal anti body that recognizes both ROP2 and ROP4 (56), or an antibody that recognizes the HA-tag on ROP17-3xHA (ROP2, ROP4, and ROP17 all have arginine-rich AH domains at their N-termini). The results showed no difference in association of ROP2/4 and ROP17 between the Control and the ∆MOSPD2 knockout cells (Fig. 12). Together with the results Monolayers were then stained with crystal violet and allowed to dry. Plaque area for each condition was measured in Fiji and plotted. Data are from one experiment consisting of three technical replicates. Significance was tested using Student's t test to do pair-wise comparisons. (D) Control and ∆MOSPD2 cells were infected with RH, ME49, or CTG parasites for 10 days (RH), or 13 days (ME49, CTG). Monolayers were then stained and analyzed as described in (C). GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HFFs, human foreskin fibroblasts; MOSPD2, motile sperm domain-containing protein 2.
Research Article mSphere described above (Fig. 6), this indicates that association with the PVM of at least these three ROP proteins is neither responsible for, nor mediated by, MOSPD2. Research Article mSphere

DISCUSSION
The PVM is a dynamic interface that allows Toxoplasma tachyzoites to co-opt host functions while remaining "hidden" from detection. Here, we dissected possible mechanisms and consequences of the dramatic presence of host MOSPD2 at this Research Article mSphere interface, a phenomenon that seems unlikely to be a chance event. Toxoplasma growth in vitro in MOSPD2 knockout cells leaned towards showing a small (~64-88%) decrease in plaque size, although the difference was not always statistically significant. This minimal effect could be a factor of the nutrient-rich conditions complete media provide to the host and parasite and so future work where the nutrients are more limiting will be important to explore. One possible role for MOSPD2's association with the PVM is that the CRAL/TRIO domain of MOSPD2 might bind to lipid substrates and move them between membranes, a function described for other proteins with similar domains (57)(58)(59). Thus, MOSPD2-PVM association could be one of the ways Toxoplasma acquires lipids from its host, including lipids needed to expand the PVM and elaborate the intravacuolar network (IVN) as the parasites grow. Future work assessing lipid scavenging in wild type and MOSPD2 knockout cells will be needed to address this possibility.
Another possibility is that MOSPD2 association at the PVM could be related to the immune response and so its function will only be revealed when infection of a cell type other than HFFs is examined or when cells are stimulated by cytokines like IFN-γ. For example, monocytes and macrophages have previously been shown to express higher levels of MOSPD2 (60). The migration of monocytes has also been shown to be dependent on this host protein (60). Our studies, however, have all been in non-motile HFFs, precluding an assessment of the role of such enrichment in cell migration. Toxoplasma is known to induce hypermigratory phenotypes in monocytes, macrophages, and dendritic cells (61)(62)(63)(64)(65)(66). One Toxoplasma protein involved in this hypermigration is ROP17 (65), a protein specifically enriched in the MOSPD2 immunopre cipitation. It is therefore appealing to hypothesize that the association of MOSPD2 at the PVM could be a mechanism by which Toxoplasma influences monocyte migration. Future work could focus on detailing whether MOSPD2 association in monocytes impacts cell migration upon infection with Toxoplasma. Ultimately, however, in vivo studies will be needed and while viable MOSPD2 knockout mice have been reported (54), discriminat ing between direct and indirect effects of such a disruption will be extremely difficult; i.e., determining whether any difference in the course of an infection with Toxoplasma in such animals is because of a specific role of MOSPD2 at the PVM or is due to a generalized effect on the host's overall metabolism, immunity or other characteristic.
In terms of the mechanism of MOSPD2-PVM association, we found that host mitochondria at the PVM preclude MOSPD2 association with this membrane. This seems most likely to relate to the fact that MOSPD2 is typically found embedded within the ER and, in any one patch of PVM at any one time, direct association can be with either host mitochondria or ER, but not both. Further work will be needed to determine whether other PVM-associating proteins show similar specificity in their location and, therefore, whether the mutually exclusive association of host mitochondria and MOSPD2 is mediated by the bulk size of mitochondria and ER or is due to more molecule-level effects such as localized differences in the lipid composition of the PV membrane. It is known, for example, that HMA facilitates lipid scavenging (67) and a possible hypothesis would be that the PVM's lipid composition or some other aspect of the PVM is altered when mitochondria associate with the vacuole. Although studies have investigated lipids important for Toxoplasma's growth (reviewed in 68), characterization of the PVM has been difficult in the context of lipids due to the fragile nature of the vacuole and arduous task of separating host and parasite membranes/lipids.
We show that there are substantial differences in the extent of MOSPD2 associa tion depending on the parasite strain used to infect HFFs. Many polymorphisms are known to exist between Types I/II/III parasites, including in expression levels and/or sequence of the PVM-localized ROP5, ROP17, and ROP18 (29)(30)(31). Any of these or other strainspecific proteins could contribute to the different patterns seen with the three Toxoplasma types examined here, but although all of these proteins co-precipitated with MOSPD2 from ME49-infected cells, none appeared individually responsible for MOSPD2's association with the PVM, at least for the RH strain. Future work with targeted deletions in the other strains could reveal these or another crucial interactor to be involved in MOSPD2 association that is not detectable in RH-infected cells. Fortunately, F1 progeny from crosses between Type I/II/III strains exist (69) and have been mapped. Therefore, phenotyping them with respect to MOSPD2 association should reveal the locus or loci involved, even if multiple such loci contribute collectively to the strainspecific differ ence; i.e., even if it is a quantitative trait.
The candidate MOSPD2-Toxoplasma interactome identified additional parasite proteins known to be at this host-parasite interface (20,25,30,(44)(45)(46)(47)(48). Among these, ROP1 and the polymorphic ROP5 and ROP17 are of particular interest due to the presence of a short linear motif that resembles an FFAT (two phenylalanine in an acidic tract) (70); such a motif in other host proteins is known to bind to the MOSPD2 MSP domain allowing membrane contact sites (MCS) to occur between this ER-resident protein and other organelles (35). Although we found that deletion of any one of these ROP loci on their own did not affect MOSPD2 recruitment, the possibility certainly remains that these and related ROP proteins collectively mediate MOSPD2 association with the PVM and that removal of multiple such proteins would markedly reduce MOSPD2 association. Alternatively, these parasite PVM proteins might associate with MOSPD2 indirectly, perhaps through interactions with other molecules, for example with specific lipid domains, as discussed above.
Mutant constructs in each of the three major domains of MOSPD2 revealed that the CRAL/TRIO and tail anchor play a role in, but are not sufficient for, MOSPD2's localization to the PVM. This strongly suggests that MOSPD2 must be anchored into a membrane for it to be enriched at the PVM, but which membrane, the PVM-proximal side of host ER or the PVM itself, is not addressed by our data. CHX treatment showed that translation is necessary for association suggesting that the MOSPD2 accumulating at the PVM is newly translated rather than being taken from pre-existing supplies in the cell, although we cannot exclude the possibility that inhibiting translation eliminates a short-lived host protein necessary for such accumulation. For example, tail-anchored proteins are known to be chaperoned by a class of proteins in the GET pathway in yeast and TRC pathway in mammals once their translation is completed (71). Perhaps one or many of these chaperones is short-lived, disrupting the pathway of tail anchor insertion into membranes. Certainly, a guided entry of tail-anchored proteins (GET)/transmembrane recognition complex (TRC) seems likely to be involved in the overall process but whether such is of parasite or host origin is not easily predicted. If it is of parasite origin, it could be polymorphic and explain the strainspecific difference. The Toxoplasma genome encodes strongly predicted GET/TRC proteins [ToxoDB.org] but there is no evidence reported that any of these are secreted outside the parasite, suggesting that they instead perform the usual duties for such proteins within the parasite (which are known to have their own tail-anchored proteins (72,73) and so in need of their own GET/TRCs). Whether or not GET/TRCs are involved in MOSPD2's association with the PVM, and whatever their source, this does not address which membrane, PVM or ER, harbors MOSPD2's tail-anchor and the immune-electron microscopy performed here does not have the resolution needed to address this crucial point. Future studies could aim at characterizing this using a combination of molecular, genetic, and structural biology techniques.
The extent of the association of MOSPD2 with the PVM is, so far, one of the greatest of any host protein studied to date in both the degree and specificity of the association. It has the potential, therefore, to reveal both important biochemistry and biology about the host-parasite interaction. The work described here reveals several important details about this phenomenon but much more work will be needed to reveal exactly how and why it is drawn to this site so strongly.
These tachyzoites and all subsequently generated lines were propagated in human foreskin fibroblasts (HFFs) cultured in complete Dulbecco's modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS; HyClone, Logan, UT), 2 mM L-glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin at 37°C with 5% CO 2 . The HFFs were obtained from the neonatal clinic at Stanford University following routine circumcisions that are performed at the request of the parents for cultural, health, or other personal medical reasons (i.e., not in any way related to research). These foreskins, which would otherwise be discarded, are fully deidentified and therefore do not constitute human subjects research.
To obtain parasites, infected monolayers were scraped and the host cells lysed by passage through a 25-gauge needle and counted using a hemocytometer.

Cycloheximide (CHX) treatment
HFF monolayers were seeded on sterile, glass coverslips 18 or more hours prior to adding 1 µg/mL or 10 µg/mL cycloheximide (CHX, Millipore Sigma). HFFs in 24-well or 6-well dishes were incubated with CHX for 1 h before adding ME49 and pulse spun at 82 g for 1 s to help the parasites make contact with the host cells. Parasites were cultured on HFFs for 1 h then washed, and control or CHX-containing media was added where appropriate. The infected monolayers were then incubated for 5 additional hours prior to carrying out immunofluorescence assay (IFA) protocols.

Immunofluorescence assay (IFA)
Infected cells grown on glass coverslips were fixed and permeabilized using 100% cold methanol for 15 min. Samples were washed three times with phosphatebuffered saline (PBS) and blocked using 3% bovine serum albumin (BSA) in PBS for 30 min at room temperature (RT). HA was detected with rat monoclonal anti-HA antibody 3F10 (Roche), SAG1 was detected with mouse anti-SAG1 monoclonal antibody DG52 (79), GRA7 was detected with rabbit anti-GRA7 antibodies (80), V5 was detected with a mouse anti-V5 tag monoclonal antibody (Invitrogen), MOSPD2 was detected with a rabbit polyclonal antibody (Millipore Sigma), calreticulin was detected with a mouse monoclonal antibody FMC 75 (Abcam), ROP2/4 was detected with a mouse monoclonal antibody (56). Primary antibodies were detected with goat polyclonal Alexa Fluor-conju gated secondary antibodies (Invitrogen). Both the primary and secondary antibodies were diluted in 3% BSA in PBS. Coverslips were incubated with the primary antibodies for 1 h at RT, washed, and incubated with secondary antibodies for 45 min at RT. Vectashield with DAPI (4′,6-diamidino-2-phenylindole) stain (Vector Laboratories) was used to mount the coverslips on slides. Fluorescence was detected using widefield epifluorescence microscopy or by confocal microscopy on an LSM 700 laser scanning confocal microscope. Images were analyzed using ImageJ software and the intensity levels of the images adjusted such that no data were removed from images. All images shown for any given condition/staining in any given comparison/data set were obtained using identical parameters unless otherwise stated.
For mitochondrial staining, HFF monolayers were infected as described. MitoTracker (Invitrogen) was used at manufacture's recommendation prior to fixing with 4% formaldehyde in DMEM lacking phenol red for 15 min at room temperature. Samples were then washed and permeabilized with 0.2% Triton X-100 for 15 min at room temperature. IFA staining was done as described above.

Partial permeabilization
Parasites were syringe released as described and used to infect HFFs for 2-3 h, at which time the cells were washed with PBS and then fixed with 4% formaldehyde at room temperature for 15 min. Formaldehydefixed samples were rinsed once with PBS, permeabilized with 0.02% digitonin solution for 1-3 min, and then blocked with 3% BSA in PBS for 1 h at RT. Staining was performed as described above.

IFA analysis
Quantification of MOSPD2 fluorescence around the PVM was performed in Fiji. To quantify fluorescence intensity, a polygonal region of interest (roi) was generated around the outside and inside of the PVM as defined by calreticulin or SAG1 staining to exclude the host cytosol, parasites, and lumen of the PV. The roi was copied onto the MOSPD2 channel and pixel intensity was measured. Average fluorescence intensity for the roi was plotted as arbitrary units.
Alternatively, a single line 1 or 5 pixels wide was generated around the PVM, again guided by calreticulin/SAG1 staining or intra-parasitic fluorescent signal. It was copied onto the MOSPD2 channel and pixel intensity measured at each point on the line. To normalize across vacuoles, each pixel intensity value was divided by the median or mean pixel intensity for that given vacuole and the ratios generated plotted.

Plasmid construction
For lentiviral plasmid constructs, standard molecular cloning techniques were used to amplify wild-type MOSPD2 from a pCDNA construct (primers A1, A2, A20, and A21 (34); a list of all primers used in this study can be found in Table S1).
For eGFP constructs, the backbone originated from pCDNA-MOSPD2 (34). The backbone for AH-TA was amplified using A9 and A25 from pCDNA-MOSPD2 (34). The eGFP insert was amplified from CAS9_sgRNA (82) using primers A23 and A24. The negative control and AH-CT plasmids were constructed using AH-TA as a template and primers A11 and A28 then A26 and A27, respectively. The full length eGFP-MOSPD2 plasmid was constructed using primers A11 and A30 from HA-TA. The MOSPD2 insert was amplified using primers A29 and A31 from pCDNA-MOSPD2 (34). Plasmids were ligated using standard molecular biology techniques.

Mammalian cell culture and stable cell line generation
All mammalian cell lines were propagated in complete Dulbecco's modified Eagle medium (DMEM) supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin at 37°C with 5% CO 2 , unless otherwise noted.
For preparation of lentiviruses, HEK 293T cells in 10 cm dishes were transfected at 80% confluence with the lentiviral plasmid pLenti-CMV-Puro (Addgene plasmid no. 17452) containing the gene of interest (2 mg) and the lentiviral packaging plasmids pVSV-G, pDVPR, and pAdvant (gifts from Jan Carette, Stanford University) using FuGENE HD transfection reagent (Promega) according to manufacturer's instructions. After about 24 h, the medium was replaced with fresh medium. Approximately 48 h after transfec tion, the cell medium containing the lentivirus was harvested and filtered through a 0.45 µm filter and supplemented with 8 mg/mL protamine sulfate. To generate stable lines, HFFs in T25 culture flasks were then infected with the virus-containing medium (1-2.5 mL). The following day, the viral-containing medium was removed and replaced with fresh, antibiotic-free medium. The HFFs were allowed to recover for 48 h and then selected with medium containing 2 µg/mL puromycin for 3 days.

Transient mammalian transfections
HFFs were grown on glass coverslips to ∼80% confluence and subsequently transfec ted with Lipofectamine LTX reagent (Invitrogen) and 500 ng of each pCDNA plasmid (with the tagged gene of interest as described above) according to the manufactur er's instructions in antibiotic-free medium. Cells were incubated with the transfection reagent for ∼6-7 h and tachyzoites were added for another 21 h before imaging or fixing.

Gene disruption in Toxoplasma
A list of all sgRNA sequences and primers used in this study can be found in Table  S1. For gene disruption plasmids, guide RNAs, designed against a protospacer adjacent motif (PAM) site of each gene of interest, were cloned into the pU6-Universal plasmid (Addgene plasmid number 52694; http://n2t.net/addgene:52694; RRID:Addgene_52694). Parasites were transfected with the pU6-sgRNA plasmid containing the guide for GRA12 (A13 and A14) or TGGT1_247350 (A32 and A33) and allowed to infect HFFs in DMEM. Linear PCRamplified hypoxanthine-guanine phosphoribosyl transferase (HPT) (primers A15 and A16 from pTKO2 (38)) was co-transfected with the GRA12 guide. After at least 18 h of recovery time, transfected cell cultures were drug selected for 8 days with 25 µg/mL mycophenolic acid (MPA) and 50 µg/mL xanthine (XAN). Single clones were selected from the transfected populations in 96-well plates using limiting dilution in MPA/XAN-supplemented medium and PCR verified for gene disruption (A17 and A18).

Western blotting
Cell lysates were prepared in Laemmli sample buffer (Bio-Rad) at the time points post-infection indicated. The samples were boiled for 5 min, separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. The membranes were blocked with 5% non-fat dry milk or 5% BSA in Trisbuffered saline supplemented with 0.5% Tween 20, and proteins were detected by incubation with primary antibodies diluted in blocking buffer, followed by incubation with secondary antibodies (raised in goat against the appropriate species) conjugated to horseradish peroxidase (HRP) and diluted in blocking buffer. HA was detected using an HRP-conjugated HA antibody (catalog no. 12013819001; Roche), SAG2A was detected using rabbit polyclonal anti-SAG2A antibodies, VAP-A was detected using mouse monoclonal 4C12 (Santa Cruz), calreticu lin was detected with a mouse monoclonal antibody FMC 75 (Abcam), MOSPD2 was detected with a rabbit polyclonal antibody (Millipore Sigma), ubiquitin was detec ted using rabbit polyclonal antibody (Thermo), α-tubulin was detected using mouse monoclonal antibody B-5-1-2 (Signa-Aldrich), and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was detected using mouse monoclonal anti-GAPDH antibody 6C5 (Calbiochem). Horseradish peroxidase (HRP) was detected using an enhanced chemilu minescence kit (Pierce).

Plaque assay
Parasites were syringe released from HFFs and added to confluent HFFs in T25 flasks. After 9-13 days of undisturbed incubation at 37°C as described above, the infected monolayers were washed with PBS, fixed with methanol, and stained with crystal violet. The plaque area was measured in arbitrary units using ImageJ software.

Transmission electron microscopy (TEM)
For ultrastructural observations of Toxoplasma-infected HFF for 24 h by thin section, samples were fixed in 2.5% glutaraldehyde in 0.1 mM sodium cacodylate and processed as described previously (83). Ultrathin sections of infected cells were stained with osmium tetroxide before examination with Hitachi 7600 EM under 80 kV equipped with a dual AMT CCD camera system. Quantitative measurement of length for the PV mem brane and host ER elements attached to this membrane using ImageJ was performed on 18 representative electron micrographs at low magnification to ensure the entire PV fit into the field of view.

Immunoelectron microscopy (IEM)
Monolayers of HFF infected with Toxoplasma from 6 or 24 h were fixed in 4% parafor maldehyde (PFA; Electron Microscopy Sciences, PA) in 0.25 M HEPES (ph 7.4) for 1 h at room temperature, then in 8% PFA in the same buffer overnight at 4°C. Samples were infiltrated, frozen, and sectioned as previously described (84). The sections were immunolabeled with rabbit anti-MOSPD2 antibody at 1/50 diluted in PBS/1% fish skin gelatin. The sections were then incubated with IgG antibodies, followed directly by 10 nm protein A-gold particles before examination with the EM.

IPs for mass spectrometry
Immunoprecipitations (IPs) to identify MOSPD2-interacting proteins in HFFs were performed as follows. One 10 cm dish of HFFs for each infection condition were seeded with 3 million HFFs 18-24 h prior to infection. HFFs were infected with 16 million ME49∆hpt parasites for 19 h. Infected cells were washed three times in cold PBS and then scraped into 1 mL cold cell lysis buffer [50 mM Tris (pH 8.0), 150 mM NaCl, 0.1% (vol/vol) Nonidet P-40 alternative (CAS no. 9016-45-9)] supplemented with complete protease inhibitor cocktail (cOmplete, EDTA free; Roche) and phosphatase inhibitor (PhosSTOP, Roche). Cell lysates were passed three times through a 25-gauge needle, followed by passage three times through a 27-gauge needle. After lysing, samples were incubated on ice for 30 additional minutes. The cell lysates were spun at 10,000 g for 10 min at 4°C to remove insoluble material and unlysed cells. Protein concentration was calculated using a Bradford Assay (Thermo Scientific). Equal amounts of protein (3,500 µg) from lysates were added to 100 µL magnetic beads conjugated to anti-V5 nanobodies (Chromotech), and the mixture was incubated rotating at 4°C for 1 h. Unbound protein lysate was removed, and the anti-V5 magnetic beads were then washed 10 times in cell lysis buffer containing protease inhibitor and lacking NP-40. V5-tagged MOSPD2 and associated proteins bound to the beads were delivered to the Stanford University Mass Spectrome try core for on-bead digestion.

Mass spectrometry sample preparation
For on-bead digestion, beads were resuspended in 100 mM S-Trap digestions buffer (TEAB) and mixed head-over-head on a ThermoLyne LabQuake shaker for 10 min. DTT was added to a final concentration of 10 mM at 55°C for 5 min followed by head-overhead mixing, at room temperature, for 25 min. Acrylamide was added at 30 mM for cysteine capping during head-over-head mixing for an additional 30 min. Trypsin/LysC (500 ng) was added for proteolysis overnight, at 37°C. Samples were then quenched with 5 µL of 50% formic acid, separated from beads, and cleaned by C18.

Mass spectrometry
For peptides from on-bead digests, the samples were analyzed either on a Orbitrap Fusion tribrid mass spectrometer (Thermo Scientific) RRID:SCR_018702 or an Orbitrap Eclipse tribrid mass spectrometer (Thermo Scientific) RRID:SCR_022212, in both cases coupled to a Acquity M-Class liquid chromatograph (Waters Corporation). In brief, peptides were injected at a flow rate of 300 nL/min with a mobile phase A of 0.2% aqueous formic acid and a mobile phase B of 0.2% formic acid in acetonitrile. Peptides were directly injected onto a ~25 cm in-house pulled-and-packed fused silica column with an I.D. of 100 µm. The column was packed with 1.8 µm C18 stationary phase, and the gradient was a 2-45% B, followed by a high B wash for a total gradient time of 180 min. The mass spectrometer was operated in a data dependent fashion using Collision-induced dissociation (CID) fragmentation in the ion trap to generate MS/MS spectra and Higher-energy C-trap dissociation (HCD) in the orbitrap following synchro nous precursor selection for detection and quantification of report ions in the MS3 step.

Mass spectrometric analysis
For a typical data analysis, peptide spectra assignments and protein inferences were performed using Byonic v.4.2.4 (Protein Metrics), assuming fullyspecific tryptic digestion and up to two missed cleavages, as well as common modifications such as methionine oxidation and cysteine alkylation.