Engineering Brain Parasites for Intracellular Delivery of Therapeutic Proteins

Protein therapy has the potential to alleviate many neurological diseases; however, delivery mechanisms for the central nervous system (CNS) are limited, and intracellular delivery poses additional hurdles. To address these challenges, we harnessed the protist parasite Toxoplasma gondii, which can migrate into the CNS and secrete proteins into cells. Using a fusion protein approach, we engineered T. gondii to secrete therapeutic proteins for human neurological disorders. We tested two secretion systems, generated fusion proteins that localized to T. gondii’s secretory organelles and assessed their intracellular targeting in various mammalian cells including neurons. We show that T. gondii expressing GRA16 fused to the Rett syndrome protein MeCP2 deliver a fusion protein that mimics the endogenous MeCP2, binding heterochromatic DNA in neurons. This demonstrates the potential of T. gondii as a therapeutic protein vector, which could provide either transient or chronic, in situ synthesis and delivery of intracellular proteins to the CNS.


Introduction
Protein therapeutics have the potential to treat various human disorders: many genetic diseases caused by mutations in a specific protein can be rescued by delivery of the functional 30 protein, treating the root of the disease. In addition, proteins with downstream or compensatory activity can provide indirect rescue or mitigation of disease symptoms (Leader, Baca, & Golan, 2008). However, owing to their macromolecular nature, the delivery of therapeutic protein to target tissues is extremely challenging. Low functional stability and rapid loss of activity following administration or during storage, along with low permeability through physiological 35 barriers, limit the delivery of active proteins (Leader et al., 2008). Efficient delivery is particularly challenging for neurological diseases due to the blood-brain barrier (BBB), which tightly regulates the transport of molecules to the brain, and blocks the transport of most large, charged and hydrophilic molecules (DiNunzio & Williams, 2008). In addition to specific delivery to the brain, intracellular delivery of proteins poses additional hurdles. Passive delivery 40 is often precluded as most proteins are unable to spontaneously penetrate through the cell membrane or become endocytosed effectively (Torchilin, 2008). Furthermore, intracellular delivery has been proven especially difficult for neuronal cells (Karra & Dahm, 2010).

Fig. 1. Generating T. gondii for delivery of heterologous therapeutic proteins.
A therapeutic protein of interest is genetically fused to an endogenously-secreted protein of the parasite, which targets it for secretion from the respective secretory organelle. The genetic 120 construct encoding for the fusion protein is introduced into T. gondii to generate transgenic lines which secrete the therapeutic protein. The transgenic T. gondii actively migrates to the CNS, bypasses the BBB and delivers the protein to neurons. The top right corner presents the three secretion systems of T. gondii, which secrete proteins using different mechanisms and at different stages of cell invasion or intracellular persistence. Mn-microneme, Rh-rhoptry, DG-125 dense granule.

Targeting proteins to the rhoptry secretory organelle for transient delivery
The first secretion system we tested was the rhoptries. As some rhoptry proteins are injected before cell invasion, their secretion does not depend on productive invasion and 130 persistence of the parasite inside the host cell. Furthermore, quantification in the CNS of mice showed that rhoptry proteins can be injected by T. gondii into 30-50 times more cells in the brain than those productively infected (Koshy et al., 2012). This suggests an advantage of rhoptry secretion for transient delivery of proteins, as each parasite can secrete rhoptry proteins into multiple cells without invading them. 135 In order to co-opt rhoptry secretion, heterologous proteins were fused to the rhoptry protein Toxofilin, which is injected into the cytosol of host cells upon initial parasite engagement (Boothroyd & Dubremetz, 2008). The fusion to Toxofillin was modelled after an ingenious reporter line of T. gondii expressing Toxofilin-fused Cre recombinase, which enabled the labeling of cells contacted by the parasite both in vitro and in vivo (Cabral et al., 2016;Koshy et 140 al., 2010;Lodoen, Gerke, & Boothroyd, 2010). We generated parasites with genomic integration of constructs encoding for Toxofilin fused to different variations of: ASPA, SMN, GDNF, PARK2, GALC, MECP2, and TFEB (Table S1). Expression was controlled by the endogenous promoter of Toxofilin, and an HA tag was added for immunolabeling of the fusion proteins. Most fusion proteins mis-localized to other organelles of the parasite, predominantly to the 145 endoplasmic reticulum (ER), golgi and micronemes. However, Toxofilin-fused GDNF, PARK2 and codon-optimized TFEB (TFEBopt) were successfully localized to the rhoptries (Fig. 2, Table S1). the rhoptries (TFEBopt, PARK2, GDNF) or in the lack of any, the most common localizations in the stable pool (full list of observed localizations in Table S1). Text above the merge images indicates the localization of the fusion protein in the image. Scale bar: 10μm. 160 Targeting proteins to the dense granule secretory organelle for continuous delivery As rhoptry secretion occurs as a discrete event upon initial parasite engagement, it provides only transient protein delivery. In contrast, dense granule proteins can be secreted continuously as the parasite persists inside a cell. Therefore, we assessed dense granule-mediated secretion for continuous delivery of intracellular proteins. For this purpose, we designed 165 translational fusions to GRA16. GRA16 is secreted from the dense granules to the parasitophorous vacuoles (PV), exported from the PV to the host cell cytosol, and localizes to the host cell nucleus (Bougdour et al., 2013). We generated parasites with genomic integration of constructs encoding for GRA16 fused to different variants of: ASPA, SMN, GALC, MECP2, and TFEB (Table S2). Expression was controlled by the endogenous promoter of GRA16, and an 170 HA tag was added for immunolabeling of the fusion proteins.
GRA16-fused GALC-TAT was not expressed in T. gondii. GRA16-fused GALCopt was expressed, but localized only to the ER of the parasites and was not secreted to the PV, nor detected in the host cell. ASPA, ASPAopt and GALC fused to GRA16 localized to the PV but were not detected in the host cell, indicating that they were successfully targeted to the dense 175 granules and secreted to the PV, but not exported in detectable levels. Importantly, the GRA16fused nuclear proteins MeCP2 (MECP2opt, total fusion protein size 110kDa), SMN (SMN1, 88kDa) and TFEB (TFEBopt, 109kDa) localized to both the PV and the host cell nucleus. This indicates that these proteins were successfully targeted to the dense granules, secreted to the PV, exported to the host cell cytosol and accumulated in their site of activity, the host nucleus ( Fig.   180 3) (Bowerman et al., 2017;Katz et al., 2016;Napolitano & Ballabio, 2016). Interestingly, by analyzing the level of intrinsic disorder for each of the tested fusion proteins, we found a strong correlation (R=0.975, P=4*10 -5 ) between the level of localization to the host cell nucleus and the average intrinsic disorder score of the fused protein (based on IUPred2A (Mészáros, Erdős, & Dosztányi, 2018)) ( Fig. 3F, Fig. S1). This provides strong 185 support for involvement of protein structural disorder in export through the PVM, and for the hypothesized translocon model of protein export (Hakimi & Bougdour, 2015;Hakimi et al., 2017).
To assess whether a shorter fragment of GRA16 could replace the full-length protein as a carrier for secretion, we tested three truncated variants of GRA16 fused to MeCP2. However, as 190 none of the truncated fusion proteins were detected in the host cell nucleus, we decided to continue with the full-length GRA16 as the carrier protein (Fig. S2).
Since GRA16-fused MeCP2 and TFEB displayed the most robust delivery and targeting to the host cell nucleus, we focused on these fusion proteins for the rest of the study. The T.
gondii lines expressing these proteins were named GRA16-MeCP2 (expressing GRA16-HA- below each scatter plot shows the mean relative nuclear localization ±95% confidence interval (relative to GRA16-HA). Significance represents the difference between the fusion protein and the parental strain ('no construct'), calculated by one-way ANOVA with multiple comparisons (Dunnett test). **** P<0.0001; * P<0.0332; ns-not significant. (F) Correlation between host nuclear localization and the level of intrinsic disorder of the fused therapeutic protein, based on 220 IUPred2A (see Fig. S1). Linear regression line drawn in grey. Scale bar: 10μm.

Kinetics of GRA16-mediated protein delivery in human fibroblasts and neurons
To characterize the kinetics of GRA16-mediated protein delivery in vitro, we used highcontent imaging to quantify infection rate and nuclear protein delivery for parasites expressing 225 GRA16 alone and GRA16 fused to MeCP2 or TFEB. We infected human foreskin fibroblast (HFF) cells using different MOI (Multiplicity Of Infection-the proportion of parasite per host cell administered to the culture) and followed nuclear fluorescence over time ( Fig. 4A-C). 13 Overall, there was no significant difference in the infection rate of the different lines, indicating that they are similar in their ability to invade host cells (by two-way ANOVA). 24 hours post-230 administration, the percentage of cells productively infected was on average 50-65% for cells infected using MOI=3, 29-48% with MOI=1 and 13-19% with MOI=0.33. 24 hours postadministration, the mean percentage of host cell nuclei labeled as positive for the proteins was 44-53% for cells infected using a MOI=3, 20-43% with MOI=1 and 8-21% with MOI=0.33.
Quantification of fluorescence intensity in the host cell nuclei showed that the fusion of the 235 mammalian proteins MeCP2 and TFEB to GRA16 reduced its accumulation in the host cell nucleus by 7.5 folds on average.

Delivery of GRA16-fused TFEB and MeCP2 to the nucleus of human neurons
Since the primary targets of protein delivery would be neurons, we tested whether the 240 transgenic T. gondii can deliver therapeutic proteins to their site of activity in human neurons.
Lund Human Mesencephalic (LUHMES) cells were differentiated in vitro to morphologically and biochemically mature dopamine-like neurons (Lotharius et al., 2002;Scholz et al., 2011;Shah et al., 2016) and infected with the T. gondii lines GRA16-HA, GRA16-MeCP2 and GRA16-TFEB. All tested lines displayed clear secretion, export and accumulation of the tagged 245 protein in the nuclei of the neurons. Analysis of the kinetics of parasite infection in these neurons showed that 24 hours post-administration, 70% of neurons were infected using MOI=2, 59% using MOI=1 and 39% using MOI=0.5 ( Fig. 4D-F).

T. gondii-delivered MeCP2 binds heterochromatic DNA in the nucleus of primary neurons
In order to test the therapeutic relevance of the system, we focused on GRA16-MeCP2.
MeCP2 is a nuclear protein which is expressed ubiquitously, but MeCP2 levels are highest in postmitotic neurons, where it is critical for the function and maintenance of the neurons (Lyst & Bird, 2015). This is demonstrated by the disease manifestation of Rett syndrome (RTT), a severe 270 neurological disorder caused by loss-of-function mutations in the MECP2 gene. The activity of the MeCP2 protein depends on its specific binding of heterochromatic DNA in the nucleus.
MeCP2 recognizes patterns of methylated DNA and recruits co-repressor complexes. Therefore, a classical readout for its functionality is its ability to specifically bind heterochromatic DNA (Guy et al., 2018). 275 To test the functionality of GRA16-fused MeCP2, we infected mouse primary neuronal cultures with T. gondii expressing GRA16-MeCP2. In mouse cells, heterochromatic DNA can be readily visualized with DAPI staining. Importantly, we found that the MeCP2 fusion protein delivered by the parasites co-localizes to the foci of heterochromatic DNA in the nucleus (Fig.   5A). This suggests that the MeCP2 fusion protein synthesized and delivered by the parasites 280 binds heterochromatic DNA in the neuronal nucleus, mimicking the functional endogenous MeCP2 (Guy et al., 2018). Similarly, mouse neuroblastoma (N2A) cells infected with T. gondii secreting GRA16-MeCP2 presented the same binding of heterochromatic DNA, as well as colocalization with MeCP2 immunostaining. By comparison, cells infected with T. gondii parasites secreting GRA16-HA did not present the same co-localization, and the protein was evenly 285 diffused in the nucleus, confirming the specificity of binding to the fusion protein containing MeCP2 (Fig. 5B).
Excess of MeCP2 can be deleterious, as exhibited by the pathogenesis of MECP2 duplication syndrome (Ramocki, Tavyev, & Peters, 2010). Furthermore, gene therapy studies have shown that MECP2 overexpression leading to 2-6 folds higher protein levels can induce 290 adverse symptoms and increased mortality in mice (Collins et al., 2004;Gadalla et al., 2013;Luikenhuis, Giacometti, Beard, & Jaenisch, 2004). To test the level of MeCP2 rescue achieved by the transgenic T. gondii, we quantified MeCP2 delivery in MeCP2-deficient neurons.

Discussion
In this study we investigated a method for therapeutic intracellular protein delivery via a 325 transgenic brain parasite -T. gondii. We assessed different approaches for heterologous protein secretion and characterized T. gondii lines that deliver therapeutic proteins to the nuclei of various mammalian cells. We showed that T. gondii expressing a GRA16-fused MeCP2 deliver a fusion protein that binds heterochromatic DNA in neurons, mimicking the functional endogenous MeCP2. 330 For both the Toxofilin and GRA16 fusion proteins, we found that mistargeted proteins were mostly arrested within the secretory pathway (ER, golgi). However, some fusion proteins resulted in unexpected mistargeting to parasite-unique organelles, such as the microneme or apicoplast (Table S1 and S2). As the fusion proteins tested in this study represent a wide and unconventional set of sequences, the targeting of the tested fusion proteins could provide an interesting resource to analyze factors affecting protein trafficking in T. gondii, as well as the regulation of protein secretion and export from the parasitophorous vacuole (PV).
Interestingly, previous work which attempted to fuse dense granule exported proteins to marker proteins such as fluorescent or other highly structured proteins found that fusion to another protein blocked its ability to be exported through the PV membrane (PVM), causing it to 340 accumulate in the PV space (Curt-Varesano, Braun, Ranquet, Hakimi, & Bougdour, 2016;Hakimi & Bougdour, 2015;Hakimi et al., 2017;Marino et al., 2018). In addition, known exported dense granule proteins of T. gondii are enriched in regions of high intrinsic disorder (Hakimi & Bougdour, 2015). In this study we tested the secretion, export and host cell nuclear localization of multiple heterologous proteins fused to the dense granule exported protein 345 GRA16. Of the expressed fusion proteins that were successfully secreted by the dense granules, some were detected in the host cell nucleus (MeCP2, SMN, TFEB) and some were not, suggesting that they were not exported from the PV (ASPA and GALC). When comparing the intrinsic disorder profiles of these proteins, we noticed a significant correlation between the level of protein disorder and the propensity to be exported and localize to the host cell nucleus (Fig. 3,   350 Fig. S1). This provides strong support for the proposed model of protein export, in which high intrinsic disorder is required for protein export through the PVM (Hakimi et al., 2017).
During this study we also established a new model for the study of T. gondii in human neurons differentiated from neuronal precursor LUHMES cells. Although for many hosts (including humans), T. gondii's dominant infection is in the CNS, most research on the biology 355 of T. gondii is performed in standard fibroblast and epithelial cell lines. On the other hand, use of primary neuronal cultures can be highly variable, expensive, labor intensive and lead to high animal use. LUHMES cells are conditionally-immortalized neuronal precursor cells for which culturing is straightforward and which can differentiate into mature dopamine-like human neurons within one week (Lotharius et al., 2002;Scholz et al., 2011;Shah et al., 2016). This 360 LUHMES neuronal model could facilitate the study of fundamental questions regarding the unique interaction of T. gondii with neurons, and especially human neurons.
T. gondii-mediated protein delivery has multiple attractive characteristics. T. gondii has an exceptionally high CNS-specificity driven by active trafficking to the CNS from peripheral tissues, even with oral administration. Likewise, active motility also drives T. gondii's 365 widespread distribution in the brainin a systematic characterization of cyst distribution in infected mice, 92% of the brain regions examined were found to contain tissue cysts (Berenreiterová, Flegr, Kuběna, & Němec, 2011). Transgenic T. gondii can provide either longterm or transient protein delivery, based on the utilized secretion organelle and subsequent persistence or clearance of the parasites. Importantly, as a biological agent, T. gondii is 370 responsive to external cues and can implement engineered genetic circuits (Jimenez-Ruiz, Wong, Pall, & Meissner, 2014;Wang et al., 2016). This could be utilized for incorporating inducible expression systems, or for generating attenuated T. gondii which can be cleared from the tissue following protein delivery. T. gondii can provide local and compartmentalized protein synthesis in situ and deliver proteins directly into the cytosol of cells. It can deliver large proteins, the size 375 limit of which is yet unknown (the largest fusion protein delivered in this study is 110kDa). For comparison, the leading vectors used for gene therapy today, AAV and self-complementary AAV (scAAV), are limited to a packaging capacity of 4.7kb and 2.2-3kb, respectively, which renders them impractical for many human proteins (Colella et al., 2018).
Rett syndrome is a debilitating neurological disorder that arises from mutations in the X-380 linked MECP2 gene and affects approximately 1 in 10,000 females. It is characterized by apparently normal early development followed by profound neurologic regression around 1-2 years of age (Carrette et al., 2018;Katz et al., 2016). Since 2007, several gene therapy studies in rodents have shown that expression of functional MeCP2 can reverse the symptoms of Rett syndrome, even after disease onset (Gadalla et al., 2013(Gadalla et al., , 2017Guy, Gan, Selfridge, Cobb, & 385 Bird, 2007;Katz et al., 2016). Furthermore, exclusive MeCP2 expression in the CNS is sufficient to rescue the majority of Rett syndrome phenotypes (Ross et al., 2016), supporting the therapeutic potential of MeCP2 delivery to the CNS. As a result, several studies are exploring MECP2 gene therapy, mostly using AAV and scAAV vectors for delivery of the MECP2 gene (Gadalla et al., 2013;Garg et al., 2013;Katz et al., 2016;Matagne et al., 2017;Sinnett et al., 390 2017). These studies have been spurring tremendous hope for gene therapy of Rett syndrome.
However, some challenges are still being overcome, such as toxicity from overexpression or mistargeting of the gene, packaging limitations of the viral vectors and limited brain transduction efficiency (Gadalla et al., 2017;Katz et al., 2016). In this study, we demonstrate intracellular delivery of the MeCP2 protein into neurons. Importantly, we also demonstrate that the delivered 395 MeCP2 protein binds heterochromatic DNA in the neurons, mimicking the functional endogenous MeCP2 (Carrette et al., 2018;Gadalla et al., 2013).
In this study we developed and characterized transgenic T. gondii lines capable of intracellular delivery of therapeutic proteins based on fusion to GRA16. This approach could be extended to the use of other endogenously-secreted T. gondii proteins which exhibit different 400 secretion dynamics and could be adapted to the therapeutic protein of interest. Furthermore, in addition to the delivery of endogenous mammalian proteins, this approach could be used for the delivery of engineered proteins, such as engineered enzymes, ligands, antibodies, transcriptional regulators or programmable nucleases (Gaj, Gersbach, & Barbas, 2013).

Acknowledgements:
We thank all members of the Rechavi and Sheiner labs for their endless help, support, enriching discussions and valuable feedback on the manuscript. We thank Prof. Daniel Frenkel and Prof.  Error bars represent mean ±SD. N=10-32 cells per condition. Take note that the y-axis of the leftmost graph is divided to two segments, to account for the high values recorded from GRA16-HA. AFU-arbitrary fluorescence units. Text below each scatter plot shows the mean relative nuclear localization ±95% confidence interval (relative to the full-length GRA16[505]-HA).

T. gondii culture and maintenance
Type I RH T. gondii were grown in human foreskin fibroblasts (

Generation of T. gondii clonal lines
To generate stable clonal lines, T. gondii were transfected as described, and genomic integration of the exogenous DNA was selected for by drug selection followed by clone isolation. After transfection, parasites were transferred onto a fresh dish with HFF cells. The next day, media was changed to fresh media containing the drug used for selection (pyrimethamine 525 for DHFR-TS selection or mycophenolic acid+Xanthine for HXGPRT positive selection).
Starting from the first detection of egressed extracellular parasites in the culture supernatant, 50-200 μl drops of the supernatant were passed daily to a second dish of HFF cells in selective media. When the parasites in the second dish started egressing from the HFF, drops of the supernatant were passed daily to a third dish, and so on. After the 3rd-4th dish started lysing 530 (about 3 weeks after transfection), the parasites were considered a "stable pool" containing parasites that integrated the exogenous DNA construct into their genome. The stable pool was immunofluorescently stained to assess the percentage of construct-expressing parasites in the stable pool as well as to evaluate protein localization upon genomic integration of the construct.
Clonal lines were isolated in 96-well plates with HFF by limiting dilutions or by FACS sorting 535 single parasites into each well. After 5-10 days of growth, clones were selected as wells containing a single plaque, and tested for integration of the construct via PCR using genomic lysate as template. Genomic lysates were prepared in lysis buffer (10% proteinase K in TE buffer) by incubation at 60 °C for 60 minutes followed by 95 °C for 10 minutes. PCR-positive clones were verified by immunofluorescent staining. Protein localization was determined 540 morphologically based on comparison to the polarized light image, and to co-staining with DAPI and anti-IMC1. Each fusion protein localization was validated over 2-5 independent transfections. At least 100 parasites were microscopically assessed by eye per transfection, and at least 20 z-stack images were acquired for each transfection, for transient and for stable expression (Tables S1 and S2).

Culture, neuronal differentiation and infection of LUHMES cells
Culturing and differentiation of LUHMES cells were carried according to Shah et al.

Preparation and infection of mice primary neuronal cultures
The cortex and hippocampi of WT P1 pups were dissociated enzymatically using papain and mechanically using gentle pipettation. Cells were counted and seeded in a 24-well plate on poly-l-lysin coated glass coverslips at a density of 100,000 cells per well, in neurobasal media (NBA, 2% B27 and 1% L-glutamine). After 5 days in culture, cells were infected with 580 tachyzoites of the GRA16-MeCP2 T. gondii line at MOI=1. 12 hours post inoculation, cells were washed with PBS, fixed and immunofluorescently stained as described.

Immunofluorescent staining (Immunofluorescence assay, IFA)
Cells were grown on 13 mm glass coverslips in a 24-well plates. At the respective time-585 point, cells were fixed in 4% PFA for 20 min at room temperature (RT). They were permeabilized and blocked by incubation with 2% blocking solution (2% bovine serum albumin, 0.2% Triton-X in PBS) for 20 minutes at RT. Blocking solution was removed, coverslips were covered with primary antibodies in blocking buffer and incubated for 1 hour at RT. Following 3 washes with PBS (5 minutes, shaking), they were covered with secondary antibodies in blocking 590 buffer and incubated for 40 minutes at RT, protected from light. Coverslips were washed in PBS 3 times, dipped briefly in water, blotted on paper to remove excess water and mounted on slides with Fluoromount-G mounting media containing DAPI (Southern Biotech, 0100-20). Slides were allowed to dry overnight at RT prior to imaging. For long term storage, slides were kept at 4 °C in the dark.

Microscopy data acquisition and analysis
Unless specified otherwise, all slides were imaged using a DeltaVision Core microscope (AppliedPrecision) using a 100x objective. Images were handled using the Fiji distribution of ImageJ, imported using the OME bio-formats plugin (Linkert et al., 2010) and deconvolved 600 using the Diffraction PSF 3D and Iterative Deconvolution 3D plugins (Dougherty, 2005). All adjustments of brightness and contrast were linear and applied to the entire image equally.
Background fluorescence was determined by sampling an "empty" area of the image, and maximum display threshold for the image was set to allow optimal visualization of the cell structures and protein localization (e.g. in the host cell nucleus). All raw images, imaging metadata, and ImageJ macros detailing the processing protocol are available on: https://github.com/shaharbr.

Analysis of T. gondii infection and protein delivery kinetics in HFF and in LUHMES neurons
Image analysis was performed using the open-source CellProfiler (Carpenter et al., 2006) software. Full protocol used for the image analysis, including the parameters chosen for identification of host cells and intracellular T. gondii, is available together with the raw data and 645 image analysis outputs on: https://github.com/shaharbr. In brief, host cell nuclei were identified using the DAPI channel and T. gondii were identified using the 594nm channel (corresponding to anti-IMC1 staining). Each T. gondii vacuole was associated to a host cell nucleus based on proximity. Fluorescence intensity on the 488nm channel, corresponding to the anti-HA staining was used to quantify the levels of GRA16 fusion-protein localization in the parasite vacuoles and 650 in the host cell nuclei. Importantly, timepoints above 24 hours post infection were removed from the dataset as we found that the parasite vacuoles were too large for efficient segmentation and host cell association by the used image analysis tool. Resulting data tables were analyzed using a custom python code. In brief, this analysis involved labeling and organization of the data,

Intrinsic disorder score and correlation to nuclear localization of the fusion proteins
The intrinsic disorder profile for the full translated open read frames of each GRA16 fusion protein was calculated using https://iupred2a.elte.hu/ , using the following parameters: Prediction type: IUPred2 long disorder (default), Context-dependent prediction: ANCHOR2 690 (Mészáros et al., 2018). A custom python code was used to calculate the averaged intrinsic disorder score of the fused therapeutic protein, calculate the correlation between the intrinsic disorder score and nuclear localization of each fusion protein and generate plots. The averaged intrinsic disorder score was calculated as the average of the ANCHOR2 score and the IUPred2 score along the span of the fused therapeutic protein (excluding GRA16-HA), minus the 0.5

Molecular cloning
For the plasmids encoding for Toxofilin-fused proteins, we used as a backbone a pGRA 700 vector containing Toxofilin cDNA fused to Cre recombinase, surrounded by the 1.1 kb genomic sequence upstream to Toxofilin ('Toxofilin promoter') and the 3' UTR of GRA2, a kind gift from A. A. Koshy (Koshy et al., 2010;Lodoen et al., 2010 To generate plasmids for the expression of Toxofilin-fused GDNF, PARK2 and MECP2, we PCR-amplified the mammalian cDNA with primers that add an EcoRV restriction site at the 5' and a PacI restriction site at the 3' of the cDNA. The Cre recombinase was replaced with the amplified mammalian cDNA by restriction-ligation (using SfoI and PacI for the backbone and 715 EcoRV and PacI for the cDNA, SfoI and EcoRV both produce blunt ends).
To generate plasmids for the expression of Toxofilin-fused ASPA, SMN1, GALC, GALC-TAT and MECP2opt, we amplified the pGRA vector and cDNA, and used NEBuilder assembly.
As GALC contains a region of exceptionally high %GC at the start of the gene, PCR amplification and NEBuilder assembly resulted in a deletion at the start of the gene. To fix this, 720 we performed an additional cloning step to add the missing fragment using the 'round-the-horn' (Stephen Floor, 2018) PCR blunt cloning method for both the Toxofilin-GALC and Toxofilin-GALC-TAT plasmids.
To generate plasmids for the expression of Toxofilin-fused TFEBopt, ASPAopt and GALCopt, we replaced the MEP2opt in the Toxofilin-MECP2opt plasmid with the respective 725 cDNA using restriction-ligation with EcoRV and PacI which were added to the sequence during DNA synthesis.
To generate plasmids for the expression of GRA16-fused genes, GRA16 and its promoter were amplified from RH T. gondii genomic DNA (primers were designed based on (Bougdour et al., 2013)). These GRA16 PCR products were used to replace the Toxofilin promoter and coding 730 sequence in the Toxofilin-ASPA plasmid, with the GRA16 promoter and coding sequence, to produce the GRA16-ASPA plasmid (by NEBuilder assembly).

Antibodies
Antibodies and the respective concentrations they were used in, are as follows: anti-HA

Statistical analysis
To determine the significance of nuclear localization of each GRA16-fused protein, we performed one-way ANOVA for the effect of the expressed construct, with multiple comparisons to the "no construct" control. To determine the significance of nuclear localization of each 750 truncated variant of GRA16, fused to HA or to HA-MeCP2, we performed two-way ANOVA for the effect of the GRA16 variant (rows) and of the fused sequence (HA or HA-MECP2, columns), with multiple comparisons to the "no construct" control. To determine the significance of the difference in infection rate between parasite lines in the kinetics analysis, for each timepoint we performed two-way ANOVA for the effect of parasite line (columns) and MOI (rows). To 755 determine the significance of nuclear levels of MeCP2 and GRA16-MeCP2, we performed twoway ANOVA for the effect of the time after inoculation (rows) and the condition (columns), with multiple comparisons to the to the "uninfected MECP2-KO" control. Multiple comparisons were performed using the following parameters: within each row, compare the means of each group to control, report multiplicity adjusted P value for each comparison (Dunnett test), one family per 760 row.