Toxoplasma gondii protease TgSUB1 is required for cell surface processing of micronemal adhesive complexes and efficient adhesion of tachyzoites

Tachyzoites lacking TgSUB1 have altered microneme protein processing

The TgSUB1 gene was disrupted by homologous recombination in two independent cell lines: RH GFP-Δhxgprt and RHΔhxgprt background as described elsewhere (Binder et al., 2008). Loss of TgSUB1 expression was confirmed by quantitative real-time PCR, Western blot and immunofluorescence analyses (Fig. 1A and Fig. S1). The expression, targeting and intracellular processing of MICs AMA1, MIC2, MIC3, MIC4, MIC5, MIC6 and M2AP were not affected in the Δsub1 strain (Fig. 1C; Binder et al., 2008; and data not shown). To determine if MIC proteolytic processing on the parasite surface was impaired, secretion of microneme proteins MIC2, MIC4 and M2AP, known substrates of MPP2 and MPP3 was analysed by Western blot. Efficiency of the secretion procedure was confirmed by Western blot analysis to show enhancement of secreted forms and absence of cytosolic proteins (data not shown). M2AP is first cleaved in its C-terminal region by MPP3 generating the fragment M2AP-1 and further cleaved by MPP2 generating three additional fragments (M2AP 2–4, Fig. 1B) (Zhou et al., 2004). MIC4 (72 kDa) is cleaved within its N-terminus to generate a 70 kDa species and is further processed by MPP2 in its C-terminus to generate 50 and 15 kDa products (Carruthers et al., 2000; Brecht et al., 2001; Zhou et al., 2004). MIC4 50 kDa and M2AP 1–4 secreted products are no longer detected in the Δsub1 strain and this loss of processed forms is associated with accumulation of the M2AP mature form (Fig. 1C). Results were similar in both Δsub1 strains (data not shown). In wild-type parasites, MIC2 is cleaved in its N-terminus by MPP2 removing a 5 kDa fragment (Fig. 1B) (Carruthers et al., 2000; Zhou et al., 2004), but in the Δsub1 strain, the MIC2-secreted product was of higher molecular weight, indicating that it too was no longer processed (Fig. 1C).

To identify other potential substrates for TgSUB1, we performed a global proteomic analysis of the ESA (excreted-secreted antigen) profile of Δsub1 (GFP background) and its sibling transfectant Clone 2 (which continues to express TgSUB1). ESA preparations were analysed by two-dimensional differential gel electrophoresis (2D-DIGE) analysis (Fig. 2A). Proteins that were diminished or absent in the Δsub1 strain are visible as green spots. As expected, TgSUB1-secreted products were not observed in the Δsub1 strain (80 kDa, 70 kDa, 40 kDa products and TgSUB1 prodomain spot D8) (Fig. 2A, Fig. S2 and Table 1). Inhibition of MIC2, MIC4 and M2AP processing was confirmed as depicted by the diminution of the MIC2 95–90 kDa products, M2AP 1–4 products and the MIC4 50 and 15 kDa products (spots D3-D4-D21, Fig. 2A, Fig. S2 and Table 1). Moreover, 2D-DIGE analysis confirmed the accumulation of the M2AP mature form and revealed the absence of MIC4 70 kDa product and accumulation of the MIC4 72 kDa (spot U2) precursor in the Δsub1 (Fig. 2A, Table 1 and Fig. S2).

Several additional proteins with differences in abundance were identified by mass spectrometry (Table 1). Among these was perforin like protein 1 (PLP1) which was originally identified in a proteomic screen of secretory products (Zhou et al., 2004). 2D-DIGE analysis revealed accumulation of a precursor form and diminution of a probable processed form (spots U1 and D9; Fig. 2A and Table 1). This was confirmed by Western blot analysis, which showed the absence of PLP1 processed product in the Δsub1 strain (Fig. 2B).

The abundance of several other proteins in ESA was changed in the Δsub1 strain including the microneme proteins MIC1 and MIC3 (Table 1, Fig. 2A). In addition to known micronemal proteins, other secretory proteins with altered abundance including the dense granule protein GRA7 and two SAG-related sequence (SRS) proteins (TGGT1_121850 and TGGT_113990; Table 1, Fig. 2A, spots D7-D16-U4). Previous studies characterizing ESA components have typically included dense granule proteins and surface antigens (Cesbron-Delauw et al., 1989; Zhou et al., 2004). The differences in ESA profile were reproducible in three independent biological replicates analysed by 2D-DIGE, as well as numerous independent ESA secretion analyses of both Δsub1 strains.

TgSUB1 GPI anchorage is necessary for MPP2 activity

The Δsub1 strain was complemented with full-length TgSUB1 (Δsub1::FLSUB1) or a TgSUB1 lacking the C-terminal GPI signal sequence (Δsub1::ΔGPISUB1). A clone where only the CAT cassette has been integrated was also selected as a control (Δsub1pCAT). TgSUB1 complementation and CAT cassette integration were confirmed by Southern blot (Fig. S1B). TgSUB1 expression was restored to wild-type levels as shown by quantitative real-time PCR and Western blot analyses (Fig. S1C and D). Both FLSUB1 and ΔGPISUB1 proteins were correctly targeted to the micronemes consistent with prior results showing the GPI anchor is not involved in TgSUB1 microneme targeting (Fig. 3A; Binder et al., 2008). Upon microneme discharge, TgSUB1 was secreted and correctly processed (Fig. S1E).

We performed IFA on extracellular parasites stimulated for microneme secretion and on invading parasites to test if TgSUB1 protein deposition upon the parasite surface after microneme secretion is dependent on the GPI anchor. In both the ΔGPISUB1- and FLSUB1-complemented strains, TgSUB1 was detected on the parasite surface, as observed for many MIC proteins lacking membrane association such as MIC5 (Brydges et al., 2006) and M2AP (Rabenau et al., 2001; Huynh et al., 2003). The expression, targeting and intracellular processing of the rhoptry and microneme proteins MIC2, MIC3, MIC4, MIC5, MIC6, M2AP and ROP 2,3,4 were not affected in the Δsub1-complemented strains (Fig. 3C, Fig. S1F and data not shown).

Complementation of the MIC protein surface processing was investigated by Western blot analysis of ESA from the complemented strains. Processing of MIC2, M2AP and MIC4 was restored in Δsub1::FLSUB1consistent with TgSUB1 being required for the proteolytic activity responsible for cell surface trimming of these proteins (Fig. 3C). Unexpectedly, only the first processing step of M2AP, which is ascribed to MPP3, was restored in the Δsub1::ΔGPISUB1 strain suggesting that SUB1 GPI surface anchorage is necessary for restoration of full proteolytic activity, including cleavage of substrates cleaved by MPP2.

Loss of TgSUB1 impairs tachyzoite attachment and invasion efficiency

Since surface proteolytic trimming is proposed to be involved in the activation of MICs for host cell interaction, we tested the ability of the different Δsub1 strains to attach to fixed host cells. A significant decrease in attachment (30–40%) was observed for Δsub1,Δsub1::ΔGPISUB1 and Δsub1pCAT strains (Fig. 4A). The ability of the different strains to invade human foreskin fibroblasts was assessed by a Red/Green invasion assay, which discriminates between parasites that can attach versus those that have successfully invaded a host cell (Huynh et al., 2003). After a 15 min incubation, invasion of Δsub1,Δsub1::ΔGPISUB1 and Δsub1pCAT strains was decreased by 45–60% compared with RHΔhxgprt and Clone 5 (Fig. 4B).

Complementation with the full-length TgSUB1 gene increased invasion (Δsub1::FLSUB1), but did not completely restore entry to parental levels, a finding that has been seen in several previous studies (Huynh et al., 2003; Cerede et al., 2005). When Δsub1 parasites were allowed to invade for 1 h, they showed comparable invasion to control parasites (data not shown). This suggests that Δsub1 parasites have a delay in invasion, a phenomenon previously seen for M2AP-deficient parasites (Huynh et al., 2003).

Strains lacking TgSUB1 have a defect in gliding motility

To determine if loss of TgSUB1-dependent MIC surface processing affects other parasite processes, we analysed parasite gliding motility using a commonly used in vitro assay that assesses trail formation and gliding motility on FBS-coated coverslips. Toxoplasma motility is characterized by three forms: helical, circular and twirling. Helical motility deposits wide-arcing or straight, long trails (non-circular) in contrast to circular motility (Hakansson et al., 1999). No trails were detected for Δsub1 and Δsub1pCAT whereas trails were seen for Δsub1::FLSUB1, Clone 5 and RHΔhxgprt, strains that have intact TgSUB1-dependent MIC surface processing (Fig. 5A). Very short preliminary trails were detected for Δsub1::ΔGPISUB1 suggestive of transient gliding movements. Similar results were obtained using long incubation times (data not shown). Parasite gliding was also examined by live video microscopy. Quantification of the different forms of gliding revealed that a greater proportion of the Δsub1 parasites were inactive compared with the Δsub1::FLSUB1 strain with significantly less helical movement (Table 2), which could account for the lack of trail deposition.

Since the defect in gliding on FBS-coated slides was more pronounced than the defect observed in invasion, we analysed trail formation on other biologically relevant substrates including heparin, chondroitin sulfate A and collagen I with the three representative strains: Δsub1::FLSUB1, Δsub1::ΔGPISUB1 and Δsub1. Although trails could be detected in all strains with all substrates, fewer and shorter trails were detected for the Δsub1::ΔGPISUB1 and Δsub1 strains when gliding on heparin and chondroitin sulfate A (Fig. 5B). To further analyse the gliding behaviour, the proportion of long trails was determined and trail length was measured (Table 3). Long trails were more frequent for the Δsub1::FLSUB1 strain when gliding on heparin or chondroitin sulfate A compared with the Δsub1::GPISUB1 and the Δsub1 strains. Moreover, there was a significant quantitative difference in gliding distance between the Δsub1::FLSUB1 versus the Δsub1::ΔGPISUB1 and Δsub1 strains on heparin and chondroitin sulfate A. Long trail per cent and gliding distance were slightly reduced on collagen I for the Δsub1::ΔGPISUB1 and Δsub1 strains compared with the Δsub1::FLSUB1; however, the differences were not statistically significant.

Live video microscopy of parasites gliding on collagen I was used to quantify the different forms of gliding (Table 2). Δsub1 parasites showed greater activity oncollagen I than on FBS, with a statistically significant increase in helical motility, explaining the recovery of trails. The videos revealed that slightly more Δsub1::FLSUB1 parasites were active on collagen I compared with Δsub1::ΔGPISUB1 parasites but the per cent of helical movement were similar, correlating with static gliding data. As gliding motility is also important for calcium-dependent induced egress (Meissner et al., 2002b), we next determined the proportion of lysed vacuoles after inducing egress with calcium ionophore. Nosignificant differences were observed in the proportion of lysed vacuoles between the strains (data not shown).

TgSUB1 expression is required for maximal virulence in the mouse model

We tested the virulence of the different strains in vivo in the mouse model. We first tested the intraperitoneal route of infection, which is the most common method used to study in vivo virulence of type I non-cyst-forming strains. There was no significant difference in the time of death between the strains using a range of inocula (10–1000 parasites, data not shown).

Because dissemination through the vasculature is important during natural T. gondii infection, we next infected the mice intravenously. Mice infected with 250 Clone 5 or Δsub1::FLSUB1 parasites died within 10 days. Mice infected with the Δsub1 and Δsub1pCAT strains died within 14 and 15 days. Unexpectedly, the Δsub1::ΔGPISUB1 strain was more attenuated in intravenous infection with 15% of the mice surviving infection (Fig. 6A). Similar results were obtained with the second Δsub1::ΔGPISUB1 clone (17% survival with 250 parasite inoculum, Fig. 6A). Infected mice exhibited clinical evidence of cerebral infection (palsy, atrophy, agitation), and all mice that survived infection were seropositive.

Inappropriate release of the active TgSUB1 into the blood and host tissues could activate the endothelium and induce a more potent inflammatory response compared with the Δsub1 strain. We therefore quantified parasite burden and host cytokine profiles after intravenous infection with Δsub1::ΔGPISUB1, Δsub1 and Clone 5. Δsub1::ΔGPISUB1 parasite burden was diminished in all organs although this difference was statistically significant in the lungs only (Fig. 6B). Cytokine profiles elicited by the three strains differed somewhat, but were not suggestive of a more robust inflammatory response to Δsub1::ΔGPISUB1 parasites (Fig. 6C). These studies are consistent with a defect in either in vivo replication or dissemination for strains lacking full TgSUB1 activity, but more detailed characterization of parasite in vivo viability and host response are required.

Parasite maintenance

Toxoplasma gondii RH strain Δhxgprt parasites (Donald and Roos, 1998), and mutants derived from this strain (all originally RH-ERP), were maintained by continual passage in human foreskin fibroblast (HFF). Cells were grown in Dulbecco's modified Eagle medium (DMEM; Gibco) containing 10% fetal bovine serum (FBS), 1% Penicillin/Streptomycin and 1% Glutamine. Parasites and cells were cultured in a 5% CO₂ atmosphere at 37°C.

Generation of TgSUB1KO- and TgSUB1KO-complemented strains

The TgSUB1 gene was disrupted by homologous recombination with a construct containing the HXGPRT minigene flanked by three kilobases of the 5′ and 3′ sequence flanking the TgSUB1 coding region. Two strains with TgSUB1 disruption were generated (RH gra1-GFP5S65T-tubcatΔhxgprt and RHΔhxgprt background) (Binder et al., 2008). In each case, a clone that had integrated the TgSUB1 disruption cassette but continued to express TgSUB1 was chosen for parallel phenotypic analysis (Clone 2 for GFP strain and Clone 5 for non-GFP strain). Complementation with TgSUB1 was performed in the RH Δsub1Δhxgprt background. Either the full-length TgSUB1 gene or the TgSUB1 gene lacking the sequence encoding for the 26 last amino acids containing the signal for GPI anchorage was amplified from cDNA. The PCR products were cloned into a plasmid containing three kilobases of 5′ and 3′TgSUB1 flanking sequence by standard cloning procedures. The TgSUB1-complemented strains were generated by co-transfecting the Δsub1 strain with 100 µg of NotI-linerarized TgSUB1 complementation construct and 10 µg of NotI-linearized plasmid containing a tub-cat selection cassette. Stable lines were obtained by selection with mycophenolic acid/xanthine and chloramphenicol and cloned by limiting dilution. For each complementation, two different clones from independent transformations were isolated and tested.

Indirect immunofluorescence microscopy

For labelling of intracellular tachyzoites: HFF monolayers were grown on glass coverslips in 24-well plates until confluent. Freshly egressed tachyzoites were allowed to infect confluent cells for 20–24 h. Cells were fixed for 20 min in 3% paraformaldehyde and permeabilized for 10 min in 0.2% Triton X-100 in PBS1X and blocked (PBS1X, BSA 3%). Antibody dilutions used were: rabbit anti-TgSUB1 AE653 (1:500; Binder et al., 2008) and mouse anti-MIC2 mAb 6D10 (1:500, gift of David Sibley, Washington University School of Medicine, USA). Alexa Fluor-conjugated secondary antibodies (Molecular Probes) were used at 1:2000 dilution. Coverslips were mounted on slides with the Vectashield mounting medium (Vector Laboratories) and viewed with an Olympus Digital Microscope (Albert Einstein College of Medicine Analytical Imaging Facility).

For labelling of extracellular tachyzoites: freshly lysed parasites were resuspended in Hank's balanced salt solution (HBSS), with or without 2 µM A23187, and incubated at 37°C for 5 min to induce microneme secretion. The parasites were then added to glass coverslips before being fixed (3% paraformaldehyde) and blocked (PBS, BSA 3%). Parasites were then labelled with a rabbit anti-TgSUB1 AE653 (1:500, PBS, BSA 3%) followed by a conjugated anti-rabbit secondary antibody (Molecular probes, 1:2000). Alternatively, freshly egressed parasites were resuspended in DMEM media and allowed to infect HFF cells for 5 min at 37°C. Invading parasites were immunolabelled with the mouse anti-SAG1 DG52 monoclonal antibody (1:500; gift from John Boothroyd, Stanford University School of Medicine) and the anti-SUB1 rabbit antiserum AE653.

Secretion assay

Freshly lysed tachyzoites (2–3 × 10⁷ parasites) were filtered, centrifuged and resuspended in DMEM 1% FBS, 20 mM Hepes. Parasites were pre-incubated for 10 min at room temperature with 1 µM cytochalasin D. Microneme secretion was induced by adding 1% ethanol and warming parasites to 37°C for 10 min as previously described (Carruthers et al., 1999b). Cells were removed by centrifugation at 1000 g for 10 min at 4°C. The supernatant containing the microneme-secreted products was used for Western blot analysis. Large-scale secretion preparations for 2D-DIGE were generated as described previously (Zhou et al., 2004).

Two-dimensional differential gel electrophoresis

Microneme-secreted contents from 1 × 10⁹ parasites were analysed by 2D-DIGE as previously described (Zhou et al., 2004). Proteins were labelled using amine reactive cyanine dyes (minimal Cy dyes, GE Healthcare) according to the manufacturer's recommendations. The microneme-secreted supernatant from Clone 2 (GFP-positive) and Δsub1GFP parasites were labelled with two different fluorochrome dyes (Cy5 or Cy3). An equal amount of the two different microneme-secreted supernatants were mixed and labelled with a third fluorochrome dye (Cy2) as an internal standard control. An equal amount of the three labelled samples were mixed together and then separated by a two-dimensional gel (pH = 4.2–6.8 first dimension and 12.5% SDS-PAGE second dimension). Spots were visualized using a Typhoon^TM 9400 imager (GE Healthcare) and analysed using DeCyder-2D (v.5.0, GE Healthcare). The software analyses the overlaid signals and detects differentially expressed proteins. Proteins on the two-dimensional gel were transferred to Immobilon membrane and stained with Coomassie blue G250 and spots of interest were excised from the gel and digested with sequencing grade modified procine trypsin (Promega, http://www.promega.com). Tryptic peptides were fractionated by reverse-phase HPLC. Eluted peptides were sequenced by tandem mass spectrometry analysis. Proteins were identified by searching the data against the EPICDB Toxoplasma proteomic database (http://toro.aecom.yu.edu/biodefense;Madrid-Aliste et al., 2009) using MASCOT and are based on two or more peptide sequences with scores above the 95% confidence threshold.

Western blotting

Protein lysates were loaded on an 8% SDS gel and then transferred to nitrocellulose membrane. Membranes were probed with rabbit antisera recognizing TgSUB1 (1:10 000 PfSUB1, gift from Michael Blackman, National Institute for Medical Research, UK), SAG1-p30 (1:10 000, gift from Lloyd Kasper, Dartmouth Medical School) and MIC5 (1:10 000) followed by anti-rabbit horseradish peroxidase (HRP; 1:15 000) antibody. Secreted microneme proteins (ESA or excreted secreted antigens) were loaded on SDS gels of different resolution according to the protein probed (from 6% to 10% acrylamide). Membranes were probed with the following antibodies: mouse anti-MIC2 6D10 1:2000 (gift of David Sibley, Washington University School of Medicine, USA); rabbit anti-M2AP 1:10 000; rabbit anti-PfSUB1 1:2500) and rabbit anti-MIC4 1:10 000 (gift from Dominique Soldati-Favre, University of Geneva, Switzerland). The membranes were then probed with a secondary anti-rabbit or anti-mouse antibody (1:5000–10 000) conjugated to HRP. Parasite lysis was evaluated by probing the membrane with an antibody directed against cytosolic proteins (mouse anti-actin or tubulin 1:2000). Signals were revealed with a peroxidase Western blot detection reagent (SuperSignal Pierce). All gels were run under reducing conditions.

Quantitative real-time PCR

RNA from freshly egressed parasites was purified using Trizol followed by chloroform extraction. Then 3.5 µg of total RNA were retrotranscribed using the SuperScript kit (Invitrogen). Quantitative real-time PCR was performed on the 7300 ABI apparatus using the Power SYBR Green Master Mix (Applied Biosystems) according to the manufacturer's instructions. PCR primers were designed with the Primer Express software (Applied Biosystems) to amplify a 100 bp target gene fragment. PCR was performed in a 10 µl volume containing 7 ng of cDNA and 0.3 µM of each primer. cDNA levels were normalized to cDNA α-tubulin levels. Experiments were performed thee times with two different RNA preparations.

Invasion and attachment assays

Red/green invasion assays were performed as previously described (Huynh et al., 2003). HFF monolayer coverslips were infected with freshly egressed filtered tachyzoites (7.5 × 10⁶ per chamber) for 15 min at 37°C. The coverslips were then fixed (paraformaldehyde 3%), and blocked (PBS1X, BSA 3%). External attached parasites were labelled with a mouse anti-SAG1 DG52 antibody (1:500; gift from John Boothroyd, Stanford University School of Medicine) followed by a conjugated anti-mouse secondary antibody. Samples were then methanol permeabilized, blocked, and all parasites were labelled with a rabbit anti-Toxoplasma antiserum (Y321 1:1500; gift from Louis Weiss, Albert Einstein College of Medicine) followed by a conjugated anti-rabbit secondary antibody. Data were compiled from four independent experiments counting 10 fields per clone per coverslip at 60 × magnification. The number of invaded parasites is the difference between the total number of parasites (green) and the number of attached parasites (red).

For attachment assays, 1 × 10⁷ parasites were loaded with 1 µM calcein-AM, 30 min at room temperature (Invitrogen) and then allowed to attach for 15 min at 37°C on fixed HFF cells (2% glutaraldehyde for 5 min at 4°C, blocking with 0.16 M ethanolamine). Data were compiled from four independent experiments counting 20 fields per clone per coverslip at 40× magnification.

Replication/growth rate assay

Freshly lysed filtered parasites (5 × 10⁴) were allowed to invade and replicate in HFF coverslip monolayers for 24 or 38 h. Cells were fixed, permeabilized and immunolabelled with the rabbit anti-Toxoplasma antiserum Y321 (1:1500) followed by a conjugated anti-rabbit secondary antibody. The number of parasites per vacuole was enumerated on duplicate coverslips counting approximately 100 vacuoles per time point. Growth rate was determined in three independent experiments.

Induced egress assay

Egress was induced with the ionophore A23187 as previously described (Black et al., 2000). Briefly, HFF 24-well coverslip monolayers were inoculated with 10⁵ parasites and allowed to grow for 38 h. Host cells were incubated for 3 min at 37°C with modified HBSS (1 mM MgCl₂, 1 mM CaCl₂, 10 mM NaHCO₃, 20 mM Hepes) containing 2 µM A23187 or DMSO (control). Cells were fixed with paraformaldehyde and vacuoles were labelled by indirect immunofluorescence microscopy.

Static gliding assay

Glass coverslips were coated overnight at 4°C with 50% FBS/50% PBS1X or 50 µg ml⁻¹ heparin, chondroitin sulfate A or collagen I (solubilized in acetic acid). Freshly lysed filtered tachyzoites were resuspended in HHE (HBSS, 10 mM Hepes, 1 mM EGTA) and 250 µl was inoculated on the coated slides for 30 min at 37°C. Slides were then fixed with 3% paraformaldehyde. Trails left by gliding parasites were visualized at 40 × by staining with the mouse anti-SAG1 DG52 antibody followed by a conjugated anti-mouse secondary antibody. Between 40 and 200 trails were enumerated per strain per treatment. Only trails associated with a parasite were counted. A trail was considered as circular if the diameter was approximately the length of a tachyzoite; trails larger in diameter or straight were considered as non-circular. Trail lengths were measured in parasite body length.

Live gliding microscopy

One T25 flask of parasites was filter-purified, pelleted and resuspended in 10 ml of HHE (HBSS, 10 mM Hepes, 1 mM EGTA). Two millilitres of parasites was placed in a 35 mm glass bottom Petri dish and allowed to settle at room temperature for 10 min, before placing on a 37°C heating block for 5 min to activate parasites. The dish was transferred to a heated ring placed on the stage of a Zeiss Axio inverted microscope. Videos were captured at one frame per second for 1 min, using the Axio time-lapse software. The type of gliding movement was enumerated by observing individual parasites for each strain on each substrate. Maximum projection images showing the motility patterns over the 1 min period were generated by exporting the individual frame and reimporting them as Z-stacks. The time-lapse movies were generated by exporting the frames from the 1 min period, increasing the speed to 5 ×, and converted the image series to an avi file.

In vivo virulence assay

Groups of four to five female CD-1 mice between 7 and 9 weeks of age were infected intraperitoneally or intravenously with different parasite inocula. The survival of mice after infection was monitored daily over a period of 8 weeks. Plaque assays were performed with the different inocula to ensure equal amounts of viable injected parasites. The seroconversion of all surviving mice was assayed 4 weeks post infection by Western blot analysis by probing mouse serum against RH tachyzoite lysate. Results represent three independent experiments. Care and handling of animals was performed in accordance with AAALAC guidelines after approval of protocols by the Albert Einstein College of Medicine Animal Use Committee.

Parasite tissue burden and cytokine tissue level quantification

Groups of two or three female CD-1 mice between 7 and 9 weeks of age were infected intravenously with 250 parasites of the Clone 5, Δsub1, Δsub1::FLSUB1 and Δsub1::ΔGPISUB1 strains. Six days after injection, spleen, lung and brain were collected and DNA was purified using TRIzol. Parasite genomic DNA load was determined by quantitative real-time PCR using parasite specific primers (Tox 9–11; Reischl et al., 2003). Number of parasites per mg was calculated using a standard curve generated with genomic DNA from RHΔhxgprt tachyzoites. Groups of three mice were infected intravenously with 250 parasites of the Clone 5, Δsub1::ΔGPISUB1 and Δsub1 strains. At days 3 and 6 after injection, RNA was purified using TRIzol. Five micrograms of total RNA was retrotranscribed using the SuperScript kit (Invitrogen) and 1 µl of cDNA was used in 10 µl of SYBR green quantitative real-time PCR reaction. Cytokine levels are shown as the relative proportion of cDNA compared with uninfected tissue and normalized to GAPDH cDNA levels.

Statistical analysis

Statistical analyses were performed using the Prism Graph pad software. Two-tailed Student's t-test was used for analysis of invasion, attachment, replication, egress and gliding assays. The level of significance for mouse virulence assays was determined with the Kaplan–Meier estimator. Differences were considered significant if P-value was < 0.05.