Identification and characterization of differentiation mutants in the protozoan parasite Toxoplasma gondii



Two forms of the protozoan parasite Toxoplasma gondii are associated with intermediate hosts such as humans: rapidly growing tachyzoites are responsible for acute illness, whereas slowly dividing encysted bradyzoites can remain latent within the tissues for the life of the host. In order to identify genetic factors associated with parasite differentiation, we have used a strong bradyzoite-specific promoter (identified by promoter trapping) to drive the expression of T. gondii hypoxanthine–xanthine–guanine phosphoribosyltransferase (HXGPRT) in stable transgenic parasites, providing a stage-specific positive/negative selectable marker. Insertional mutagenesis has been carried out on this parental line, followed by bradyzoite induction in vitro and selection in 6-thioxanthine to identify misregulation mutants. Two different mutants fail to induce the HXGPRT gene efficiently during bradyzoite differentiation. These mutants are also defective in other aspects of differentiation: they replicate well under bradyzoite growth conditions, lysing the host cell monolayer as effectively as tachyzoites. Expression of the major bradyzoite antigen BAG1 is reduced, and staining with Dolichos biflorus lectin shows reduced cyst wall formation. Microarray hybridizations show that these mutants behave more like tachyzoites at a global level, even under bradyzoite differentiation conditions.


The protozoan parasite Toxoplasma gondii is an important human and veterinary pathogen. This parasite can infect virtually any nucleated animal cell, but is rapidly controlled by the cellular immune response, leaving only a latent infection that can re-emerge periodically to provide a natural boost to host immunity (McLeod and Remington, 1987). The promiscuity of this parasite accounts for its prominence as a congenital pathogen, in both humans (McLeod and Remington, 1987) and animals (Dubey, 1986). In recent years, Toxoplasma has also achieved notoriety as a cause of life-threatening opportunistic disease in immunocompromised individ-uals, including patients with AIDS, primarily because of the reactivation of latent infection (Luft and Remington, 1992).

Sexual replication of T. gondii takes place only in cats, but the sexual cycle is not obligatory for parasite proliferation, and population genetic studies argue that sex is relatively rare in the field (Howe and Sibley, 1995). Asexual replication – in a wide variety of intermediate host species and tissues – is characterized by two forms: rapidly growing ‘tachyzoites’ and latent ‘bradyzoite’ tissue cysts (Dubey et al., 1998). Tachyzoites are responsible for acute illness and congenital neurological birth defects, whereas the more slowly dividing bradyzoite form can remain latent within the tissues for many years, representing a threat to immunocompromised patients. Infection is commonly acquired by ingestion of undercooked meat harbouring bradyzoite tissue cysts. After passage through the gut, rupture of the cyst releases infectious parasites capable of invading across the intestinal epithelium. Bradyzoites rapidly dedifferentiate to form tachyzoites, which disseminate throughout the body and can invade across the blood–brain barrier. During the course of infection, a small fraction of tachyzoites differentiates to form bradyzoites, perhaps in response to host cell stimuli, such as immune attack. Replication of these parasites is greatly slowed as they begin to express differentiation-specific markers and establish a cyst wall. An understanding of the mechanisms governing intercon-version between tachyzoites and bradyzoites might be expected to lead to new strategies for preventing tissue cyst formation and/or parasite re-emergence in immunocompromised patients.

Tachyzoites are readily cultured in the laboratory, and recent work by many groups has provided methods for studying bradyzoite differentiation in vitro (Popiel et al., 1994; Soete et al., 1994; Weiss et al., 1995; Bohne and Roos, 1997). Several stage-specific markers are now available to follow this process (reviewed by Weiss and Kim, 2000), and a variety of stress conditions (heat shock, alkaline shock, oxidative stress, pyrimidine starvation, etc.) have been shown to induce tachyzoite-to-bradyzoite differentiation (reviewed by Dubey et al., 1998; Weiss and Kim, 2000). An expressed sequence tag (EST) sequencing project has identified numerous T. gondii cDNAs (Ajioka et al., 1998;, and sequencing of stage-specific libraries has identified several transcripts that are specific to either the tachyzoite or the bradyzoite form (Manger et al., 1998).

A wide variety of molecular tools is now available for the genetic manipulation of T. gondii parasites (Donald and Roos, 1993; 1994; 1995; Kim et al., 1993; Boothroyd et al., 1995; Roos et al., 1997; Striepen et al., 1998). In particular, insertional tagging permits saturation mutagenesis of the haploid parasite genome, allowing the identification of any non-essential gene whose function can be assayed (Roos et al., 1997). Such approaches have been used to identify several bradyzoite-specific genes (Knoll and Boothroyd, 1998; R. G. K. Donald and D. S. Roos, unpublished). A similar approach can be considered for the identification of regulatory genes that control parasite differentiation: assuming that it is possible to generate a parasite line that can be selected against as bradyzoites, insertional mutagenesis should permit the identification of mutants that are unable to convert from tachyzoites to bradyzoites under differentiation conditions (Bohne and Roos, 1997).

We have taken advantage of a bradyzoite-specific promoter to drive stage-specific expression of the combined positive/negative selectable marker hypoxanthine– xanthine–guanine phosphoribosyltransferase (HXGPRT) (Donald et al., 1996) and identified two mutants that fail to induce this gene efficiently during bradyzoite differen-tiation in vitro. Further studies demonstrate that these mutants are also defective in other aspects of the differentiation pathway.


A selectable marker for identifying bradyzoite differentiation mutants

The first step in the identification of bradyzoite differen-tiation mutants requires developing a screen for selecting parasites that are unable to differentiate. Although various stage-specific markers are known (including both protein and sugar antigens), the intracellular nature of T. gondii infection means that most of these can only be visualized in fixed samples. We therefore sought to engineer a transgenic parasite that expresses a negative selectable marker (i.e. a marker that can be selected against) only under differentiation conditions.

The purine salvage enzyme HXGPRT has previously been shown to serve as a useful marker for either positive or negative selection, using mycophenolic acid (MPA) or 6-thioxanthine (6TX) respectively (Donald et al., 1996). In the first step of this analysis, a promoterless HXGPRT gene was fused upstream of a pyrimethamine resistance marker in vector pDHFR*-TSc3 (Donald and Roos, 1993) in order to trap numerous bradyzoite-specific promoters after insertional mutagenesis (R. G. K. Donald and D. S. Roos, unpublished results). The most potent of these promoters, designated pt7, was fused upstream of HXGPRT as shown in Fig. 1A. This construct was transfected into RH strain parasites in which both HXGPRT and uracil phosphoribosyltransferase (UPRT) genes had been deleted (RH ΔHXGPRT ΔUPRT; Bohne and Roos, 1997). The HXGPRT deletion in these parasites provides an appropriate host for selections, whereas the UPRT deletion results in pyrimidine starvation under low [CO2] conditions, leading to efficient bradyzoite induction (Bohne and Roos, 1997). Parasites were selected for high HXGPRT expression (mycophenolic acid resistance) under bradyzoite growth conditions, and low HXGPRT expression (6-thioxanthine resistance) under tachyzoite growth conditions, and several stable transgenic clones were isolated for further characterization. Parasite line 7-1 exhibits a highly regulated HXGPRT expression (see below) and was selected as the parental strain for further experiments.

Figure 1.

Strategy for isolating bradyzoite differentiation mutants.

A. Construction of the parental parasite line 7-1, harbouring a combined positive/negative selectable marker (HXGPRT) that is specifically induced under bradyzoite differentiation conditions.

B. Selection strategy used to isolate parasites harbouring regulatory gene mutations. This study reports on the isolation of putative mutants defective in positive regulation of bradyzoite gene expression. HXGPRT OFF indicates resistance to 6-thioxhantine (6-TX) and sensitivity to mycophenolic acid (MPA); HXGPRT ON is 6-TXS and MPAR.

C. Strategy for the identification of bradyzoite differentiation mutants by insertional mutagenesis. See text for further details.

A series of reconstitution studies was carried out in order to determine whether a mutant defective in the induction of bradyzoite-specific genes could be isolated from the overwhelming background of 7-1 parental parasites. 7-1 parasites were mixed with a mock mutant line (RH ΔHXGPRT ΔUPRT tachyzoites expressing β-galactosidase; Seeber and Boothroyd, 1996) at a ratio 50:1. This mixture was inoculated into a host cell monolayer, which was immediately shifted to bradyzoite differentiation conditions (low [CO2]) and subjected to selection in 6-thioxanthine for 3 days. Parasites were then switched back to tachyzoite growth conditions until they lysed out of the host cell monolayer, inoculated into six-well plates with coverslips, fixed and stained for β-gal expression (Seeber and Boothroyd, 1996). The results, summarized in Table 1, show that it is possible to obtain a good enrichment in the ‘mock mutant’ (10- to 30-fold) after a single cycle of selection under bradyzoite conditions, confirming that 7-1 parasites are suitable for mutagenesis experiments.

Table 1. Enrichment of mock differentiation mutants after a single cycle of selection under bradyzoite conditions.a
Parasite colonies expressing LacZFold enrichment
(after selection)
No selection6-thioxanthine
BlueWhiteLacZ+ (%)BlueWhiteLacZ+ (%)
  1. a. Parental parasites (ΔHXGPRT ΔUPRT, pt7HXGPRT) were mixed at a ratio of 50:1 with LacZ+ mock mutants (ΔHXGPRT ΔUPRT) in order to mimic rare misregulation mutants that have lost the ability to turn on HXGPRT under bradyzoite growth conditions. The mixed population was inoculated into HFF cells grown in T25 flasks, induced to differentiate into bradyzoites in low [CO 2] (see Experimental procedures) and cultivated for 3 days in the presence of 6-thioxanthine to select against the parental line (or without drug in controls). Parasites emerging after a single cycle of selection were inoculated into six-well plates containing HFF cells grown on coverslips, cultured for 24 h, fixed and stained for β-galactosidase expression (Seeber and Boothroyd, 1996).


As shown in Fig. 1B, one would expect that a mutant parasite in which a positive regulator of the differentiation cascade is inactivated would not be able to turn on pt7 (HXGPRT OFF) and would therefore be resistant to 6-thioxanthine in the bradyzoite stage. Conversely, a knock-out of a negative regulator is expected to produce parasites that express HXGPRT in the tachyzoite stage. In this report, we describe the isolation of putative positive regulator knock-out parasites.

Pyrimethamine-resistant DHFR-TS vectors (Donald and Roos, 1993; 1994) have been shown to integrate into the parasite genome effectively at random. The high frequency of transformation using DHFR-TS vectors permits the entire parasite genome to be tagged in a single electroporation cuvette (Roos et al., 1997). This strategy has been applied successfully to clone several genetic loci (Donald and Roos, 1995; Donald et al., 1996; Roos et al., 1997; Sullivan et al., 1999). 7-1 parasites were transfected with insertional mutagenesis vectors pDHFR*-TSc3 and pDHFR*-TSc3ABP (Roos et al., 1997), followed by selection in pyrimethamine. After lysis of the host cell monolayer, tachyzoites were inoculated in fresh cultures, shifted to bradyzoite growth conditions using CO2 starvation and selected in a combination of pyrimethamine and 6-thioxanthine for 3 days. Parasites were then returned to tachyzoite growth conditions in order to recover the surviving parasite population for another round of selection (Fig. 1C). After each round of selection, parasites were switched to bradyzoite growth conditions in six-well plates containing human foreskin fibroblast (HFF) cells grown on coverslips, and immunofluorescence assays were performed to check for HXGPRT expression. After three rounds of selection, several vacuoles exhibited unusually low levels of HXGPRT expression as bradyzoites. A fourth round of selection was then performed, and the parasites were cloned by limiting dilution (Roos et al., 1994). Each individual clone was tested for HXGPRT expression under bradyzoite conditions, and 40 clones (derived from two independent transfections) showing low HXGPRT expression were analysed further.

Southern blotting confirms tagging loci distinct from the HXGPRT

To examine the possibility that HXGPRT expression was lost as a result of inactivation of the parental HXGPRT gene(s), a Southern blot was hybridized with an HXGPRT probe as shown in Fig. 2A. Wild-type parasites (lane 1) show a single 6.5 kb restriction fragment corresponding to the genomic locus (Donald et al., 1996). The HXGPRT knock-out (lane 2) shows the expected smaller band, resulting from deletion of a 1.5 kb SalI fragment con-taining coding sequence essential for HXGPRT function (Donald et al., 1996). The 7-1 parental strain (lane 3) shows additional bands, corresponding to multiple copies of the pt7-HXGPRT transgene integrated into the parasite genome. None of these genes was disrupted in any of the 10 mutant clones examined (lanes 4–13).

Figure 2.

Genomic integration of the insertional mutagenesis vector. Genomic DNA was isolated from RH strain parasites, RH ΔHXGPRT ΔUPRT parasites (labelled RH ΔHΔU), the parental strain (RH ΔHXGPRT ΔUPRT parasites expressing HXGPRT under the control of a bradyzoite-specific promoter) and various mutant clones derived from two independent transfections (B- and P-series mutants). DNA was digested with BamHI and subjected to Southern analysis using a 32P-labeled HXGPRT cDNA fragment (A) or a fragment of DHFR cDNA derived from the insertional mutagenesis vector pDHFR*-TSc3 (B).

The same filter was rehybridized with probe derived from the insertional mutagenesis vector as shown in Fig. 2B. A high-molecular-weight band corresponding to the endogenous DHFR genomic locus can be seen in all lanes. As expected, only the mutants show additional bands derived from chromosomal integration of the mutagenesis vector. Clones designated ‘B’ or ‘P’ derive from independent transfections. Because all the mutants from each transfection appear to be siblings, only one clone was selected from each set for further studies.

Mutant parasites show increased replication rates under bradyzoite growth conditions

In order to characterize these putative bradyzoite differentiation mutants, we first examined parasite growth rates under tachyzoite and bradyzoite conditions. Wild-type bradyzoites exhibit a marked reduction in replication rate compared with tachyzoites (Dubey et al., 1998). When parasites were grown under tachyzoite conditions, no differences were observed between the parental and mutant strains (Fig. 3A). All cultures continued to replicate, lysing the entire host cell monolayer within 4 days. Under differentiation conditions, the parental strain grew slowly: most vacuoles in these cultures contained only four parasites after 4 days of growth (black bars in Fig. 3B), indicating a cell cycle of ≈ 48 h. (Each infectious T. gondii parasite establishes a distinct intracellular vacuole within which its progeny replicate, synchronously, for the duration of the infectious cycle (Fichera et al., 1995): four parasites in a vacuole result from two cell divisions over this 96 h period.) Parental parasites never managed to lyse out of the host cell monolayer – a typical observation when CO2 starvation is used to induce bradyzoite formation (Bohne and Roos, 1997). In contrast, both the mutant parasites grew much more rapidly than the parental strain. Four days after the shift to low [CO2], most of the intracellular parasitophorous vacuoles in flasks inoculated with the P11 mutant contained 16 parasites, indicating that four cell cycles had been completed during this time. Most B7 vacuoles contained 32 parasites (five cell cycles). Thus, the B7 and P11 mutants replicate at least twice as rapidly as the parental strain under differentiation conditions (cell cycle time of 12–24 h versus 48 h). The mutant parasites continued to proliferate, ultimately lysing the entire culture flask ≈ 8 days after the CO2 shift. Of note, although both P11 and B7 lysed out the monolayer, P11 grew more slowly than B7. This result has been reproducibly observed in all replicates of multiple experiments using different inducing conditions, suggesting different phenotypes for the two mutant lines under study.

Figure 3.

Growth rate studies. Parental (WT) and mutant (P11, B7) tachyzoites were inoculated into host cell monolayers and grown under tachyzoite conditions (A) or immediately shifted to bradyzoite differentiation conditions using either low CO2 to induce pyrimidine starvation (B) or alkaline stress (C). The number of parasites per vacuole was counted for 100 vacuoles in at least two experiments for each time point; error bars indicate average ± standard deviation.

Mutants were also induced to differentiate in response to alkaline stress – another method that had been used successfully to obtain bradyzoites in vitro (Soete et al., 1994). Similar results were obtained, as shown in Fig. 3C, although it should be noted that the alkaline stress induction procedure used involves CO2 starvation as well as pH shift, producing more stressful conditions than CO2 starvation alone (this probably explains why both mutants grew more slowly in Fig. 3C; note the change in scale on the y-axis). Experiments designed to induce differentiation without CO2 starvation (heat shock, alkaline stress in high [CO2], etc.) suggested similar effects (see below), but we have not been able to induce differentiation in the RH strain sufficiently well to obtain reliable measurements under these conditions.

Expression of bradyzoite markers is significantly reduced in the differentiation mutants

In order to determine whether downregulation of bradyzoite-specific expression is restricted to the HXGPRT transgene in these mutant parasites, we examined bradyzoite antigen expression levels, as shown in Fig. 4. Based on staining with an affinity-purified polyclonal antiserum raised against recombinant HXGPRT (Donald et al., 1996), expression in the parental strain was evident in only 10% of all vacuoles under tachyzoite growth conditions, but in >90% of vacuoles under bradyzoite conditions. In contrast, HXGPRT was observed in <20% of the mutant strains under all cultivation conditions tested (Fig. 4A). The major bradyzoite antigen BAG1 – a low-molecular-weight heat shock protein (Bohne et al., 1995) – behaved similarly, exhibiting stage-specific expression in parental parasites, but more limited induction in both mutants (Fig. 4B). Dolichos biflorus lectin – a marker that specifically stains the cyst wall (Boothroyd et al., 1997) – also showed a dramatic reduction in the mutants relative to parental parasites (Fig. 4C).

Figure 4.

Bradyzoite marker expression. Parental (WT) and mutant (P11, B7) tachyzoites were inoculated into host cell monolayers and grown under tachyzoite conditions for 24 h or shifted to bradyzoite differentiation conditions in low CO2 for 4 days. Fluorescence assays were carried out as described in Experimental procedures, using anti-HXGPRT serum (A), anti-BAG1 serum (B) or TRITC-labelled Dolichos biflorus lectin (C). Insets show examples of positive and negative staining for each marker from parental parasites (overlay of phase and fluorescence images). At least 100 parasitophorous vacuoles were examined for each sample in at least two replicate experiments; error bars indicate average ± standard deviation.

To examine bradyzoite expression using induction conditions that do not involve CO2 starvation, heat shock was applied as described previously (Soete et al., 1994). Bradyzoite induction in the RH strain is relatively poor under these conditions: only 30% of parental parasites showed positive staining for Dolichos biflorus lectin, 50% exhibited detectable expression of the bradyzoite surface antigen BSR4 (p36) (Knoll and Boothroyd, 1998), and <2% were positive for BAG1 (data not shown). Nevertheless, both mutants exhibited decreased expression of BSR4 after heat shock. Similar results were obtained using alkaline stress in high [CO2] to induce bradyzoite differentiation (not shown).

Microarray hybridizations suggest a global lack of bradyzoite gene induction

In order to examine the phenotype of putative differentiation mutants on a genomic scale, RNA from the parental 7-1 strain and one mutant (B7) was hybridized to a glass slide microarray containing 4224 ESTs from an in vivo bradyzoite EST library (Manger et al., 1998), 96 highly represented ESTs from a tachyzoite library (Ajioka et al., 1998) and various controls. The 2353 ESTs for which quality sequence is available ( form 613 contigs (including 235 singletons and 378 represented by two or more ESTs). The redundancy of this library allows for the expression levels of genes represented by multiple ESTs to be measured several times on a single array.

Two replicate induction experiments were performed for each parasite strain (parental and B7 mutant) by cultivation in low [CO2] for 4 days. RNA from each sample was extracted and hybridized with duplicate microarrays as described in Experimental procedures. All microarray data are freely available (see Supplementary material). Comparison of gene expression profiles in replicate experiments indicates high experimental reproducibility, as shown in Table 2: correlation coefficients were ≈ 0.86 for all replicates (a correlation coefficient = 1 would indicate complete identity). A lower correlation between parental tachyzoite and bradyzoite samples (0.64) suggests that there are many differences in gene expression when wild-type parasites are grown under tachyzoite versus bradyzoite conditions, as expected from the dramatic differences in their biology (Weiss and Kim, 2000).

Table 2. Correlation coefficients for microarray hybridization experiments.a
 Parental (7-1)Mutant (B7)
  • a.

    cDNA obtained from parasites grown under tachyzoite or bradyzoite conditions was hybridized to glass slide microarrays containing ≈ 4300 features, as described in Experimental procedures and discussed in the text. Correlation coefficients between individual experiments were calculated as the covariance divided by the product of standard deviations for each feature in the array.

  • b.

    Experiments performed on a different date.

Parental (7-1)Tachyzoite0.86b0.640.860.81
 Bradyzoite 0.870.640.61
Mutant (B7)Tachyzoite  0.86b0.82
 Bradyzoite   0.85

A high correlation between parental and mutant tachyzoites (0.86) indicates few differences during growth as tachyzoites, as shown in Table 2 and Fig. 5A. At a global level, gene expression in parental and mutant parasites was as similar as the expression observed in replicate experiments using parental (or mutant) tachyzoites alone. In contrast, the correlation coefficient for parental versus mutant parasites grown under bradyzoite conditions was very low (0.61), indicating substantial differences in gene expression (Table 2 and Fig. 5B). B7 mutant parasites grown under bradyzoite conditions were significantly more similar to tachyzoites, of either parental or mutant strains (correlation coefficients of 0.81 and 0.82 respectively).

Figure 5.

Microarray analysis of gene expression in parental and mutant parasites. RNA from parental and mutant parasites cultivated under tachyzoite or bradyzoite conditions was hybridized with an array containing 4224 ESTs, representing 613 contigs from an in vivo bradyzoite library and 96 highly expressed ESTs from a tachyzoite library, as described in Experimental procedures. Results were normalized based on hybridization to a DNA control and plotted for parental versus mutant parasites grown under tachyzoite conditions (A) or bradyzoite conditions (B). Axes indicate relative expression levels; the diagonal line shows the pattern expected for identical samples (see also Table 2). Fold induction (or repression) was calculated as the ratio of bradyzoite/tachyzoite expression levels and plotted for mutant B7 versus parental parasites (C). Concentric red boxes indicate two-, three- or fivefold changes in expression. ESTs belonging to the same cluster were grouped together to provided replicate assays for the same gene, and average values were plotted as squares in (D). Additional tachyzoite genes known to be highly expressed (from EST studies) but represented on the microarray by only a single feature are also shown (triangles). Horizontal lines indicate twofold induction or repression in the parental strain; yellow and green symbols represent genes suspected to be induced or repressed (respectively) in bradyzoites based on abundance in the EST database (Ajioka et al., 1998).

To examine gene expression profiles in greater detail, these data were transformed as described in Experimental procedures, and bradyzoite-to-tachyzoite ratios for each feature were plotted for mutant versus parental parasites (Fig. 5C). Note that this presentation displays levels of induction rather than expression, i.e. constitutively expressed genes cluster in the middle of the graph, regardless of their absolute expression level. The expression of most genes does not change during the shift from tachyzoite to bradyzoite conditions, as indicated by the large central cloud (smallest red box delineates <twofold induction or repression). Among those genes whose expression was significantly altered, a wide range of induction/repression was seen in parental parasites, from a 10:1 ratio favouring bradyzoites to a 6:1 ratio favouring tachyzoites (the paucity of highly repressed features is attributable to the relative abundance of bradyzoite cDNAs on the microarray, as noted above).

Focusing on the most reliable results from these studies, genes for which at least three known ESTs are represented on the microarray were grouped together, and average values for the expression of each gene were plotted for mutant versus parental parasites (squares in Fig. 5D). The identity of genes whose expression is induced more than twofold in parental parasites under bradyzoite conditions is shown in Table 3. Because tachyzoite-specific genes are under-represented on the microarray, singletons corresponding to genes known to be tachyzoite specific from EST studies and Northern analysis (Ajioka et al., 1998; Manger et al., 1998) are also included in Fig. 5D (triangles). Stage-specific transcripts identified in the T. gondii EST sequencing project ( are shown in yellow or green, representing bradyzoite- or tachyzoite-specific genes respectively. There is a strong correlation between the results obtained from EST sequencing and expression analysis in the parental parasites, validating the reliability of the microarray approach. The validity of these microarrays has also been confirmed by probing Northern blots with known constitutive, repressed and induced genes (Cleary et al., 2002).

Table 3. Transcripts represented by multiple replicates on the microarray that were induced more than twofold in wild-type parasites under in vitro bradyzoite conditions.a
 Parental parasites (wild-type)B7 mutant parasites
Cluster ID
(no. of features)
Best hitTachyBradyB/T ratioTachyBradyB/T ratio
  1. a. ‘Cluster ID’ refers to ‘Toxoqual3’ clusters describing overlapping EST clones ( The number of independent ESTs from each cluster spotted on the microarray is indicated in parentheses. ‘Best hit’ refers to the best BLASTX hit in GenBank for each cluster (italics indicates putative identification based on match with non-T. gondii sequences). Expression levels indicate fraction of the total hybridization signal on each array × 103.

  2. b. Also identified as upregulated in bradyzoites in vivo by analysis of EST abundance ( Ajioka et al., 1998; Manger et al., 1998).

  3. c. Also identified as upregulated in bradyzoites in the Prugniaud strain ( Singh et al., 2002).

3993 (3)b 5244.865142.91
4196 (3)bcSAG48354.129121.23
4054 (5)bcLDH24143.39440.97
3906 (11)bcBAG19202.638101.17
3908 (4)bcENO124522.5833220.73
3949 (4) 4122.53661.25
4229 (6)b 7132.34891.07
4303 (8)c 6132.158141.87
4133 (4) 5122.14661.06
4276 (4) RPS3a 29592.0235612.10

Of the 10 genes represented in Fig. 5D that were most highly induced in the parental strain (Table 3), five were previously identified as upregulated by EST analysis (Manger et al., 1998), and four of these encode known bradyzoite-specific proteins (SAG4, LDH2, BAG1 and ENO1). At least seven of the 10 genes induced in parental parasites grown under bradyzoite conditions were not significantly induced in the B7 mutant parasites, however, and the expression of one of the remaining genes (the most dramatically induced transcript, represented by EST cluster Ctoxoqual3_3993) was significantly less highly induced in the mutants.

It is clear that induction and repression levels in parental parasites were greater than in the mutants, i.e. the distribution in Fig. 5D tends towards the vertical. Taken together with the analysis of other indicators of bradyzoite differentiation (Figs 3 and 4), Tables 2 and 3 and Fig. 5 suggest that the B7 mutant is unable to pursue the tachyzoite-to-bradyzoite differentiation programme as efficiently as wild-type (parental) parasites.


The genetic basis and mechanisms underlying inter-conversion between tachyzoites and bradyzoites ranks among the most important – and least understood – aspects of T. gondii biology. In this report, we describe a selection scheme for generating mutants deficient in tachyzoite-to-bradyzoite differentiation. This strategy has permitted the isolation of two distinct mutants that misregulate the expression of a bradyzoite-specific selectable marker under differentiation conditions and show significantly increased replication rates and reduced expression of bradyzoite genes and other markers of differentiation. We are currently in the process of scaling up this strategy to isolate many more such mutants.

Given the availability of a large set of partially characterized ESTs for T. gondii (Ajioka et al., 1998), microarray approaches provide the means to examine stage-specific expression on a genome-wide scale. In general, the results obtained from microarray studies were concordant with data on EST abundance, as discussed above. Discrepancies, such as genes that appear to be bradyzoite specific in the EST database but not in the microarrays (yellow squares in Fig. 5D that lie towards the centre of the distribution) may be explained in some cases by the relatively small number of bradyzoite ESTs that have been sequenced to date. Other ESTs may represent genes that are induced late during bradyzoite differentiation. The EST project sequenced cDNAs obtained from mature bradyzoite tissue cysts generated in vivo, whereas the bradyzoites used for microarray hybridizations were generated by 4 days of induction in vitro. Four days of cultivation under differentiation conditions is probably not sufficient for maximal induction or repression of many stage-specific transcripts, as is known to be the case for the bradyzoite-specific antigen p21 and the tachyzoite-specific antigen SAG1 (Soete et al., 1994). Most of the genes identified as induced under bradyzoite conditions in this study were also identified as induced in the Prugniaud strain (Singh et al., 2002). Differences may be attributable to strain-specific variation, differences in the timing of differentiation or differences in the experimental conditions used (CO2 starvation versus alkaline conditions).

Microarray hybridizations suggest a global failure of the bradyzoite differentiation process in the B7 mutant: most of the stage-specific genes induced (or repressed) in parental parasites were not affected to the same extent in these mutants. Those genes that were induced in both parental and mutant lines may lie upstream of the B7 mutation in the differentiation pathway. Alternatively, some of the induced genes may be related to general stress, or to CO2 starvation, rather than bradyzoite differentiation per se. Interestingly, reduced induction of bradyzoite antigens LDH2 and BAG1 (Table 3), as well as SAG2C/D (not shown in Table 3, because this gene is represented by only a single feature in the microarray), was also observed in independent tachyzoite-to-bradyzoite differentiation mutants generated by chemical mutagenesis in the Prugniaud strain (Singh et al., 2002). It will be interesting to compare the B7 and P11 mutants, given their differences in replication rate.

The RH strain of T. gondii is unusually accessible to molecular genetic manipulation (Roos et al., 1994; Boothroyd et al., 1995), permitting saturation insertional mutagenesis of the entire parasite genome. Production of mutants by transgene insertion greatly facilitates identification and cloning of the relevant loci, using either plasmid rescue or inverse polymerase chain reaction (PCR) (Roos et al., 1997). Sequences flanking the plasmid integration site have been rescued, but show no similarity to anything currently in the database (data not shown); characterization will require more extensive cloning, isolation of full-length cDNA, complementation of the B7 mutant and reconstitution of the mutant phenotype via targeted gene knock-out. Unfortunately, the RH strain is not ideal for bradyzoite differentiation studies, because of the relative difficulty in producing bradyzoite cysts in vivo. These parasites appear to be very similar to less virulent strains when differentiation is carried out in vitro, however (Soete et al., 1994; Bohne and Roos, 1997), so we anticipate that results obtained using RH will be valid for other T. gondii strains as well. Tagged loci rescued from mutant RH parasites can readily be knocked out in avirulent strains (e.g. Prugniaud) to permit testing of their phenotype in vivo.

Experimental procedures

Parasite growth and differentiation

RH strain T. gondii tachyzoites (wild type) and ΔHXGPRT ΔUPRT mutants (deleted at both HXGPRT and UPRT loci; Bohne and Roos, 1997) were maintained by serial passage in primary cultures of HFF cells, as described previously (Roos et al., 1994). RH strain parasites do not form bradyzoite cysts in mice because of their high virulence, but have been reported to differentiate in rats (De Champs et al., 1997) and are capable of differentiation in vitro (Soete et al., 1994). The RH strain is also particularly well suited to molecular genetic manipulation (Roos et al., 1994; Boothroyd et al., 1995).

UPRT knock-out parasites can be induced to differentiate into bradyzoites in low [CO2], resulting in pyrimidine starvation (Bohne and Roos, 1997). CO2 depletion was accomplished by inoculating tachyzoites into a host cell monolayer in minimal essential medium without NaHCO3 (Gibco) but containing 25 mm HEPES (Gibco). Cultures were equilibrated at pH 7 and incubated at 37°C at ambient CO2 (0.03%). Alkaline induction of bradyzoite differentiation (Soete et al., 1994) was carried out by inoculating tachyzoites into a host cell monolayer, allowing the parasites to invade for 2 h and then adding fresh medium containing 1 g l−1 NaHCO3 and 50 mM tricine that had been adjusted to pH 8.1 with NaOH. These cultures were incubated at 37°C and ambient CO2.

Transfection, selection and insertional mutagenesis

The production of parental parasites expressing a bradyzoite-specific HXGPRT selectable marker, and the insertional mutagenesis of this strain to identify candidate differentiation mutants, is outlined in Fig. 1 and described under Results. Technical details are as follows. Transfections were carried out by electroporation (Roos et al., 1994) using 107 freshly lysed-out tachyzoites and 50 μg of plasmid DNA in a 2 mm gap cuvette (BTX): 1.5 kEV pulse, resistance 24 Ω. Selection conditions for HXGPRT expression (Donald et al., 1996) used 25 μg ml−1 mycophenolic acid plus 50 μg ml−1 xanthine. Selection against HXGPRT used 6-thioxanthine at 320 μg ml−1. Parasite clones were isolated by limiting dilution as described previously (Roos et al., 1994). Insertional mutagenesis was carried out by transfection using linearized plasmid pDHFR*-TSc3 or pDHFR*-TSc3ABP (Donald et al., 1996; Roos et al., 1997), both of which harbour a cDNA-derived T. gondii DHFR-TS minigene mutated to produce a pyrimethamine-resistant allele (Donald and Roos, 1993; 1994). These vectors integrate throughout the parasite genome by non-homologous recombination, conferring resistance to 1 μM pyrimethamine.


HFF cells were grown to confluence on coverslips in six-well plates, infected with parasites and examined at different times after infection using a Zeiss Axiovert 35 inverted microscope equipped with a 100 W Hg-vapour lamp and epifluorescence filter sets. For replication rate studies, the number of parasites per parasitophorous vacuole was determined for at least 100 vacuoles per time point in replicate experiments (Fichera et al., 1995). For immunofluorescence assays, parasites were fixed for 10 min in phosphate-buffered saline (PBS) containing 3% paraformaldehyde, permeabilized for 10 min in 0.25% Triton X-100 and blocked for 30 min in bovine serum albumin (BSA). Samples were incubated for 1 h with either polyclonal rabbit anti-T. gondii HXGPRT or monoclonal mouse anti-BAG1 (kindly provided by Dr W. Bohne). After washing three times, samples were incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody and tetramethyl rhodamine isothiocyanate (TRITC)-labelled Dolichos biflorus lectin (Sigma) (Boothroyd et al., 1997). At least 100 parasitophorous vacuoles were examined per sample.

Nucleic acid hybridizations

Southern blotting was carried out using genomic DNA from 5 × 107 parasites (6–7 μg), digested with BamHI, separated on 0.7% agarose gel and blotted to nylon membranes according to standard protocols. Blots were hybridized overnight with a 32P-labelled HXGPRT cDNA clone (Donald et al., 1996) or a HindIII–BglII fragment containing the promoter region and T. gondii DHFR coding sequence derived from the insertional mutagenesis vector pDHFR*-TSc3.

Microarray experiments were performed using a T. gondii cDNA microarray as described by Singh et al. (2002). This array contains 4224 ESTs representing at least 613 contigs from an in vivo bradyzoite library (Manger et al., 1998), 96 highly expressed ESTs from a tachyzoite library (Ajioka et al., 1998) and various parasite, human and vector control features. Many of the ESTs on this array derive from the same contig clusters (, providing extensive internal replicates for analysis (see Results).

Total RNA was isolated from parental and B7 mutant tachyzoites or bradyzoites (grown under differentiation conditions for 4 days) using the RNeasy midi kit (Qiagen). A minimum of two independent experimental inductions was carried out for each parasite strain. The percentage of T. gondii and human RNA in each preparation was determined by measuring the fluorescence intensity of each species’ large ribosomal subunit RNA in agarose gels stained with SYBR green (Molecular Probes). T. gondii RNA (3 μg) from each sample was reverse transcribed (Superscript II), and the resulting cDNA was labelled by random priming using Cy5-labelled nucleotides (AP Biotech). Labelled cDNAs were mixed with 500 ng of a Cy3-labelled reference standard consisting of PCR-amplified vector DNA (present in all ESTs spotted on the array), and this mixture was hybridized to the microarray overnight at 65°C. Microarrays were washed, scanned (GenePix scanner; Axon Instruments), and fluorescence intensity was determined using SCANALYZE software ( and the Stanford Microarray Database (Sherlock, 2001). All microarray data are freely available (see Supplementary material).

To compare gene expression levels, values for each sample (parental and mutant parasites, grown under tachyzoite and bradyzoite conditions) were determined by first calculating Cy5 (T. gondii)/Cy3 (reference) ratios for each spot to normalize for hybridization efficiency. No value was entered for poor-quality spots (based on measurements of pixel distribution and background fluorescence). Experimental reproducibility (Table 2) was assessed by calculating the correlation coefficient for replicate samples (covariance divided by the product of standard deviations), and spots with high variability (coefficient of variation ≥40%) were excluded, leaving 3356 experimental samples. Expres-sion levels for mutant versus parental parasites grown under tachyzoite or bradyzoite conditions are shown in Fig. 5A and B respectively.

Samples were normalized further by converting the Cy5/Cy3 ratios to the percentage of total hybridization [(Cy5/Cy3)/total hybridization of the sample ×100]. Fold induction (or repression) under bradyzoite conditions was calculated as the ratio of bradyzoite/tachyzoite expression levels for 2475 spots per array (excluding those where fold induction could not be calculated because hybridization in one or more samples was below background). Figure 5C plots fold induction under bradyzoite conditions for parental versus mutant parasites for the 1138 spots for which EST sequence is available. In cases in which at least three ESTs from the same ‘Toxoqual3’ cluster ( were available, these replicate samples were averaged (Fig. 5D).

Supplementary material

The following material is available from

Microarray data

The ‘Raw data’ file presents Cy5/Cy3 ratios indicating the hybridization of experimental samples over DNA reference controls. Clone # refers to dbEST identifier; ID refers to the gene name or EST cluster <>. The ‘Processed data’ file shows bradyzoite/tachyzoite expression ratios for those features used to generate Fig. 5C (see Experimental procedures for a description of data filtering).


We wish to thank Arzoo Iqbal for assistance in cell culture, Warren Ewens for advice on statistical analysis of micro-array data, John Boothroyd for his continued encouragement, and members of the Roos and Boothroyd laboratories for many helpful discussions. This research was supported by NIH grant R37-AI28724 to D.S.R. and a fellowship from the National Research Council of Argentina to M.M. D.S.R. is a Burroughs Wellcome Scholar in Molecular Parasitology. U.S. is supported by the NIH (K08 AI01453) and is a Burroughs Wellcome Fund Career Development awardee.