|Home | About | Journals | Submit | Contact Us | Français|
Apicomplexan parasites are obligate intracellular pathogens that rely on actin-based motility to drive host cell invasion. Prior in vitro studies implicated aldolase in coupling actin filaments to the cytoplasmic domains of surface adhesins in the parasite. Here, we tested the essentiality of this interaction in host cell invasion. Homology modeling indicated a partial overlap of the binding surfaces between the enzyme active site and the region responsible for interaction with the microneme protein 2 cytoplasmic tail domain (MIC2t). Targeted mutagenesis delineated residues unique to each activity based on in vitro studies. Complementation of a conditional knockout (cKO) of the T. gondii aldolase gene (TgALD1) with mutants defective in either distinct function was used to test their respective roles. Our studies demonstrate aldolase is not only required for energy production, but is also essential for efficient host cell invasion based on its ability to bridge adhesin-cytoskeleton interactions in the parasite.
Toxoplasma gondii is the causative agent of toxoplasmosis, an opportunistic infection of serious concern to immunocompromised patients and the developing fetus (Hill and Dubey 2002). As an obligate intracellular pathogen, T. gondii must penetrate host cells to replicate and survive, a process shared by related apicomplexans such as malaria (Plasmodium spp.) (Sibley, 2004). To power host cell invasion, apicomplexans utilize gliding motility, a substrate-dependent mode of locomotion unique to these organisms and an essential component of parasite dissemination (Sibley, 2004). Gliding motility relies on formation of actin filaments and disruption with chemical agents renders the parasite immotile (Wetzel et al., 2003), and also prevents host cell invasion (Dobrowolski and Sibley, 1996). Motility also relies on TgMyoA, a class XIV myosin that provides the motive force for parasite motility and invasion (Meissner et al., 2002). TgMyoA, with its regulatory light chain (TgMLC1), is anchored into the inner membrane complex through association with the integral membrane protein TgGAP50 (Gaskins et al., 2004; Opitz and Soldati, 2002). Thus anchored, the nonprocessive motor walks along the actin filament toward the barbed end, while contact with the substratum is provided by micronemal proteins that are deposited at the parasite apex following secretion (Sibley, 2004).
One important family of micronemal proteins that contributes to motility and invasion is the thrombospondin-related anonymous protein (TRAP) family (Robson et al., 1995). Originally identified in P. falciparum based on homology to conserved domains, paralogues were subsequently identified in other motile stages of the malarial life cycle (Dessens et al., 1999; Yuda et al., 1999), and in other apicomplexans such as Eimeria and Cryptosporidium (Spano et al., 1998; Tomley et al., 1991). TRAP was shown to be essential in gliding motility and invasion of the mosquito salivary glands by P. berghei sporozoites (Sultan et al., 1997). The TRAP orthologue in T. gondii is known as MIC2 (Wan et al., 1997), which is also essential to parasite motility, adherence and efficient host cell invasion (Huynh and Carruthers, 2006). Like other TRAP proteins, MIC2 is a type-I transmembrane protein with extracellular domains related to integrin A domain and thrombospondin type 1 (TSR1) repeats and a short cytoplasmic domain (Wan et al., 1997). The extracellular A-domains are thought to bind receptors on host cells or substratum, including glycosaminoglycans and heparin-like molecules, thus providing anchorage for traction (Harper et al., 2004; Ménard, 2001). The cytoplasmic tails of MIC2 (MIC2t) and TRAP (TRAPt) were found to interact with the glycolytic enzyme aldolase based on in vitro pull-down assays (Buscaglia et al., 2003; Jewett and Sibley, 2003). This interaction depends upon a conserved penultimate tryptophan and a collection of acidic residues in the extreme C-terminus of the cytoplasmic domain (Buscaglia et al., 2003; Starnes et al., 2006). The generation of a co-crystal structure of aldolase with a TRAPt peptide highlighted the importance of the conserved penultimate tryptophan and confirmed the importance of the acidic residues, which are believed to establish a series of hydrogen bonds with charged residues near the enzyme active site (Bosch et al., 2007). Based on these in vitro studies, it was proposed that aldolase may serve as the bridge between the adhesin and the cytoskeleton in vivo, thereby connecting external adhesins to the actin-myosin motor (Jewett and Sibley, 2003).
In eukaryotes, class I aldolase exists as a homotetrameric enzyme (Penhoet et al., 1967) that participates in glycolysis by cleaving fructose-1,6-bisphosphate into two products: glyceraldehyde-3-phosphate and dihydroxy-acetone phosphate (Meyerhof et al., 1936). Each subunit of aldolase contains an active site, situated in the center of the established (α/β)8 barrel fold (Sygusch et al., 1987). The tetrameric state of the enzyme provides a platform for aldolase to cross-link actin filaments through a conserved set of residues that overlaps the catalytic pocket (Wang et al., 1996). The ability of aldolase to bind both F-actin and the cytoplasmic tails of adhesins such as MIC2 (MIC2t) and TRAP (TRAPt), would facilitate bridging of the adhesin to the cytoskeleton during gliding motility in apicomplexans. Despite compelling in vitro data supporting aldolase-MIC2t interactions and evidence for co-localization and co-precipitation from parasite lysates (Buscaglia et al., 2003; Jewett and Sibley, 2003), the essentiality of aldolase in vivo remains untested. Moreover, for this model to be correct, aldolase must be able to separately participate in glycolysis and bridge to the cytoskeleton, raising the issue of how these disparate activities are controlled.
In the current study, we sought to directly test the role of aldolase in bridging adhesin-cytoskeletal interactions during motility and invasion in the parasite. To this end, we generated a series of aldolase mutations that delineated MIC2t binding from enzyme activity and tested these by complementing a conditional knockout (cKO) of T. gondii aldolase (TgALD1).
Previous studies demonstrated that a stretch of acidic residues in the MIC2 cytoplasmic domain (MIC2t) that flank a tryptophan residue (MIC2-W767) are critical for binding to TgALD1 in vitro (Fig. S1) (Starnes et al., 2006). Homology modeling of TgALD1 revealed a basic groove that surrounds the substrate-binding pocket (Fig. S1). Residues previously implicated in F-actin binding were also encompassed in this same basic cleft (Fig. S1) (Wang et al., 1996). The basic nature of this cleft suggested it may interact with the acidic MIC2t through electrostatic interactions. Modeling the interaction with aldolase revealed the indole ring of MIC2-W767 was sandwiched between the hydrophobic side chains of aldolase arginine 303 (ALD-R303) and arginine 42 (ALD-R42), while the carboxyl group of MIC2-D765 makes several key hydrogen bonds with basic side groups in ALD-R148, ALD-K41, and ALD-R42 (Fig. 1A). The interactions predicted through these modeling experiments indicate the importance of both electrostatic and hydrophobic interactions in mediating the binding of the MIC2t to TgALD1.
To determine if the basic cleft of aldolase contributes to the interaction with the MIC2t, we generated a series of alanine point mutants in TgALD1. The ability of these TgALD1 mutants to bind to a GST-MIC2t fusion protein was assessed using an in vitro pull-down assay. These experiments demonstrated that residues within the basic cleft of TgALD1 contributed to the interaction with the MIC2t to varying degrees (Fig. 1B). For example, the mutations ALD-K146A and ALD-R148A demonstrated the greatest decreases in binding (11% and 2% of wild type, respectively) (Fig. 1C, Table 1), while the mutations ALD-K41A, ALD-R42A and ALD-K106A showed greater than 60% reductions compared to the wild type enzyme (Fig. 1C, Table 1). The double mutant ALD-K41E, R42G exhibited a loss in binding nearly equal to the single mutant at position ALD-R148A. Other mutations had modest to no effects, except ALD-E34A, which showed a 2-fold increase in binding. Collectively, these results demonstrate the MIC2t interacts with key residues in the basic groove of aldolase with the most significant contributions residing deep within the pocket and along a positively charged ridge defined by K41 and R42.
Since residues implicated in binding to the MIC2t were close to the substrate-binding site, the same point mutants were also evaluated for enzyme activity. Most mutants that exhibited decreased binding also showed defects in enzyme activity (Table 1). In contrast, ALD-K41A and ALD-R42A, retained similar levels to wild type enzyme activity as shown by kcat levels. While the double mutant ALD-K41E, R42G retained partial activity, it showed a 5-fold reduction in kcat. The results from the purified pull-down and activity assays were used to assign the aldolase mutations into one of three classes (Fig. 1D, Table 1). The first class includes those residues with minimal decrease in the interaction with the MIC2t but that cannot cleave substrate (orange residues in Fig. 1D, typified by ALD-D33A, Table 1). The second and largest group was unable to carry out either function (blue residues in Fig. 1D, typified by ALD-R148A, Table 1). The final class of mutations, including ALD-K41A and ALD-R42A and the double mutant ALD-K41E, R42G, retained normal or partial enzyme activity, but had significant decreases in MIC2t binding (green residues in Fig. 1D, Table 1). Collectively, these results demonstrate an overlap between the enzyme active site and the MIC2t binding surfaces, yet also indicate the two functional interfaces can be separated by specific mutations.
The aldolase-bridging model predicts that aldolase also interacts with F-actin in order to connect the cytoplasmic tails of adhesins with the cytoskeleton, thus assuring force transduction. To investigate this interaction, we selected a representative from each of the above classes of aldolase mutations and evaluated F-actin binding in a co-sedimentation assay. Approximately 55% of the wild type enzyme was associated with F-actin, while the ALD-D33A mutation demonstrated a 2-fold decrease in binding. All other mutations examined exhibited an approximately 5-fold decrease in their binding capacity (Fig. 1E, F). These results imply that the aldolase surfaces responsible for binding to F-actin and MIC2t largely overlap.
Previous studies using heterologous antisera to rabbit aldolase indicated that TgALD1 was apically localized and deposited in the trails of gliding parasites (Jewett and Sibley, 2003). However, this finding has been recently questioned by a study reporting that aldolase, along with other components of glycolysis, are translocated to the periphery in extracellular vs. intracellular T. gondii parasites (Pomel et al., 2008). The original lot of this commercial antisera is no longer available. Hence, to further evaluate the location of aldolase, a polyclonal antisera raised against TgALD1 was used to visualize aldolase during intracellular growth and extracellular gliding. In parasites growing intracellularly, aldolase was localized to the cytoplasm and demonstrated a partial apical co-localization with MIC2, consistent with previous reports (Jewett and Sibley, 2003). A second cytoplasmic protein, pyruvate kinase (PK), exhibited a similar cytoplasmic localization, but without a concentration at the apical end (Fig. 2). During gliding motility, aldolase remained primarily cytoplasmic in a pattern that was concentrated at the apical end, with a small amount being deposited the trails (Fig. 2B, arrowheads). Quantification of these distributions using a profile plot along the anterior-posterior axis, confirmed that aldolase shows a strong apical to posterior gradient, unlike PK that was uniformly distributed (Fig. 2B). When the profile was drawn perpendicular to the long axis, aldolase was not concentrated appreciably at the cell periphery, but instead was consistently elevated throughout the cytosol, resulting in a plateau effect (Fig. 2B). These analysis were performed on deconvolved images, and while only a single layer is shown in Fig. 2B, the same pattern was observed throughout the Z-stack in multiple examples. In parasites that were undergoing gliding, aldolase, as well as the surface protein SAG1, was deposited in trails (Fig. 2B bottom panel). These results support our previous findings obtained with the heterologous antibody to aldolase (Jewett and Sibley, 2003), and are similar to reports of the distribution of aldolase in malaria sporozoites (Buscaglia et al., 2003). However, under no circumstance did we observe relocalization of aldolase to the periphery in extracellular parasites, as previously suggested (Pomel et al., 2008). The reason for this discrepancy is unclear but may reflect differences in the antisera or experimental conditions used. Nonetheless, under the conditions studied here, aldolase was found in the cytosol, concentrated at the anterior end of the parasite, and deposited in the trails during gliding motility.
To test the role of aldolase in vivo, we utilized the tetracycline (Tet)-repressible system to generate a conditional knock-out (cKO) (Meissner et al., 2002). The genome of T. gondii contains two genes that are annotated as fructose-1,6-bisphosphate aldolase, although one of these is expressed in tachyzoites at the mRNA level (GLS, unpublished data), as reported previously (Pomel et al., 2008). The functional TgALD1 locus was disrupted by homologous recombination in a merodiploid line expressing a regulatable, HA-9-epitope tagged copy of TgALD1 (Fig. 3A). Multiple cKO clones exhibited equivalent levels of down-regulation and from this pool a single clone was selected for further analysis. The cKO parasites were initially characterized by immunofluoresence using the anti-TgALD1 antibody following growth for 24 hr in the presence of 1.5 μg/ml ATc (Fig. 3B). The cKO parasites expressed less aldolase than the parental TATi-1 line in the absence of ATc, and expression was strongly reduced when cultured in ATc. Down-regulation of aldolase was determined by quantitative Western blot following 48 hr of growth in 1.5 μg/ml ATc, and found to be decreased by ~95% (Fig. 3C). As expected, the presence of ATc had no effect on TgALD1 levels in the TATi-1 line. These experiments demonstrate successful disruption of the chromosomal copy of TgALD1, as well as a significant down-regulation of the Tet-regulatable copy of TgALD1 in the cKO.
To test the aldolase-bridging model in vivo, it was paramount to differentiate between parasites compromised in glycolysis versus those with selective disruption of the motor complex. To independently address these functions, a wild type, C-myc epitope tagged TgALD1 and a representative from each class of aldolase mutation described above, was used to complement the cKO. Approximately equal levels of protein expression were determined by Western blot analysis (Fig. 4A), with the exception of the double mutant ALD-K41E, R42G, which showed elevated levels. Proper cytoplasmic localization was confirmed by immunofluoresence against the C-myc epitope (Fig. 4B). The resulting complemented parasite clones were used to test the role of aldolase in bridging to actin and MIC2t versus its contribution to glycolysis.
The contribution of aldolase to parasite survival was initially evaluated using a monolayer lysis assay, which captures invasion, replication, egress, and re-invasion of a host cell monolayer. Addition of ATc prevented the cKO and lines complemented with the enzyme null mutants (i.e. ALD-D33A and ALD-R148A) from disrupting the monolayer (Fig. 5A), reflecting decreased growth of these lines. In contrast, complementations with mutants with normal enzyme activity (i.e. ALD-K41A and ALD-R42A) lead to disruption of the monolayer, even in the presence of ATc (Fig. 5A). The one exception to this trend was the double mutant ALD-K41E, R42G, which showed a partial reduction in host cell monolayer lysis when cultured in the presence of ATc (Fig. 5A). In the absence of ATc, the cKO and all of the complemented clones were able to lyse the host monolayer efficiently, thus demonstrating that none of the constructs exerted a dominant negative phenotype (Fig. 5A).
To determine the contribution of aldolase to parasite replication, the average number of parasites per parasitophorous vacuole was determined by microscopic examination. Following culture for 36 hr in ATc, the cKO demonstrated a significant growth defect compared to the absence of ATc (Fig. 5B). In agreement with the lytic assay, the cKO complemented with wild type aldolase and the mutants that retained enzyme activity, rescued the growth defect in the presence of ATc, while those without detectable enzyme activity did not (Fig. 5B). In contrast to the lytic assay, the double mutant ALD-K41E, R42G grew normally in the presence of ATc, indicating that it was capable of supporting normal parasite replication. The appreciated growth defects were a direct result of the presence of ATc as all parasite lines grew equally well in its absence (data not shown).
In addition, we determined whether aldolase mutants could produce levels of ATP in the parasite equivalent to the wild type enzyme. Cellular ATP concentrations were determined with a luciferase assay following growth for 48 hr in the presence of ATc. Comparison of the cKO in the absence ([ATP] ≥ 236 nM) vs. presence ([ATP] ≥ 92.7 nM) of ATc indicated that cellular ATP levels were significantly decreased when aldolase expression was repressed. We tested a subset of aldolase mutants that were used to complement the cKO line, again based on the different functional classes defined above. The mutants that retained enzyme activity and restored parasite growth in vivo (i.e. ALD-K41A, ALD-R42A), produced equivalent, if not greater levels of ATP to the wild type complemented line (see above)(Table 1). Normal levels of ATP were also produced by the double mutant ALD-K41E, R42G, despite its lower efficiency in vitro (Table 1). Consistent with the growth assay, the enzyme null mutants (i.e. ALD-D33A and ALD-R148A) produced significantly less ATP ([ATP] ≤ 135, P ≤ 0.005) than wild type levels. These results indicate that the enzymatic activity of aldolase is crucial for ATP generation and normal parasite growth. Additionally, these results demonstrate that the defect of the double mutant ALD-K41E, R42G in the monolayer lysis assay was not due to defective ATP production.
We next sought to determine if aldolase serves a critical role in bridging the MIC2-cytoskeletal interaction during parasite motility. Following growth for 48 hr in ATc, parasites were allowed to glide on BSA-coated coverslips and motility was evaluated as the number of trails deposited per area. In the cKO, the down-regulation of aldolase resulted in a significant decrease in the number of trails produced (Fig. 5C, P ≤ 0.001). Complementation with the wild type enzyme rescued the phenotype, while the cKO complemented with enzyme dead mutants failed to restore normal gliding (Fig. 5C). When complemented with mutants that displayed decreased binding to the MIC2t but retained enzyme activity (i.e. ALD-K41A, ALD-R42A, and ALD-K41E, R42G), an appreciable difference in parasite motility was not observed. This is in contrast to the observation that reduced expression of MIC2 compromises helical, but not circular gliding (Huynh and Carruthers, 2006). Collectively, these results imply that the primary contribution of aldolase to parasite gliding motility may be due to its role in energy production, although other explanations are also possible (see below).
Additionally, we utilized the complemented aldolase mutants to determine the contribution of aldolase to host cell invasion. Following growth in ATc, the cKO and complemented clones expressing the enzyme dead mutants (i.e. ALD- D34A, ALD-R148A) exhibited a significant decrease in host cell attachment and invasion (P ≤ 0.001), compared to parasites complemented with the wild type enzyme (Fig. 5D). This is consistent with the lower levels of ATP generated by these mutants (Table 1) and likely indicates an energy requirement for efficient invasion. While we have not tested the K146A mutant in these assays, we would expect it to behave similarly to R148A, given their similarly compromised enzyme activity observed in vitro (Table 1). The single point mutants with normal enzyme activity but impaired binding to the MI2Ct (i.e. ALD-K41A and ALD-R42A), showed normal host cell attachment, but were significantly less efficient at host cell invasion (P ≤ 0.05, compared to the wild type complement) (Fig. 5D). The double mutant ALD-K41E, R42G also showed normal adherence to the host cell, but demonstrated a far greater host cell invasion defect (P ≤ 0.005) (Fig. 5D). To more directly compare the different lines, the capacity to invade host cells following attachment was expressed as percentage total cell-associated parasites. The single mutants with impaired binding to the MIC2t showed partial reduction in the efficiency of invasion (Fig. 5D). The invasive efficiency of the cKO complemented with the double mutant ALD-K41E, R42G was reduced to 41.0%, which was substantially lower than either of the enzyme dead mutants or the wild type cKO cultured in the presence of ATc (Fig. 5D). Collectively, these data indicate that aldolase enzyme activity is necessary for efficient cell invasion, in part due to a requirement for ATP. Additionally, independent of enzyme function, aldolase serves a critical role in bridging interactions between the MIC2t and the cytoskeleton as shown by the mutants ALD-K41E, R42G.
We generated a conditional knockout (cKO) of T. gondii aldolase (TgALD1) to address whether this glycolytic enzyme also serves an essential role in bridging the cytoplasmic domain of MIC2 to the cytoskeleton in vivo. Structural modeling and point mutagenesis were used to delineate mutations that separately affected enzyme activity versus bridging of MIC2t and F-actin. Homology modeling highlighted residues in aldolase that form a basic groove and participate in electrostatic interactions with acidic residues in the MIC2t. Mutational analysis revealed that many of the charged residues in aldolase that contribute to the MIC2t interaction were also required for enzymatic activity; however, the two activities could be separated into distinct sub-domains enabling independent testing of these two functions. Complementation of the aldolase cKO with various point mutants demonstrated that aldolase contributes to parasite growth and efficient host cell invasion. A requirement for ATP generation affected motility, invasion and growth, while independent of glycolytic activity, the binding of aldolase to MIC2t was essential for efficient invasion of host cells. This requirement was shown most convincingly by the double mutant ALD-K41E, R42G, which retained normal ATP levels, replicated normally, and yet was significantly impaired in cell invasion. Collectively, these findings demonstrate an essential role in vivo for aldolase to serve as the physical connection or bridge between adhesins and the cytoskeleton.
Prior mutational studies have shown the penultimate tryptophan in the cytoplasmic tails of MIC2 and TRAP contribute to the interaction with aldolase in vitro and in vivo (Buscaglia et al., 2003; Jewett and Sibley, 2003; Starnes et al., 2006). Our modeling studies revealed that the indole ring of MIC2-W767 was stably anchored between the ALD-R42 and ALD-R303 side chains. However, this was not due to a cation-π interaction, which usually predominate between aromatic residues and arginine side chains during protein-protein interactions (Ma and Dougherty, 1997). Instead, the association was hydrophobic in nature as the indole ring is stabilized by interactions with the alkyl carbons in the arginine side chains. A similar interaction was observed in co-crystal structures between rabbit muscle aldolase and a WASP peptide (St-Jean et al., 2007), as well as in a TRAPt peptide and Plasmodium aldolase (Bosch et al., 2007). Our modeling studies extend these findings to the interaction between aldolase and MIC2t, and reveal that the binding site is conserved among these peptides, based on shared acidic residues and a tryptophan residue.
Homology modeling also predicted a series of electrostatic interactions between the MIC2-D765 and basic residues within the aldolase cleft (i.e. ALD-K41, ALD-R42 and ALD-R148). Individual mutations of these charged residues partially decreased binding and a much more dramatic phenotype was appreciated in the double mutant ALD-K41E, R42G, which demonstrated a ~20-fold decrease in MIC2t binding in the pull-down assay. Interestingly, we found that the residues in aldolase that were crucial for MIC2t binding were also required for an interaction with F-actin, a subset of which were previously identified in rabbit aldolase (Wang et al., 1996). Aldolase has a penchant for participating in associations unique from its role in glycolysis as evidenced by the reported interactions with the glucose transporter GLUT-4 (Kao et al., 1999), phospholipase D (Kim et al., 2002), and the Wiskott-Aldrich Syndrome peptide WASp (St-Jean et al., 2007). The interaction with actin and MIC2t may be a case of fortuitous adaptation as the parasite developed a mechanism to utilize multivalent interactions to provide increased power for host cell invasion.
Previous crystallographic studies were used to predict that the residues contributing to TRAPt binding were not distinct from residues contributing to enzyme activity (Bosch et al., 2007). In contrast, our extensive mutational analysis combined with biochemical testing performed here defined two residues (i.e. ALD-K41 and ALD-R42) found at the edge of the basic cleft that uniquely contributed to the MIC2t interaction. A double mutant in these residues (i.e. K41E, R42G) demonstrated an even greater decrease in binding to MIC2t. All three of these mutations maintained in vitro enzyme activity, generated normal levels of ATP, and were able to sustain normal parasite growth. The retention of normal enzyme activity in these mutants is likely attributed to the position of the ALD-K41 and ALD-R42 residues, which lie along the edge of the basic groove, jutting out at the periphery of the enzyme active site. Importantly, the role of these residues was not previously appreciated and their identification made it possible to test the separate roles of aldolase in binding to MIC2t versus glycolysis in vivo.
In order to test the contribution of the aldolase-bridging model to parasite motility and invasion in vivo, it was first necessary to establish a conditional knock out (cKO) line for T. gondii. A series of aldolase mutants that were screened in vitro for their ability to disrupt in either glycolysis or tail binding, as well as both functions, were then used to complement the cKO and to define the role of aldolase in parasite survival. These studies demonstrated that the predominant contribution of aldolase to parasite motility was its role in glycolysis and energy production; as all mutants that affected glycolysis also reduced motility. In contrast, mutants only affecting binding to MIC2t had no motility phenotype. This result was in contrast to what was expected and to results of a previous study examining a conditional knockout of TgMIC2, in which gliding motility was impaired and the residual gliding occurred in a nonproductive circular pattern (Huynh and Carruthers, 2006). In contrast to its role in gliding, aldolase binding to the MIC2t and/or F-actin was required for efficient host cell invasion. The single mutants (i.e. ALD-K41A and ALD-R42A) and the double mutant ALD-K41E, R42G exhibited an ~25% and 50% decrease in invasion efficiency, respectively. The pronounced decrease in host cell invasion was not due to decreased energy production. Given the observed decreases in the efficiency of binding to the MIC2t in vitro, these findings indicate that the phenotype of these mutants is most likely attributable to a disruption of the bridging activity required for linking the adhesin to the cytoskeleton in vivo. Collectively, these experiments suggest that MIC2t linkage to the cytoskeleton by aldolase plays a greater role when the parasite encounters resistance and traction is required (i.e. invasion), but may be less crucial during gliding under the in vitro conditions tested here. In the context of an infection, movement across and through tissues may require additional traction that cannot be appreciated using in vitro assay. Hence, migration in vivo may have a greater reliance on aldolase for bridging. Testing this hypothesis would require an in vivo model of infection, but unfortunately serial culture passages have completely attenuated the TATi-1 line used here for the cKO, making this experiment unfeasible.
Our findings and previous studies indicate the same region of aldolase mediates binding of substrate, actin and the cytoplasmic tails of apicomplexan adhesins (Bosch et al., 2007; St-Jean et al., 2007; Wang et al., 1996). While F-actin and MIC2t binding are mutually exclusive on a single monomer, they may interact with individual subunits in a tetramer to form a productive complex. The aldolase tetramer forms two planar dimers that are offset on the horizontal axis, positioning each dimer to interact separately. Thus, a single aldolase tetramer may present one dimer for MIC2t binding, leaving the remaining dimer for F-actin binding. Such a conformation could also potentially stabilize the filamentous state of T. gondii actin, which is inherently unstable (Sahoo et al., 2006).
In addition to being important for bridging to MIC2t, aldolase provides an essential function in glycolysis. Repression of the wild type enzyme in the cKO revealed a substantial defect in cell motility and cell invasion. This likely reflects the combined effects of reduced cellular ATP and disruption in binding to the MIC2t. It has recently been reported that aldolase and other glycolytic enzymes redistribute to the inner surface of the inner membrane complex (IMC) in extracellular parasites (Pomel et al., 2008). While the role of this relocalization has not been clearly established, it was suggested that it might locally enhance energy production and hence facilitate motility. While we did not observe this peripheral localization of aldolase in extracellular parasites, the role of aldolase in glycolysis is not at odds with our findings that aldolase participates in bridging to the adhesin MIC2t. Rather, aldolase that is engaged with the tail of MIC2 is presumably partitioned between the plasma membrane and outer leaflet of the IMC, and hence segregated from the pool of aldolase in the cytosol that participates in glycolysis.
The binding residues that mediate the interaction of aldolase with MIC2t partially overlap with enzyme activity, and yet our data indicate that both functions are separately required for parasite survival. This raises important questions about how the parasite is capable of coordinating these two mutually exclusive functions under conditions where they might co-exist in the cell. Given the estimated binding affinity in physiologic salt (KD 0.375 μM (GLS, unpublished data)) and relative stochiometry (3 μM MIC2 vs. 30 μM TgALD1 (GLS, unpublished data)), the majority of the MIC2t is predicted to be complexed with aldolase in the cytoplasm without significantly decreasing the amount of enzyme available for glycolysis (i.e. in the absence of substrate, the predicted MIC2t occupancy is ~98%, leaving 90% of aldolase free). While not a direct measure of interaction affinities, the results from the purified protein pull-down assay can be used to estimate the amount of MIC2t that is occupied in the cKO complemented with various mutants. For example, in the ALD-K41E, R42G double mutant the interaction is decreased by 20-fold, but the cellular concentration of MIC2 does not change and thus 30% of the cellular stores of MIC2 will likely be occupied by aldolase as a result. This level of occupancy may explain the residual invasion seen with the double mutant. The residual invasion and motility phenotypes of the single and double mutants may also be due to incomplete shutdown by ATc and the resulting residual activity of wild type aldolase (estimated to be ~ 5 %). As a consequence, partial occupancy of MIC2-ALD may be sufficient to transiently engage the cytoskeleton and mask any potential role of aldolase in motility. Additionally, our previous studies demonstrated that a second, upstream cluster of acidic residues in the MIC2t is also essential in vivo yet does not participate in aldolase binding (Starnes et al., 2006). Hence, it remains possible that other proteins contribute to linking MIC2 and possibly related adhesins, to the cytoskeleton. Despite these limitations, our studies clearly define an important role for aldolase binding to MIC2t in facilitating efficient parasite invasion. In the absence of this activity, parasite invasion was significantly decreased despite maintaining normal levels of ATP.
Our studies reveal that aldolase impacts not only energy production but also stabilizes and increases the efficiency of invasion. Detailed biochemical and mutational analyses delineated specific residues critical for linking the cytoplasmic tail of MIC2 to the cytoskeleton and separated this function from residues participating in cellular energy production. In vivo complementation of the cKO further elucidated the requirement of aldolase bridging in apicomplexan host cell invasion. Our findings demonstrate that aldolase functions in a complex to bridge MIC2 to the cytoskeleton in vivo. Disruption of this interaction, combined with inhibition of other key components in the motor complex, may allow new avenues of treatment to prevent infection by apicomplexan parasites.
Toxoplasma gondii tachyzoites were maintained by growth in monolayers of human foreskin fibroblasts (HFFs), as described previously (Starnes et al., 2006). Chloramphenicol (20 μg/ml) (Sigma-Aldrich, St. Louis, MO), phleomycin (5 μg/ml) (InvivoGen), anhydrotetracycline (ATc) (1.5 μg/ml) (Clontech, Palo Alto, CA) and pyrimethamine (3 μM) (Sigma-Aldrich) were added to the media as indicated. The cKO and complemented strains were propagated in HFF cells cultured in complete medium with tetracycline free fetal bovine serum (HyClone, Logan, UT).
GST-MIC2t (containing residues 721–769 of MIC2) was purified from the E. coli BL21 strain, concentrations determined using the BCA assay (Thermo, Rockford, IL) and used for pulldown assays, as described previously (Starnes et al., 2006). Recombinant T. gondii aldolase was co-incubated with GST-MIC2t or GST alone and bound proteins were eluted in sample buffer, resolved by SDS-PAGE, stained with SYPRO Ruby protein gel stain (Molecular Probes, Eugene, OR) and quantified using a FLA-5000 phosphorimager (Fuji Film Medical Systems, Stamford, CT). Values given are mean % of control ± SEM for three independent experiments. Co-Sedimentation Assay-- Purified rabbit skeletal muscle actin (Cytoskeleton Inc., Denver, CO) was diluted to 10 μM in 1X G-Buffer (5 mM Tris (pH 8.0), 0.2 mM CaCl2, 0.5 mM DTT, 0.2 mM ATP), and centrifuged at100,00 × g for 45 min using a TL100 rotor in a Beckman Optima TL ultracentrifuge (Becton Coulter, Fullerton CA) to remove protein aggregates. Actin was polymerized by the addition of 1/10 total volume 10X F-Buffer (500 mM KCl, 20 mM MgCl2, 10 mM ATP) at room temperature for 1 hr. Purified aldolase was added to polymerized actin in a 1:10 molar ratio, allowed to bind for 30 min at room temperature, and centrifuged at 100,000 × g for 1 hr at 25°C. The pellet was resuspended in sample buffer, resolved by 10% SDS-PAGE and aldolase detected by Western blot analysis and quantified using a FLA-5000 phosphorimager, as described previously (Starnes et al., 2006).
In vitro aldolase activity was determined by measuring NADH oxidation as a function of decreasing absorption/minute at 340 nm, as described previously (St-Jean M, 2005). Activity was measured in triplicate for different substrate concentrations using a Beckman Coulter DU 640 Spectrophotometer (Fullerton, CA). The KM and Vmax values were determined using the Michaelis-Menton equation in the KaleidaGraph 4.0 graphic software package (Synergy Software, Reading PA).
Aldolase was detected using rabbit polyclonal antisera to TgALD1 (Starnes et al., 2006). MIC2 was detected using rat polyclonal sera raised against the MIC2 ectodomain. T. gondii β-tubulin was detected using rabbit polyclonal antisera (Morrissette and Sibley, 2002). The surface antigen, SAG1, was detected with the monoclonal antibody DG52. Pyruvate kinase was detected using rabbit polyclonal antisera, generously contributed by Takashi Asai (Keio University, Tokyo, Japan). C-myc was detected with monoclonal antibody 9E10 (Invitrogen, Carlsbad, CA).
For immunofluorescence (IF) localization in intracellular parasites, monolayers of HFF cells were infected with freshly egressed parasites, cultured overnight, fixed with 4% formaldehyde in PBS for 20 min at 4°C and permeabilized with 0.25% Triton X-100. IF localization using antibodies to T. gondii proteins was performed as described previously (Starnes et al., 2006). IF localization in extracellular parasites was performed as described above, except cells were permeabilized with 0.1% Saponin. Images were acquired in wide field as Z-stack series (0.25 um steps) using a Zeiss Axioskop 2 MOT Plus microscope equipped with a 63 × 1.3 numerical aperture lens and a Axiocam Mrm camera (Carl Zeiss Inc, Thornwood, NY). Images were deconvolved using Axiovision v4.2 using the nearest neighbor algorithm. Profiles were generated from representative z-slices in AxioVision v4.2. Images were processed using linear settings in Adobe Photoshop.
Amino-terminal HIS6-tagged TgALD1 was generated by PCR amplification from a previously described clone (Jewett and Sibley, 2003) and directionally cloned into the pET16b vector (Novagen, Madison, WI) using the NdeI and BamHI restriction sites. The aldolase point mutants were generated using the QuickChange Site-Directed Mutagenesis Kit from Stratagene (Cedar Creek, TX). Proteins were induced in BL21 E. coli, purified using nickel affinity chromatography, and purity checked by SDS PAGE, as described previously (Starnes et al., 2006).
To generate constructs for regulated expression, N-terminally HA9-tagged TgALD1 was PCR amplified as above, cloned into the regulated expression vector p7TetOS4 (Meissner et al., 2002) using the restriction sites EcoRI and PacI to generate the plasmid pS4.HA9.ALD. To provide selection, the CAT gene driven by the SAG1 promoter, (Kim et al., 1993), was cloned into the SacII site in pS4.HA9.ALD.
To generate plasmids for creating a genomic knock out of aldolase, the Ble selectable marker conferring phleoymycin resistance (Messina et al., 1995) was flanked by 2kb of upstream and downstream genomic sequence of TgALD1 (www.Toxo.db.org, 46.m00002). To provide negative selection, a tandem YFP cassette driven by the α-tubulin promoter (kindly provided by Boris Striepen), was inserted at the SacI restriction site, to generate the plasmid pKOAld:YFP.
To complement the aldolase cKO strain, a plasmid expressing N-terminally, C-myc-tagged aldolase was generated by replacing the Ble gene in the vector SAG1/Ble/SAG1 (Messina et al., 1995) with wild type TgALD1. The DHFR selectable marker, conferring resistance to pyrimethamine (Donald and Roos, 1993), was inserted to generate the plasmid: pC-Myc.WTALD.DHFR.
The TATi-1 line was transfected by electroporation with the plasmid pS4.HA9.ALD and stable transformants selected using chloramphenicol (Kim et al., 1993). Clones were isolated via limiting dilution in 96-well plates seeded with HFF cells. A representative clone was transfected with linearized pKOAld:YFP plasmid and subjected to two independent rounds of phleomycin selection (Messina et al., 1995). Drug resistant, YFP-negative parasites, were recovered using a Dako MoFlo (Carpinteria, CA) gated to collect negative cells. The YFP-negative pool was single cell cloned and plaques were replica plated in 96-well plates with and without 1.5 μg/ml ATc. Potential cKO clones were isolated based on absence of growth in the presence of drug. The cKO was complemented with wild type or mutant aldolase genes by transfection, clones selected with 3 μM pyrimethamine, and expression confirmed by staining for the c-myc epitope.
Growth was monitored using a monolayer lysis assay, as described previously (Brossier et al., 2008). Parasites were grown in the absence or presence of ATc (1.5 ug/ml) for 48hrs, harvested and used to infect HFF monolayers in 96-well plates. Monolayers were incubated an additional 48 hr (± ATc), fixed with 70% EtOH, stained with 0.1% crystal violet and absorbance read at 570 nm using the EL800 multiwell plate reader (Bio-Tek Instruments, VT). Parasite lines were evaluated in three independent experiments with quadruplicate wells for each line and condition. Results are expressed as mean ± SD.
A second assessment of parasite growth was determined by counting the number of parasites per vacuole at 36 hrs post-infection, as described previously (Taylor et al., 2006). Parasites were harvested and used to infect HFF cells grown on glass coverslips. After culture at 37°C 5% CO2 for 36 hrs (± ATc), monolayers were fixed for IF, stained for the surface antigen SAG1 using mAb DG52, and mounted with Vectashield containing DAPI. Coverslips were examined by epifluorescence microscopy and the average number of parasites per vacuole (± SEM) was determined for each parasite line from 30 randomly selected vacuoles per coverslip (n=3) in three independent experiments.
Cellular concentrations of ATP were determined using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI) according to manufacturer’s protocol and luminescence was determined using a BioTek Synergy 2 Multi-Mode Microplate reader coordinated with Gen5 reader control and data analysis software (Winooski, VT). Cellular concentrations were estimated by comparison to a standard curve using purified ATP. Selective transgenes expressed in the cKO were grown in the presence of ATc for 48hrs, harvested when they egressed, and tested in triplicate. A representative of three similar experiments is shown as mean ± SD.
Parasites were grown for 48hr in the presence or absence of 1.5 μg/ml ATc. Freshly harvested parasites were allowed to glide on BSA-coated coverslips for 25 min at 37°C. Coverslips were fixed for IF and stained with antisera to the surface antigen SAG1 to detect trails, as described previously (Brossier et al., 2008). The total number of trails (circular and helical) per field for 5 fields per coverslip (n=3) were determined in triplicate; means ± SEM of three independent experiments.
Invasion of HFF cells cultured on coverslips was performed using a two-color IF staining protcol to distinguish extracellular from intracellular parasites, as previously described (Brossier et al., 2008). Intracellular and extracellular parasites were determined by microscopic examination of 5 fields per coverslip (n=3) and expressed as function of total host cells per field. Data are presented as means ± SEM of three independent experiments.
The structure of fructose-bisphosphate aldolase from Plasmodium falciparum in complex with the TRAP-tail (PDB entry: 2PC4) was used for building a homology model of T. gondii aldolase. A subunit of the muscle aldolase tetramer was modeled with the T. gondii aldolase sequence using Modeller software (Eswar et al., 2007) based on 10 trial models. The model with lowest energy and best Molprobity score (Lovell SC et al., 2003) was retained for modeling the TgALD1-MIC2 interaction.
The MIC2 tail sequence in T. gondii aldolase was then threaded according to the WASP conformation obtained in the structure of rabbit muscle aldolase in complex with a C-terminal peptide of Wiskott-Aldrich syndrome protein (WASp) (PDB entry: 2OT0) after superposition of the two aldolase structures. The resultant model was minimized with GROMACS (Berendsen et al., 1995) package using the all-atom force field GROMOS96 (parameter set 43a1). Minimization consisted of 600 steps of steepest descent followed by 1000 cycles of conjugate gradient minimization. A 2ns molecular dynamic simulation was then run using the position restrained option in GROMACS to optimize fit of the MIC2 peptide in the T. gondii aldolase-binding site. The final model geometry was checked by Molprobity and gave a score of 2.05.
Table S1 List of strains and plasmids used in this study.
Table S2 List of primers used for generating plasmid constructs.
We thank J.P. Vogel and S. Lourido, for careful reading of the manuscript and thoughtful discussions. We also thank J. Nawas and J. LaFrance-Vanasse for technical assistance and B. Eades, Siteman Cancer Center, for assistance with cell sorting. This work was funded by an NIH grant AI034036 (to LDS) with partial support from an institutional training grant AI07172-26 (to GLS) and a grant from the Natural Science and Engineering Research Council of Canada (to JS).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.