Toxoplasma gondii myosin A and its light chain: a fast, single‐headed, plus‐end‐directed motor

To determine the biochemical and biophysical properties of this motor, full‐length TgMyoA and TgMyoAΔtail (a mutant lacking 53 C‐terminal residues) were purified from recombinant parasites. The apparent molecular mass of the two proteins was determined by glycerol gradient centrifugation (Figure 1A). As expected, TgMyoAΔtail fractionated close to its molecular size of ∼100 kDa. In contrast, TgMyoA had an apparent size of 160 kDa, clearly smaller than a dimer, but nevertheless bigger than was predicted by the 53 residues difference between the two proteins. Despite the lack of clearly identifiable binding site in the neck/tail of TgMyoA, we searched for the presence of a light chain. The neck domains of many unconventional myosins bind calmodulin, therefore we raised specific antibodies against T.gondii calmodulin (TgCaM; previously described by Seeber et al., 1999). Association of TgCaM with TgMyoA or TgMyoAΔtail could be ruled out as they did not co‐purify (Figure 1C) and did not co‐fractionate on glycerol gradients (data not shown). Beside calmodulin, a calmodulin‐like protein is present in the P.falciparum genome database as well as in several expressed sequence tag (EST) databases of other apicomplexan parasites including T.gondii. We cloned and characterized the gene coding for this T.gondii calmodulin‐like protein and raised polyclonal antibodies against the first 75 amino acid residues of the protein fused to GST and expressed in Escherichia coli. The antisera detected a 31 kDa protein that, in contrast to TgCaM, co‐fractionated with TgMyoA on the glycerol gradient (Figure 1A).

We conclude that, as predicted from the absence of a coiled‐coil domain in the tail region, these experiments indicate that native TgMyoA is single headed (Figure 1A).

Further analysis by western blotting revealed that the 31 kDa protein co‐purified with TgMyoA heavy chain in several native preparations but was absent when TgMyoA was purified under denaturing conditions or when native TgMyoAΔtail was purified (Figure 1B). Based on the characteristics defined in this study, we named this protein T.gondii myosin light chain 1 (TgMLC1). Consistent with the results obtained above, we also showed that the last 53 amino acid residues in the tail of TgMyoA were necessary and sufficient to bind TgMLC1 (Figure 1D). A stretch of His₈ tag was introduced at the N‐terminus of green fluorescent protein (GFP) in the pTGFP and pTGFPAtail constructs (Hettmann et al., 2000). HisGFP and HisGFPAtail proteins were expressed stably in T.gondii and purified from parasite lysates on Ni‐NTA resin. TgMLC1 co‐purified specifically with HisGFPAtail but not with HisGFP as detected by western blotting of the material eluted from the batch purification. Finally, additional data from another group allowed us to establish that the TgMyoA complex is at least heterotrimeric. It is composed of one TgMyoA heavy chain (93 kDa), one light chain (TgMLC1; 31 kDa) and a recently identified, 50 kDa myristoylated protein (TgMADP) which anchors TgMyoA at the plasma membrane (C.Beckers, International Meeting on Toxoplasmosis, Freising, 2001). Using anti‐TgMADP antibodies, we observed that the docking protein co‐purified with TgMyoA under exactly the same conditions as TgMLC1 (data not shown).

As anticipated for a TgMyoA‐associated protein, TgMLC1 localized at the parasite plasma membrane under static conditions (intracellular parasites) (Figure 2C). In addition, during host cell penetration, TgMLC1 was slightly concentrated at the moving junction and in the part of the parasite penetrating the host cell (Figure 2A). Less than 10 min later, TgMLC1 was still distributed at the parasite periphery but also revealed a circle at the posterior end, delineating an open ‘appendage’ at the parasite posterior extremity (Figure 2B). This structure, visible only shortly after invasion, is likely to correspond to the transient posterior invagination where multi‐lamellar vesicles are released to form the intravacuolar network (Sibley et al 1995).

TgMLC1 comprises 213 amino acids and appears to be composed of four degenerate EF hand domains (Figure 3A). Structural homology searches indicated that only the third one (residues 149–179) fits the consensus sequence well; the first one is relatively conserved but the other two carry multiple deletions. Very closely related proteins of similar sizes are present in the genome of several Apicomplexa and most probably correspond to the light chain of MyoA homologues. The N‐terminal 60 residues are unique, not present in calmodulin or other myosin light chains and highly conserved in the Apicomplexa. TgMyoAΔtail was shown to be cytosolic, and the determinant of plasma membrane localization of TgMyoA mapped within its last 22 residues and was strictly dependent on an RR motif (Hettmann et al., 2000) (see Figure 3B, bar and double arrowhead). Interestingly, deletion of the tail prevents co‐purification of TgMLC1 and TgMADP. Close examination of the last 22 residues reveals the presence of a very degenerate IQ motif (QXXXR, Figure 3B). The precise topology of the TgMyoA–TgMLC1–TgMADP complex remains to be determined.

The heterotrimeric TgMyoA complex could only be purified in very small amounts, which were insufficient for kinetic analysis using stopped‐flow methods but were ample for use in the flash photolysis system developed for limited quantities of myosins (Weiss et al., 2000). In contrast, transient kinetic data for monomeric TgMyoAΔtail were acquired by both stopped‐flow and flash photolysis methods. The maximum velocity at which myosins move over actin is normally limited by the net rate of cross‐bridge detachment at the end of the power stroke (Siemankowski and White, 1984; Weiss et al., 2001). This detachment rate can be limited by the rates of both ADP release and ATP binding, measurements of which can therefore be diagnostic of maximum gliding velocity. The rate constant of the ATP‐induced dissociation of both full‐length TgMyoA and TgMyoAΔtail from actin was determined by the flash photolysis method as shown in Figure 4A. A known concentration of ATP was released from 0.5 mM caged ATP in a sample 0.5 μM actin and 1 μM TgMyoA and the dissociation of the complex monitored by the decrease of the light scattering signal. A single exponential fit to the light scattering trace gave the k_obs of the reaction (Figure 4A). The experiment was repeated at various ATP concentrations and a plot of k_obs versus [ATP] was linear over the range of concentrations used. The slope of the linear regression defines the apparent second‐order rate constant K₁k₊₂ = 0.28 and 0.46/μM/s for full‐length TgMyoA and TgMyoAΔtail, respectively (Figure 4B).

The affinity of ADP for acto·TgMyoA was measured from the ADP inhibition of the ATP dissociation experiment. A constant 50 μM ATP was photoreleased in the presence of increasing amounts of ADP (0 μM–2 mM). The value of k_obs decreased as the ADP concentration increased, and the resulting data were described by a hyperbola (Scheme I, see Table I footnotes), suggesting that ADP binding and dissociation is in rapid equilibrium with acto·TgMyoA preceding ATP‐induced dissociation of the complex (Scheme II). The analysis defined the ADP affinity, K_AD, as 840 and 720 μM for TgMyoA and TgMyoAΔtail, respectively (Figure 4C). The flash photolysis experiments showed that the values of both K₁k₊₂ and K_AD were similar for TgMyoA and TgMyoAΔtail and demonstrate that the truncation of the tail by 53 residues does not alter the kinetic properties of the motor domain of TgMyoA.

A more complete kinetic characterization by stopped flow was then undertaken with the more readily available TgMyoAΔtail. The ATP‐induced dissociation and the ADP inhibition of the dissociation were measured using 400 nM acto·TgMyoA complex. Unlike any other myosin so far examined, TgMyoA (and TgMyoAΔtail) did not quench the fluorescence of a pyrene label covalently attached to actin. Therefore, dissociation of the complex was measured by light scattering. The results are given in Table I and although the stopped‐flow measurements resulted in a higher value of K₁k₊₂ and a lower value of K_AD than flash photolysis, the values are of the same order for both methods. The data confirm that TgMyoA is dissociated quickly from actin by ATP and that the TgMyoA·actin complex has a low affinity for ADP, properties that TgMyoA shares with fast muscle myosins.

The rate of ATP binding to TgMyoAΔtail was determined by measuring the intrinsic tryptophan fluorescence change which occurs on ATP binding. A 3% decrease in fluorescence was observed compared with the usual fluorescence increase obtained for other myosins, probably due to the absence of the tryptophan normally thought to be responsible (Batra and Manstein, 1999). The slope of the linear fit to a plot of k_obs versus [ATP] gives K₁k₊₂ = 0.72/μM/s (Table I). The measurement was repeated using the fluorescent analogues mantATP and mantADP and gave similar second‐order binding rate constants. For ADP, the dissociation rate constant was estimated by displacing bound mantADP with a large excess of ATP, resulting in a value of k₊₆ = 0.95/s. The affinity of TgMyoA for mantADP was estimated as 2 μM from the relationship K_D = K₇k₊₆/k₋₆. The affinity of actin for TgMyoA was measured in a competition experiment with rabbit myosin S1. Under the conditions used, TgMyoA had a 2‐ to 3‐fold lower affinity for actin than S1. Under physiological conditions, the dissociation equilibrium constant (K_A) of S1 for actin is 30 nM, suggesting a value of 60–90 nM for TgMyoA·actin. To determine ATP turnover activity of TgMyoAΔtail, 50 nM TgMyoAΔtail with 50 or 200 μM ATP were quenched at various time points. The analysis of [ATP]/[ADP] ratios with an HPLC system gave an ATP turnover rate of 0.07/s in the absence and 1.5/s in the presence of 20 μM actin. A >20‐fold activation in the presence of actin is typical of fast myosins.

To complement this analysis, we directly determined some of the functional properties of this motor. The sliding velocity of actin filaments was measured in a standard in vitro motility assay (Kron and Spudich, 1986). We measured an average velocity of 5.2 μm/s for TgMyoA (Figure 5A). TgMyoAΔtail showed no motility under the same conditions, although the protein displayed ATP‐dependent binding to actin. This lack of motility is probably due to a functional defect caused by deletion of the last 53 C‐terminal residues, which hampers association with TgMLC1. The sequences of class XIV myosins diverge from that of other myosins, starting at a position corresponding to the SH1–SH2 region, just downstream of the 50k–20k domain junction (Figure 3C). Thus, TgMyoA displays little sequence similarity to other types of myosin in regions that normally form the converter and lever arm. The recent finding that myosin VI, exhibiting an unusual converter domain, is a minus‐end‐directed motor (Wells et al., 1999) prompted us to determine the directionality of TgMyoA. For this purpose, the in vitro motility assay was performed with actin filaments labelled with Oregon green at their plus (barbed) ends. To block actin polymerization at the plus end of the labelled tips, they were generated in the presence of gelsolin (Figure 5B). TgMyoA moved these polarity‐marked filaments so that the labelled end was at the back of the moving filament, demonstrating that TgMyoA, like myosin II, moves towards the plus end of F‐actin.

Furthermore, we employed a combined microneedle– laser trap transducer (Ruff et al., 2001) and directly measured the step displacements generated by single TgMyoA molecules. In agreement with our kinetic analysis, TgMyoA interacted transiently with actin filaments. A histogram analysis of 598 such interactions revealed an average step displacement of ∼5.3 nm (Figure 6), comparable with the step size reported for myosin II (Veigel et al., 1999). In addition, these measurements confirm at the single molecule level that TgMyoA is a non‐processive motor. Taken together, the sliding velocity and step size measurements of full‐length TgMyoA are in perfect agreement with, and thus validate, the kinetic values.

Transgenic T.gondii parasites expressing the constructs pS·TgMyoA‐tail and pS·TgMyoA were used as a source of recombinant proteins (Hettmann et al., 2000). Toxoplasma gondii tachyzoites were grown in Vero cells (African green monkey kidney cells) in tissue culture flasks, maintained in Dulbecco‘s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS), 2 mM glutamine and 25 μg/ml gentamicin. Rabbit polyclonal antibodies were raised against a bacterially expressed GST fusion of 75 amino acids corresponding to the N‐terminus of TgMLC1.

TgMyoAΔtail was purified as described previously (Hettmann et al., 2000). Full‐length His‐tagged TgMyoA was purified similarly with the following modifications. The parasite pellet (10¹⁰ parasites) was lysed by sonification in a buffer with detergent [2% Triton X‐100, 4 mM ATP, 2 mM MgCl₂, 0.5 mM dithiothreitol (DTT), pH 7.6 and 1× Complete protease inhibitor cocktail (Roche)]. The suspension was clarified by ultracentrifugation at 65 000 r.p.m. and the supernatant was applied to an Ni‐NTA resin column and eluted according to the manufacturer (The QIAexpressionist) (Janknecht et al., 1991). For the purification under denaturating conditions, the cells were lysed in the presence of 6 M guanidinium chloride and the clarified lysate was dialysed in the presence of 8 M urea before application on the column. Protein concentration was determined by a Bradford assay (Bio‐Rad). SDS–PAGE was performed by standard methods. Average yields of 30 μg of TgMyoA and 250 μg of TgMyoAΔtail were obtained per gram of parasites. Active forms of TgMyoAΔtail and TgMyoA were purified further by co‐precipitation with F‐actin and release in the presence of ATP (Anson, 1992). Myosin subfragment 1 (S1) was prepared by chymotryptic digestion of rabbit myosin as described by Weeds and Taylor (1975). The molar concentration was determined from the extinction coefficient at 280 nm of ϵ^1% = 7.9/cm and a molecular weight of 115 kDa. Rabbit skeletal actin was purified by the method of Spudich and Watt (1971). Its molar concentration was determined by absorbance at 280 nm (ϵ^1% = 1.104/cm) and a molecular weight of 42 kDa (West et al., 1967). The preparation of pyrene‐labelled actin was described previously (Criddle et al., 1985). To prepare phalloidin‐stabilized F‐actin (RhPh/actin), a stock solution of 10 μM F‐actin and 10 μM phalloidin was incubated in experimental buffer overnight (Kurzawa and Geeves, 1996). Unless noted otherwise, all experiments with myosin S1 were performed in experimental buffer containing 100 mM KCl, 20 mM MOPS and 5 mM MgCl₂ at pH 7.0 and 20°C. In all flash photolysis experiments, fresh DTT was added to a concentration of 10 mM.

The centrifugation of 12 ml of glycerol density gradients was performed as described previously (Mitchell et al., 1997). The 10 and 30% glycerol solutions contained 50 mM NaH₂PO₄ and 300 mM NaCl pH 7.6. TgMyoA purified through the Ni‐NTA column was loaded onto gradients after ultracentrifugation and overnight dialysis at 4°C against the 10% glycerol solution. A gradient was run in parallel with size standards: bovine serum albumin (BSA; 70 kDa), bovine catalase (240 kDa; Sigma) and beef heart lactate dehydrogenase (140 kDa; Roche). Fractions of 500 μl were collected and analysed by SDS–PAGE and western blot analysis.

Stopped‐flow experiments were performed with a Hi‐Tech Scientific SF‐61 DX2 stopped‐flow spectrophotometer (Hi‐Tech Scientific, Salisbury, UK) using procedures described previously (Ritchie et al., 1993; Cremo and Geeves, 1998; Furch et al., 1999). The affinity of TgMyoA for pyrene–actin was measured by competition with S1 (Coluccio and Geeves, 1999). The amplitude of the pyrene fluorescence change observed on dissociation of pyrA·S1 by addition of ATP gave a direct measure of the fraction of actin occupied by S1. Added TgMyoA competes with S1 for binding sites on actin, and analysis of the loss of amplitude defines the relative affinity of S1 and TgMyoA for actin.

The optical bench for flash photolysis is designed for simultaneous time‐resolved detection of transmission changes and either fluorescence or light scattering changes (Weiss et al., 2000). This system handles very small amounts of protein (<1 μg) in volumes of 10–20 μl, making experiments with the less efficiently purified TgMyoA possible.

The ATP turnover reaction for TgMyoA and acto·TgMyoA was measured by assaying the ATP/ADP ratio as a function of incubation time using HPLC analysis (Äkta Basic 10, Amersham Pharmacia Biotech with a 15 cm ODS‐Hypersil column, run in 20 mM MOPS, 30 mM KCl and 5 mM MgCl₂ at pH 7.0). Samples in standard buffer (except 30 mM KCl) were quenched at various times by the addition of 2 μl of 6 M HClO₄ to 100 μl of sample and applied to the column after neutralization.

Standard in vitro motility assays were performed as described previously (Furch et al., 1999). Assays were performed at 30°C in 25 mM imidazole·HCl pH 7.4, 25 mM KCl, 4 mM MgCl₂, 1 mM EGTA, 2 mM ATP and 10 mM DTT. All solutions were kept on ice prior to use and the microscope stage was stabilized at 30°C. A Zeiss Axiovert 135 inverted microscope with a 1.3 NA 40× oil immersion lens was used for epifluorescence with 540–550 nm excitation (100 W HBO) and 580 nm emission band‐pass (Omega Optical filter set XF 37). Labelled actin filaments were imaged with a 1.3 NA 40× lens on an intensified video camera (Photonic Science Darkstar) and recorded, via an image processor (Hamamatsu Photonics, Argus 20) and time–date generator, by an S‐VHS VCR (JVC BRS800‐E). After focusing and observation of the actin filaments in rigor, motion was initiated by infusing 100 μl of buffer containing 2 mM ATP. For manual analyses, tapes were replayed from the VCR to the Argus 20 and individual filaments tracked frame‐by‐frame using a mouse. Velocity was calculated from the positional change and time code recorded on the tape. Experiments with TgMyoA and TgMyoAΔtail were performed with two freshly made protein preparations each. Chicken fast skeletal muscle HMM (8.5 ± 1.5 μm/s) and Dictyostelium discoideum myosin II (2.4 ± 0.3 μm/s) were used as positive controls and >100 filaments were analysed in each case. For protein preparation, see Ruppel et al. (1994) and Kron and Spudich (1986).

Oregon green‐labelled G‐actin was mixed with gelsolin at a ratio of 33:1 and incubated on ice for 20 min. Adding 100 mM KCl at room temperature started actin polymerization. The resulting F‐actin seeds were stabilized with phalloidin. Elongation of these seeds is restricted to the pointed end, because their barbed ends are blocked by gelsolin. For elongation, a 10‐fold excess of unlabelled G‐actin was added in Tris buffer containing 100 mM KCl, and the reaction was allowed to proceed for 15 min at room temperature. Finally, the polarity‐labelled F‐actin was diluted and the elongated part of the filaments was stabilized with rhodamine–phalloidin. An Olympus IX70 inverted microscope with a Zeiss 1.4 NA 100× oil immersion objective was used to image the filaments by epifluorescence. A Polychrom II monochromator (Till Photonics, Munich, Germany) in combination with a dual‐band filter set (Chroma, Brattlesborough, USA) was used for excitation at 492 and 554 nm. Filaments were imaged using a Till Imago CCD camera, and TillVision was used for wavelength selection and image acquisition.

Single actomyosin interactions were recorded with a combined microneedle–laser trap system as described previously (Ruff et al., 2001). This setup is similar to the double‐laser trap system utilizing a three bead geometry (Finer et al., 1994), except that we replaced one of the trapped beads with a fine glass microneedle (stiffness 0.01–0.03 pN/nm, time constant ∼1 ms). A magnified brightfield image of the needle projected onto a quadrant photo‐diode (Hamamatsu, S1557) allows position detection with subnanometer resolution and a bandwidth of >8 kHz. Signals were low‐pass filtered (1 kHz), sampled at 5 kHz with a microcomputer and stored on disk for off‐line analysis. Single molecule in vitro assays were carried out in standard motility buffer (25 mM imidazole pH 7.4, 25 mM KCl, 1 mM EGTA, 4 mM MgCl₂, 1 μM ATP, 10 mM DTT, 100 μg/ml glucose oxidase, 18 μg/ml catalase and 3 mg/ml glucose). To record from a single TgMyoA motor, we constructed a recording chamber from two coverglasses that were separated by a 1 mm Teflon spacer. The lower coverglass was coated with a 0.1% nitrocellulose solution (Fullam Latham, NY) in amyl acetate containing 2.5 μm diameter silica beads (Bangs Laboratories, Fishers, IN) which served as pedestals. TgMgoA (1.5 μg/ml) was directly adsorbed onto nitrocellulose‐coated surface by incubating for 15 min. We inserted a microneedle from the open side into the recording chamber and, using NEM‐myosin as glue, attached a single actin filament at one end to the tip of the microneedle and at the other end to a 1 μm latex bead captured in the laser trap. The entire arrangement of microneedle, actin filament and latex bead was pulled taut, lowered carefully onto the myosin‐coated bead and the ensuing actomyosin interactions were recorded. To determine the step size, we followed the strategy of Molloy et al. (1995). We identified single actomyosin interactions by a reduction of the thermal fluctuation of the microneedle sensor during binding to myosin. From the average position of a large number of such events, we calculated a histogram and determined the step size from the shift of the mean position of the distribution.