Identification and Cloning of Two Members of the Class XIV Myosins
To identify myosins in T. gondii,
we designed a PCR screening strategy taking advantage of the highly conserved motifs present in the head domain (Schwarz et al., 1999
). The approach identified TgM-A, a myosin already described by Heintzelman and Schwartzman (1997)
, and TgM-D, an additional member of the apicomplexan-specific class XIV of myosins (Figure B). The entire TgM-D gene was sequenced. The protein predicted from the open reading frame has 823 amino acid residues (Figure A) with a mass of 91 kDa (accession number AF105117
The sequences of the short and conserved, positively charged tail domains of TgM-A and TgM-D showed a significant homology with an expressed sequence tag derived from the closely related organism P. falciparum. This allowed us to identify and subsequently clone a complete myosin corresponding to a partial P. falciparum sequence present in the database (PfMyo-1; see MATERIALS AND METHODS). We named this myosin PfM-A (Figure A) because its predicted sequence of 90 kDa has highest homology to TgM-A (Figure B). The degree of conservation at the amino acid level between the myosins is high and extends throughout the entire coding region. There is 55.7% identity between TgM-A and TgM-D, 63.1% between TgM-A and PfM-A, and 52.0% between TgM-D and PfM-A. Pairwise comparisons between any of these myosins with the other two, alternatively spliced, class XIV myosins from T. gondii, TgM-B and -C, give lower identity scores of ~47–49%.
Like the other apicomplexan myosins TgM-D and PfM-A have a very short tail, do not contain the strictly conserved glycine residue at the proposed fulcrum point of the lever arm, and appear to lack conserved “IQ” motifs that bind calmodulin and calmodulin-related proteins. Among their particularities, and contrary to the P. falciparum
molecule, all the T. gondii
myosins do not follow the TEDS rule (Heintzelman and Schwartzman, 1997
; Figure A), which describes the presence of an acidic or phosphorylatable residue at a precise position close to the actin-binding region (originally mapped by Brzeska et al., 1989
; for review, see Bement and Mooseker, 1995
). In lower eukaryotes, phosphorylation of this conserved serine or threonine was shown to be crucial for the stimulation of the ATPase activity of class I myosins (Bement and Mooseker, 1995
; Carragher et al., 1998
; Novak and Titus, 1998
). It is unclear at the moment whether and how these class XIV motors are activated and how conformational changes in the molecules occur.
The different recombinant constructs of TgM-A and TgM-D used in this study are presented schematically in Figure C.
Antibodies to the Tail of TgM-A Are Isoform Specific
To investigate the expression and distribution of TgM-A, we raised polyclonal sera against two peptides covering its entire tail. The specificity of the antibodies was assessed by Western blotting of parasites expressing GFP·TgM-Atail and GFP·TgM-Dtail (Figure A). The anti-myc antibodies detected both GFP chimeras, but despite high overall sequence homologies, the anti-tail antibodies recognized only the TgM-A tail (Figure A). In wild-type parasites, the affinity-purified antibody recognized a single band migrating above 90 kDa, the predicted size of TgM-A (Figure B). Recombinant mycTgM-A was detected by the anti-tail antibodies and was slightly bigger than endogenous TgM-A. Confirming the antibody specificity, only the endogenous protein was detectable in parasites expressing mycTgM-AΔtail (Figure B).
Figure 2 Immunoblot analysis of wild-type and recombinant parasites. (A and B) The antibody against the tail of TgM-A does not recognize the tail of TgM-D. Extracts of parasites expressing GFP·TgM-Atail or GFP·TgM-Dtail were blotted with anti-myc (more ...)
The Expression of Endogenous TgM-A Is Down-regulated in the Transgenic Parasites Expressing Full mycTgM-A
Surprisingly, in recombinant parasites expressing full-length mycTgM-A, the endogenous TgM-A was barely detectable (Figure B). To obtain a quantitative impression of the phenomenon, equal numbers of wild-type and different recombinant parasites were analyzed (Figure B). MycTgM-A appeared to be expressed at approximately two- to threefold the level of endogenous protein found in wild-type parasites and parasites expressing mycTgM-AΔtail. We conclude that expression of full-length mycTgM-A, but not of the head domain alone, led to down-regulation of the endogenous TgM-A protein and expression. In parasites expressing GFP·TgM-Atail, the anti TgM-A tail antibodies simultaneously detected the 40-kDa GFP chimera and the endogenous TgM-A. Expression of the endogenous protein appeared slightly reduced compared with wild-type cells, indicating that the tail alone mimicked the down-regulation effect observed with the full-length myosin. Because the signal for the GFP chimera was approximately half as intense as for mycTgM-A (Figure B, first lane), we suggest that the effect is partial probably because of a lower expression level.
The Presence of Its Tail Influences the Expression of TgM-D
TgM-D was well expressed transiently, but our initial efforts to obtain stable transformants were unsuccessful, potentially because mycTgM-D overexpression was not well tolerated by the parasites. To circumvent this problem, we introduced the selectable marker directly in the expression vector and obtained a few stable cell lines expressing mycTgM-D. Expression of transgenic mycTgM-D (Figure B) was significantly reduced compared with mycTgM-A (Figure A), although the promoter used to drive its expression was previously shown to be stronger (Soldati and Boothroyd, 1995
). The construct pT·TgM-DΔtail could readily be introduced into the parasites by simple cotransfection with the selection marker plasmid. Immunoblot analysis of total cell lysates with an antibody against the c-myc epitope showed that parasites transformed with pS·TgM-A or pT·TgM-D expressed a single protein of the expected sizes of 93 and 91 kDa (Figure ). Deletion of their tails resulted in slightly higher mobility. A semiquantitative analysis was performed by comparing the signal intensities resulting from loading lanes with a precise number of parasites. This revealed that mycTgM-A and mycTgM-AΔtail were expressed at comparable levels (Figure A), whereas mycTgM-DΔtail was expressed at an ~5- to 10-fold higher level than mycTgM-D (Figure B). These results provided evidence that the presence of the tail ofTgM-D limits the expression level of the recombinant protein.
Figure 3 The presence of its tail limits the expression level of recombinant TgM-D. Western blot analysis used the anti-myc antibody on parasites expressing recombinant mycTgM-A and mycTgM-AΔtail (A) or mycTgM-D and mycTgM-DΔtail (B). The respective (more ...)
TgM-A Localization Is Confined to the Parasite Periphery
TgM-A and TgM-D share a high degree of homology, and it is not yet known whether T. gondii
expresses additional similar myosins (in addition to the related TgM-B and -C, which harbor longer and different tail domains). To determine unambiguously, without risk of cross-reaction, the subcellular localization of TgM-A, TgM-D and mutants thereof, two approaches were used. First we generated isoform-specific antibodies. Second, we used an epitope-tagging strategy, taking advantage of the easy accessibility of T. gondii
to genetic manipulation. The anti-myc tag antibody revealed that, aside from a relatively diffuse cytoplasmic distribution, a proportion of mycTgM-A localized precisely beneath the plasma membrane (Figure A, a), as visualized by the almost perfect colocalization with SAG1, the major surface antigen of T. gondii
tachyzoites (Figure A, b). Because SAG1 is anchored at the plasma membrane via a glycosylphosphatidylinositol, the staining suggests that TgM-A associates closely with this membrane, although the resolution of light microscopy cannot exclude that the myosin interacts with the inner membrane complex. This latter structure is composed of flattened membrane cisternae found in close apposition with the plasma membrane of all Apicomplexa. It covers the elaborate basket of microtubules and contributes to the maintenance of cell shape and polarity (Morrissette et al., 1997
Figure 4 Immunofluorescence localization of TgM-A and TgM-D and their respective tail-less constructs. Classical (A and D) and confocal (B and C) immunofluorescence microscopic analysis of wild-type parasites (C, b) or parasites expressing mycTgM-A (A, B, a–c, (more ...)
Surprisingly, despite strong and specific signals obtained on Western blots (Figure ), imunofluorescence using the anti-TgM-A tail antibodies failed to detect endogenous TgM-A in wild-type parasites (our unpublished results) and parasites expressing mycTgM-AΔtail (Figure B, e) and also failed to corroborate the peripheral localization of mycTgM-A (Figure B, compare a and b). Nevertheless, in the latter case a signal was clearly perceived and was similar in intensity (but not in staining pattern) to the cytoplasmic signal obtained for mycTgM-AΔtail (Figure B, compare d and b). It is reasonable to assume that the tail of TgM-A, being very rich in arginine and lysine residues, aldehyde cross-linking to plasma membrane proteins, may lead to complete epitope masking. Therefore, we used an alternative fixation method involving rapid freezing fixation and permeabilization in ultra-cold methanol, an ideal method to preserve both structure and antigenicity (Neuhaus et al., 1998
). Figure C illustrates that both the endogenous TgM-A (Figure C, b) and the myc-tagged protein (Figure C, a) were now detected by the anti-tail antibody. As expected, the signal for TgM-A was slightly lower in wild-type parasites than in recombinants expressing the myc-tagged protein. The cytoplasmic staining for mycTgM-A (Figure B, b) was stronger than for endogenous TgM-A (Figure B, e). Finally, irrespective of the expression level, both proteins localized similarly all around the parasite periphery, confirming the epitope-tagging data.
The Short Tail Domains Are Necessary for Plasma Membrane Localization of the T. gondii Myosins
Myosin molecules are modular motors made up of three domains. The N-terminal domain is the actin binding motor unit per se. The middle neck domain bears the light chains and acts as a lever arm. The tail domain is exceptionally divergent and reflects the diversity in myosin functions. The tail is thought to target a given myosin to its cargo or site of action and thereby to determine the specific task of the motor. To assess the role of the short tail domains in the subcellular distribution of the two proteins, we constructed vectors expressing myosins lacking their tail. Recombinant parasites expressing mycTgM-AΔtail and mycTgM-DΔtail were analyzed by confocal microscopy (Figure , B and D). As already observed above, absence of tail in mycTgM-AΔtail caused the disappearance of the membrane-associated pool (Figure B, d).
In a way similar to mycTgM-A, mycTgM-D appeared to distribute predominantly at the parasite periphery (Figure D, a), even though not as sharply as mycTgM-A. The subtle difference in distribution may indicate a different mode or mechanism of localization. As in the case of mycTgM-A, the absence of tail caused an apparent redistribution of mycTgM-D to the cytoplasm (Figure D, b), and, as mentioned above (Figure B), it also led to a significant increase in the level of the recombinant product compared with the full-length protein.
The Tail of TgM-A Is Necessary for Distribution in the Particulate Fraction
To corroborate the localization data with biochemical fractionation, cells were first lysed in PBS, separated into soluble and sedimentable fractions by high-speed centrifugation, and analyzed by Western blotting. A comprehensive analysis was undertaken on cells expressing GFP·TgM-Atail, allowing for simultaneous investigation of endogenous TgM-A and of the GFP tail chimera. In agreement with the imunofluorescence, TgM-A partitioned with the particulate fraction (Figure A). The nature of this interaction was further investigated. TgM-A behaved as a strongly associated peripheral membrane protein, because it was resistant to solubilization by high salt. Full solubilization was achieved only by carbonate treatment (pH 11.5) and extraction by detergent (1% Triton X-100) or 3 M urea (our unpublished results).
Even though mycTgM-AΔtail appeared cytoplasmic (Figure B, d), it is formally possible that TgM-A localized at the periphery and, despite inclusion of ATP during fractionation, interacted with the particulate fraction through binding of its head domain to the actin cytoskeleton. Therefore, the role of the tail domain was investigated for recombinant TgM-A (Figure C). When parasites were lysed in PBS, about one-third to one-half of mycTgM-A sedimented with the particulate, membrane fraction whereas mycTgM-AΔtail partitioned essentially with the soluble cytosolic fraction, in perfect agreement with the imunofluorescence data. This indicated that the sedimentation of TgM-A was not due to interactions through the motor domain but was likely mediated by the tail domain (also see Figure and accompanying text). Similar results were obtained for mycTgM-D (Figure C). Moreover, whereas endogenous TgM-A is quantitatively recovered with the particulate fraction (Figure , A and B), approximately half of mycTgM-A, which is overexpressed two- to threefold, is found in the soluble fraction (Figure C), possibly because a plasma membrane receptor is saturated.
TgM-A Binds to F-actin in an ATP-dependent Manner
The peripheral targeting of TgM-A appears not to depend on the interaction of the head domain with the actin cytoskeleton. This could potentially be caused by the fact that the motor domains of class XIV myosins are predicted to exhibit important structural divergences. Therefore, it was crucial to assess experimentally whether these proteins truly act as ATP-driven, actin-dependent motors. In addition, according to the capping model of invasion, the myosin involved would have to be able to grab onto F-actin and exert ATP-dependent traction. The complete solubility of recombinant mycTgM-AΔtail in tachyzoites allowed us to purify biochemical quantities of the protein under native conditions. The degree of purity of the recombinant mycTgM-AΔtail was examined by SDS-PAGE and Coomassie blue staining (Figure A). The identity of two low-molecular-mass bands (~30 kDa) has not been fully determined yet (our unpublished results), but they were distinct from degradation products of TgM-A and could potentially represent copurifying light chain(s). To gain first insights into the biochemical characteristics of this mycTgM-AΔtail, a cosedimentation assay with F-actin was performed (Figure B). The results showed that mycTgM-AΔtail coprecipitated with actin only in the absence of ATP and that the sedimented myosin was subsequently, even though only incompletely, released by the addition of 10 mM ATP, demonstrating that this purified myosin was able to reversibly bind actin in an ATP-dependent manner. Similar results have been obtained with full-length mycTgM-A (our unpublished results). Together, our results also demonstrated the functionality of both recombinant TgM-A constructs.
Figure 6 MycTgM-AΔtail sediments with F-actin in an ATP-dependent manner. (A) SDS-PAGE and Coomassie blue staining of mycTgM-AΔtail purified from an extract of recombinant parasites. (B) SDS-PAGE analysis and Coomassie blue staining of the supernatant (more ...)
The Short TgM-A and TgM-D Tail Domains Are Sufficient to Target a Reporter Protein to the Cell Periphery
The short basic tails of TgM-A, TgM-D and PfM-A are very similar in length and in amino acid composition. A comparison of the myosin tails is depicted in Figure A. The conserved spots of basic residues are highlighted. Analysis of the predicted secondary structure of these domains also revealed a conserved overall organization in three helices interrupted by short loops. To determine whether the short tail domains are sufficient to confer specific cellular distribution to the myosins, the tails of TgM-A (last 82 amino acids) TgM-D (last 57 amino acids), and PfM-A (last 85 amino acids) were fused to GFP. The plasmids pT·GFP and pT·GFP·TgM-Atail were stably integrated into tachyzoites by cotransfection with HXGPRT expression vector. We had to introduce the selectable marker gene on pT·GFP·TgM-Dtail and pT·GFP·PfM-Atail to obtain stable transformants. The transgenic parasites expressing GFP and GFP·TgM-Atail were analyzed by direct GFP fluorescence, whereas GFP·TgM-Dtail and GFP·PfM-Atail required indirect immunofluorescence analysis using anti-myc antibodies to definitely visualize the recombinant protein (Figure B). The nonfusion GFP was abundantly expressed, homogeneously distributed in the cytosol with a significant accumulation in the nucleus (Figure B, a). In contrast, GFP·TgM-Atail was found almost exclusively closely associated with the plasma membrane, with barely detectable cytosolic staining (Figure B, b). In a way reminiscent of the respective localization of the full-length proteins, GFP·TgM-Dtail showed a slightly more diffuse and speckled peripheral signal than the GFP·TgM-Atail chimera (Figure B, c). The difficulty reported above to express high amounts of mycTgM-D appeared indeed to be dependent of the tail domain, because parasites expressing GFP·TgM-Dtail were also difficult to obtain, and the fusion was expressed at a very reduced level compared with both GFP and GFP·TgM-Atail. This was visible on the immunofluorescence stainings and was confirmed by Western blotting (our unpublished results). Similarly, the GFP·PfM-Atail fusion was expressed at a low level and localized essentially to the parasite cytosol (Figure B, d).
The GFP·TgM-Atail and GFP·TgM-Dtail Chimeras Partition with the Membrane Fraction
The transgenic parasites expressing GFP and chimeras were lysed in PBS, separated into soluble and particulate fractions, and subsequently analyzed by Western blot. As illustrated in Figure C, the migration on SDS-PAGE of the GFP chimera was in agreement with their predicted sizes. GFP·TgM-Atail and GFP·TgM-Dtail completely partitioned with the particulate fraction, whereas GFP and GFP·PfM-Atail were entirely recovered in the supernatant. The quantitative association of GFP with the particulate fraction was comparable with the behavior of endogenous TgM-A. In contrast, GFP·TgM-Dtail sedimented despite the fact that the main part of mycTgM-D was soluble. This could potentially indicate that both GFP chimeras were expressed at levels that did not saturate their respective peripheral binding sites. In addition, it is important to note that both chimeras were fully solubilized by carbonate treatment, demonstrating that they were not simply aggregated in the cell (Figure C). In fact, the solubilization characteristics of GFP·TgM-Atail were extremely similar to the ones of endogenous TgM-A (see Figure A).
The Localization of TgM-A Is Not Due to Interactions with the Actin Cytoskeleton
Altogether, the data indicate a strong interaction of the tail of TgM-A with components of the parasite cortex, likely with proteins associated peripherally with the plasma membrane, even though an interaction with the closely apposed inner membrane complex cannot be excluded. To definitively eliminate the potential targeting role of the cortical actin cytoskeleton, F-actin was severely compromised by cytochalasin D incubation. It has been previously reported that phalloidin fails to stain the actin filaments in T. gondii
(Dobrowolski et al., 1997b
). Therefore, as indication of the treatment efficacy, the peripheral localizations of mycTgM-A (Figure A, a and b) and GFP·TgM-Atail (Figure A, c and d) are shown simultaneously with a phalloidin staining of the host actin cytoskeleton. No noticeable change of localization of both mycTgM-A and GFP·TgM-Atail was observed despite severe disturbance of the actin cytoskeleton (Figure A, compare b with a and d with c).
The Tail of TgM-A Does Not Localize to the Plasma Membrane of HeLa Cells
Stretches of basic residues in the tail of class I myosins have been shown to contain high-affinity phospholipid binding sites (Doberstein and Pollard, 1992
). If TgM-A were to interact with the plasma membrane solely through unspecific interaction with lipids, the peripheral localization should be observed in unrelated cellular systems. When expressed in HeLa cells (Figure B), GFP·TgM-Atail is homogeneously distributed in the cytoplasm, confirming that the association with the particulate fraction in T. gondii
was not due to improper folding. More importantly, the chimera showed no sign of membrane association. Indeed, its cytoplasmic localization (Figure B, d) was indistinguishable from the one of GFP (Figure B, b).
Identification of the Molecular Determinants of Peripheral Membrane Localization
To define precisely, at the amino acid level, the determinants of membrane localization, deletion analysis and site-specific mutagenesis of the basic residues in the tail of TgM-A fused to GFP were undertaken (Figure A). Faithful expression and stability of the respective constructs were confirmed by immunoblotting (Figure C). The GFP fusions with mutants of TgM-A tail were also examined by indirect immunofluorescence (Figure B). Deletion of the C-terminal extension composed of the last 14 amino acids (TgM-AtailΔ14) did not alter the association of mycTgM-A with the cell membrane. In contrast, a further deletion encompassing the last 22 amino acids (TgM-AtailΔ22) caused a complete loss of membrane localization. The mutagenesis of two arginine residues within the last 22 amino acids into alanine residues (TgM-Atail mut III) completely abolished localization. However, the conversion of the three other sets of basic residues into neutral amino acids (TgM-Atail mut I and TgM-Atail mut II) did not alter significantly the plasma membrane distribution. These data indicate that the localization of TgM-A is directed by two precise residues rather than by the overall positive charge of the tail domain.