Export of Plasmodium falciparum calcium-dependent protein kinase 1 to the parasitophorous vacuole is dependent on three N-terminal membrane anchor motifs



Calcium-dependent protein kinases play a pivotal role in calcium signalling in plants and some protozoa, including the malaria parasites. They are found in various subcellular locations, suggesting an involvement in multiple signal transduction pathways. Recently, Plasmodium falciparum calcium-dependent protein kinase 1 (PfCDPK1) has been found in the membrane and organelle fraction of the parasite. The kinase contains three motifs for membrane binding at its N-terminus, a consensus sequence for myristoylation, a putative palmitoylation site and a basic motif. Endogenous PfCDPK1 and the in vitro translated kinase were both shown to be myristoylated. The supposed membrane attachment function of the basic cluster was experimentally verified and shown to participate together with N-myristoylation in membrane anchoring of the kinase. Using immunogold electron microscopy, the protein was detected in the parasitophorous vacuole and the tubovesicular system of the parasite. Mutagenesis of the predicted acylated residues and the basic motif confirmed that dual acylation and the basic cluster are required for correct targeting of Aequorea victoria green fluorescent protein to the parasitophorous vacuole, suggesting that PfCDPK1 as the leishmanial hydrophilic acylated surface protein B is a representative of a novel class of proteins whose export is dependent on a ‘non-classical’ pathway involving N-myristoylation/palmitoylation.


Calcium-dependent, calmodulin-independent protein kinases (CDPKs) form multigene families in plants and some protozoans and appear to constitute the major calcium-regulated protein kinase family active in signal transduction in these organisms (Kappes et al., 1999; Harmon et al., 2000; Hrabak et al., 2003). The unique calmodulin-independent nature of these kinases derives from a characteristic and highly conserved N-terminal serine/threonine protein kinase domain, which is contiguous with a C-terminal calmodulin-like calcium-binding domain (Roberts and Harmon, 1992).

In plants, CDPKs translate Ca2+ signals generated by external stimuli into cellular responses, thereby regulating cell division and differentiation, host symbiont interaction, the development of tolerance to stress stimuli and of defence responses to pathogens (Harmon et al., 2000; Romeis et al., 2001; Ludwig et al., 2004). The genome of Plasmodium falciparum encodes a family of five classical CDPKs named PfCDPK1–5 sharing between 39% and 56% amino acid identity and a CDPK-related protein kinase (Farber et al., 1997; Li et al., 2000; B. Kappes, unpubl. data). Whereas the functions of PfCDPK2, 3 and 5 are unknown, PfCDPK1 seems to be involved in merozoite invasion and membrane biogenesis processes of blood stage parasites (B. Kappes, unpubl. results). The Plasmodium berghei equivalent of PfCDPK4, PbCDPK4, has recently been identified as the molecular switch that translates the xanthurenic acid-induced calcium signal into a cellular response by regulating cell cycle progression in the male gametocyte (Billker et al., 2004). In addition, it may have a second function in ookinete infectivity (Billker et al., 2004).

Apart from their calcium requirement, very little is known about the mechanisms that regulate their activities and their biological roles. CDPKs are, however, found in several subcellular localizations (membranes, cytoplasm) and some isoforms are located in more than one compartment (Satterlee and Sussman, 1998; Lu and Hrabak, 2002; Dammann et al., 2003).

PfCDPK1 has been shown to be present in the membrane and organelle fraction of the blood-stage parasites and in the membrane fraction of ring-stage infected erythrocytes (Zhao et al., 1994a,b). Membrane localization of a rice CDPK results from dual acylation with myristate and palmitate, which is most probably the mechanism by which the majority of the membrane-associated plant CDPKs are bound to membranes, as all of them contain a consensus sequence for N-myristoylation and a cysteine in close proximity (Martin and Busconi, 2000; Dammann et al., 2003).

Proteins destined to become myristoylated possess the sequence Met–Gly at their N-terminus. The initiating methionine is removed co-translationally and myristate is linked to Gly2 via an amide bond. The requirement for Gly at the N-terminus is essential and no other amino acid will substitute. However, not all proteins with an N-terminal glycine are N-myristoylated. The ability to be recognized by N-myristoyl transferase depends on the downstream amino acid sequence. Palmitoylation usually occurs via attachment of palmitate through a thioester linkage to the sulphydryl group of a cysteine. The location of these palmitoylations varies. In dually fatty-acylated proteins, this modification occurs when the cysteine residue is located near to a previously myristoylated Gly2 at the N-terminus, whereby the preferred location for the cysteine is position 3 (Resh, 1996, 1999).

The binding energy provided by myristate is relatively weak and not sufficient to fully anchor a protein to a cellular membrane (Peitzsch and McLaughlin, 1993). The second signal for membrane binding has been defined as either a polybasic cluster of amino acids forming electrostatic interactions with the head groups of acidic phospholipids or a palmitate moiety. For proteins with a ‘myristate plus basic’ motif, the presence of both the hydrophobic and electrostatic forces leads to a synergism and may result in a nearly 3000-fold enhancement of binding to membranes as it has been reported for Src (Resh, 1999). The basic clusters of some members of the myristate plus basic motif group, such as MARCKS, the myristoylated alanine-rich C kinase substrate, contain phosphorylatable amino acids. For MARCKS phosphorylation of serines within its basic cluster reduces the electrostatic attraction, producing translocation of the protein from the membrane to the cytosol, a mechanism that is called the myristoyl-electrostatic switch (McLaughlin and Aderem, 1995).

Inspection of the N-terminus of PfCDPK1 revealed the presence of three motifs which may contribute to its membrane attachment: a myristoylation motif, a cysteine at position 3, which may serve as a palmitoylation site and a polybasic cluster of amino acids close to the N-terminus (MGCSQSSNVKDFKTRRSKFNT). To our knowledge, PfCDPK1 is thus the first example of a protein that contains two supplementary signals for membrane binding in addition to N-myristoylation being studied. The presence of this triple membrane attachment motif led us to the assumption that one of the signals or a combination thereof might serve different functions than just simply membrane anchoring, which could be cellular targeting.

Thus the goals of the present study were to determine the  precise  subcellular  localization  of  the  kinase  within the malaria parasite, to analyse whether PfCDPK1 is N-myristoylated, to characterize the contribution of N-myristoylation and the polybasic cluster to the membrane binding properties of the enzyme and to investigate whether one of the three motifs or a combination thereof is involved in targeting PfCDPK1 to its correct localization within the infected red blood cell.


Endogenous PfCDPK1 is myristoylated

To determine whether N-myristoylation of PfCDPK1 occurs within the parasite, parasite cultures were metabolically labelled with [3H]-myristate. After the labelling, parasites were harvested, lysed and PfCDPK1 immunoprecipitated using two affinity-purified PfCDPK1-specific antibodies. Lysates and immunoprecipitates were analysed by SDS-PAGE and fluorography. With both antibodies a pronounced signal at ≈ 60 kDa corresponding to the size of PfCDPK1 was obtained. In addition, a fainter band with a slightly lower electrophoretic mobility was observed. This pattern is typical for PfCDPK1, which in total parasite lysate always appears as a double band of which the upper is of much lower intensity. Generally, this effect is more pronounced if the antibody 2129 is used (Zhao et al., 1994a). This result clearly demonstrates that endogenous PfCDPK1 underwent myristoylation (Fig. 1).

Figure 1.

PfCDPK1 is myristoylated in vivo. Parasite cultures were labelled with [3H]-myristate and cell lysates were prepared. PfCDPK1 was immunoprecipitated with two PfCDPK1-specific anti-sera, electrophoretically fractionated and analysed by SDS-PAGE and fluorography. Lane 1, [3H]-myristate-labelled parasite lysate; lanes 2 and 3, immunoprecipitates with the PfCDPK1-specific antibodies 2979 and 2129 respectively.

The PfCDPK1 gene product is myristoylated in vitro

To analyse the contribution of N-myristoylation and the polybasic cluster to membrane anchoring of PfCDPK1, three mutants were generated: a myristoylation-negative mutant (VG2), in which Gly at position 2 was replaced by a Val; a deletion mutant, Δ10-20, lacking the polybasic cluster and a double mutant (DM) lacking both features (Table 1). The products of the cDNAs of wild-type PfCDPK1 and the mutants VG2, Δ10-20 and DM were [35S]-methionine labelled in a coupled in vitro transcription/translation system under the conditions given in Experimental procedures. After the reaction, the products were separated on an 8% SDS-PAGE and analysed by fluorography. Figure 2A shows the results of the T7-directed transcription and translation of the constructs in the presence of [35S]-methionine. The translational products of all plasmids were of the expected molecular size of about 60 kDa. The deletion mutant Δ10-20 and the double mutant which both lack 10 amino acids clearly have a slightly higher electrophoretic mobility as the wild-type and VG2 mutant (Fig. 2, lanes 3 and 4). A band of a reduced molecular mass appeared in all coupled in vitro transcription/translation reactions (Fig. 2A), which may either originate from an internal translation initiation or from proteolytic degradation. The control reaction without plasmid did not yield any bands of comparable intensity (Fig. 2A, lane 5).

Table 1. Membrane attachment motifs of PfCDPK1 and mutants.
  1. The membrane attachment motifs of PfCDPK1 and mutant proteins are in bold face. Phosphorylatable amino acids within the basic cluster are underlined.

Figure 2.

In vitro synthesis and myristoylation of wild-type PfCDPK1 and mutant proteins. Wild-type PfCDPK1 and mutant proteins were [35S]-methionine labelled (A) or [3H]-myristate labelled (B) in a coupled in vitro transcription/translation as described in Experimental procedures. Samples were incubated at 30°C for 90 min and analysed by SDS-PAGE and fluorography. The gels were exposed for either 1 h without enhancing screen ([35S]-methionine labelling) or for 14 days with enhancing screen ([3H]-myristate labelling) at −70°C. Lane 1, wild-type PfCDPK1; lane 2, VG2; lane 3, Δ10-20; lane 4, DM; and lane 5, haemolysate control. The molecular weight standards are given on the right side of the figure.

To test whether the mutations had the assumed effect on myristoylation, the coupled in vitro transcription/translation reactions were carried out in the presence of [3H]-myristate under the conditions given in Experimental procedures. As expected the products of the cDNA coding for wild-type PfCDPK1 and the Δ10-20 mutant, which contain the Gly residue at position 2, were myristoylated (Fig. 2B, lanes 1 and 3). In contrast, no myristoylation occurred on those gene products, where Gly2 had been replaced (VG2 and DM) (Fig. 2B, lanes 2 and 4). The product with the reduced molecular mass, which appeared upon [35S]-methionine labelling in all four reactions, was not myristoylated, indicating that its N-terminus is missing. N-myristoylation of other products appeared in all reactions including the reticulocyte lysate control (Fig. 2B, lane  5). Thus, these signals result from N-myristoylation of endogenous reticulocyte lysate proteins.

Acylation and the basic motif contribute to the membrane attachment of PfCDPK1

In general, N-myristoylation plus either palmitoylation or a polybasic cluster of amino acids is sufficient to fully anchor a peptide or protein to a cellular membrane. PfCDPK1, however, possesses all three motifs. Palmitoylation of cysteine residues does not occur in a coupled in vitro transcription/translation reaction probably due to the absence of palmitoyl acyltransferases in the haemolysate (Busconi et al., 1997). This observation was confirmed for the conditions applied in our experiments (data not shown). Therefore, coupled in vitro transcription/translation enables membrane binding to be studied in the absence of palmitoylation and allows to address the question whether both the myristate moiety and the basic motif confer membrane attachment to PfCDPK1 and if so to dissect their individual contribution.

To determine the effects of N-myristoylation and the basic motif on membrane association, cDNAs of wild-type and the mutants VG2, Δ10-20 and DM were in vitro transcribed/translated in the presence of microsomal membranes and [35S]-methionine. After the reactions, the samples were pelleted by ultracentrifugation at 100 000 g to yield a soluble and a membrane fraction. The presence of the gene products in either fraction was analysed by SDS-PAGE and fluorography. After fluorography, the individual bands were cut out of the gel and the incorporated radioactivity was determined.

Membrane association of PfCDPK1 increased with the amount of microsomal membranes added to the coupled in vitro transcription/translation reaction. However, an increase in microsomal membranes had an opposite effect on the efficiency of the in vitro transcription/translation resulting in a reduction in the yield of translational product (data not shown). We therefore chose those conditions that provided optimal translation efficiency to membrane binding ratio, which in our hands was the addition of 1.4 µl of canine pancreatic microsomal membranes. All experiments described hereafter were therefore performed with this amount of microsomal membranes. Fifty-six per cent of the full-length translational product of PfCDPK1 was bound to microsomal membranes under these conditions. For calculation of the membrane binding efficiency of the mutant proteins, this value was set to 100%. Unspecific binding of PfCDPK1 was ruled out by the fact that the binding of the truncated product, which is lacking the N-terminus and thus all three membrane-anchoring motifs, to microsomal membranes was almost negligible in the presence of full-length PfCDPK1.

The membrane binding efficiency of the deletion mutant, missing the basic cluster, should solely be dependent on N-myristoylation and should be impaired if compared with the binding efficiency of the wild-type protein. Indeed, the membrane binding of the Δ10-20 was reduced to 32% when compared with wild-type (Fig. 3). This result verifies the postulated membrane binding function of the basic cluster of PfCDPK1. Membrane anchoring of the myristoylation-negative VG2 mutant should exclusively result from the polybasic cluster and should be decreased if compared with wild-type binding. As a matter of fact, the portion of VG2 present in the membrane bound fraction was reduced to 28% (Fig. 3). With 7.5% the double mutant, DM, exhibited only residual membrane binding (Fig. 3). These data clearly prove that both N-myristoylation and the polybasic cluster promote membrane binding of PfCDPK1 in vitro and demonstrate that their contribution is virtually equal under the applied conditions.

Figure 3.

Binding of in vitro translated wild-type PfCDPK1 and mutant proteins to acceptor membranes. Wild-type PfCDPK1 and mutant proteins were [35S]-methionine labelled in a coupled in vitro transcription/translation in the presence of acceptor membranes as described in Experimental procedures. Samples were ultracentrifugated into a soluble and particulate fraction, which were analysed by SDS-PAGE and fluorography. To determine the amount of PfCDPK1 in the individual fractions, the respective bands were cut out of the gel and counted. The experiment was repeated at least three times with similar results.

To investigate the membrane binding properties of the above-used mutants in the living parasite, we decided to create fusions of the mutants and PfCDPK1 wild-type further on referred to as PfCDPK1-wt with luciferase as reporter gene. However, instead of DM, which is in principal a triple mutant, as palmitoylation requires preceding N-myristoylation, an actual triple mutant (TM) was used, where all three membrane binding motifs had been altered (Table 1). P. falciparum was transfected with the respective constructs and analysed for the expression of the recombinant genes 72 h after transfection. Infected erythrocytes were treated with increasing concentrations of saponin and separated into a pellet and a supernatant fraction. The distribution of the PfCDPK1-wt–luciferase fusion protein in both fractions was quantified by the enzymatic activity of the luciferase as described by Burghaus and Lingelbach (2001). Up to a saponin concentration of 0.8%, 85% of the luciferase activity of the PfCDPK1-wt construct was detected in the pellet fraction (Fig. 4A). Complete release of aldolase, which we used as a marker for soluble proteins of the parasite cytoplasm, is observed at 0.5% saponin and a complete release of PfBiP, a heat shock protein of the endoplasmic reticulum and our marker for proteins of internal compartments of the parasite, at 0.8% (Burghaus and Lingelbach, 2001). These findings suggest that the high portion of luciferase activity of the fusion construct in the pellet fractions can be ascribed to the membrane binding properties of PfCDPK1 and is not caused by compartmentalizational artefacts. To obtain indications for the membrane binding properties of the mutants, we decided to analyse the effect of the mutations on membrane attachment at a saponin concentration of 0.6%, which is above the concentration at which the complete release of soluble cytosolic proteins occurs. To keep experimental variations at a minimum, transfections, lysis and analysis of PfCDPK1-wt and mutant constructs were performed in parallel. Under the applied conditions, about 94% of the luciferase activity of the triple mutant construct, which is lacking all membrane binding motifs, was detected in the supernatant fraction, suggesting that membrane attachment of PfCDPK1 in vivo is indeed mediated by the N-terminal motifs. Thirty-nine per cent of the luciferase activity of the VG2 construct, which should be deficient for N-myristoylation and palmitoylation but is still able to attach to membranes by means of the basic cluster, was found in the pellet fraction, indicating that the basic stretch indeed exhibits a membrane binding function in the living parasite. Whereas VG2 and the Δ10-20 displayed comparable membrane binding properties when analysed in the in vitro system (Fig. 3), they quite substantially differed when analysed in vivo. Only 15% of the luciferase activity of the Δ10-20 construct was detected in the supernatant fraction at a saponin concentration of 0.6%. As Δ10-20 contains both acylation sites, the difference in the in vitro and in vivo behaviour of this mutant can easily be explained by a palmitoylation event, which should take place in the parasite but does not occur in the in vitro system used. Thus, the stronger membrane attachment of the Δ10-20 mutant provides indirect evidence for a palmitoylation of PfCDPK1 in the living parasite.

Figure 4.

Differential solubility of luciferase-tagged PfCDPK1 wild-type and mutant proteins in infected red blood cells.
A. Erythrocytes infected with parasites transfected with the PfCDPK1-wt–luciferase construct were lysed with increasing concentrations of saponin. Samples were centrifuged and the activity of luciferase was determined in the pellet and supernatant fractions.
B. Erythrocytes infected with parasites transfected with the luciferase-tagged PfCDPK1-wt and mutant constructs (Δ10-20, VG2 and TM) were lysed at a saponin concentration of 0.6%. Samples were centrifuged and luciferase activity of the pellet and supernatant fractions was determined.
Black bars represent the pellet and grey bars the supernatant fractions. Each bar represents the arithmetic mean of three independent experiments. The standard deviation is given on top of the bar.

Autophosphorylation of membrane-associated PfCDPK1 does not affect its membrane attachment

The basic cluster of PfCDPK1 contains amino acid motifs, which fulfil the requirements of a phosphorylation recognition motif of PfCDPK1 (K/R-X-X-X-S/T). The serine and threonine residues within the basic cluster could theoretically be autophosphorylated by the kinase itself resulting in a weakened electrostatic interaction of the basic stretch with the membrane phospholipids. To investigate the influence of autophosphorylation on membrane binding of PfCDPK1, its cDNA was in vitro transcribed/translated in the presence of microsomal membranes and [35S]-methionine as described above. The 100 000 g pellets, containing the membrane-bound kinase, were resuspended in kinase assay buffer and the reactions started by addition of ATP. Autophosphorylation was allowed to proceed for 30 min at 30°C in the presence or absence of Ca2+. After the kinase reaction, samples were pelleted by ultracentrifugation and the soluble and membrane fractions analysed by SDS-PAGE and fluorography. In the presence of Ca2+ under conditions where autophosphorylation should occur, PfCDPK1 remained bound to the microsomal membranes (Fig. 5, lane 3). No difference was observed to conditions where autophosphorylation was abolished by the absence of Ca2+ and presence of EGTA in the reaction (Fig. 5, lane 1). To exclude that lack of autophosphorylation within the basic effector domain results from the fact that the in vitro translated PfCDPK1 is not active, recombinantly expressed active PfCDPK1 was added to the assays. Addition of the recombinant enzyme to the membrane bound in vitro transcribed/translated kinase had no effect on its membrane attachment (data not shown). Consequently, under the conditions applied, autophosphorylation had no effect on the membrane anchoring of PfCDPK1.

Figure 5.

Effect of autophosphorylation on the binding of in vitro translated wild-type PfCDPK1 to acceptor membranes. Wild-type PfCDPK1 was [35S]-methionine labelled in a coupled in vitro transcription/translation in the presence of acceptor membranes as described in Experimental procedures. Particulate fractions were resuspended in kinase assay buffer in the presence (+) or absence (–) of Ca2+ and incubated at 30°C for 30 min in the presence of ATP. These reactions were again fractionated by ultracentrifugation into the soluble (S) and particulate (P) fractions, which were then analysed by SDS-PAGE and fluorography. The gel was exposed for 2 h at −70°C. The molecular weight standards are given at the right side of the figure.

PfCDPK1 is located in the parasitophorous vacuole and the tubovesicular system of the parasite

To determine the precise subcellular localization of PfCDPK1 within the parasite, the affinity-purified antibody 2129 was used in immunoelectron microscopic examinations of sections of red blood cells infected with P. falciparum. Antibody binding was visualized by using collodial gold-conjugated protein A. In trophozoites and schizonts, intraerythrocytic stages of the parasite, gold labelling of PfCDPK1 was most prominent in the compartment of the parasitophorous vacuole and in all membranous systems derived from the parasitophorous vacuole (Fig. 6A–D). These are intraparasitic vacuoles (IPVs) of trophozoites and schizonts (Fig. 6A and B) and the tubovesicular system (TVS), an extension of the parasitophorous vacuole membrane (PVM) into the host cell cytoplasm, which builds an enormous network of membranes in this compartment (Fig. 6C and D). The Golgi-like structure formed by the TVS in Fig. 6C is also known as Maurer's cleft. In the parasitophorous vacuole, the majority of the gold particles were either found in association with the plasma membrane of the parasite or with the PVM (Fig. 6A, B and D). Rarely, gold label associated with PfCDPK1 was found free in lumenal spaces (see lumenal spaces of the IPV in Fig. 6A). Only very low levels of gold particles were observed in the cytoplasm of trophozoite or schizonts. No labelling was present on uninfected red blood cells. Preimmune serum from the rabbit showed little or no labelling at the same concentration as the affinity-purified antibody 2129. These results confirm our recent cell fractionation results and provide further evidence that the majority of PfCDPK1 is membrane associated in the living parasite.

Figure 6.

Subcellular localization of PfCDPK1. Sections of P. falciparum (clone K1) erythrocytic stages were incubated with the affinity-purified PfCDPK1-specific antibody 2129 (dilution 1:5, Experimental procedures) and binding was visualized using gold-conjugated protein A.
A and B. Gold particles indicative of antibody binding associated with the parasitophorous vacuole and intracellular vacuoles (intraparasitic vacuole) of schizont stages. Intracellular vacuoles originating from the plasma membrane of the parasite (B).
C and D. Sections of erythrocytes infected with two schizonts and one schizont, respectively, showing labelling of the tubovesicular system in the host erythrocyte. The membranous structure visible in the erythrocyte cytosol in (C) is also known as Maurer's clefts.
EC, erythrocyte cytosol; MC, Maurer's clefts; IPV, intraparasitic vacuole; N, nucleus; PM, parasite plasma membrane; PV, parasitophorous vacuole; PVM, parasitophorous vacuole membrane; TVS, tubovesicular system. Size bars (200 nm) have been inserted in each panel.

Dual acylation and the basic cluster are required to target PfCDPK1 to the parasitophorous vacuole

According to the two signal model, one membrane binding motif in addition to N-myristoylation is sufficient to fully anchor a protein to a membrane (Resh, 1999). The luciferase-tagged Δ10-20 mutant, which still possesses two signals for membrane anchoring that are the two acylation motifs, displayed only slightly reduced membrane binding properties when compared with wild-type PfCDPK1 (Fig. 4B). This finding suggested that one of the membrane binding motifs of PfCDPK1 or combinations of them may exert an additional function, which we thought might be targeting to the parasitophorous vacuole. We sought to test this model by individually testing the targeting properties of each motif. To examine the role of each motif and combinations thereof in targeting, gene constructs of PfCDPK1 and mutants were made to create a fusion of GFP with the 48 N-terminal amino acids of PfCDPK1. Upon expression in the parasite, the WTsignal-GFP accumulated outside of the parasite within the parasitophorous vacuole. This secreted form of GFP was evident as continuous fluorescence surrounding the parasite in trophozoite (data not shown) and early schizont stages (Fig. 7A, panels a and b) and as a cobblestone-patterned fluorescence surrounding the developing merozoites of schizont stages (Fig. 7A, panel c). Similar patterns have been observed with the ACPsignal-GFP construct, which accumulates within the parasitophorous vacuole (Waller et al., 2000). Some GFP was associated with the food vacuole and hemozoin, which might be caused by re-entrance of the GFP into the parasite with erythrocyte cytoplasm ingestion. The export of GFP out of the parasite confirms that the first 48 amino acids of PfCDPK1 have targeting function.

Figure 7.

Plasmodium falciparum expressing WTsignal- (A) and mutantsignal-GFPs (B).
A. WTsignal-GFP expression results in export to the parasitophorous vacuolar space: schizont stages showing continuous fluorescence surrounding the parasite (panels a and b), which ultimately delineates the products of schizogony (panel c).
B. VG2signal-, AC3signal-, Δ10–20signal- and DMsignal-GFP expression is apparently cytosolic and excluded from the food vacuole.
First column, transmitted light; second column, GFP-fluorescence; third column, DNA staining with Hoechst 33342 (fc 1.25 µM); fourth column, overlay of the first three columns. The scale bars represent 2 µm with the exception of panel a in A where it represents 1 µm.

To delineate the postulated targeting function of the triple motif, the following mutants were analysed: VG2 lacking the myristoylation site, AC3 missing the palmitoylation signal, Δ10-20 containing a deletion of the basic cluster, DM lacking the myristoylation site and the basic cluster and finally TM a triple mutant lacking all three motifs. In contrast to WTsignal-GFP, GFP-fluorescence of all mutants was restricted to the parasite cytosol (Fig. 7B). None of the mutants showed a pattern which was reminiscent of a parasitophorous vacuole staining or a cobblestone-patterned fluorescence in developing merozoites of schizont stages as observed for WTsignal-GFP. GFP-fluorescence of all mutant proteins was found to be evenly distributed within the parasite cytosol (Fig. 7B, panels VG2, Δ10-20 and DM) except for AC3, where some GFP-fluorescence appears to concentrate in diffuse foci (Fig. 7B, panel AC3). As palmitoylation requires a preceding myristoylation, DM and TM are both supposed to lack the membrane binding properties of all three motifs and therefore should have the same effect on the localization of the GFP fusion proteins. As expected, both mutants showed a similar GFP-fluorescence, which is the reason why TM was not included in the Fig. 7B. The GFP-fluorescence of the GFP control construct, containing no additional sequence, was detected in the parasite cytosol. Together these results demonstrate that the triple motif exhibits a targeting function for the export of proteins to the parasitophorous vacuole. Moreover, all three motifs seem to be necessary to exit the parasite.


Many acylated proteins contain a combination of two membrane-anchoring features. This ‘two signal model’ for membrane binding is highlighted by proteins belonging to the Src family and α-subunits of heterotrimeric G-protein complexes, which are modified at their N-termini with both myristate and palmitate. In addition, for Src itself, HIV-Gag-1 and the MARCKS protein N-myristoylation has been observed in combination with N-terminal clusters of basic amino acids (Resh, 1999). In contrast to these proteins, PfCDPK1 possesses a combination of three membrane-anchoring features at its N-terminus, which is so to say one signal too much to simply confer membrane binding. To dissect the function of this tripartite motif in membrane addressing and attachment of PfCDPK1, the present work focused on the characterization of two of its molecular determinants, the myristoylation site and the basic effector domain.

PfCDPK1 is myristoylated in vivo and in vitro, proving that its myristoylation site is functional (Figs 1 and 2B). In the rabbit reticulocyte lysate, a protein of the predicted molecular size (60 kDa) was synthesized. This full-length translational product of wild-type PfCDPK1 underwent myristoylation (Fig. 2B). As expected, N-myristoylation was abolished in those mutants in which the myristoylation site had been replaced by a valine (Fig. 2B). In addition to the full-length translational product, a minor band of reduced molecular mass appeared after in vitro transcription/translation. No N-myristoylation of this truncated product was observed, indicating that it arose either from proteolytic degradation of the N-terminus or from internal translation initiation (Fig. 2A and B).

Palmitoylation does not occur in a coupled in vitro transcription/translation reaction (Busconi et al., 1997), which was confirmed for the experimental conditions we used (data not shown). This allows membrane binding in vitro to be studied in the absence of palmitoylation and thus to address the question whether and if yes to which extent myristoylation and/or the basic effector domain contribute to the membrane attachment of PfCDPK1. Membrane binding of the mutant lacking the polybasic amino acid stretch was reduced to 32%, that of the myristoylation-negative mutant to 28% and that of the double mutant to 7.5% of that of wild-type PfCDPK1 demonstrating that both myristoylation and the basic motif promote membrane binding of the kinase. Moreover, in the system used, N-myristoylation and the polybasic cluster contributed almost equally to the membrane attachment of the enzyme. With 7.5%, the double mutant displayed only a residual membrane binding efficiency.

The influence of the above mutations on the membrane binding properties of PfCDPK1 in vivo was analysed with luciferase-tagged PfCDPK1-wt and mutant proteins. To exclude that the mutants might be trapped within the parasite, a saponin concentration of 0.6% was chosen, at which the release of aldolase, our marker for soluble proteins of the parasite cytosol, was complete. The membrane binding efficiency of the VG2 mutant, which should be neither myristoylated nor palmityolated and simply attaches to membranes by means of the basic cluster, is 40% under the experimental conditions used and thus is even higher than the one observed in our in vitro system (28%; compare Figs 3 and 4B). The Δ10-20 mutant, which possesses both acylation sites and theoretically should attach via myristate and palmitate in vivo, displays roughly 85% of the membrane binding efficiency of PfCDPK1-wt. This latter finding indicates that (i) palmitoylation indeed occurs in vivo and contributes to membrane anchoring of PfCDPK1 and (ii) all three membrane binding motifs of PfCDPK1 are required to achieve the full membrane binding capacity of the wild-type kinase. A further support that palmitoylation at Cys3 takes place in vivo is provided by the effect of its mutation on the localization of the GFP-tagged AC3 mutant, which resides within the parasite cytosol (Fig. 7B). The triple mutant, which lacks all three membrane binding motifs and should behave in the same manner as the double mutant used in the in vitro assay, displayed only a residual membrane binding activity.

Whereas for Src, the synergistic effect of hydrophobic and electrostatic interactions results in a 3000-fold enhancement of membrane binding efficiency, the interactions of the membrane anchors of PfCDPK1 appear to be rather additive than synergistic. Our in vitro and in vivo data suggest that each membrane anchor contributes about equally to the membrane binding properties of PfCDPK1.

MARCKS associates with membranes by the cooperative interaction of the myristoyl moiety and the basic effector domain. Protein kinase C phosphorylation of serine residues within the basic domain disrupts the electrostatic interaction with acidic phospholipids because of electrostatic repulsion and releases the protein from the membrane. This cycle of reversible membrane attachment has been called the myristoyl-electrostatic switch (McLaughlin and Aderem, 1995). PfCDPK1 possesses serine and threonine residues embedded in its basic effector domain, suggesting that it may be subjected to a myristoyl-electrostatic switch mechanism. Some of the serine and threonine residues of this basic cluster meet the requirements of a phosphorylation motif of PfCDPK1, which prompted us to investigate whether autophosphorylation may play a role in reversible membrane attachment of the kinase. To elucidate the function of autophosphorylation in a postulated myristoyl-electrostatic switch mechanism, the membrane bound enzyme was subjected to a kinase reaction either in the presence or in the absence of recombinant PfCDPK1. Data presented in Fig. 5 show that autophosphorylation did not weaken the interaction of PfCDPK1 with microsomal membranes (lane 3) and consequently had no influence on its membrane attachment. This also proved to be true when the active recombinant enzyme was added to the reaction. If a myristoyl-electrostatic switch mechanism acts on reversible membrane binding of PfCDPK1 as suggested by the presence of serine and threonine residues in the basic effector domain, it will most probably be triggered by another kinase or kinases.

The presence of an additive rather than a synergistic effect of the membrane anchors of PfCDPK1 might indeed result from a myristoyl-electrostatic switch mechanism operating in the living parasite. In light of three instead of two membrane anchors, a strong synergistic effect would probably hamper the presumed reversible membrane attachment of PfCDPK1 induced by the introduction and removal of phosphorylations in the basic cluster and thus may interfere with PfCDPK1 function.

Up to now, the plasmodial CDPKs are the only characterized parasite kinases which contain a basic cluster and acylation sites as membrane anchoring motifs. Other parasite kinases, such as PfPK2, PfPK4 and PfFEST, which are known to be membrane associated as well, have none  of  the  above  motifs,  but  contain  several  lysines and/or arginines in their N-terminal region, which may contribute to membrane attachment (Zhao et al., 1992; Kun et al., 1997; Mohrle et al., 1997). Inspection of putative protein kinases identified in the Plasmodium genome database PlasmoDB (http://PlasmoDB.org; The Plasmodium Genome Database Collaborative, 2001) revealed the presence of several kinases that contain motifs for myristoylation and palmitoylation at their N-termini. It is, however, striking that the other two supposably membrane attached plasmodial CDPKs, PfCDPK2 and 4, contain a basic cluster and a myristoylation but no palmitoylation motif. In contrast, the majority of the membrane attached plant CDPKs harbour sites for myristoylation and palmitoylation (Farber et al., 1997; Dammann et al., 2003).

Because the location of proteins certainly conditions their functional role, we determined the precise localization of PfCDPK1 within the infected red blood cell using immunoelectron microscopy. Although the enyzme does not possess any motif, which could be regarded as a conventional or known targeting signal sequence to the vacuolar space, PfCDPK1 was found in ‘extracellular compartments’ of the parasite, the parasitophorous vacuole and the TVS, in ring, trophozoite and schizont stages.

The location of PfCDPK1 in the parasitophorous vacuole is in good agreement with its Ca2+ requirements. The mean calcium dissociation constant of PfCDPK1 is 80 µM and thus is about 20-fold higher than the one of calmodulin (Zhao et al., 1994b). The lower affinity of PfCDPK1 to Ca2+ can easily be explained by its localization in the vacuolar compartment. The Ca2+ concentration in the parasite's parasitophorous vacuole is in the order of 40 µM and thus about 100- to 1000-fold higher than the one in the infected erythrocyte or parasite cytoplasm (Gazarini et al., 2003). Exerting its functions in an environment of high Ca2+ concentration, the calcium affinity of PfCDK1 has to be appropriately adjusted and fine-tuned to function in the Ca2+ signalling system of the parasitophorous vacuole.

What function PfCDPK1 could exhibit within the parasitophorous vacuole and parasitophorous vacuole-derived compartments? Based on some experimental evidence, our current working hypothesis suggests that PfCDPK1 is involved in the control of membrane biogenesis processes of the parasite. The fact that PfCDPK1 is found in intra- and extracellular compartments of the parasite where dynamic changes of membranes, such as the growth of the PVM, the construction and remodelling of the tubovesicular network and the formation and changes of IPVs, are taking place is in good congruence with the proposed working hypothesis. Moreover, the kinase accumulates in those regions where active membrane biogenesis processes, e.g. merozoite membrane formation, are taking place (B. Kappes, unpubl. obs.).

One of the most exciting findings of the present work is the identification of a novel export signal for the transport of proteins past the parasite's plasma membrane. The key signature of this novel export signal seems to be a combination of three membrane anchors, which are N-myristoylation, palmitoylation and electrostatic interactions by means of the basic cluster. All three membrane attachment sites appear to be required for the correct targeting of PfCDPK1 to the parasitophorous vacuole. As none of the mutant proteins, in which a single membrane binding motif had been altered, was correctly targeted, it may as well be that the triple motif is required and sufficient for the export of the kinase to the vacuolar space (Fig. 7B). In addition, this investigation provides indirect evidence for a palmitoylation of Cys3, as its replacement by alanin results in an export-deficient mutant (Fig. 7B).

Secretory proteins typically contain N-terminal signal peptides that direct them to the translocation apparatus of the endoplasmic reticulum (Walter et al., 1984). After vesicular transport from the endoplasmic reticulum (ER) via the Golgi to the cell surface, lumenal proteins are released into the extracellular space whereas transmembrane proteins and proteins with a membrane anchor remain at the cellular surface by fusion of Golgi-derived secretory vesicles with the plasma membrane. This pathway of protein export from eukaryotic cells is known as the classical ER/Golgi-dependent transport (Rothman and Wieland, 1996; Mellman and Warren, 2000). In Plasmodium the situation is more complex, because the parasite has to target some of its proteins to the red blood cell cytoplasm and membrane. To do so, the parasite has to transport them past its own plasma membrane, across the parasitophorous vacuole and possibly across the erythrocyte cytosol. As the mature erythrocyte is devoid of the biosynthetic/secretory machinery present in other cells, the parasite has developed an unusual trafficking system. As a part of this system, it appears to transpose components of its classical secretory machinery outside the confines of its own plasma membrane into the host cytoplasm (Albano et al., 1999; Adisa et al., 2001; Hayashi et al., 2001; Wickham et al., 2001; Wickert et al., 2003). The available body of evidence suggests that transport from the parasite to the host erythrocyte is a multistep process, requiring at least a bipartite signal. The first N-terminal sequence, which may or may not be a canonical signal sequence, appears to mediate entry into the secretory pathway and transport into the parasitophorous vacuolar lumen. A recessed trafficking signal then allows proteins to pass the PVM and enter the erythrocyte cytosol (Burghaus and Lingelbach, 2001; Wickham et al., 2001; Lopez-Estrano et al., 2003).

In contrast to the classical ER/Golgi-dependent transport, other non-classical transport routes for protein secretion have been described, which in the majority of the cases are operational in the absence of a functional ER/Golgi system and have therefore been named ER/Golgi-independent protein secretion (Nickel, 2003). A quite remarkable example of non-classical protein export from eukaryotic cells is the mechanism of cell surface expression of Leishmania hydrophilic acylated surface protein B (HASPB), which is found on the surface only in the infectious stages of the parasite life cycle. Like PfCDPK1, Leishmania HASPB is a dually acylated protein modified by N-myristoylation and palmitoylation. Despite lacking any identifiable signal that could lead to translocation by the exocytic pathway, HASPB is surface localized in Leishmania major (Flinn et al., 1994; Pimenta et al., 1994). Mutational analysis revealed that the first 18 amino acids are required and sufficient to target HASPB to the cell surface. A mutant that retains the N-terminus but lacks the myristoylation site and a mutant lacking both acylation sites were redistributed into the cytoplasm (Denny et al., 2000). These findings are in good agreement with our data on PfCDPK1, where the mutants lacking either the myristoylation site (Fig. 7B, panel VG2) or both acylations sites (data not shown) are completely delocalized to the cytosol. A palmitoylation-deficient but myristoylation-competent mutant of HASPB was found to be associated with the outer leaflet of the Golgi (Denny et al., 2000). As AC3 appears to be the only mutant, which is not evenly distributed in the parasite cytosol, this might be the case for palmitoylation-negative mutant of PfCDPK1 as well. However, an association of PfCDPK1 with Golgi-like structures of the P. falciparum awaits further experimental validation. Based on their observations, Denny et al. (2000) proposed a model in which HASPB is transferred from the cytoplasm to the outer leaflet of the Golgi membrane, from where it is transported to the plasma membrane via conventional vesicular transport, a process which inserts HASPB into the inner leaflet of the plasma membrane. The same export route may be used by PfCDPK1. The mechanism how HASPB is then translocated across the membrane is completely unknown; however, a flip flop mechanism has been proposed (Nickel, 2003).

The results obtained with the luciferase- and the GFP-tagged mutant proteins may at first glance partly contradict each other. Both mutants, Δ10-20 and VG2, are partly attached to membranes and display 85% and 39% membrane binding activity, respectively, in the in vivo membrane binding studies, whereas the respective GFP mutants are evenly distributed within the parasite. It has, however, to be taken into account that the analysis of the GFP-fluorescence was performed with living parasites, whereas the analysis of the membrane binding properties of the luciferase-tagged mutants had to be performed after lysis of the transfected parasites under conditions at which lysis of the parasite plasma membrane, as measured by the release of aldolase, was complete. Thus, the conditions of both studies differ substantially. The interaction of the mutants with the membranes of the actual destination of PfCDPK1, namely the membranes confining the parasitophorous vacuole or microdomains thereof, which are freely accessible after saponin lysis, could be much stronger than their attachment to internal membranes of the parasite and could explain the different behaviour of the mutants in both studies.

Our findings obtained with either the luciferase- or the GFP-tagged fusion proteins provide indirect evidence that palmitoylation of Cys3 occurs within the living parasite. Similar to other lipid modifications, palmitoylation promotes membrane binding. The function of palmitoylation, however, ranges usually beyond that of a simple membrane anchor. Trafficking of lipidated proteins to the plasma membrane is in many cases dependent on palmitoylation (Bijlmakers and Marsh, 2003; Smotrys and Linder, 2004). In addition, modification with fatty acids impacts the lateral distribution of proteins on the plasma membrane by targeting them to lipid rafts. Many palmitoylated signalling proteins, including Src family kinases and some Gα-subunits, associate with lipid rafts. It has been suggested that localization to rafts increases the local protein concentration and facilitates protein–protein interactions especially of signalling proteins. Moreover, it has been proposed that signal transduction could be regulated, in part, by sequestering signalling proteins into different plasma membrane domains until they are brought together by an activating signal (Bijlmakers and Marsh, 2003; Smotrys and Linder, 2004). The presence of lipid rafts in the PVM and the apparent palmitoylation of PfCDPK1 suggest that the kinase might be localized to those vacuolar rafts (Lauer et al., 2000).

Is targeting via the triple motif of dual acylation and the basic cluster restricted to PfCPDK1? In course of the intraerythrocytic growth cycle, the parasite is forced to reconstruct its host cell according to its requirements. As a part of this reconstruction, not only proteins involved in nutrient uptake, immune evasion and trafficking are transposed outside the confines of the parasite's plasma membrane but also components of signalling pathways, such as PfCDPK1 and PfFEST, which is associated with parasite knobs (Kun et al., 1997). In principal, one could imagine that signalling proteins operating at the same cellular destination outside the parasite might simply facilitate their interaction and coordinate the signalling processes using similar or related targeting signals. Thus signalling proteins having their tasks within the vacuolar space or compartments derived from the parasitophorous vacuole might use similar export signals for targeting beyond the parasite's plasma membrane, particularly if they are directed to vacuolar rafts. Thinking one step further, a direct targeting signal to vacuolar rafts could even assists the interaction of those proteins significantly. The catalytic domain of a second plasmodial member of the PP2C protein phosphatase family, which so far has not been characterized, contains a myristoylation motif, a palmitoylation site and a basic effector domain at its N-terminus (B. Kappes, unpubl. findings) and may therefore be transported to the parasitophorous vacuole by means of the triple motif. Thus, we propose a model according to that proteins taking on signalling functions in the parasitophorous vacuolar compartment are targeted to the vacuolar space by a mechanism equal or related to the one described in the present publication.

Experimental procedures


Plasmodium falciparum isolate K1 (Thailand), adapted to growth in horse serum, was a gift of Dr Matile (F. Hoffmann-La Roche, Basel). P. falciparum HB3 and 3D7 parasites were kindly provided in house by Professor Lanzer. Professor Alan Cowman and Dr Tim Gilberger (The Walter and Eliza Hall Institute for Medical Research, Melbourne, Australia) kindly provided the transfection plasmid pARL-1a+.

Parasites and transfection

Parasite culturing.

P. falciparum K1 was cultured as described by Graeser et al. (1996). P. falciparum HB3 and 3D7 were cultivated according to Trager and Jensen (1976) in RPMI 1640 medium containing 5% human serum A+ under reduced oxygen.

Stable transfection of P.  falciparum isolate 3D7.

 Before transfection, 3D7 parasites were synchronized by sorbitol lysis (Lambros and Vanderberg, 1979). Transfection was carried out with 150 µg of circular DNA of WTsignal-GPF, VG2signal-GPF, AC3signal-GPF, Δ10–20signal-GPF, DMsignal-GPF, TMsignal-GPF or GFPcontrol, the GFP control construct lacking any additional sequences. Transfected parasites were cultured in 90 mm Petri dishes in the above-described culture medium supplemented with 0.25% Albumax II for 48 h without drug pressure before the medium was supplemented with 5 nM WR 99210. Transfectants were selected under drug pressure until 5% parasitemia was reached. Depending on the construct, the first parasites were observed after 15–20 days of selection. Green fluorescence of GFP-expressing transfectant parasites was observed and captured in live cells using a Zeiss LSM510 confocal microscope.

Transient transfection of P. falciparum isolate FCBR.

The P. falciparum isolate FCBR was cultured in RPMI 1640 (Sigma) supplemented with 20 mM glutamine (Sigma), 10% human plasma (Marburg Blood Bank) and 40 µg ml−1 gentamicin (Invitrogen). For transfection, human erythrocytes of blood group A+ (Marburg Blood Bank) were loaded with 50–75 µg of circular plasmid DNA as described by Deitsch et al. (2001). DNA-loaded erythrocytes were transferred immediately into 40 ml of culture medium and ring-stage parasites added to 2–3% parasitemia. Culture medium was changed once daily. On day four human erythrocytes were added to the cultures to reduce parasitemia to 4% and the medium was changed twice daily. A mock transfection was performed with cytomix and used as a control.

Harvest and fractionation of infected erythrocytes

Infected erythrocytes were harvested on day 6 and trophozoites stages enriched to > 50% by gel floatation as described previously (Pasvol et al., 1978). Saponin treatment was carried out as detailed by Burghaus and Lingelbach (2001). Briefly, 10% saponin (grade pure, Serva) in ice-cold phosphate-buffered saline (PBS) was added to 2–3 × 108 infected red blood cells to a final concentration ranging from 0.15% to 0.8% in a total volume of 100 µl. Samples were incubated on ice for 10 min, diluted with 50 µl of PBS and centrifuged immediately at 16 500 g for 25 min at 8°C. Supernatants were transferred to a fresh tube containing 25× Complete™ protease inhibitor mixture (Sigma). Pellets were washed three times with PBS. Both fractions were stored at −20°C until further processed. Pellets were thawed in two volumes of lysis buffer (25 mM Tris-phosphate, pH 7.8, 2 mM dithiothreitol, 2 mM EDTA, 10% glycerol and 1% Triton X-100) containing Complete™ protease inhibitor mixture (Sigma), incubated for 15 min at room temperature and centrifuged for at 16 500 g for 15 min at 8°C. The clear lysates were kept on ice until luciferase activity was determined.

Detection of luciferase activity

Reagents for the luciferase assay were purchased from Promega, and a Lumat type LB9501 (Berthold) was used to measure the enzyme activity as relative light units. Recombinant Photinus pyralis luciferase produced in Escherichia coli was purchased from Promega. Luciferase activity of pellet and supernatant fractions was determined and standard curves were measured as described previously (Burghaus and Lingelbach, 2001). Briefly, 100 µl of substrate solution was automatically injected into a 10 µl aliquot of the sample and chemiluminescence measured for 60 s. Either lysis buffer or the supernatant of the saponin lysis of infected red blood cells of the mock transfection was used to generate the standard curves of relative light units versus luciferase concentration for pellet or supernatant fractions respectively. For each sample the enzyme activity was measured in triplicate and the mean value calculated. The total amount of luciferase contained in each fraction was determined and its distribution between pellet and supernatant is presented in per cent.

Immunoaffinity purification of the PfCDPK1-specific antibodies

PfCDPK1 was purified as described (Zhao et al., 1994b). PfCDPK1  storage  puffer  was  changed  against  coupling buffer (0.1 M NaHCO3, pH 8.3, 0.5 M NaCl) on a HR10/10 desalting column (Pharmacia). Coupling to CNBr-activated Sepharose® 6MB (Pharmacia) was carried out according to the manufacturers instructions. Antibodies specifically bound to PfCDPK1 beads were eluted in 100 mM glycine, pH 2.5, neutralized by the addition of 1 M Tris-HCl, pH 8.8, dialysed against PBS and concentrated.

Immunogold electron microscopy

Parasite cultures of P. falciparum isolate K1 with 10% parasitemia were fixed in 3% paraformaldehyde and 0.5% glutaraldehyde in PBS puffer (pH 7.2) for 1 h at room temperature. After two washes in PBS, the pellet was enclosed into 2% agarose dissolved in H2O. The agar blocks were dissected into smaller pieces and washed in PBS twice. After dehydration with graded ethanol at decreasing temperature, the agarose particles were embedded in Lowicryl K4M according to the protocol supplied (Chemische Werke Lowi, FRG). Sections (50 nm) were obtained by cutting with a diamond knife followed by mounting on carbon-coated, formvar-supported copper grids.

The specimens were blocked for 15 min in PBS buffer containing 1% non-fat dry milk powder. The grids were then incubated for 2 h with a 1:5 dilution of the affinity-purified PfCDPK1-specific antibody 2129 in the above-mentioned blocking puffer followed by five washes in PBS containing 0.01% Tween 20. The antigen–antibody complexes were labelled for 1 h with protein A-coupled gold particles (15 nm; purchased from the Department of Cell Biology, Utrecht University) suspended in blocking puffer according to the instructions of the suppliers. After three washes in PBS containing 0.01% Tween 20 for 5 min each, the samples were fixed for 10 min in 1% glutaraldehyde in PBS followed by several washes in PBS. After staining with 2% uranylacetate (10 min) and 2% lead citrate (1 min), the samples were examined by electron microscopy.

Polymerase chain reaction (PCR)-based mutagenesis

Two oligonucleotides were designed to replace or delete amino acid residues potentially involved in membrane attachment of PfCDPK1. The putative myristoylation site, Gly2, was changed to valine with the VG2 primer (5′-AGAACTAGTGGA TCCATGGTGTGTTCACAAAGTTCAAACGTGAAAGATT TC-3′). The supposed palmitoylation site, Cys3, was replaced by an alanine with the AC3 primer (5′-ACTAGTGGATCCATG GGGGCTTCACAAAGTTCAAACGTGAAAGATTTC-3′). The nucleotides that were exchanged to obtain the desired mutation are underlined and marked in bold. The basic cluster close to the N-terminus (10KDFKTRRSKFT20) involved in the assumed ‘myristoyl plus basic’ binding was deleted with the following oligonucleotide: 5′-GGGTGTTCACAAAGTTCAA ACGTGAATGGAAATAATTATGGGAAAAGT-3′. Nucleotides which mark the border of the deletion (Δ10-20) are underlined and in bold letters. The DM was generated from the Δ10-20 mutant using the VG2-specific primers and the TM from DM with the AC3-specific primers. The replacement of the original amino acids with the desired ones and the deletion of the basic cluster were performed by an in vitro site-directed mutagenesis method according to Gilberger et al. (1997). All mutants were analysed performing the Sanger dideoxy chain termination reaction for double-stranded DNA (Sambrook et al., 1988). One hundred per cent of the colonies contained the desired mutation in the absence of other unwanted mutations.

Luciferase and GFP expression constructs

Luciferase constructs.

The luciferase gene was amplified from the pHx1luc vector (Burghaus and Lingelbach, 2001) and subcloned into pBS II SK+. To generate C-terminal luciferase fusions with PfCDPK1 and the mutants VG2, Δ10-20 and TM, new restriction sites had to be introduced at the 5′ and 3′ ends of the PfCDPK1 and luciferase gene by polymerase chain reaction (PCR) amplification. XbaI and XhoI were introduced in front of the NcoI site at the 5′ end and BamHI at the 3′ end of PfCDPK1. At the same time the stop codon was removed. Accordingly, a BamHI site was introduced at the 5′ end of the luciferase gene and a XhoI site at its 3′ end. A primary luciferase fusion with PfCDPK1 was made by inserting a NotI/BamHI digested PfCDPK1-wt fragment in front of the luciferase gene in pBS II SK+. The fusion constructs of the mutants were generated by replacing a NcoI/PpuMI 5′ fragment of the PfCDPK1-wt–luciferase fusion construct with the respective fragment of the mutant clones. The PfCDPK1- and mutant-luciferase genes were then excised and inserted as a XhoI fragment into the P. falciparum expression vector pHC1 (Crabb et al., 1997).

GFP constructs.

The GFP gene was kindly provided by Jude Przyborski and the P. falciparum expression vector pARL-1a+ by Tim Gilberger and Alan Cowman (Crabb et al., 2004). The strategy for the construction of the PfCDPK1-wt- and mutant-GFP constructs was in principal similar to the one used for the generation of the luciferase fusion constructs in that XhoI sites were introduced at the 5′ ends of PfCDPK1 and the mutant genes and at the 3′ end of GFP. The fusion of the two partners occurred via BamHI sites at the 3′ end of the PfCDPK1 and mutant constructs and at the 5′ end of GFP. We were, however, not able to obtain transfectants with the full-length genes of either PfCDPK1 or its mutants, supposably because overexpression of the active kinase harms the parasite. For this reason, the N-terminal 48 amino acids preceding the protein kinase catalytic domain further on referred to as ‘signal’ were used for the second generation of PfCDPK1-wt- and mutant-GFP constructs. PCR amplification was used to generate this shortened versions of PfCDPK1-wt and its mutants and to introduce XhoI and BamHI sites at their 5′ and 3′ ends respectively. The fusion constructs were cloned into the XhoI restricted pARL-1a+.

Metabolic labelling and immunoprecipitation

[9,10(n)-3H]myristic acid (10–60 Ci mmol−1; New England Nuclear) in ethanol was allowed to dry on a tissue culture plate surface in quantities that brings the concentration of the isotope to 200 µCi ml−1 after addition of 5 ml of parasite culture (10% haematocrit). Labelling was performed at 37°C for 4 h. Labelled parasites were isolated by saponin lysis as described by Suetterlin et al. (1992). The parasites were then lysed in an appropriate volume of RIPA buffer [50 mM Tris-HCl pH 8, 150 mM NaCl, 0.5% sodium deoxycholate, 50 mM NaF, 0.1% SDS, 1% Triton X-100, 5 mM EDTA supplemented with freshly added DTT (1 mM), PMSF (phenylmethylsulphonyl fluoride; 1 mM), pepstatin A (1 µg ml−1) and leupeptin (1 µg ml−1)]. The lysate was cleared twice by centrifugation at 15 000 g at 4°C for 15 min. To immunoprecipitate the radiolabelled PfCDPK1, the cleared lysate was incubated with 10 µl of the affinity-purified antibody 2129 or antibody 2979 coupled to protein A beads (Affi-Prep® protein A matrix, Bio-Rad). The beads were washed five times with 1 ml of RIPA buffer, suspended in 5× Laemmli sample buffer and boiled for 5 min. Samples were analysed on an 8% SDS-PAGE. Gels were stained with Coomassie blue, destained with acetic acid and methanol, treated for fluorography (Amplify; Amersham), dried and exposed to Hyperfilm-MP (Amersham) for 3 months.

In vitro transcription/translation and membrane binding assays

The cDNAs encoding either wild-type or the mutant PfCDPK1 in pGEM®-3Z (0.4 µg) were in vitro transcribed and translated in a coupled reticulocyte lysate system (TNT Quick Coupled Transcription/Translation Systems, Promega). Transcription used the T7 promoter upstream of the PfCDPK1 cDNA. Radiolabelling of the products of the transcription/translation reaction was carried out in a 25 µl reaction in the presence of either 20 µCi [9, 10(n)-3H]myristic acid (10–60 Ci mmol−1; New England Nuclear),  20 µCi [9,10(n)-3H]palmitic  acid (30–60 Ci mmol−1; New England Nuclear) or 17 µCi l-[35S]-methionine (1000 Ci mmol−1; Amersham) at 30°C. [3H]-labelled myristic or palmitic acid were previously dried and resuspended in DEPC-treated water. Membrane binding assays were performed in the presence of 1.4 µl of canine pancreatic microsomal membranes (Promega). After the in vitro transcription/translation reaction, microsomal membranes were pelleted from the supernatant by ultracentrifugation for 1 h at 100 000 g. The pelleted fraction was washed twice in PBS and resuspended in sample buffer. All procedures were performed at 4°C. Labelled translation products were analysed on 8% SDS-PAGE (Sambrook, 1988). Gels were stained with Coomassie blue, destained with acetic acid and methanol, treated for fluorography (Amplify; Amersham), dried and exposed to Hyperfilm-MP (Amersham) for 14 days for the myristate and 3 months for the palmitate labelling experiment and to Kodak X-Omat AR-5 for 1–2 h for the methionine labelling.


After the in vitro transcription/translation reaction, microsomal membranes were pelleted as described above and washed twice in 1× kinase assay buffer (20 mM Tris-HCl, pH 7.4, 10 mM MgCl2). Kinase assays of the particulate fraction were performed by 30 min incubations at 30°C in a final volume of 25 µl of a mixture containing 20 mM Tris-HCl, pH 7.4, 10 mM MgCl2, 1 mM DTT, 88 µM ATP and 1 mM Ca2+ or 100 µM EGTA in the presence or absence of 50 ng of the recombinant kinase respectively. The recombinant PfCDPK1 used in the assay had a specific activity of 279 nmol min−1 mg−1 protein with casein as substrate.


We thank Kirsten Plückhahn for technical support, Dr Sylke Müller and Dr Ralph Graeser for technical advice, Dr Gilberger and Professor Cowman for the generous gift of pARL-1a+, Jude Przyborski and Yvonne Kuhn for the helpful exchanges of experiences concerning the transfection technologies and Dr Sylke Müller as well for critical reading of the manuscript. The work was supported in part by funds of the Swiss National Foundation (Grant 31-39122.93) and the Deutsche Forschungsgemeinschaft (DFG-project KA  1491/2-1) and includes parts of the doctoral thesis of Christian Möskes. Christian Möskes was in part supported by a PhD scholarship programme ‘Pathogenic microorganisms: molecular mechanisms and genome’ of the Deutsche Forschungsgemeinschaft.