Biochemical Society Transactions

Recent Advances in Membrane Biochemistry

The Plasmodium falciparum Ca2+-ATPase PfATP6: insensitive to artemisinin, but a potential drug target

Bertrand Arnou, Cédric Montigny, Jens Preben Morth, Poul Nissen, Christine Jaxel, Jesper V. Møller, Marc le Maire

Abstract

The disease malaria, caused by the parasite Plasmodium falciparum, remains one of the most important causes of morbidity and mortality in sub-Saharan Africa. In the absence of an efficient vaccine, the medical treatment of malaria is dependent on the use of drugs. Since artemisinin is a powerful anti-malarial drug which has been proposed to target a particular Ca2+-ATPase (PfATP6) in the parasite, it has been important to characterize the molecular properties of this enzyme. PfATP6 is a 139 kDa protein composed of 1228 amino acids with a 39% overall identity with rabbit SERCA1a (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase 1a). PfATP6 conserves all sequences and motifs that are important for the function and/or structure of a SERCA, such as two high-affinity Ca2+-binding sites, a nucleotide-binding site and a phosphorylation site. We have been successful in isolating PfATP6 after heterologous expression in yeast and affinity chromatography in a pure, active and stable detergent-solubilized form. With this preparation, we have characterized and compared with the eukaryotic SERCA1a isoform the substrate (Ca2+ and ATP) -dependency for PfATP6 activity as well as the specific inhibition/interaction of the protein with drugs. Our data fully confirm that PfATP6 is a SERCA, but with a distinct pharmacological profile: compared with SERCA1a, it has a lower affinity for thapsigargin and much higher affinity for cyclopiazonic acid. On the other hand, we were not able to demonstrate any inhibition by artemisinin and were also not able to monitor any binding of the drug to the isolated enzyme. Thus it is unlikely that PfATP6 plays an important role as a target for artemisinin in the parasite P. falciparum.

  • artemisinin
  • Ca2+-ATPase
  • malaria
  • Plasmodium falciparum Ca2+-ATPase (PfATP6)
  • P-type ATPase
  • sarcoplasmic/endoplasmic reticulum Ca2+-ATPase (SERCA)

Background

Artemisinin-based therapies are of utmost importance in the treatment of human malaria caused by the eukaryotic protist Plasmodium falciparum, one of the most widespread diseases on Earth and a leading cause of death. Artemisinin is a sesquiterperne trioxane lactone, containing an endoperoxide bridge, that was originally extracted from the plant Artemisia annua, a herb described in Chinese traditional medicine. Over the last few decades, several semi-synthetic derivatives and analogues have been developed in order to increase the solubility and the half-life of the drug, to improve therapeutic efficiency. The activity of artemisinin and its derivatives is directly related to their endoperoxide bridge, since deoxyartemisinin, a derivative in which a single oxygen replaces the endoperoxide bridge, is biologically inactive [1].

Some years ago, the hypothesis that a SERCA (sarcoplasmic/endoplasmic reticulum Ca2+-ATPase) from P. falciparum, PfATP6, might act as an important target for artemisinin was proposed on the basis of biochemical assays performed on oocytes heterologously expressing PfATP6 and on parasite cultures. Ca2+-ATPase activity attributed to PfATP6 in oocyte membranes was found to be inhibited by artemisinin, and thapsigargin, a specific inhibitor of SERCA-type proteins, was described as antagonizing the parasiticidal activity of artemisinin [2]. Moreover, mutation studies performed with the same expression system suggested that the sensitivity of PfATP6 to artemisinin was critically dependent on Leu263 in predicted transmembrane segment, TM3, at the putative thapsigargin-binding site [3]. Thus the mutation L263E (leucine corresponding to the orthologous Glu255 in mammalian SERCA1a) was found to abolish artemisinin inhibition, whereas the reciprocal E255L replacement on SERCA1a was found to transform SERCA1a into an artemisinin-sensitive ATPase.

Elsewhere, recent experimental evidence does not support PfATP6 as an important target of artemisinins in P. falciparum. Even if it has been noticed that there is correlation between the occurrence of certain point mutations in PfATP6 in field isolates and reduced in vitro sensitivity to artemisinins [4,5], artemisinin resistance is not strictly related to these mutations but revealed some polymorphism [69]. Furthermore, biochemical data obtained with the purified enzyme after yeast expression showed that Ca2+-dependent activity of PfATP6 was not inhibited by artemisinins [10]. The same was true for the E255L SERCA1a mutant: irrespective of whether it had been expressed in COS cells or purified after yeast expression, no inhibition of Ca2+-dependent activity by artemisinin was noticed [10]. Furthermore, the L263E PfATP6 mutant, assumed to be artemisinin-insensitive, gave no significant difference with wild-type ATPase in response to artemisinin or its derivatives when introduced into P. falciparum by allelic exchange [11].

In the present paper, we describe the comparison of the properties of PfATP6 and the eukaryotic SERCA1a isoform, both with respect to their substrate-dependence and susceptibility to pharmacological inhibition. We find that, whereas there are quantitative differences in the susceptibility to inhibitors such as CPA (cyclopiazonic acid) and thapsigargin, the essential characteristics of the two enzymes are the same and do not include sensitivity to artemisinin.

PfATP6, a SERCA-type pump

SERCA-type pumps are essential membrane proteins involved in cellular Ca2+ homoeostasis, controlling an impressive and varied array of cellular functions. Unlike vertebrates that possess three SERCA genes [12], in P. falciparum the genome contains a single SERCA gene, originally described in 1993 [13]. The coding region of the gene consists of 3684 nucleotides and corresponds to a protein with a molecular mass of 139.4 kDa. From amino acid sequence alignments, it is clear that PfATP6 conserves all of the motifs and residues that are important for the function and/or structure of a SERCA pump such as two high-affinity Ca2+-binding sites, a nucleotide-binding site and a phosphorylation site [13]. Moreover, PfATP6 shares with SERCA1a present in skeletal muscle, serving as the paradigm of SERCA-type Ca2+-ATPases [14], an overall 39% sequence identity that is increased to 53% if the sequences corresponding to the N- (nucleotide-binding) domain are excluded. Thus it can be concluded with confidence that PfATP6 folds the same way as does SERCA1a, with a membrane domain containing ten transmembrane helices and a cytosolic part with well-defined catalytic domains (Figure 1). In summary, PfATP6 possesses all of the characteristic features of the catalytic unit of the P-type ATPase family in which SERCA pumps are assembled in a subgroup IIA according to their specificity for Ca2+ transport across sarcoplasmic/endoplasmic reticulum membranes.

Figure 1 Comparison of SERCA1a and PfATP6 structures

The structure of SERCA1a in E2 MgFl2 complex with bound CPA and AMPPCP (A) (PDB code 3FGO [23]) and the homology model of PfATP6 in the same state (B) [23] are shown as cartoon representation. A- (actuator) domain is yellow, the N-domain is red, the P- (phoshorylation) domain is blue and the membrane domain is wheat. Unique amino acid sequence features found only in PfATP6 are depicted as black loops with random coil, since no corresponding template structures were found to process homology modelling. Structural representations were prepared using PyMOL software (http://pymol.sourceforge.org/).

The main difference between PfATP6 and SERCA1a resides in their N-domains: whereas the core of the domain is remarkably well conserved, especially with regard to the nucleotide-binding region, there are, in the sequence of PfATP6, 200 additional residues in this domain (Figure 1B). Those insertions, for which the fold remains unknown, are not found in any other P-type ATPases, making PfATP6 a distinctive pump. Moreover, the sequences of the insertions contain an abnormally high frequency of asparagine residue plus two polyasparagine motifs (nine and ten homopolymeric runs). The presence of such extraordinary stretches is a characteristic feature of the proteins present in the parasite, compared with most eukaryotic ones, thus orthologous proteins can be up to 50% longer in P. falciparum as compared with yeast for example [15]. Those Plasmodium-specific inserts, are predicted to form non-globular structures oriented towards external surfaces of proteins, as shown in Figure 1(B), that probably do not interfere with the function of the rest of the proteins [15]. However, even if their exact functional role remains a mystery, it has been postulated that the inserts could facilitate the immune evasion by the parasite by generating a non-productive antibody response against the asparagine-rich regions of Plasmodium proteins [16]. Thus we believe that PfATP6 could harbour, in its N-domain, unstructured and solvent-exposed regions that do not interfere in any important way with enzymatic activity; however, the possibility cannot be excluded that such a large insert could also be combined to form a peculiar domain with an unknown function.

Protein production, purification and enzymatic properties

We prepared PfATP6 by heterologous overexpression in the yeast Saccharomyces cerevisiae as a fusion protein, with a BAD (biotin-acceptor domain) linked to the C-terminus by a thrombin cleavage site and its purification by affinity chromatography as described previously [10,17]. Briefly, yeast cells were first cultured for 36 h at 28°C under high aerobic conditions. After that, expression of recombinant protein was induced by the addition of galactose and stopped after 18 h of culture under anaerobic conditions. Then yeast cells were harvested, broken with glass beads and an LM (light membrane) fraction was isolated by differential centrifugation. The LM fraction was washed twice to remove soluble biotinylated contaminants and was solubilized with DDM (dodecyl maltoside) at a ratio of 3:1 (w/w) between DDM and total protein in the presence of 20% glycerol together with 1 mM CaCl2 as required to retain full enzymatic activity of PfATP6. After high-speed centrifugation, the solubilizate was loaded on to a streptavidin–Sepharose resin, and contaminants were washed away, and, after thrombin cleavage, PfATP6 devoid of BAD tag was eluted from the resin as pure protein (Figure 2A). The final sample was subsequently concentrated on a filter with a 30000 Da molecular-mass cut-off, flash-cooled in liquid nitrogen and stored at −80°C. At this final stage, the purified PfATP6 (~0.4 mg/ml) was present in a medium composed of 36 mM KCl, 36 mM Mops/Tris (pH 7), 1.8 mM CaCl2, 40% glycerol and 1.3 mg/ml DDM.

Figure 2 Properties of solubilized PfATP6 and SERCA1a

(A) SDS/PAGE followed by Coomassie Blue staining: 2 μg of purified PfATP6 (PA6), 2 μg of SERCA1a (S1a) preparation and molecular-mass marker (M, sizes are in kDa) were loaded. (B) Ca2+-dependent ATPase activity measured using a coupled-enzyme assay. Reaction medium contained 50 mM Tes/Tris (pH 7.5), 0.1 M KCl, 1 mM MgCl2, 20% (v/v) glycerol, 0.17 mM NADH, 1 mM PEP, 0.05 mg/m; LDH, 0.083 mg/ml PK, 0.1 mM Ca2+ and 1:0.25 C12E8/DOPC at 23°C. Reactions were triggered by adding protein (0.004 mg/ml) and stopped by the addition of 500 μM EGTA. Absorbance at 340 nm (maximum absorption for NADH) was recorded as a function of time, and the difference between the slopes obtained before and after addition of EGTA was considered to be due to the Ca2+-ATPase activity. The results are plotted as percentage of maximal activity as a function of MgATP concentration. (C) Binding of Ca2+ to PfATP6 and SERCA1a as deduced from tryptophan fluorescence changes. Fluorescence was measured with a Fluorolog-3 spectrofluorimeter with excitation and emission wavelengths set at 290 and 330 nm respectively (bandwidths were 5 and 10 nm respectively). Reaction medium contains 50 mM Mops/Tris (pH 7), 0.1 M KCl, 0.75 mM EGTA, 70 μM Ca2+, 20% (v/v) glycerol, 1:0.25 C12E8/DOPC and 0.01 mg/ml protein. For each experiment, the initial [Ca2+]free was approximately 0.04 μM, and the fluorescence intensity at this step was arbitrarily taken as 0%. Then, extra total Ca2+ was increased stepwise to 370, 670, 770, 870, 1170, 1470 and 2470 μM, resulting in [Ca2+]free of 0.3, 3, 30, 120, 420, 720 and 1720 μM respectively. Changes in tryptophan fluorescence as a function of free Ca2+ were corrected for the intensity drift, and plotted as percentage values after normalization to 100% of the total change in fluorescence and normalization for their (opposite) sign. Maxchelator software (http://maxchelator.stanford.edu/) was used to calculate [Ca2+]free from total Ca2+ and EGTA concentrations.

SERCA1a was prepared from sarcoplasmic reticulum vesicles isolated from rabbit skeletal muscle and purified by extraction with a low concentration of sodium deoxycholate as described previously [18] (Figure 2A). The purified membranes were solubilized in 36 mM KCl, 36 mM Mops/Tris (pH 7), 1.8 mM CaCl2, 40% glycerol and 1.3 mg/ml DDM at 0.4 mg/ml protein concentration before use in order to resemble the conditions used for studying the yeast-expressed and purified PfATP6.

The enzymatic properties of the purified proteins have been examined using a spectrophotometric enzyme-coupled ATPase activity assay as described previously [19]. During turnover, the ATPase alternates between two major conformations: ‘E1’, which presents high-affinity binding sites for Ca2+ towards the cytosolic space, and ‘E2’ with a low affinity for Ca2+. A specific aspartate residue can be phosphorylated after binding of Ca2+ and reaction with ATP to form an energy-rich aspartyl-phosphorylated intermediate. Phosphorylation and subsequent dephosphorylation drive the active transport of cations (for a review, see [14]). The enzymatic assay is based on the conversion of PEP (phosphoenolpyruvate) into pyruvate by PK (pyruvate kinase), using ADP (produced by the ATPase from ATP) as a substrate. This reaction is coupled to the conversion of pyruvate into lactate by LDH (lactate dehydrogenase). The latter step requires NADH which is oxidized to NAD+. Then, because NADH absorbs at 340 nm and NAD+ does not, the decrease in the A340 can be converted into ATPase activity, taking into account that 1 molecule of NADH oxidized corresponds to the hydrolysis of 1 molecule of ATP by the ATPase. We carried out the enzyme-coupled assay in the presence of 20% glycerol together with 1:0.25 C12E8 [dodecyl octa(ethylene glycol) monoether]/DOPC (dioleyl phosphatidylcholine) since the presence of phospholipid is a requirement for maintaining a stable ATP hydrolysis rate by solubilized PfATP6 [10]. Under those conditions, we recorded the specific activity of detergent-solubilized PfATP6 and SERCA1a as a function of ATP concentration. In these experiments, PfATP6 is activated by MgATP in a manner that can be described by two phases. In the first phase (0–100 μM MgATP), activity rises very rapidly as a function of MgATP concentration, whereas at higher MgATP concentrations (0.1–5 mM), there is a second phase with a much slower rise (Figure 2B). This behaviour is very similar to the activation by MgATP of SERCA1a under our experimental conditions (Figure 2B), and it is also largely in agreement with previous observations [20]. At 5 mM MgATP, we recorded a maximal activity of 1.5 μmol of hydrolysed ATP per min per mg of enzyme for PfATP6 and 7 μmol of hydrolysed ATP per min per mg of enzyme for SERCA1a. Thus, even if the two enzymes have similar properties in terms of ATP concentration-dependence, PfATP6 seems to be a slower enzyme compared with SERCA1a.

We also measured, by intrinsic tryptophan fluorescence, the Ca2+-binding affinity for PfATP6 and SERCA1a under the same solubilizing conditions as described above for the activity measurements (i.e. 20% glycerol together with 1:0.25 C12E8/DOPC). This method, optimized for detergent-solubilized recombinant Ca2+-ATPases [21], is often used to monitor E1/E2 structural changes in SERCA1a in response to Ca2+ binding. We took advantage of the specific signal induced by Ca2+ binding to PfATP6 or SERCA1a to titrate Ca2+ binding as a function of the concentration of free Ca2+. As can be seen from Figure 2(C), the Ca2+-binding profile of PfATP6 fits pretty well with that of SERCA1a. Under our experimental conditions (see also [21]), both enzymes are characterized by a half-maximal Ca2+ concentration of ~10 μM and a Hill coefficient close to 1.8, indicating that binding of the two Ca2+ ions is a highly co-operative process. Thus PfATP6 and SERCA1a also exhibit very high similarities regarding Ca2+ binding, and the lower specific activity of PfATP6, as compared with SERCA1a, cannot be ascribed to a different mode of Ca2+ binding or to a reduced Ca2+ affinity, rather it is certainly the result of other intrinsic properties of the protein.

Drug inhibition

We tested first the effect of three specific inhibitors of mammalian SERCA proteins, namely CPA, BHQ [2,5-di-(t-butyl)-1,4-hydroquinone] and thapsigargin on Ca2+-dependent ATPase activities. Under our experimental conditions, IC50 values for detergent-solubilized SERCA1a were ~10 μM for CPA, ~25 μM for BHQ and ~0.05 μM for thapsigargin as deduced from Figures 3(A)–3(C) (closed squares and dotted lines). For PfATP6, the IC50 value for BHQ is slightly higher (~65 μM) compared with SERCA1a (Figure 3B), whereas the IC50 value for CPA is 25 times lower (~0.4 μM, Figure 3A) and for thapsigargin more than 3000 times higher (>150 μM, Figure 3C) on PfATP6 as compared with SERCA1a. In other words, BHQ has comparable inhibitory effects on PfATP6 and SERCA1a, whereas CPA inhibition is much more effective on PfATP6, and thapsigargin only inhibits PfATP6 at such high concentrations (≥150 μM) that do not allow us to determine properly the IC50 because of solubility problems. Although PfATP6, under our experimental conditions, qualitatively behaves the same way as a mammalian SERCA-type enzyme, there are notable differences as compared with SERCA1a.

Figure 3 Effect of inhibitors on PfATP6 and SERCA1a Ca2+-dependent ATPase activity

Inhibitors were first pre-incubated for 5 min at 23°C together with the protein diluted to 0.004 mg/ml in a reaction buffer containing 50 mM Tes/Tris (pH 7.5), 0.1 M KCl, 0.1 mM Ca2+, 1 mM MgCl2, 20% (v/v) glycerol, 0.17 mM NADH, 1 mM PEP, 0.05 mg/ml LDH, 0.083 mg/ml PK and 1:0.25 C12E8/DOPC at 23°C. Reactions were triggered by the addition of 5 mM MgATP and stopped by the addition of EGTA to a final concentration of 500 μM. Absorbance at 340 nm was recorded as a function of time, and the difference between the slopes obtained before and after addition of EGTA was considered to be due to the Ca2+-ATPase activity. Results are percentages of maximal activity as a function of inhibitor concentration: (A) CPA; (B) BHQ; (C) thapsigargin (Tg); (D) artemisinin. Because inhibitor was added from a stock solution dissolved in DMSO, resulting in a final DMSO concentration of 1% (v/v), the maximal 100% Ca2+-ATPase activity was determined when the same volume of DMSO was added without inhibitor.

Subsequently, we tested the effect of artemisinin on PfATP6 Ca2+-dependent ATPase activity. As can be seen, artemisinin had no significant inhibitory effect on PfATP6 enzymatic activity even at concentrations as high as 500 μM (Figure 3D), whereas it was reported previously to be completely inhibited by 10 μM artemisinin [2,3]. The same inability to inhibit PfATP6 activity was also observed under a few other experimental conditions, such as in the presence of Fe2+, which has been suggested to be required for artemisinin to be active [2], and under detergent-solubilizing conditions or on DOPC-reconstituted PfATP6, and also for several artemisinin derivatives such as artemisone, dihydroartemisinin or artesunate [10]. Thus the enzymatic activity of the isolated PfATP6, in the same manner as is the case for SERCA1a, was not affected by artemisinin and its derivatives [10].

Drug binding

In addition to activity inhibition assays that relate the ability of molecules to stop the enzymes during turnover, we used intrinsic fluorescence measurements as a tool to examine the binding of drugs by the two ATPases. Initially, CPA has been used to calibrate the method, not only because it was already reported to give a specific signal by binding to SERCA1a [22], but also because it inhibits efficiently both PfATP6 and SERCA1a activity (BHQ could not be used because it gives a large unspecific signal in our experimental conditions; results not shown). Furthermore, homology modelling based on the CPA-bound structure of SERCA1a indicates that the CPA-binding site is conserved in PfATP6 [23]. The experimental set-up is the following: we first start the measurement with the enzyme in an E1 Ca2+-bound state (i.e. in the presence of 100 μM Ca2+), then bring it to E2 Ca2+-free state by adding 4 mM EGTA ([Ca2+]free <1 nM) and successively add in four steps the same concentration of inhibitor (SERCA-type inhibitors are known to bind to Ca2+-free states; for a review, see [24]). It should be noticed that the transition from Ca2+-bound to Ca2+-free state induces a fluorescence signal of PfATP6 opposite in sign to SERCA1a (compare Figures 4A, 4C and 4E with Figures 4B, 4D and 4F, EGTA addition). This certainly reflects the non-conserved position of tryptophan residues along the polypeptide chain of the two ATPases, which results in an individually different tryptophan fluorescence response. Then, proceeding to the first addition, CPA induces a fall in tryptophan fluorescence, whereas the three following ones give no deviation from the intensity drift. The fall in fluorescence induced by the first addition of CPA (0.2 μM) specifically reports on a conformational change in response to inhibitor binding by both SERCA1a and PfATP6 (Figures 4A and 4B respectively). At this point, we can also notice that 0.2 μM is enough to saturate CPA binding on both Ca2+ pumps, whereas higher concentrations were required to totally inhibit their Ca2+-dependent ATPase activity (~10 and ~200 μM for PfATP6 and SERCA1a respectively, Figure 3A). This means that PfATP6 affinity for CPA is much higher for the E2 Ca2+-free state as compared with enzyme during turnover, confirming that this inhibitor binds preferably to the Ca2+-free form of the PfATP6 pump as observed previously for SERCA1a [25].

Figure 4 Inhibitor binding on PfATP6 and SERCA1a

The intrinsic fluorescence of SERCA1a (A, C and E) and PfATP6 (B, D and F) was recorded at 0.01 mg/ml protein in a temperature-regulated (20°C) and continuously stirred 2 ml cuvette. Fluorescence was measured with a Fluorolog-3 spectrofluorimeter with excitation and emission wavelengths set at 290 and 330 nm respectively (bandwidths were 5 and 10 nm respectively). The medium used for each experiment was 50 mM Tes/Tris (pH 7.5), 0.1 M KCl, 5 mM MgCl2, 20% (v/v) glycerol, 0.1 mM Ca2+ and 1:0.25 C12E8/DOPC. Sequential additions were made, indicated by arrows: EGTA (4 mM) in all panels, 0.2 μM CPA in (A) and (B), 15 μM thapsigargin (Tg) in (C) and (D) and 10 μM artemisinin (ART) in (E) and (F).

For SERCA1a, thapsigargin behaves qualitatively very similarly to CPA in terms of tryptophan fluorescence response, with a somewhat lower amplitude for the binding-specific signal (Figure 4C). PfATP6, meanwhile, gives a very weak signal in response to the first 15 μM thapsigargin addition, and the next additions do not give any observable changes in fluorescence (Figure 4D). Thus, even if the amplitude is weak, the specific signal induced by thapsigargin indicates the occurrence of a conformational change related to binding of the drug on PfATP6. However, artemisinin does not induce any specific changes on intrinsic tryptophan fluorescence either on SERCA1a or on PfATP6 in Ca2+-free state (Figures 4E and 4F respectively). Considering the fact that artemisinin also had no effect on ATPase activity, we can assume that artemisinin probably does not bind on PfATP6 with high affinity.

Discussion

In the present paper, we have described the relationship between artemisinin and the isolated PfATP6 examined in detail using biophysical and enzymatic approaches. The expression and purification methods that we set up some years ago for SERCA1a [17,26] already paved the way to study mammalian membrane proteins after heterologous expression in S. cerevisiae and affinity purification taking advantage of the BAD tag. Using similar conditions, we were able to extend the process for expressing and for isolating the plasmodial PfATP6 with a high degree of purity in an active and stable form [10] (Figures 2A and 2B). The purified PfATP6 reveals Ca2+-binding properties (Figure 2C), and Ca2+- or pH-dependence for activity [10], highly similar to that of mammalian SERCA1a, but with a slower turnover. In the same way as the mammalian SERCA isoforms, with different C-terminal tails, show differences in their pumping velocity related to specific sequence differences [27], the lower activity of PfATP6 compared with SERCA1a could be induced by the ten-residue extension at its C-terminus (Figure 1). One could also envisage that the long inserts in the N- domain, by restricting the mobility of the domain, slow down the conformational changes occurring during the reaction cycle and thereby induce a lower activity. Nonetheless, under our experimental conditions, PfATP6 behaves like a SERCA-type Ca2+ pump with characteristics comparable with the mammalian isoforms.

Having overcome the difficulties of heterologous expression and purification of a membrane protein, we used our PfATP6 preparation to characterize the inhibition profile of the pump. Interestingly, we found that CPA is a very powerful inhibitor of the Plasmodium Ca2+-ATPase with an IC50 value in the submicromolar range (~0.4 μM) under our experimental conditions. BHQ and thapsigargin, meanwhile, have IC50 values for PfATP6 within a much higher concentration range (~65 and >150 μM respectively). Moreover, we demonstrated that it is possible to monitor structural changes induced by drug binding on PfATP6 using tryptophan fluorescence measurement techniques. Thereby, we can conclude from our results that CPA and thapsigargin stabilize the enzyme in a dead-end conformation in an E2-like state, in a similar way as they do with SERCA1a. Thus PfATP6 is sensitive to SERCA-type inhibitors with somewhat pronounced quantitative differences as compared with SERCA1a. The residues in immediate proximity to thapsigargin and that from the neck of the thapsigargin-binding cleft found in SERCA1a reveal both substantial and conservative differences in chemical properties of the residue side chains compared with the equivalent binding cleft in PfATP6, based on sequence alignment with SERCA1a (Figure 5A and Table 1). Thus the drastically diminished sensitivity for thapsigargin is likely to be the result of these modifications to the binding site. On the other hand, the BHQ/CPA-binding pocket is remarkably well conserved (Figures 5B and 5C, and Table 1), in agreement with the conserved inhibitory effect of the two drugs. However, the reason the IC50 value for CPA is significantly lower for PfATP6 compared with SERCA1a is more complex and might be the result of a favoured access of the molecule through the channel linking the surface of the protein to the binding site. These particularities highlight an interesting pharmacological profile of the Plasmodium Ca2+ pump.

Figure 5 Residues involved in inhibitor binding on SERCA1a and conservation in PfATP6

Surface representation of SERCA1a-binding sites for (A) thapsigargin (PDB code 2AGV [34]), (B) CPA (PDB code 3FGO [23]) and (C) BHQ (PDB code 2AGV [34]). Inhibitors are depicted as sticks with carbon atoms coloured grey, nitrogen coloured blue and oxygen coloured red. Conservative changes between SERCA1a and PfATP6 (isoleucine/leucine/valine) are indicated in green and substantial changes are indicated in red (see also Table 1). Black labels indicate the SERCA1a residues, and white labels indicate the equivalent PfATP6 residues.

View this table:
Table 1 Conservation of residues in the CPA/BHQ and thapsigargin drug-binding pocket

Binding pockets of SERCA1a for the inhibitors CPA, BHQ and thapsigargin compared with the equivalent residues in PfATP6. Residues within 4 Å (1 Å = 0.1 nm) of the inhibitors are listed with conserved residues in normal type, conservative changes in bold (isoleucine/valine/leucine) and other changes in italic.

However, contrary to assumptions arising from previous studies in the Xenopus oocyte membranes (Krishna and co-workers [2]), we were not able to demonstrate any effect of artemisinin on the isolated PfATP6: no inhibitory effect on the Ca2+-dependent ATPase activity was measured and no conformational change of the protein was observed when the drug was added, even at concentrations as high as 500 μM. As already described in our previous work [10], the inefficiency of artemisinin to inhibit the specific activity of the isolated PfATP6 was also true in the presence of Fe2+ that induces the formation of free radicals by reaction with peroxide (and by extension with the endoperoxide bridge of artemisinin) and in the presence of various artemisinin derivatives such as artemisone, dihydroartemisinin or artesunate. Moreover, Krishna and co-workers, using the same oocyte system, also suggested that resistance of SERCA1a to artemisinin could be eliminated by mutation of a single amino acid (E255L), to induce a high sensitivity to artemisinin [3]. However, the Ca2+-dependent ATPase activity of the same SERCA1a E255L mutant, when analysed either in the COS cell system or after expression in yeast and purification, was not affected by artemisinin [10]. Therefore the sensitivity of SERCAs to artemisinin in Xenopus oocytes could be a system-specific observation.

Since the two studies in Xenopus oocytes cited above were published, several studies raising doubts concerning the hypothesis that PfATP6 could be a target for artemisinin compounds have been reported. First, it was not possible to confirm an antagonism between artemisinin and thapsigargin [28] that was one of the cornerstones for hypothesizing PfATP6 as an important target of artemisinin in P. falciparum. Secondly, allelic-exchange of wild-type PfATP6 to a Leu263 mutant, assumed to be artemisinin-insensitive, gave no significant difference from wild-type in IC50 values for artemisinin or its derivatives [11]. Thirdly, after homology modelling of the PfATP6 three-dimensional structure and prediction of the affinities for artemisinins and other anti-malarials to the model, no correlation was found between affinity of the compounds for PfATP6 and in vitro anti-malarial activity [29]. Finally, it may be added that, in contrast with thapsigargin, which has a major effect on endoplasmic reticulum morphology, artemisinin has no significant specific effect on the endoplasmic reticulum [28], despite causing widespread damage to the parasite membranes via oxidation reactions [30]. Therefore, being embedded in a lipid bilayer, the Plasmodium Ca2+-ATPase is likely to be harmed by the induced membrane damages and then inhibited indirectly. Thus, even if it was anticipated previously to be an important target of artemisinins in the parasite P. falciparum, PfATP6 might play only a minor and indirect role in the mechanism of action of artemisinins (concerning the debate on the mechanism of action of artemisinin, see the recent review by O'Neill et al. [31]).

It is interesting to note, however, that another P-type ATPase of P. falciparum, PfATP4, is likely to have an important role in relation to new anti-malarial agents, the spiroindolones [32], and, taking the lead of well-characterized SERCA1a inhibitors and their clinical potential [24,33], both PfATP4 and PfATP6 appear to be favourable targets for new anti-malarial drugs in the future.

Funding

This work was supported by the Danish Medical Research Council, the Danish Natural Science Research Council (Center for Structural Biology, the Dansync program), the Aarhus University Research Foundation, the Novo-Nordisk Foundation (Denmark) and by the Centre National de la Recherche Scientifique (CNRS), the Commissariat à l'Énergie Atomique (CEA) and the DIM Maladies Infectieuses, Parasitaires et Nosocomiales Emergentes/Ile de France (France).

Acknowledgments

We are grateful to Delphine Cardi, Estelle Marchal and Alexandre Pozza for valuable input during the initial phase of the project and to Birte Nielsen for expert technical assistance.

Footnotes

  • Recent Advances in Membrane Biochemistry: Biochemical Society Annual Symposium No. 78 held at Robinson College, Cambridge, U.K., 5–7 January 2011. Organized and Edited by J. Malcolm East (Southampton, U.K.) and Frank Michelangeli (Birmingham, U.K.).

Abbreviations: BAD, biotin-acceptor domain; BHQ, 2,5-di-(t-butyl)-1,4-hydroquinone; C12E8, dodecyl octa(ethylene glycol) monoether; CPA, cyclopiazonic acid; DDM, dodecyl maltoside; DOPC, dioleyl phosphatidylcholine; LDH, lactate dehydrogenase; LM, light membrane; N-domain, nucleotide-binding domain; PEP, phosphoenolpyruvate; PfATP6, Plasmodium falciparum Ca2+-ATPase; PK, pyruvate kinase; SERCA, sarcoplasmic/endoplasmic reticulum Ca2+-ATPase

References

View Abstract