Heat Shock Protein 90 Function Is Essential for Plasmodium falciparum Growth in Human Erythrocytes*

Hsp90 is important for normal growth and development in eukaryotes. Together with Hsp70 and other accessory proteins, Hsp90 not only helps newly synthesized proteins to fold but also regulates activities of transcription factors and protein kinases. Although the gene coding for heat shock protein 90 fromPlasmodium falciparum (PfHsp90) has been characterized previously, there is very little known regarding its function in the parasite. We have analyzed PfHsp90 complexes and addressed its role in parasite life cycle using Geldanamycin (GA), a drug known to interfere with Hsp90 function. Sedimentation analysis and size exclusion chromatography showed PfHsp90 to be in 11s 20, w complexes of ∼300 kDa in size. Similar to the hetero-oligomeric complexes of Hsp90 in mammals, PfHsp70 was found to be present in PfHsp90 complexes. Homology modeling revealed a putative GA-binding pocket at the amino terminus of PfHsp90. The addition of GA inhibited parasite growth with LD50 of 0.2 μm. GA inhibited parasite growth by arresting transition from Ring to trophozoite. Transition from trophozoite to schizonts and reinvasion of new erythrocytes were less significantly affected. While inducing the synthesis of PfHsp70 and PfHsp90, GA did not significantly alter the pattern of newly synthesized proteins. Pre-exposure to heat shock attenuated GA-mediated growth inhibition, suggesting the involvement of heat shock proteins. Specificity of GA action on PfHsp90 was evident from selective inhibition of PfHsp90 phosphorylation in GA-treated cultures. In addition to suggesting an essential role for PfHsp90 during parasite growth, our results highlight PfHsp90 as a potential drug target to control malaria.

Plasmodium falciparum is responsible for the most severe form of malaria in humans, causing approximately 2 million deaths every year. During its asexual life cycle in human erythrocytes, the parasite progresses through three growth phases (1). The early form following invasion called the Ring stage is the phase of establishment in the erythrocyte. Trophozoite stage is the metabolically most active biosynthetic phase, whereas the schizont stage represents the phase of nuclear division before release of merozoites from the erythrocyte. Heat shock proteins of the class Hsp60, Hsp70, and Hsp90 are known to be expressed by the parasite during the intraerythrocytic stages in the vertebrate host (2)(3)(4). Although these heat shock proteins share significant homologies with their mammalian counterparts, there is very limited information available regarding their functional roles in parasite development. We have focused our study on the role of parasite-heat shock protein 90 expressed during erythrocytic cycle in humans.
Previous reports have shown that the gene coding for Hsp90 from P. falciparum is present on chromosome 7, has a single intron of 800 bp, and encodes protein with 745 amino acids, giving a molecular mass of 86 kDa (5,6). Sequence comparison shows 59% identity and 69% similarity to human Hsp90. There is significant similarity at the NH 2 -terminal nucleotide-binding domain in the central acidic hinge region as well as at the COOH-terminal substrate-binding domain. The presence of EEVD motif at the COOH terminus suggests cytosolic localization of heat shock protein 90 from P. falciparum (PfHsp90) 1 (7).
In higher eukaryotes, cytosolic Hsp90 is a highly abundant protein organized in the form of a multi-chaperone complex. Two obvious roles have been ascribed to Hsp90. 1) Together with Hsp70 and Hsp60, Hsp90 helps newly synthesized proteins to fold and 2) it helps modulate the activities of transcription factors (steroid hormone receptors and nuclear receptors) and protein kinases (8 -10). The latter activity of Hsp90 puts it at the center stage of signal transduction events, crucial for cell survival and growth. Indeed, experiments performed in yeast, fruit fly, plant, and animal systems support the idea that in addition to helping newly synthesized proteins to fold, Hsp90 also regulates cell cycle and development (11)(12)(13).
In addition to the common approaches of gene disruption in yeast and mutational analysis in Drosophila, a new approach of pharmacologically interfering with Hsp90 function has been described in the literature (14 -16). The approach involves the use of a benzoquinone ansamycin drug called Geldanamycin (GA), which specifically binds to Hsp90 and interferes with its ATP-dependent chaperone function (14). The specificity of GA interaction with Hsp90 has been described at the molecular level through the crystal structure of human Hsp90 complexed to GA (17). The use of GA in Drosophila and Arabidopsis has provided important insights into the profound involvement of Hsp90 in various cellular functions.
We have examined the complexes of PfHsp90 and analyzed its role in parasite growth in human erythrocytes using GA. Using an antibody specific to PfHsp90 and by employing sucrose gradient sedimentation as well as gel filtration chromatography, we have characterized the complexes of PfHsp90 in the parasite cytoplasm. In addition to demonstrating its pres-ence in a complex similar in size to mammalian Hsp90, we also show PfHsp70 to be a part of the PfHsp90 multi-chaperone complex. Most importantly, we show that GA inhibits parasite growth in human erythrocytes. Modeling the NH 2 -terminal domain of PfHsp90 based on its high degree of sequence identity to human Hsp90 uncovered a putative GA-binding pocket at its amino terminus. The addition of GA to parasite cultures inhibited parasite growth with an LD 50 of 0.2 M. The involvement of heat shock proteins in GA-mediated growth inhibition was evident from the observation that parasites pre-exposed to heat shock resisted growth inhibitory effects of GA. The specificity of GA action on PfHsp90 was evident from a selective inhibition of PfHsp90 phosphorylation in GA-treated cultures without a significant change in overall protein phosphorylation. Our results show that GA inhibits parasite growth through its interaction with PfHsp90 and suggests an essential role for PfHsp90 in parasite growth in human erythrocytes.

MATERIALS AND METHODS
Reagents and Antibodies-Antisera to PfHsp70 and PfHsp90 were generated against the recombinant COOH-terminal fragments of proteins in rabbit (18). DSP was obtained from Pierce. GA, herbimycin A (HA), and apyrase were purchased from Sigma.
Sedimentation on Sucrose Gradient and Gel Filtration Analysis-[ 35 S]Cys-and [ 35 S]Met-labeled trophozoites were cross-linked in vivo using DSP (9). Cells were hypotonically lysed (18), and the lysates were layered onto 5-25% continuous sucrose gradient. The gradient was spun at 40,000 rpm for 15 h at 4°C in SW41 rotor. 500-l fractions were immunoprecipitated with PfHsp90 antiserum and analyzed by SDS-PAGE and fluorography. For gel filtration analysis, saponin-released parasites (18) were sonicated briefly in phosphate-buffered saline containing protease inhibitors on ice. The lysate was clarified by centrifugation at 20,000 ϫ g for 20 min at 4°C and passed through Superdex 200 column (Amersham Biosciences) at a flow rate of 0.5 ml/min. Manually collected fractions (500 l each) were trichloroacetic acidprecipitated and immunoblotted for PfHsp90. Thyroglobulin (669 kDa), alcohol dehydrogenase (150 kDa), and bovine serum albumin (66 kDa) were used as molecular size standards.
Metabolic Labeling, Radioiodination, and Immunoprecipitation-The peak fraction of PfHsp90 (corresponding to elution volume 10.5 ml) was collected, and protein complexes were cross-linked using DSP (see above). Iodination of cross-linked proteins was carried out using [ 125 I] by chloramine T method (19). The iodinated fraction was precleared using protein A-agarose beads and immunoprecipitated using anti-PfHsp90, anti-PfHsp70, or preimmune serum. The immune pellets were solubilized in Laemmli buffer and analyzed as above.
Parasites were metabolically labeled with [ 35 S]Cys and Met and lysed in 20 volumes of NETT buffer (150 mM sodium chloride, 1 mM EDTA, 10 mM Tris, and 1% Triton X-100) (18) containing 20 units/ml apyrase, protease inhibitors, and phosphatase inhibitors. This lysate was used for non-denaturing immunoprecipitation (IP). For denaturing IP, cells were directly lysed in Laemmli buffer, heated at 95°C for 5 min, and diluted to 0.1% SDS concentration with NETT buffer. PfHsp70 or PfHsp90 immunoprecipitated from these lysates were solubilized in two-dimensional lysis buffer or Laemmli buffer for analysis. To examine induction of heat shock proteins following GA treatment, cells were treated with Me 2 SO or 10 M GA for 14h and metabolically labeled in the last 2 h. Cells were lysed in 2ϫ Laemmli buffer, boiled at 95°C for 5 min, and diluted with NETT buffer to 0.1% SDS concentration. Clarified lysate was used for immunoprecipitation of PfHsp70 and PfHsp90.
Early Ring-infected erythrocytes at 2% hematocrit and with 1-2% parasitemia were treated with Me 2 SO (as a control) or various concentrations of GA (0.05, 0.5, 1, 2, and 5 M) for 24 h (21). At the end of 24 h, [ 3 H]hypoxanthine at 10 Ci/ml was added and incubated for an additional 3 h. At the end of labeling, cells were lysed and DNA was collected in glass fiber filters using a cell harvester and the radioactivity incorporated was measured by liquid scintillation counting.
For phosphate labeling of proteins, parasites were cultured in phosphate-free medium supplemented with 0.5% human serum, [ 32 P]orthophosphoric acid, and 1 mCi/ml medium at 37°C for 6 h in the absence or presence of GA. Cells were lysed at the end of treatment in Laemmli buffer in presence of 2 mM NaF and 1 mM sodium orthovanadate and heated at 95°C for 5 min. The lysate was treated with DNase and RNase in the presence of protease inhibitors. The lysate was diluted to 0.1% SDS concentration with NETT buffer, which was used for IP or measurement of cpm in a liquid scintillation counter.
Modeling the Amino Terminus of PfHsp90 -Amino-terminal sequence of PfHsp90 (1-177 amino acids) was submitted to SWISS MODEL program (www.expasy.ch) to obtain the three-dimensional structure using the crystal structure of amino-terminal human Hsp90 (Protein Data Bank code 1YET) as a template. This model was superimposed over the structure of human Hsp90⅐GA complex using STAMP (Structure Alignment of Multiple Protein) program (20).
Effect of GA on Parasite Growth-Sorbitol-synchronized Ring-infected cells were suspended in RPMI 1640 medium to 5% hematocrit. 200-l cell suspension was treated with Me 2 SO or with GA in Me 2 SO at various concentrations for 24 h. Smears were taken at the end of treatment, Giemsa-stained, and viewed under microscope. The same experiment was performed with early trophozoites and also schizonts. To examine the effect of GA and HA on parasitemia, Ring-infected erythrocytes were either treated with Me 2 SO or with 5 M drug for 48 h (culture was replenished with medium and GA every 24 h) and parasitemia was calculated. A change in percentage of parasitemia with respect to control was plotted against the duration of drug treatment. To analyze stage-specific expression of PfHsp90, synchronous cultures of infected erythrocytes were harvested at 10 h (Ring), 20 h (early trophozoite), 30 h (late trophozoite), and 40 h (schizont) and lysed in 10 volumes of NETT buffer. The protein content of the lysates was estimated by the Bradford method. Equal protein from the different stages was analyzed for the presence of PfHsp90 by Western blotting.
Prior Heat Shock and GA Treatment-An asynchronous culture of infected erythrocytes at ϳ20% parasitemia was given heat shock at 41°C for 1 h in complete RPMI 1640 medium and then recovered at 37°C for 2 h. As a control, an equal number of cells were kept at 37°C for 3 h. Parasitemia was calculated at the end of recovery period in both control and heat-shocked cells following which the cells were treated with either Me 2 SO alone or 1 M GA for 24 h. Parasitemia was calculated again at the end of treatment period.
In Vitro Kinase Assay-Ring stage parasites were treated with Me 2 SO or 0.5 M GA for 6 h. Parasites were released using 0.01% saponin in phosphate-buffered saline, and hypotonic lysates were prepared from these parasites. 6 g of lysate from GA-treated or untreated parasites was used in a kinase assay using 2 Ci of [␥-32 P]ATP with or without 5 g of dephosphorylated bovine milk casein as substrate in 1ϫ kinase buffer (10 mM Tris, pH 7.5, 10 mM NaCl, 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, and 5 M ZnSO 4 ) in the presence of 1 mM sodium orthovanadate and 1 mM NaF. The kinase assay was performed at 37°C for 30 min and terminated by the addition of Laemmli buffer and boiling. Phosphorylation of casein was analyzed by 10% SDS-PAGE and fluorography.
Two-dimensional Gel Electrophoresis-Synchronous Ring-infected erythrocytes were treated with 10 M GA for 16 h at 37°C followed by the addition of [ 35 S]Cys and Met at 100 Ci/ml for 3 h. As a control, Ring stage parasites (normal Rings) were labeled with [ 35 S]Met at 100 Ci/ml for 3 h. The cells were washed twice in phosphate-buffered saline, and parasites were released by saponin lysis. Parasites were sonicated in 50 volumes of 20 mM Tris buffer containing 1% Triton X-100 supplemented with protease inhibitors. Lysates containing equal cpm were acetone-precipitated, solubilized in two-dimensional lysis buffer, and analyzed on a pH 3.5-9.5 tube gel in the first dimension and 7.5% SDS-PAGE in the second dimension followed by fluorography.

Complexes of PfHsp90 -Mammalian
Hsp90 is known to be organized in a 11 s 20, w hetero-oligomeric complex consisting of Hsp70 and other accessory co-chaperones. To examine whether PfHsp90 is organized in a complex similar to that of mammalian Hsp90, we employed techniques of sucrose density gradient ultracentrifugation and size exclusion chromatography. Synchronous cultures of P. falciparum were metabolically labeled with radiolabeled amino acids for 24 h, and the lysates were subjected to centrifugation on sucrose gradients as described under "Materials and Methods." Fractions collected after run were immunoprecipitated with antibodies to PfHsp90 and analyzed by SDS-PAGE and fluorography. As shown in Fig. 1A (top panel), PfHsp90 was found in complexes ranging in size from 4.5 to 9 s 20, w , indicative of heterogeneous populations ranging from monomeric forms to complexes as high as 150 kDa in size. To examine whether our lysis protocol was dis-rupting complexes of PfHsp90 in the cell, we used an approach of in vivo cross-linking to freeze complexes of PfHsp90 before preparing parasite lysates. We used DSP, which is a membrane-permeable, reversible cross-linker to cross-link PfHsp90 complexes. As seen in Fig. 1A, upon cross-linking (ϩDSP), PfHsp90 was found in complexes ranging from 5 s 20, w to as high as 11 s 20, w , corresponding to approximately 300 kDa in size. Fig. 1A, bottom panel, shows quantitation of these results.
We also employed size exclusion chromatography using HPLC to analyze PfHsp90 complexes. Lysates were prepared from synchronous cultures of the parasite as described under "Materials and Methods" and analyzed on a gel filtration column Superdex 200 interfaced to HPLC. Fractions were collected up to a one-bed volume of the column and analyzed for the presence of PfHsp90 and PfHsp70 by trichloroacetic acid precipitation of fractions and Western blotting using specific antibodies. As shown in Fig. 1B (top panel), PfHsp90 eluted as a single peak at a volume of 10 -11 ml. A comparison with the elution volumes of standards indicated that the complex corresponded to ϳ300 kDa in size. This was in agreement with the 11 s 20, w complexes observed on sucrose gradient analysis of cross-linked samples. On the other hand, PfHsp70 was found to To analyze the composition of the 300-kDa complex seen on HPLC, we iodinated the peak fractions corresponding to elution volumes 10 -10.5 ml after cross-linking and immunoprecipitated the labeled complex using antibodies to PfHsp90 and PfHsp70. An equal volume of the labeled fraction was also incubated with protein A beads and pre-immune serum to ascertain the specificity of immunoprecipitation. Fig. 1C shows an analysis of the immunoprecipitate by SDS-PAGE and fluorography. As expected, a clear band corresponding to PfHsp90 (lane 2) was visible on SDS-PAGE, which was absent in the control (lane 1). In addition to PfHsp90, three other bands corresponding in size to ϳ75-, 60-, and 50-kDa proteins were also seen in PfHsp90 immunoprecipitate. To examine whether the 75-kDa protein was PfHsp70, an equal volume of iodinated fraction was immunoprecipitated using antibodies to PfHsp70. As shown in Fig. 1C (lane 3) in addition to a signal corresponding to PfHsp70, we also found a signal for PfHsp90. The result indicated that PfHsp70 was included in the 300-kDa complex of PfHsp90 in the parasite cytoplasm.
Interaction of PfHsp90 and PfHsp70 -To further examine possible interactions between PfHsp70 and PfHsp90, we used a more direct approach of co-immunoprecipitation from labeled parasite lysates. Synchronous cultures of P. falciparum were metabolically labeled for 2 h with radiolabeled amino acids and lysed with detergent containing buffer in the presence of apyrase as described under "Materials and Methods." Apyrase was included in the lysis buffer to stabilize possible association of PfHsp70 and PfHsp90 (9). The lysates were divided into three equal aliquots. One aliquot was immunoprecipitated with antibodies to PfHsp90. The second aliquot was immunoprecipitated with antibodies to PfHsp70; whereas the third aliquot was incubated with preimmune serum alone to serve as a control. The immunoprecipitates were analyzed by SDS-PAGE and fluorography as described under "Materials and Methods." As seen in Fig. 2A, in samples immunoprecipitated with antibodies to PfHsp70 (lane 2), we could detect a band corresponding to itself but, in addition, a band corresponding in size to PfHsp90 was also evident. Similarly, in sample immunoprecipitated with antibodies to PfHsp90, we could find a band corresponding to PfHsp70 in addition to itself (lane 3). Preim-mune serum control (lanes 1 and 4) did not show any specific signal. Immunoprecipitation done under denaturing conditions for Pfhsp70 and PfHsp90 pulled down only the respective proteins ( lanes 5 and 6).
To ascertain that the co-precipitating bands indeed correspond to PfHsp70 and PfHsp90, we also analyzed the immunoprecipitates by two-dimensional gel electrophoresis. We have previously defined the positions of PfHsp70 and PfHsp90 spots on two-dimensional gels (18). As shown in Fig. 2B, in PfHsp70 immunoprecipitates, we found a spot of PfHsp70 (arrowhead). In PfHsp90 immunoprecipitate (Fig. 2C) in addition to a spot corresponding to PfHsp90 (arrow) at a position expected from its size (90 kDa) and pI (4.8), we also found a distinct 75-kDa spot (arrowhead) corresponding in size (74.3 kDa) and pI (5.4) to PfHsp70. The preimmune serum control on the other hand did not show the presence of either of these spots on twodimensional gels (data not shown). When a small aliquot of PfHsp90 IP (apyrase) was mixed with denaturing PfHsp70 IP (which pulls down only PfHsp70 as shown in Fig. 2A, lane 5), there was an increase in the intensity of the 75-kDa spot (compare PfHsp90:PfHsp70 ratios between panels C and D) showing that PfHsp70 is co-precipitated with PfHsp90 under non-denaturing conditions. The results confirmed that PfHsp90 and PfHsp70 were present in a common complex.
A Putative GA-binding Domain at the Amino Terminus of PfHsp90 -GA, a benzoquinone ansamycin drug with antitumor properties, is known to specifically bind the amino-terminal domain of human Hsp90. The structure of the complex of human Hsp90 with GA has shown that the drug interacts with the amino-terminal ATP-binding domain of Hsp90 (17). To examine the presence of putative GA binding site in PfHsp90, we aligned sequences corresponding to the amino-terminal GAbinding domain of human Hsp90 with PfHsp90 (see Fig. 3A). The amino-terminal domain of PfHsp90 shows 69% identity with human Hsp90 and contains a GXXGXG motif essential for ATP binding. The contact-making residues, critical in the binding to GA, were conserved in PfHsp90 (indicated by asterisk). Based on the high degree of sequence similarity between human Hsp90 and PfHsp90 in this region, we modeled the structure of the PfHsp90. We used the structure of human Hsp90 in complex with GA as a template for this purpose (Protein Data Bank code 1YET). Modeling the amino-terminal domain of PfHsp90 (residues 3-177) indicated that the overall fold in this region was highly similar to that in human Hsp90 (Fig. 3B). The contact-making residues, namely Lys-58, Asp-93, Gly-97, Lys-112, Phe-138, and Gly-183, were similarly positioned in PfHsp90 (Fig. 3C). The structures could be superimposed with a root mean square deviation of 0.07. The analysis indicated that GA is likely to interact with PfHsp90 in a manner similar to its association with human Hsp90. GA Inhibits Parasite Growth-To examine the effect of GA on parasitemia, synchronous Ring stage parasites were treated with 5 M GA as described under "Materials and Methods" and percentage parasitemia was determined every 12 h until 48 h. We also examined the effect of a closely related benzoquinone ansamycin, HA, on parasite growth. The results were plotted as percentage parasitemia with respect to control versus hours of drug treatment. As shown in Fig. 4A, there was a progressive decline in the percentage parasitemia in cultures exposed to 5 M GA. Although 5 M HA treatment also resulted in reduction in parasitemia, the effect was less drastic. [ 3 H]Hypoxanthine incorporation in GA-treated cultures (Fig. 4B) confirmed growth inhibitory effects of GA, and the LD 50 value was determined to be 0.2 M.
GA Blocks Ring to Trophozoite Stage Progression-To examine the effect of GA on growth of the parasite, we used highly synchronous cultures of P. falciparum in Ring stage and treated them with different concentrations of GA. At the end of 24 h of GA treatment, smears were taken to determine the stage of the parasite and percent parasitemia. The results were compared, in parallel, with mock-treated Ring stage culture. As expected, in the absence of GA treatment, Ring stage cultures progressed to the trophozoite stage (see Fig. 5A, left panel,  arrow). Fig. 5A, middle panel, also shows a smear from culture treated with 10 M GA. Most of the parasites persisted in the Ring stage (see arrow) even at the end of 24 h in the presence of the drug. In cultures treated with GA at a concentration of 0.5, 1, 5, or 10 M for 24 h, we found a progressive decline in the number of parasites progressing to the trophozoite stage. Right panel shows the extent of progression from Ring to trophozoite stage at increasing concentrations of GA. A similar experiment was carried out with synchronous cultures in the trophozoite stage progressing to schizont stage in the presence of different concentrations of GA. As shown in Fig. 5B, in the absence of GA treatment (left panel), trophozoite stage culture effectively progressed to schizonts (arrow), but in the presence of GA (middle panel), there was persistence of trophozoites (arrow) even after 24 h. The degree of inhibition in stage progression increased with an increase in GA concentration. Fig. 5B, right panel, shows quantitation of the inhibitory effect of GA on progression from trophozoite to schizont stages. The inhibitory effect of GA was more pronounced for progression from Ring to trophozoite stage than from trophozoite to schizont stage. Importantly, GA did not significantly affect the release of merozoites from schizonts, and their reinvasion to gave rise to new Rings (5C). To examine the levels of PfHsp90 present in different stages of parasite growth, we prepared lysates from equal number of Rings, early trophozoites, late trophozoites, and schizonts and examined equal amounts of protein from different stages by SDS-PAGE and Western blotting. Fig. 5D, top panel,  GA Induces Heat Shock Protein Synthesis-To examine the effect of GA treatment on overall protein synthesis and heat shock protein induction, we used two aliquots of synchronous Ring stage culture. One was cultured for 14 h in the presence of GA and labeled for 2 h (see "Materials and Methods"). Giemsa staining of smears from the cultures confirmed that GA-treated aliquot was blocked in the Ring stage (as also described above). The second Ring stage culture was labeled for 2 h without any treatment and lysed as described under "Materials and Methods." Aliquots of lysates from both the samples were precipitated with acetone, and radioactivity incorporated was counted. As shown in Fig. 6A, the amount of radioactivity incorporated in GA-treated culture was not drastically different from that seen in untreated Ring stage culture (normal Rings). When lysates of normal Ring and GA-blocked Ring containing equal cpm were analyzed by two-dimensional gel electrophoresis, overall protein profile looked similar in both (Fig. 6B). The result suggested that the overall profile of proteins made in the presence of GA was qualitatively and quantitatively similar to that in controls.
To examine the effect of GA on the levels of PfHsp70 and PfHsp90, we immunoprecipitated these proteins from lysates containing equal cpm from GA-treated and untreated cultures using specific antibodies. As shown in Fig. 6C, top panel, higher amounts of PfHsp70 and PfHsp90 were seen in GA-treated parasites (lanes 2 and 4) compared with untreated cultures (lanes 1 and 3). Bottom panel shows quantitation of these data, indicating ϳ3-fold induction of PfHsp70 and 5-fold induction of PfHsp90 upon GA treatment.
Prior Heat Shock Counteracts GA-mediated Growth Inhibition-To analyze the involvement of heat shock proteins in GA-mediated parasite growth inhibition, we examined whether prior heat shock is able to attenuate GA-mediated growth arrest. An asynchronous culture of P. falciparum was divided into two aliquots. One aliquot was exposed to heat shock for 1 h at 41°C, whereas the other aliquot was used as a control at 37°C. The culture was allowed to recover for 2 h following heat shock, and percent parasitemia was determined in both of the aliquots. Following recovery, both of the aliquots were incubated with 1 M GA or Me 2 SO for 24 h. At the end of 24 h of GA treatment, parasitemia was again determined in both the cultures. Fig. 7 shows quantitation of the drop in parasitemia upon GA treatment in control culture and culture exposed to prior heat shock. Although there was Ͼ50% drop in parasitemia upon GA treatment in control cells, the decrease was significantly attenuated in culture pre-exposed to heat shock. The result suggests that pre-induction of heat shock proteins as a result of prior heat shock may confer protection against GA-mediated growth inhibition.
GA Blocks PfHsp90 in a Dephosphorylated State-Chaperoning function of mammalian Hsp90 has been linked to its phosphorylation state in the cell, and GA has been shown to abrogate Hsp90 phosphorylation in mammalian cells (22)(23)(24). To examine whether phosphorylation state of PfHsp90 was affected in GA-treated parasites, we labeled Ring stage culture with [ 32 P]phosphoric acid with or without GA treatment (0.5 M) as described under "Materials and Methods" and measured total cpm incorporated in the cell lysates. Lysates corresponding to equal cpm were immunoprecipitated with antibodies to PfHsp90 or PfHsp70, and the immunoprecipitates were analyzed by SDS-PAGE and fluorography. As shown in Fig. 8A  the signal corresponding to PfHsp70 (lane 4) was similar to control that for PfHsp90 were significantly reduced (lane 6). Quantitation of bands indicated a 60% drop in signal for phosphorylated PfHsp90 in GA-treated cultures (Fig. 8A, bottom  panel).
To rule out the possibility that GA was inhibiting overall kinase activity in the parasite, we also examined the ability of lysates from GA-treated parasites to phosphorylate exogenously added bovine milk casein. Ring stage parasites were treated with 0.5 M GA or Me 2 SO for 6 h. Hypotonic lysates prepared from saponin-released parasites were incubated with casein in the presence of [␥-32 P]ATP for 30 min (see "Materials and Methods"). Casein in the reaction mixture was then analyzed by SDS-PAGE and fluorography. As shown in Fig. 8B, casein was phosphorylated to a similar extent by lysates from both GA-treated and untreated parasites. These results indicated that GA did not affect total kinase activity in the infected erythrocytes and that decrease in phosphorylation of PfHsp90 seen in in vivo phosphate-labeling experiment was a result of a specific effect of GA on PfHsp90. DISCUSSION Among the different heat shock proteins described in biological systems, heat shock protein 90 plays a particularly important role in cell growth and development. In addition to participating in the folding and assembly events of newly synthesized proteins, Hsp90 also regulates signal transduction events by interacting with transcription factors and protein kinases. In mammalian cells, Hsp90 is one of the most abundant proteins in the cytoplasm, existing as a multichaperone complex associated with Hsp70, Hsp60, p23, and cyclophilin (25). The lethal phenotype of Hsp90 disruption in yeast underlines its pivotal role in cell function (26). Mutations in Hsp90 from Drosophila melanogaster are known to result in a variety of developmental abnormalities (12). Similarly, defects in Hsp90 function in Arabidopsis thaliana result in phenotypic changes (13). The results have been interpreted to suggest that normal functioning of Hsp90 is essential to buffer phenotypic variations.
Although the gene coding for Hsp90 in P. falciparum has been cloned (27), there is very limited information available regarding its function in the parasite. At the level of primary structure (as predicted from the nucleotide sequence), Hsp90 from P. falciparum is highly homologous to Hsp90 in mammalian cells. Previous studies indicate that PfHsp90 is predominantly cytoplasmic and is expressed in all three intraerythrocytic stages of human host (4). To begin to understand the biochemical role of PfHsp90, we have initiated studies on its size and composition of its complexes in the parasite cytoplasm. Furthermore, by using GA, a drug known to inhibit Hsp90 function (14,15), we show that PfHsp90 is essential for parasite stage progression during intraerythrocytic growth.
Sedimentation analysis on sucrose gradients from in vivo cross-linked samples indicated that PfHsp90 was present in complexes ranging in size from 5 to 11 s 20, w in the cell. The sedimentation coefficient of 5 s 20, w corresponded to monomeric 90-kDa form of PfHsp90, whereas 11 s 20, w corresponded to complexes of size 300 kDa. This was in agreement with the size of multi-chaperone complex reported for mammalian Hsp90 (28). Gel filtration analysis confirmed a size range up to ϳ300 kDa for PfHsp90. Similar to the Hsp90 complex reported in mammalian cells, we found PfHsp70 to be a part of the PfHsp90 complex. In addition, at least two other proteins of size 60 and 50 kDa could be seen in the PfHsp90 complexes. The identities of these proteins remain to be established. It is probable that these are parasite counterparts of Hop60 and cyclophilin known to be present in the mammalian Hsp90 multi-chaperone complex. Recently published genome sequences of the parasite indeed show the presence of Hop60 and cyclophilin homologs (29).
GA has been reported to interact with the nucleotide-binding domain of human Hsp90 (30). The crystal structure of the complex between amino-terminal domain of Hsp90 and GA has been reported previously (17). The high degree of sequence similarity between human Hsp90 and PfHsp90 allowed us to model the amino-terminal domain of PfHsp90 and compare its structure with that of human Hsp90. Superimposition of PfHsp90 amino-terminal domain with Hsp90 in complex with GA using STAMP program revealed that all of the points of contact could be juxtaposed. The model of PfHsp90 suggested that it possessed a putative GA-binding domain at the amino terminus.
The addition of GA to P. falciparum culture resulted in inhibition of its intraerythrocytic stage progression. Inhibition of parasite growth occurred in a narrow time window, affecting progression from Ring to trophozoite stage maximally. Progression from trophozoites to schizont stage release of merozoites from schizonts and reinvasion by newly released merozoites were affected less drastically by GA. Trophozoite being biosynthetically most active phase, the progression from Ring to trophozoite, may depend heavily on PfHsp90 function, thereby showing an obvious sensitivity to GA. Indeed, Hsp90 synthesis was found to be maximal during Ring to trophozoite transition. Examination of cultures beyond one generation (Ͼ48h) revealed a significant decrease in the number of parasites in GA-treated cultures. This was probably a result of lysis of Ring stage-arrested parasites upon GA treatment.
Attenuation of GA-mediated growth arrest in parasites preexposed to heat shock provided direct evidence for the involvement of heat shock proteins in GA-mediated growth arrest. Such "cross-tolerance" conferred by pre-exposure to heat stress has been well documented in the literature (31). That GAmediated arrest in parasite growth was a result of its effect on PfHsp90 function was evident from a specific drop in the levels of phosphorylated PfHsp90 without any decrease in overall protein phosphorylation in GA-treated parasites. A similar GAmediated drop in Hsp90 phosphorylation has been reported in mammalian cells. Such a decrease in mammalian Hsp90 phosphorylation has been shown to be linked to inhibition of its chaperoning function (22).
GA has been shown to affect the growth of another protozoan parasite belonging to the genus Leishmania (32). GA arrests the growth of the promastigote stage (insect stage) of Leishmania donovani in culture. GA is also thought to mimick heat stress experienced by the parasite during transmission from the insect to the vertebrate host and induce its progression to amastigote stage, which is normally found only in the vertebrate host. Overexpression of cytosolic Hsp90 has been shown to overcome such growth inhibitory effect of GA in Leishmania. GA has also been reported to inhibit tumor growth in animal cells by interfering with tumor-inducing factors like mutated p53, Raf kinase src kinase, and steroid receptors, which are known substrates of cytosolic Hsp90 (14,33,34). GA is thought to bring about tumor inhibitory effects by abrogating interactions of the above tumor-promoting factors with Hsp90. The anti-tumor potential of GA is currently being evaluated under clinical trials in humans (33).
In all, our study shows PfHsp90 to be present in heterooligomeric complexes essential for parasite growth. Specific mechanisms of parasite growth inhibition by GA need to be addressed in future. Identification of PfHsp90 client proteins, i.e. nuclear receptors and protein kinase orthologs in the parasite, becomes an important priority for malaria biologists. The study of GA effects on gene expression profiles during stage progression will help uncover detailed mechanisms of parasite stage progression and may lead to identification of novel drug targets against malaria.