Suramin and Suramin Analogues Inhibit Merozoite Surface Protein-1 Secondary Processing and Erythrocyte Invasion by the Malaria Parasite Plasmodium falciparum*

Malarial merozoites invade erythrocytes; and as an essential step in this invasion process, the 42-kDa fragment of Plasmodium falciparum merozoite surface protein-1 (MSP142) is further cleaved to a 33-kDa N-terminal polypeptide (MSP133) and an 19-kDa C-terminal fragment (MSP119) in a secondary processing step. Suramin was shown to inhibit both merozoite invasion and MSP142 proteolytic cleavage. This polysulfonated naphthylurea bound directly to recombinant P. falciparum MSP142 (Kd = 0.2 μm) and to Plasmodium vivax MSP142 (Kd = 0.3 μm) as measured by fluorescence enhancement in the presence of the protein and by isothermal titration calorimetry. Suramin bound only slightly less tightly to the P. vivax MSP133 (Kd = 1.5 μm) secondary processing product (fluorescence measurements), but very weakly to MSP119 (Kd ∼ 15 mm) (NMR measurements). Several residues in MSP119 were implicated in the interaction with suramin using NMR measurements. A series of symmetrical suramin analogues that differ in the number of aromatic rings and substitution patterns of the terminal naphthylamine groups was examined in invasion and processing assays. Two classes of analogue with either two or four bridging rings were found to be active in both assays, whereas two other classes without bridging rings were inactive. We propose that suramin and related compounds inhibit erythrocyte invasion by binding to MSP1 and by preventing its cleavage by the secondary processing protease. The results indicate that enzymatic events during invasion are suitable targets for drug development and validate the novel concept of an inhibitor binding to a macromolecular substrate to prevent its proteolysis by a protease.

Malarial merozoites invade erythrocytes; and as an essential step in this invasion process, the 42-kDa fragment of Plasmodium falciparum merozoite surface protein-1 (MSP1 42 ) is further cleaved to a 33-kDa N-terminal polypeptide (MSP1 33 ) and an 19-kDa C-terminal fragment (MSP1 19 ) in a secondary processing step. Suramin was shown to inhibit both merozoite invasion and MSP1 42 proteolytic cleavage. This polysulfonated naphthylurea bound directly to recombinant P. falciparum MSP1 42 (K d ‫؍‬ 0.2 M) and to Plasmodium vivax MSP1 42 (K d ‫؍‬ 0.3 M) as measured by fluorescence enhancement in the presence of the protein and by isothermal titration calorimetry. Suramin bound only slightly less tightly to the P. vivax MSP1 33 (K d ‫؍‬ 1.5 M) secondary processing product (fluorescence measurements), but very weakly to MSP1 19 (K d ϳ 15 mM) (NMR measurements). Several residues in MSP1 19 were implicated in the interaction with suramin using NMR measurements. A series of symmetrical suramin analogues that differ in the number of aromatic rings and substitution patterns of the terminal naphthylamine groups was examined in invasion and processing assays. Two classes of analogue with either two or four bridging rings were found to be active in both assays, whereas two other classes without bridging rings were inactive. We propose that suramin and related compounds inhibit erythrocyte invasion by binding to MSP1 and by preventing its cleavage by the secondary processing protease. The results indicate that enzymatic events during invasion are suitable targets for drug development and validate the novel concept of an inhibitor binding to a macromolecular substrate to prevent its proteolysis by a protease.
Plasmodium falciparum develops and replicates within erythrocytes, releasing merozoites that invade new red blood cells. This stage of the parasite's life cycle is responsible for the disease malaria, and inhibition of merozoite invasion reduces parasitemia, with beneficial outcome for the host. Invasion is a complex process, involving interaction between parasite and host molecules; and although poorly understood at the molecular level, it may represent a new chemotherapeutic target. Current antimalarial drugs are becoming increasingly ineffective, and few new treatments have been developed in recent years (1). New drugs are required against established and novel targets, together with strategies to reverse resistance and to bypass toxicity of known antimalarial compounds (2). The development of compounds that are significantly different from those already in use or directed against new targets may delay the onset of resistance (3).
Several proteins have been identified on the surface of the merozoite (4). For example, merozoite surface protein-1 (MSP1) 1 is synthesized as an ϳ200-kDa precursor and is present on the surface of the late stage parasite within the erythrocyte. At or immediately prior to merozoite release, MSP1 is cleaved (primary processing) into four fragments that form part of a protein complex on the surface of the free merozoite. One of these fragments, the 42-kDa C-terminal polypeptide (MSP1 42 ), has a glycosylphosphatidylinositol anchor holding the complex to the parasite surface. Upon erythrocyte invasion, the protein complex is released from the merozoite surface following secondary processing, which involves a single proteolytic cleavage within MSP1 42 . The 33-kDa N-terminal part of MSP1 42 is shed with the complex, whereas MSP1 19 , the Cterminal part of MSP1 42 , remains on the surface of the invading merozoite. MSP1 19 contains two epidermal growth factor (EGF) domains (5). Certain MSP1 19 -specific monoclonal antibodies inhibit both secondary processing and erythrocyte invasion (6), suggesting that inhibitors of the protease responsible for MSP1 secondary processing would inhibit invasion. Prevention of secondary processing of MSP1 may thus be a good chemotherapeutic target.
We have chosen to examine suramin, a symmetrical hexasulfated naphthylurea, to explore its ability to bind to MSP1 and to inhibit invasion because it has been shown that this highly charged compound interferes with interaction of growth factors containing EGF domains with their receptors (7-10). Suramin has also been shown to inhibit invasion of HepG2 cells by P. falciparum sporozoites (11). Other polyanionic compounds such as sulfated polysaccharides are known to inhibit merozoite invasion (12). Suramin has been used to treat African trypanosomiasis (13) and filariasis (14). Although suramin is toxic, some of its analogues (15) have been found to be much less toxic than suramin itself when tested in mice (16). Here, we show that suramin binds to MSP1 42 and that such binding inhibits both MSP1 secondary processing and merozoite invasion of erythrocytes. Some suramin analogues are also shown to inhibit processing and invasion.

In Vitro Culture and Synchronization of P. falciparum
Asexual blood stages of P. falciparum (isolate FCB-1) were maintained at 37°C in RPMI 1640/Albumax medium (Invitrogen) supplemented with 2 mM L-glutamine as previously described (17). Cultures were gassed with 7% CO 2 , 5% O 2 , and 88% N 2 and maintained by routine passage in fresh human erythrocytes. Parasites were synchronized by Percoll and sorbitol treatment as described previously (18). Briefly, schizonts were purified by centrifugation over Percoll and then returned to culture in the presence of fresh erythrocytes. After 4 h, during which time released merozoites invaded erythrocytes, the cells were treated with 5% sorbitol for 10 min to lyse the residual schizonts before returning the parasites to culture.

In Vitro Invasion Inhibition Assays
Compounds were tested for their capacity to inhibit erythrocyte invasion by P. falciparum in vitro using two approaches: a short-term assay in which the number of newly invaded erythrocytes was counted using microscopy and an assay measuring uptake of [ 3 H]hypoxanthine. In the short-term assay (19), compounds were incubated in triplicate with cultures containing highly synchronous P. falciparum schizonts at ϳ2% parasitemia and ϳ2% hematocrit. After between 6 and 24 h, thin films were prepared, stained with Giemsa reagent, and examined by microscopy. The number of newly invaded ring stages was counted, and inhibition of invasion was expressed as percent invasion relative to a control untreated culture (calculated using the formula ((I C /(I C ϩ U C ))/ (I A /(I A ϩ U A ))) ϫ 100, where I C is the number of erythrocytes infected with ring stages, U C is the number of uninfected erythrocytes in the presence of the compound, I A is the number of erythrocytes infected with ring stages, and U A is the number of uninfected erythrocytes in the absence of the compound).
Suramin and NTS were also assayed for P. falciparum invasion inhibition using [ 3 H]hypoxanthine uptake (20). Serial dilutions of the compounds were added in triplicate to cultures containing mature schizonts at a final parasitemia of ϳ0.5% and a hematocrit of ϳ2% in 96-well plates. Following incubation for 24 h to allow schizont rupture and erythrocyte invasion, [ 3 H]hypoxanthine (0.5 Ci/well) was added. After a further 18 h of culture, cells were harvested onto glass-fiber filters (Filtermat A, Wallac, Turku, Finland) using a cell harvester. Filters were wetted with scintillation mixture, and bound radioactivity was quantified in a ␤-counter. Control incubations without compound or without parasites were included in each experiment. The amount of radioactivity in each sample was expressed as a percentage of activity in the control wells containing no compound. Three independent experiments were performed for each compound.

MSP1 Secondary Processing Assay
P. falciparum merozoites were purified as described previously (17) after release from mature schizonts into growth medium supplemented with 2 mM EGTA. Following sequential passage of culture supernatants through pre-wetted filters of 3-and 1.2-m pore size, merozoites were harvested by centrifugation. Merozoites were washed with Ca 2ϩ -and Mg 2ϩ -free phosphate-buffered saline in the presence of protease inhibitors (leupeptin, antipain, and aprotinin, each at 10 g/ml). The merozoites were then resuspended in cold 50 mM Tris-HCl (pH 7.2), 5 mM CaCl 2 , and 1 mM MgCl 2 plus 10 g/ml each leupeptin, antipain, and aprotinin (processing buffer) and divided into 18-l aliquots on ice. Two microliters of either diluted compound or reaction buffer were added, and the samples were incubated for 1 h at 37°C. Control assays included those in which processing was prevented by immediate addition of SDS, 1 mM phenylmethylsulfonyl fluoride (PMSF), or 5 mM EGTA and those incubated in the absence of compound or in the presence of 0.2 mM N ␣ -p-tosyl-L-lysine chloromethyl ketone. Processing was stopped by rapid solubilization into SDS, and MSP1 processing was analyzed by Western blotting (17). Blots were probed with rabbit anti-MSP1 33 /MSP1 42 antibody, and the bands corresponding to MSP1 42 and MSP1 33 polypeptides were visualized by enhanced chemiluminescence.

MSP1-based Recombinant Proteins
P. falciparum MSP1 42 -The pET(AT)Pf-MSP1 42 plasmid was used to express a His 6 -tagged MSP1 42 protein (P. falciparum 3D7 clone, Gen-Bank TM /EBI accession number Z35327) essentially as described (21). Following bacterial growth, protein expression was induced by addition of 0.1 mM isopropyl-␤-D-thiogalactopyranoside at 25°C. After 2 h of additional growth, the bacterial cells were harvested by centrifugation, resuspended in buffer, and lysed by microfluidization. The protein was purified by binding and elution from a column of nickel-nitrilotriacetic acid Superflow resin (QIAGEN Inc.), followed by passage through two ion exchange columns (SuperQ 650M and CM 650M, TosoHaas).
P. falciparum MSP1 19 -P. falciparum MSP1 19 (Swiss-Prot accession number P04933) was expressed from a synthetic gene optimized for Pichia pastoris expression (36). This sequence, inserted at the SnaBI site of vector pPIC9K (Invitrogen), contained the N-terminal sequence YHHHHHHIEGRH preceding the MSP1 19 sequence. A point mutation (Ser 3 to Ala) was also introduced to eliminate the potential N-glycosylation site at Asn 1 . Following purification of MSP1 19 as described (5), the N-terminal tag was cleaved with factor Xa (New England Biolabs Inc.), and MSP1 19 was purified by gel filtration (Superdex 200). The final product consisted of the 96-amino acid MSP1 19 fragment preceded by a single His residue.
P. vivax MSP1 42 -The sequence encoding MSP1 42 from P. vivax was cloned into plasmid jmp28, a modified pET28 vector that encodes the N-terminal MHHHHHHIEGRWIL peptide immediately upstream of the inserted sequence. P. vivax (Belem strain) DNA was used as a template for PCR-based cloning. The sequence of the expressed protein (following the vector-encoded N-terminal peptide) corresponds to Asp 1325 to Ser 1704 (GenBank TM /EBI accession number A39401).
The protein was expressed in Escherichia coli BL21(DE3) pLysS cells by induction of a culture at A 600 ϭ 0.6 -0.8 with isopropyl-␤-D-thioga-lactopyranoside at a final concentration of 1 mM. The cells were induced for 3 h at 37°C and then harvested by centrifugation. The cells were lysed using the Bugbuster protein extraction reagent (Novagen). After centrifugation of the cell lysate, the pellet containing MSP1 42 was dissolved in 6 M guanidine HCl, 100 mM Na 2 HPO 4 , and 10 mM Tris-HCl (pH 8.0). This was applied to a nickel-nitrilotriacetic acid column and washed with 10 column volumes of the loading buffer, followed by 6 column volumes of 8 M urea, 100 mM Na 2 HPO 4 , and 10 mM Tris-HCl at pH 7.0 and 6 column volumes of the same buffer at pH 6.1. The protein was eluted in 3 column volumes of 8 M urea, 100 mM Na 2 HPO 4 , and 10 mM Tris-HCl (pH 4.3).
P. vivax MSP1 33 -P. vivax MSP1 33 was expressed using the P. vivax MSP1 42 clone described above as a template for PCR, followed by ligation of the product into the pET30Xa/LIC vector. The sequence expressed runs from Asp 1325 to Ser 1618 (GenBank TM /EBI accession number A39401) following the N-terminal purification tag. The protein was expressed in BL21(DE3) pLysS cells as described above for P. vivax MSP1 42 . The cell lysate was cleared by centrifugation, and the supernatant was loaded directly onto a nickel-nitrilotriacetic acid column. The column was washed with 10 volumes of 50 mM NaH 2 PO 4 , 300 mM NaCl, and 10 mM imidazole (pH 8.0) and with 6 column volumes of the same buffer containing 30 mM imidazole and finally eluted with 3 column volumes of the same buffer with 250 mM imidazole. The eluted protein was then loaded directly onto a Superdex 200 column (26 ϫ 600 mm) equilibrated in 20 mM Tris-HCl and 250 mM NaCl (pH 8.0) and purified by gel filtration. The fractions containing MSP1 33 were pooled and dialyzed extensively against phosphate-buffered saline.

Binding of Suramin to MSP1 Assayed by Fluorometry
The intrinsic fluorescence of suramin was used to determine its capacity to bind to MSP1-derived proteins. Fluorescence was measured using a PerkinElmer Life Sciences LS-3B or Spex Industries Fluoro Max-2 fluorometer. Suramin was excited at a wavelength of 315 or 330 nm at 2.5-nm resolution, and emission spectra were measured at wavelengths ranging from 350 to 450 nm. Suramin and the MSP1 proteins were diluted in 20 mM NaH 2 PO 4 and 150 mM NaCl at pH 7.2 and analyzed at 10 or 20°C. Titrations were performed by adding aliquots of the suramin solution to the MSP1 solution. In controls, suramin was titrated into buffer alone. The K d for MSP1/suramin binding was determined from three independent experiments.

Isothermal Titration Calorimetry
Isothermal calorimetric titrations were performed using a Microcal Omega VP isothermal titration calorimeter. Tested proteins were dialyzed extensively against isothermal titration calorimetry buffer (phosphate-buffered saline at pH 7.4) before use. All experiments were performed at 25°C. The calorimeter cell, which has an internal volume of 1.425 ml, contained 30 M MSP1 42 or MSP1 33 , which was titrated by injection of a total of 290 l of 600 M suramin. The heat of dilution of suramin into buffer alone was determined in control experiments. Data were fitted by least-squares methods using the Microcal Origin Version 5.0 evaluation software provided by the manufacturer. Each experiment was performed twice.

NMR Studies on MSP1 19
NMR experiments were carried out on Varian spectrometers operating at proton frequencies of 500, 600, and 800 MHz. Suramin (0.03-16.0 mM) 1 H spectra were recorded at 5-35°C. The assignments of the 1 H signals of suramin (except for the NH signals) were made by analysis of the two-dimensional gradient selected double quantum filtered COSY spectrum at 500 MHz at 25°C on a sample containing 1 mM suramin in 50 mM potassium phosphate and 100 mM KCl in 90% H 2 O and 10% D 2 O at pH 6.5. The NH assignments were made on the basis of nuclear Overhauser effect correlation spectroscopy experiments (Fig. 2). The assignments agree with those reported previously (22). MSP1 19 /sura-min samples were examined with either unlabeled or 15 N-labeled MSP1 19 in 50 mM sodium phosphate and 100 mM NaCl in 90% H 2 O and 10% D 2 O at pH 6.5 (pH values are pH* meter readings uncorrected for deuterium isotope effects). Titration was carried out by mixing two samples each containing 0.1 mM MSP1 19 and with one also containing 16 mM suramin. One-dimensional 1 H and two-dimensional 1 H-15 N heteronuclear single quantum correlation NMR spectra were recorded for each concentration of suramin. Nuclear Overhauser effect correlation spectra were recorded on MSP1 19 /suramin samples in D 2 O (1.76 mM protein and 6 mM suramin in 50 mM sodium phosphate and 100 mM NaCl at pH 6.5); these spectra were compared with MSP1 19 spectra recorded in the absence of suramin to detect any suramin-induced changes in 1 H chemical shifts for protein side chain resonances.

Suramin Inhibits Erythrocyte Invasion by Merozoites and MSP1 Secondary
Processing-Suramin inhibition of erythrocyte invasion was measured both by microscopic examination of cultures following staining with Giemsa reagent and by biosynthetic incorporation of [ 3 H]hypoxanthine. As shown in Fig. 3, suramin inhibited invasion in a dose-dependent manner with an IC 50 of 60 Ϯ 9 M, whereas NTS did not inhibit invasion at 200 M, the highest concentration tested. Similar results were obtained in both assays. Free merozoites were observed in the stained samples, suggesting that suramin does not inhibit merozoite release from schizonts, but does prevent invasion, confirming our earlier findings. 2 At these concentrations, suramin had no effect on the intracellular growth of the parasite (data not shown).
Suramin was shown to inhibit secondary processing of MSP1 42 in extracts of merozoites (Fig. 4A). Following a 1-h incubation of merozoites in the presence of suramin or control compounds, no MSP1 33 product was detected in the sample incubated with 1 mM PMSF (a potent inhibitor of processing) or 200 M suramin. In contrast, 200 M NTS had no effect on MSP1 secondary processing, with clear formation of MSP1 33 occurring in the NTS-treated sample. Significant MSP1 cleavage was detected only in samples incubated with 12.5 M or lower concentrations of suramin (Fig. 4B).
Suramin Binds to MSP1 42 and MSP1 33 -The intrinsic fluorescence of suramin is low when excited at 315 nm; but when it bound to MSP1 42 , it exhibited a pronounced increase in emission intensity, with the maximum emission wavelength shifted slightly from 408 to 411 nm (Fig. 5, inset). This fluorescence enhancement was used to measure binding of suramin to MSP1 42 from P. falciparum. Following titration of a solution of MSP1 42 with suramin, the marked increase in the fluorescence intensity was measured as a function of suramin concentration (Fig. 5). Analysis of the binding data revealed that suramin bound to P. falciparum MSP1 42 with a K d of 0.22 Ϯ 0.04 M.  concentrations, preventing measurements of suramin binding in vitro. However MSP1 33 from P. vivax is soluble to at least 20 mg/ml and hence amenable to in vitro methods of measuring suramin binding. When analyzed by fluorometry, P. vivax MSP1 42 and MSP1 33 showed large binding-induced enhancements of the intrinsic suramin fluorescence similar to that observed with P. falciparum MSP1 42 . P. vivax MSP1 42 bound suramin with a K d of 0.3 Ϯ 0.1 M compared with 0.22 Ϯ 0.04 M for P. falciparum MSP1 42 . MSP1 33 from P. vivax showed a high affinity site that was 5-fold weaker compared with MSP1 42 (K d ϭ 1.5 Ϯ 0.5 M). The ⌬H values analyzed by isothermal titration calorimetry (11 Ϯ 1 kcal/mol for MSP1 42 and 12 Ϯ 1 kcal/mol for MSP1 33 ) are also very similar. At the high concentrations (Ͼ30 M) required for isothermal titration calorimetric analysis, there was evidence for some nonspecific binding. The suramin binding for MSP1 33 was only 5-fold weaker than that for MSP1 42 ; and this, taken together with the similar large induced fluorescence enhancements, indicates that there is a similar hydrophobic suramin-binding pocket in the two proteins.
Suramin Binds to MSP1 19 -No fluorescence enhancement was observed when suramin was added to MSP1 19 from P. falciparum; and thus, suramin binding to MSP1 19 could not be determined by fluorescence measurements. However, using NMR, it was possible to follow changes in chemical shifts of the 1 H and 15 N signals in 1 H-15 N heteronuclear single quantum correlation spectroscopy experiments when titrating 15 N-labeled MSP1 19 with suramin. At the maximum concentration of suramin used (16 mM), ϳ50% of MSP1 19 was complexed with the ligand. The residues that showed the largest shifts upon addition of suramin were Ile 2 , His 5 , Phe 19 , His 21 , Leu 22 , and Arg 25 . The shifts were fitted by nonlinear regression analysis to a single binding curve and gave an average K d of ϳ15 Ϯ 5 mM (Fig. 6). Suramin seems to bind weakly to a site or sites near the N terminus of the protein, as indicated in Fig. 7, which shows the positions of the affected MSP1 19 residues in a threedimensional structure. All of the observed shifts were very small, Ͻ0.15 ppm for the NH 1 H signals. The chemical shift of the protein aromatic ring protons showed almost no change, and the protons on suramin itself showed shifts of Ͻ0.05 ppm. These small changes indicate that suramin does not bind to MSP1 19 in a hydrophobic pocket (see "Discussion").
Suramin Analogues Inhibit Erythrocyte Invasion and MSP1 Secondary Processing-A number of suramin analogues were examined to probe the features of the molecule necessary for binding to MSP1 and for inhibiting P. falciparum MSP1 processing. Four series of symmetrical compounds (Groups A-D)  (Fig. 8, A and D). However, all Group B and C compounds inhibited invasion in vitro (Fig. 8, B and C), with IC 50 values similar to that of suramin. The suramin analogues differ in toxicity (16), with the least toxic inhibitory compound (B1) being ϳ10 times less toxic than suramin. 3 All of the compounds that inhibited invasion also inhibited MSP1 42 processing (Fig. 9, B and C), whereas those that did not inhibit invasion also did not inhibit processing (Fig. 9, A and D) when tested at a concentration of 200 M. Similar parasite invasion and MSP1 processing assays could not be carried out for P. vivax because this species cannot be cultured in vitro. DISCUSSION Sulfated polyanions have been shown to have a number of inhibitory effects on malaria parasites in vitro. For example, merozoite invasion of erythrocytes and sporozoite invasion of hepatocytes are inhibited by sulfated glycans (11,12,(23)(24)(25), and suramin has been shown to inhibit invasion of HepG2 cells by P. falciparum sporozoites with an IC 50 of 50 M (11). The molecular targets of these agents are unclear, although the 135-kDa Duffy-binding ligand of Plasmodium knowlesi has been shown to bind to fucoidan (26), and TRAP (thrombospondin-related anonymous protein) has been implicated in the effects of suramin on sporozoite invasion (11).
Here, we have demonstrated that suramin prevented merozoite invasion in a dose-dependent manner with an IC 50 (ϳ60 M), which is similar to that at which the compound inhibits sporozoite invasion (11). In addition, suramin bound to MSP1 42 and inhibited secondary processing of MSP1 on the merozoite surface at the time of invasion. Previous studies have also 3 J. Walker, unpublished data. shown that inhibition of MSP1 secondary processing results in inhibition of erythrocyte invasion. For example, the protease responsible for MSP1 secondary processing is inhibited by calcium-chelating agents such as EGTA and EDTA and by PMSF (a broad-range serine protease inhibitor) (27,28) and by a specific peptidyl chloromethyl ketone based on the cleavage site within MSP1 42 . 4 All of these agents also inhibit erythrocyte invasion. Furthermore, certain monoclonal antibodies that bind to the C-terminal region of MSP1 42 inhibit both processing and erythrocyte invasion (17,19). To date, all of the data obtained using either small inhibitor molecules or antibodies support the proposition that MSP1 secondary processing is essential for invasion, validating the processing as a chemotherapeutic target. The secondary processing protease has not yet been identified, but has been characterized as a calciumdependent, membrane-bound serine protease (28) with candidates including members of the subtilisin family (29 -31). There is no direct evidence that suramin can act as an inhibitor of the specific protease, but suramin is known to inhibit other proteases, e.g. neutrophil serine proteinase (32). It is unlikely that suramin acted by chelating calcium, as calcium was present in the processing assays at a concentration at least 10 times that of the compounds. In the absence of any direct information about protease inhibition, our characterization of the direct binding of suramin to MSP1 42 , inhibition of secondary processing, and erythrocyte invasion offers the most plausible mechanism for the observed antimalarial action.
To investigate the mechanism by which suramin inhibits processing and invasion, it was necessary to examine two species of plasmodia, P. falciparum and P. vivax, because all of the relevant experiments could not be done with either species alone. Fluorescence measurements were used to examine suramin binding to MSP1 42 and to the two polypeptide fragments (MSP1 19 and MSP1 33 ) that result from its secondary processing. Binding experiments using MSP1 42 from both P. falciparum and P. vivax indicated that suramin has a similar binding constant for the two species. Binding studies were also carried out with P. vivax MSP1 33 (aggregation problems prevented such studies with P. falciparum MSP1 33 ). Suramin bound to MSP1 42 with a K d of 2 ϫ 10 Ϫ7 M and only five times more weakly to the MSP1 33 fragment (K d ϭ 1.5 M), but bound ϳ10 5 times more weakly to the MSP1 19 fragment (K d ϳ 15 Ϯ 5 mM). The estimate of binding to MSP1 19 was from NMR studies, as it was not possible to use the fluorescence approach.
The NMR spectral assignments and solution structure of MSP1 19 have been determined previously (5). The structure contains two EGF domains folded into a U-shaped structure such that the N and C termini are close to each other. Some of the residues affected by suramin binding, notably His 5 , Phe 19 , and Leu 22 , are located in the region between the two EGF domains and near the closely apposed N and C termini of MSP1 19 (5). Suramin contains many aromatic rings, and these would be expected to cause large ring current chemical shifts of the signals from protein residues in any hydrophobic binding site. Such shifts were not observed. The very small changes in chemical shift observed both for the NH signals from the protein as well as the suramin signals themselves indicate that suramin does not bind to a hydrophobic pocket in MSP1 19 . The most likely mechanism for suramin binding involves electrostatic interaction of its negatively charged groups with positively charged residues on MSP1 19 (Fig. 7). Such electrostatic interactions could be affected without the naphthyl aromatic rings approaching protein nuclei too closely. Candidate residues for such electrostatic interaction are His 5 , His 21 , and Arg 25 . They exhibit perturbed chemical shifts in the complex and are located near the other affected residues. All of the perturbed residues are from the first domain of MSP1 19 and lie close to the processing site at the N terminus (Fig. 7). Two monoclonal antibodies that inhibit parasite invasion and secondary processing in vitro also interact with binding sites, including the suramin-binding region of MSP1 19 (33).
It is unlikely that the binding of suramin to MSP1 depends exclusively upon electrostatic interactions because other ligands with negatively charged sulfonate groups (such as NTS and the Group A and D compounds) are not effective inhibitors of MSP1 processing. The large fluorescence enhancements seen upon suramin binding to P. falciparum MSP1 42 and P. vivax MSP1 42 and MSP1 33 suggest that suramin binds to a hydrophobic region of MSP1 42 and MSP1 33 . The similar K d and ⌬H values obtained for MSP1 42 and MSP1 33 suggest that the hydrophobic binding site for suramin is similar in the two proteins. Although suramin binding to MSP1 42 requires residues from both regions of MSP1 42 , it is clear that the EGF domains of MSP1 19 make only a small contribution to the overall strength of the interaction. A working model of the structure for the complex of suramin with MSP1 42 is shown in Fig. 10, which indicates the general features of a strong hydrophobic binding site on the MSP1 33 fragment and a weaker electrostatic interaction with MSP1 19 .
Although suramin and some related compounds inhibit secondary processing and invasion, several related molecules do not. For example, whereas all Group B and C compounds inhibited both MSP1 secondary processing and merozoite invasion, NTS as well as the Group A and D compounds did not. We observed no examples of compounds that inhibited processing but not invasion or vice versa. This direct correlation between the two activities provides further evidence that MSP1 secondary processing is an essential requirement for successful erythrocyte invasion.
The most active suramin-like compounds are those in Groups B and C with two or four bridging aminobenzoylurea groups. These molecules have rigid symmetrical structures, and it is possible that either the presence or length of these aminobenzoylurea groups is essential for binding to MSP1. The position of the sulfonate groups and other substitutions of the naphthyl rings have little effect on the activity relationships of the compounds. A study of basic fibroblast growth factor inhibition in vitro and anti-angiogenic activity in vivo using a similar series of compounds (16) concluded that the extended multiple ring structure is important for activity and that the substitution pattern of the naphthyl rings is irrelevant. From the series of 4  19 compounds we examined, only suramin, NTS, and compound A6 possess three sulfonate groups/naphthyl ring, whereas all of the other compounds have two; therefore, it is unlikely that the exact number of sulfonate groups is important for activity. The methyl groups of suramin are absent in the analogues we examined, suggesting that they are not important for inhibiting processing or invasion.
In vivo studies with suramin have shown that Ͼ99% of suramin is bound to plasma proteins such as albumin, which reduces the effective free suramin in circulation; its slow release from these complexes accounts for its prolonged half-life of 40 -50 days (34). Plasma concentrations of suramin as high as 300 g/ml (ϳ200 M) are clinically achievable without significant toxicity (35) This concentration is only 3-4-fold higher than the IC 50 for suramin inhibition of merozoite invasion in vitro; and therefore, achievable levels of suramin are unlikely to be sufficient to completely prevent merozoite invasion in humans. However, some of the suramin analogues examined here are 10 times less toxic than suramin, and further analysis of structure-activity relationships may identify more active compounds with a more favorable therapeutic index.
In summary, this work provides evidence that MSP1 secondary processing is essential for invasion and can be inhibited by suramin and its analogues. Secondary processing has the potential to be a new target area for antimalarial drug development. Our results also exemplify the novel concept of an inhibitor binding to the substrate of a protease to prevent proteolysis. This may be exploited to develop a fluorescencebased competitive high throughput screen to identify other potent inhibitors of MSP1 processing.