Antiplasmodial Properties of Acyl-Lysyl Oligomers in Culture and Animal Models of Malaria

Abstract

Our previous analysis of antiplasmodial properties exhibited by dodecanoyl-based oligo-acyl-lysyls (OAKs) has outlined basic attributes implicated in potent inhibition of parasite growth and underlined the critical role of excess hydrophobicity in hemotoxicity. To dissociate hemolysis from antiplasmodial effect, we screened >50 OAKs for in vitro growth inhibition of Plasmodium falciparum strains, thus revealing the minimal requirements for antiplasmodial potency in terms of sequence and composition, as confirmed by efficacy studies in vivo. The most active sequence, dodecanoyllysyl-bis(aminooctanoyllysyl)-amide (C₁₂K-2α₈), inhibited parasite growth at submicromolar concentrations (50% inhibitory concentration [IC₅₀], 0.3 ± 0.1 μM) and was devoid of hemolytic activity (<0.4% hemolysis at 150 μM). Unlike the case of dodecanoyl-based analogs, which equally affect ring and trophozoite stages of the parasite developmental cycle, the ability of various octanoyl-based OAKs to distinctively affect these stages (rings were 4- to 5-fold more sensitive) suggests a distinct antiplasmodial mechanism, nonmembranolytic to host red blood cells (RBCs). Upon intraperitoneal administration to mice, C₁₂K-2α₈ demonstrated sustainable high concentrations in blood (e.g., 0.1 mM at 25 mg/kg of body weight). In Plasmodium vinckei-infected mice, C₁₂K-2α₈ significantly affected parasite growth (50% effective dose [ED₅₀], 22 mg/kg) but also caused mortality in 2/3 mice at high doses (50 mg/kg/day × 4).

INTRODUCTION

Malaria is a widespread infectious disease with an estimated 250 million to 300 million clinical cases annually and 3.3 billion people at risk (13, 46). The increasing resistance of Plasmodium falciparum, the most deadly parasite, to the available antimalarial drugs further emphasizes the need to develop new strategies for antimalarial therapeutics. Naturally occurring antimicrobial peptides (AMPs) and their synthetic analogs have recently emerged as interesting tools for exploration of new antimalarial weapons (5, 14, 16, 29, 49, 50). These peptides can vary considerably in their structural and mechanistic properties (3, 22, 23, 32, 54) but typically display pronounced amphipathic and distinctly basic characters (47, 52). Their antimicrobial action is rarely mediated by interaction with stereospecific targets, such as receptors or enzymes (9, 50). Rather, nonspecific interactions, including charge and hydrophobicity, represent the main features governing their cytotoxic properties (4, 9, 10). Thus, while some peptides seem to target vital intracellular processes through direct interaction with nucleic acids (31, 42) and proteins (8, 33), most are believed to cause cell death by interfering with cytoplasmic membrane functions (2, 11, 26) by causing structural defects (27, 45, 51) and eventually allow uncontrolled leakage of essential ions and metabolites. Despite their nonspecific mode of action, such peptide-based antimicrobials can be relatively inactive on normal animal cells and erythrocytes (17, 30, 36). The basis for this discrimination appears to be related to the composition of the target cell membrane, including differences in fluidity, negative charge density, the absence or presence of cholesterol, and the presence in the peptide-susceptible organisms of a large negative transmembrane electrical potential (28, 52). Such antimicrobials present attractive advantages over most anti-infectious agents because it is extremely difficult for microbial organisms to develop effective resistance mechanisms against nonprotein targets (34, 35, 52). Moreover, chemical mimics of AMPs may overcome classical shortcomings of peptide pharmaceuticals, such as susceptibility to rapid inactivation in vivo, which significantly impede their potential systemic applications (12, 41).

Among the various classes of AMP mimics proposed, the oligoacyllysyl (OAK) class is quite interesting mainly due to its simplicity (39, 41). OAKs are composed of a relatively small number of building blocks, acyl-lysyl subunits (referred to as α_n subunits, where n specifies the number of carbon atoms in the acyl chain). Various lines of evidence suggest that the OAK approach is likely to help the development of useful anti-infective agents and could also generate valuable scientific information along the way: OAKs were shown to exert antibacterial activity in vitro and in vivo, regardless of secondary structure considerations (25, 39, 44). Their flexible chain arrangements can be made to vary in a myriad of alternatives (25, 38), which have so far generated specific antitumor (15) and antiplasmodial (37) sequences, in addition to antibacterial sequences (39, 42, 53). They also provided insights into the complex mechanisms by which AMPs affect cell viability, having the ability to target either intracellular (15, 42, 43) or membrane-linked (6, 7, 44) components.

Here we have extended our previous study (37) by attempting to generate improved antiplasmodial sequences, thus challenging the OAK methodology for its ability to help define requirements for selective antimalarial activity. We have characterized new OAKs that appear more potent and that possess substantial in vivo activity in a rodent model of malaria.

MATERIALS AND METHODS

Peptide synthesis.

The OAKs were synthesized by the solid-phase method, applying the Fmoc (9-fluorenylmethyloxycarbonyl) active ester chemistry (Applied Biosystems model 433A) essentially as described previously (38). 4-Methylbenzhydrylamine resin was used to obtain amidated compounds. 4-Aminobutyric, 8-aminocaprylic, and 12-aminododecanoic acids were protected with an Fmoc group at the N terminus prior to synthesis. The crude compounds were purified to chromatographic homogeneity in the range of >95% by reverse-phase high-performance liquid chromatography (HPLC) using a mass spectrometer (MS) (Alliance-ZQ Waters). HPLC runs were performed on a C₁₈ column (Vydac) with a linear gradient of acetonitrile (AcN) in water (1%/min); both solvents contained 0.1% trifluoroacetic acid. The purified compounds were subjected to MS analysis in order to confirm their composition and stocked as lyophilized powder at −20°C. Prior to testing, fresh solutions were prepared in water (mQ; Millipore), briefly vortexed, sonicated, centrifuged, and then diluted in the appropriate medium.

Parasite cultivation.

The K1, FCR3, and NF54 strains of P. falciparum were cultivated in complete medium (RPMI 1640 supplemented with 25 mM HEPES and 10% human serum) as described previously (20) with human red blood cells (hRBCs). The culture was synchronized by the sorbitol method (21).

Determination of IC₅₀.

Synchronized cultures at the ring stage were cultured at 1% hematocrit and 2% parasitemia in the presence or absence of increasing concentrations of the test compounds. After 18 h of incubation, parasite viability was determined by measurement of [³H]hypoxanthine (final concentration, 2 μCi/ml) incorporation into parasite nucleic acids for 6 h. Thereafter, parasite-associated radioactivity was determined using the Filtermate/Matrix 96 Direct Beta counter. The amount of [³H]hypoxanthine incorporated into the parasites' nucleic acids was compared to the amount taken up by the controls (without OAK), used to calculate the 50% inhibitory concentration (IC₅₀) by nonlinear regression fitting of the data by using the Sigmaplot software program. Statistical data for each experiment were obtained from at least two independent assays, each performed in duplicate.

Time and stage dependence of action.

Cultures at the ring stage were seeded in 24-well plates at 1% hematocrit and 2% parasitemia in plate medium (growth medium without hypoxanthine, 10 mM NaHCO₃, and 7% heat-inactivated human plasma). The test compounds were added at different concentrations and were removed after 6, 24, or 48 h of contact by washing cells once with 2 ml of complete medium. Cultures without an OAK were used as control. Parasite viability was measured by adding 2 μCi of [³H]hypoxanthine/well at time 30 h and pursuing incubation with the radioactive precursors for 24 h. Two independent experiments were performed in duplicate.

Testing of hemolytic effect.

To assess the hemolysis of infected cells, cultures were exposed to increasing concentrations of the test compounds for 24 h. The optical density in the supernatant was determined after centrifugation, and the percent lysis compared to the amount of full lysis (by water) of the cells present in the culture was calculated. The hemolysis of normal (uninfected) RBCs (see Fig. 3A) was assessed against human red blood cells after 3 h of incubation in phosphate-buffered saline (PBS) (50 mM sodium phosphate and 150 mM NaCl [pH 7.4]) at 37°C in the presence of 1% hematocrit, as described previously (38). Alternatively, hemolysis of normal RBCs was assessed at a single concentration of 150 μM tested compound according to the Antibacterial Peptides Protocols (48), where hemoglobin leakage was determined after 1 h of incubation in PBS at 37°C using a 10% hematocrit. Hemolytic activity data were obtained from at least two independent experiments.

Fig. 3.

Hemolysis and pharmacokinetics. (A) Hemolysis of human erythrocytes (1% hematocrit) after 3 h of incubation in PBS at 37°C with the specified OAK analogs. (B) Mean blood concentrations of C₁₂K-2α₈ determined by LC-MS after intraperitoneal administration of the specified OAK doses. Each time point represents values obtained from 2 mice. Limit of detection, ∼0.2 μM.

OAK organization in solution.

Eventual self-assembly of OAKs in solution was investigated by obtaining static light-scattering measurements (19) by using a Cary Eclipse fluorescence spectrophotometer (Varian Inc.). Briefly, 2-fold dilutions of the OAKs were prepared in PBS (50 mM sodium phosphate, 150 mM NaCl, pH 7.3), and the light scattering of each dilution was measured by holding both the excitation and the emission at 400 nm (slit width, 5 nm). The data represent the averages of two separate recordings.

In vivo studies.

With the exception of the antimalarial assay, all procedures, care, and handling of the animals were reviewed and approved by the Technion Institutional Animal Care and Use Committees. The antimalarial assay was designed according to French (law no. 87-848, dated 19 October 1987) and European (EEC directive 86/609, dated 24 November 1986) regulations and approved by the local ethical committee. Adult male ICR mice (weight range, 20 ± 2 g) used for pharmacokinetic and toxicity studies were obtained from Harlan Laboratories (Rehovot, Israel), housed six per cage within individually ventilated cages (IVC) (Techniplast, Italy), and acclimatized for at least 5 days before each experiment.

Toxicity.

The maximal tolerated dose (MTD) was examined after intraperitoneal (i.p.) and intravenous (i.v.) injection of C₁₂K-2α₈ into normal ICR mice. Animals were directly inspected for adverse effects for 4 h and once daily for 6 days thereafter.

Quantitative analysis in plasma.

The OAK was administered in a single i.p. injection (0.3 ml PBS) to normal pathogen-free, male ICR 4-week-old mice. The specified doses were administrated to two mice per data point in two independent experiments. At the specified time points, mice were euthanized by CO₂ asphyxiation and blood samples were collected from the portal vein and centrifuged for 2 min at 6,000 × g. From each blood sample, 0.2 ml plasma was collected and added to 1 ml of the extraction buffer (acetonitrile-formic acid, 9:1 [wt/wt], fortified with 50 mM ammonium formiate), shaken for 30 min, and recentrifuged for 2 min at 14,000 × g. Extract supernatants (150 μl) were diluted 2-fold in water containing 0.1% trifluoroacetic acid (TFA) and analyzed by liquid chromatography (LC)-MS (Waters/Micromass, Manchester, United Kingdom).

The column used was a Vydac P54 C₄ (5-μm particles, 250 by 4.6 mm). Elution was obtained using a gradient of acetonitrile (from 25 to 55% in 30 min). MS setup was as follows. The quadrupole instrument was run in ESI+ mode using the following conditions: capillary voltage of 3.5 kV, cone voltage of 32 V, source temperature at 117°C, desolvation gas temperature at 350°C, and desolvation gas flow (N₂) of 450 liters/h. Data acquisition was done in selected ion recording (SIR) mode using transitions corresponding to z = +2 and +3. For quantitative calibration, the identical procedure was performed to establish standard curves in water and mouse plasma using 0.4 to 20 μg of C₁₂K-2α₈ in 0.2 ml water or plasma. Extraction yields varied between 92% and 108%. Low limits of quantitation were 100 ng/ml in plasma/blood and 20 ng/ml in water.

Antimalarial activity.

In vivo antimalarial activities against the Plasmodium vinckei petteri (279BY; provided by I. Landau, Paris, France) strain in female Swiss mice OF1 (22 to 26g, Charles River Laboratories France) were determined. Mice were infected at day 0 (D₀) by intravenous injection in the caudal vein of 10⁷ infected erythrocytes in 200 μl 0.9% NaCl, leading to an initial parasitemia at D₁ between 0.5 and 1.5%. Mice were intraperitoneally treated once a day for 4 days at D₁, D₂, D₃, and D₄. On D₅, parasitemia levels were monitored in Giemsa-stained blood smears and by flow cytometry of blood samples (Yoyo-1 iodide [491/509]; Invitrogen) (1) to determine the dose leading to the inhibition of 50% of parasite growth (50% efficient dose [ED₅₀]). Survival of mice was monitored until at least 1 month after the end of treatment with measurement of parasitemia (recrudescence).

For treatment, the OAKs were administered by the intraperitoneal route in ≈100 μl PBS containing either 0, 5, 25, or 50 mg/kg of body weight to 3 or 4 mice/dose, as specified. Each animal was weighted the first day of the experiment in order to adjust dosage to weight. The evolution of parasitemia during the OAK treatment was evaluated by performing thin blood smears each day from D₁ or D₂, as specified (results represent the mean parasitemia/3 or 4 mice/dose).

RESULTS

Study design and structure-activity relationships.

Figure 1A depicts the primary structure of a prototypical OAK, composed of a dodecanoyllysyl subunit at the amino end linked to tandem repeats of various numbers of aminooctanoyl-lysyl subunits, all linked by amide bonds. A total of 55 derivatives were tested to assess the importance of backbone length, charge, and hydrophobicity. The sequences and their biophysical attributes are listed in Tables 1 and 2. Individual values for IC₅₀ and their hydrophobicity/charge (H/Q) characteristics are shown. While sequences listed in Table 1 were previously documented for antibacterial (6, 25, 38, 39) or antiplasmodial (37) activities, Table 2 displays new antiplasmodial sequences. All compounds were initially subjected to a screen for antiplasmodial and hemolytic activities, while a few representatives were further characterized in cultures of mammalian cells and in animal models of malaria as detailed below.

Fig. 1.

Molecular structures and structure-activity relationships of OAKs. (A) Molecular structure of a prototypical OAK sequence (C₁₂K-nα₈). Brackets define an α₈ subunit (aminooctanoyl-lysyl). (B) Effect of the α₈ subunit number on the IC₅₀ of OAK series where the N-terminal subunit ends with a hydrogen (H) or is expanded with an aminooctanoyl (NC₈), octanoyl (C₈), or dodecanoyl (C₁₂) residue. (C) Relationships between HQ values and antiplasmodial activity (H stands for hydrophobicity as determined by elution time in reversed-phase HPLC. Q stands for molecular charge at physiological pH). The HQ map of 55 OAKs was constructed by using the compounds listed in Tables 1 and 2 and their antimalarial potencies. Solid symbols (circles and triangles) represent the most potent OAKs (IC₅₀ ∼ 1 μM) as assessed against P. falciparum (highlighted in bold characters in the tables). Solid triangles represent hemolytic sequences (displaying ≥ 10% hemolysis after 3 h of incubation of 100 μM OAK with 1% RBC in PBS at 37°C). Solid circles represent selective nonhemolytic compounds.

Table 1.

Structure-activity relationships of five OAK series, tested for their inhibitory activities with cultured human RBC infected with P. falciparum FCR3^e

OAK group	OAK designation	Hb	Qc	IC50 (μM)a
1d	HK-1α12	32	3	99
	HK-2α12	40	4	15
	HK-3α12	44	5	1.6
	C12K-1α12	55	2	1.5
	C12K-2α12	53	3	0.9
	C12K-3α12	53.5	4	0.7
	NC12K-1α12	39	3	21
	NC12K-2α12	44.5	4	1.4
	NC12K-3α12	47	5	2.2
2	HK-1α8	22	3	260
	HK-2α8	27	4	180
	HK-3α8	30	5	130
	HK-4α8	32	6	90
	HK-5α8	33	7	84
	HK-6α8	33	8	71
	HK-7α8	34	9	47
3	C12K-1α8	51	2	0.4
	C12K-2α8	48	3	0.2
	C12K-3α8	46	4	1.2
	C12K-4α8	49	5	2.4
	C12K-5α8	50	6	5.6
	C12K-6α8	50	7	9.8
	C12K-7α8	48	8	16
4	C8K-1α8	39	2	16
	C8K-2α8	36	3	2.7
	C8K-3α8	40	4	15
	C8K-4α8	41	5	49
	C8K-5α8	41	6	44
	C8K-6α8	40	7	31
	C8K-7α8	40	8	37
5	NC8K-1α8	30	3	120
	NC8K-2α8	35	4	100
	NC8K-3α8	34	5	94
	NC8K-4α8	36	6	73
	NC8K-5α8	37	7	71
	NC8K-6α8	37	8	67
	NC8K-7α8	37	9	90

^aIC₅₀ represents the peptide concentration that produced 50% inhibition of hypoxanthine uptake of P. falciparum after 18 h in culture (chloroquine IC₅₀ was 0.1 μM).

^bH, hydrophobicity as determined by elution time in reversed-phase HPLC. Values were rounded to the nearest half-unit.

^cQ, molecular charge at physiological pH.

^dData were taken from reference 37.

^eThe most potent OAKs (IC₅₀ of ∼1 μM) are highlighted in bold.

Table 2.

Structure-activity relationships of five OAK series, tested for their inhibitory activities with cultured human RBC infected with P. falciparum NF54^d

OAK group	OAK designation	Hb	Qc	IC50 (μM)a
1	C12K-1α4	50	2	2.1
	C12K-2α4	47.5	3	>10
	C12K-3α4	46.5	4	>10
2	C12K-C6-α6	50	2	1
	C6-C6K-α12	42	2	5.6
	C6-C6K-C6-α6	33	2	>10
3	C12K-K-C6-α6	46	3	>10
	C6-C6K-K-α12	38	3	>10
	C6-C6K-K-C6-α6	23	3	>10
4	C12K-K -α12K	48	4	>10
	C12(ω7)K-K-α12K	47	4	>10
	C12(ω7)K-K-α12	50	3	2.7
	C12(ω7)K -α12K	50	3	1.2
5	C12K-Kα10	49	3	>10
	C12K-(Kα10)2	46	5	>10
	C12K-Kα8	47	3	>10
	C12-(Kα8)2	48.5	4	>10
	C12K-(Kα8)2	45	5	8.9

^aIC₅₀ represents the peptide concentration that produced 50% inhibition of hypoxanthine uptake of P. falciparum after 18 h in culture.

^bH, hydrophobicity as determined by elution time in reversed-phase HPLC. Values were rounded to the nearest half-unit.

^cQ, molecular charge at physiological pH.

^dThe most potent OAKs (IC₅₀ of ∼1 μM) are highlighted in bold.

The series listed in Table 1 involve OAKs based on α₁₂ and α₈ subunits. α₁₂-OAKs (group 1) were previously studied for antiplasmodial activity (37) and are shown for the purpose of comparison only. The series of α₈-OAKs (groups 2 to 5) includes variations aiming to assess the importance of the N-terminal residue and/or the importance of increasing the number of subunits (from 1 to 7) when the N terminus is constant. Various members (e.g., C₁₂K-7α₈) of this group are known to be endowed with potent antibacterial properties, yet all are devoid of hemolytic activity or of a tendency to aggregate (38, 39, 42) and were therefore particularly interesting to assess for antiplasmodial activity. The data show that the presence of an octanoyl residue at the amino end increases the antiplasmodial potency considerably (by up to 67-fold) compared with its absence (compare group 4 with group 2), but this amplification dwindles as the total length of the compound increases. Substitution of the octanoic with dodecanoic acid (group 3) resulted in a similar trend but further increased potency by up to an additional 40-fold. In contrast, substitutions with aminooctanoic acids (group 5) or aminododecanoic acids (not shown) have consistently reverted/reduced the activity considerably. It is thus interesting to notice how the nature of N-terminal acyl dictates the behavior of each series and the critical importance of a hydrophobic N terminus. Figure 1B shows the plot of IC₅₀ versus the subunit number, with an identical N terminus for each series. The plots highlight the occurrence of an optimal number of subunits, much as has been observed for antibacterial OAKs (25). For instance, comparing C₈K-nα₈ and C₁₂K-nα₈ (having identical charges), these series yielded a perfectly parallel outcome, shifted only with respect to IC₅₀s, thus emphasizing the role of hydrophobicity in defining requirements for optimal activity. These results also underline the requirement for a relatively low positive charge (+2 to +4) for potency. It is interesting to note that the antibacterial α₈ sequences (n = 5 to 7) displayed a rather weak antiplasmodial activity, and conversely, the most active sequences were devoid of antibacterial activity. This might bear mechanistic relevance, as discussed below, suggesting differential requirements for affecting viability of parasites versus bacteria.

Table 2 includes various series designed to assess or verify a particular feature. To probe the importance of charge density, we tested a subseries of analogs of the most promising α₈ series, where 8-carbon (aminooctanoyl) acyls were replaced with 4-carbon analogs (aminobutyroyl). This series (group 1) seemed promising at first (the IC₅₀ of the smallest member was 2.1 μM). However, unlike the case of the α₈ series, further increases in the number of α₄ subunits were consistently detrimental to antiplasmodial activity. The fact that hydrophobicities and overall charge were quite equivalent to the α₈ counterparts (compare with the first 3 compounds in group 3, Table 1) suggested the importance of an optimal charge density and/or overall backbone length as discussed below. In contrast to the data summarized in Fig. 1B, one can see from inspection of Table 1 that the minimal number of subunits required for potency is displaced as the intercalating acyl length increases from C₄ to C₈ to C₁₂ (i.e., optima were at n = 1, 2, and 3, respectively).

To probe the relative role of each acyl and lysyl residue, we produced and tested a new series of analogs of C₁₂K-α₁₂. Thus, the 12-carbon-chain dodecanoyl at the amino end and/or the aminododecanoyl that is intercalated between lysines was each substituted with two 6-carbon acyls linked by an amide bond, which, significantly, reduces hydrophobicity. This series (group 2) revealed/confirmed the importance of higher hydrophobicity at the N terminus, since only dodecanoyllysyl-bis(aminohexanoyl)lysyl-amide (C₁₂K-C₆-α₆) conserved the potency of C₁₂K-α₁₂, whereas both C₆-C₆-K-α₁₂ and C₆-c₆-K-c₆-α₆ displayed reduced antiplasmodial potencies. Contrarily, in a similar series (group 3) where a lysyl residue was added to the first and/or the second lysyl, none of these modifications enhanced antiplasmodial activity. Moreover, adding a lysyl to a closely related analog (group 4) where the N-terminal dodecanoyl was replaced with its unsaturated version (slightly less hydrophobic) revealed that a double lysyl (KK) motif is least tolerated when it is closest to the N terminus. Finally, a variety of additional examples based on α₁₀ or α₈ OAKs are shown in group 5, supporting the notion that an increase in charge is detrimental to antiplasmodial activity (although less so when the KK motif is far enough from the N terminus).

Collectively, these results point to a number of trends defining the minimal requirements for antiplasmodial potency and teach how the antiplasmodial activities of OAKs depend on optimal combinations of a set of properties that includes hydrophobicity, charge density, and overall backbone length. Figure 1C displays the HQ plot of all 55 compounds, showing that the most active sequences have localized to a defined HQ window. This issue will be resumed in Discussion.

Antiplasmodial properties of a representative OAK.

Dodecanoyllysyl-bis(aminooctanoyllysyl)-amide (designated C₁₂K-2α₈) was among the most active compounds, with an IC₅₀ of 0.2 μM, and was therefore selected for further characterization. The dose dependence curve of the antiplasmodial effect as tested against the strain FCR3 was highly comparable to that of the classical antimalarial drug chloroquine (Fig. 2A). Essentially similar plots were obtained against two additional strains of P. falciparum, NF54 and K1 (IC₅₀ of 0.45 and 0.18 μM, respectively; data not shown). In contrast, this OAK (and all members of this series) was highly selective against the malaria parasite in that cytotoxicity to cultured fibroblasts was extremely low, as tested at 100 μM (data not shown).

Fig. 2.

Antiplasmodial properties of selected OAKs. (A) Dose dependence curves of C₁₂K-2α₈ and chloroquine for P. falciparum FCR3, assessed by [³H]hypoxanthine incorporation as described in Materials and Methods. (B) Effects of selected OAKs on parasite viability as a function of developmental stage and time of exposure assessed by [³H]hypoxanthine incorporation. Parasites at the ring stage exposed for 6 or 24 h (white or gray background, respectively) and at the trophozoite stage exposed for 6 or 24 h (left or right slashed columns, respectively) are analyzed. The black column represents the control experiment for 48 h of exposure of rings. The IC₅₀s were calculated from 3 repeated experiments performed in triplicate.

Time and stage dependence.

The pharmacodynamic properties of drugs are essential aspects of chemotherapy. It is therefore imperative to know how long the pathogen must be exposed to the drug in order to elicit a maximal irreversible toxic effect and what is the developmental stage that is the most susceptible to the drug. The effects of three analogs were tested by exposing parasite cultures to different OAK concentrations for 6 or 24 h either at the ring or at the trophozoite stage. The OAK was then removed, and parasite viability was measured using the hypoxanthine incorporation test applied at the time interval 30 to 48 h postinvasion. The IC₅₀ for each treatment was calculated, and results for a representative strain (FCR3) are summarized in Fig. 2B. When compounds are added for 6 or 24 h to synchronized parasite cultures, IC₅₀s were lower when they were added at the ring stage than when added at the trophozoite stage, indicating that the ring stage was more sensitive to the OAK. Importantly, the toxic effect against Plasmodium was not reversible, in that removal of the drug after incubation did not allow the parasites to recover. Remarkably, adding the drug at the ring stage for 24 h leads to an IC₅₀ similar to the IC₅₀ observed after incubation of the parasite for a whole cycle of 48 h, showing that most of the antiplasmodial effect has occurred at the ring stage. A similar trend was observed with additional derivatives of this series, namely, C₁₂K-1α₈ and C₁₂K-3α₈ (Fig. 2B), or when using different strains (data not shown).

Hemolysis.

Various antiplasmodial AMPs exert potent hemolytic activity (18, 29, 40). To verify if inhibition of parasite growth is due to hemolysis, leakage of hemoglobin was assessed under the same conditions used for growth inhibition, i.e., after 24 h of incubation with the different OAKs at different concentrations. The previously characterized antiplasmodial OAK, C₁₂K-1α₁₂ (used here as a positive control) was relatively highly hemolytic, and it is plausible that it acts through lysis of the host cell, thus destroying the intracellular habitat of the parasite. In contrast, no significant lysis was caused by C₁₂K-1α₈, C₁₂K-2α₈, or C₁₂K-3α₈ at concentrations that were 2 orders of magnitude higher than their respective IC₅₀s (Fig. 3A). This suggests that the mode of parasite growth inhibition is unlikely to be mediated by destruction of the host cell. Additionally, hemolysis of normal (uninfected) RBCs was also assessed at a single very high concentration of 150 μM. Since none of the OAKs attained 50% hemolysis, their 50% lytic concentrations (LC₅₀s) were estimated by assuming a linear correlation between hemolysis and the OAK concentration. All OAKs exhibited little if any hemolytic activity in the concentration range of their antiplasmodial effect. Specifically, C₁₂K-2α₈, which displayed an average IC₅₀ of 0.3 ± 0.1 μM, presented <0.4% hemolysis at 150 μM. The ratio of the LC₅₀ for RBCs to the IC₅₀ is an indicator of drug selectivity; i.e., the higher that it is, the more selective the compound is to the parasite and the less likely it is to cause the lysis of RBCs. For the most active compound, the ratio was 18,750, implying that at the relevant antiplasmodial concentrations this compound should not be hemolytic.

Self-assembly in solution.

The aggregation properties of representative OAKs were determined by light scattering in PBS. Unlike C₁₂K-3α₁₂, which was highly aggregative, displaying a critical aggregation concentration (CAC) in the submicromolar range (38), the C₁₂K-nα₈ series were not aggregative at 100-fold-higher concentrations (data not shown), indicating that the observed activities were not biased by aggregation phenomena.

In vivo studies.

As shown in Table 3, the maximal tolerated dose (MTD) in mice for single i.v. administrations was estimated at 5 mg/kg, and mice treated with 10 mg/kg exhibited a brief period of malaise, as reflected by an observed relative immobility. In contrast, mice given i.p. administrations were all symptom free even at the dose of 25 mg/kg and therefore were selected for assessing pharmacokinetics and antimalarial efficacy.

Table 3.

Determination of MTD of C₁₂K-2α₈ in ICR mice

Dose (mg/kg)	Administration route	Observation	Survivala	MTD (mg/kg)
5	i.v.	No signs of toxicity	7/7	5
12.5	i.v.	Brief malaise signs	7/7	5
12.5	i.p.	No signs of toxicity	7/7	>25
25	i.p.	No signs of toxicity	7/7	>25

^aNo. of surviving mice/no. tested.

Figure 3B shows the rates of blood accumulation and elimination of the OAK upon i.p. administration of 3 single doses. Administration of 100 μg/mouse (i.e., 5 mg/kg) resulted in rapid accumulation in blood that reached a maximal concentration of ∼5 μg/ml (∼5 μM), which was rapidly eliminated, as suggested by the time course curve. In contrast, administration of 250 and 500 μg/mouse (corresponding to 12.5 and 25 mg/kg) resulted in higher blood concentrations (∼10 and 100 μg/ml, respectively) that were also sustained for a relatively longer time.

Antimalarial activity was evaluated against P. vinckei in mice that were infected on day 0 (D₀) and received one daily dose of OAK for 4 days (D₁ to D₄). Parasitemia was followed by using thin blood smears of treated and untreated mice each day from D₁ or D₂. The outcome of a pilot experiment scouting for dose-response is shown in Fig. 4A (n = 3 mice per group). Treatment with C₁₂K-2α₈ showed no antimalarial activity at the dose of 5 mg/kg (all mice died on D₇ with 100% parasitemia, as in the control group). At 25 mg/kg, parasite proliferation was significantly inhibited (50% parasitemia was reached at D₆, instead of D₃ as for the controls). This dose was not sufficient to clear the parasitemia but managed to delay mouse death. At 50 mg/kg, the OAK induced a potent antimalarial effect, always keeping parasitemia at a very low level (4% ± 3% at most) until D₅. On D₅, however, two mice died, suggesting chronic toxicity of the compound. Nonetheless, on day 5, the surviving mouse had a low parasitemia (2.5%, compared with 88% for the controls) before being cleared. The D₂₀ thin blood smear of the surviving mouse showed no parasite, indicating that the 4-day treatment with C₁₂K-2α₈ at 50 mg/kg had the capacity to induce a complete cure of P. vinckei infection in mice. Based on determination of parasitemia on D₅, the ED₅₀ and ED₉₀ (doses leading to the inhibition of 50 and 90% of parasites growth, respectively) were around 22 and 49 mg/kg.

Fig. 4.

Efficacy studies. Evolution of parasitemia in P. vinckei-infected mice during and after i.p. treatments. (A) Four days of treatment (D₁ to D₄) with C₁₂K-2α₈. Symbols: dashed line, vehicle; circle, triangle, and rectangle, doses of 5, 25, and 50 mg/kg, respectively (n = 3 mice/group). Note: 2 of the 50-mg/kg-treated mice died on D₅, and the third was parasite-free on D₂₀. (B) Effect of 2 days' treatment with 50 mg/kg/day of C₁₂K-2α₈ (n = 4 mice/group). Symbols: dashed line, vehicle; circle, OAK treatment (D₁ and D₂). All mice died with high parasitemia. (C) Effect of 3 days' treatment with 50 mg/kg/day of C₁₂K-2α₈ (n = 4 mice/group). Symbols: dashed line, vehicle; circle, OAK treatment (D₁ to D₃). Note: three treated mice died on D₅, D₉, and D₁₁, and the fourth was parasite free on D₂₀. (D) Four days of treatment (D₁ to D₄) with OAK analogs using 50 mg/kg/day (n = 4). Symbols: dashed line, vehicle; circle, treatment with C₈K-2α₈; rectangle, treatment with C₁₂K-1α₈. All results are expressed as means ± SEM.

An essentially similar outcome was observed in a second experiment, where the treatment regimen was reduced from 4 to 3 and 2 days of administration of 50 mg/kg/day (n = 4 mice/group). Although infection in this experiment turned out to be somewhat more virulent than in the previous one, as evident in the faster death of control mice (i.e., D₅ versus D₇), the relative efficacy and toxicity of the OAK could still be established. Thus, treatment with three administrations (D₁ to D₃) again allowed the total cure of one mouse, but its toxic effects may have been responsible for the deaths of at least two mice (Fig. 4B), whereas treatment with two administrations (D₁ and D₂) did not provide total cure but seemed to prevent toxicity (Fig. 4C).

Figure 4D shows the outcome of similar experiments where two additional groups of infected mice were treated (4 times, 50 mg/kg/day; n = 4 mice/group) with one of two analogs of C₁₂K-2α₈: C₁₂K-1α₈ or C₈K-2α₈. As shown in Table 1, these analogs displayed lower in vitro efficacies (IC₅₀s were 2-fold and >10 fold higher, respectively) and were tested here in order to corroborate the importance of the sequence length and of hydrophobicity at the N terminus, respectively. Both analogs were less efficient in vivo as well, but it is interesting to note that the potency order was maintained. Thus, parasitemia in mice treated with C₈K-2α₈ followed the same pattern as that in untreated mice, whereas parasitemia was significantly reduced after administration of C₁₂K-1α₈ (e.g., from 37 to 23% on D₃ and from 84 to 50% on D₄).

DISCUSSION

Minimal requirements for antiplasmodial OAKs.

In this investigation, we extended our study of the antiplasmodial effects of dodecanoyl-based OAKs (37) and have identified several lead compounds worth further investigation. Some of the compounds displayed very high in vitro selectivity for parasites compared to their effect on mammalian cells. The outstanding feature of the present series of compounds is their lack of hemolytic effect within the range of relevant antiplasmodial concentrations. In achieving this, we have circumvented the major handicap of dodecanoyl-based OAKs, i.e., hemolytic activity and its mechanistic repercussion on pathogen selectivity.

The most active compound, C₁₂K-2α₈, is proposed to illustrate the global minimal requirements for selective antiplasmodial activity in terms of charge density, molecular hydrophobicity, and backbone length in the sense that all compounds with similar attributes displayed potent (IC₅₀ ∼ 1 μM) and selective (nonhemolytic) activity. When the HQ values of the 55 compounds tested are plotted, this group of active compounds occupies a defined HQ area (Fig. 1C). This representation implies that OAK sequences with similar characteristics should demonstrate antiplasmodial activity. However, since this representation accounts only for the number of charges but not for their density, various inactive compounds seem to occupy the “active window,” whereas Tables 1 and 2 readily demonstrate that all the inactive compounds within this active window invariably display high charge density (e.g., they present a KK motif, or the charges are separated by short acyls). Furthermore, the active compounds within this window can be separated on the basis of selectivity. Thus, nonselective potency (IC₅₀ ∼ 1 μM) tolerates a relatively wider range of hydrophobicity and charge values (H = 40 to 55; Q = 2 to 4), but selective potency is restricted to H = 50 to 51 and Q = 2, although lower hydrophobicity values (H = 46 to 48) can be compensated with higher charge values (Q = 3 to 4 but not beyond) and as long as they remain within the confines of an optimal separation (probably around 8 carbons). By comparison, it is interesting to note that antibacterial activity was found to occupy a distinct HQ window (e.g., H = 50 ± 2 and Q = 6.5 ± 2 for E. coli) (38), reflecting specific attributes of bacteria, such as the presence of a cell wall and higher anionicity. These attributes are also probably linked to the differential optima in subunit number exhibited by each antiplasmodial series (Tables 1 and 2) compared with antibacterial OAKs (25), although exactly how is yet to be determined.

Mechanistic considerations.

The mechanism of antiplasmodial action is not understood. However, it is likely to resemble that postulated for host defense peptides, even though the data presented argue against the likelihood of membrane disruption as a mechanism for antimalarial activity. Thus, previously investigated OAKs were shown to affect cell viability by a variety of mechanisms, including targeting the membrane and intracellular organelles (as specified below). Furthermore, interference with the function of a target membrane also can be achieved in a number of ways. For instance, the antibacterial OAK sequences C₁₂K-7α₈ (39) and C₁₂K-3β₁₀ (25) abruptly disrupted the plasma membrane, leading to quasi-immediate membrane depolarization/permeabilization, and death was typically observed within minutes. In contrast, C_12(ω7)K-β₁₂ (44) and NC_12-2β₁₂ (53) were clearly devoid of membrane-disruptive properties despite their high binding affinity for reasonably suitable model membranes. They rather adhered to phospholipids in a shallow manner, thereby altering some of the fundamental membrane properties (e.g., charge density and fluidity). These modifications are likely to critically modify normal membrane-mediated interactions and functions that are important for maintaining homeostasis. Moreover, α₁₂-3β₁₂ was localized to the cytoplasmic compartment of cancer cells and interfered with mitochondrial functions (15), whereas C₁₂K-5α₈ (42) and C_16(ω7)K-β₁₂ (43) have shown the ability to interact directly with nucleic acids and to kill bacteria at lower rates. Therefore, based on the mechanistic studies performed with antibacterial/antitumor OAK analogs, it is possible that the antiplasmodial mechanism of C₁₂K-2α₈ involves interaction with intracellular targets and/or the modification of membrane properties that interfere with homeostasis. Consistent with such a mechanism are the observed relatively slow antiplasmodial kinetics (Fig. 2B), as well as the differential stage-dependent sensitivity, as opposed to the abrupt disruption of the host or of parasite membranes, characterized by faster kinetics and a similar sensitivity of ring and trophozoite stages (5, 18, 37).

Potential for systemic antimalarial efficacy.

The in vivo data suggest a potential usefulness of the OAK approach for generating systemically efficient antimalarial compounds. Development of antimalarial drugs in general and of antimalarial AMPs in particular, especially those aiming to target the blood stage of the malaria parasite cycle, must logically satisfy two preliminary requirements: they should demonstrate a lack of hemolytic activity and a sustained high concentration in the blood compartment, particularly if they are slow acting. In this regard, C₁₂K-2α₈ and probably several other OAK sequences generated during this study could theoretically be considered suitable candidates for development as systemic antimalarials. In particular, both in vitro and in vivo data support the lack of hemotoxicity of C₁₂K-2α₈, while at least upon i.p. administration, the OAK was able to sustain high concentrations in blood for at least several hours (Fig. 3B). The pharmacokinetic data revealed the rates of OAK accumulation/elimination in blood, suggesting that beyond a certain threshold of administrated dose, the resulting blood levels increased in a nonlinear fashion and were sustained for a longer time period. The relatively rapid OAK elimination at a low administration dose is unlikely to stem from susceptibility to proteases, as verified for various OAK sequences (39, 42). Thus, elimination is likely attributable to renal function or an unknown detoxification mechanism that probably underwent saturation at the high doses, hence the observed slow OAK elimination thereafter. While this phenomenon certainly requires further characterization studies to be understood, it nonetheless seems to suggest a clear correlation between concentrations in blood and antimalarial activity in that at 5 mg/kg, where the OAK blood levels were the lowest, there was no reduction in parasitemia; in contrast, a significant reduction in parasitemia was observed at ≥25 mg/kg, which correlated with the sustained high blood levels. Future optimization studies might achieve a better efficacy/toxicity profile, for instance, through OAK encapsulation in liposomal systems, which was shown to reduce systemic toxicity while increasing antibacterial efficacy (24).

In conclusion, a representative OAK, C₁₂K-2α₈, showed antimalarial activity in P. vinckei-infected mice, with a potential for complete cure of parasitemia, but presented serious toxicity signs at the highest doses. Future studies will verify whether different treatment regimens, using this OAK or a derivative thereof, will be able to reduce toxicity.