Synthesis and Antiplasmodial Activity of Bisindolylcyclobutenediones

Malaria is one of the most dangerous infectious diseases. Because the causative Plasmodium parasites have developed resistances against virtually all established antimalarial drugs, novel antiplasmodial agents are required. In order to target plasmodial kinases, novel N-unsubstituted bisindolylcyclobutenediones were designed as analogs to the kinase inhibitory bisindolylmaleimides. Molecular docking experiments produced favorable poses of the unsubstituted bisindolylcyclobutenedione in the ATP binding pocket of various plasmodial protein kinases. The synthesis of the title compounds was accomplished by sequential Friedel-Crafts acylation procedures. In vitro screening of the new compounds against transgenic NF54-luc P. falciparum parasites revealed a set of derivatives with submicromolar activity, of which some displayed a reasonable selectivity profile against a human cell line. Although the molecular docking studies suggested the plasmodial protein kinase PfGSK-3 as the putative biological target, the title compounds failed to inhibit the isolated enzyme in vitro. As selective submicromolar antiplasmodial agents, the N-unsubstituted bisindolylcyclobutenediones are promising starting structures in the search for antimalarial drugs, albeit for a rational development, the biological target addressed by these compounds has yet to be identified.


Introduction
Occurring mainly in tropical and subtropical areas, malaria remains one of the most threatening infectious diseases on earth. Transmitted by the bite of an infected female Anopheles mosquito, malaria is caused by pathogenic protozoa of the genus Plasmodium, wherein the species P. falciparum is the most dangerous pathogen. It is responsible for malaria tropica, the most severe form of the disease, which can be fatal if left untreated. It is estimated that there were 229 million malaria cases in 2019, of which about 409,000 were fatal, most of them in sub-Saharan Africa [1]. Besides its impact on individual and public health, malaria causes enormous economic damage and is linked to the endemic poverty of the affected countries [2].
Plasmodia are intracellular protozoa whose complex life cycles are divided into two phases: asexual reproduction that occurs in vertebrates and sexual reproduction in the vector, the Anopheles mosquito [3]. As vector control and prevention measures, the World Health Organization promotes insecticide-treated bednets and the use of insecticides for regular indoor spraying. Implementation of these procedures has resulted in a significant decrease in infection number in the recent two decades [4]. Unfortunately, this progress has come to a standstill in recent years [1,5,6]. To treat severe malaria in countries with high resistance rates to conventional malaria drugs, the WHO recommends artemisininbased combination therapies as the medication of choice. Increasing resistance to all classes of antimalarial agents, including artemisinin derivatives, poses a threat to endemic regions. A reason for the emerging resistance is the improper use of antimalarial drugs, which exerts selection pressure on the parasites [4,[7][8][9][10].
For the treatment of infections with resistant parasites, new drugs are needed. To minimize the risk of cross-resistance, these new medications should be distinguished from established chemotherapeutics by mode of action and chemical structure. In this regard, it would be favorable to address alternative biological targets in the parasites. The plasmodial kinome offers such potential alternative targets [11][12][13]. In the genome of P. falciparum at least 65 kinases can be identified that belong to established eukaryotic protein kinase groups, many of which are similar to human protein kinases [12,14]. Nevertheless, significant phylogenetic divergence between the mammalian and the plasmodium kinome results in mechanistic and structural differences of the enzymes which may be exploited for drug design strategies. Basic reverse genetics experiments and phosphoproteomic analyses resulted in a list of plasmodial kinases that are essential either in erythrocytic schizogonia or in the sexual developmental stage of parasites [14]. Thus, plasmodial protein kinases have been suggested as drug targets [11][12][13]. However, while the inhibitor of the plasmodial phosphatidylinositol 4-kinase (PfP4K) MMV390048 is currently studied in clinical trials [15,16], compounds acting on protein kinases of the parasites have not yet reached this status [6].
Based on the docking studies, the protein kinase Pf GSK-3 [27][28][29][30] was assumed as a putative target structure for the antiplasmodial compounds within the parasites. However, in vitro enzyme inhibition assays refuted this assumption. The relevant biological target structure for bisindolylcyclobutenediones in plasmodia is therefore still unknown and has yet to be elucidated.

Molecular Docking
The unsubstituted 3,4-bis(indol-3-yl)cyclobut-3-ene-1,2-dione 2a was docked into known structures of plasmodial protein kinases using the GOLD program [31], whereby no constraints were defined and the presence or absence of water molecules at suitable positions of the ATP binding pocket was permitted by the application of the "toggle" function. The plasmodial protein kinase structures were downloaded from the protein data bank [32], if available. The protein kinases Pf GSK-3 (homology model [33], based on template 1J1B [34]), Pf map-2 (PDB: 3NIE), Pf PK5 (PDB: 1V0P [35]), Pf CDPK2 (PDB: 4MVF) Pf lammer (PDB: 3LLT) and Pf PK7 (PDB: 2PML) were used as templates. The obtained docking results were assessed by visual inspection, and the most probable orientations are shown below. The structure of 2a was modified taking into account the suggested orientations to optimize the bioactivity. Furthermore, it was intended to generate clues regarding the relevant biological targets by comparing the structure-activity relationships in the compound class with the orientations in the six plasmodial protein kinases suggested by the docking analyses [36].
Since no experimentally determined spatial structure of Pf GSK-3 exists at present, a homology model of the Pf GSK-3 was selected for the investigations described here, which had been created using a crystal structure of human GSK-3 [33]. In the depicted pose, an indole ring of 2a occupies the adenine region and interacts with the hinge region via a hydrogen bond. The other indole ring is orthogonally arranged and forms a hydrogen bridge to the protein. Furthermore, a keto group of the ligand acts as an H-bond acceptor and interacts with the side chain of Lys166. The cyclobutenedione ring is positioned in the sugar region. In this docking pose, the ligand fills the binding pocket very well ( Figure 2a).

Pf map-2
In the docking experiment with 2a in Pf map-2 (plasmodial mitogen-activated protein kinase 2; PDB: 3NIE), only one orientation of the molecule in the ATP binding pocket was identified, in which, however, no hydrogen bonds are formed to the hinge region ( Figure 2b). An indole ring occupies parts of the ribose and adenine areas and forms a hydrogen bond to an aspartyl residue in the front region of the binding pocket. The second indole element is located in the front part of the binding pocket. One of the carbonyl groups forms a hydrogen bridge to catalytic lysine. Overall, this predicted pose appears unfavorable for a tight binding of the ligand.

PfPK7
The docking orientation of 2a in PfPK7 (2PML) shows that an indole ring is planar to the cyclobutenedione element and lies deep in the binding pocket. The other indole ring is inclined by about 45° relative to the cyclobutenedione element and is placed in the sugar-binding region of the active center. 2a forms two hydrogen bonds in this pose, but these are not directed towards the hinge region. The ligand does occupy the adenine binding site (Figure 2f).

Design Considerations
The structure of 2a was modified, taking into account the docking experiments' orientations to optimize the bioactivity. These structure variations were also intended to generate clues regarding the relevant biological targets from the observed structure-activity relationships. In this context, symmetrical and nonsymmetrical analogs of 2a were designed by a formal introduction of substituents in both indole units. The known bisindolylmaleimide 8 [37] was included in the study for comparison.

Chemistry
The synthesis of the bisindolylcyclobutenediones 2a-2ar was based on 3,4-dichlorocy-  The docking of 2a in the plasmodial kinase Pf lammer (PDB: 3LLT) resulted in an orientation of the cyclobutenedione ring towards the Lys-Glu salt bridge. One of the two carbonyl groups forms a hydrogen bridge to the conserved lysine in the rear area of the binding pocket. While direct bonds to the hinge region are lacking, one of the indole nitrogen atoms forms an interaction with this region mediated by a water molecule (Figure 2e).
The docking orientation of 2a in Pf PK7 (2PML) shows that an indole ring is planar to the cyclobutenedione element and lies deep in the binding pocket. The other indole ring is inclined by about 45 • relative to the cyclobutenedione element and is placed in the sugar-binding region of the active center. 2a forms two hydrogen bonds in this pose, but these are not directed towards the hinge region. The ligand does occupy the adenine binding site (Figure 2f).

Design Considerations
The structure of 2a was modified, taking into account the docking experiments' orientations to optimize the bioactivity. These structure variations were also intended to generate clues regarding the relevant biological targets from the observed structure-activity relationships. In this context, symmetrical and nonsymmetrical analogs of 2a were designed by a formal introduction of substituents in both indole units. The known bisindolylmaleimide 8 [37] was included in the study for comparison.

Chemistry
The synthesis of the bisindolylcyclobutenediones 2a-2ar was based on 3,4-dichlorocyclobut-3-ene-1,2-dione (squaric acid dichloride, 4), which is readily obtained from squaric acid (3) and oxalyl chloride [38]. It has been reported that 3,4-bisarylcyclobutene-1,2-diones can be prepared by a reaction of 4 with two equivalents of an aromatic compound in the presence of aluminum chloride [39]. However, this method is only suitable for symmetrical compounds and resulted in very low yields of product 2a. Better results were gained by a sequential process. For this method, the dichloride 4 was first reacted with one indole equivalent [40]. Subsequent reaction with a second equivalent of the same or another indole resulted in better yields and enabled the production of asymmetric compounds with two differently substituted indole units. A further improvement was achieved by strengthening the nucleophilicity of the indoles 6 by conversion into magnesium derivatives 7, a procedure that has already been described for the production of asymmetric bisindolylmaleimides [41]. The indole magnesium intermediates 7 were then transformed into the title compounds by reaction either with squaric acid dichloride 4 or monoarylated intermediate 5 (Scheme 1). The hydroxylated compound 2ap was produced by ether cleavage with boron tribromide from the methoxy-substituted analog 2ad (Scheme 2).

Biological Evaluation
All synthesized new bisindolylcyclobutenediones 2a-2ar and the related bisindolylmaleimide 8 [37] were screened for inhibition of erythrocytic stages of Plasmodium falciparum employing a bioluminescence-based viability assay on transgenic parasites that constitutively express luciferase (Pf NF54-luc) [30]. Initially, the compounds were evaluated in single-dose test runs (3 µM). If >20% viability inhibition was observed, dose-response curves were generated and IC 50 values were calculated (Table 1). Depending on the substitution pattern on the indole subunits, several congeners proved to be rather potent, exhibiting considerable inhibition in single-digit micromolar or in submicromolar concentrations. The bisindolylmaleimide 8 [37] inhibited the parasites in a single-digit micromolar concentration, similar to many of the bisindolylcyclobutenediones in the study. Because the bisindolylcyclobutenediones 2 display a higher degree of structural novelty, only this compound class was further pursued for its antiplasmodial activity. Some among the active compounds displayed substituents at the indole scaffolds in 1, 2, 5-or 7-position, respectively. These structural features were considered unfavorable for orientations as illustrated in Figure 2 with 2a and Pf map-2, Pf PK5, Pf CDPK2, Pf lammer, and Pf PK7, respectively. However, a structure generated for 2a in Pf GSK-3 in which the indole units are oriented perpendicular to each other (Figure 2a) was considered compatible with the indicated substitution pattern. Consequently, the potent antiplasmodial derivatives 2k, 2ad, and 2ae were tested in a kinase Glo assay [42] for inhibition of recombinant Pf GSK-3. However, these exemplary structures were inactive up to a concentration of 10 µM, so the biological target responsible for the antiplasmodial activity of the title compounds has not yet been identified. For the compounds with particularly potent antiplasmodial activity, cytotoxicity was also determined using the human cell line THP-1 to assess selectivity with respect to parasite cells ( Table 2). In these studies, representative 2ai was found to be particularly beneficial with a selectivity index of 259.

Discussion
In view of the increasingly critical resistance situation in drug therapy for malaria, the development of additional active agents is urgently required. These agents should be structurally distinct from established antiplasmodial drugs and directed against previously unaddressed biological targets. Representing a well-studied enzyme class, the plasmodial protein kinases comprise entities which are essential for parasite development and replication. Given the high number of protein kinase inhibitors approved as anticancer drugs, protein kinases are generally accepted as "druggable". Hitherto, the compound class of 3,4-bis(indol-3-yl)cyclobut-3-ene-1,2-diones 2 has been neglected in the search for biologically active compounds. This is surprising since the compounds show structural

Discussion
In view of the increasingly critical resistance situation in drug therapy for malaria, the development of additional active agents is urgently required. These agents should be structurally distinct from established antiplasmodial drugs and directed against previously unaddressed biological targets. Representing a well-studied enzyme class, the plasmodial protein kinases comprise entities which are essential for parasite development and replication. Given the high number of protein kinase inhibitors approved as anticancer drugs, protein kinases are generally accepted as "druggable". Hitherto, the compound class of 3,4-bis(indol-3-yl)cyclobut-3-ene-1,2-diones 2 has been neglected in the search for biologically active compounds. This is surprising since the compounds show structural similarity to bisindolylmaleimides 1, which are established protein kinase inhibitors. Therefore, we investigated whether the unsubstituted prototype 2a could be fitted into the ATP-binding pockets of plasmodial protein kinases by a docking routine. In these experiments, structures of plasmodial kinases from the Protein Data Bank were used as templates. A commonly observed binding motif generated by docking showed in some cases (Pf GSK-3, Pf Pk5, Pf CDPK2) hydrogen bonding between the hinge region of the kinase and the potential ligands, with at least one of the indole nitrogen atoms acting as a hydrogen bond donor ( Figure 2). A literature survey revealed that bisindolylcyclobutenediones without substituents on one or both nitrogen atoms were previously unknown. A series of corresponding compounds was therefore synthesized. Sequential Friedel-Crafts reactions of squaric acid dichloride (4) with two differently substituted indole systems provided access to unsymmetrically substituted title compounds. Since the biological target in plasmodia was not a foregone conclusion, the new compounds were tested in phenotypic screening. Phenotypic screenings have the advantage over target-oriented methods in that the antiparasitic effect can be achieved by attacking different, even initially unknown, target structures. In addition, positive results from phenotypic screenings ensure that the active ingredients can overcome the cell membrane of the parasites and remain there intracellularly long enough to exert their effect [43][44][45][46][47][48][49].
Testing of the unsubstituted prototype 2a in a viability assay on genetically modified parasites constitutively expressing luciferase resulted in inhibition of the parasites in the single-digit micromolar concentration range. Although bisindolylmaleimide 8 [37] inhibited the parasites at comparable concentrations, we decided rather to further investigate the structurally novel 3,4-bis(indol-3-yl)cyclobut-3-ene-1,2-diones 2. Substitution at the 5-position on one of the indole rings of 2a with halogen or cyano substituents led to slightly improved results (2b, 2c, 2f). The introduction of a 5-iodo substituent even shifted the antiplasmodial activity into the submicromolar range. In contrast, a symmetric 5,5disubstitution with halogen, cyano, or methoxy substituents resulted in lower inhibitory activities (2n, 2o, 2p, 2aq).
While methyl substitution at both indole nitrogen atoms (compound 2i) significantly decreased the antiplasmodial activity, single N-methylation gave improved results in some cases (2q, 2u, 2y). The otherwise unsubstituted representative 2k with a basic side chain on one of the two indole nitrogen atoms also showed a solid antiplasmodial efficacy. In contrast, diminished activity was observed when N-methylation at one of the indole nitrogen atoms was combined with a substitution at the 5-position of the other indole ring (2t, 2af, 2ag). While most representatives with 2-substituted indole rings were hardly active (2g, 2l, 2s, 2z, 2aa), some congeners with 2-phenylindole substitution still showed activity (2h, 2m, 2u). A benzyloxy substituent on one of the two indole rings prevented potent antiplasmodial activity of the corresponding derivatives 2e, 2ar.
The influence of a methoxy substituent in the 5-position at one of the two indole elements connected with the substitution at the other indole ring was striking but inconsistent. While the bromo substitution at the second indole ring in 4-and 6-position reduced the antiparasitic effect (2am, 2an), the 5-or 7-bromo isomers exhibited strong antiplasmodial activity in the submicromolar concentration range (2ad, 2aj). Similarly, the 5-methoxy-substituted derivatives with chloro (2ac, 2ai) or iodo substitution (2ae, 2ak) in 5-or 7-position on the other indole ring also showed strong antiparasitic activity. The 7-substituent could also be replaced by an ethyl residue (2al).
The question arose whether a hypothesis about a protein kinase as a biological target in plasmodia could be derived from the observed structure-activity relationships. Substitution in the 2-position on one or both indole rings prevents a planar arrangement of these structural elements and would be incompatible with the predicted orientations of the compounds in Pf PK5, Pf CDPK2, and Pf lammer (Figure 2). The structure-activity relationships derived from these observations present an inconsistent picture when viewed as a whole, making it doubtful whether plasmodial kinases are indeed the biological targets of this new class of drugs. However, it is clear that monosubstitution at one of the two nitrogen atoms is possible, whereas substitution at both indole nitrogen atoms reduces the activity. This argues against both indole nitrogen atoms interacting simultaneously with the hinge region of a plasmodial kinase. The combined substitution pattern on both indole rings also indicates that in the biologically active conformation, the two indole rings are most likely not aligned planar to each other.
Of the enzyme-ligand complexes shown in Figure 2, the structure of Pf GSK-3 is most consistent with the above observations. Using some 5-bromosubstituted analogs (2b, 2t,  2ag, 2ao) as examples, this possibility was further investigated by docking experiments. The 5-monobrominated compound 2b (IC 50 P. falciparum = 3 µM) can be unconstrainedly docked into the homology model of Pf GSK-3, with the bromine atom filling a hydrophobic pocket in the backspace of the ATP-binding site. An additional 5-methoxy group on the second indole ring advantageously occupies a small binding pocket in the upper region of the ATP-binding site, which could explain the enhanced inhibitory activity of this compound (2ad, IC 50 P.f = 0.47 µM) ( Figure S1). In this orientation, N-methylation on the methoxy-substituted indole ring would also be compatible (2ao, IC 50 P.f = 0.54 µM).
In contrast, N-methylation on the bromine-substituted indole ring would prevent its interaction with the hinge region, which could account for the inactivity of this compound (2ag, IC 50 P.f > 10 µM).
In light of these considerations, it was hypothesized that Pf GSK-3 could represent the biologically relevant target structure in plasmodia. However, this assumption was refuted by testing three particularly potent compounds on isolated recombinant Pf GSK-3 enzyme. At a concentration of 10 µM, the particularly potent antiplasmodial agents 2k, 2ad and 2ae did not inhibit Pf GSK-3. Therefore, the target structure responsible for the antiplasmodial activity of the substance class must be identified in further studies. In this regard, other mechanisms of action and other targets (e.g. hemoglobin degradation, hemozoin formation, β-hematin inhibition) should be considered which already have been identified for indole-derived antimalarial agents (recently reviewed by Li et al. [50]).
When used as drugs, antimicrobial agents should, as far as possible, only damage the pathogen cells without affecting the viability of the host cells. Therefore, to obtain an overview of selectivity in this class of compounds, a selection of particularly potent antiplasmodially active compounds were evaluated using the human cell line THP-1 ( Table 2). While most compounds showed only weak selectivity, the three compounds 2ad, 2ae, and 2ai turned out to be highly selective with selectivity indices ranging from two to three digits.
The bisindolylcyclobutenediones are compounds with multiple aromatic ring systems and a high fraction of sp 2 -hybridized carbon atoms, which also typically exhibit low solubility in aqueous media due to the solid crystal lattice structure. For compounds intended to be used as drugs, this often results in poor absorption after oral administration and thus low bioavailability. Further studies of this class of compounds must therefore be directed not only at elucidating the molecular mechanism of action of the antiplasmodial activity. Rather, derivatives must be prepared that have a higher fraction of sp 3 -hybridized carbon atoms and exhibit optimized solubility.
Nevertheless, the pronounced antiparasitic activity, novelty value of the structural class, and selectivity versus human cells warrant further exploration of these antiplasmodial agents.

General Procedure A for the Syntheses of 3-Chloro-4-(1H-indol-3-yl)cyclobut-3 -ene-1,2-dione Derivatives 5a-f
The syntheses were performed following an adapted protocol initially developed by Schmidt et al. [40]. To squaric acid dichloride (4) (1.51 g, 10.0 mmol) [38] dissolved in dry diethyl ether (25 mL) and cooled to 0 • C, a solution of the appropriate indole (12.0 mmol) in the same solvent (25 mL) is added dropwise. The mixture is allowed to warm to room temperature and stirred for twenty-four hours. The resulting precipitate is filtered, washed with diethyl ether, and crystallized from chloroform/N,N-dimethylformamide. The product is dried at 110 • C in vacuo.

General Procedure B for the Syntheses of 3,4-Bis(indol-3-yl)cyclobut-3-ene-1,2-diones 2l, 2m, 2p
With reference to Green et al. [39] and Matsuoka et al. [26], the 3,4-bis(indol-3yl)cyclobut-3-ene-1,2-diones 2l, 2m, 2p are synthesized by the following procedure: To a suspension of 3,4-dichlorocyclobut-3-ene-1,2-dione (4, 0.300 g, 2.00 mmol) and anhydrous aluminum chloride (800 mg, 6.00 mmol) in dry dichloromethane, the appropriate indole (4.00 mmol) is added. The mixture is stirred for 10-65 h at room temperature with moisture protection. Subsequently, the reaction suspension is stirred in a mixture of water (50 mL) and five drops of 37% aqueous hydrochloric acid. The resulting precipitate is separated and washed with water. After drying, the residue is dissolved in acetone (10 mL) and eluted through a column of acidic aluminum oxide with acetone. The yellow or tawny fraction is collected and evaporated under reduced pressure. The solid residue is crystallized from an appropriate solvent or purified by silica gel column chromatography. The resulting product is dried at 80-90 • C in vacuo.
The aqueous layer is extracted with ethyl acetate (3 × 20 mL). After drying the combined organic layers with anhydrous sodium sulfate, the solvent is removed under reduced pressure. Subsequently, the resulting residue is crystallized from a suitable solvent and dried at 80-90 • C in vacuo.

General Procedure E for the Syntheses of 3,4-Bis(indol-3-yl)cyclobut-3-ene-1,2dione 2aq
With reference to Faul et al. [41] the 3,4-bis(indol-3-yl)cyclobut-3-ene-1,2-dione derivative 2aq is synthesized by the following procedure. To a stirred solution of the appropriate indole (5 mmol) in anhydrous toluene (8.35 mL) is added ethyl magnesium bromide (3.0 M in diethyl ether, 1.67 mL) under nitrogen. After heating to 60 • C for one hour, a solution of squaric acid dichloride (4) in dry tetrahydrofuran (1.64 mL) is added slowly. Subsequently, the reaction mixture is refluxed for four h under nitrogen, cooled to room temperature, and diluted with ethyl acetate (25 mL). The organic layer is washed successively with 1 M aqueous hydrochloric acid (15 mL), water (2 × 15 mL), and brine (15 mL) and dried with anhydrous sodium sulfate. After evaporation under reduced pressure, the product is crystallized from the given solvent and then dried at 80-90 • C in vacuo.

Molecular Docking
For docking studies, the plasmodial protein kinase crystal structures 1V0P, 4MVF, 2PML, 3LLT, and 3NIE were downloaded from the Protein Data Bank [32] and three homology models of Pf GSK-3 were selected, which had been created using the crystal structures of human GSK-3 (pdb-id: 1J1B, 2OW3 and 1R0E, respectively) [33]. After protonation optimization and energy minimization (defaults) conducted with MOE (Molecular Operating Environment, 2013.08 ed., Chemical Computing Group Inc. Montreal, QB, Canada), the .pdb files of the protein structures were used for the subsequent docking experiments. In the same way, the ligands were prepared with MOE and saved as mol2 files. The docking experiments were carried out with GOLD Suite 2.2.5 [31] without the definition of constraints. The spatial extent of the ATP binding pocket was defined as a 10 Å zone with respect to either the originally enclosed ligand or, in the homology models, with respect to the gatekeeper residue. After the relevant water molecules in the binding pocket for the automatic toggle were established, the remaining water molecules were removed. For the respective ligands, ten genetic algorithm runs were generated, the options "save lone pairs" and "allow early termination" being disabled. The docking accuracy was set to 200%. The evaluated docking solutions by the integrated ChemScore.kinase templates were stored with ChemScore values in descending order as .mol2 or .sdf files. The thus rated poses were mapped and analyzed using the visualization program Chimera [51]. By additional visual inspection, the poses were interpreted regarding meaningful orientation, considering important polar interactions, occupation of the hydrophobic back pocket, and optional formation of halogen bonds. Chimera is developed by the Resource for Biocomputing, Visualization, and Informatics at the UCSF (supported by NIFMS P41-GM103311).

Luciferase-Based Viability Screening for Antiplasmodial Activity
Screening for the antiplasmodial activity of test compounds was performed by luciferasebased viability assays as reported previously [30]. In brief, asexual erythrocytic stages of transgenic NF54-luc P. falciparum were cultured as described elsewhere [28,52]. Parasite cultures with parasitaemia of 0.5-1% were incubated in the presence of 3 µM test compounds for 48 h. After the subsequent addition of Bright-Glo substrate solution (Promega, Madison, WI), the resulting luminescence was measured. Untreated cultures (0.01% DMSO) were used as negative controls. Experiments were performed in triplicate and were repeated as a whole at least twice. Blasticidin (Sigma-Aldrich, St. Louis, MO, USA) was included as a positive control [53]. Test compounds exhibiting satisfactorily inhibition activity in 3 µM concentration (in most cases >20% inhibition of viability) were rated as actives of which IC 50 values were determined from dose-response curves.

Cytotoxicity Assay on Human THP-1 Cell Line
The cytotoxicity of selected antiplasmodial bisindolylcyclobutenediones was determined against THP-1 cells following a protocol described previously [54][55][56]. In brief, THP-1 cells were incubated with test compounds in serial compound dilutions in the presence of 0.01% DMSO at 37 • C. After 48 h, Alamar Blue was added, and after 4 h the fluorescence was read (λ ex = 544 nm, λ em = 590 nm) using a microplate reader. THP-1 cells in wells containing medium plus 0.01% DMSO served as negative controls. Experiments were performed in triplicate and were repeated as a whole at least twice.

PfGSK-3 Inhibition Assay
The Pf GSK-3 inhibition assay was carried out using the commercial KinaseGlo Plus assay (KinaseGlo Plus, Promega) as described before [42]. In brief, the kinase and a corresponding substrate (GS-1 substrate, Promega) in the presence of ATP were used to perform substrate phosphorylation. Subsequently, luminescence is generated utilizing luciferase which is a component of the KinaseGlo Reagent. The intensity of the luminescence signal corresponds to the amount of remaining ATP in the reaction mixture. For the investigation, potential kinase inhibitors were dissolved in DMSO and were added directly to the kinase reaction. The final DMSO concentration in the kinase reaction was at most 1%. As the negative control, 1% DMSO without test compound was measured. A control reaction without kinase was measured simultaneously. The difference between kinase reaction and control reaction corresponds to the kinase activity. While the published selective Pf GSK-3 inhibitor 5v [28] exhibited the expected inhibitory activity in this assay, the potent antiplasmodial bisindolylcyclobutenediones 2k, 2ad, and 2ae failed to show kinase inhibitory activity up to a concentration of 10 µM.

Conclusions
In view of the high incidence and the spreading resistance to approved drugs, new active agents against the malaria pathogens are urgently needed. Such new drugs should hit previously unaddressed targets and thus be based on novel mechanisms of action. Although plasmodial protein kinases have been frequently proposed as biological targets for malaria therapy, no drugs have yet been developed on this basis. Against this background, novel N-unsubstituted bisindolylcyclobutenediones were synthesized as analogs to the kinase-inhibiting bisindolylmaleimides using a sequential Friedel-Crafts acylation starting from squaric acid dichloride. Docking experiments on various plasmodial kinases revealed plausible orientations in the ATP-binding pocket of the enzymes. When the new compounds were screened in vitro on erythrocytic P. falciparum forms, several derivatives showed good inhibition values with favorable selectivity profiles. The 7-chloro-5'-methoxysubstituted compound 2ai proved to be the most interesting representative, showing submicromolar antiparasitic activity and a three-digit selectivity index against a human cell line. Whether inhibition of plasmodial protein kinases is responsible for the antiparasitic activity of the new compounds remains unclear for the time being. The protein kinase Pf GSK-3, initially assumed to be the target structure, was not inhibited by particularly active representatives of the new substance class. As selective submicromolar antiplasmodial agents, the N-unsubstituted bisindolylcyclobutenediones are promising starting structures in searching for therapeutic antimalarials, although for a rational development the biological target addressed by these compounds remains to be identified.

Patents
The syntheses and antiplasmodial activities of compounds reported in this article have been claimed by some of the authors in a patent application [58].
Supplementary Materials: The following are available online. Table S1: Structures of all test compounds, Table S2: 1 H NMR-spectra of test compounds displayed in Table 2, Figure S1: Docking pose of 2b (orange) and 2ad (blue) in the ATP binding pocket of a Pf GSK-3 homology model.