Several Human Cyclin-Dependent Kinase Inhibitors, Structurally Related to Roscovitine, As New Anti-Malarial Agents

In Africa, malaria kills one child each minute. It is also responsible for about one million deaths worldwide each year. Plasmodium falciparum, is the protozoan responsible for the most lethal form of the disease, with resistance developing against the available anti-malarial drugs. Among newly proposed anti-malaria targets, are the P. falciparum cyclin-dependent kinases (PfCDKs). There are involved in different stages of the protozoan growth and development but share high sequence homology with human cyclin-dependent kinases (CDKs). We previously reported the synthesis of CDKs inhibitors that are structurally-related to (R)-roscovitine, a 2,6,9-trisubstituted purine, and they showed activity against neuronal diseases and cancers. In this report, we describe the synthesis and the characterization of new CDK inhibitors, active in reducing the in vitro growth of P. falciparum (3D7 and 7G8 strains). Six compounds are more potent inhibitors than roscovitine, and three exhibited IC50 values close to 1 µM for both 3D7 and 7G8 strains. Although, such molecules do inhibit P. falciparum growth, they require further studies to improve their selectivity for PfCDKs.


Introduction
Malaria remains today one of the most devastating infectious diseases in the world. Despite remarkable progress in the global fight against malaria, this deadly protozoan infection takes an estimated 700,000 lives per year, mostly African children under five years of age [1].
Among the different species of protozoan parasites responsible for malaria, Plasmodium falciparum, which is transmitted to humans through the bite of infected Anopheles mosquitoes, is the most lethal [2]. Current treatments effective for malaria include chloroquine, mefloquine and artemisinin, but these drugs becoming less effective due to the gradual emergence of drug-resistant strains [3]. Artemisinin-based combination therapies have been adopted as the first-line antimalarial agents of choice against these resistant Plasmodium parasites. Recently, however, there has been increasing concern regarding the development of resistance to the artemisinins further emphasizing the need for new antimalarial agents with different mechanisms of action [4]. As a consequence of the resistance, an alarming resurgence of malaria has occurred, and the anti-malarial drug space should be urgently extended. New targets, such as the apicoplast [5], the protozoan proteases [6], or the specific mitochondrial electron transport chain are currently being investigated [7]. Protein kinases, which regulate protozoan growth and differentiation during its life cycle, have also emerged to be among the most promising new anti-malarial targets [8][9][10][11]. A short survey of the recent literature highlights the success of targeting P. falciparum kinases such as thymidinate kinase (KI = 20 µM) [12], cGMP-dependent protein kinase (IC50 = 8 nM) [13], calcium-dependent kinase 1 (IC50 in the 10-20 nM range) [14,15], and PfRIO-2 kinase, which regulates the plasmodial ribosome biogenesis [16].
The purine scaffold is widely used for the development of therapeutic agents [39,40], and it has therefore provided several CDKs inhibitors such as (R)-roscovitine (1) or purvalanol A (2a) and B (2b) ( Figure 1) [41]. Basically, (R)-roscovitine, a 2,6,9-trisubstituted purine, is a very promising molecule developed by Cyclacel Pharmaceuticals which reached phases 2 and 2b clinical trials against different types of cancers, and phase 1 clinical trial against glomerulonephritis. Interestingly, the (R)-roscovitine clinical trials pointed out the low cytotoxicity of this molecule. To summarize, a daily oral dose of 2 g was found as the maximal tolerated dose in human. Pharmacokinetics studies showed a rapid elimination of this drug with a half-life between 60 and 90 min. In humans, following oral administration, (R)-roscovitine offers good oral bioavailability and undergoes a rapid passage into the blood, distribution in tissues, and metabolism. Altogether, these results suggest that oral administration is possible for extended periods of treatment [41]. This molecule binds CDKs in the ATP pocket and is rather selective for CDK1, CDK2, CDK5, CDK7 and CDK9, but interacts with other kinases although with lower affinity (e.g., casein kinase 1 (CK1), dual specificity, tyrosine phosphorylation regulated kinases (DYRKs), pyridoxal kinase) [42,43].
In the present paper, we report the synthesis of a new series of 2,6,9-trisubstituted purines structurally-related to (R)-roscovitine, and bearing amino-heterocyclic motives in the C 2 purine ring, such as 2-aminopyrimidine (4a-d) and 5-aminopyrimidine (5) mimicking the pyrimidine core found in the anti-cancer drug imatinib (Gleevec ® ), and an aminopyrazine (6) (Figure 2). In addition, we report the characterization of new 2,6,9-trisubstituted purines belonging to previously reported series of HsCDKs inhibitors. All these molecules were evaluated against a panel of human kinases including CDK1, CDK2, CDK5, glycogen synthase kinase-3 (GSK-3), CK1 and DYRK1A. Finally, these newly-synthesized molecules and several other compounds belonging to our library of purines were screened as potential growth inhibitors of P. falciparum.

Scheme 1. Synthetic route to compounds 4a-d.
We report here the optimization of this catalytic system in order to obtain exclusively this di-amination product in good yield. As shown in Scheme 2, we used aminopyrazine as a nucleophile for this optimization, and our study revealed that the use of 8% of Pd(OAc)2 pre-catalyst and 16% of racemic BINAP ligand gave compound 6 with the highest conversion rate. In contrast to our previous study on amination with aminopyridines, we do not observe here the formation of a triarylamine product based upon the aminopyridazine scaffold [47]. Surprisingly, the replacement of the Pd(II) pre-catalyst by a Pd(0) pre-catalyst (Pd2dba3) leads only to decomposition products. Finally the use of the Pd(OAc)2 catalytic system allowed the formation of compound 5 in 70% yield (Scheme 2).

Molecular Assays on Mammalian Kinases
All newly synthesized compounds were tested as potential inhibitors on a panel of several mammalian kinases. The effects on CDK1/cyclin B, CDK2/cyclin A, CDK5/p25, GSK-3αβ, CK1δ/ε and DYRK1A kinase activity are reported as the IC50 values calculated from dose-response curves. They are summarized in Table 1. The inhibition was determined covering a range of concentrations of the compounds by means of a [γ-33 P] assay as described in the experimental section.
Globally, compounds 4a-d, with a 2-aminopyrimidine in the C 2 position, and compound 6 with a 2-aminopyrazine in the C 2 position, are poor kinase inhibitors when compared to (R)-roscovitine (1) or DRF053 (3). Moreover, we recently reported that 4a-d analogs bearing an aminoalcohol in the purine C 2 position inhibits CDKs with IC50 values ranging from 0.01 µM to 1.00 µM. Altogether, these results underline the relevance of the C 2 aminoalcohol to improve the CDKs ATP-pocket binding, since the hydroxyl group of the amino-alcohol is engaged in a H-bonding network involving water molecules. Therefore this hydroxyl group plays a critical role for improved inhibitor binding [49,50]. Table 1. Newly-synthesized trisubstituted purines effects on several mammalian protein kinases activity. Purine derivatives were tested at various concentrations as described in the experimental section; IC50 values are provided in µM were estimated, dose-response curves and are expressed in µM. All assays were run in triplicates and data points were within less than a 10% range. 50 value reported as >10 indicates that the compound display no activity at the highest concentration tested (10 µM); -: not tested.
Compound 11 is a potent inhibitor of CK1, with an IC50 value of 0.04 µM. This kinase is involved in multiple physiological events, such as circadian rhythm regulation, and its implication in Alzheimer's disease was strongly suggested [43,57]. With the significant exception of our previously reported series or trisubstituted purines including DRF053 [34], known CK1 inhibitors encompass generally polycyclic complex structures, including natural products such as hymenialdisine (IC50 = 0.03 µM) [58], a pyrrole-imidazole alkaloid extracted from marine sponges, and original synthetic molecules [59].

Evaluation on P. falciparum
A selection of 15 molecules 4b-c, 10-13 representing the different 2,6,9-trisubstituted purine series, including newly synthesized products was screened as anti-malarial agents on two different P. falciparum strains, Pf3D7, a standard drug-sensitive African stain, and Pf7G8, a brazilian chloroquino-resistant strain (Figure 3). The compound activity was evaluated during the parasitic erythocytic cycle, by the means of a [ 3 H] hypoxanthine incorporation assay [69]. It is directly correlated to parasite viability, as described in the experimental section. IC50 values on mammalian kinases for the P. falciparum screened compounds are summarized in supplementary material, Table A1, appendix section.
At first, the activity of the fifteen compounds was screened at three different concentrations (10, 50 and 100 µM) on Pf3D7. Chloroquine which is not a kinase inhibitor, but which is the antimalarial reference drug was used as the positive control. This allowed the selection of the six most active compounds, which totally inhibit the parasite proliferation during its erythrocytic cycle at a concentration of 100 µM. Then, in a second assay, these six molecules plus (R)-roscovitine (1) and purvalanol A (2a) were evaluated in a dose-dependent assay, in order to determine their IC50 values on the two P. falciparum strains. Chloroquine was again used as a positive control.
As expected, each P. falciparum strain growth is partly inhibited by (R)-roscovitine and purvalanol A, and the latter exhibited the strongest activities. Nevertheless, previous studies reported that IC50 values for purvalanol A on recombinant PfPK5 and Pfmrk are respectively of 8 µM and 26 µM, suggesting the likely existence of alternative in vivo molecular targets [32,70].
With the marked exception of compound 20, the screened molecules in the second round of assays exhibited significant IC50 values ranging from 0.7 µM to 7 µM on both strains ( Table 2) and therefore appeared to be more potent than (R)-roscovitine and purvalanol A [71], and also more potent than other adenine and adenosine derivatives with a purine ring [72]. Growth inhibition potency a Inactive compounds are drawn on a white background; these molecules were not selected for further IC 50 measurement. Compounds with IC 50 > 6 µM on Pf3D7 are drawn with light grey background. The most potent P. falciparum growth inhibitors are drawn with dark grey background. Table 2. IC50 values for Pf3D7 and Pf7G8 for the six mammalian kinase inhibitors and the three control molecules. These values were estimated by non-linear regression analysis using an inhibitory sigmoid Emax model. All assays were run in triplicates.

Compounds
Growth IC 50  Interestingly, these six inhibitors appeared to be 3-to 5-fold more potent against Pf7G8 than against Pf3D7 (e.g., compound 21: Pf3D7 IC50 = 3.05 µM vs. Pf7G8 IC50 = 0.69 µM). Such marked sensitivity of the chloroquino-resistant strain has been frequently observed, and is generally due to an acute internalization of the drug into the protozoan. Indeed, protozoans such as P. falciparum are unable to synthesize purines de novo, and all their purines are provided by the host. To this end, these organisms have developed very effective and specific systems to take up and internalize purines. Thus, these uptake pathways might differ from one strain to another, and therefore might be the reason of these slight differences observed for P. falciparum growth inhibition. However, these differences observed between P. falciparum growth inhibition according to the chloroquine sensitivity could also depend on the specific PfCDK inhibited by the tested compounds which remain to be studied.
Compounds 11, 12, 21 and 22 which exhibited sub-micromolar IC50 for the mammalian kinases appeared to be promising inhibitors of P. falciparum growth with IC50 values around 1-5 µM for both strains. These results are probably correlated with the similar sequence and structure between mammalian and protozoan kinases. Nevertheless, further "hit to lead" improvement will be necessary to improve selectivity towards protozoan growth inhibition. To address this optimization, it might be interesting to consider molecules 18 and 22 IC50 values. Indeed, for these compounds, the protozoan growth inhibition potency is directly linked to the purine C 2 pyridine core nitrogen position, suggesting a potential H-bond involving this nitrogen atom and the protozoan cellular target. This is a major difference with CDK inhibition, as we previously reported very close IC50 values for both of them on CDK1, CDK2 and CK1 (compound 18: CDK1 : IC50 = 0.41 µM; CDK5: IC50 = 0.73 µM; CK1: IC50 = 0.10 µM; compound 22 : CDK1: IC50 = 0.59 µM; CDK5: IC50 = 0.60 µM; CK1: IC50 = 0.08 µM [44]). Starting from this, a specific anti-malaria pharmacophore introduction in another position of the purine scaffold might pave the way to the design of a more specific and potent "hit".

General Procedures
Chemical reagents and solvents were purchased from Sigma-Aldrich (Lyon, France), Fluka (Lyon, France) and Carlo Erba (Val de Reuil, France). Reactions were monitored by TLC using Merck (Fontenay sous Bois, France) silica gel 60F-254 thin layer plates. Column chromatographies were performed on SDS Chromagel 60 ACC 40-63 µM. Melting points were determined on a Reichert Köfler hot-stage (Depew, NY, USA) and are uncorrected. NMR spectra were recorded on Bruker (Wissembourg, France) Avance 400 MHz (100 MHz for 13 C-NMR) at 300 K. Chemical shifts were reported as δ values (ppm) indirectly referenced to the solvent signal or to tetramethylsilane (TMS) as internal standards. Data are reported in the conventional form. Mass spectra were recorded on a ZQ 2000 Waters using a Z-spray (ESI-MS).
General Procedure for Buchwald-Hartwig Amination. A solution of Pd(OAc)2 and BINAP in dry toluene was warmed at 45 °C for 5 min. The leaving group-containing purine was then added under N2 bubbling; the mixture was kept at 45 °C for 10 min. and KOtBu was added. After 10 min, the appropriate nucleophile was added. The mixture was heated at 100 °C under N2 until reaction completion (3 h to 2 days depending upon the nucleophile used). After cooling to room temperature, the mixture was filtered through Celite, and concentrated. The residue was dissolved in CH2Cl2 (75 mL) and washed (1 × 10 mL) with water and brine (2 × 10 mL). The organic layer was dried and concentrated under vacuum. The residue was purified by chromatography on silica gel using various amounts of EtOAc/cyclohexane/ethanol as eluants.
We already reported chemical characterization for compounds 5, 6 and 6'; 1 H-NMR, 13 C-NMR and mass spectra are in accordance with previously reported data [44]. (8). 1    Chemical Characterizations of Compounds 10-13. These compounds have been obtained following a previously reported synthesis strategy. See references [43] and [46] for the details of corresponding experimental procedures and characterizations of intermediates.

Biology-In Vitro Drug Susceptibility Assays
Compounds were tested against synchronous ring-stage parasites of Pf3D7 and Pf7G8 strains. Drugs testing were carried out in 96-well microtiter plates. The chloroquine (CQ) diphosphate was purchased from Sigma (Lyon, France). The CQ was dissolved and diluted in water to obtain final concentrations ranging from 12.5 to 3,200 nM. The compounds were dissolved in DMSO and diluted in RPMI 1640 to obtain final concentrations ranging 0.1 µM to 100 µM. For each assay, each drug dilution was analyzed in triplicate.
For the in vitro isotopic microtest, 200 µL of the suspension of synchronous parasitized red blood cells with >90% of the parasites at the ring stage (final parasitaemia, 0.3%; final haematocrit, 4%) per well were plated in 96-well plates that contained serial CQ or compounds concentrations. Parasite growth was measured by the incorporation of radiolabeled [ 3 H]hypoxanthine with a specific activity of 14.1 Ci/mmol (Perkin-Elmer, Courtaboeuf, France) (1 µCi per well) to each well at time zero. The plates were incubated at 37 °C, in an atmosphere of 5% O2, 5% CO2, and 90% N2 for 48 h. Immediately after incubation, plates were frozen and then thawed to lyse the erythrocytes. Cultures were harvested onto glass fiber filters and washed using a cell harvester (FilterMAT; Skatron Instruments, Dalsletta, Norway). The dried fiber filter papers were mixed with 2 mL of scintillation fluid (OptiScint; Perkin-Elmer, Waltham, UK). The radioactivity was counted using a liquid scintillation counter (Wallac 1410; Perkin-Elmer, Waltham, UK). The results were recorded as counts per minute (cpm) per well at each drug concentration. The drug concentration that could inhibit 50% of the parasite growth (IC50) were estimated by non-linear regression analysis using an inhibitory sigmoid Emax model (percent) = 100 − [(100 × C γ )/(C γ + IC50 γ )], where C corresponds to the drug concentration and γ is the sigmoidicity factor. The initial value and the asymptotic result for high concentrations were fixed to 0% and 100%, respectively [75], available on the web [76].

Conclusions
Using a convergent synthesis route which includes as a key step a Buchwald-Hartwig amination, we report the synthesis of new series of 2,6,9-trisubstituted purines structurally-related to (R)-roscovitine. The evaluation of these molecules as inhibitors on a mammalian kinases panel revealed that several compounds (5, 6, 10 and 11) exhibit marked activities against CDK5, CK1 and DYRK1A, a set of kinases involved in several neuronal pathologies such as Down Syndrome and Alzheimer's disease. These molecules might be considered as starting "hits" for further structural optimization.
These newly-synthesized purines were also evaluated as potential inhibitors of P. falciparum growth in vitro. By means of a two-step screening assay, we identified six molecules which are more potent against the two P. falciparum clones (Pf3D7 and Pf7G8) than (R)-roscovitine and purvalanol A. By comparison with chloroquine, these growth inhibitions are still modest, but further structural optimizations of the purine scaffold are possible. Thus, further studies should allow us to determine whether potent anti-malarial motives might be introduced into the purine scaffold. Another key point to address will be the improvement of the compound selectivity, as our strongest P. falciparum growth inhibitors also appeared among the best mammalian CDK inhibitors we designed.
Altogether, these results open the way to further synthesis of new drugs targeting one of the most deadly disease in the world, malaria.

Conflicts of Interest
All the authors have read the manuscript, concur with its content, and state that its content has not been submitted elsewhere. The authors declare no competing financial interests and no conflict of interest.