Search

Malaria is a leading cause of morbidity and mortality, causing more than 400,000 deaths per year. Malaria is caused by parasites of the Plasmodium genus with most deaths due to P. falciparum infection. The control of malaria is complicated by the lack of a widely effective vaccine, the spread of mosquito resistance to insecticides and Plasmodium parasite resistance to available drugs, including the gold standard artemisinin-combination therapies. Thus, there is an urgent requirement for the development of new antimalarials, in particular those with different modes of action to existing drugs to limit potential problems of cross-resistance. Plasmodium species have a complex lifecycle that includes transmission from the female Anopheles mosquito vector to a human host requiring significant morphological changes. These morphological changes are associated with stage-specific changes in transcription regulated by epigenetic mechanisms. The proteins involved in these processes are potential new therapeutic targets for malaria. This includes histone deacetylases (HDACs), which together with histone acetyltransferases (HATs), are involved in reversible posttranslational acetylation of histone and non-histone proteins, regulating transcription and other cellular processes. To date, over 650 HDAC inhibitors have been investigated for in vitro activity against malaria parasites. Some inhibitors, particularly those with a hydroxamic acid zinc-binding group that targets inhibitors to the HDAC active site, have demonstrated low nM in vitro potency against P. falciparum and selectivity for the parasite over human cells. However, antiplasmodial HDAC inhibitor drug development has been hindered by factors including the lack of recombinant P. falciparum HDACs (only one available and purity is low), the lack of HDAC crystal structures (none available) and low throughput activity assays that are largely indirect measures of HDAC inhibition. Without these tools, mode of action studies, the rational design of new and improved inhibitors and the prioritisation of compounds for preclinical testing remains difficult. To address some of these challenges and further progress the development of antimalarial HDAC inhibitors, the current study employed a multi-pronged approach, including: (i) investigating the in vitro and in vivo activity of new HDAC inhibitors; (ii) establishing a higher throughput ELISA method to analyse P. falciparum lysine acetylation alterations and; (iii) developing a quantitative structure-activity relationship (QSAR) model based on classification algorithms. HDAC inhibitors typically have a pharmacophore comprising a zinc-binding group that interacts with the zinc ion in the active site of the enzyme, a linker unit and a cap group promoting hydrophobic interaction with amino acid residues at the entry of the active site. Here, a set of 26 new HDAC inhibitors with a peptoid-based scaffold was tested in vitro against drug sensitive asexual intraerythrocytic-stage P. falciparum 3D7 parasites. The set are analogues of compounds that have previously shown in vitro dual-stage antiplasmodial activity against asexual intraerythrocytic and exoerythrocytic stages and includes 16 compounds with a hydroxamic acid zinc-binding group and 10 prodrugs of this compound class. The unprotected hydroxamate-based inhibitors demonstrated growth inhibition of P. falciparum 3D7 asexual intraerythrocytic-stage parasites in the nanomolar to micromolar range (50% growth inhibition values (PfIC50) 0.008-1.04 μM) and up to 1,250-fold selectivity (selectivity indices (SI; PfIC50/human cell IC50): 10-1,250) for the parasite compared to human cells. Structure-activity relationship (SAR) analysis of cap region residues (carbonyl region, carboxylic region and isocyanide region) indicated that benzyl groups in the isocyanide region and alkyl groups in the para position of the carboxylic region are associated with increased antiplasmodial activity. In addition, methyl groups in the carbonyl region of the cap group demonstrated reduced cytotoxicity against neonatal foreskin fibroblasts (NFF), however, also somewhat reduced activity against asexual blood-stage parasites. Work by collaborators demonstrated micromolar in vitro activity of several compounds of this set against exoerythrocytic P. berghei parasite forms indicating dual-stage activity. The compound with the greatest dual-stage activity displayed an IC50 of 8 nM against asexual blood-stage P. falciparum and an IC50 of 60 nM against exoerythrocytic P. berghei in vitro. Compounds with PfIC50 of 100 nM or lower were tested against the multi-drug resistant P. falciparum Dd2 line (resistant to chloroquine, pyrimethamine, mefloquine, and other antimalarial drugs), and demonstrated a resistance index (RI) 100 and may therefore be of interest in further studies. Based on the in vitro antiplasmodial activity, selectivity and chemical diversity in the cap region, five peptoid-based compounds (3a, 3c, 3f, 3m, 3n, Pf3D7 IC50 0.008-0.034 μM, SI 97-625) were further investigated for in vivo efficacy against Plasmodium parasites. In addition, four analogues of the tethered phenylbutyrate-based HDAC inhibitor AR42 (Pf3D7 IC50 0.02 μM, SI 39) were also investigated in vivo (JT21b, JT83, JT92a, JT94; Pf3D7 IC50 0.005-0.21 μM, SI 55-118, (data generated by Dr MJ Chua, personal communication)). AR42 is currently in phase 1 clinical trials against various types of cancer and demonstrates an improved pharmacokinetic profile compared to a number of clinically approved HDAC inhibitors (e.g. AR42 Cmax 14.7 μM compared to vorinostat Cmax 1.9 μM, AR42 t1/2 11.1 h compared to vorinostat t1/2 0.75 h; tested in mice). AR42 analogues were of interest as AR42 has previously been shown to cure Plasmodium infections in mice (Dr MJ Chua, Griffith Institute for Drug Discovery; unpublished). While the two analogue sets differ significantly in linker and cap group, both bear a hydroxamic acid zinc-binding group. Compounds were tested in groups of two female BALB/c mice infected with P. berghei ANKA infected erythrocytes. Dosing was via oral gavage at 25 mg/kg twice daily with four hours between dosing (beginning 2 h post infection) for four consecutive days. Peripheral blood parasitemia was monitored by microscopic evaluation of stained thin blood films from day four post infection. None of the peptoid-based HDAC inhibitors attenuated P. berghei growth in BALB/c mice by more than 33% (3f (31%) and 3n (33%) on day 6 post infection). Data from collaborators demonstrated 3n to have the best metabolic stability (t1/2 271 min, Clint 6 μL/min/mg in mice; Prof Finn Hansen, University of Bonn, Germany) which may have contributed to this compound’s improved activity compared to some other analogues. In comparison, AR42 and two if its analogues cured mice of infection (AR42, 1 of 2 mice; JT21b, 2 of 2 mice; JT83 2 of 2 mice), up until day 24 post infection, at which point the mice were euthanised. AR42 and analogues are the first demonstration of oral cures in mice with a HDAC inhibitor (manuscript in preparation) and these data will be pursued in future work to further develop this HDAC inhibitor chemotype for malaria. One of the current limitations in the field is the lack of recombinant P. falciparum HDACs and the need to rely on low throughput assays to demonstrate HDAC inhibitor action via reduced total deacetylase activity or in situ lysine acetylation alterations. While deacetylase assays do not allow the differentiation of compound effects, Western blot using different acetyl-lysine antibodies can reveal compound specific acetylation profiles. Here, two higher throughput methods, dot blot and ELISA, were investigated to assess the effects of HDAC inhibitors on lysine acetylation. Using the control hydroxamate HDAC inhibitor vorinostat (first HDAC inhibitor clinically approved for cancer), the ELISA method was demonstrated to be more reliable than dot blot in detecting acetylation changes in protein lysates from P. falciparum trophozoites exposed to compound for 3 h. ELISA was therefore used to investigate histone H3 and H4 lysine acetylation alterations following exposure of P. falciparum to six commercially available anti-cancer HDAC inhibitors (vorinostat, panobinostat, trichostatin A, romidepsin, entinostat and tubastatin A). All compounds have in vitro activity against asexual intraerythrocytic P. falciparum parasites (Pf3D7), with tubastatin A activity reported for the first time here (PfIC50 0.15 ± 0.03 μM). All compounds were also shown to inhibit >84% deacetylase activity using P. falciparum protein lysates in an in vitro assay at 1 μM, with the exception of entinostat (~50% inhibition at 1 μM); this compound was also the least active against the parasite (PfIC50 11.5 μM). Using ELISA, vorinostat, panobinostat, trichostatin A, romidepsin and entinostat were all found to cause a ~3-fold increase in the signal detected using an anti-tetra-acetyl-lysine antibody. In comparison, the only human HDAC6-specific inhibitor tested, tubastatin A, caused 1.8-fold histone H4 hyperacetylation compared to the control. Further investigations of the individual N-terminal H4 lysine residues using antibodies specific to acetylated lysine 5, 8, 12 or 16 revealed that all compounds, except tubastatin A, caused hyperacetylation using each antibody. No differential effect was observed for histone H3 acetylation, with all compounds causing an ~1.8-fold increased signal using an acetyl-H3 antibody. The new ELISA method developed here provides a higher throughput way to assess differential compound induced lysine acetylation alterations in P. falciparum and therefore represents a valuable new tool to aid the investigation of HDAC inhibitors for malaria. As discussed above, the lack of tools, such as recombinant P. falciparum HDAC proteins, crystal structures and homology models, has meant that the identification of antiplasmodial HDAC inhibitors has been limited to whole-cell screening approaches which can be time-consuming and costly. To begin to address this problem, quantitative structure-activity relationship (QSAR) models were developed based on logistic algorithms with the aim of providing a new tool to triage compounds for in vitro testing. A database of 457 antiplasmodial HDAC inhibitors was assembled with published data on PfIC50 and, for 292 of those compounds with data on plasmodial selectivity. Two independent prediction algorithms based on logistic regression were developed to classify (1) antiplasmodial activity or (2) selectivity of hydroxamate-based HDAC inhibitors. Seven different activity and five different selectivity models were built, each with individual decision cut-offs defining active/selective and non-active/unselective compounds (e.g. PfIC50: active compound non-active compound; SI: selective compound >100< unselective compound). Activity model A7 revealed the highest prediction performance by predicting 93% of the training compound set and 87% of the external test compound set correctly. Cross validation revealed a prediction accuracy of 91%. The most accurate selectivity model S4 demonstrated a slightly poorer prediction performance due to a much smaller initial data set as not all the HDAC inhibitors had reported selectivity information (64%). Despite this, the selectivity model demonstrated an internal prediction accuracy of 91%, a cross-validated (internal) prediction accuracy of 82% and an external prediction accuracy of moderate 72%. To validate the prediction performance of the activity model further, they were applied to a set of 22 experimentally untested compounds (validation set) and the prediction performance compared to their experimental antiplasmodial activity. Applying prediction model A7 to this compound set predicted three hit compounds (two of which were confirmed by experimental assay data) and 12 non-actives (confirmed for 11 based on experimental assay data). The experimental PfIC50 assessment revealed asexual blood-stage PfIC50s for the whole set in the nanomolar to micromolar range (PfIC50 0.006-8.45 μM; data from Dr MJ Chua), with the correctly predicted hits (S2_E10 and LD016) having PfIC50