1,3-Diphenylureido hydroxamate as a promising scaffold for generation of potent antimalarial histone deacetylase inhibitors

We report a series of 1,3-diphenylureido hydroxamate HDAC inhibitors evaluated against sensitive and drug-resistant P. falciparum strains. Compounds 8a–d show potent antiplasmodial activity, indicating that a phenyl spacer allows improved potency relative to cinnamyl and di-hydrocinnamyl linkers. In vitro, mechanistic studies demonstrated target activity for PfHDAC1 on a recombinant level, which agreed with cell quantification of the acetylated histone levels. Compounds 6c, 7c, and 8c, identified as the most active in phenotypic assays and PfHDAC1 enzymatic inhibition. Compound 8c stands out as a remarkable inhibitor, displaying an impressive 85% inhibition of PfHDAC1, with an IC50 value of 0.74 µM in the phenotypic screening on Pf3D7 and 0.8 µM against multidrug-resistant PfDd2 parasites. Despite its potent inhibition of PfHDAC1, 8c remains the least active on human HDAC1, displaying remarkable selectivity. In silico studies suggest that the phenyl linker has an ideal length in the series for permitting effective interactions of the hydroxamate with PfHDAC1 and that this compound series could bind as well as in HsHDAC1. Taken together, these results highlight the potential of diphenylurea hydroxamates as a privileged scaffold for the generation of potent antimalarial HDAC inhibitors with improved selectivity over human HDACs.

Infectious diseases are among the leading causes of death globally, collectively second only to cardiovascular diseases.Malaria, a parasitic infection caused by Plasmodium spp., has been among mankind's deadliest diseases, affecting over 247 million people in 2021 in endemic countries and causing 619,000 deaths worldwide 1 .Despite large investments to develop alternative pharmacological interventions such as malaria vaccines, effective malaria control likely relies upon the continued development of new small-molecule antimalarial drugs 2 .To date, the antimalarial agents can be grouped into seven main classes: arylaminoalcohols (quinine derivatives), 4-and 8-aminoquinolines (chloroquine and primaquine derivatives, respectively), analogues of artemisinin, antifolates (e.g.pyrimethamine), antibiotics (e.g.tetracycline), and other agents (e.g.atovaquone) 3 .A major limitation of current antimalarial chemotherapy is the rapid spread of drug-resistant Plasmodium falciparum parasites following first-line treatments (artemisinin-based combination therapies) or following alternative drug combination protocols [4][5][6][7] .A strategy to combat malaria drug resistance is to identify and develop new antimalarial drugs acting on novel parasite targets, likely to be used to complement the existing treatment options.
The transcriptional control in malaria parasites is complex and has been the subject of extensive investigation.Moreover, there is increasing evidence that targeting transcriptional regulation represents a potential new therapeutic approach for malaria 8 .In this sense, histone deacetylase enzymes (HDACs) are well-known key regulators of transcription and human HDACs are validated targets for some types of hematological cancers 9 .In eukaryotes, a homeostatic balance of the acetylation state of histones is modulated by the coordinated activity of histone acetyltransferases (HATs) and HDACs 10 .HATs increase the acetylation of certain amino acid residues on histone proteins, resulting in a less condensed segment of DNA that can be transcribed.On the other hand, HDACs regulate target genes through the deacetylation of key lysine residues in histones and non-histone substrates, thus promoting DNA condensation and suppression of gene expression 11 .This epigenetic regulation is observed in all stages of the Plasmodium life cycle, is critical to parasite stress response, and is thought to contribute to the transcriptional regulation of drug resistance 12,13 .Human HDACs can be classified into three classes of zinc-dependent enzymes and one NAD + -dependent class also called sirtuins 14 .To date, three classes of HDACs have been identified in P. falciparum: (1) PfHDAC1 (PFI1260c) is a predominant nuclear class I HDAC enzyme; (2) PfHDAC2 (PF14_0690) and PfHDAC3 (PF10_0078) are assigned to class II HDACs; (3) and two sirtuins PfSir2A (PF13_0152) and PfSir2B (PF14_0489) 15 .Noteworthy, PfSir2A and 2B are recognized as nonessential enzymes and are primarily involved in the regulation of var gene expression, involved with antigenic variation making them less appealing as drug targets 16 .On the other hand, PfHDAC1 proved to be the highly conserved isozyme among all species and is involved in gametocytogenesis, schizogony, and hepatocyte invasion besides exhibiting moderate similarity (~ 61%) to human HDACs (hHDACs) [17][18][19] .

Design and synthesis of the urea-derived HDAC inhibitors
The original concept that supported the synthesis of the series considered the cinnamyl hydroxamate of 2 as the primary linker-ZBG moiety (Fig. 1B) which was attached to four different para-phenylureas (compounds 6a-d, Fig. 2).Subsequently, the cinnamyl double bond was hydrogenated (compounds 7a-c, Fig. 2) to evaluate the influence of a saturated linker over the biological activity (Fig. 1B).Moreover, the double bond was removed from the original scaffold (Fig. 1B), generating the www.nature.com/scientificreports/preliminary phenyl-hydroxamates 8a-d (Fig. 2).After our first round of antiplasmodial screening, we synthesized additional compounds 8e-k and 15 to generate SAR insights with respect to the optimization of the capping moiety and the identification of ways to increase productive interactions in the target binding pocket (Fig. 2).The synthesis started with the addition of ethyl (E)-3-(4-aminophenyl)acrylate ( 9) to different phenyl isocyanates in dichloromethane (DCM), generating intermediates 10a-d.Intermediates 11a-c have been prepared through catalytic hydrogenation of 10a-c with palladium on activated charcoal in ethanol as solvent.The final products 6a-d and 7a-c were obtained by reaction of intermediates 10a-d and 11a-c with aqueous hydroxylamine under basic conditions.The synthesis of compounds 8a-i started with the addition of methyl 4-aminobenzoate (12) to appropriate phenyl isocyanates generating intermediates 13a-i.The final compounds 8a-i were prepared under same conditions as described above for 6a-d and 7a-c.The amino compound 8j was obtained through standard catalytic hydrogenation of 8d with palladium.Compound 8k has been prepared by coupling 4-(boc-amino)benzoic acid (14) and O-(tert-butyldimethylsilyl)hydroxylamine under HATU conditions.Deprotection with trifluoroacetic acid (TFA) provided 15, which was converted into 8 k after addition to 4-cyanophenyl isocyanate in DCM 23 .
Antiplasmodial activity and effects on cell viability of compounds 6a-d, 7a-c, 8a-k A preliminary screen for compounds 6a-d, 7a-c, and 8a-d against P. falciparum 3D7 in three different concentrations (200, 20, and 2 μM) was used to investigate their antiplasmodial potential (Fig. S1).Subsequently, compounds have been submitted to dose-response assays to determine their IC 50 values (Table 1).To evaluate the therapeutic window as well as the selectivity index (SI) of the compounds, the cytotoxic effect against human hepatocarcinoma cells (HepG2) has been also determined (Supporting information, Figs.S1, S2).The main finding of this series is that the use of a diphenylurea cap-linker moiety correlates with higher antiplasmodial potency relative to both cinnamyl and di-hydrocinnamyl cap linkers.Avoiding the use of a cinnamyl hydroxamates gave a benefit with respect to target selectivity, as reflected by mild (IC 50 > 50 µM) or no measurable cytotoxicity on HepG2 cells (IC 50 > 200 µM), which differed from the more toxic cinnamyl hydroxamates, corroborating previous findings of the cinnamyl series on HS-5 human hematological cells 26 .
Because the emergence of drug resistance is an important consideration, we assessed the potency of our best compounds against the Plasmodium falciparum Dd2 strain (herein named Dd2).While PfDd2 was ~ 28 × less susceptible to chloroquine (CQ, Fig. 3A, B), SAHA (1), and nexturastat A (NextA, 16, a selective human HDAC6 inhibitor), we observed mostly retained efficacy against this strain with our best compounds (Fig. 3A-D).In particular, the HDACi 8c, our best non-toxic compound in terms of antiparasitic effects (Fig. 3E, F), and a close analogue (8d) show minimal loss in effectiveness against PfDd2 (~ 1.6 × drop in potency).Interestingly, both NextA (16) and SAHA (1) have some degree of toxicity vs. HepG2 cells (IC 50 = 21.5 and 96.8 µM, respectively, see Supporting Information, Fig. S4), which is not observed for 8c (IC 50 > 200 µM), suggesting that 8c has the potential to be well-tolerated at therapeutic doses.

Molecular modelling
The potential binding mode for representative compounds of each synthesized scaffold (6c, 7c, and 8c) to PfH-DAC1, HsHDAC1 and HsHDAC6 was studied using molecular modelling.Briefly, we generated a homology model of PfHDAC1 and HsHDAC6 (based on the PDB ID: 6DVO), while the HsHDAC1 structure was retrieved (PDB 5ICN), which was used for the docking of compounds.Additionally, for PfHDAC1 and HDAC6, we generated mono and bidentate models for the methoxy-substituted compounds, in order to compare their influence on the binding mode, while HDAC1 was exclusively simulated as monodentate.Those restrictions were imposed in order to isolate the effects rising from the different cap changes.Those binding mode models underwent classical molecular dynamics simulation.The predicted binding energy and frequency of interactions along the trajectory were used as parameters for the binding discussion.The model suggests that the overall binding mode of our compounds within PfHDAC1 is similar to the previously suggested for HsHDAC1 and 6 (Supporting Information, Fig. S5 and Table S1), in several aspects but not the linker.This is supported by the high similarity between the amino acids composing each pocket (Supporting Information, Fig. S6).
From the ZBG perspective, the HsHDAC6 structures usually display monodentate Zn 2+ coordination mode for sterically bulky HDAC6-selective phenyl hydroxamate inhibitors, while flexible saturated acyl groups retain bidentate coordination 31 .This monodentate binding mode is energetically accessible (0.5 kcal/mol difference between mono and bi) 31 and common for bulky inhibitors that cannot bind deeper in the HDAC6 narrow pocket, a requirement for effective bidentate coordination.In our simulations, we analyzed both mono and bidenticity and opted to discuss the later as it generated stabler binding conformations (Table S2) and better agreed with the Plasmodium counterparts.
Further, the linker interactions with Ser90/Asp97 (HsHDAC6/PfHDAC1, Table 2 and Supporting information, Table S1), seem to be more frequent than previously observed in HsHDAC1 29 , are exemplified in the model.We suggest that this interaction, together with tighter hydrophobic contacts, contributes to the overall lower binding energy in PfHDAC1 when compared to the human counterpart (Supporting information, Table S2).A more comprehensive study of the interactions performed by the cap fragments suggests that stability relies on hydrophobic and pi-mediated contacts.Among the hydrophobic contacts contributed to stabilizing the interaction of our compounds with either HsHDAC1 or PfHDAC1 (such as Phe202/203, Leu269/271 and Tyr301/303, present in all studied isoforms) and towards selectivity against HDAC6.Specifically, interactions between the cap group and the amino acids Pro23 (numbering follows HsHDAC6), Phe142 and Tyr304 seem more frequent than the HsHDAC1 and PfHDAC1 counterparts.HsHDAC6 also engages more frequently in pi-mediated interactions with Phe142 and His173, while Plasmodium's HDAC only uses these residues as polar contacts to the linker.

Table 2. Summary of Protein-ligand interaction frequency during the analyzed trajectory for each compound.
Frequency is displayed as (%) of the hydrogen bond, water-mediated interactions or pi-pi interactions, and separated according to the compound moiety performing it.The full description can be found in Table Table S1 31 .However, HDAC6 specific inhibitor's cap displays a preference for the shallow portion of the pocket, meaning the two covering loops (L1 and L2).Nonbranched cap inhibitors point toward loop L1, in crystal structures, whereas the branched ligands would interact with both loops L1 and L2, accordingly, our best inhibitors engage more frequently with the L1.

Target engagement validation via in vitro enzymatic assay and Western Blot analysis
Our compounds were biochemically evaluated for human HDAC1/8 (Class I), and HDAC6 (Class IIb) inhibition.It is evident that compounds from the cinnamyl series 6a-d (IC 50  = 9.0-14.3nM, Table 3) and phenyl series 8a-k (IC 50  = 8.5-21.3nM) demonstrated significant activity against the human HDAC6, markedly outperforming their activity against other isoforms.In contrast, the dihydrocinnamyl series 7a-c (IC 50 = 151.4-184.9nM) proved to be the least active/selective against HDAC6.It is noteworthy that all three series exhibited a degree of selectivity for this isoform compared to HDAC1/8.Interestingly, all series were less active against HDAC1 (6a-6d, IC 50 = 1.1-2.7 μM; 7a-7c, IC 50 = 3.5-4.8μM; 8a-k, IC 50 = 3.7-8.2μM), indicating a high degree of selectivity, Table 3. Potential PfHDAC inhibitors tested by means of previously reported biochemical in vitro deacetylation assays in human isoforms 32,33  especially against HDAC1, as indicated in Table 3.However, it is relevant to mention that all compounds showed moderate activity against class I HDAC8, with 7a-c being four times less active in this isoform compared to other compounds of the series (Table 3 and respective binding mode are depicted in the Fig. 4).
Given the high potency of 6c, 7c, and 8c in the HsHDAC panel and the malaria cell-based data, we decided to further characterize their mechanism of action.In this sense, the mode of action of our compounds (6c, 7c, and 8c) was addressed by an in vitro PfHDAC1 biochemical assay, monitoring their ability to inhibit the lysine acetylation of a model peptide (Fig. 5A).All tested compounds significantly inhibited PfHDAC1 with SAHA 1 being the most potent displaying IC 50 values between 0.1 and 1 µM.Compounds 7c and 8c displayed similar potency to NextA (16), inhibiting roughly 50% of the enzyme activity at 1 µM, while 6c was the most potent with residual activity of 26% on 1 µM.Further, immunoblotting analysis for acetylated histones (H3 and H4, Fig. 5B and Supporting Information, Fig. S7, Fig. S8) in infected erythrocytes suggest a non-statistically significant increase in acetylated H3 and H4 for SAHA (1) and NextA (16) treated cells.
Increased H3 acylation did reach statistical significance in the case of compounds 8c and 7c (two-fold increase in acetylated-H3), while compounds 8c and 6c demonstrated a significant elevation in acetylated-H4 (Fig. 5C, D).
This difference in potency might be related to the better interaction among the diphenylurea compounds and the PfHDAC binding cavity.Noteworthy, previous findings for this set of compounds made over a panel of solid and hematological cancer cells have indicated that the cinnamyl linker was the best one to explore human HDACs, thus presenting the most potent activities 26 .Taken together, these findings might indicate that 1,3-diphenylureido hydroxamate is a relevant scaffold to design of potent yet more selective antimalarial HDAC inhibitors, a combination that has been elusive, especially with a hydroxamate ZBG.Curiously, both electron-donating (EDG) and electron-withdrawing (EWG) groups at the para position of the capping ring generated potent compounds as 8c and 8d, although 8d was the most potent antiplasmodial compound of the series and the methoxy derivative 8c presented the highest selectivity index (SI) over human HepG2 cells (SI > 271, Table 1).Noteworthy, the second round of optimized inhibitors gave a significant improvement in selectivity (SI up to > 769, Table 1) without sacrificing potency, as observed for 8e-f, 8h, and 8j-k.On the other hand, compounds 8g and 8i had impaired potency/selectivity that could be caused by the presence of the trifluoromethoxy and trifluoromethyl groups, respectively, at the para position (Table 1).The importance of the phenylurea cap to the antimalarial potential of the series can be observed by compound 15, which is the linker-ZBG portion of compound 8a-k.Even though 15 preserved some activity over 3D7 parasites, lacking the capping motif caused a 24-fold reduction in potency compared to 8k, thus indicating that the phenylurea moiety is indeed playing a significant role in the way that these compounds interact with the target.
On-target activity confirmation against recombinant PfHDAC1 shows that compounds from all tested scaffolds have potency on the same level as NextA (16) but are less potent than SAHA (1).While the increased ratio of acetylated-H4 upon treatment confirms 8c and 6c target engagement on a cellular level, further improvements in compound permeability are desirable.It is worth noting, however, that compound 8c exhibited a relevant inhibitory activity against PfHDAC1, while conversely displaying marginal inhibitory action against HsHDAC1.
Our modelling results support the idea that these compounds would bind on both PfHDAC1 and HsHDAC1, with higher predicted binding affinity to PfHDAC1 independently from the scaffold.In HsHDAC1, compounds with a methoxylated cap had weaker hydrophobic interactions than chlorinated ones, which is consistent with their decreased potency in this isoform.Compounds with the Cl-substituted cap perform the worst in terms of predicted energy for the HsHDAC1, in comparison to the Methoxy and unsubstituted counterparts, which agrees with the on-target IC 50 values (Supporting information, Table S2).HsHDAC6 and PfHDAC1 display higher potency, with no significant differences in the interactions of their cap moieties.The bidentate interaction pattern of the hydroxamate moiety appears to slightly favor hydrophobic interactions in PfHDAC1, which would need to be further supported by calculations that allow polarization and/or QM integration.Additionally, in terms of comparison between predicted binding energy and on-target IC 50 values, HDAC6 bidentate interaction's predicted binding energy reflects much better the determined IC 50 s than the monodentate.
The PfHDAC1 is closely related to HsHDAC1, whose main targets are the acetylated H3 and H4, while PfH-DAC2 and PfHDAC3 belong to HDAC class II, such as HsHDAC6 18,36 .Our docking models using a PfHDAC6like model supports that longer acyl linkers could function better by occupying its larger pocket, which is in line with the previous HsHDAC activity of these scaffolds.Interestingly, despite our best efforts, no acetylated-αtubulin, as evidence for HDAC6-like inhibition (i.e.HDAC class II inhibition), was detected.We hypothesized that this fact is due to the non-conservation of the acetylation site of the plasmodial homologue, which could result in the no recognition by the human-targeting acetyl-α-tubulin antibody employed.
The unavailability of a recombinant HDAC class II from Plasmodium falciparum, together with the discussion that catalytic activity without endogenous cofactors is controversial 37 , supports our concerns that PfHDAC2-3 would be experimentally challenging.Other groups have used human HDAC1 and 6 activity assays as surrogates www.nature.com/scientificreports/ to assess Plasmodium's HDAC activity 38 .The comparison between our PfHDAC1 inhibitory data (~ 50% inhibition at 1 µM) with this previously published HsHDAC1 dataset (IC 50 values ranging from 0.9 to 3.2 µM) 29 would support this correlation.Moreover, our initial compounds 29 are privileged against HsHDAC6 (with at least ~ tenfold selectivity against HDAC1), which would encourage further studies on Plasmodium's class II HDACs.Nardella and co-workers have disclosed novel HDAC-DNA methyltransferase (DNMT) inhibitors, designed by derivatizing the pan-histone deacetylase inhibitor SAHA (1) with procainamide 39 .These compounds have acyl linkers with different lengths, with hexyl linkers (n = 6) yielding the most potent derivatives in combination with a basic cap.Shortening of this linker, between the phenyl group and the hydroxamic acid, resulted in a complete loss of antimalarial activity, which disagree with our results, where cinnamyl hydroxamates were less potent antimalarials.It is also important to highlight that Nardella's design was restricted to saturated acyl linkers, which are flexible and less bulky than our best compounds.
Interestingly, our lead compounds generally retain high potency against the CQ-resistant laboratory Plasmodium falciparum strain Dd2.Previous studies, testing HDACi against field-isolated parasites 25,40 reported higher ex vivo IC 50 values, when compared to the laboratory 3D7 strain.Despite being tested in similar conditions, strains from Gaboa reported around a three-fold increase in IC 50 values compared to the laboratory strain 40 , www.nature.com/scientificreports/whereas Indonesian isolates were even more resistant with median IC 50 values ranging from 20-35 times the value found with the 3D7 strain 25 .
Regarding the DMPK assessment, all tested compounds (6c, 7c, and 8a-d, Table 3) started to be depleted in the absence of NADPH.It is known that NADPH is a required cofactor for cytochrome P450 (CYP450) and flavin monooxygenase enzymatic functions.Moreover, depletion in the absence of NADPH is a strong indication of hydrolysis mediated by other enzymes like proteases and esterases, which is a well-known metabolic target observed for hydroxamates 40 .Besides primary degradation by hydrolysis, the rate of disappearance of compounds 8a-c was significantly higher upon the addition of NADPH, suggesting the involvement of parallel clearance mechanisms for these compounds.Among all tested CYP450 isozymes, all compounds were mainly susceptible to CYP1A2 which agrees with previous findings for 1 41 .
Previous screening of a small library of HDAC inhibitors (180 chemotypes 42 ) identified the compound FNDR-20123, among others enriched phenylethyltriazoles, with IC 50 values against P. falciparum culture in the low nanomolar range.FNDR-20123's phenylethyltriazole cap-linker moiety is a bioisostere from our best-performing phenylurea scaffold, suggesting that a bulkier linker is tolerated, as long as its flexibility is considered.The inhibitor has a high half-life (t 1/2 2-9.21 h), higher than 8c's by an order of magnitude, which together with their high treatment dosage (50 mg/kg) would have contributed to the consistent reduced parasitaemia.This points out that despite the high antiplasmodial potency of our compounds and excellent target engagement profile, further optimization should rely on minimizing degradation by primary metabolism and favouring ADME properties, mainly solubility.Our best compounds, however, attend to the Malaria Venture lead-like criteria 43 for antimalarial potency (IC 50 < 0.1 mM) and parasite selectivity (SI > 100).

Conclusions
Herein, a series of 1,3-diphenylureido hydroxamates with known HDAC inhibitory activity have been synthesized and evaluated against sensitive and drug-resistant P. falciparum strains.Compounds 8a-d presented potent antiplasmodial activity indicating that the phenyl linker generated compounds with improved potency compared to cinnamyl and di-hydrocinnamyl linkers.Compound 8c presented the highest SI of the first round of screened compounds and a stable preliminary metabolic profile.In vitro mechanistic studies confirmed PfHDAC1 on-target activity at a recombinant level, consistent with cellular quantification of acetylated histone levels.Notably, compound 8c demonstrated potent inhibition of PfHDAC1, contrasting with marginal activity against HsHDAC1.In silico studies suggest that the phenyl linker would have the ideal length among the series for interaction with the PfHDAC1 catalytic cavity and that our compound series could bind as well as in HsHDAC1.Taken together, these results highlight the potential of diphenylurea hydroxamates as a privileged scaffold for the generation of potent antimalarial HDAC inhibitors with improved selectivity over human cells.
Synthesis of compound 10c.Ethyl (E)-3-(4-(3-(4-methoxyphenyl)ureido)phenyl)acrylate (10c).General Procedure A was followed using ethyl 4-aminocinnamate (9) and 4-methoxyphenyl isocyanate to afford the title compound as a white solid (1.62 g, 95%). 1 H NMR (300 MHz, DMSO-d 6 ) δ 8.85 (s, 1H), 8.55 (s, 1H), 7.71-7.46(m, 5H), 7.38 (d, J = 8.9 Hz, 2H), 6.89 (d, J = 8.9 Hz, 2H), 6.48 (d, J = 16.09 . Synthesis of compound 11a.Ethyl 3-(4-(3-phenylureido)phenyl)propanoate (11a).General Procedure C In an argonated solution of intermediate 10a (0.31 g, 1 mmol, 1 eq.) in ethanol (30 mL), 0.24 g of 10% palladium on activated charcoal (10% Pd/C) was added at 0 °C.The resulting mixture was stirred at room temperature under H 2 atmosphere for 16 h.The product was filtered through a small pad of Celite, and concentrated to afford the product as a white solid (0.312 g, > 99%). 1  In vitro PfHDAC1 enzymatic assay.PfHDAC inhibition assays were performed by BPS Bioscience (San Diego, CA).Compounds 6c-8c, and nexturastat A (NextA, 16) were dissolved in DMSO with the highest concentration at 5 mM.Then, each DMSO solution was directly diluted 10 × fold into the HDAC assay buffer for an intermediate dilution of 10% DMSO in HDAC assay buffer.Then, 5 µL of the intermediate dilution was added to a 50 µL reaction so that the final concentration of DMSO is 1% in all the reactions.The enzymatic reactions for the PfHDAC1 were conducted at 37 ºC for 2 h in a 50 µL mixture containing HDAC assay buffer in duplicate, 5 µg BSA, HDAC substrate (peptide BOC-Ac-Lys-AMC, catalogue number: 50063), PfHDAC1 enzyme, and the tested compound.Enzymatic reactions were stopped by 50 μL/well of 2 × HDAC and the plate was incubated for further 15 min (room temperature).Fluorescence intensity was measured at an excitation of 360 nm and an emission of 460 nm using a Tecan Infinite M1000 microplate reader.All PfHDAC1 activity assays have been performed in duplicates at each tested concentration (1 and 10 μM), besides Vorinostat (SAHA, 1), which was tested at 0.01, 0.1, and 1.0 μM concentrations.The fluorescent intensity data were analyzed using GraphPad Prism (v9.5.2).In the absence of the compound, the fluorescent intensity (F t ) in each data set was defined as 100% activity.In the absence of HDAC, the fluorescent intensity (F b ) in each data set was defined as 0% activity.The percent activity in the presence of each compound was calculated according to the following equation: % activity = (F-F b )/(F t -F b ), where "F" is the fluorescent intensity in the presence of the compound.
Western Blot analysis.Equal amounts of proteins from each extract were solubilized in sample buffer (50 mM Tris-HCl (pH 6.8), 2% SDS, 32% glycerol, 1.5 mM bromophenol blue) and subjected to SDS-PAGE (20%).Proteins were transferred to PVDF membranes, 5% non-fat dry milk in TBS with Tween 20 (0.1%) was used as blocking agent for 1 h at room temperature.After, incubated with the antibodies overnight at 4 °C (Acetyl Histone H3 Lys9 C5B11 Cell Signaling, Acetyl Histone H4 Lys16 E2B8W Cell Signaling, Histone H3 96C10 Cell Signaling and Histone H4 D2X4V Cell Signaling).For the analysis of protein acetylation levels, the membranes were stripped and re-probed with the corresponding anti-total protein.Mouse monoclonal anti-α-Tubulin (B512 Sigma-Aldrich) was used as loading control.Detection was performed by enhanced chemiluminescence using horseradish peroxidase-conjugated secondary antibodies (Vector Laboratories, Burlingame, CA, USA) and SuperSignal TM West Pico PLUS Chemiluminescent substrate kit (Thermo Scientific).Images were acquired using ChemiDoc TM Imaging System (BioRad Laboratories, CA, USA).Quantitative densitometry was carried out using ImageLab software (Bio-Rad Laboratories, CA, USA).The volume density of the chemiluminescent bands was calculated as an integrated optical density × mm 2 after background correction from each independent experiment (N = 3).
DMPK evaluation.To determine stability in hepatic microsomes, the compound (1 μM) was incubated with 1 mg/mL human or mouse hepatic microsomes at 37 °C with continuous shaking 48 .At 0, 5, 10, 20, 40, and 60 min time points, aliquots were removed and acetonitrile was added to quench the reactions and precipitate the proteins.Samples were then centrifuged through 0.45 μm filter plates and half-lives (T 1/2 s) were determined by LC-MS/MS.To determine cytochrome P450 (CYP450) inhibition, 10 μM compound was incubated with human liver microsomes and selective marker substrates (1A2, phenacetin demethylation to acetaminophen; 2C9, tolbutamide hydroxylation to hydroxytolbutamide; 2D6, bufuralol hydroxylation to 4′-hydroxybufuralol; 3A4, midazolam hydroxylation to 1′-hydroxymidazolam).After a 10 min incubation, the reaction was terminated and the per cent inhibition was determined.

Molecular modelling
Homology model and protein preparation.The human HDAC1 was retrieved from a representative simulation frame of our previous work 29 .The Plasmodium falciparum 3D7 HDAC1 homology model was generated from the (UniProt: Q7K6A1_PLAF7, full sequence) using Phyre2 on intensive mode with standard options 49 .Model was validated by checking its Ramachandran plot and overall energy levels, showing low confidence for the C-terminal after His375.Human model of HDAC6 was generated after the Danio rerio structure (PDB ID: 6DV0) similarly as described above.All protein structures were prepared using the Protein Wizard Preparation tool, with standard options and the homology model was further refined to remove sterical clashes.
Molecular docking.Three-dimensional ligand structures were generated with LigPrep, using Epik to predict their protonation in pH 7.0 ± 1.0, diastereoisomers configuration were derived from the synthesis.The OPLS4 force field was employed for structure generation.Docking was performed using Glide 50,51 using the Zn 2+ ion to orient the binding pocket center, employing XP scoring function.Since redocking of vorinostat was poorly performed, for each ligand up to 10 poses were generated, from which we then selected the conformation for MD based on relevant interactions.
Molecular dynamics simulations.MD simulations were carried out by using the Desmond engine 52 with the OPLS4 force-field 53 .The system encompassed the protein-ligand/cofactor complex, a predefined water model (TIP3P) 54 as a solvent and counterions (Na + or Cl -adjusted to neutralize the overall system charge).The system was treated in a cubic box (13 Å) with a periodic boundary condition (PBC) specifying the size of the box from the box edges to any atom of the protein.Short-range coulombic interactions were calculated using 1 fs time steps and 9.0 Å cut-off value, whereas long-range coulombic interactions were estimated using the Smooth Particle

Figure 5 .
Figure 5. Hydroxamic acid derivatives as potent in vitro PfHDAC1 inhibitors.(A) Residual activity (%) of a set of representative inhibitors (6c-8c) against recombinant PfHDAC1 tested in two concentrations and compared against the controls NextA (16) and SAHA(1).Enzyme activity was calculated by normalizing the data against the DMSO treated samples and discounting the background.Data are presented as shown above, with mean ± S.D. (n = 2).Differences to this value were analyzed by one way ANOVA using Dunnett's method for multiple comparison (asterisks), where ****P < 0.001, comparing against the SAHA 1 µM treatment.Comparison of group/concentration against the DMSO control is represented as † for P < 0.0001.(B) Immunoblotting analysis for histones (H3 and H4, unmodified and acetylated), and α-tubulin (αTub) in infected erythrocytes (trophozoites), incubated with DMSO (0.05%, control), SAHA (1), NextA(16) or compounds 6c-8c, with 10 × IC 50 for 4 h, representative gel from N = 3 (see Supporting Informationfor all gels).Numbers on side of the bands represent the predicted mass (kDa) for each protein.(C,D) Band quantification of histone H3 (C) and H4 (D), respectively, normalizing their acetylated detection against their unmodified version.Differences to this value were analyzed by one way ANOVA using the Dunnett's post correction, comparing each group against the DMSO control, as rank tests where P values are explicitly depicted.

Table 1 .
Antiplasmodial activity and cytotoxicity of compounds . TFA d a Figure 2. Design and synthesis of analogues 6a-d, 7a-c, and 8a-k.Reagents and conditions
b n.d.: not determined.c Values in brackets indicate 95% confidence intervals.d n.a.: not applicable.

8c-Na + 8c-K + 8d-Na +
the background from both.Curves were plotted using non-linear regression based on means from three independent experiments measured in triplicate.Error bars indicate standard deviation, whose values are provided in the respective table.For curves that do not reach the zero-point IC50was not calculated and is indicated as an approximation.Vol.:(0123456789) Scientific Reports | (2023) 13:21006 | https://doi.org/10.1038/s41598-023-47959-zwww.nature.com/scientificreports/ .

Table 4 .
29crosome stability assessment and CYP inhibition profile of selected HDACis.aHalf-life(t1/2 ) in 1 mg/mL hepatic microsomes.bInhibition<10%.cTestedconcentration:40 µM.dTested concentration: 1.0 µM. e n.a.: not applicable.UV detection at 254 nm.Purity of tested compounds was > 95%, determined through analytical HPLC.All tested compounds were analyzed using a high liquid chromatograph (Shimadzu) coupled to an accurate Q-TOF mass spectrometer, Compact model (Bruker Daltonics), and electrospray ionization interface.Isolated compounds were dissolved in DMSO and subjected to separation using a Kinetex 1.7 μm EVO C18 100 Å (100 × 2.1 mm; Phenomenex Ltd.), with a mobile phase composed of 0.1% formic acid in a mixture of water and acetonitrile.The flow rate was 0.4 mL/min with a gradient program: initial 10% B, 100% B at 5 min, 25% B at 7 min, and a 5 min post-run at 10% B. Injection volume was 20 μL, and column temperature was maintained at 40 °C.The Q-TOF/MS operated in positive mode with specific parameters: ion gas source (N 2 ) temperature 200 ℃; nebulizer pressure 45 psi; and capillary voltage of 2,800 V. Mass spectrometer was operated in MS scan mode with internal mass calibration using sodium formate29.