Identification of covalent fragment inhibitors for Plasmodium falciparum UCHL3 with anti-malarial efficacy

Abstract

Malaria continues to be a major burden on global health, responsible for 619,000 deaths in 2021. The causative agent of malaria is the eukaryotic parasite Plasmodium. Resistance to artemisinin-based combination therapies (ACTs), the current first-line treatment for malaria, has emerged in Asia, South America, and more recently Africa, where >90% of all malaria-related deaths occur. This has necessitated the identification and investigation of novel parasite proteins and pathways as antimalarial targets, including components of the ubiquitin proteasome system. Here, we investigate Plasmodium falciparum deubiquitinase ubiquitin C-terminal hydrolase L3 (PfUCHL3) as one such target. We carried out a high-throughput screen with covalent fragments and identified seven scaffolds that selectively inhibit the plasmodial UCHL3, but not human UCHL3 or the closely related human UCHL1. After assessing toxicity in human cells, we identified four promising hits and demonstrated their efficacy against asexual P. falciparum blood stages and P. berghei sporozoite stages.

Graphical Abstract

In 2021, there were an estimated 247 million cases of malaria worldwide, resulting in 619,000 deaths.¹ Malaria is caused by the eukaryotic protozoan parasite, Plasmodium, which is spread through the bite of an infected Anopheles mosquito.¹ Of the five species of Plasmodium that cause disease in humans, P. falciparum is the most prevalent and the deadliest. The current first-line treatment for P. falciparum malaria are artemisinin-based combination therapies (ACTs), which consist of a short-lived artemisinin that rapidly reduces parasite biomass paired with a longer-lived partner drug to clear remaining parasites. Partial resistance to artemisinin has emerged in Asia, South America, and Africa.^2–4 This has spurred a call to action by the global community to address the significant need for novel therapeutic targets and molecular scaffolds for antimalarial drug discovery.

The ubiquitin-proteasome system (UPS) is conserved in eukaryotes, responsible for regulating several vital cell processes such as cell-cycle progression, protein trafficking, signaling, DNA repair, and protein quality control.⁵ Ubiquitin is attached to substrate proteins through a series of E1 ubiquitin activating enzymes, E2 ubiquitin conjugating enzymes, E3 ubiquitin ligating enzymes, and sometimes E4 ubiquitin chain elongation factors.⁶ The type and length of ubiquitin linkage(s) dictates substrate fate.⁷ Monoubiquitin and polyubiquitin chains can be edited or removed from substrate proteins via deubiquitinases (DUBs), which are also responsible for generating free ubiquitin from ubiquitin precursors.^8,9

In humans, the UPS is an established target for cancer, and several components of the UPS including the proteasome, E3 ubiquitin ligases, and DUBs have been validated as therapeutic targets.^10,11 The UPS has also been gaining increasing interest as a therapeutic target in malaria. Proteasome inhibitors have been shown to kill parasites at the liver, blood, and mosquito stages, as well as block parasite transmission to mosquitos.^12–15 Multiple classes of Plasmodium-specific proteasome inhibitors have now been developed, several of which have been shown to overcome artemisinin resistance.^16–18 In addition, several classes of mammalian DUB inhibitors were demonstrated to have antimalarial activity against Plasmodium both in vitro and in vivo.¹⁹

Here, we investigate P. falciparum ubiquitin C-terminal hydrolase L3 (PfUCHL3; PF3D7_1460400) as a potential antimalarial target. PfUCHL3 is a cysteine protease that was shown to have both deubiquitinase and deneddylase activity in P. falciparum.^20,21 However, little is known regarding the natural substrate(s) of this enzyme within the parasite.²² PfUCHL3 is expressed in ookinetes, oocysts, gametocytes, and throughout asexual development.^23–26 Researchers were able to create transgenic parasites overexpressing wild type PfUCHL3, but not a catalytically dead mutant, suggesting that the enzyme may be essential in asexual blood stages.²⁰ PfUCHL3 shares only 36% sequence identity with human UCHL3 (hUCHL3) and crystal structures of the human and parasite enzymes reveal differences in the active site that could be exploited to selectively target PfUCHL3.²⁰ By conducting a structure-based virtual screen of the pathogen box from Medicines for Malaria Venture, one group recently identified two putative PfUCHL3 inhibitors that have anti-malarial activity against asexual blood stage parasites.²⁷ To our knowledge, no other PfUCHL3 inhibitors have been reported and we sought to further explore this DUB as a potential anti-malarial therapeutic target.

To this end, we performed a high-throughput screen on a library of fragments that contain cysteine reactive electrophiles. We identified seven covalent-fragment inhibitors of PfUCHL3 that have selectivity over human UCHL3 (hUCHL3) and the closely related human DUB, UCHL1 (hUCHL1). Four molecules were found to exhibit low toxicity to human HEK293 cells and have anti-parasitic activity against P. falciparum asexual blood stages and P. berghei sporozoite stages. The results of this study suggest targeting PfUCHL3 may be a viable anti-plasmodial therapeutic approach.

To begin, we screened a 1700-member covalent fragment library from Enamine (Kyiv, Ukraine) containing diverse scaffolds and four known cysteine electrophiles: activated nitriles, epoxides, chloroacetamides and acrylamides. For the primary PfUCHL3 screen, we utilized the fluorescence-based ubiquitin rhodamine-110 (Ub-Rho) enzymatic activity assay.²⁸ Iodoacetamide served as a positive control while DMSO served as the negative control. The Z’-factor analysis for the controls was determined to be 0.76 indicating the assay exhibited a sufficiently wide window between signal-to-noise to reliably identify hit molecules.²⁹ Fragments were first screened in singlet at a single-concentration of 500 μM, with a 1 h preincubation prior to the addition of Ub-Rho substrate. Percent inhibition was determined compared to a DMSO control set to 100% activity. The hit criteria were set at 90% inhibition of PfUCHL3, which yielded 76 hits (Figure S1A). The 76 hits were then validated in triplicate against PfUCHL3 in the same assay at which point 12 did not validate and were triaged leaving 64 to move forward (Figure S1B). The 64 hits represented two electrophile classes: 9 acrylamides and 55 chloroacetamides. We prioritized the acrylamide hits, due to acrylamide being a pharmaceutically acceptable cysteine electrophilic moiety targeting many human kinases.^30–33 Eight of the nine were commercially available as dry powders, re-ordered, and evaluated for off-target inhibition against the human DUB orthologs, hUCHL1 and hUCHL3. One hit, UCH0084, did not pass quality control characterization and was triaged. The remaining seven molecules were evaluated in dose-response assays against PfUCHL3, hUCHL3, and hUCHL1 to determine half-maximal inhibitory concentrations (Figure 1 and Table 1).

Figure 1.

PfUCHL3 inhibition dose-response curves for seven validated acrylamide hits.

Table 1.

PfUCHL3 high-throughput screen hits with selectivity over human DUBs

Cmpd	Structure	PfUCHL3 IC50 (μM)a	hUCHL3 IC50 (μM)a	hUCHL1 IC50 (μM)a	Selectivity ratiob
UCH0080		27.9 (24.6 - 31.6)	>500	>500	>17.8
UCH0081		120 (99.8 - 143.2)	>500	>500	>4.2
UCH0082		68.6 (57.4 - 81.9)	>500	>500	>7.2
UCH0083		89.8 (74.5 - 108.3)	282 (193.8 – 419.9)	>500	3.1
UCH0085		328 (274.2 - 397.2)	>500	>500	>1.3
UCH0086		110 (102.5 - 136.4)	>500	>500	>4.5
UCH0087		23.2 (20.8 - 31.0)	>500	>500	>21.6

^aExperiments were performed in technical triplicate after 1 hour incubation of analog with DUB and averages are reported as IC₅₀ values. Errors are reported as 95% confidence intervals in parentheses. N.D. = not determined.

^bSelectivity ratio = hUCHL3/PfUCHL3.

Of the seven hits, UCH0080, UCH0082, UCH0083, and UCH0087 all displayed sub-100 μM (23 – 90 μM range) potency against PfUCHL3, with UCH0080 and UCH0087 being the most potent. Six of the seven compounds did not exhibit any inhibition of hUCHL3 even up to 500 μM. Only UCH0083 displayed sub-500 μM hUCHL3 potency, with IC₅₀ values of 282 μM. No compound displayed inhibitory activity against hUCHL1 at 500 μM. At the hit stage, the high-throughput screen produced validated PfUCHL3 hits with a range of activity against the desired target and varying selectivity against the two closest human DUB off-targets.

Molecules were next evaluated for efficacy in P. falciparum asexual blood-stage parasites. Asynchronous Cam3.II C580Y parasites were exposed to a single dose (500 μM) of UCH0080, UCH0081, UCH0082, UCH0083, UCH0085, UCH0086, or UCH0087 and parasite viability was assessed in the following replication cycle at 72 h by high content imaging.³⁴ Parasite DNA was stained with Hoechst 33342, and respirating parasite mitochondria were detected with MitoTracker^™ Deep Red FM. Red Blood Cells (RBCs) were visualized by staining with Wheat Germ Agglutinin Alexa Fluor^™ 488 (Figure S2). Of the seven compounds evaluated, five displayed >99% inhibitory activity against P. falciparum at 500 μM (Figure 2). The other two molecules, UCH0086 and UCH0087, displayed < 10% inhibitory activity against parasites despite exhibiting comparable potency to the other five analogs against the purified PfUCHL3. Assessment of the scaffolds for UCH0086 and UCH0087 does not provide any clear physicochemical or structural differences between these two molecules and the hits with anti-parasitic activity that would explain the lack of potency against P. falciparum. Given the lack of anti-plasmodial activity these molecules were deprioritized from further evaluation (Figure 2).

Figure 2.

Screen of lead compounds against asexual P. falciparum asexual blood stages. Asynchronous Cam3.II K13 C580Y parasites were exposed to 500 μM of the indicated compound. 150 nM dihydroartemisinin (DHA) was used as positive control for 100% parasite killing. DMSO (0.25% v/v), equivalent to the highest level of DMSO in compound-treated samples, was used as a negative control. Parasite viability was assessed after 72 h by high-content imaging. (A) Shown are representative images from high content imaging. Hoechst (blue) was used to stain for parasite DNA and Mitotracker Deep Red (pink), which stains respirating parasite mitochondria, was used as a viability marker. Red blood cell density is shown in Figure S2. Scale bar = 50 μm. (B) Shown are mean % inhibition ± standard error of the mean (S.E.M.) from 3 biological replicates, each performed in technical duplicates.

Dose response assays were then conducted with UCH0080, UCH0081, UCH0082, UCH0083, and UCH0085 against Cam3.II K13 C580Y, Cam3.II K13 WT, and 3D7 strain parasites. These parasite strains have numerous differences genetically and consequently in drug resistance profiles (Table S1). Specifically, Cam3.II parasites have mutations in the known drug resistance genes dhfr (PF3D7_0417200), dhps (PF3D7_0810800), and pfcrt (PF3D7_0709000), which confer resistance to the antimalarials pyrimethamine, sulfadoxine, and chloroquine, respectively. In addition, Cam3.II K13 C580Y parasites harbor an additional mutation in kelch 13 (K13; PF3D7_1343700), which confers resistance to artemisinin.³⁵ Asynchronous parasites were exposed to a range of compound concentrations from 0.5 μM to 500 μM and parasite viability was assessed 72 h later (Figure 3, images provided in Figures S3 and S4).³⁴

Figure 3.

Potency of hit compounds against P. falciparum asexual blood stages. Asynchronous parasites were exposed to a range of compound concentrations (0.5 μM to 500 μM). Parasite viability was assessed 72 h later. (A) IC₅₀ curves against Cam3.II K13 C580Y (B) Shown are mean IC₅₀ values ± S.E.M. for Cam3.II K13 C580Y(blue), Cam3.II K13 WT (grey), and 3D7 strain parasites (teal) from 3 biological replicates. Statistical significance was examined using a Mann-Whitney U test. *p < 0.05; ns = not significant.

Analogs provided a range in IC₅₀ values from 2 μM to 180 μM for the five tested molecules, with some compounds exhibiting pronounced parasite strain-specific effects (Figure 3 and Table 2). For example, relative to 3D7 parasites, Cam3.II parasites had 7-fold reduced susceptibility to UCH0085 (Cam3.II K13 C580Y: 146.1 μM and Cam3.II K13 WT: 140.0 μM vs 3D7: 21.2 μM, p = 0.0159 and p = 0.0286; Figure 3B). In addition, sensitivity to UCH0082 was dependent on K13 genotype. In the isogenic parasites differing only at the K13 locus, K13 C580Y parasites displayed 2-fold higher IC₅₀ values than K13 WT parasites (Cam3.II K13 C580Y: 106.6 μM vs Cam3.II K13 WT: 50.0 μM and 3D7: 54.9 μM, p = 0.0159). Note that Cam3.II parasites have mutations in the known drug resistance genes dhfr, dhps, mdr1, and pfcrt that are common in clinical isolates from Southeast Asia, as well as numerous mutations in unstudied genes which may contribute to antimalarial drug resistance. These data demonstrate the importance of using multidrug-resistant parasites for drug discovery purposes, as parasites in the field may already be highly resistant to certain compounds, even if these compounds have high potency against multi-drug sensitive parasites.

Table 2.

In vitro inhibition constants (IC₅₀) against P. falciparum strains

Cmpd	Structure	3D7 (μM)a	Cam3.II K13 WT (μM)a	Cam3.II K13 C580Y (μM)a
UCH0080		98.1 ± 12.0	59.5 ± 3.9	90.9 ± 9.1
UCH0081		2.3 ± 0.1	2.6 ± 0.2	3.4 ± 0.2
UCH0082		54.9 ±4.8	50.0 ± 3.9	106.6 ± 17.7
UCH0083		183.2 ± 12.9	109.7 ± 3.3	160.3 ± 26.1
UCH0085		21.2 ± 1.2	140.0 ± 19.3	146.1 ± 3.3
UCH0114		nt	nt	>500
MMV688704b		0.27b	nt	32.4 ± 1.3

^aMean IC₅₀ values ± S.E.M. from at least 3 biological replicates.

^bMMV688704 reported by Bharti et al.²⁷ nt = not tested.

Of the five analogs, UCH0081 exhibited the greatest potency with an IC₅₀ of 3.4 μM for Cam3.II K13 C580Y, 2.6 μM for Cam3.II K13 WT, and 2.3 μM for 3D7 (Cam3.II K13 C580Y vs Cam3.II K13 WT p = 0.1111 and 3D7 p = 0.0317; Figure 3B). Interestingly, UCH0081 is 30-fold more potent against the parasite than against the recombinant PfUCHL3. This likely suggests that either PfUHCL3 is not the primary target or that there are multiple targets leading to polypharmacology. To assess whether the inhibition exhibited by UCH0081 was dependent on a covalent inhibition mechanism we synthesized the non-electrophilic analog UCH0114 (Scheme S1) in which the acrylamide was replaced with an isosteric ethyl amide moiety, then evaluated the potency against asynchronous Cam3.II K13 C580Y (Figure 4 and Table 2). The antimalarial activity for UCH0114 was abrogated and we were unable to determine an IC₅₀ value (Figure 4 and Table 2), indicating that the mode of action for UCH0081 is likely covalent. We also carried out a dialysis assay^36–38 to confirm irreversible covalent inhibition with the top four hits in which PfUCHL3 was pre-incubated with each hit then the enzyme-inhibitor complexes were subject to dialysis. After dialysis, if PfUCHL3 is reversibly modified by the inhibitors then the dialyzed enzyme should exhibit recovered enzymatic activity. This was not the case for the four molecules tested (Figure S5; UCH0080, UCH0081, UCH0082, UCH0083) as there was no difference in enzyme activity pre- and post-dialysis confirming the mechanism of inhibition, at least for these four and likely the remaining hits, is via irreversible covalent inhibition. Additionally, we compared UCH0081 activity to MMV688704, which was recently identified as a putative PfUCHL3 inhibitor.²⁷ Notably, under the assay conditions used here, UCH0081 was eightfold more potent than MMV688704 (IC₅₀ value = 32.8 μM, p = 0.0357; Figure 4 and Table 2) while the molecule was reported to have an IC₅₀ value of 0.27 μM against 3D7 strain.

Figure 4.

UCH0081 anti-plasmodium activity is mediated through covalent inhibition with greater potency than MMV688704. (A) Dose-response curves for UCH0081, UCH0114, MMV688704 versus asynchronous Cam3.II K13 C580Y parasites. (B) Mean IC₅₀ values ± S.E.M. from at least 3 biological replicates. Statistical significance was examined using a Mann-Whitney U test. *p < 0.05.

To de-risk scaffolds, the toxicity of the top five compounds was evaluated in human embryonic kidney 293 (HEK293) cells using CellTiter-Glo^® assay (Promega, Madison, WI). For this assay, cells were treated with 500 μM of each inhibitor, as well as doxyrubicin as a positive control, and viability was assessed 24 h later. Cell viability was quantified by luminescence and normalized to the DMSO control set at 100% viability. Results indicated that four of the five molecules – UCH0080, UCH0081, UCH0082, UCH0083 – displayed low toxicity to the mammalian cell line with greater than 80% cell viability at 500 μM. In contrast, UCH0085 demonstrated < 50% cell viability at the same concentration after 24 h incubation (Figure 5). Based on these results, UCH0080, UCH0081, UCH0082, and UCH0083 were prioritized for additional antiparasitic evaluation against the sporozoite stage of P. berghei. To further de-risk UCH0080, UCH0081, UCH0082 and UCH0083, glutathione stability assays were performed. There was some observed glutathione conjugation observed, UCH0083 had an observed t_1/2 of 1.03 hours while the other three molecules had observed half-lives greater than 6 hours (Table S2). The most stable analog appeared to be UCH0082 with 92% of molecule remaining after completion of the incubation period. These data are useful for future pharmacokinetic evaluation of the hits.

Figure 5.

Cell viability of HEK293 cells for molecules tested at 500 μM. Data shown as percent viability relative to DMSO control.

The parasite’s clinically asymptomatic development in the liver is a desirable target for chemotherapy aimed at disease prophylaxis. Therefore, activity of the four prioritized analogs were tested for activity in blocking sporozoite infection of and development within hepatocytes, using the rodent-infective P. berghei as a model. All molecules were effective in inhibiting parasite growth in the human hepatoma cell line HepG2 at non-toxic concentrations (Figure 6). UCH0081 showed the greatest efficacy (IC₅₀ of 6.4 μM) which was comparable to the efficacy against the asexual blood stage of P. falciparum. This was followed by UCH0080 (IC₅₀ of 49 μM). UCH0082 and UCH0083 displayed comparable IC₅₀ of 80 μM and 87 μM, respectively.

Figure 6.

Anti-P. berghei sporozoite activity of UCH0080 (blue), UCH0081 (red), UCH0082 (green), and UCH0083 (purple). The calculated IC₅₀ values for each analog are shown at right.

In conclusion, PfUCHL3 represents a potential anti-malarial therapeutic target and few inhibitors have been reported to date. In this study we conducted a high-throughput screen to identify covalent fragment hits against the recombinant PfUCHL3 enzyme and hits were counterscreened for activity versus two human DUB homologs, hUCHL3 and hUCHL1. Seven hits displayed generally good selectivity with only two molecules exhibiting less than 500 μM potency against hUCHL3. These seven molecules were then evaluated for activity against P. falciparum asexual blood and hepatocyte stages. Analog UCH0081 exhibited the most potent anti-Plasmodium activity in the 2 – 4 μM range against the three strains tested. Additionally, this analog exhibited comparable potency in blocking sporozoite stage infection of P. berghei in a hepatocyte model. These stages are the targets of causal prophylactic strategies that aim to eradicate or substantially decrease parasite burden before the infection turns symptomatic and parasites form transmissible gametocytes in the bloodstream. It is crucial for controlling malaria caused by P. vivax and P. ovale as these species form hypnozoites that stay dormant in the liver and when reactivated cause disease relapse. Since there are no diagnostic tests for liver stage infection, drugs that prevent sporozoite infection and/or liver stage development will be crucial for malaria eradication.

While these molecules were identified from PfUCHL3 HTS, there is still work to be completed to link the PfUCHL3 inhibition to the anti-plasmodial activity. The anti-parasitic activity for UCH0081 is 30-fold more potent in the anti-parasite assay compared to the in vitro inhibition of PfUCHL3, a trend that would suggest that either there are multiple targets for UCH0081, which may or may not include PfUCHL3, or that the antiparasitic activity for UCH0081 is through covalent inhibition of a different target altogether. The remaining analogs all exhibited anti-Plasmodium activity that was less potent than activity against recombinant PfUCHL3, which would be expected if PfUCHL3 is indeed the target of these compounds. Due to the lack of a PfUCHL3 antibody it is difficult to confirm on target engagement of these molecules with PfUCHL3 through standard ubiquitin activity-based probe assays.^39–42 Experiments using both chemoproteomics and resistant mutant isolation followed by genome wide sequencing approaches are currently in progress to assess intracellular target engagement for all molecules. Nonetheless, these hits represent viable starting points for medicinal chemistry optimization with the potential to be translated into new anti-malarial therapeutics.