Inhibition of Fatty Acid Oxidation as a New Target To Treat Primary Amoebic Meningoencephalitis

Primary amoebic meningoencephalitis (PAM) is a rapidly fatal infection caused by the free-living amoeba Naegleria fowleri. The amoeba migrates along the olfactory nerve to the brain, resulting in seizures, coma, and, eventually, death. Previous research has shown that Naegleria gruberi, a close relative of N. fowleri, prefers lipids over glucose as an energy source. Therefore, we tested several already-approved inhibitors of fatty acid oxidation alongside the currently used drugs amphotericin B and miltefosine.

Inhibition of metabolic processes essential to microorganisms is a fruitful strategy for the development of effective drugs (16). Several widely used drugs, such as the antimalarials atovaquone and proguanil and the broad-spectrum antiprotozoal, antihelminthic, and antiviral drug nitazoxanide, target the metabolism of the pathogen to exert their killing effect (17,18).
Previous research by our group showed that N. gruberi, a close relative of N. fowleri, prefers fatty acids as a food source (19). This led us to the hypothesis that inhibiting fatty acid oxidation (FAO) could inhibit growth of or even kill the amoeba. We identified several drugs that are currently used or have been used to inhibit fatty acid metabolism in different parts of this pathway. All of those drugs target enzymes that are present in the N. gruberi and N. fowleri genome (19) (see also Fig. 1 and Discussion). As the fatty acid preference was shown in N. gruberi, we first determined the effects of these compounds on N. gruberi. We then tested promising compounds on the actual pathogen, N. fowleri, and finally determined whether there was synergy present when the compounds were combined in a checkerboard assay.

RESULTS
The inhibitory effects of all compounds on N. gruberi determined through area under the curve (AUC) calculation are represented in Fig. 2. Thioridazine (TDZ) inhibited growth of N. gruberi in a concentration-dependent manner, inhibiting 50% of growth at approximately 10 M (Fig. 2A). Addition of perhexiline (PHX) resulted in an inhibition level of about 50% at concentrations above 50 M (Fig. 2B). Etomoxir (ETO) addition resulted in clear inhibition at concentrations above 600 M (Fig. 2C), and addition of valproic acid (VPA) resulted in inhibition of growth in a concentration-dependent manner, with inhibition of 50% of growth at around 700 M (Fig. 2D). Orlistat (ORL) inhibited circa 50% of growth at concentrations of 50 M and higher (Fig. 2E), and amphotericin B (AMB) was very effective at inhibiting growth, inhibiting circa 75% at concentrations of 0.2 M and higher (Fig. 2F). Addition of miltefosine (MIL) resulted in

FIG 2
Growth curves of Naegleria gruberi were obtained in the presence or absence of inhibitors of fatty acid oxidation or drugs currently used to treat primary amoebic meningoencephalitis. Optical density was measured daily over a 5-day period. Data shown represent the area under the growth curve (AUC) determined for the indicated compounds and the respective controls, including lines representing 100% and 50% of the control AUC. At the top of the graph, data representing the capacity for regrowth are shown. ϩ, clear regrowth; ϩ/Ϫ, inconsistent or little regrowth; Ϫ, never any regrowth; N.D., not done. Experiments were performed twice in triplicate wells; error bars represent standard deviations (SD). inhibition in a concentration-dependent manner, with efficient inhibition of growth at 80 M (Fig. 2G). The capacities for regrowth of the amoebae after 5 days of exposure differed for the examined compounds, as can be seen in the bars above the individual graphs in Fig. 2. Amoebae incubated with VPA and ORL showed regrowth for all concentrations used, while PHX consistently prevented regrowth at 90 M. The amoebae showed regrowth after TDZ exposure at up to a concentration of 20 M, while the amoebae showed regrowth after ETO exposure at up to a concentration of 600 M. Amoebae incubated with MIL showed regrowth after exposure to concentrations below 80 M and inconsistent regrowth at the examined concentrations over 80 M. AMB was most effective in preventing regrowth, always blocking regrowth at a concentration of 0.4 M or higher (Fig. 2).
Next, the compounds were also assessed for their effect on N. fowleri. This was done in two ways: via viability staining with CellTiter-GLO and with direct cell counting using a flow cytometer. The 50% inhibitory (IC 50 ) concentrations of all drugs on both organisms are listed in Table 1. The IC 50 results reported here for compounds tested on N. gruberi represent approximations, as the range of concentrations tested was narrow. Results of the CellTiter-GLO viability stain assay can be seen in Fig. 3, and results of cell counting can be seen in Fig. S1 in the supplemental material. These figures show that absence of viability and absence of cell growth were observed after exposure to TDZ, PHX, and ETO, revealing that these compounds are effective against N. fowleri as well as against N. gruberi. The efficacy of ETO and PHX was higher against N. fowleri than against N. gruberi, and the IC 50 levels of both drugs were about 5-fold lower against N. fowleri than against N. gruberi. VPA and ORL showed some inhibition of N. fowleri growth at high concentrations, but their efficacy was much lower against N. fowleri than against N. gruberi. The effects of AMB and MIL were roughly similar against N. gruberi and N. fowleri. The IC 50 calculations show concordance between the two methods, confirming the efficacy of the compounds against N. fowleri in two ways (Table 1). Next, compounds were combined in these IC 50 concentrations to screen for a possible synergistic effect of combinations of compounds against N. fowleri (see Table S1 in the supplemental material). Checkerboard assays were performed for the best six combinations of the drugs, after which F min was calculated. The combinations MIL plus PHX, MIL plus TDZ, MIL plus VPA, PHX plus TDZ, and TDZ plus VPA showed additivity. The combination of ETO and MIL resulted in an F of 0.5, indicating that synergy was present when these drugs were combined (Table S1). Further analysis of the synergistic effect of the combination ETO and MIL with the program Combenefit resulted in a mapped surface analysis whose results can be seen in Fig. 4. The map shows that the synergy was most pronounced when 12.5 and 25 M concentrations of MIL were combined with 25 to 200 M concentrations of ETO.

DISCUSSION
Our study showed that fatty acid oxidation (FAO) inhibitors clearly inhibited growth of both N. gruberi and N. fowleri in vitro. Hence, not only are lipids the preferred food source for N. gruberi, but oxidation of fatty acids seems to be essential also for growth of N. fowleri. The current treatment regimen using miltefosine (MIL) and amphotericin B (AMB) was confirmed to be effective at inhibiting growth in vitro, which is in agreement with previous reports (20-23) and validates our assays performed to detect compounds that inhibit growth of Naegleria. The importance of fatty acid oxidation for N. fowleri was demonstrated in a recent in vivo study, as Herman et al. observed upregulation of genes of N. fowleri involved in FAO after mouse passage (24). Our results now show that FAO inhibition is a valid target for new PAM therapy options, as etomoxir (ETO), perhexiline (PHX), and thioridazine (TDZ) showed total growth inhibition of N. fowleri. Furthermore, our results show that there is additivity of MIL combined with PHX, TDZ, and valproic acid (VPA) and synergy between ETO and MIL, providing evidence that inhibition of fatty acid oxidation can be a valuable addition to the current treatment regimen.
We observed some differences between N. gruberi and N. fowleri in the levels of efficacy of the FAO inhibitors. All FAO inhibitors affected N. gruberi growth, but the effects of VPA and orlistat (ORL) were much less profound on N. fowleri than on N. gruberi. In contrast, ETO and PHX were more effective at inhibiting growth of N. fowleri than N. gruberi. Taking the targets of the FAO inhibitors into account, we can hypothesize on the reasons for these differences. The investigated FAO inhibitors affect different enzymes involved in lipid metabolism (depicted in Fig. 1). TDZ inhibits peroxisomal oxidation of lipids (25,26). ORL inhibits lipases, enzymes that hydrolyze triacylglycerol, thereby obstructing the first step in the breakdown of lipids (27). ETO and PHX inhibit carnitine palmitoyltransferase-1 (CPT-1), blocking transport of fatty acids into mitochondria (28,29). Among other activities, VPA interferes mainly with mitochondrial ␤-oxidation (30). The targets of the FAO inhibitors are present in the N. gruberi genome as well as in the N. fowleri genome, showing that regarding the metabolic properties, the two organisms are very much alike (19, 31, 32). However, this does not imply that the enzymes are exactly identical in amino acid sequence. Minor amino acid differences could result in small structural differences and hence in differences in the effects of the various drugs. Furthermore, unavoidable differences between N. fowleri and N. gruberi under in vitro growth conditions could play a role, as the optimal culture media (PYNFH versus Bacto Casitone) and culture temperatures (25°C versus 37°C) are different between the two, resulting in different metabolic rates.
We observed that several FAO inhibitors show additivity (PHX, TDZ, and VPA) or synergy (ETO) when combined with MIL. Unfortunately, synergy between AMB and the FAO inhibitors was not observed. This would be of importance, as AMB has serious side effects (14). There is also a risk for serious adverse events when using MIL (33). We did observe synergy between ETO and MIL, which is a promising result as this could potentially lead to lowering of MIL dosages and subsequent reduction in the risk of serious adverse events. ETO has been in use for some time but was retired due to its adverse side effects in the liver. Currently, it is being repurposed as an anticancer agent (34). Unfortunately, there are no data on the pharmacokinetics of ETO, so we do not know the clinical applicability of ETO. We found a relatively high IC 50 against N. fowleri of approximately 100 to 150 M, but the synergy shown with MIL deserves further investigation in an animal model.
Of all FAO inhibitors tested, TDZ showed the lowest IC 50 (approximately 6 to 10 M) against N. fowleri in our study. TDZ has been in use as an antipsychotic drug since the early 1950s and was originally identified as a dopamine receptor 2 antagonist. Later, TDZ was also shown to be a selective inhibitor of peroxisomal ␤-oxidation (25,26). It is now being repurposed as an anticancer, anti-inflammatory, and antimicrobial agent (35)(36)(37)(38). The pharmacokinetics of TDZ are well studied. In a recent clinical study, the sum of TDZ and its metabolites in serum approached 10 M (37). Furthermore, TDZ has been shown to accumulate in brain tissue of chronically treated patients, resulting in concentrations 10-fold higher than that in serum (39). Although AMB is effective at nanomolar concentrations, AMB and MIL are known to have poor CNS penetration (11)(12)(13)(14). This could possibly explain the poor prospects for treatment of patients with PAM and emphasizes the possible benefit of TDZ.
Development and testing of new drugs for PAM are difficult, as randomized controlled trials for the treatment of PAM are impossible due to the rapidly fatal nature of the disease and its relatively rare occurrence. Repurposing existing drugs is therefore the most promising way to obtain additional drugs to combat PAM. There are numerous examples of drugs that have been successfully repurposed to treat rare diseases (40). All tested FAO inhibitors (including TDZ and ETO) have been or are still in clinical use. TDZ inhibits N. fowleri growth at concentrations that can be reached at the site of infection, and checkerboard assays revealed synergy between MIL and ETO. Further animal testing should be performed to confirm the added value of these inhibitors.
Inhibition assays for N. gruberi. To screen the effects of fatty acid oxidation inhibitors and current therapies for PAM on N. gruberi, we determined compound efficacy with optical density (OD) measurements. Drugs were prepared as stock solutions as follows: TDZ, 10 mM in water; PHX, 50 mM in dimethyl sulfoxide (DMSO); ETO, 10 mM in water; VPA, 20 mM in water; ORL, 35 mM in DMSO; AMB, 10 mM in DMSO; MIL, 10 mM in water. Stock solutions were diluted in water or PYNFH medium and added as a volume of 10 l to 1 ϫ 10 4 N. gruberi trophozoites in 90 l of PYNFH. Compounds were tested per plate in triplicate in at least two independent experiments; controls contained equivalent concentrations of compound solvents (water, PYNFH or DMSO). Plates were incubated at 25°C, and OD measurements of the contents of the 96 wells were performed every 24 h using a FLUOstar Optima microplate reader. Regrowth capacity was assessed by collecting the whole contents of the wells at day 5 by vigorous pipetting, after which the content (100 l) was added to Eppendorf tubes containing 1 ml PYNFH medium. Samples were washed by centrifugation at 1,000 relative centrifugal force (rcf) and subsequent careful replacement of the supernatant with fresh PYNFH medium. After this washing cycle was repeated three times, 1 ml supernatant was discarded and the remaining contents were added to a new plate. Controls were diluted 10ϫ after the washing step to allow proper detection of regrowth in these samples. OD was measured for a further period of 9 days of incubation at 25°C.
Inhibition assays for N. fowleri. For N. fowleri, two independent methods were used to determine compound efficacy: a CellTiter-GLO luminescence-based ATP stain method (less sensitive to the number of amoeba but more sensitive to viability) and cell counting by flow cytometry (sensitive to the number of amoeba but with count determined irrespective of viability). Similar stock solutions were prepared for N. fowleri experiments as those used for N. gruberi experiments. For CellTiter-GLO experiments, stock solutions were diluted and added to a black 96-well plate as 10 l compound-80 l Bacto Casitone, after which 300 N. fowleri cells in 10 l Bacto Casitone were added to each well for a total volume of 100 l. After 24 h of incubation at 37°C, CellTiter-GLO reagent was added and luminescence was determined by the use of a CLARIOstar microplate reader. For flow cytometry experiments, stock solutions were diluted and added to translucent 96-well plate wells as 20 l compound-178 l Bacto Casitone, after which 60 N. fowleri cells in 2 l Bacto Casitone were added to each well for a total volume of 200 l. After 72 h of incubation at 37°C, paraformaldehyde was added to obtain a 1.5% concentration, after which cell counting was performed by the use of a Guava EasyCyte flow cytometer. Appropriate gating was applied to all samples. All experiments were performed in triplicate; all plates contained positive controls in triplicate wells with equivalent concentrations of compound solvents (water, Bacto Casitone, or DMSO) and negative controls without amoebae.
Checkerboards assay. Twofold dilutions were prepared of MIL (100 to 3.13 M), ETO (400 to 25 M), PHX (25 to 1.61 M), TDZ (25 to 0.8 M), and VPA (2,000 to 62.5 M). Drugs were added to black 96-well plates in a 5-by-6 checkerboard design as described before (41). The 96-well plates were inoculated with 300 N. fowleri cells in Bacto Casitone per well. After 24 h of incubation at 37°C, CellTiter-GLO reagent was added and luminescence was determined by the use of a CLARIOstar microplate reader. All experiments were performed in triplicate; all plates contained positive controls in triplicate wells with equivalent concentrations of compound solvents (water, Bacto Casitone, or DMSO) and negative controls without amoebae. Data analysis. GraphPad Prism 8 was used to process data. For N. gruberi, graphs of separate wells were constructed with OD values represented on the y axis and time (in days) on the x axis. Area under the curve (AUC) was then calculated by the use of GraphPad Prism 8, and the calculated data were combined into a bar chart. To determine IC 50 values, results were normalized and nonlinear regression with variable slope was performed. For N. fowleri, luminescence data and cell counts were normalized as a percentage of the respective control, after which nonlinear regression with variable slope was performed to determine IC 50 values. To determine synergy, analysis was performed using the fractional inhibitory concentration index (FIC i ), as described before (41). Briefly, FIC i values were determined for the wells with the lowest concentration of drugs that resulted in Ͻ10% luminescence of the growth control without drugs. FIC i was calculated as ⌺FIC ϭ C a MIC a ϩ C b MIC b , with C a and C b being the concentrations of the drugs in the well and MIC a and MIC b the lowest concentrations of separate drugs that resulted in Ͻ10% of luminescence. F min was defined as the lowest ΑFIC i value. Additivity was determined as 0.5 Ͻ ΑFIC i Ͻ 2, synergy as ΑFIC i ϭ Յ0.5, and antagonism as ΑFIC i ϭ Ͼ2. Surface response analysis was performed with the Combenefit program (42).