Characterization of two early lead anti-amoebic agents with activity against Naegleria fowleri

Abstract

Brain infection by the pathogenic free-living amoebae Naegleria fowleri is a life-threatening illness, with a mortality rate >95% in the US. One of the challenges faced in treating these infections is the lack of efficacious drugs. Here, we further characterize two leads. The first, a N. fowleri enolase inhibitor (HEX ((1-hydroxy-2-oxopiperidin-3-yl) phosphonic acid)), emerged from a target-based approach while the other, BDGR-20191 (rac-(4aR,13bS)-10,11-dichloro-4-(oxetan-3-ylmethyl)-1,2,3,4,4a,5,6,13b-octahydro-8H-[1,6]naphthyridino[5,6-b]quinazolin-8-one), was identified in a phenotypic screen for anti-amoebic agents. Chronic low dose exposure to these agents led to cell lines that were tolerant to each. However, the tolerance phenotypes were reversed in the absence of selection with the agent. Differential gene expression analyses suggest that HEX tolerant cells upregulate cellular components that could impact levels of accumulated glycolytic intermediates, while BDGR-20191 tolerant cells could be targeting membrane biosynthesis. Unbiased metabolomics of the lines supported these possibilities. While HEX and BDGR-20191 are not synergistic with one another or with other standard-of-care drugs, their additive nature suggests they could be a useful addition to current therapeutic cocktails.

Graphical abstract

Highlights

• •Resistant Naegleria fowleri were generated to two promising anti-amoebic leads.
• •Stable resistance was not observed to either HEX or BDGR-20191.
• •RNAseq and metabolomics revealed alterations that could impact glycolytic intermediates and membrane biosynthesis.
• •Synergy analyses suggest the leads were additive with currently used drugs.

1. Introduction

Naegleria fowleri is a free-living ameboflagellate that can cause a serious infection in humans called primary amoebic meningoencephalitis (PAM). The amoeba is typically found in freshwater environments including lakes, ponds, and rivers but can survive in poorly chlorinated pools, splash pads, and other municipal water sources. If inadvertently introduced into the nasal passages, the amoeba can migrate along the olfactory nerves, breach the cribriform plate, and infect the central nervous system (CNS) to cause disease.

The typical sequelae of early infection describe many possible infections, with headache and fever dominating the symptoms. As the disease progresses, neurological symptoms including confusion and seizures can occur (Kofman and Guarner, 2022). Even with the onset of these more severe symptoms, diagnosis of PAM remains challenging. Current clinical approaches used to recognize infections typically involve either microscopic visualization or PCR-based detection of amoebae in the cerebral spinal fluid (CSF). In either case, the test requires collection of CSF, but this invasive sampling is generally deferred until other etiological agents can be ruled out, critically delaying diagnosis (Yoder et al., 2010).

In the event of a diagnosis, infection is typically treated with a combination of antifungal, antiparasitic, and antimicrobial agents. These combinations usually include amphotericin B and miltefosine, with fluconazole, rifampin, and azithromycin often included as well. To attempt to limit damage caused by immune inflammation, dexamethasone and therapeutic hypothermia may be included in the treatment plan. Despite these efforts, infection outcomes remain poor, with fatality rates >95%.

To identify new and efficacious agents, we and others have used target-based and phenotypic screens to identify new anti-amoebic agents. Target-based drug discovery (also called rationale drug discovery) campaigns are designed to identify and optimize inhibitors of critical cellular targets. For example, homologs to DrugBank protein targets have been considered (Tillery et al., 2021). Multiple phenotypic screens, which have scored the impact of a molecule on pathogen viability without a priori knowledge about mode of action, have been completed. These typically differ by the makeup and number of small molecules included in the screen. Screens of an FDA-approved drug collection and other very large collections have yielded promising leads (Debnath et al., 2012; Colon et al., 2019; Rice et al., 2020).

Here, we describe the further characterization of a target-based early lead and a compound discovered in a phenotypic screen for anti-amoebic activity. The target-based lead, (1-hydroxy-2-oxopiperidin-3-yl) phosphonic acid (HEX), is an inhibitor of N. fowleri enolase (NfENO) and is a potent anti-amoebic compound (EC₅₀ = 0.21 ± 0.02 μM) that is well tolerated in mammals (Lin et al., 2020; Milanes et al., 2024). Further, intranasal delivery of this agent was found to significantly extend the life of treated rodents (Milanes et al., 2024). The compound identified in a phenotypic screen, rac-(4aR,13bS)-10,11-dichloro-4-(oxetan-3-ylmethyl)-1,2,3,4,4a,5,6,13b-octahydro-8H-[1,6]naphthyridino[5,6-b]quinazolin-8-one (BDGR-20191), is representative of a group of anti-amoebic quinazolinones (Lish et al., 2024) that were very potent against the pathogen in vitro. BDGR-20191 was active at single-digit nanomolar concentrations (EC₅₀ = 5.6 ± 0.74 nM) and did not have detectable toxicity against human neuronal SH-SY5Y cells (CC₅₀ value > 20 μM).

HEX likely impacts amoeba glucose metabolism, a supposition further supported by accumulation of glycolytic intermediates upstream of NfENO in treated cells (Milanes et al., 2024). However, it is unclear how the charged phosphonate of HEX affects uptake by the cell to engage the target. In the case of BDGR-20191, the mode of action of the compound remains unresolved. To further dissect these issues, and understand possible amoebic responses to treatment, we have generated cell lines that are tolerant to each compound but have not yet identified cells with stably resistant phenotypes to either agent. Gene expression analyses of the lines suggest that ∼35% of the genes with altered expression were shared between cells tolerant to either compound. These likely reflect general stress responses. Combination studies using the HEX and BDGR-20191 together or with other drugs currently part of the standard of care indicated that the combinations had an additive impact and could be used to limit cells that persist.

2. Materials and methods

2.1. Amoebae culturing

N. fowleri TY strain (NfTY, ATCC 30170), originally isolated from a patient in the United States (Virginia) in 1969, were generously provided by Dr. Dennis Kyle (University of Georgia). Trophozoites were cultured axenically at 37°C in Nelson's Complete Medium (NCM; 0.17% liver infusion broth, 0.17% glucose, 0.012% sodium chloride, 0.0136% potassium phosphate monobasic, 0.0142% sodium phosphate dibasic, 0.0004% calcium chloride, 0.0002% magnesium sulfate, 10% heat-inactivated fetal bovine serum, 1% penicillin-streptomycin) in treated tissue culture flasks. For cell passage, cells were placed on ice for 20-30 min, centrifuged (3000 x g, 4°C, 5 min), resuspended in 1 mL NCM, and passed into fresh media.

2.2. Development of drug-tolerant N. fowleri

NfTY trophozoites were grown to confluency and then passed into fresh NCM containing either HEX at 400 nM, about twice the EC₅₀ value concentration (Milanes et al., 2024), or BDGR-20191 at 6 nM, about the EC₅₀ value concentration for the agent. As the cells reached confluency, they were passed into fresh NCM with increased concentrations of compound in four replicates. For HEX, this involved stepwise passages into media with five- and ten-fold increments of the EC₅₀ value concentration. The stepwise passages used to generate tolerance to BDGR-20191 included increments of two-, five-, and ten-fold the EC₅₀ value concentration of the agent.

2.3. Library preparation and sequencing

Upon reaching confluency in media containing ten times the EC₅₀ value concentration of the parental line, 10⁶ cells from each replicate were collected by centrifugation (3000 x g, 4°C, 5 min) and RNA was extracted using Total RNA Mini Kit (IBI Scientific IB47320). RNA sample integrity was evaluated using a TapeStation 4200 (Agilent, Santa Clara, CA) and RNA ScreenTape to generate an eRIN, and all samples were >8. Total RNA (1 μg per sample) was used to construct an mRNA library using the Illumina Stranded RNA Prep kit with Poly-A selection (Illumina, San Diego, CA) according to manufacturer recommendations. Library quality was evaluated with a TapeStation 4200 and High Sensitivity D1000 ScreenTape (Agilent). Library quantity was evaluated with qPCR using the NEBNext Library Quant Kit for Illumina (NEB, Ipswich, MA). Multiplexed libraries were pooled and sequenced on an Illumina NextSeq 2000 to a read length of 2 x 76 and an average depth of 43,301,441 read pairs per sample.

2.4. RNA analysis

The experimental design included two treated and one control group with four biological replicates each. Quality metrics or raw and trimmed data were analyzed using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Trimming of low-quality bases and adapter sequences was performed using TrimGalore (https://github.com/FelixKrueger/TrimGalore). Trimmed reads were aligned using gsnap v2023-12-01 to the AmoebaDB-68_NfowleriTy genome. Subread's featureCounts v2.0.1 counted by transcript including only uniquely mapped, concordantly aligned, reversely stranded paired end reads (Liao et al., 2014). Raw gene counts (Nf.Ty.samples.gene.counts.renamed.txt) were then used as input to edgeR for differential gene expression (DGE) analysis using genewise negative binomial generalized linear models (Robinson et al., 2010). Genes with low coverage across all samples were filtered out, and library sizes were normalized using the trimmed mean of M-values method. Sample comparisons were set up as quasi-likelihood ratio tests, and genes having a false discovery rate (FDR) of <0.05, |log2FC| ≥ 1.0 were considered to have significant expression abundance changes. (See Supplemental Tables 1 and 2 for differentially expressed genes (DEGs) in HEX or BDGR-20191 tolerant cells, respectively.)

2.5. Gene Ontology and pathway analyses

The R package topGo and Gene Ontology (GO) terms, downloaded from AmoebaDB for the N. fowleri TY genome, were used to identify GO and Pathway enrichment in the DGE analyses. The classic algorithm and fisher's test were used. Enriched GO terms were investigated using Quick GO (https://www.ebi.ac.uk/QuickGO).

2.6. Analysis of metabolites from HEX or BDGR-20191 tolerant lines

For unbiased metabolomics analyses, N. fowleri drug-tolerant trophozoites were grown to near 100% confluency in 6-well TC-treated plates (Thermo Scientific Nunclon Delta Surface 140685) and washed three times in PBS prior to the addition of ice-cold CH₃CN. After 20 min at −20°C, ice-cold H₂O was added and the resulting mixture transferred to sterile microcentrifuge tubes. This process was repeated twice, and the total volume of extraction material was adjusted to 2.5 x 10⁶ cell equivalents per sample.

Each cell sample was homogenized with glass beads using a mixer mill (MM200 Mix Grinder, Retsch GmbH, Haan, North Rhine-Westphalia, Germany) for 5 min at 30 Hz. The homogenate was centrifuged at 14,000 rpm for 20 min at 4°C. The supernatant was collected and lyophilized overnight. The dried extracts were then reconstituted in 50% acetonitrile for parallel two-dimensional liquid chromatography-mass spectrometry (2DLC–MS) analysis. The parallel 2DLC-MS system was configured using a Thermo Q Exactive HF Hybrid Quadrupole-Orbitrap Mass Spectrometer coupled with a Thermo DIONEX UltiMate 3000 HPLC system (Thermo Fisher Scientific, Waltham, MA, USA). The parallel 2DLC system was equipped with a reversed phase chromatography column (RPC, a Waters Acquity UPLC HSS T3 column, 2.1 × 150 mm, 1.8 μm) and a hydrophilic interaction chromatography column (HILIC, a Millipore SeQuant ZIC-cHILIC column, 2.1 × 150 mm, 3 μm). Each LC column was connected to a 2-μL sample loop, and the column temperature was set to 40°C. The gradient methods of the parallel 2DLC have been reported before (Xu et al., 2023). Each sample was analyzed in random order by 2DLC-MS in both positive (+) and negative (−) modes to obtain full MS data for metabolite quantification. For metabolite identification, pooled samples, a mixture of small amounts of every sample in the same group, were analyzed by 2DLC-MS/MS in positive and negative modes to acquire MS/MS spectra at three collision energies (20, 40, and 60 eV). Metabolite identification was first performed by matching the retention time and MS/MS spectrum with the corresponding information in our in-house database. The unmatched MS/MS spectra were subsequently searched against the Compound Discovery software (version 3.5, Thermo Fisher Scientific, Inc., Dreieich, Germany). The identification criteria were set as follows: m/z variation ≤5 ppm, retention time variation ≤0.15 min, and spectrum similarity score ≥0.4. Metabolite relative quantification was achieved using the method described in our previous publication (He et al., 2024).

2.7. In vitro synergy assay

NfTY trophozoites were seeded at 10⁴ cells/mL in NCM in white tissue culture-treated 96-well plates. The concentration ranges used were 0 – 200 ng/mL (Amphotericin B), 0 – 2.5 μM (HEX), and 0 – 40 nM (BDGR-20191). These concentrations were based on previously solved EC₅₀ concentration values (Lish et al., 2024; Milanes et al., 2024). Cells were cultured in the presence of both drugs for 48 h (37°C, 5% CO₂), followed by incubation of the plates at RT (10 min), and addition of CellTiter Glo reagent (Promega G7570). Plates were mixed by orbital shaking (2 min, RT) and incubated 10 min at RT. Luminescence was scored on a BioTek Synergy H1 microplate reader and further analyzed by SynergyFinder Plus (Zheng et al., 2022) to generate heat plots, dose-response curves, and synergy maps of drug combinations.

2.8. Synthesis of BDGR-20191

The synthesis and characterization of HEX and BDGR-20191 have been previously described (Lin et al., 2020; Lish et al., 2024; Milanes et al., 2024). The compounds were synthesized and provided by the laboratories of D. Whitehead and J. E. Golden at Clemson University and the University of Wisconsin-Madison as a powder sample, respectively. The compounds were analyzed by ¹H and ¹³C NMR for structural integrity and UV/HPLC/MS for purity analysis and shown to be >95% pure.

3. Results

3.1. Drug tolerant cells have reduced growth rate

N. fowleri trophozoites continuously cultured in the presence of sub-lethal doses of HEX or BDGR-20191 resulted in parasites with increased tolerance to the agents, although their growth rate slowed. In media supplemented with toxic levels of HEX or BDGR-20191 (2.1 μM or 70 nM, about ten times the EC₅₀ values for HEX or BDGR-20191, respectively), viable parental cells were undetectable. The tolerant cells continued to grow, albeit slowly, under these conditions. The HEX and BDGR-20191 tolerant amoebae had doubling times of ∼18.9 h and ∼24.3 h, respectively (Fig. 1). Parental cells in normal media have a doubling time of ∼8.6 h. The tolerant nature of the cells was not due to a permanent change in the amoebae. After one week of culture in the absence of agent, the cell growth rate returned to near normal (Fig. 1), and EC₅₀ values for the cells were indistinguishable from the untreated controls. (Please note, we were unable to score the tolerant cell line EC₅₀ values when they were grown under drug pressure, as the slow growth rate impacted our viability assay protocol.)

Fig. 1.

Agent tolerant N. fowleri cells have an impaired growth phenotype in the presence of compound. Growth analysis of HEX (A) and BDGR-20191 (B) tolerant lines compared to WT cells. Cells were seeded at 1 x 10⁴/mL and densities were monitored. Amoebae doubling times were 8.6 h for NfTY (•), 19 h for HEX tolerant NfTY in the presence of HEX (■), 7.8 h for HEX tolerant NfTY in the absence of HEX, 24 h for BDGR-20191 tolerant NfTY in the presence of BDGR-20191 and 8.9 h for BDGR-20191 tolerant NfTY in the absence of BDGR-20191.

3.2. Significant gene expression changes in the HEX and BDGR-20191 tolerant lines differ, suggesting potential different cellular impacts of the agents

RNA analysis of the drug tolerant cell lines indicates that there were 1731 and 2004 significantly differentially expressed genes (DEGs) in the HEX and BDGR-20191 tolerant lines, respectively. Some of the DEGs were shared between the two cell lines, suggesting these resulted from general stress responses (Fig. 2; Supplemental Tables 1–4). (Please note: all significant gene expression changes are defined by an FDR <0.05 and a −1 ≥ Log2FC ≥ 1.) The HEX and BDGR-20191 tolerant lines shared 973 significantly DEGs compared to WT, accounting for ∼35% of all DEGs (Fig. 2, Supplemental Tables 3 and 4). Of the 973 common DEGs, 453 shared upregulation, while 488 were downregulated in both. Notably, 32 genes were significantly differentially expressed in different directions between the samples. Most of the upregulated genes that were shared between the compound-treated lines are involved in tRNA metabolic processes (GO:0006399); how these genes are involved in tolerance remains unclear. The top upregulated gene, based on Log2FC, shared between both datasets is a BTB/POZ domain containing protein (NfTy_025200), with fold change values of 5.5 for the HEX tolerant line and 2.9 for the BDGR tolerant line. Most of the shared downregulated genes share the same GO term (GO:0055086), being involved in nucleobase-containing small molecule metabolic processes. These tend to be energy-intensive pathways, requiring large amounts of ATP and reducing equivalents. It is not surprising that HEX inhibition of enolase leads to suppression of energy demanding anabolic pathways. BDGR-20191 may similarly impact the cell based on these impacted pathways.

Fig. 2.

Venn diagram comparing DEGs from HEX and BDGR tolerant cell lines. The overlap includes shared differentially expressed genes. Expression changes in the tolerant cell lines were validated by RT-qPCR on a subset of upregulated and downregulated genes in each condition (Fig. S12).

The HEX tolerant lines had a total of 963 and 768 significantly upregulated and downregulated genes, respectively (Fig. S1). Of these, there were 758 genes unique to the HEX tolerant lines that were significantly differentially expressed in comparison to the untreated line. In this group of genes unique to HEX tolerant amoebae, 479 were upregulated and 279 downregulated. The ten most significantly upregulated unique genes based on Log2FC were a leucine rich repeat family protein (NfTy_058810), a UbiA prenyltransferase (NfTy_002410), a calcium binding protein (NfTy_011750), a zinc ion binding protein (NfTy_057780), and six genes annotated as unspecified products (Figs. S2 and S3). Upregulated genes included, based on GO term enrichment, those involved in nucleic acid metabolic process, RNA metabolic process, and nucleic acid biosynthetic process, while downregulated genes were involved in phosphorus metabolic process and intracellular signal transduction (Fig. 3, Fig. S2 and S3). HEX inhibition of enolase is known to trigger accumulation of glycolytic intermediates upstream of enolase. Through gluconeogenesis these intermediates (through glucose-6-phosphate) could fuel nucleic acid synthesis in the pentose phosphate pathway, acting as a sink for potentially toxic glycolytic intermediates. This may explain the observed upregulation of nucleic acid biosynthesis pathways in the HEX tolerant amoebae. Among the upregulated genes was triphosphate isomerase, which is upstream of enolase, and could be acting to convert accumulated intermediates in gluconeogenesis.

Fig. 3.

Gene enrichment analysis of differentially expressed genes in HEX and BDGR tolerant cell lines. The frequency of upregulated (left panel) or downregulated (right panel) GO terms in HEX and BDGR-20191 tolerant cell lines is also indicated.

HEX is a phosphonate-based enolase inhibitor. We have speculated that amoebae efficiently acquire the inhibitor through the action of a putative phosphonate transporter (Milanes et al., 2024). Notably, the expression of this putative phosphonate transporter gene (NfTy_081940) was not significantly altered, suggesting that any role it has in uptake may be either minimal or the cell cannot tolerate changes in its expression. There were, however, six ABC transporters that were significantly upregulated and had a fold change values greater than one. These include NfTy_041620, NfTy_036610, NfTy_013120, NfTy_082260, NfTy_012080, and NfTy_002740.

The cells that persisted in the presence of BDGR-20191 had significant expression changes in 1031 unique genes, with 447 upregulated and 584 downregulated. GO term analysis of all significantly differentially expressed genes in the BDGR-tolerant lines indicated that of the 899 upregulated genes, 439 were associated with biosynthetic and cellular biosynthetic processes (Fig. 3, Fig. S2 and S4). Among the most upregulated genes, based on Log2FC, from the BDGR-20191 treated cells were HMG-CoA synthase (NfTy_038620), a ras homolog (NfTy_076940), and sterol 24-c-methyltransferase (NfTy_017510). These suggest that cholesterol biosynthesis could be connected to tolerance of the agent. Of the 1105 downregulated genes, 338 were described by the phosphorus metabolic processes and phosphate-containing compound metabolic processes GO terms (Fig. 3, Fig. S2 and S4). Specifically, the most significant included an ATP-binding cassette protein (NfTy_086550) and a crinkler (CRN) family protein (NfTy_090810) with fold changes of −6.3 and −5.6, respectively.

3.3. Metabolomic analyses points to distinct mechanisms of tolerance to HEX and BDGR20191

When compared to untreated amoebae, HEX- and BDGR-20191-tolerant cells had significant changes in 43 and 125 metabolites, respectively. For the HEX-tolerant cells, there were 11 metabolites that were significantly more abundant and 32 that were reduced compared to normal cells. The BDGR-20191 tolerant line had 37 significantly more abundant and 88 significantly less abundant metabolites compared to the control amoebae (Fig. 4, Supplemental Tables 5 and 6 for a complete list of metabolites altered in either HEX- or BDGR-20191-tolerant cells).

Fig. 4.

Unbiased metabolomic analysis of compound-tolerant amoebae compared to untreated controls. (A) Differential metabolite volcano plot for the HEX and BDGR tolerant cell lines compared to parent (WT) amoebae. Downregulated and upregulated metabolites are in purple and orange, respectively, while those in grey are unchanged. (B) Comparative metabolite intensities from amoebae tolerant to HEX and BDGR for fructose-6-phosphate/glucose-6-phosphate (F6P/G6P), pyruvate, phosphatidylinositol O-32:2 (PI (O-32:2)), and methionine. Statistical analysis was performed using pairwise two-tail tests. ∗∗∗p value < 0.001, ∗∗p value < 0.01 ∗ p value < 0.05. Error bars represent SEM.

Many of the same metabolites were altered in both tolerant cell lines. Eight were commonly more abundant, while 30 were found in both groups to be reduced. Metabolites that were overrepresented compared to the parental line in both include toxins and potential antifungals (2-deoxyglucose, aurantiamine, and undecylenic acid), suggesting that the general stress response includes upregulation of machinery to generate these molecules. The compounds that were reduced in both included amino acids and nucleic acid biosynthetic components, though it is unclear if these are reduced due to changes in metabolism or reduced biosynthesis.

The metabolites that were uniquely changed in the two tolerant cell lines suggest that different mechanisms were at work to prevent the lethal effect of HEX and BDGR-20191. Compounds more abundant in the HEX-tolerant lines included F6P/G6P and seduheptulose 7-phosphate (SH7P) (Fig. 4B). These may have been more abundant because of increased funneling of potentially toxic glycolytic intermediates that result from ENO inhibition into the pentose phosphate pathway (Milanes et al., 2024). Pyruvate levels were also increased, possibly reflecting increased production through the glutaminolytic pathway (many of its components were reduced in the HEX lines) or other amino acid metabolism (Fig. 4B). Additionally, it could be generated from malate, as the amoeba genome encodes a malic enzyme (NfTy_045880) that could convert malate to pyruvate. Glutathione and lysophosphatidylcholice (18:2) were the only metabolites uniquely down in the HEX lines. While the cause of these differences is unresolved, glutathione biosynthesis requires glutamine, which is reduced in the HEX-tolerant lines. The BDGR-20191-tolerant lines had more uniquely altered metabolites than the HEX lines. Increased metabolites include fatty acid and lipid components, while compounds that were uniquely down included many possible carbon sources, including amino acids and nucleic acid intermediates (Fig. 4). This could reflect a globally suppressed metabolic state and a decreased demand for building blocks, as the BDGR-20191 cells were noted to grow very slowly (Fig. 1).

3.4. Exploring drug synergy with HEX

HEX is a potent killer of N. fowleri in vitro and has demonstrated promising life-extending potential in a rodent model of infection (Milanes et al., 2024). Given that current therapeutic approaches are not monotherapies, we assessed the impact of HEX in combination with one of the current components of the standard of care, amphotericin B. Analysis of combinations of HEX or BDGR-20191 with other compounds using four synergy reference models (HSA, Bliss independence, Loewe additivity, and ZIP) revealed that HEX was additive with all the other treatments with synergy scores of −10 to 10 (Table 1, Fig. 5, Figs. S5–S11). Additionally, combinations using HEX and BDGR-20191 were additive in their effect.

Table 1.

HEX activity in combination with front-line treatment agents and other metabolic inhibitors.

Compound	N. fowleri TY EC50 Value (μM) (Published value)	Impact with HEX? (Score)a	N. fowleri EC50 value + HEX (μM)b	HEK-293 CC50 value + HEX (μM)c	S.I. of compound + HEXd
HEX	0.87 ± 0.23 (0.21 ± 0.02d (Milanes et al., 2024))	-	-	-	-
Amphotericin B	0.02 ± 0.01 (0.06 ± 0.00d (Escrig et al., 2020))	Additive (2.84)	0.02 ± 0.01	>10	∼500
Miltefosine	36 ± 9.1 (38 ± 0.02d (Escrig et al., 2020))	Additive (1.72)	44 ± 10	>10	ND
Atovaquone	14 ± 2.2 (NA)	Additive (5.82)	13 ± 6.3	>50	∼3.8
BDGR-20191	0.01 ± 0.00 (0.01 ± 0.00d (Lish et al., 2024))	Additive (6.73)	0.01 ± 0.00	>2.0	∼200
ELQ-300	11 ± 3.4 (NA)	Additive (2.36)	2.3 ± 0.53	>10	∼4.3
Nitroxoline	0.85 ± 0.22 (1.2 ± 0.21d (Chao-Pellicer et al., 2023))	Additive (1.36)	1.9 ± 0.58	>12.5	∼6.6
Posaconazole	0.04 ± 0.03 (0.17 ± 0.05d (Russell and Kyle, 2022))	Additive (3.99)	0.01 ± 0.00	>0.7	∼70

^dSI = EC₅₀/CC₅₀ values. Please note, these are conservative estimates based on the CC₅₀ values equaling the greatest concentration of agent + HEX tested. Therefore, the SI values may be greater.

^dPublished strain tested was Nf69.

^aSynergy scores listed are based on the HSA synergy calculation; >10 would suggest synergy.

^bHEX was held constant at 0.04 μM.

^cValues represent the highest achievable drug concentrations administered under conditions that maintain the minimal DMSO concentration (<0.5% v/v) required to avoid adverse effects on cell viability. HEX was held constant at 0.04 μM. Please note that HEX alone at up to 10 μM has no impact on HEK-293 growth.

^dPublished strain tested was TY.

Fig. 5.

Synergy Analysis by HSA model. Synergy map highlighting synergistic and antagonistic dose regions in red and green colors, respectively. The HSA model states that the expected combination effect is the maximum of the single drug response at the corresponding concentration. (Top panel) The HEX and BDGR-20191 combination has a HSA synergy score of 6.73 (p = 2.15 x 10⁻¹⁵). (Bottom panel) HEX and Amphotericin B combination has an HSA synergy score of 2.84 (p = 1.72 x 10⁻¹). Synergy scores from −10 to 10 suggest the interaction between the two drugs is likely to be additive. All analyses were performed using SynergyFinder Plus.

4. Discussion

In the United States, N. fowleri infections are typically treated with cocktails of antibiotics, antifungals, and antineoplastic agents. The components of these multidrug combinations typically have in vitro anti-amoebic activity, with some (posaconazole, for example) being potent anti-amoebic agents. In our hands, posaconazole has an EC₅₀ value of 0.04 ± 0.03 μM, and others have reported similar values (Hargrove et al., 2021; Russell and Kyle, 2022). Some of the other components have more modest activity against the amoeba, with in vitro EC₅₀ values in the tens of micromolar range. For example, miltefosine has reported EC₅₀ values ranging from ∼20 to 100 μM, depending on strain of N. fowleri tested (Russell and Kyle, 2022). While there is some strain variation in sensitivity to agents, the drugs included typically have some historical and empirical support for their inclusion in the cocktail, despite only a handful of infection survivors.

As a first step to better understand the bioactivity of two new leads against the amoebae, we developed cells lines that were not eliminated by lethal doses of the agents. In both cases, we were able through time to develop cell lines that were refractory to elimination by doses of up to ten-fold the EC₅₀ values of the parental lines for HEX and BDGR-20191. Our preliminary assessment suggests that neither cell can establish an infection in a rodent model.

In vitro, the HEX and BDGR-20191 tolerant cells reverted to parental level of sensitivity if drug pressure was removed for one week. This rapid return of sensitivity suggests that heritable mutations, a classic hallmark of resistance, were not the basis for the refractory nature. Instead, the phenotype is better explained by tolerance to HEX and BDGR-20191. Cells that are tolerant are not eliminated following a treatment period that is sufficient to kill susceptible strains. This increase in “minimum duration for killing” (MDK) usually does not require acquisition of mutations associated with resistance. An increase in the MDK can also describe persistent cells; however, the persistence phenotype is usually found in only a few members of a population whereas tolerance can be a more general phenomenon. Here, tolerance is a more suitable classification, as we did not detect appreciable population reduction in the generation of the cell line (Brauner et al., 2016).

The HEX and BDGR-20191 tolerant cell lines significantly altered expression of ∼14% and ∼17% of their predicted gene complement (based on the amoebae having ∼12,000 protein coding genes). This level of differentially expressed genes is similar to resistant cell lines from different organisms. In the American trypanosome, ∼9% of the predicted genome is differentially expressed in cells resistant to benznidazole (Lima et al., 2023), while T-cell lymphoma specimens that are resistant to chemotherapy alter expression of ∼20% of their gene complement (Lima et al., 2023).

The HEX tolerant cells notably did not upregulate NfENO, the proposed target of the agent (Milanes et al., 2024). NfENO is central to both glycolysis and gluconeogenesis, interconverting 3-phosphoglycerate and phosphoenolpyruvate, key metabolic processes that likely generate essential products. We hypothesize that the relatively slow adaptation process used to generate the tolerant cells provided sufficient time for the amoebae to modulate their metabolism to increase lipid consumption, a capacity they are known to have (Herman et al., 2021).

HEX-treated cells are known to accumulate glycolytic intermediates upstream of NfENO, including 3-phosphoglycerate, 2-phosphoglycerate, fructose-6-phosphate (F6P), and glucose-6-phosphate (G6P). The management of these intermediates could play a key role in the tolerance to HEX. The upregulation of triphosphate isomerase, the enzyme responsible for the interconversion of fructose-6-phosphate and G6P could drive the accumulated substrates towards G6P, which has many possible fates. Indeed, metabolomics analysis indicated that F6P/G6P levels were increased. From there, it could enter the pentose phosphate pathway to generate substrates required to feed upregulated nucleic acid biosynthetic processes, hinted at by the increase in S7P. Ultimately, funneling G6P into various biosynthetic paths could serve as a sink for the HEX-induced increase in intermediates, reducing the consequences of their accumulation.

The mechanism of action of BDGR-20191 remains unresolved but alteration of genes potentially associated with oxidative stress that could limit damage to proteins and membranes may play a role. Notably, glutathione levels were reduced, and lipid and membrane components were increased in the BDGR-20191 tolerant cells, perhaps reflecting attempts to address the toxicity of the compound. We anticipate that future efforts, including thermal proteome profiling studies and pulldowns using the agent modified with tags as bait, will guide our understanding of its modes of action and targets. Nevertheless, the transient nature of the tolerance and potency of this class of agents (Lish et al., 2024) suggests that the target pathway(s) are critical to amoebae viability.

Since the current treatment for PAM involves a multi-drug cocktail, there is potential for synergistic interactions that could enhance overall therapeutic efficacy. Synergistic effects are ideal, as they occur when two agents produce a greater effect together than the sum of their individual effects. However, additive interactions can still be clinically valuable.

In vitro, HEX in combination with either BDGR-20191 or Amphotericin B produced an additive response. This indicates that the effect of the two drugs delivered together was equivalent to the sum of their individual effects, which is an expected and informative outcome.

Although additive behavior does not enhance potency beyond what each agent can achieve alone, it still holds translational promise. In a clinical or in vivo context, an additive relationship may allow for similar therapeutic efficacy at lower individual drug doses, potentially reducing toxicity or improving tolerability. Additive interactions also suggest that the compounds are targeting the amoebae through distinct mechanisms, aligning with our current understanding of their modes of action. This supports the rationale for maintaining them together as a combination therapy, as their independent contributions can collectively improve overall treatment effectiveness without introducing mechanistic redundancy. However, it is important to note that Amphotericin B has limited penetration across the blood-brain barrier, and this pharmacokinetic limitation may influence how in vitro behavior translates to effectiveness in an in vivo model of disease. As a result, future in vivo work will need to take into consideration the poor brain permeability of Amphotericin B and how that corresponds to therapeutic outcomes.

Successful management of N. fowleri infections remains a major challenge, a situation that reflects difficulties in diagnosis and treatment. First, infections are often not recognized early because of misdiagnosis for other, more common, maladies that share similar symptoms. Once recognized, infections are difficult to treat. The brain is an immune privileged compartment, and the blood-brain barrier interferes with delivery of many potential therapeutics. These factors, combined with the limited number of potent drug-like agents that have shown promising in vitro activity, have limited successful outcomes of infection. Here, we have further characterized two potent anti-amoebic lead compounds. While we noted that we could generate amoebae that were tolerant to both agents, this is likely not a concern if these were to be incorporated into a treatment protocol. First, the slow increase in tolerance required days of exposure, far longer than a typical infection lasts. Second, we anticipate that acute dosing at much higher levels will not be tolerated by the amoebae, particularly in combination with other agents known to impact the pathogen.

Note

Supplementary data is associated with this article.

Funding sources

This work was supported by National Institutes of Health awards R21 AI175463, R21 AI171217, R21 TR005502, and the NIH Center for Biomedical Excellence (COBRE) grant under award number P20GM146584 to JCM. Chemistry was supported through the National Institutes of Health (R21AI161232, J.E.G.) and instrumentation at the UW−Madison Analytical Instrumentation Center and the Medicinal Chemistry Center, both within the UW-Madison School of Pharmacy.

CRediT authorship contribution statement

Caroline M. Palmentiero: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Jillian E.M. McKeon: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. Colm P. Roster: Data curation, Investigation, Writing – review & editing. Samuel Kwain: Investigation, Writing – review & editing. Daniel C. Whitehead: Investigation, Writing – review & editing. Matthew S. Lish: Investigation, Writing – review & editing. Jennifer E. Golden: Investigation, Writing – review & editing. Rooksana E. Noorai: Data curation, Formal analysis, Writing – review & editing. Maslyn A. Greene: Investigation, Writing – original draft. Khairul I. Khan: Data curation, Formal analysis, Investigation, Methodology. Xinmin Yin: Data curation, Formal analysis, Investigation, Methodology. Xiang Zhang: Resources, Supervision. James C. Morris: Conceptualization, Funding acquisition, Resources, Supervision, Writing – review & editing.

Conflict of interest statement

The authors declare no financial or non-financial competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplemental Figure 1. Volcano plots of differentially expressed genes in the HEX- and BDGR-tolerant cell lines relative to control. Orange-significantly up regulated. Purple – significantly down regulated. Parameters include a FDR of <0.05 and Log2FC ≥ 1 and ≤ −1 for up- and down regulated genes, respectively. (A) Control vs HEX. (B) Control vs. BDGR-20191. Multimedia component 8

Supplemental Figure 2. GO Enrichment Plot of top 20 processes enriched in all DEGs from HEX (top panel) and BDGR-20191 (bottom panel) tolerant cell lines. Multimedia component 9

Supplemental Figure 3. GO Enrichment Plot of top 10 processes enriched in upregulated (top panel) and downregulated (bottom panel) genes from HEX-tolerant cell lines. Multimedia component 10

Supplemental Figure 4. GO Enrichment Plot of top 10 processes enriched in upregulated (top panel) and downregulated (bottom panel) genes from BDGR-20191 tolerant cell lines. Multimedia component 11

Supplemental Figure 5. Synergy Analysis by HSA model. 3D synergy maps highlighting synergistic and antagonistic dose regions in red and green colors, respectively. The HSA model states that the expected combination effect is the maximum of the single drug response at the corresponding concentration. (Top panel) HEX and BDGR-20191 combination has a HSA synergy score of 6.73 (p = 2.15e-15). (Bottom panel) HEX and Amphotericin B combination has a HSA synergy score of 2.84 (p = 1.72e-1). Synergy scores from −10 to 10 suggest the interaction between the two drugs is likely to be additive. All analyses were performed using SynergyFinder Plus. Multimedia component 12

Supplemental Figure 6. Synergy analysis of HEX and BDGR-20191 by Bliss independence reference model. 3D (top panel) and 2D (bottom panel) synergy maps. (A) Bliss independence model assumes a stochastic process in which two drugs elicit their effects independently. The Bliss synergy score of this combination is −2.03 (p = 1.30e-01). Synergy scores of −10 to 10 suggest two drugs are likely to be additive. Multimedia component 13

Supplemental Figure 7. Synergy analysis of HEX and BDGR-20191 by Loewe additivity reference model. 3D (top panel) and 2D (bottom panel) synergy maps. The Loewe additivity model calculates the expected response as if both drugs are the same. The Loewe synergy score was calculated to be 4.53 (p = 2.62e-07). Multimedia component 14

Supplemental Figure 8. Synergy analysis of HEX and BDGR-20191 by ZIP reference model. 3D (top panel) and 2D (bottom panel) synergy maps. The Zero interaction potency (ZIP) model captures the drug interaction relationships by comparing the change in the potency of the response curves between the drugs individually and their combinations. The ZIP synergy score for this combination is −1.68 (p = 2.10e-01). Multimedia component 15

Supplemental Figure 9. Synergy analysis of HEX and Amphotericin B by Bliss independence reference model. 3D (top panel) and 2D (bottom panel) synergy maps. The Bliss synergy score of this combination is 1.13 (p = 7.24e-01). Multimedia component 16

Supplemental Figure 10. Synergy analysis of HEX and Amphotericin B by Loewe additivity reference model. 3D (top panel) and 2D (bottom panel) synergy maps. The Loewe synergy score was calculated to be 0.85 (p = 7.31e-01). Multimedia component 17

Supplemental Figure 11. Synergy analysis of HEX and Amphotericin B by ZIP reference model. 3D (top panel) and 2D (bottom panel) synergy maps. The ZIP synergy score for this combination is 1.32 (p = 5.89e-01). Multimedia component 18

Supplemental Figure 12. RNA-seq validation using RT-qPCR. Three differentially expressed genes (DEGs) from HEX tolerant (A) and BDGR-20191 tolerant cells (B) were quantified using RT-qPCR. Total RNA was extracted from wild-type, HEX-tolerant, and BDGR-20191 tolerant N. fowleri TY strain trophozoites (approximately 5 × 10⁶ cells per sample) using the IBI Total RNA Mini Kit. qRT-PCR was performed using the Luna Universal One-Step RT-qPCR kit in technical triplicate on a CFX Opus 96 real-time PCR machine. Gene expression levels were normalized to 28S rRNA as an endogenous control, and all samples were expressed as relative abundance compared to wild-type controls. The genes are NfTy_025200 (BTB/POZ domain), NfTy_058810 (leucine rich repeat family protein), NfTy_06430 (unspecified product), NfTy_038620 (synthase, hydroxymethylglutaryl-CoA synthase activity), and NfTy_086550 (ATP-binding cassette, sub-family C (CFTR MRP)). Multimedia component 19