Characterisation of sterol biosynthesis and validation of 14α-demethylase as a drug target in Acanthamoeba

The soil amoebae Acanthamoeba causes Acanthamoeba keratitis, a severe sight-threatening infection of the eye and the almost universally fatal granulomatous amoebic encephalitis. More effective treatments are required. Sterol biosynthesis has been effectively targeted in numerous fungi using azole compounds that inhibit the cytochrome P450 enzyme sterol 14α-demethylase. Herein, using Gas Chromatography Mass Spectrometry (GCMS), we demonstrate that the major sterol of Acanthamoeba castellanii is ergosterol and identify novel putative precursors and intermediate sterols in its production. Unlike previously reported, we find no evidence of 7-dehydrostigmasterol or any other phytosterol in Acanthamoeba. Of five azoles tested, we demonstrate that tioconazole and voriconazole have the greatest overall inhibition for all isolates of Acanthamoeba castellanii and Acanthamoeba polyphaga tested. While miconazole and sulconazole have intermediate activity econazole is least effective. Through GCMS, we demonstrate that voriconazole inhibits 14α-demethylase as treatment inhibits the production of ergosterol, but results in the accumulation of the lanosterol substrate. These data provide the most complete description of sterol metabolism in Acanthamoeba, provide a putative framework for their further study and validate 14α-demethylase as the target for azoles in Acanthamoeba.

Acanthamoeba is a genus of normally free-living amoeba found in the environment throughout the world. Amoebozoa are single celled organisms and are generally believed to be phylogenetically close to fungi and metazoans, together forming the Amorphea 1 . Acanthamoeba are mixotrophic and have the ability to encyst when the environment becomes hostile. Acanthamoeba can be a facultative pathogen with the ability to cause serious infections. The most common of these is Acanthamoeba keratitis (AK) that can lead to corneal damage and eventual blindness. Clinical cases of AK are most commonly found in contact lens wearers or following trauma to the eye. It is estimated that AK occurs in 1 in 30,000 contact lens wearers in the UK 2 . Acanthamoeba can also be an opportunistic pathogen causing granulomatous amoebic encephalitis (GAE) in people with suppressed immune systems.
Treatment of AK is currently arduous and extremely difficult for the patient requiring hospitalisation due to the requirement for alternating eye drops containing either propamidine isethionate and polyhexamethyl biguanide, or propamidine isethionate and neomycin to be applied on an hourly basis for prolonged periods. To compound matters resistance to polyhexamethyl biguanide is developing 3,4 . An additional problem is that many drugs do not work so well against cysts or even induce Acanthamoeba to encyst. GAE is almost always fatal despite extreme measures including multiple drug combinations combined with surgical procedures or cryotherapy. A potential exploitable difference between Acanthamoeba and their human host is sterol metabolism. While the principal membrane sterol in humans is cholesterol, previous studies have reported that ergosterol and 7-dehydrostigmasterol as the major sterols of Acanthamoeba 5 . Furthermore, sterol metabolism has been successfully targeted to treat a number of fungal infections of humans. Notably, a number of azole antifungal drugs are known to curtail sterol production through the inhibition of sterol 14α-demethylase. While this would have the obvious effect of reducing the integrity of Acanthamoeba membranes alternative mechanisms of action include the build up of toxic precursors of ergosterol. It is encouraging that a number of azoles designed to target fungal sterol 14α-demethylase have proven effective in treatment of patients with Acanthamoeba infections [6][7][8][9] .
This study was initiated to characterise the sterol composition of Acanthamoeba and correlate this with information in the transcriptome project to construct a scheme for sterol biosynthesis in this organism as well as determine the susceptibilities of a number of A. castellanii and A. polyphaga strains to a variety of azoles (econazole, miconazole, sulconazole, tioconazole and voriconazole) using a previously described AlamarBlue assay 10,11 . We demonstrate that ergosterol as previously reported is the major sterol of Acanthamoeba, but find no evidence of 7-dehydrostigmasterol as previously reported 5 and through the use of GCMS provide the first direct evidence that voriconazole inhibits 14α-demethylase in Acanthamoeba.
Comparative susceptibilities of Acanthamoeba castellanii and Acanthamoeba polyphaga isolates to a selection of azoles. The growth of 5 different isolates of Acanthamoeba species (A. castellanii: T4 clinical isolate, ATCC 50370, Neff strain; A. polyphaga: ATCC50371 and CCAP 1501/18) was assessed at 24 hours and 96 hours in the presence of five different azoles (econazole, miconazole, sulconazole, tioconazole and voriconazole), using a modified AlamarBlue microtiter plate assay as described by McBride et al. 10 . A representative set of dose response curves, depicting voriconazole treatment of the Acanthamoeba isolates is shown in Fig. 3. Data for all azoles tested is shown in Table 2. At 24 hours econazole had no effect on the growth of any of the Acanthamoeba tested, whereas miconazole and sulconazole only inhibited ATCC 50370 (IC 50 s 47.7 µM ± 3.5; 46.85 µM ± 6.23, respectively) at 24 hours and tioconazole inhibited ATCC 50370 and ATCC 50371 (IC 50 s 66.66 µM ± 12.12; 8.5 µM ± 1.1, respectively). On the other hand, at 24 hours voriconazole inhibited all isolates with exception of A. castellanii Neff strain (Fig. 2E). All azoles tested inhibited Acanthamoeba growth at 96 hours ( Table 2).

GCMS demonstrates voriconazole specifically inhibits 14-α -demethylase.
As the drug assay demonstrated that voriconazole was the most effective of the drugs tested, being shown to be capable of inducing actual cell death as compared to stasis its method of action was confirmed via GCMS. Lanosterol was undetectable in untreated Acanthamoeba at all time points examined, but accumulated to detectable levels cells treated with voriconazole at concentrations equivalent to a concentration that inhibited 50% of Acanthamoeba growth (IC 50 ) (Fig. 4). In contrast 4,4-dimethyl-cholesta-8,14,24-trienol, a downstream product of 14α-demethylase was abundantly detected in non-treated control cultures, but not voriconazole-treated cultures (data not shown). Ergosterol levels were reduced in voriconazole treated cultures relative to control cultures at all time points examined (Fig. 4). These results therefore demonstrate that voriconazole inhibits Acanthamoeba 14α-demethylase and results in inhibition of ergosterol production.

Discussion
Treatments for diseases caused by Acanthamoeba are far from ideal and rely on a combination of drugs that have shown some efficacy normally in historical or recent clinical cases. In cases where chlorhexidine resistant Acanthamoeba have been encountered topical voriconazole has been used 7 . There have also been case reports of successful treatment of AK and GAE patients using orally administered voriconazole [6][7][8][9]12 . The choice of azole used in each of these cases is not based on laboratory or clinical controlled studies, but rather some positive The structures on column 4 represent the 'fragment ion peak' , while the structure on column 5 represent the 'molecular ion peak' . The fragment ion peak is the base peak which confirms the positions of the double bonds on rings B and C, this information is combined with the molecular ion peak to elucidate the complete structures.
outcomes. Clearly laboratory studies have the ability to inform clinical studies and have the potential to improve treatments. Previous studies have examined relatively few azoles against relatively few isolates of Acanthamoeba making direct comparisons and generalisations regarding efficacy difficult 13,14 . The situation is further compounded by the use of diverse laboratory assay systems previously in common use. Stepping back from these empirical studies, there have been relatively few studies of sterol biosynthesis in Acanthamoeba and those available hampered by the technologies available at the time. Nonetheless there are a number of intriguing aspects of the sterol composition reported in the literature. Firstly, Acanthamoeba castellanii (Neff strain) was reported to have 7-dehydrostigmasterol (60%) and ergosterol (40%) as its major sterols, using infrared and ultraviolet spectroscopy gas chromatography and mass spectral analysis 5 . A further study found that Acanthamoeba culbertsoni in addition to 7-dehydrostigmasterol and ergosterol also had stigmasta-5,7,22,25-tetraene-3β-ol and stigmasta-5, 7,14,22,25-pentaene-3β-ol 15 . The availability of modern GCMS techniques, a microtitre inhibition assay and the recent availability of the Acanthamoeba transcriptome project provides an opportunity to address the aforementioned shortcomings in knowledge 10,16 .
GCMS was first used to determine the major sterol composition of Acanthamoeba castellanii (Neff strain) and a recently isolated A. castellanii clinical T4 isolate described by Mattana et al. 17 . The results demonstrate that the sterol composition is conserved between the two strains examined. We find that ergosterol is the major end product of sterol biosynthesis in Acanthamoeba in contrast with previously published work that reported the phytosterol 7-dehydrostigmasterol as the major sterols. In fact, we find no evidence of any of the phytosterols, including 7-dehydrostigmasterol, in Acanthamoeba. Moreover, using GCMS we have identified a number canonical putative precursors of ergosterol as well as a number of non-canonical precursors. Interrogation of the transcriptome project has allowed us to propose a hypothetical pathway for sterol biosynthesis in Acanthamoeba that incorporates the sterols identified by GCMS. This pathway has elements of the canonical pathway, but also contains elaborations that ultimately feed back into the canonical pathway to provide putative alternative routes to ergosterol. Future studies, potentially using labelled precursors will be necessary to test the unique aspects of this candidate pathway. We have interrogated the transcriptome project for Acanthamoeba and have identified a set of putative predicted genes that we propose provide the means to produce the identified sterols.
In addition, through testing a number of azoles against a set number of isolates of A. castellanii and A. polyphaga we establish that there are relatively small differences between A. castellanii and A. polyphaga isolates in their susceptibility to specific azoles. Consistent with this we demonstrate that the assumed target of these azoles 14α-demethylase is well conserved between all 5 isolates examined. However, we demonstrate that voriconazole is far superior to the other azoles tested as it inhibits parasite growth earlier (at 24 hours) and at lower concentrations. The superior effects of voriconazole are consistent with a previous report that demonstrated this azole Figure 2. Proposed scheme of sterol biosynthesis in Acanthamoeba. Products and intermediates in black were identified by GCMS while those in grey were not. Candidate enzymes, named according to their GenBank Accession numbers, are annotated in red. **Denotes that this protein is not an enzyme but an anchor protein.
and itraconazole were superior to fluconazole in inhibiting A. castellanii and A. polyphaga in vitro 18 . Importantly, this study also noted that voriconazole and itraconazole had lower dissociation constants for Acanthamoeba 14α-demethylase than fluconazole 18 . The results from previous studies that found voriconazole to have an IC 50 of 1.24 µM and 7.16 µM against a variety of Acanthamoeba species are also broadly consistent with the ranges found in this study 6,19 . The fact that voriconazole has also been reported to show efficacy against cysts over a period of 7 days is encouraging 20 .
Although the above studies strongly imply that 14α-demethylase is the target of voriconazole in Acanthamoeba this had not been formally demonstrated. Furthermore, target validation of this enzyme, or any other for that matter, would normally be achieved through gene-deletion experiments which are not currently possible in Acanthamoeba as it is highly polyploid. Thus we employed GCMS to determine the mode of action of voriconazole. Our results demonstrate that voriconazole at IC 50 levels significantly inhibits the production of ergosterol biosynthesis and the ergosterol intermediate 4,4-dimethyl-cholesta-8,14,24 trienol in Acanthamoeba. In addition, lanosterol, the substrate for 14α-demethylase and an ergosterol precursor accumulated in voriconazole treated Acanthmoeaba. These observations validate 14-α-demethylase as the target of voriconazole.
Overall the multidisciplinary studies described herein, making use of the Acanthamoeba transcriptome, analytical GCMS and functional cellular assays yield the most comprehensive description of sterol biosynthesis and composition in Acanthamoeba to date and a candidate biosynthetic pathway for future validation. The results demonstrate the potential benefits of testing diverse azoles in Acanthamoeba and the GCMS technique described herein provides the means to verify their mode of action in Acanthamoeba.

Materials and Methods
Acanthamoeba strains and culture conditions. Five isolates of Acanthamoeba spp. were used for comparative results. An Acanthamoeba castellanii of the T4 genotype isolated from a case of Acanthamoeba keratitis (referred to A. castellanii T4 clinical) was donated by Dr. Antonella Mattana (Sassari, Italy). A. castellanii (Neff) was originally obtained from Dr Keith Vickerman (Glasgow, United Kingdom). A. castellanii ATCC 50370 and Acanthamoeba polyphaga ATCC 50371, American clinical keratitis isolates, were originally obtained from the American Type Culture Collection (ATCC, United States) and A. polyphaga CCAP 1501/18 was originally obtained from the Culture Collection of Algae and Protozoa (CCAP, United Kingdom).  GC-MS sample preparation. 8 × 10 6 Acanthamoeba were pelleted by centrifugation at 1,200 g for 5 minutes. The media was then discarded and 5 ml of a 1/1 chloroform/(90% methanol, 10% water) mix was added. The cells were then transferred to a clean glass vial and left overnight at 4 °C. The cells were homogenised and the lower chloroform containing layer of approximately 2.5 ml was transferred to a clean glass tube. Additional chloroform was then added and the cells homogenised again, this was repeated twice more accounting for a total of 10 ml of chloroform being extracted. The chloroform solvent was then evaporated and the lipids reconstituted with 500 µl of chloroform prior to being analysed. The presence of steroids was analysed on a Thermo Finnigan Polaris Plus GCMS. Initial temperature was at 100 °C, initial time of 1.00 min, with 1 ramp at a rate of 20.0 deg/min to a final temperature of 320 °C. EIMS is set at detector voltage of 500.0 V with a source temperature of 220 °C while the GC interface temperature is at 220 °C with an emission current of 150 µV. Full scan acquisition is at 2.5 scans per second where the scan range starts at 40 amu to end at 650 amu. Syringe volume is at 10 µl. The column was a Factory Four TM capillary (VF-1ms 30 m × 0.25mm, 0.25 µm). Compounds were identified with NIST online database library. Bioinformatical analysis. Genes encoding the enzymes belonging to the sterol synthesis pathway were identified in GenBank ™ (http://www.ncbi.nlm.nih.gov/genbank) both by word search (specific enzyme name AND Acanthamoeba), verified in amoebaDB 21 and BLASTx for those enzymes not yet specifically identified in the Acanthamoeba genome/transcriptome using combination of Saccharomyces cerevisiae and Arabidopsis thaliana sterol pathway enzymes. The pathway was constructed using the KEGG metabolic pathway (http://www. genome.jp/kegg/pathway/map/map00100.html).
PCR for 14α-demethylase. RNA was extracted and complementary DNA was synthesised as described in ref. 22. Oligonucleotides (5′ CCTCCATCTTCTGAATCTGTCGAGG 3′ and 5′ GGTGTGATTGCTCACGTTCGCCTTC 3′) were designed using PCR primer tool in Genbank and they were synthesised commercially by Life Technologies Ltd. (Paisley, UK). Automated sequencing was performed commercially by Source Bioscience (Nottingham UK).  AlamarBlue Assay. The AlamarBlue, resazurin assay was carried out as previously described 10 . Seeding densities for each of the isolates at each of the time points was determined empirically as previously described. Each drug was tested over a wide range of concentrations for efficacy against the 5 Acanthamoeba isolates at 23.3 °C for 24 hours or 96 hours. Six hours prior to the end of the incubation period 10 μl of AlamarBlue reagent (Abd Serotec, UK) was added to all wells except wells that act as a blank. The plate was then incubated for a further six hours at 23.3 °C in the dark. Controls consisted of a media blank, trophozoites with media alone, trophozoites with the highest concentration of solvent found in the drug samples, trophozoites with 12.5 μM chlorhexidine and media with the highest concentration of drug alone. The final volume for all wells was 100 μl.