Evolution of the Sterol Biosynthetic Pathway of Pythium insidiosum and Related Oomycetes Contributes to Antifungal Drug Resistance

ABSTRACT

Pythiosis is a life-threatening infectious disease caused by the oomycete Pythium insidiosum. Direct exposure to Py. insidiosum zoospores can initiate infections of the eye, limb, gastrointestinal tract, or skin/subcutaneous tissue. Treatments for pythiosis have mostly relied on surgery. Antifungal drugs are generally ineffective against Py. insidiosum. However, one patient with an invasive Py. insidiosum infection recovered completely following treatment with terbinafine and itraconazole. Additionally, the drug target sterol biosynthetic enzymes have been identified in the oomycete Aphanomyces euteiches. It remains an open question whether Py. insidiosum is susceptible to the antifungal drugs and harbors any of the known drug target enzymes. Here, we determined the in vitro susceptibilities of terbinafine and itraconazole against 30 isolates of Py. insidiosum. We also analyzed endogenous sterols and searched for genes encoding the sterol biosynthetic enzymes in the genomes of Py. insidiosum and related oomycetes. The susceptibility assay showed that the growth of each of the Py. insidiosum isolates was inhibited by the antifungal agents, but only at difficult-to-achieve concentrations, which explains the clinical resistance of the drugs in the treatment of pythiosis patients. Genome searches of Py. insidiosum and related oomycetes demonstrated that these organisms contained an incomplete set of sterol biosynthetic enzymes. Gas chromatographic mass spectrometry did not detect any sterol end products in Py. insidiosum. In conclusion, Py. insidiosum possesses an incomplete sterol biosynthetic pathway. Resistance to antifungal drugs targeting enzymes in the ergosterol biosynthetic pathway in Py. insidiosum was due to modifications or losses of some of the genes encoding the drug target enzymes.

INTRODUCTION

Pythiosis is a life-threatening infectious disease and has been increasingly reported from tropical, subtropical, and temperate countries (1–4). The etiologic agent is the fungus-like organism Pythium insidiosum (1). The microscopic morphology of Py. insidiosum is similar to that of filamentous fungi (1, 3). However, based on phylogenetic analyses, Py. insidiosum belongs to the oomycetes, a unique group of microorganisms in several genera, such as Pythium, Phytophthora, Hyaloperonospora, Albugo, Saprolegnia, and Aphanomyces (5, 6). Unlike most other pathogenic oomycetes, which infect plants, Py. insidiosum infects humans and animals (i.e., horses and dogs) (1–3, 6). In the environment, Py. insidiosum presents as mycelia and motile zoospores (7). Py. insidiosum infection occurs when a patient contacts or ingests pathogen-contaminated water. Common sites of the Py. insidiosum infection include the skin, eye, artery, and gastrointestinal tract (1–3).

Due to the aggressive nature of the infections, patients or animals with pythiosis often die if proper management is not provided. The treatment of choice for pythiosis has mostly relied on extensive surgery to remove the infected organ (2, 3). Use of an immunotherapeutic vaccine (which is a crude protein extract of Py. insidiosum) exhibits various efficacies, depending on the affected host, with 97% favorable responses in cattle, 60 to 85% in horses, 56% in humans, and only 33% in dogs (8, 9). Conventional antifungal drugs, inhibiting the production of fungal ergosterol, have been used for treatment of pythiosis, but drug resistance is usually observed (2, 4, 10). The oomycetes, including Phytophthora and Pythium species, are sterol auxotrophs and thus require exogenous sterols for basic metabolism (11–13). Lack of the drug target sterol biosynthetic enzymes in these organisms explains treatment failure of the antifungal drugs (12, 14).

Contrary to the above findings, Shenep et al. reported on successful medical treatment of a patient with invasive Py. insidiosum infection using a combination of two antifungal drugs: terbinafine and itraconazole (15). In addition, Madoui et al. analyzed the transcriptomic data of Aphanomyces euteiches (16, 17) and showed that this oomycete microorganism contains several sterol biosynthetic enzymes responsible for production of endogenous sterols (12). Such discordant results raise questions about Py. insidiosum's susceptibility to the antifungal drugs and whether the pathogen produces the drug target enzymes in the sterol biosynthetic pathway.

A better understanding of the drug resistance mechanism(s) of Py. insidiosum could lead to the development of more efficient agents for the control of this and related pathogens. Recently, genome sequences and transcriptomic data of many oomycetes (i.e., Py. insidiosum [18–21], non-insidiosum Pythium species [22, 23], Phytophthora species [14, 24, 25], Hyaloperonospora species [26], Albugo species [27], Saprolegnia species [28], and Aphanomyces species [17]) have been made publicly available. These genetic resources were used to explore the sterol biosynthetic pathway in Py. insidiosum and related oomycetes. In this study, we aimed to demonstrate the in vitro growth inhibitory effects of two key antifungal drugs (terbinafine and itraconazole) against clinical isolates of Py. insidiosum. We also sought to identify endogenous sterols and search for genes encoding sterol biosynthetic enzymes in this pathogen.

RESULTS

Antifungal drug susceptibility of 30 Py. insidiosum isolates.

Two key antifungal drugs, terbinafine and itraconazole (concentrations of 0 [no drug], 0.5, 2, 8, 32, 64, and 128 mg/liter), were tested against 30 isolates of Py. insidiosum. In relation to the no-drug control, terbinafine at 2 mg/liter modestly, but significantly, increased the average growth of Py. insidiosum to 116% (Fig. 1A). In contrast, at the concentrations 32, 64, and 128 mg/liter, terbinafine significantly reduced the average growth of Py. insidiosum to 39%, 7%, and 0.7%, respectively (Fig. 1A). Terbinafine at concentrations of 32, 64, and 128 mg/liter completely inhibited the growth of 1, 8, and 13 of the Py. insidiosum isolates, respectively.

FIG 1.

In vitro drug susceptibility assay of 30 Py. insidiosum isolates against antifungal drugs. The drug concentrations of terbinafine (A) and itraconazole (B) used in the in vitro susceptibility assay are 0.5, 2, 8, 32, 64, and 128 mg/liter. A dot represents the percentage of radial growth of an individual isolate in relation to the growth of the same isolate exposed to no drug (control). *, statistical significance compared to the no-drug control.

When Py. insidiosum was tested against 0.5 mg/liter itraconazole, the average growth was modestly, but not significantly, increased to 110% of that of the no-drug control (Fig. 1B). However, when the organism was tested against the higher drug concentrations, the average growth of Py. insidiosum was significantly decreased to 78% (at 8 mg/liter itraconazole), 65% (32 mg/liter), 61% (64 mg/liter), or 59% (128 mg/liter) of that of the no-drug control (Fig. 1B). None of the Py. insidiosum isolates was completely inhibited by itraconazole.

Identification of Py. insidiosum sterols.

The sterol extract of Py. insidiosum (strain Pi-S) was analyzed for the presence of the sterol end products, such as cholesterol, stigmasterol, ergosterol, and fucosterol. Commercially available sterol standards (cholesterol, stigmasterol, ergosterol, and fucosterol) were used to generate the reference gas chromatography-mass spectrometry (GC-MS) peaks. The sterol extract from Candida albicans strain ATCC 90028 was also prepared and analyzed in parallel with the extract from Py. insidiosum. To serve as the internal control, 5α-cholestane was added to the sterol extracts of Py. insidiosum and Ca. albicans. Two hundred milligrams of lyophilized cells from either Py. insidiosum or Ca. albicans was derivatized for GC-MS analysis. The retention times of the reference sterols were 15.34 min for 5α-cholestane (peak 1), 25.80 min for cholesterol (peak 2), 29.26 min for ergosterol (peak 3), 31.56 min for stigmasterol (peak 4), and 33.88 min for fucosterol (peak 5) (Fig. 2A). Two GC-MS peaks, corresponding to 5α-cholestane (peak 1) and ergosterol (peak 3), were identified in Ca. albicans (Fig. 2B), whereas only one GC-MS peak, corresponding to the internal control 5α-cholestane, was detected in Py. insidiosum (data not shown). Even when the amount of lyophilized cells from Py. insidiosum was increased 7-fold (1,400 mg), only the 5α-cholestane peak was observed (Fig. 2C).

FIG 2.

Total ion chromatogram of the sterol extracts generated by GC-MS analysis: (A) Standard mixture of sterol end products, (B) sterol extract from the Ca. albicans strain ATCC 90028 (200 mg lyophilized cells), and (C) sterol extract from Py. insidiosum strain Pi-S (1,400 mg lyophilized cells). Peaks 1 to 5 indicate 5α-cholestane (internal control), cholesterol, ergosterol, stigmasterol, and fucosterol, respectively.

Identification of enzymes involved in sterol biosynthesis.

The full-length deduced protein sequences of 22 enzymes involved in the mevalonate pathway (HMG1, HMG2, ERG8, ERG10, ERG12, ERG13, ERG19, and IDI1) and the sterol biosynthetic pathway (ERG1 to -7, ERG9, ERG11, ERG20, and ERG24 to -27) of the model organism Saccharomyces cerevisiae (Fig. 3; Tables 1 to 3), were subject to BLAST search against the translated genomes or transcriptomes of 17 oomycetes (including Py. insidiosum), three algae, two fungi, and one diatom (Table 4). Both fungi (Ca. albicans and Aspergillus fumigatus), two algae (Aurantiochytrium limacinum and Ectocarpus siliculosus), and only one oomycete (Saprolegnia parasitica) harbored the complete set of 8 enzymes in the mevalonate pathway (Table 2). Only the fungi possessed the complete set of 14 enzymes in the sterol biosynthetic pathway (Table 3). ERG5, ERG10, ERG11, and ERG26 homologs were identified in all species included in this study. ERG2 homologs were absent in all nonfungal microorganisms.

FIG 3.

Schematic diagram showing enzymes and substrates in the mevalonate (upper part) and sterol biosynthesis (lower part) pathways. ERG1 and ERG11 are the drug targets of terbinafine and itraconazole, respectively.

TABLE 1.

The best BLAST hits and accession numbers of orthologous enzymes in the mevalonate and sterol biosynthetic pathways presented in the Py. insidiosum genome^a

Protein	Enzymeb	S. cerevisiae enzyme accession no.	Py. insidiosum genome
			Protein ID	Accession no.	E value
ERG1	Squalene epoxidase	NP_011691
ERG2	C-8 sterol isomerase	NP_013929
ERG3	C-5 sterol desaturase	NP_013157	PINS02380006C	GAQ12368.1	3.0E−63
ERG4	C-24(28) sterol reductase	NP_011503
ERG5	C-22 sterol desaturase	NP_013728	PINS02520001C	GAQ12369.1	2.0E−17
ERG6	Sterol 24-C-methyltransferase	NP_013706
ERG7	Lanosterol synthase	NP_011939
ERG8	Phosphomevalonate kinase	NP_013947	PINS00060035C	GAQ12361.1	6.0E−54
ERG9	Squalene synthase	NP_012060
ERG10	Acetyl-CoA acyltransferase	NP_015297	PINS00050004C	GAQ12360.1	4.0E−133
ERG11	Lanosterol 14-α-demethylase	NP_011871	PINS02520001C	GAQ12369.1	2.0E−11
ERG12	Mevalonate kinase	NP_013935
ERG13	HMG-CoA synthase	NP_013580	PINS00120058C	GAQ12362.1	2.0E−120
ERG19	Diphosphomevalonate decarboxylase	NP_014441	PINS01150035C	GAQ12364.1	2.0E−92
ERG20	Farnesyl diphosphate synthase	NP_012368	PINS01980016C	GAQ12366.1	6.0E−102
ERG24	Delta14-sterol reductase	NP_014119	PINS01230016C	GAQ12365.1	2.0E−38
ERG25	Methylsterol monooxygenase	NP_011574
ERG26	Sterol-4-α-carboxylate 3-dehydrogenase	NP_011514	PINS00200056A	GAQ12363.1	3.0E−16
ERG27	3-Keto-steroid reductase	NP_013201
HMG1	HMG-CoA reductase 1	NP_013636
HMG2	HMG-CoA reductase 2	NP_013555
IDI1	Isopentenyl-diphosphate delta-isomerase	NP_015208	PINS02140012C	GAQ12367.1	1.0E−49

^aThe best BLAST hits were defined as having a cutoff E value of less than −6. Shown are accession numbers of orthologous enzymes in the mevalonate and sterol biosynthetic pathways presented in the genome of Py. insidiosum strain Pi-S, isolated from a human patient with vascular pythiosis (19). The query sequences were derived from Saccharomyces cerevisiae.

^bCoA, coenzyme A; HMG, hydroxymethylglutaryl.

TABLE 3.

The best BLAST hits of orthologous enzymes in the sterol biosynthetic pathway presented in the genomes of 17 oomycetes, 3 algae, 2 fungi, and a diatom^a

Organism	Group	Subgroup	E value for enzyme in the sterol biosynthetic pathway
			ERG20	ERG9	ERG1	ERG7	ERG11	ERG24	ERG25	ERG26	ERG27	ERG6	ERG2	ERG3	ERG5	ERG4
Ca. albicans	Fungi	Ascomycota	5.0E−155	1.0E−138	4.0E−143	0.0E+00	0.0E+00	3.0E−78	2.0E−118	1.0E−139	4.0E−111	8.0E−144	5.0E−66	6.0E−118	0.0E+00	7.0E−166
As. fumigatus	Fungi	Ascomycota	2.0E−118	8.0E−107	2.0E−89	0.0E+00	5.0E−147	2.0E−146	2.0E−100	2.0E−56	5.0E−36	1.0E−116	8.0E−52	3.0E−91	2.0E−171	2.0E−145
Au. limacinum	Algae	Labyrinthulida	3.0E−99	5.0E−65		4.0E−138	1.0E−73	4.0E−103	4.0E−17	9.0E−34	1.0E−08	5.0E−68		1.0E−62	6.0E−33	4.0E−53
Au. anophagefferens	Algae	Pelagophyceae	6.0E−94	1.0E−70		1.0E−129	6.0E−58	1.0E−73	1.0E−21	3.0E−25		4.0E−83		4.0E−10	1.0E−10	3.0E−102
Ec. siliculosus	Algae	PX clade	1.0E−118	8.0E−91	4.0E−65	4.0E−79	2.0E−66	2.0E−100	5.0E−09	1.0E−37		1.0E−99		1.0E−16	2.0E−17	3.0E−55
Ph. tricornutum	Diatom	Bacillariophyta	7.0E−115	1.0E−70		5.0E−126	3.0E−69	1.0E−100	6.0E−10	1.0E−33		2.0E−81		1.0E−19	1.0E−26	1.0E−52
Ap. euteicheb	Oomycete	Saprolegniales	3.0E−94	2.0E−69	4.0E−56	4.0E−66	8.0E−68	2.0E−73	1.0E−61	2.0E−25		2.0E−78			6.0E−16	1.0E−38
Sa. parasitica	Oomycete	Saprolegniales	9.0E−118	3.0E−79	4.0E−65	2.0E−158	1.0E−80	2.0E−84	2.0E−62	2.0E−50		5.0E−84		1.0E−62	1.0E−17	3.0E−50
Py. insidiosumc	Oomycete	Pythiales	6.0E−102				2.0E−11	2.0E−38		3.0E−16				3.0E−63	2.0E−17
Py. ultimum	Oomycete	Pythiales	1.0E−119				6.0E−12	3.0E−36	8.0E−08	6.0E−18		2.0E−10		3.0E−79	3.0E−14	6.0E−26
Py. arrhenomanes	Oomycete	Pythiales	6.0E−105				3.0E−14	1.0E−32		2.0E−19		7.0E−10			3.0E−17	2.0E−30
Py. irregulare	Oomycete	Pythiales	4.0E−123				4.0E−15	4.0E−40	1.0E−08	1.0E−14		9.0E−12		4.0E−81	3.0E−16	1.0E−29
Py. iwayamai	Oomycete	Pythiales	2.0E−120				6.0E−17	4.0E−31	4.0E−07	5.0E−14		3.0E−10		1.0E−71	1.0E−15	4.0E−22
Py. vexans	Oomycete	Pythiales	5.0E−95				1.0E−11	2.0E−36		4.0E−14		5.0E−12		1.0E−78	1.0E−14	1.0E−26
Py. aphanidermatum	Oomycete	Pythiales					6.0E−15	5.0E−35	6.0E−11	2.0E−19		6.0E−14		2.0E−81	9.0E−16	4.0E−31
Ph. sojae	Oomycete	Peronosporales	1.0E−118				3.0E−13	4.0E−38	4.0E−07	4.0E−12		9.0E−12		2.0E−82	2.0E−16	6.0E−29
Ph. ramorum	Oomycete	Peronosporales	4.0E−119				2.0E−14	6.0E−38		3.0E−13		2.0E−11		7.0E−79	2.0E−16	5.0E−26
Ph. parasitica	Oomycete	Peronosporales	1.0E−115				4.0E−16	1.0E−39	2.0E−07	3.0E−14		4.0E−14		4.0E−79	3.0E−16	2.0E−26
Ph. capsici	Oomycete	Peronosporales	1.0E−117				7.0E−12	8.0E−38	2.0E−08	9.0E−16		4.0E−14		8.0E−79	7.0E−14	1.0E−27
Ph. cinnamomi	Oomycete	Peronosporales	4.0E−117				5.0E−16	4.0E−37	8.0E−07	7.0E−13		6.0E−10		3.0E−80	5.0E−18	2.0E−27
Ph. infestans	Oomycete	Peronosporales	1.0E−115				2.0E−13	4.0E−39	9.0E−08	1.0E−14		2.0E−13		1.0E−74	3.0E−14	7.0E−28
Hy. arabidopsis	Oomycete	Peronosporales	3.0E−105				2.0E−14			1.0E−12		4.0E−11			5.0E−14
Al. laibachii	Oomycete	Albuginales	2.0E−107				2.0E−10			6.0E−16		4.0E−08			2.0E−11

^aThe best BLAST hits of orthologous enzymes in the sterol biosynthetic pathway are defined as having a cutoff E value of less than −6. The query sequences were derived from Saccharomyces cerevisiae.

^bOnly transcriptome data were available and used for analysis.

^cPy. insidiosum (strain Pi-S) was isolated from a human patient with vascular pythiosis (19).

TABLE 4.

Results from genome and transcriptome databases for 17 oomycetes, 3 algae, 2 fungi, and a diatom used for the sterol biosynthetic enzyme search

Organism	Group	Genome size (Mb)	Accession no. or source
Pythium insidiosuma	Oomycetes	53.24	BBXB01000001 to BBXB01001192
Pythium ultimum	Oomycetes	42.8	http://pythium.plantbiology.msu.edu/
Pythium aphanidermatum	Oomycetes	35.88	http://pythium.plantbiology.msu.edu/
Pythium arrhenomanes	Oomycetes	44.67	http://pythium.plantbiology.msu.edu/
Pythium irregulare	Oomycetes	42.97	http://pythium.plantbiology.msu.edu/
Pythium iwayamai	Oomycetes	43.2	http://pythium.plantbiology.msu.edu/
Pythium vexans	Oomycetes	33.84	http://pythium.plantbiology.msu.edu/
Phytophthora sojae	Oomycetes	82.6	http://www.broadinstitute.org/
Phytophthora ramorum	Oomycetes	66.65	http://www.broadinstitute.org/
Phytophthora parasitica	Oomycetes	82.39	http://www.broadinstitute.org/
Phytophthora capsici	Oomycetes	56.04	http://genome.jgi.doe.gov/
Phytophthora cinnamomi	Oomycetes	77.97	http://genome.jgi.doe.gov/
Phytophthora infestans	Oomycetes	228.54	http://www.broadinstitute.org/
Hyaloperonospora arabidopsis	Oomycetes	73.89	http://protists.ensembl.org/
Albugo laibachii	Oomycetes	32.76	http://protists.ensembl.org/
Aphanomyces euteichesb	Oomycetes		http://www.polebio.lrsv.ups-tlse.fr/aphano/
Saprolegnia parasitica	Oomycetes	53.13	http://www.broadinstitute.org/
Phaeodactylum tricornutum	Diatoms	25.05	http://genome.jgi.doe.gov/
Aureococcus anophagefferens	Brown tide algae	56.66	http://genome.jgi.doe.gov/
Ectocarpus siliculosus	Brown algae	391.72	http://bioinformatics.psb.ugent.be/genomes/
Aurantiochytrium limacinum	Microalgae	60.93	http://genome.jgi.doe.gov/
Candida albicans	Fungi	27.56	http://www.candidagenome.org/
Aspergillus fumigatus	Fungi	29.39	http://www.aspgd.org/

^aPy. insidiosum (strain Pi-S) was isolated from a human patient with vascular pythiosis (19).

^bOnly transcriptome data were available and used for analysis.

TABLE 2.

The best BLAST hits of orthologous enzymes in the mevalonate pathway presented in the genomes of 17 oomycetes, 3 algae, 2 fungi, and a diatom^a

Organism	Group	Subgroup	E value for enzyme in the mevalonate pathway
			ERG10	ERG13	HMG1	HMG2	ERG12	ERG8	ERG19	IDI1
Ca. albicans	Fungi	Ascomycota	2.0E−154	0.0E+00	0.0E+00	0.0E+00	8.0E−88	6.0E−81	6.0E−135	1.0E−85
As. fumigatus	Fungi	Ascomycota	1.0E−137	1.0E−164	0.0E+00	0.0E+00	8.0E−68	3.0E−53	1.0E−115	9.0E−73
Ec. siliculosus	Algae	PX clade	3.0E−136	2.0E−125	9.0E−98	4.0E−101	4.0E−45	4.0E−45	1.0E−100	6.0E−52
Au. limacinum	Algae	Labyrinthulida	2.0E−118	8.0E−119	3.0E−139	1.0E−136	2.0E−15	3.0E−36	2.0E−116	3.0E−62
Au. anophagefferens	Algae	Pelagophyceae	3.0E−114	3.0E−109			6.0E−09		5.0E−41	3.0E−53
Ph. tricornutum	Diatoms	Bacillariophyta	3.0E−136	6.0E−105	5.0E−131	2.0E−127	8.0E−44			7.0E−57
Ap. euteicheb	Oomycetes	Saprolegniales	1.0E−101		1.0E−117	1.0E−101		9.0E−30
Sa. parasitica	Oomycetes	Saprolegniales	7.0E−137	2.0E−119	1.0E−134	2.0E−133	4.0E−34	1.0E−45	2.0E−95	2.0E−59
Py. insidiosumc	Oomycetes	Pythiales	4.0E−133	2.0E−120				6.0E−54	2.0E−92	1.0E−49
Py. ultimum	Oomycetes	Pythiales	2.0E−133				2.0E−40	4.0E−60	4.0E−99	3.0E−55
Py. arrhenomanes	Oomycetes	Pythiales	1.0E−133	3.0E−118			2.0E−35	6.0E−07	2.0E−106	2.0E−52
Py. irregulare	Oomycetes	Pythiales	1.0E−126	6.0E−120			4.0E−34	7.0E−48	7.0E−93	1.0E−54
Py. iwayamai	Oomycete	Pythiales	1.0E−120	3.0E−75			7.0E−39	3.0E−57	6.0E−26	4.0E−55
Py. vexans	Oomycetes	Pythiales	4.0E−85	3.0E−121			2.0E−43	2.0E−36	2.0E−77	1.0E−52
Py. aphanidermatum	Oomycetes	Pythiales	5.0E−133	8.0E−121			6.0E−42	8.0E−14	6.0E−89	1.0E−51
Ph. sojae	Oomycetes	Peronosporales	9.0E−128	1.0E−119				7.0E−40	7.0E−103	1.0E−50
Ph. ramorum	Oomycetes	Peronosporales	1.0E−128	9.0E−120			2.0E−40	7.0E−37	6.0E−104	5.0E−49
Ph. parasitica	Oomycetes	Peronosporales	5.0E−135	1.0E−120			5.0E−41	2.0E−56	6.0E−34	1.0E−56
Ph. capsici	Oomycetes	Peronosporales	8.0E−132	1.0E−123			1.0E−43	4.0E−39	7.0E−105	2.0E−57
Ph. cinnamomi	Oomycetes	Peronosporales	9.0E−129	5.0E−117			1.0E−38	9.0E−38	1.0E−100	3.0E−56
Ph. infestans	Oomycetes	Peronosporales	4.0E−127	3.0E−117			9.0E−42	3.0E−36	1.0E−103	2.0E−53
Hy. arabidopsis	Oomycetes	Peronosporales	2.0E−127	3.0E−110			1.0E−37		1.0E−104	8.0E−54
Al. laibachii	Oomycetes	Albuginales	1.0E−124	6.0E−119			6.0E−17	4.0E−51	2.0E−92	3.0E−50

^aThe best BLAST hits of orthologous enzymes in the mevalonate pathway are defined as having a cutoff E value of less than −6. The query sequences were derived from Saccharomyces cerevisiae.

^bOnly transcriptome data were available and used for analysis.

^cPy. insidiosum (strain Pi-S) was isolated from a human patient with vascular pythiosis (19).

Regarding the oomycetes, two enzymes in the mevalonate pathway (HMG1 and HMG2) and three enzymes in the sterol biosynthetic pathway (ERG1, ERG7, and ERG9) were absent in all organisms of the non-Saprolegniales subgroups (i.e., Pythiales, Peronosporales, and Albuginales) but present in all organisms of the Saprolegniales subgroup (Tables 2 and 3). Py. insidiosum contained 5 out of 8 enzymes in the mevalonate pathway, including ERG8, ERG10, ERG13, ERG19, and IDI1 (Table 2). In addition, Py. insidiosum harbored genes for only 6 out of 14 sterol biosynthetic enzymes: ERG3, ERG5, ERG11, ERG20, ERG24, and ERG26 (Table 3).

Phylogenetic analysis of ERG11.

ERG11-encoding sequences of all microorganisms included in this study were retrieved for phylogenetic analysis (Table 3). The resulting tree showed two major phylogenetic groups (Fig. 4): group 1 included 3 fungi (S. cerevisiae, Ca. albicans, and Aspergillus fumigatus), 3 algae (Au. limacinum, Aureococcus anophagefferens, and Ec. siliculosus), 2 oomycetes (Aphanomyces euteiches, and Sa. parasitica), and one diatom (Phaeodactylum tricornutum), and group 2 contained the 15 remaining oomycetes (7 Pythium species [Py. insidiosum, Py. ultimum, Py. aphanidermatum, Py. arrhenomanes, Py. irregulare, Py. iwayamai, and Py. vexans], 6 Phytophthora species [Ph. capsici, Ph. infestans, Ph. sojae, Ph. ramorum, Ph. parasitica, and Ph. cinnamomi], Hyaloperonospora arabidopsis, and Albugo laibachii).

FIG 4.

Phylogenetic analysis of ERG11 orthologs of 17 oomycetes, 3 fungi, 3 algae, and a diatom. A phylogenetic tree was reconstructed based on the neighbor-joining algorithm (see Materials and Methods). The branch support values of at least 70% are shown at corresponding nodes. *, oomycete; #, fungus.

DISCUSSION

Inconsistent effects of antifungal drugs in the treatment of pythiosis patients have been reported (2, 12, 15). We have sought to determine whether 30 isolates of Py. insidiosum are susceptible to the antifungal agents and whether the organism harbors the drug target enzymes of the sterol biosynthetic pathway. There is no standardized antibiotic susceptibility test for Py. insidiosum. In our preliminary work to develop such a test, we considered using zoospores or hyphal suspensions as inocula (29, 30). However, in our hands, generation of zoospores from Thai strains is difficult and labor-intensive and did not provide enough zoospores for the susceptibility testing. Likewise, hyphal suspensions, prepared as described by other investigators (15, 30, 31) were not reproducibly uniform in our hands (data not shown). As an alternative, we have used the method of Brown and coworkers (32), starting with agar plugs of active hyphal growth and measuring radial growth of the hyphae. They demonstrated that the radial growth assay was a simple and reproducible method for susceptibility testing of Py. insidiosum and argued that the submerged-growth nature of Py. insidiosum in agar is an advantage in that the organism is kept in direct contact with the antifungal drug. We have found that this assay is simple, reliable, and reproducible. To control for differences in the growth of the different Py. insidiosum isolates, we calculated the percentage of radial growth of each isolate as the growth with exposure to the antifungal drug divided by growth of the same isolate with no drug × 100.

Although RPMI is a commonly used medium for the in vitro susceptibility testing of Py. insidiosum (30), we have chosen to use Sabouraud dextrose (SD) agar for these assays. Since a particular medium could interfere with antifungal drug activity, this may be a concern. However, Stopiglia et al. compared the performances of RPMI and SD agar for in vitro susceptibility testing of a pathogenic fungus against various antifungal drugs (33). They reported no significant differences between the MICs obtained in the two media for many antifungal drugs tested (including itraconazole and terbinafine), with one exception—i.e., amphotericin B. Thus, in the present study, we used SD agar (which is lower cost and easier to prepare) as the medium for in vitro susceptibility testing of Py. insidiosum against itraconazole and terbinafine.

As shown here, two key antifungal drugs (terbinafine and itraconazole) had different impacts on growth of Py. insidiosum. At relatively low concentrations (0.5 to 2.0 mg/liter), both drugs paradoxically slightly enhanced the growth in the majority of Py. insidiosum isolates (Fig. 1). At higher concentrations (≥8 mg/liter), both drugs inhibited the pathogen in a dose-dependent manner (Fig. 1). Terbinafine was more inhibitory than itraconazole against the organism, which is consistent with the published results of Argenta et al. (34) and Fonseca et al. (30). Both of these groups performed comprehensive in vitro susceptibility testing of terbinafine and itraconazole against 30 and 22 Brazilian isolates of Py. insidiosum, respectively. To completely inhibit the growth of 50% of the isolates tested required at least 16 mg/liter of itraconazole and 4 mg/liter of terbinafine in the studies by Argenta et al. (34) and at least 64 mg/liter of itraconazole and 8 mg/liter of terbinafine in those by Fonseca et al. (30), while our group found that more than 128 mg/liter of both drugs was required. The different inhibitory effects of the two drugs against Py. insidiosum may be due to the differences in the strains of Py. insidiosum tested (i.e., Thai versus Brazilian strains) or the differences in the susceptibility methods used. (There is no standardized susceptibility assay available for the organism.)

In the present study, complete growth inhibition of some Py. insidiosum isolates was observed only with terbinafine at extremely high concentrations (64 to 128 mg/liter [Fig. 1A]). Although Py. insidiosum can be inhibited using the high drug concentrations, such concentrations are difficult to achieve in vivo. Kovarik and coworkers studied the pharmacokinetics and tissue distribution of terbinafine and showed that the peak plasma concentration of the drug is ∼1.7 mg/liter, while the peak tissue concentrations in sebum, hair, nail, and skin are ∼20, ∼10, ∼1, and ∼1 mg/liter, respectively (35). These plasma and tissue concentrations (especially for the skin, which is a target site of Py. insidiosum infection) are lower than the minimal drug concentration required to inhibit the organism in our study. Our results explain why Py. insidiosum clinically resists this antifungal drug as observed in most pythiosis patients (2).

In fungi, there are 14 enzymes in the sterol biosynthetic pathway (Fig. 3). The mechanisms of action of terbinafine and itraconazole rely on interfering with the synthesis of sterol at the ERG1 step (converting squalene to squalene epoxide) and the ERG11 step (catalyzing lanosterol to dimethylcholestatrienol), respectively. Based on the genome search, Py. insidiosum contained only six enzymes (ERG3, ERG5, ERG11, ERG20, ERG24, and ERG26) (Table 3) in the sterol biosynthetic pathway. The presence of the drug target ERG11, with an E value of −11, may explain the limited inhibitory effect of itraconazole against Py. insidiosum (Fig. 1B). However, the lack of the drug target ERG1 in the Py. insidiosum genome contradicts the fact that the organism was notably sensitive to terbinafine. There might be another target protein responsible for terbinafine-mediated growth inhibition in Py. insidiosum. This possibility is especially likely in light of the high in vitro concentrations needed to see growth inhibition by terbinafine.

We used GC-MS to detect the sterol end products identified in various organisms—i.e., cholesterol (animals), stigmasterol (plants), ergosterol (fungi), and fucosterol (algae and oomycetes) (13)—to find evidence to support the presence of the sterol biosynthetic pathway in Py. insidiosum. We were able to detect ergosterol in the fungus Ca. albicans. However, no sterol end products were detected in Py. insidiosum, even when the starting amount of sterol extract was 7-fold increased versus that used for Ca. albicans (Fig. 2). Taken together, the bioinformatic and GC-MS findings present a coherent picture that Py. insidiosum lacks a complete set of sterol biosynthetic enzymes, which leads to its failure to produce the sterol end products.

The mevalonate pathway generates dimethylallyldiphosphate, which is a precursor of the sterol biosynthetic pathway. Based on the oomycete genome searches for the homologous enzymes present in the mevalonate and sterol biosynthetic pathways (Fig. 3), we found that the oomycetes in the Saprolegniales subgroup (i.e., Ap. euteiches and Sa. parasitica) possess a nearly complete set of the enzymes in such pathways, whereas the oomycetes in the subgroups Pythiales, Peronosporales, and Albuginales (i.e., Pythium species, Phytophthora species, Hyaloperonospora species, and Albugo species) contain a restricted number of such enzymes (Tables 2 and 3). This finding is consistent with the findings that many oomycetes (including Py. insidiosum) of the non-Saprolegniales subgroups are sterol auxotrophs (11, 13), while the oomycetes of the subgroup Saprolegniales can synthesize their own sterols (i.e., cholesterol and fucosterol), as clearly shown for Ap. euteiches (12). The ability to synthesize sterols in the oomycetes of the Saprolegniales subgroup and the fungi could indicate their derivation from an ancient common ancestor. However, even if so, during the course of evolution, some critical enzymes involved in sterol synthesis (i.e., HMG1, HMG2, ERG1, ERG7, and ERG9) have been lost in the oomycetes of the non-Saprolegniales subgroups. As a compensatory mechanism, the non-Saprolegniales oomycetes increased their capacity to acquire exogenous sterol by greatly expanding the number of genes encoding sterol-binding and -uptake proteins—the elicitin gene family (36–38). In contrast, this elicitin gene expansion is absent in the Saprolegniales oomycetes (36, 38).

In the sterol biosynthetic pathway, ERG1 and ERG11 are the target enzymes of terbinafine and itraconazole, respectively. The enzyme-encoding sequence of ERG11, but not ERG1, was identified in the genomes of all organisms included in this study (Table 3). The ERG11 sequences were then used for investigation of the phylogenetic relationship among oomycetes, fungi, algae, and a diatom. The resulting ERG11-based tree located most of the oomycetes in group 2 and the remaining oomycetes (Ap. euteiches and Sa. parasitica) and all other organisms (fungi, algae, and diatom) in group 1 (Fig. 4). Based on the BLAST search using the query sequence from S. cerevisiae, the E values among the sterol-producing Saprolegniales oomycetes (E value range, −80 to −68) were much lower than that of the sterol auxotrophic, non-Saprolegniales oomycetes (E value range, −17 to −10) (Table 3). This finding suggests that the rate of gene modification (i.e., mutation, insertion, or deletion) of ERG11 since divergence of the oomycetes from the common ancestor with the fungi (represented by S. cerevisiae) was higher for the non-Saprolegniales oomycetes than that for the Saprolegniales oomycetes. This relationship is also found for other enzymes, i.e., ERG6, ERG24, ERG25, and ERG26 (Table 3), and it suggests that there was less selective pressure on the non-Saprolegniales oomycetes to maintain the sterol biosynthetic pathway.

In conclusion, the in vitro susceptibility testing demonstrated that the Py. insidiosum isolates were sensitive to the antifungal agents, but only at concentrations that are difficult or impossible to achieve in vivo. These results explain the clinical resistance of the drugs in the treatment of most pythiosis patients. The genome analysis of Py. insidiosum demonstrated that the organism contains an incomplete set of sterol biosynthetic enzymes. GC-MS analysis did not detect any sterol end products in Py. insidiosum. Clinical resistance of antifungal drugs in Py. insidiosum results from modifications or losses of some of the genes encoding the drug target sterol biosynthetic enzymes.

MATERIALS AND METHODS

Ethics statement.

This study was approved by the Committee on Human Rights Related to Research Involving Human Subjects, at the Faculty of Medicine, Ramathibodi Hospital, Mahidol University (approval no. MURA2011/412/S3). Informed consent was not obtained from patients (from whom microorganisms were obtained) because the data were analyzed anonymously, and the institutional ethics committee waived the need for written informed consent from the participants.

Microorganisms and growth conditions.

Thirty isolates of Py. insidiosum from patients with vascular pythiosis (n = 10), ocular pythiosis (n = 7), cutaneous pythiosis (n = 3), and undefined forms of pythiosis (n = 10) were recruited in this study. The identity of each Py. insidiosum isolate was confirmed by zoospore induction (39, 40) and ribosomal DNA (rDNA) sequence homology analysis (41). All organisms were maintained on SD agar (pH 7.2) at room temperature.

Ten small pieces of SD agar containing a 7-day-old actively growing colony of Py. insidiosum (strain Pi-S) were transferred into 100 ml of SD broth (pH 7.2) and incubated, with shaking (150 rpm), at 37°C for 10 days. The organism was killed with 0.02% (wt/vol) thimerosal (Sigma). The hyphal mat was collected by filtration of the cultured broth through a 0.22-μm-pore-size membrane (Durapore; Merck Millipore). Candida albicans strain ATCC 90028 was cultured in yeast extract-peptone-dextrose medium, with shaking (250 rpm), at 30°C for 3 days. The yeast cells were collected by centrifugation at 6,000 × g for 10 min. Harvested organisms were lyophilized using a Genevac EZ-2 Elite machine and stored at −30°C until used for sterol extraction.

In vitro antifungal susceptibility assay.

The radial growth of Py. insidiosum was determined using the methods described by Brown et al. and Krajaejun et al. (32, 42), with some modifications. Briefly, Py. insidiosum was subcultured on SD agar at 37°C for 5 days. A 5-mm-diameter SD agar plug with actively growing mycelium was cut from the edge of a young Py. insidiosum colony and placed on SD agar (pH 7.2) containing various concentrations (128, 64, 32, 8, 2, 0.5, and 0 [no-drug control] mg/liter) of terbinafine (purity, 100% [Novartis]) or itraconazole (purity, 100% [Janssen Pharmaceutical]). The agar plate was incubated at 37°C for 2 days. Two colony diameters of each Py. insidiosum isolate were measured at day 2. All isolates were tested in duplicate. The average colony diameter was subtracted by 5 mm (the diameter in millimeters of an agar plug) and then divided by 2 to obtain a mean radial growth. The percentage of radial growth of each isolate, after exposure to the antifungal drug, was calculated based on the growth of the same isolate with no drug (control). The percentages of radial growths of Py. insidiosum exposed to different drug concentrations were statistically compared using the GraphPad Prism program version 5.00 (GraphPad Software, USA) and one-way analysis of variance (ANOVA) with Dunnett's posttest.

Sterol extraction.

Sterol extraction was performed using the methods of Axelsson et al. (43) and Madoui et al. (12), with some modifications. Briefly, the lyophilized samples (∼200 mg of either Ca. albicans yeasts or Py. insidiosum hyphae), 10 μg of 5α-cholestane (Sigma [used as an internal control]), and 3 ml of 1% (wt/vol) KOH in methanol were mixed and incubated at 80°C for 90 min. After the reaction mixture was cooled to room temperature, 1 ml of water and 2 ml of hexane were sequentially added to the mixture and incubated on a rocker (BioSan MR-1) overnight at room temperature. The organic phase (upper layer), containing sterols, was transferred to a new glass tube and evaporated using nitrogen gas. The residual material was derivatized at 60°C for 30 min using 20 μl of pyridine (Sigma) and 50 μl of N, O-bis (trimethylsilyl) trifluoroacetamide (Thermo Scientific). The derivatized material was dissolved in 500 μl toluene and subjected to gas chromatography-mass spectrometry (GC-MS).

Four commercially available sterols (including cholesterol [Tokyo Chemical Industry], ergosterol [Acros Organics], stigmasterol [Tokyo Chemical Industry], and fucosterol [Sigma]) were mixed and served as the standard for GC-MS analysis. After addition of 5α-cholestane, the sterol standard mixture was derivatized (as described above) and dissolved in 500 μl toluene.

GC-MS.

The sterol extracts and the sterol standard mixture were analyzed using GC-MS (model 7890A-5975C [Agilent Technologies]; capillary column [model 19091S-433HP-5MS] with 5% phenyl methyl siloxane [size, 30 m by 250 μm by 0.25 μm]). The electro-ionization mass spectrometry was set to the scan mode with an m/z range of 50 to 500. The selected-ion monitoring (SIM) mode was used to analyze trace amounts of sterol compounds: 5α-cholestane masses (m/z of 217, 357, and 372), cholesterol masses (m/z of 129, 329, 368, and 458), ergosterol masses (m/z of 131, 337, 363, and 468), stigmasterol masses (m/z of 83, 255, 394, and 484), and fucosterol masses (m/z of 257, 296, 386, and 484). Temperatures were set as follows: injector, 260°C; oven, 3°C/min increments from 230°C to 239°C, 1°C/min increments to 280°C, and then 280°C for 2 min; MS source, 230°C; and MS quadrupole, 150°C.

Sequence homology analysis.

Genomes or transcriptomes of 17 oomycetes (14, 17, 19, 22–28, 44, 45), two fungi (Candida albicans strain SC5314 and Aspergillus fumigatus strain Af293) (46, 47), and one each of diatom (48), microalga (45), brown tide alga (49), and brown alga (50) were retrieved from public databases (Table 4). The BLAST 2.2.28+ program (http://www.ncbi.nlm.nih.gov/) was locally installed for genome search analysis. The sequences of 22 enzymes (including ERG1 to -13, ERG19, ERG20, ERG24 to -27, HMG1, HMG2, and IDI1) involved in endogenous sterol production of the Saccharomyces cerevisiae strain S288c (51) were obtained from the NCBI database (Fig. 3; Tables 1 to 3) and used to BLAST search against all genomes or transcriptomes included in this study. The cutoff E value was 10⁻⁶.

Phylogenetic analysis.

Phylogenetic analysis of the ERG11 homologs, coding for the lanosterol 14-α demethylase (which is the drug target of itraconazole) from all microorganisms, was performed online at www.phylogeny.fr (Table 3) (52). Briefly, the sequences were aligned using the MUSCLE program (53). The phylogenetic relationship was analyzed by neighbor joining with 1,000 bootstraps (54) and the Jones-Taylor-Thornton matrix substitution model (55). A phylogenetic tree was generated by TreeDyn (56).

Sequence accession numbers.

All sequences of orthologous enzymes in the mevalonate and sterol biosynthetic pathways presented in the genome of Py. insidiosum (strain Pi-S) have been submitted to the DNA Data Bank of Japan database, under the accession numbers indicated in Table 1.