Water samples from multiple sites along the Nakdong River in Korea were collected in sterilized 1-L bottles (Winners, Busan, Korea). Each site was characterized by a specific water grade (Fig. 1): Grade I (Ia and Ib)–Daecheon, Cheolma, Hoedong, Baenaegol, and Yusan; Grade II–Geumho, Samnak, Yangsan, and Uksu; Grade III–Mulgeum, Seonakdong, and Suyeong; Grade IV–Pyeonggang, Seokdae, Jangnim, Nakdong, and Samrak (wetland); and Grade V (V and VI)–Hwamyung and Gamjeon [18,19].

Agar medium for culturing the amoebae was prepared by initially autoclaving a 2% suspension of Bacto-agar (Biosciences, San Jose, CA, USA), which was cooled to a temperature of approximately 50°C–60°C and supplemented with amphotericin B solution (Sigma, St. Louis, MO, USA) to inhibit fungal growth. Agar plates were prepared by pouring 12 ml aliquots of the agar into Petri dishes (SPL, Pocheon, Korea); the plates were smeared with a 20% suspension of heated Escherichia coli [20].

Filtering apparatus (filter flask, glass base, graduated funnel, and aluminum clamp) that had initially been autoclaved and sterilized were used to filter the collected water samples. The water samples were filtered through 0.45-μm pore filters (Advantec membrane filters, Tokyo, Japan). The filters were lifted from the device using a scalpel, and forceps were inverted and placed on the surfaces of the E. coli-inoculated agar plates, which were then incubated at 25°C. The filtering apparatus was resterilized between the filtering of each sample to prevent cross-contamination.

Fungal growth and amoebae were microscopically checked every day after incubation. The amoebae were isolated, if detected. Regions on plates with dense amoebal growth were marked with a pen, excised, and placed on the surfaces of fresh agar plates. After obtaining the monoxenic cultures, we excised gel segments with good amoebal growth and placed them in a solution of 1% HCl, followed by incubation overnight at 25°C for axenization. The following day, after centrifuging at 1,500 rpm for 5 min, the resulting supernatants were removed, and 7 ml of peptone yeast extract glucose medium was added to the remaining pellets; the suspensions were then transferred to T-25 flasks, followed by incubation at 25°C [20].

Genomic DNA was extracted from the cultured amoeba to identify the species of FLA and endogenous bacteria. The lysis buffer (SNET buffer) used to extract the DNA contained 20 mM Tris-HCl (pH 8.0, 1 M Tris-HCl; Dongin Biotech, Korea), 5 mM EDTA (pH 8.0, 0.5 M EDTA; Dongin Biotech), 400 mM NaCl (Duksan Pure Chemical, Ansan, Korea), 1% SDS (BiosesangTM, Yongin, Korea), and distilled water (DW), which was filtered before use using a 50-ml syringe and a 0.45-μm syringe filter (Sartorius Stedim Biotech, Göttingen, Germany). After extracting the amoebae from the 1.5% agar gels using 1× PBS, the suspensions were centrifuged at 15,000 rpm for 10 min. Following the removal of the supernatants, the remaining pellets were resuspended in protease K-supplemented SNET buffer at a concentration of 0.1 g/4 ml. The samples were incubated overnight in a 55°C-incubator using a rocker. The following day, an equal volume of phenol: chloroform: isoamyl alcohol (25:24:1) reagent was added to the samples, followed by incubation on a rocker for 30 min at room temperature. The mixtures were centrifuged at 15,000 rpm for 5 min at room temperature, and the resulting supernatants were transferred to fresh Eppendorf tubes. An equal volume of isopropanol was added, and the samples were centrifuged at 8,000 rpm for 15 min at 4°C. The resulting supernatants were discarded, and the pellets were washed with 1 ml of 70% ethanol by centrifuging at 8,000 rpm for 5 min at 4°C. The ethanol was discarded, following which the pellets were air-dried for 20 min and dissolved in diluted 1×TE buffer (10×TE buffer; Dongin Biotech).

Polymerase chain reaction (PCR) was performed to identify the amoebae isolates using reaction systems containing 2 μg of template DNA, 0.5 μl of Taq buffer, 3 μl of 10× buffer, 2 μl of dNTPs, and 2 μl each of the forward and reverse 18S rDNA primers (V4-1F 5′-GCGGTAATTCCAGCTC-3′ and V4-4R 5′-GCCMTTCCGTCAATTCC-3′) [21]. The amplifications were performed with forward and reverse 16s rDNA primers (27F 5′-AGAGTTTGATCCTGGCTCAG-3′ and 1492R 5′-GGTTACCTTGTTACGACTT-3′) to identify the bacterial endosymbiont, and the total volume was adjusted to 30 μl using DW. The PCR conditions were as follows: initial denaturation at 95°C for 5 min, followed by 34 cycles of denaturation at 95°C for 30 sec, annealing at 55°C for 30 sec, extension at 72°C for 45 sec, and a final extension step at 72°C for 5 min, after which the samples were held at 4°C. The amplified products obtained were loaded onto 1% agar gels and subsequently sequenced commercially by Macrogene (Seoul, Korea).

The nucleotide sequences were compared with those of published strains via BLAST searches. Clustal X was used for pairwise alignment and to calculate the percent sequence identities. The percentage similarity corresponds to the number of identical sites between 2 isolates aligned against each other, as in the master alignment. The phylogeny was determined using the Clustal W2 program (http://www.ebi.ac.uk/Tools/clustalw2/index.html) with a low gap penalty, and phylogenetic trees were constructed using TreeView software. All sequences obtained in this study have been registered in GenBank.

Samples were prepared for transmission electron microscopy by initially prefixing with 2.5% glutaraldehyde at 4°C, in a phosphate buffer (pH 7.2) and postfixing with 1% osmium tetroxide using the same buffer. The material was then dehydrated through a graded series of ethanol and embedded in epoxy resin (Epon 812 mixture). Sections (1 μm thick) were stained with 1% toluidine blue for examination under a light microscope. Thin sections (50–60 nm) were prepared using an ultramicrotome (EM UC7; Leica) and double-stained with uranyl acetate and lead citrate. The stained sections were observed under a JEM-1200EXII transmission electron microscope (JEOL).

The filters through which 500 ml of fresh water had been passed were observed at daily intervals under a light microscope to determine differences in the populations of FLA based on the water quality. The time to initial detection of FLA decreased with the deterioration in the quality of the collected water sample (Fig. 2). In the case of the Grade I water samples, FLA were initially observed over a protracted period of 5–14 days after inoculation, whereas in the Grade II and III samples, the FLA were detected within a week; the FLA were detected within 2 days in the Grade IV and V water samples. These observations indicated the increase in the level of FLA contamination as the quality of water deteriorated. Thus, amoeba in water samples of poor quality can be detected more rapidly when cultured on agar plates.

FLA were detected in all 18 water samples collected and cultured under monogenic and axenic conditions in this study. FLA were readily detected and isolated in most Grade I and II water samples. However, in poorer quality water (Grades III–V), the detection and isolation of the amoeba proved to be more difficult, owing to the rapid and prolific growth of bacteria and fungi. Consequently, only 3 FLA were isolated from the Grade III to V quality water samples (Table 1). Two types of FLA were detected in samples collected from Cheolma and Hoedong, and monoxenic cultures of amoeba were obtained for all Grade II water samples, except those collected from Baenaegol and Yusan. In total, 4 species of amoebae were isolated in the genus Acanthamoeba and one in the genus Vermamoeba (Hartmannella) from the Grade I water samples (KA/DC, KA/CM1, KA/CM2, KA/HD1, and KA/HD2). We obtained monoxenic cultures for all isolates detected in the Grade II water samples based on which 4 types of FLA (KA/GH, KA/SR, KA/YS, and KA/US) were identified; however, not all were successfully cultured in axenic medium. Three of these were identified as species of Acanthamoeba, whereas the remaining isolate (KA/YS) was identified as Vannella croatica, via molecular analysis. All 3 isolates obtained from the Grade III to V water samples were identified as species of Acanthamoeba (KA/MG, KAHM, and KAGJ).

For the morphological analyses, a microscope was used to take photographs of amoebae derived from monogenic cultures at ×400 magnification (Fig. 3). Among the identified isolates, the cysts of KA/DC appeared as small spheres, but no trophozoites were detected. KA/YS (a species of Vannella), presented as notably large trophozoites (approximately 34 μm in length) that were not readily distinguishable from the cysts of this amoeba. Similarly, KA/GJ, a group I species of Acanthamoeba, was characterized by large trophozoites (approximately 30 μm in length), whereas the morphology of the cysts in the KA/MG isolate (a group III species of Acanthamoeba) differed notably from that of the cysts of other group II species of Acanthamoeba. The remaining isolates, all group II species of Acanthamoeba, consisted of both ecto- and endotype cysts (Supplementary Fig. S1).

The 18S rDNA sequences of the 12 isolates were compared with those of the reference strains of 21 Acanthamoeba, one Vermamoeba, and one Vannella species using ClustalW. The genetic characteristics of most of these isolates were similar to those of the group II Acanthamoeba species, with the sequences of KA/HD2, KA/CM1, and KA/CM2 being similar to those of type A. polyphaga and A. triangularis; alternatively, KA/GH appeared to be closely related to A. castellani Neff and species in the A. lugdunensis L3a complex. The KA/HM5 and KA/SR sequences were similar to that of A. quina. The genetic distance between KA/HD1 and KA/MG was less than 0.01%, placing these isolates in the same group as the group III Acanthamoeba species A. pustulosa and A. palesteinsis. The KA/DC isolate was closely related to species in the Vermamoeba genus, and KA/YS was closely related to Vannella croatica. In the case of KA/US, the sequence was genetically similar to the Acanthamoeba species; however, a large genetic distance was observed between this isolate and most of Acanthamoeba. Moreover, the KA/US sequence did not exhibit a close match with any of the reference strains; hence, it could be a previously unrecorded FLA (Fig. 3).

In this study, 8 FLA isolates were successfully cultured in axenic medium, and endosymbionts based on 16S rDNA PCR analysis, for which similar sequences have been reported previously (Fig. 4), were detected in all of them, except for KA/CM1. KA/HD1 showed the highest similarity (97%) to the recently published 16S rDNA sequence of a Holosporaceae bacterium Serialkilleuse 9,403,403 strain. The 16S rDNA sequences of the endosymbionts of KA/SR and KA/HM were found to be closely related to Sinorichettsia chlamys and an endosymbiont of Acanthamoeba sp. KA/E9, respectively. The putative endosymbionts in KA/US, KA/CM2, and KA/MG were closely related, with sequences similar to those of the Candidatus Paracaedibacter acanthamoebae PRA3 strain.

The morphological and ultrastructural characters of the detected endosymbionts were elucidated via transmission electron microscopy (Fig. 5). All the rod-shaped endosymbionts detected in the FLA isolates were estimated to be approximately 1.72 (1.70–1.75)×0.27 (0.20–0.30) μm in size (Fig. 5A–E). In each case, the bacterial membrane was in close contact with the cytoplasm of the host FLA isolates. In isolate KA/SR, we detected a further bacteria-like structure within the cytoplasm, which lacked a membrane structure and appeared to be surrounded by a food vesicle (red arrowhead, Fig. 5F). Although 16S rDNA was detected in the KA/GJ isolate, we could not identify any endosymbiont-like structure with the cytoplasm of these amoebae.