Analysis of Acanthamoeba genotypes from public freshwater sources in Thailand reveals a new genotype, T23 Acanthamoeba bangkokensis sp. nov.

A survey of Acanthamoeba in 100 public freshwater sources in 28 provinces across Thailand has identified 9 genotypes comprising T2/6, T3-T5, T9, T11, T12, T18 and a novel ‘T23’ among 131 isolates. Sequencing of the near complete 18S rRNA gene of Acanthamoeba of all isolates has shown that the most predominant genotype T4 found in 87 isolates (66.4%) contained 4 subtypes, i.e. T4A, T4B, T4C and T4F, while all isolates assigned to genotype T2/6 belonged to subtype B. Among intron-bearing genotypes, most isolates harbouring genotype T3 contained S516 introns, characterised by 3 distinct variants whilst all genotypes T4A and T5 were intronless. Identical 18S rRNA sequences of Acanthamoeba were identified across regions of the country and four isolates in this study shared the same sequences with those from remote nations, suggesting that some strains have reproductive success in diverse ecological niche. Nucleotide diversity of genotypes T2/6B, T3, T4, T9 and T11 in this study was significantly less than that among global isolates outside Thailand, implying that limited sequence diversity occurred within local populations. A remarkably higher level of nucleotide diversity in genotype T11 than those of other genotypes (0.041 vs. 0.012–0.024) could be due to cryptic subtypes. Recombination breakpoints have been detected within genotypes and subtypes as well as within isolates despite no evidence for sexual and parasexual cycles in the genus Acanthamoeba. Tajima’s D, Fu & Li’s D* and F* statistics revealed significantly negative deviation from neutrality across genotypes and subtypes, implying purifying selection in this locus. The 18S rRNA gene of the novel genotype ‘T23’ displayed 7.82% to 28.44% sequence differences in comparison with all known genotypes. Both Bayesian and maximum likelihood phylogenetic trees have placed genotype T23 as sister to the clade comprising genotypes T10, T12 and T14, all of these possess cyst structure belonging to morphological group III. Hence, Acanthamoeba bangkokensis sp. nov. is proposed for this novel genotype. It is likely that more genotypes of Acanthamoeba remain to be discovered while the evolution of the 18S rRNA gene of this pathogenic-free living amoeba seems to be ongoing.

www.nature.com/scientificreports/ positive rate of Acanthamoeba seems not to be associated with pH and temperature of water samples. However, Acanthamoeba was more likely to be isolated from still-water than running-water sources ( Table 2).
Distribution of Acanthamoeba genotypes. Of 66 positive sampling sites, genotypes of Acanthamoeba could be assigned for 131 isolates from subculture samples by sequence analysis of the 18S rRNA gene spanning > 2.2 kb. Single genotypes were obtained from 31 sites whilst double, triple, quadruple and quintuple genotypes were found in 16, 10, 7 and 2 sites, respectively (Supplemental Table S1). Six isolates contained two variants of the same genotypes determined by sequencing of secondary PCR products from limiting dilution of primary PCR templates and/or sequencing of recombinant plasmid clones. In total, 137 sequences were obtained. In this survey, 9 genotypes were identified, comprising genotypes T2/6, T3-T5, T9, T11, T12, T18 and a genotype whose sequence differs from other known genotypes. Of these, genotype T4 was most commonly found, occurring in 87 isolates (66.4%). Based on phylogenetic classification proposed by Fuest and colleagues, isolates bearing genotype T4 in this survey could be further assigned to subtypes T4A (n = 4), T4B (n = 33), T4C (n = 27) and T4F (n = 28) whereas all sequences assigned to genotype T2/6 belonged to supertype B (n = 5) 22 (Fig. 2). With previously reported genotype T10 from a Thai keratitis patient and genotype T17 from water sources in Bangkok, altogether 11 genotypes of Acanthamoeba have been identified in Thailand 18 . Meanwhile, the cyst structures based on Pussard and Pond's classification of 130 isolates, excluding an isolate containing an unknown genotype, were congruent with their genotypes 4 . The most common morphological group of 130 Acanthamoeba isolates belonged to group II, accounting for 86.9% while groups I and III could be detected in 7.7% and 5.4%, respectively.

Sequence diversity within genotypes and subtypes.
To analyse the extent of sequence variation in the 18S rRNA gene within genotypes and subtypes of Thai isolates, the corresponding gene sequences from diverse origins available in the GenBank database were retrieved for comparison (Supplemental Table S2). Results revealed that the levels of nucleotide diversity of genotypes T2/6B, T3, T9 and T11 from Thai isolates were significantly less than those from the GenBank database (Table 3). Likewise, the subtypes T4A, T4B, T4C and T4F of Thai isolates had significantly lower levels of nucleotide diversity than those found in more diverse geographic areas. Despite a small number of the T5 genotype sequences among Thai isolates, the level of nucleotide diversity was comparable to those from diverse origins. On the other hand, 12 distinct sequences (alleles) were shared among isolates from two to four different provinces in Thailand ( Fig. 1B and Supplemental Table S3). It is noteworthy that these shared alleles belonged to genotype T3 and subtypes T4A, T4B, T4C and T4F, the prevalent genotypes/subtypes of Acanthamoeba in this survey. Meanwhile, the predominant alleles in subtypes T4B and T4F in this study possessed identical sequences with isolates from our previous survey in 2009 18 . Furthermore, Table 1. Distribution of Acanthamoeba from water sources by region. *No significant differences in the positive rates by pairwise comparison between regions using Fisher exact test (P > 0.05).  Table S3).
Intron-containing genotypes. To date, introns have been documented in genotypes T3, T4, T5 and T15, characterised by 4 unique insertion sites, i.e. S516, S943, S956 and S1389 13,[25][26][27][28][29] . Sequence analysis has shown that 15 of 17 Thai isolates belonging to genotype T3 possessed an intron insertion at S516 with size variation from 480 to 523 bp. Together with previously reported sequences, a total of 5 intronic variants occurred in genotype T3 in which 4 of these were found in Thailand. Variation in intron sequences were mostly due to insertion or deletion (indel), occurring in the predicted helices P2 and P9, and P1, P2, P5 and P9 loops of the predicted secondary structure 27 (Supplemental Figs. S1 and S2). Of these, 4 indels were located in short tandem repeat regions: CA repeats in the P2 helix and homopolymeric C, AT and CT repeats in the P9 helix (Supplemental Figs. S1 and S2). The most prevalent intronic variant (n = 11) containing 523 bp shared identical sequence with an isolate from a keratitis patient in Scotland (GenBank accession no. S81337) and 4 of these Thai isolates share identical sequences across the gene. Likewise, a Thai isolate (AcWT29) from northeast region had the same sequence as the isolates from Slovakia and Czech Republic (GenBank accession nos. GQ397468 and GQ905490). Mean- Figure 2. Distribution of genotypes and subtypes of Acanthamoeba in Thailand. Black, blue and red bars indicate genotypes in this study, previous survey in Bangkok and the novel genotype, respectively. Table 3. Nucleotide diversity (π) within genotypes and subtypes of Acanthamoeba from Thai and non-Thai origins. Genotypes T1, T2/6A, T2/6C, T6-T8, T10, T12-T14, T16, T19, T21 and T22 are excluded (n < 5 sequences spanning > 2 kb). *Parentheses denote number of sequences from Thai isolates in GenBank database. Dash indicates not applicable. Test of the hypothesis that π Thai = π non-Thai ; *P < 0.05, **P < 0.005, ***P < 0.0001. www.nature.com/scientificreports/ while, no intron was detected among Thai isolates bearing genotypes T4 (n = 131) and T5 (n = 5) while genotype T15 was not found in this survey.

Intragenic recombination.
To investigate whether intragenic recombination contributed to sequence diversity in the extant 18S rRNA gene of Acanthamoeba, potential recombination breakpoints were analysed for each genotype and subtype of Thai isolates and previously reported sequences by using the Recombination Detection Program version 4 (RDP4) 30 . Genotypes T2/6C, T8, T14, T19 and T21 were excluded from analysis due to insufficient number of available sequences (n = 2 for each genotype with > 2 kb). Results reveal that potential recombination sites have been identified within genotypes/subtypes T2, T2/6B, T2/6C, T3, T4A-T4G, T5, T6, T9, T11-T13, T15-T18 and T20 ( Fig. 3 and Supplemental Table S4). The potential recombination breakpoints were widely distributed across the 18S rRNA locus (Fig. 3). Meanwhile, analysis of 7 isolates bearing three or more variant sequences from our previous study has shown that 4 of these isolates belonging to genotypes/ subtypes T4B, T4C, T9 and T17 contained recombination breakpoints in this locus 18 (Fig. 3).

Neutrality test.
To determine whether the pattern of sequence diversity in the 18S rRNA gene of Acanthamoeba has been evolved neutrality or under natural selection, Tajima's D and its related statistics, i.e. Fu & Li's D* and F*, were deployed to analyse genotypes/subtypes T3, T4A-T4C, T4D, T4F and T5. Results revealed that almost all genotypes/subtypes yielded significant negative Tajima's D values except subtype T4D. However, the Fu and Li's D* and F* tests gave significant negative deviation from selective neutrality for all genotypes and subtypes ( Table 4).
The novel genotype 'T23'. Isolate AcW61 sampling from a pond inside Chatuchak District in Bangkok had unreadable electropherogram from nucleotide position 95 of the primary PCR amplicon onward upon initial direct sequencing of the PCR product. After limiting dilutions of the primary PCR templates, 2 distinct sequences (AcW61A and AcW61B) were obtained from direct sequencing of the secondary PCR products with clear electropherogram signals throughout the amplicons. A nucleotide deletion at position 96 of the AcW61B sequence relative to the AcW61A sequence seems to be responsible for the unreadable sequences of the primary PCR products. Re-cloning by single cyst isolation of isolate AcW61 followed by cultivation yielded the same results, suggesting allelic variation in the 18S rRNA gene copies of this isolate. Likewise, results from sequencing of recombinant plasmid clones from the primary PCR products of this isolate have reaffirmed AcW61A and AcW61B sequences. Importantly, the 18S rRNA sequences of AcW61A displayed 7.8% to 28.4% sequence dissimilarity in comparison with other known genotypes while AcW61A and AcW61B sequences shared 99.0% identity ( Fig. 4 and Supplemental Table S4). Meanwhile, both Bayesian and maximum likelihood phylogenetic www.nature.com/scientificreports/ trees gave consistent results in which both AcW61A and AcW61B sequences seem to have diverged from genotype T14 while the percentages of sequence similarity/identity were highest when compared with genotype T1 (92.2%), followed by genotypes T20 (91.5%) and T14 ((91.3%), respectively (Figs. 4, 5 and Supplemental Fig. S3). The trophozoites of genotype T23 grows well on 1.5% non-nutrient agar seeded with heat-inactivated E. coli at ambient temperature ranging from 25 °C to 30 °C as well as at 37 °C while no growth is observed at 40 °C and 42 °C. The trophozoites are variable in shape and measure approximately 20-40 μm, with typical hyaline acanthopodia during movement. The endoplasm contains finely granular materials including bacteria upon feeding and one or a few large vacuoles. Each trophozoite possesses a single vesicular nucleus with a prominent centrally placed nucleolus while multiple (usually 2) nuclei can be observed during cellular division. Based on measurement of 100 cysts, the widest diameters range from 10 to 17 μm and the shortest from 9 to 16 μm with mean (± standard deviation) diameters of 12.95 ± 1.77 and 13.26 ± 1.82 μm, respectively (Fig. 6). The ectocyst is rather thin with seemingly smooth or slightly wrinkled surface. The endocyst displays relatively round shape while rounded corner pentagon and hexagon shapes are also observed. It is noteworthy that the cysts of isolate AcW61 belongs to morphological group III akin to those found in genotypes T10, T12 and T14. Taken together, the isolate AcW61 is considered to be a novel genotype of Acanthamoeba, designated 'T23' .

Discussion
Searching for Acanthamoeba in 100 public freshwater sources across 28 provinces of Thailand has identified 9 genotypes, comprising genotypes T2/6, T3-T5, T9, T11, T12, T18 and a novel genotype T23. Consistent with previous reports, genotype T4 predominated in this study, suggesting that Acanthamoeba carrying this genotype could be well adapted for survival in diverse ecological niche 4,15,31 . Despite the fact that different genotypes of Acanthamoeba exhibited variation in pathogenic traits, e.g. presence of serine protease, osmotolerance, thermotolerance and cytopathic effects, the ubiquitous existence of the T4 genotype in environment could partly be responsible for the high prevalence of this genotype as an important causative agent of human infections 31 . Further analysis has shown that the most predominant subtype within T4 in this survey belonged to T4B, followed by www.nature.com/scientificreports/ T4F, T4C and T4A whilst T4D, T4E and T4G were not detected. Although subtype T4A was observed in 4.6% of genotype T4, it is the most predominant subtype of worldwide isolates analysed so far, suggesting differential distribution of T4 subtypes in diverse environment 4,15,31 . It is noteworthy that genotype T10 occurring in a keratitis patient in Thailand could not be isolated from all water sources in this survey, suggesting that freshwater might not be a suitable habitat for this genotype or it could be rare in nature. The occurrence of identical near fulllength 18S rRNA sequences of Acanthamoeba across regions of Thailand and from different period of sampling as well as the presence of identical sequences of isolates from Thailand with those from distant nations could suggest that some strains may successfully grow and occupy diverse ecological niche, albeit more comprehensive genetic analysis is required to address this issue (Supplemental Table S3). Meanwhile, intragenotypic nucleotide diversity of this locus among isolates in this study was significantly less than that among isolates elsewhere outside Thailand for genotypes T3, T9, T11 and subtypes T4A-T4C, T4F and T2/6B, implying that the magnitude of sequence diversity within local populations seems to be relatively limited (Table 3).
To date, a subset of the 18S rRNA sequences of genotypes T3, T4A, T 5 and T15 contains group I intronic insertions at specific locations in conserved domains 13,[25][26][27]32 . Among genotypes T3, T4A and T5 identified in this study, only genotype T3 contained intronic insertion corresponding to S516. Together with sequences available in GenBank database, 2 of 5 intronic variants in genotype T3 have been newly identified in this study (Supplemental Figs. S1 and S2). It is noteworthy that insertion or deletion of nucleotides mainly contributed to sequence variation in intronic sequences and some of which were located in the regions containing homo-or dimeric short tandem repeats (Supplemental Fig. S1). It is likely that slippage strand mispairing mechanism could generate sequence variation in the intron of genotype T3. Meanwhile, the origin of introns has been suggested to be ancient and could have co-evolved with rRNAs. The presence of introns in the same locations of the rRNA genes across domains of life could imply that their existence predated evolutionary divergence and speciation 33 . It has been suggested that the wide spread occurrence of the S516 intron in various eukaryotes could have arisen from vertical transfer while subsequent loss of the intron could occur later in the evolution 29,33 . Although intron gain and loss could occur in rRNAs, the lack of intron in all isolates bearing genotypes T4 and T5 in this study could suggest that the former process may not be common among Acanthamoeba bearing these genotypes in Thailand.
Analysis of the 18S rRNA gene of isolate AcW61 has revealed 2 related sequences, i.e. AcW61A and AcW61B, that could not be assigned to any known genotype. Both sequences shared 99.0% sequence identity, indicating that they belonged to the same genotype. The 18S rRNA sequences of isolate AcW61 exhibited more than 7.8% sequence differences from 22 known genotypes. The Bayesian phylogenetic tree has shown that the AcW61A and www.nature.com/scientificreports/ AcW61B sequences were located in a separate branch with posterior probability support approaching 0.9 from genotype T14 while both sequences were more related with genotypes T10, T12 and T14 than other genotypes. Furthermore, the cyst structure of isolate AcW61 shared the same morphological group III as those in the clade for genotypes T10, T12 and T14, suggesting that these genotypes could have arisen from the common ancestor.
Although the thermotolerance property of Acanthamoeba seems not to be directly related with its pathogenicity, several clinical isolates were capable of growing at 37 °C or at higher temperature 23,[34][35][36] . Although the novel genotype T23 grew well at 37 °C, it remains to be explored whether this genotype is pathogenic to humans. One of the complex characteristics of protozoans in the genus Acanthamoeba is the presence of polyploid genome that contain as many as 25n. Each haploid genome of Acanthamoeba contains 24 copies of the 18S rRNA gene or ~ 600 gene copies per cell 37 . The 18S rRNA gene of Acanthamoeba spanning > 2 kb in this study exhibited extensive sequence diversity among genotypes that stemmed from both nucleotide substitutions and insertions/ deletions. Despite extensive variation in the available number of sequences in each genotype analysed (Table 3), it seems that the levels of nucleotide diversity within genotypes were > 0.01 except genotype 17 which could be due to the low number of available sequences (n = 6). Interestingly, genotype T11 displayed 1.7-4.8 times higher level of nucleotide diversity than other geneotypes, suggesting the existence of subtypes among isolates bearing this genotype. Accordingly, maximum likelihood tree has shown 2 distinct clades within genotype T11 akin to those found as subtypes in genotype T4, implying that genetic diversity and evolutionary divergence within genotypes of Acanthamoeba could be more than currently perceived if more isolates are analysed (Supplemental Fig. S4).
The organization of the multicopy rRNA gene clusters of most prokaryotes and eukaryotes are found in tandem arrays with head-to-tail arrangement. Several lines of evidence have suggested that each gene copy has been generated by the process of concerted evolution in which unequal crossover occurs randomly among gene members 38,39 . According to the model, repeated events of unequal crossover or gene conversion have homogenized the gene members. Therefore, identical or near identical sequences among the rRNA gene members can be found when erroneous incorporation of nucleotide during nuclear division is absence or low. The clear non-superimposed signals in the electropherograms of most Acanthamoeba isolates in this study have suggested that the majority of gene members in the rRNA gene shared identical sequences. Mixed sequences observed in 6 isolates in this study could be from coexistence of subpopulations bearing distinct haplotypes or intragenomic heterogeneity of the 18S rRNA sequences. However, the extent of sequence dissimilarity between distinct haplotypes within isolates upon pairwise comparison was consistently minimal (less than 2%), suggesting intragenomic allelic heterogeneity of this locus. Importantly, the presence of multiple 18S rRNA sequences within cloned Acanthamoeba samples has been previously observed by Stothard and colleagues who found 7 of 53 www.nature.com/scientificreports/ Acanthamoeba strains harboured different alleles within cloned samples, implying that individual Acanthamoeba cell could have multiple 18S rRNA haplotypes albeit belonging to the same genotype 12 . In this study, the 18S rRNA sequences of most isolates (125 of 131) could be determined by direct sequencing of the PCR product, suggesting that the majority of gene members in this locus of most isolates harboured identical sequences or the presence of predominant haplotypes among members in this gene family precluding detection of minor variants upon direct sequencing of the amplicons. Heterogeneity of sequences among 18S rRNA gene members could have arisen during the process of DNA replication by point mutations or slippage strand mispairing between repetitive sequences leading to insertion or deletion of some nucleotides. Meanwhile, analysis of the available > 2 kb 18S rRNA sequences of genotypes T2, T3-T5, T6, T9, T11-T13, T15-T18 and T20, and subtypes T2/6B and T2/6C has revealed that intragenic recombination occurred within these genotypes and subtypes. Furthermore, recombination breakpoints have been identified across the 18S rRNA gene, suggesting recombination has occurred at random. It is noteworthy that Acanthamoeba reproduces asexually by binary fission although highly expressed meiotic genes have been discovered in this organism, leading to a suspicion of cryptic sexual lineages 40,41 . Recombination in the 18S rRNA gene of Acanthamoeba could be generated between gene members carrying distinct sequences; thereby, recombinant sequences were observed in some sequences within genotypes and subtypes. It is noteworthy that the total rate of spontaneous intrachromosomal unequal crossovers in Saccharomyces cerevisiae has been estimated to be 1% per mitotic division which was more frequent than interchromosomal exchanges during meiosis [42][43][44] . Therefore, the lack of sexual stage in Acanthamoeba seems not to preclude genetic recombination in the 18S rRNA locus. The presence of recombination breakpoints in individual gene members within isolates bearing genotypes T4B, T4C, T9 and T17 from our previous study 18 was in accord with intragenomic recombination in the rRNA locus as reported in other organisms [45][46][47][48] . The high copy number of the 18S rRNA gene per genome, as in Acanthamoeba, could promote intrachromosomal crossovers while emerging mutations, albeit at a slow rate in each unit, could spread across all gene members or be eliminated by unequal crossovers or gene conversion 37,49 . Furthermore, the number of repeat units in the rRNA locus could also be increased or decreased after multiple rounds of recombination events 49,50 .
Intragenomic sequence variation among individual member of the rRNA gene has been discovered in various organisms, implying that natural selection may also operate on this gene family 45,[51][52][53] . Sequence variation in the 18S rRNA locus seems to have evolved under different directions of natural selection, probably depending on the functional requirement of organisms or their life histories. For example, intragenomic rRNA polymorphisms in nematodes have been suggested to evolve neutrally 52 whereas sequence variation in the rRNAs associated with 'stress-response ribosome' in Escherichia coli seems to be in line with positive selection 54 . Accordingly, rRNA variants conferring fitness would be maintained by positive selection whilst genetic drift producing deleterious effects could be eliminated by purifying selection 39,51 . Meanwhile, Tajima's D and its related statistics, Fu and Li's D* and F* tests have detected significant negative direction of deviation from selective neutrality in the 18S rRNA gene of Acanthamoeba (Table 4), implying that negative or purifying selection has influenced sequence diversity at this locus. The uniformity of significant negative values of Tajima's D and/or its related statistics across genotypes and subtypes has further suggested that purifying selection could predate genotypic diversification of Acanthamoeba. The role of purifying selection could be an important mechanism to preserve a specific nucleotide sequence as well as its functional property of the rRNA transcripts while deleterious mutations would be subject to elimination.
In conclusion, surveys of freshwater sources across diverse regions of Thailand along with our previous report have revealed at least 11 genotypes of Acanthamoeba in this country 18 . Of these, a novel genotype, designated 'T23' , bearing cyst structure conforming to morphological group III has been identified. Phylogenetic tree has placed the 'T23' genotype within the cluster of genotypes T10, T12 and T14, all of which possess morphological group III cysts. Among intron-containing genotypes, the majority of Thai isolates (15 out of 17) belonging to genotype T3 possessed 4 variants of group I intron located at S516. The presence of identical near complete 18S rRNA sequences across regions of Thailand and across nations has suggested that some strains have reproductive success to occupy diverse ecological niche. The extent of nucleotide diversity of genotype T11 exceeded those of other genotypes, implying the presence of subtypes within this genotype. Evidence for intragenic recombination in the 18S rRNA gene has been detected across genotypes of Acanthamoeba whilst purifying selection seems to operate on this locus. It is likely that more novel genotypes of Acanthamoeba remain to be discovered either from environmental sources or clinical specimens. Genus Acanthamoeba (Volkonsky, 1931). Genotype: T23. Species: Acanthamoeba bangkokensis sp. nov. Diagnosis. The near complete 18S rRNA sequences of isolate AcW61 contains 2 highly related sequences and are sister to genotype T14 on a phylogenetic tree. The trophozoites are not distinguishable from other genotypes while the cyst structure conforms with Pussard and Pond's morphological group III with an average diameters 13.0 by 12.7 μm (range 10-17 × 9-16 μm). The phylogenetic tree inferred from the ASA.S1 fragment of the 18S rRNA gene can differentiate and assign this genotype similar to the tree constructed by using the near complete Etymology. Acanthamoeba, a genus name of pathogenic free-living amoeba with cosmopolitan distribution;

Taxonomic summary
bangkokensis, originating in Bangkok, the capital city of Thailand where this species was first isolated from a public freshwater pond in Chatuchak District.
Gene sequence data. The 18S rRNA sequence has been deposited in GenBank database under accession numbers MZ272148 and MZ272149.

Materials and methods
Samples. Water samples were collected from 100 public water sources including natural or artificial ponds and lakes (n = 72), rivers (n = 5), water falls (n = 11), irrigation canals (n = 7), moats (n = 3) and fountains (n = 2) in 28 provinces of Thailand ( Fig. 1A and Supplemental Table S1). These water sources were selected because they were near residential areas or in public places where people could easily gain access to them. Water samples were taken approximately 20 to 50 cm from the edge of the water sources and 10 to 20 cm in depth using 500 ml sterile polypropylene bottles. The number of samples collected depending on the approximate surface area of each site, i.e. 5 and 10 samples for those with surface area < 500 and > 500 square meters, respectively. For each site, the water samples were collected covering a distance of approximately 5 m interval. The survey period was from January o March 2015. Temperature and pH of water surface were recorded at the time of sampling.
Cultivation of Acanthamoeba. Each freshwater sample was centrifuged at 750 × g for 10 min. The supernatant was discarded and the remaining pellet was applied onto 1.5% non-nutrient agar overlaid with 2 ml of milky suspension of heat inactivated (60 °C, 30 min) Escherichia coli. The culture was incubated at room temperature, varying from 25 °C to 30 °C, and examined daily for 7 consecutive days under an inverted microscope (Olympus IX71, Tokyo, Japan). Positive cultures for trophozoites of Acanthamoeba were incubated further until the characteristic cysts appeared. Acanthamoeba was isolated by cutting off small pieces of agar surface containing the characteristic cysts and transferred them into 1 ml of 1% sulphuric acid and kept over night at ambient temperature to remove other free-living organisms in water. After centrifugation at 750 × g for 10 min, the sediment was used for additional rounds of the same culture condition until Acanthamoeba became overpopulated in the culture. Each Acanthamoeba isolate was obtained from single cyst isolation and subject to the same culture condition. For thermotolerance testing of some isolates, the cultures were incubated at ambient temperature, 37 °C, 40 °C and 42 °C.
Amplification of the 18S rRNA gene. Trophozoites of Acanthamoebae were harvested from in vitro culture originated from established isolates and centrifuged at 750 × g for 10 min. DNA was extracted from the remaining pellet using QIAamp DNA mini kit (Qiagen, Hilden, Germany). The nuclear small subunit rRNA gene encompassing > 2 kb was amplified by PCR using primers ACAN18SF0: 5'-TCC TGC CAG TAG TCA TAT GC-3' (nucleotides 9-28 of A. castellanii Neff strain, GenBank accession number U07416) and ACAN18SR0: 5'-CTT CTC CTT CCT CTA AAT GGT-3' (nucleotides 2236-2256) as described previously 18 . Although the primers are not specific for Acanthamoeba 18S rRNA, the abundant genomic DNA templates from the pre-cultured Acanthamoeba isolates have allowed efficient amplification of the target gene. DNA amplification was carried out in a total volume of 30 µl of the reaction mixture containing template DNA, 2. Limiting dilution of PCR template and sequencing. For isolates containing superimposed signals on electropherogram from direct sequencing of the purified PCR products which could suggest either mixed genotypes or the presence of heterogeneous 18S rRNA templates within each isolate 12,18 . The PCR-amplified 18S rRNA gene products of these isolates were subject to tenfold serial dilutions and used as DNA templates for secondary PCR. The secondary PCR was performed essentially the same as the primary PCR except that the seminested PCR primers ACAN18SF1: 5'-GCT TGT CTC AAA GAT TAA GC-3' (nucleotides  and ACAN18SR0 were used and the amplification was done for 25 cycles. Secondary PCR products generated from the highest dilution of primary PCR products were used as templates for direct sequencing. These serial dilutions, secondary PCR and sequencing were performed in 5 separate batches for each isolates whose sequence contained superimposed signal on electropherogram. www.nature.com/scientificreports/ Subcloning and sequencing. The PCR products of isolate AcW61 were excised from agarose gel, purified by using a QIAquick PCR purification kit (Qiagen), and ligated into pGEM-T-Easy Vector (Promega, Madison WI, USA) as described previously 18,55 . After incubation at 4 °C for overnight, the reaction mixture was precipitated, dissolved in 10 µl of double-distilled water, and transformed into Escherichia coli strain JM109. DNA sequences were determined from at least 5 recombinant subclones of each isolate.

Data analysis.
For genotype assignment, the sequences were aligned according to their secondary structure by using the R-COFFEE program 56,57  . Phylogenetic trees were constructed using maximum-likelihood method and Bayesian approach. Based on analysis of the best model for the sequence data, the maximum likelihood tree was constructed using Tamura-Nei model of nucleotide substitution in which gamma model was used for correction of multiple hits by taking into account the variation of nucleotide frequencies and differences in substitution rate between nucleotides 58 . Reliability of the tree branching patterns was analysed by 1,000 bootstrap pseudoreplicates using the MEGA 6.0 program 59 . Bayesian phylogenetic analysis was performed based on Markov Chain Monte Carlo algorithm with uncorrelated lognormal relaxed clock, coalescent constant population and a 4 category gamma site heterogeneity model. Simulations were run for 20,000,000 cycles and logged at every 1,000 cycles using the Bayesian Evolutionary Analysis by Sampling Trees 2 (BEAST 2) package 60 . Maximum clade credibility tree and mean heights were used to determine posterior probability of trees implemented in the TreeAnnotator program. Tree topology was visualized using the FigTree version 1.4.0 program in the BEAST 2 package 61 . For analysis of haplotype and nucleotide diversity, sequences of each genotype or subtype were aligned separately by using the MAFFT version 7 with manual adjustment 62 . Nucleotide diversity was computed by using the Tamura-Nei substitution model and gamma distribution of evolutionary rate differences among sites whilst the standard error was determined from 1000 bootstrap pseudoreplicates using the MEGA version 6.0 program 59 . Differences between nucleotide diversity values were tested by a two-tailed Z-test and the significance level was set at P < 0.05. Percentage of similarity/identity between sequences was calculated by using the Sequence Manipulation Suite 63 . Analysis of intragenic recombination was performed by using the Recombination Detection Program version 4 (RDP4) 64 that included 3SEQ 65 , Bootscan/Recscan 30 , CHIMAERA 66 , GENCONV 67 , the Maximum Chi Square 68 , RDP4 30 and Sister Scanning 69 methods. To compare the extent of sequence variation within genotypes between Thai and global isolates, 350 sequences containing > 2 kb deposited in the GenBank database were retrieved for comparison. For genotypes with few sequences available, sequences > 1.4 kb were also included for analysis (Supplemental Table S2).
Accession numbers. One hundred and thirty seven near complete sequences of the 18S rRNA gene of Acanthamoeba have been deposited in NCBI GenBank under accession numbers MZ272072-MZ272208.

Data availability
The datasets generated during and/or analyses during the current study are available from the corresponding author upon request.