ARTICLE IN PRESS

http://www.elsevier.de/protisPublished online date 20 January 2006

1

CorrespondinFax: +44 20 76e-mail graham2Present addrHospital NHSGB, UK.

& 2005 Elsevdoi:10.1016/j

157, 77—85, January 2006

Protist, Vol.

ORIGINAL PAPER

DNA Barcoding of Blastocystis

Stephanie M. Scicluna2, Blessing Tawari, and C. Graham Clark1

Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine,Keppel Street, London WC1E 7HT, UK

Submitted October 11, 2005; Accepted December 4, 2005Monitoring Editor: Michael Melkonian

We have developed a simple method for subtyping the intestinal protistan parasite Blastocystis usingan approach equivalent to DNA barcoding in animals. Amplification of a 600 bp region of the smallsubunit ribosomal RNA gene followed by single primer sequencing of the PCR product providesenough data to assign isolates to specific subtypes unambiguously. We believe that this approach willprove useful in future epidemiological studies.& 2005 Elsevier GmbH. All rights reserved.

Key words: Blastocystis; phylogeny; ribosomal RNA; subtyping.

Introduction

‘DNA barcoding’ is the term used to describe amethod proposed for producing a unique identifierfor all living species (Hebert et al. 2003). To datethis effort has focussed on identification ofanimals and the use of a 648 bp segment nearthe 50 end of the mitochondrial gene encodingcytochrome C oxidase subunit I (COI-50). Therationale for using this particular sequence is four-fold: mitochondrial DNA is present in multiplecopies per cell making PCR amplification of thetarget DNA relatively easy; there already exists alarge database of mitochondrial sequencesagainst which novel sequences can be compared;the gene is usually well conserved within speciesbut sufficiently variable that interspecific differ-ences can be detected easily; and the gene

g author;36 8739.clark@lshtm.ac.uk (C.G. Clark)

ess: Department of Microbiology, HammersmithTrust, Du Cane Road, London W12 0HS,

ier GmbH. All rights reserved..protis.2005.12.001

segment is of a size that can be easily amplifiedby PCR then sequenced in a single reaction. Theresulting sequences can be used for both speciesidentification and for limited phylogenetic analysisto aid in placing those organisms not alreadyrepresented in the database. This approach hasalready led to a number of publications represent-ing studies of a variety of animal groups (Hebert etal. 2004; Hogg and Hebert 2004).

The proponents of this approach recognise,however, that this gene is not suitable for studyingall groups of organisms (http://www.barcodingli-fe.org/). For example, rates of evolution of COI aretoo slow to allow this gene to be useful in plantstudies and other sequences have been proposed(Kress et al. 2005). In addition, anaerobic eukar-yotes do not normally have a mitochondrialgenome and therefore lack the target genealtogether. For DNA barcoding in such organismsan alternative target is required, a gene with therequisite characteristics but that is present in theorganism of interest.

The protist Blastocystis hominis is the onlystramenopile parasitic in humans (Silberman

ARTICLE IN PRESS

78 S.M. Scicluna et al.

et al. 1996). It is also not really a true species butrather a species complex consisting of manydifferent variants that can differ by over 7% intheir small subunit ribosomal RNA gene (SSU-rDNA) sequence (Clark 1997). Other animals canbe infected with the same range of variantsmaking the taxonomy of this group uncertain atbest. In the remainder of this article the organismswill be referred to for simplicity as Blastocystisirrespective of its host origin.

Blastocystis is of uncertain pathogenicity inhumans (Stenzel and Boreham 1996). Mostpopulation-based studies find no difference be-tween rates of infection in symptomatic andasymptomatic individuals. In contrast, in individualinfections a strong case can be made for theorganism being the cause of disease. The recon-ciling of these two disparate views may lie inunderstanding more about the different geneticvariants. If only one type of Blastocystis causesdisease in humans, for example, this would nothave been uncovered by any population survey todate as they have relied solely on microscopy and/or growth in culture for species identification.

At least 7 major subtypes of Blastocystis havebeen isolated from humans and other mammals(Noel et al. 2005; Yoshikawa et al. 2004). Thesecan be considered to be molecular operationaltaxonomic units (MOTU; Blaxter 2004). Most ofthe methods used to detect these subtypes haverelied on finding differences in the nuclear SSU-rDNA, either indirectly using restriction enzymepatterns (Clark 1997; Hoevers et al. 2000; Riveraand Tan 2005; Snowden et al. 2000) or by directlysequencing the gene (Abe 2004; Arisue et al.2003; Noel et al. 2005; Thathaisong et al. 2003).Restriction enzyme patterns may miss or over-emphasise differences in sequence and are quitelabour intensive to produce. However, for epide-miological studies sequencing of the completegene is unrealistic. It is ca. 1800 bp in length and inthe most recent publication on diversity theamplified gene was also cloned before beingsequenced (Noel et al. 2005). Realistically, a rapid‘DNA barcoding’ approach needs to be developedbefore evidence for a link between subtype andthe outcome of infection can be studied in largenumbers of Blastocystis isolates.

In the present study we have investigated theusefulness of a 600 bp segment of the BlastocystisSSU-rDNA for detecting the established subtypesof Blastocystis. We show that this smaller region ofthe gene can be sequenced in one reaction andthat this allows all the subtypes to be identified.Most of the established subtype relationships can

also be recovered in phylogenetic analyses usingthis region alone. We believe that, using thisapproach, large-scale epidemiological surveyscan now be undertaken with the aim of settlingthe question of Blastocystis pathogenicity in hu-mans once and for all.

Results

Primer design and testing

A selection of 25 Blastocystis SSU-rDNA se-quences representing all known mammalian sub-types were aligned, including two or moresequences of each subtype and all ‘unclassified’sequences. SSU-rDNA sequences representing arange of other human intestinal protozoan para-sites were also included. Regions were identifiedin which conservation across all Blastocystissubtypes was seen but that were not wellconserved in the other human parasites. Onesuch region is at around 600 bases from the 50 endof the Blastocystis SSU-rDNA. A primer comple-mentary to this sequence (BhRDr) was designedand used subsequently in all PCR amplificationand sequencing. Although its sequence doesmatch that of some other eukaryotes (especiallystramenopiles) it should not amplify the SSU-rDNA from other organisms likely to be found inthe mammalian intestinal tract.

The primer was tested using Blastocystis DNAextracted from short-term cultures established aspart of the diagnostic routine at two laboratories.Amplification of the 50 600 bp region of theBlasotocystis SSU-rDNA was efficient and gavelittle background. PCR product yield was high andno additional bands were seen. Subsequently, thelack of background was confirmed by our ability tosequence the amplicon directly and obtain un-ambiguous sequence in most instances.

Phylogenetic testing

All of the unique Blastocystis rDNA sequencesincluded in the analysis of Noel et al. (2005) wereedited to include only the region under study hereand a Bayesian phylogenetic analyses was under-taken. The aligned region was 626 bp in lengthand had 235 unique site patterns. Bayesiananalysis of the 50 one-third of the gene producesa tree similar to that of Noel et al. (2005) butsupported by lower posterior probabilities. How-ever, the sequences form clearly defined cladesthat are the same as those seen with the complete

ARTICLE IN PRESS

DNA Barcoding of Blastocystis 79

gene (Fig. 1). Minor differences in the branch orderamong subgroups were observed. Clustering ofsubtypes I+II is recovered in both trees as is aspecific relationship between subtypes VI and VII,but while the latter pair emerges near the base ofour tree it is specifically related to I+II+V in the treeof Noel et al. (2005). Nevertheless, strong boot-strap support was observed for all subtype cladesand for two clusters that were identified by Noel etal. as of ‘uncertain classification’. For simplicity,we have named these clusters IVa and VIa toindicate their closest well-supported subtyperelationship in the tree of Noel et al. (2005) (Fig. 1).In conclusion, because the same clustering ofsequences into subtype clades was observed,sufficient phylogenetic signal clearly exists in the50 one-third of the gene to allow accurate subtypeattribution of new sequences.

One significant difference observed in ouranalysis was with an outlying pig isolate (GenBankAccession number: AF538348). This clustersstrongly with subtype V sequences in our treebut is basal to I+II in that of Noel et al. (Fig. 1).Further inspection of this sequence indicates thatit is likely to be a chimaera, as although the 50 endclusters with subtype V the 30 end clustersstrongly with subtype I (data not shown). Untilthe correct origin of this sequence is confirmed wehave designated it as subtype ‘X’.

DNA barcoding

To investigate the stability of the subtype cladeswhen a large number of new sequences are addedto the analysis, we sequenced the 50 one-third ofthe SSU-rDNA from Blastocystis isolates obtainedfrom two diagnostic laboratories. DNA was pur-ified from culture lysates and PCR products wereobtained using primers RD5 and BhrRDr. Productswere gel purified and sequenced with bothprimers to evaluate the relative efficacy andquality of the sequence produced from each of53 Blastocystis samples (Table 1A). Retrospec-tively it was found that a number of these hadbeen isolated from non-human primates; howeverno monkey-specific subtypes have been de-scribed to date. DNA barcoding protocols useonly a single primer for sequencing in order toreduce time and expense. Of the two primers weused the quality and reliability of sequencesobtained using BhRDr were consistently greaterthan those generated using RD5. Peaks werehigher and background was lower with BhRDrwhen traces for the two strands were comparedfor the same sample. Such differences between

primers are quite commonly observed. We there-fore recommend BhRDr for Blastocystis DNAbarcoding. Sequences obtained using either pri-mer cover all the polymorphic positions in theregion of the gene under study so that noinformation is lost by using only one primer forsequencing. An additional 20 sequences weregenerated using primer BhRDr alone (Table 1B).

One DNA sample was clearly derived from amixed infection containing two subtypes of Blas-tocystis and was excluded from further analysis;the sequence was unreadable in the polymorphicregions. For all other samples, once again assign-ment of strains to subtypes was straightforward.In some samples certain positions in the sequencewere unresolved in the traces with two basesappearing to be present in equal amounts, similarto what is seen when there is heterozygosity atsingle copy gene loci. This was equally true ofthose samples sequenced on both strands andthose sequenced on one. Two explanations arepossible—a coinfection with two variants withinthe same subtype could be present, or alterna-tively variation could exist between SSU-rDNAcopies within the same isolate. It is not possible atpresent to differentiate the two possibilities.

Phylogenetic analysis

All 72 new sequences were aligned with a subset of30 sequences selected to represent the full rangeof variation observed in the study by Noel et al.(Fig. 1). Phylogenetic analysis was performed usingBayesian, Maximum Likelihood and Distance ap-proaches. All of the subtype groupings and anumber of the minor clusters observed in Fig. 1were recovered with strong support in all threeanalyses (Fig. 2). A majority of the new sequencesbelonged to subtypes III and IV. All sequences wereable to be attributed with confidence to a pre-viously established specific subtype.

Discussion

It is clear that the complete SSU-rDNA sequence ofBlastocystis is not required for assignment of anisolate to its appropriate subtype. We believe thatthe amplicon described here when sequenced withprimer BhRDr will provide sufficient information foraccurate subtyping of new samples in a mannerequivalent to the DNA barcoding approach alreadydescribed for animal species. Although not formallydescribed as species, the subtypes of Blastocystisare OTUs with a substantial degree of divergence,

ARTICLE IN PRESS

Fig

ure

1.

Co

mp

ariso

nb

etw

een

phylo

genetic

trees

pro

duced

fro

mth

eco

mp

lete

Bla

sto

cys

tis

SS

U-r

DN

Aand

the

50o

ne-t

hird

of

the

gene.

The

do

tted

line

inp

anelA

(the

tree

pro

duced

fro

mth

e50o

ne-t

hird

ofth

eg

ene)id

entifies

the

majo

rto

po

log

icald

iffe

rence

betw

een

the

two

trees.T

he

tree

ofN

oele

tal.

(2005)(B

)w

as

red

raw

nand

dup

licate

seq

uences

rem

oved

but

oth

erw

ise

itis

identicalt

oth

eo

rig

inalfi

gure

.O

nly

the

GenB

ank

accessio

nnum

bers

are

giv

en

toid

entify

the

seq

uences.

Num

bers

ad

jacent

tono

des

rep

resent

the

Bayesia

np

oste

rio

rp

rob

ab

ilities

for

each

cla

de.O

nly

the

valu

es

for

sub

typ

ecla

des

and

oth

er

majo

rb

ranchp

oin

tsare

sho

wn.T

ho

se

seq

uences

sele

cte

dfo

ruse

insub

seq

uent

analy

ses

are

ind

icate

db

y�

ad

jacent

toth

eaccessio

nnum

ber.

The

bar

ind

icate

sth

enum

ber

of

sub

stitu

tio

ns

per

site.

80 S.M. Scicluna et al.

ARTICLE IN PRESS

Table 1. Origin of Blastocystis samples used for sequencing in this study. (A) Samples for which the 600 bpregion was sequenced on both strands. (B) Samples for which the 600 bp region was used for single primersequencing.

Identification no. Host Subtype Accession no.

(A)01/893 Woolly monkey I DQ23280001/905 Woolly monkey I DQ232807JJW16 Human I DQ23277701/676 Woolly monkey II DQ23280601/970 Human# II DQ23281402/309 Human II DQ23280502/521 Woolly monkey II DQ23279902/1024 Human II DQ23278205/387 Human II DQ23282401/898 Human II DQ23280202/532 Human II DQ232794JJW5 Human II DQ232775JJW18 Human II DQ23277802/027 Unidentified primate III DQ23278802/028 Unidentified primate III DQ23278902/029 Unidentified primate III DQ23279002/030 Unidentified primate III DQ23279102/033 Unidentified primate III DQ23279202/444 Stump tailed macaque III DQ23279702/1002 Woolly monkey III DQ23278502/682 Human# III DQ23282302/284 Human III DQ23280302/311 Human III DQ23279802/440 Human III DQ23280902/550 Human III DQ23280102/604 Human III DQ23281702/615 Human III DQ23280402/638 Human III DQ23281902/640 Human III DQ23282002/668 Human III DQ23282202/1086 Human III DQ23278005/266 Human III DQ23282505/386 Human III DQ23282702/796 Human III DQ23281002/039 Human III DQ23279302/826 Human III DQ23281102/1020 Human III DQ232784JJW19 Human III DQ23277902/919 Human IV DQ23278702/1017 Human IV DQ23278102/636 Human IV DQ23281805/379 Human IV DQ23282602/797 Human IV DQ23281202/813 Human IV DQ23281502/949 Human IV DQ23278602/951 Human IV DQ23281602/127 Human IV DQ23281302/825 Human IV DQ232808JJW6 Human IV DQ23277602/393 Woolly monkey IVa DQ232795

DNA Barcoding of Blastocystis 81

ARTICLE IN PRESS

Table 1. (continued )

Identification no. Host Subtype Accession no.

02/517 Woolly monkey IVa DQ23279602/1003 Woolly monkey IVa DQ23278302/642 Human VII DQ232821

(B)00/267 Human# I DQ23283200/469 Human I DQ23284300/396 Human II DQ23282900/1005 Unidentified primate II DQ23284500/117 Human III DQ23284400/299 Human# III DQ23283400/389 Human III DQ23284000/440 Human III DQ23283900/368 Human IV DQ23283700/472 Human IV DQ23284600/369 Human# IV DQ23283100/501 Human IV DQ23283500/400 Human IV DQ23283800/387 Human IV DQ23284100/266 Human# IVa DQ23283300/266A Human# IVa DQ23283000/392 Human IVa DQ23282800/1009 Unidentified primate IVa DQ232842Mabel Pig V DQ23283600/136A Human Mixed —

#Monkey handler; Unidentified primate ¼ species information not supplied. Isolates with the prefix JJW wereprovided by Jeffrey J. Windsor (Bronglais Hospital), the rest are from the Diagnostic Parasitology Laboratory(London School of Hygiene and Tropical Medicine).

82 S.M. Scicluna et al.

and it is likely that they will be elevated to specificstatus at some time in the future.

Phylogenetic analysis of the sort describedabove is not necessary for correct assignment ofsequences to subtypes. Certainly, less computa-tionally complex programmes produce the sameclusters seen in Figures 1 and 2. Indeed, in mostcases visual inspection of an alignment containingnew sequences and a mixture of sequences ofknown subtype will allow correct subtype identi-fication to be made. This greatly simplifies theprocess. However, it is likely that subtyping ofsome sequences by alignment alone will beambiguous; in this case phylogenetic analysismay well resolve the problem.

Although we have tested a limited number ofsamples to date, we have uncovered one clearcase of a mixed subtype infection. How prevalentthis situation will prove to be is at this stageunknown, but it appears based on the current datato be relatively infrequent. Unless our samples arenot representative of Blastocystis infectionsacross the world it is unlikely that mixed infections

would greatly hamper epidemiological investiga-tions. The need for cloning of PCR products priorto sequencing would eliminate the advantages intime and expense that DNA barcoding offers.However, if apparently new subtypes are identifiedin the future it will still be possible for the entiregene to be sequenced and a more completephylogenetic analysis undertaken.

The aim of this work was not to re-examine thephylogenetic relationships among Blastocystis butrather to develop a more rapid and accurate methodfor classifying new isolates into subtypes. Thesubtype identification method described here will,we believe, allow much larger epidemiological studiesto be undertaken in the future as it dramaticallyreduces the time and expense involved whencompared to methods in use at the present time.

Methods

DNA samples: Most Blastocystis culture lysateswere provided by the staff of the Diagnostic

ARTICLE IN PRESS

Figure 2. Unrooted tree showing relationships among 102 Blastocystis sequences. The numbers adjacent tobranchpoints indicate the Bayesian posterior probabilities and percentage of support for the clade based onbootstrap resampling in maximum likelihood and distance analyses (from left to right, respectively). Onlywhere all three analyses agreed on topology and support (greater than 50% in the case of maximumlikelihood and distance) are the numbers given. Those sequences selected from the data set of Noel et al areindicated by � to aid identification and the Roman numerals indicate the subtypes to which the sequencesare assigned.

DNA Barcoding of Blastocystis 83

Parasitology Laboratory, London School of Hy-giene and Tropical Medicine, while a few wereprovided by Jeffrey J. Windsor, Bronglais GeneralHospital, Aberystwyth, Wales. They all derivefrom stool samples sent for ova and parasitediagnosis (Table 1). Such samples are routinelyinoculated into Robinson’s medium at both in-stitutions and those cultures in which Blastocystis

grew were harvested by centrifugation and lysedin 0.25% SDS/0.1 M EDTA pH 8 before beingdiscontinued. DNA was purified by the modifiedcetyl trimethylammonium bromide (CTAB) extrac-tion method (Ali et al. 2005; Clark and Diamond1991). Briefly, lysates were treated with ProteinaseK before incubation with CTAB under high saltconditions. The CTAB—carbohydrate complex

ARTICLE IN PRESS

84 S.M. Scicluna et al.

was precipitated with chloroform and the aqueousphase extracted with phenol—chloroform—isoa-myl alcohol before precipitation of the DNA withethanol. After washing in 70% ethanol andresuspension in water, the DNA was passed overa Sephacryl S200 spin column (GE Healthcare,UK) before being used in PCR.

Primers and sequencing: The Blasto-cystis-specific primer BhRDr (GAGCTTTT-TAACTGCAACAACG) and the broad-specificityeukaryote-specific primer RD5 (ATCTGGTT-GATCCTGCCAGT; Clark 1997) were used in astandard PCR reaction with Taq DNA polymerase(BIOTAQ, Bioline, UK). Amplification conditionsconsisted of 30 cycles of 1 min each at 94, 59 and72 1C, with an additional 2 min final extension.PCR products were separated in 1.2% agarosegels and purified using the Qiaquicks gel extrac-tion kit (Qiagen) according to the manufacturer’sinstructions. Purified products were sequencedusing the amplification primer(s) and ABI BigDyeTM

sequencing kit version 3.1 (Applied Biosystems,Inc.), and analysed on an ABI 3730 sequencer.Sequences were deposited in GenBank with theaccession numbers DQ232775—DQ232846.

Phylogenetic analysis: Sequences down-loaded from GenBank as well as those generatedin this study were aligned using Multalin (Corpet1988) and the alignments edited manually (avail-able on request). Only the unique sequences fromthe study by Noel et al. were included.

Bayesian analysis used MrBayes 3.0 (Huelsen-beck and Ronquist 2001) with 4 MCMC strands,100,000 generations and an initial burn in of 1000.Four categories of among site rate variation wereused and trees were sampled every 10 genera-tions. In the reanalysis of the Noel data set,Proteromonas lacertae was used to root the tree.

Maximum likelihood and distance analysesused the phylogenetic software package PHYLIPv. 3.6 (Felsenstein 1989) and a model with 4categories of among site rate variation and theproportion of invariant sites. These parameters(a ¼ 0.16; pinvar ¼ 0.00) and the transition/trans-version ratio (1.10) were estimated using Tree-Puzzle 5.0 (Schmidt et al. 2002). The alignment filewas resampled 100 times (SEQBOOT) beforeanalysis using DNAML for maximum likelihoodanalysis. For distance analysis the alignment filewas resampled 1000 times before analysis usingDNADIST (F84 model) and NEIGHBOR. Themajority rule consensus trees were producedusing CONSENSE. Blastocystis subtype nomen-clature follows that of Yoshikawa et al. (2004)(also used by Noel et al. (2005)).

Acknowledgements

We thank John E. Williams and the staff of theDiagnostic Parasitology Laboratory (LondonSchool of Hygiene and Tropical Medicine) andJeff Windsor (Bronglais Hospital) for providing thesamples for analysis, and Konstantina Eleme andAlida Javier Molina for doing some of the DNAextractions. The majority of this work formed partof the thesis submitted as part of the M.Sc. inMedical Microbiology by SMS.

References

Abe N (2004) Molecular and phylogenetic analysis ofBlastocystis isolates from various hosts. Vet Para-sitol 120: 235—242

Ali IKM, Zaki M, Clark CG (2005) A genotypingmethod for Entamoeba histolytica based on PCRamplification of tRNA gene-linked short tandemrepeats. J Clin Microbiol 43: 5842—5847

Arisue N, Hashimoto T, Yoshikawa H (2003)Sequence heterogeneity of the small subunit ribo-somal RNA genes among Blastocystis isolates.Parasitology 126: 1—9

Blaxter ML (2004) The promise of a DNA taxonomy.Philos Trans R Soc London B 359: 669—679

Clark CG (1997) Extensive genetic diversity inBlastocystis hominis. Mol Biochem Parasitol 87:79—83

Clark CG, Diamond LS (1991) The Laredo strainand other Entamoeba histolytica—like amoebae areEntamoeba moshkovskii. Mol Biochem Parasitol 46:11—18

Corpet F (1988) Multiple sequence alignment withhierarchical clustering. Nucleic Acids Res 16:10881—10890

Felsenstein J (1989) PHYLIP—Phylogeny InferencePackage (Version 3.2). Cladistics 5: 164—166

Hebert PD, Cywinska A, Ball SL, de Waard JR(2003) Biological identifications through DNA bar-codes. Proc R Soc London B 270: 313—321

Hebert PD, Penton EH, Burns JM, Janzen DH,Hallwachs W (2004) Ten species in one: DNAbarcoding reveals cryptic species in the neotropicalskipper butterfly Astraptes fulgerator. Proc Natl AcadSci USA 101: 14812—14817

Hoevers J, Holman P, Logan K, Hommel M,Ashford R, Snowden K (2000) Restriction-frag-ment-length polymorphism analysis of small-subunit

ARTICLE IN PRESS

DNA Barcoding of Blastocystis 85

rRNA genes of Blastocystis hominis isolates fromgeographically diverse human hosts. Parasitol Res86: 57—61

Hogg ID, Hebert PDN (2004) Biological identifica-tion of springtails (Collembola: Hexapoda) from theCanadian Arctic, using mitochondrial DNA barcodes.Can J Zool 82: 749—754

Huelsenbeck JP, Ronquist F (2001) MRBAYES:Bayesian inference of phylogenetic trees. Bioinfor-matics 17: 754—755

Kress WJ, Wurdack KJ, Zimmer EA, Weigt LA,Janzen DH (2005) Use of DNA barcodes to identifyflowering plants. Proc Natl Acad Sci USA 102:8369—8374

Noel C, Dufernez F, Gerbod D, Edgcomb VP,Delgado-Viscogliosi P, Ho LC, Singh M, WintjensR, Sogin ML, Capron M, Pierce R, Zenner L,Viscogliosi E (2005) Molecular phylogenies ofBlastocystis isolates from different hosts: implica-tions for genetic diversity, identification of species,and zoonosis. J Clin Microbiol 43: 348—355

Rivera WL, Tan MA (2005) Molecular characteriza-tion of Blastocystis isolates in the Philippines byriboprinting. Parasitol Res 96: 253—257

Schmidt HA, Strimmer K, Vingron M, von Hae-seler A (2002) TREE-PUZZLE: maximum likelihoodphylogenetic analysis using quartets and parallelcomputing. Bioinformatics 18: 502—504

Silberman JD, Sogin ML, Leipe DD, Clark CG(1996) Human parasite finds taxonomic home.Nature 380: 398

Snowden K, Logan K, Blozinski C, Hoevers J,Holman P (2000) Restriction fragment-length poly-morphism analysis of small-subunit rRNA genes ofBlastocystis isolates from animal hosts. ParasitolRes 86: 62—66

Stenzel DJ, Boreham PFL (1996) Blastocystishominis revisited. Clin Microbiol Rev 9: 563—584

Thathaisong U, Worapong J, Mungthin M, Tan-Ariya P, Viputtigul K, Sudatis A, Noonai A,Leelayoova S (2003) Blastocystis isolates from apig and a horse are closely related to Blastocystishominis. J Clin Microbiol 41: 967—975

Yoshikawa H, Morimoto K, Wu Z, Singh M,Hashimoto T (2004) Problems in speciation in thegenus Blastocystis. Trends Parasitol 20: 251—255