Molecular & Biochemical Parasitology 130 (2003) 13–22

Identification of genotypically mixedCryptosporidium parvumpopulations in humans and calves

Sultan Tanriverdia,c, M. Özkan Arslanb, Donna E. Akiyoshic,Saul Tziporic, Giovanni Widmerc,∗

a Faculty of Medicine, Mustafa Kemal University, Antakya, Turkeyb Department of Parasitology, Faculty of Veterinary Medicine, Kafkas University, Kars, Turkey

c Division of Infectious Diseases, Tufts University School of Veterinary Medicine, 200 Westboro Road, North Grafton, MA 01536, USA

Received 10 March 2003; received in revised form 8 May 2003; accepted 9 May 2003

Abstract

Genotypic analyses ofCryptosporidium parvumoocysts have divided the species into two genotypes, referred to as type 1 and type 2.Although humans are susceptible to both types, mixed type 1/type 2 infections have rarely been identified. The paucity of mixed infectionscould be explained by the predominance of one type over the other in mixed infections, or by the poor sensitivity of restriction fragmentlength polymorphism (RFLP) analyses for detecting subpopulations. Using a type-specific real-time PCR assay capable of detecting type1 or type 2 constituting as little as 0.01% of the population, archived and new isolates of human, bovine, and mouse origin were genotyped.Mixed type 1/type 2 infections were identified in humans and calves, including in samples previously found to be homogeneous by RFLP.Isopycnic fractionation of mixed isolates revealed that type 1 and type 2 oocysts differ in their sedimentation properties. The detection of atype 1 subpopulation in serially-propagated bovine isolates indicates that type 1 and type 2 are stably maintained during long-term passage.Together with recently reported experimental bovine and ovine type 1 infections, the persistence of type 1 subpopulation in experimentallyinfected animals suggests that animals may play a previously unrecognized role in the maintenance ofC. parvumtype 1.© 2003 Elsevier B.V. All rights reserved.

Keywords: Cryptosporidium parvum; Cryptosporidiosis; Subpopulation; Mixed infection

1. Introduction

Various Cryptosporidiumspecies have been associatedwith human and animal infections, butCryptosporidiumparvumis the most frequent cause of human cryptosporid-iosis. C. parvum infects primarily immunocompromizedindividuals and children and causes chronic or acute di-arrhea. Several studies have investigated the genotypicdiversity within this species[1] and found that the vastmajority of human infections are caused by one of twogenotypes, varyingly designated as type 1/type 2, anthro-ponotic/zoonotic and recently also assigned to differentspecies,Cryptosporidium hominisandC. parvum[2].

It has been known for some time thatC. parvumtype 1 andtype 2 differ in their host range[3,4]. With two exceptions[5,6], in nature,C. parvumtype 1 has only been detected inhumans, whereas type 2 infects many mammalian species,

∗ Corresponding author. Tel.:+1-508-8397944; fax:+1-508-8397911.E-mail address:giovanni.widmer@tufts.edu (G. Widmer).

including ruminants and rodents. Differential host specificityis the only phenotypic property known to differentiate type1 from type 2. However, experimental infections of pigs,calves[7–10]and lambs[11] indicate that host specificity isnot absolute. Although type 1 animal infections have so farbeen restricted to the laboratory, these studies demonstratethat, under specific conditions, ruminants and pigs are sus-ceptible to type 1 infection, which raises the possibility ofa zoonotic cycle for type 1.

Although humans are susceptible to both types, surpris-ingly, only a few mixed infections have been reported. Forinstance, using restriction fragment length polymorphism(RFLP) markers, 1 mixed infection was reported among49 human cases from a cryptosporidiosis outbreak in theUnited Kingdom[12], and no mixed infections were re-ported among 50 human isolates from various outbreaksand sporadic cases[13]. In contrast, a recent survey usingmicro- and minisatellite length polymorphism found 12%type 1/type 2 mixed infections among 135 human cryp-tosporidosis cases from Scotland[14]. In addition to thelimitation of the RFLP method, an alternative explanation

0166-6851/$ – see front matter © 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0166-6851(03)00138-5

14 S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22

for the paucity of mixed infections in humans is a differen-tial rate of development of type 1 and type 2, leading to apredominance of one type over the other. This view is sup-ported by experimental infections of pigs with mixed type1/type 2 oocyst populations (Feng, Akiyoshi, Tzipori andWidmer, unpublished) and accidental type 2 contaminationsof calves infected with type 1[7].

A real-time PCR protocol capable of detecting smallC.parvum subpopulations was developed to investigate thepresence of type 1 and type 2 oocysts in samples originatingfrom various host species. These studies show that mixedinfections are more common than assumed on the basis ofRFLP analyses.

2. Materials and methods

2.1. Cryptosporidium samples

Table 1lists the geographic and host origin of theCryp-tosporidiumsamples. Except for aC. murisand aC. me-leagridis isolate used as negative controls, all samples were

Table 1Cryptosporidiumisolates

Isolate code Geographic origin Date Genotype Reference

Lib13 RFLP

Human isolatesUG 405-3035a (25 isolates) Uganda 1999–2001 1, 2 1, 2 [15]ECHIV1b Texas 1998 1 1PCHIVb Texas 1998 1 19897 Massachusetts 1998 1 1OH Ohio 1997 2 2Peru1 Peru 1998 2 2 [22]Peru2 Peru 1998 2 2 [22]JRLHIVb Texas 2 2GCH2b Conn. 1995 Mixed 1 [24]GCH3b Conn. 1995 Mixed Mixed [24]0676Ib 1997 2 2 [18]2064Db 1997 Mixed 1 [18]2458Lb 1997 1 1 [18]C. meleagridisTU1867 Uganda 2000 Negative Negative [26]

Calf isolatesGCH1 Lab isolate 4/1994 Mixed 2GCH1 Lab isolate 1/1995 Mixed 2GCH1 Lab isolate 3/1995 Mixed 2 [23]GCH1 Lab isolate 11/1998 2 2GCH1 Lab isolate 2/2003 Mixed 2IOWA Arizona 2 2MDc Lab isolate 1999 Mixed 2 [16]UCP Maine Mixed 2 [23]TK 8-220 (15 isolates) Turkey 2002 2 2

Other hostsTU502, pig Lab isolate 1999 1 1 [7]MDm, mouse Lab isolate 2000 2 2 [17]7C, mouse Lab isolate 2000 2 2 [17]C. muris, mouse Lab isolate Negative Negative

aHIV status unknown; 24 isolates are type 1 and 1 isolate is type 2.bHIV positive.

C. parvum. Twenty five isolates were randomly selectedfrom a collection of 533Cryptosporidiumpositive samplesoriginating from children from Uganda with diarrhea[15].Human samples from HIV positive individuals were kindlyprovided by Cynthia Chappell and Pablo Okhuysen, Uni-versity of Texas, Houston. Fifteen isolates were collectedfrom naturally infected calves from farms surroundingKars City in northeastern Turkey. Several isolates (GCH1,IOWA, MD, UCP, TU502, 7C,C. muris RN66 are labo-ratory maintained, as indicated inTable 1. The calf- andmouse-propagated lines of isolate MD[16,17]are indicatedas MDc and MDm, respectively. MDm was derived from asingle oocyst expanded in an immunodeficient mouse.

2.2. Oocyst purification and isopycnicNycodenz gradients

Oocysts were purified from stool as described[18].Briefly, stool samples were homogenized and the oocystsfloated on a saturated sodium chloride solution, fol-lowed by sedimentation on Percoll or a 15%/30% (w/v)Nycodenz (N,N′-bis(2,3-dihydroxypropyl)acetamido-2,4,

S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22 15

6-tri-iodo-isophthalamide) (Sigma, St. Louis, MO, USA)step gradient. The Nycodenz solutions were prepared inwater.

For fractionating type 1 and type 2 oocysts from mixedinfections, purified oocysts were loaded onto a 15–30% or15–25% linear Nycodenz gradient. Oocyst suspensions of0.5–1 ml were loaded on top of the gradient and the gra-dients centrifuged in a SW41 rotor (Beckman Instruments,Palo Alto, CA) for 1 h at 50,000×g and 10◦C. Fractions of0.5–1 ml were aspirated from the bottom of the tube usinga peristaltic pump. Fractions were numbered consecutivelystarting with the bottom fraction. Fractions were diluted witha minimum of 0.5 ml water, the oocysts pelleted by centrifu-gation (2500× g, 5 min) and resuspended in 1 ml of water.

2.3. DNA isolation

Purified oocysts were counted with a hemocytometer. Fortesting the sensitivity of the Lib13 assays, each isolate wasdiluted to 106 oocysts per 100�l and DNA was extractedwith the High Pure Template Preparation Kit (Roche Diag-nostics) as previously described[19]. DNA was eluted in20�l TE buffer, which is equivalent to 5× 104 oocysts/�l.Different quantities of DNA containing between 5 and104 oocyst equivalents of type 1 (TU502) and type 2(MDm) were prepared by serial dilution. Archived DNAsamples were obtained as previously described[18] us-ing three cycles of freeze-thawing followed by proteinaseK (200�g/ml)—SDS (1%) digestion of the lysate andphenol–chloroform extraction.

2.4. PCR

The type-specific Lib13 PCR assay is based on a pre-viously described sequence polymorphism located in aC. parvum genomic sequence of unknown coding func-tion (GenBank accession number AF190627)[9]. A sin-gle upstream primer Lib13SF (5′-acctgattttgagttc-3′) waspaired with one of two downstream primers, either a type1-specific primer Lib13SRT-1 (5′-atttattaatttatctcttactt-3′)or the homologous type 2-specific primer Lib13SRT-2(5′-atttattaatttatctcttcg-3′) (Fig. 1). The length of the am-plicon amplified by these primers is 160 bp for type 1 and156 bp for type 2. The 4-bp difference in length is due

Fig. 1. Lib13 PCR assay. The upstream primer Lib13SF is common toC. parvumtype 1 and type 2. The specificity of the reaction for type 1 and type2 is based on a 4-bp deletion in type 2. Downstream primers Lib13SRT-1 and Lib13SRT-2 are specific for the corresponding type. Primer sequencesare shown in capital and boldface. Horizontal arrows show primer orientation. Dashes in type 2 sequence indicate deletion. Numbering according toGenBank accession number AF190627 (type 1) and type 2 sequence from genomic coding 1952 from the National Center for Biotechnology genomicdatabase (www.ncbi.nlm.nih.gov).

to a deletion in the type 2 sequence (Fig. 1) spanned bythe type-specific downstream primers. Amplifications wereperformed in a LightCycler (Roche Diagnostics, IN, USA)using FastStart SYBR Green I master mix (Roche Diagnos-tics). Initial denaturation was performed at 95◦C for 10 min.The denaturation step was followed by 40 cycles of 95◦Cfor 1 s, 55◦C (Lib13SF/SRT1) or 58◦C (Lib13SF/SRT2)for 5 s, and 72◦C for 15 s. Following amplification, meltingcurve analysis was performed as follows: the PCR reactionswere denatured at 95◦C for 1 s, annealed at 55◦C for 15 sand re-heated to 90◦C at a rate of 0.1◦C/s with continuousfluorescence monitoring at 530 nm (F1 channel). Crossingpoints were determined using the LightCycler software.The crossing point, or threshold cycle (Ct) is defined asthe number of temperature cycles required for generatinga specific level of fluorescence. This value is a functionof the initial template concentration, since low templateconcentrations require more temperature cycles to reach aspecific level of amplification.Ct values can thus be usedas a measure of the initial template concentration, where alower Ct value indicates a higher concentration.

2.5. RFLP analysis

Three different RFLP assays specific for the polyT[20],COWP [21] and �-tubulin locus [22] were performedas previously described. Briefly, PCR products were di-gested with restriction enzymeRsaI in PCR buffer (10 mMTris–HCl [pH 9], 50 mM KCl, 0.1% Triton X-100, 2 mMMgCl2) for 90 min at 37◦C. The 428-bp fragment of theC. parvum �-tubulin gene (GenBank accession numberY12615) was digested with restriction enzymeTsp509I for60 min at 65◦C, also in PCR buffer.

3. Results

3.1. Type-specific Lib13 PCR assay

To assess the sensitivity and specificity of the Lib13 PCRassays, and the ability of each primer pair to amplify aC.parvum type 1 or type 2 subpopulation in the presence ofexcess DNA of type 2 or type 1, constant amounts of type2 DNA, equivalent to 5× 104 oocysts (isolate MDm), were

16 S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22

mixed in 1-, 10-, 50-, 100-, 1000-, and 10,000-fold excessof type 1 DNA (isolate TU502) and vice versa. PCR am-plifications using the type 1-specific (Lib13SF/Lib13SRT-1)and type 2-specific (Lib13SF/Lib13SRT-2) primer combi-nations were performed as described above and the ampli-

Fig. 2. Sensitivity and specificity of Lib13 type 1 PCR assay. DNA samples consisting of experimental mixtures of DNA extracted from isolate TU502(type 1) and MD (type 2) were amplified using the Lib13 primer pair specific for type 1 (Lib13SF/Lib13SRT-1). Amplification (A) and melting curves(B) were monitored in real-time using SYBR Green I fluorescence. DNA extracted from 5000 MDm oocysts was used as a control to demonstrate thespecificity of the assay for type 1. Note the absence of diagnostic melting peaks in the water (negative) and MDm controls. Analogous results wereobtained with the type 2-specific assay (not shown).

fication reactions monitored by measuring SYBR Green Ifluorescence (Fig. 2A). The identity of each amplicon wasconfirmed by melting curve analysis (Fig. 2B). Agarose gelanalysis was performed on the same samples to confirmthe real-time PCR data (Fig. 3). Type 1 and type 2 DNA

S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22 17

Fig. 3. Sensitivity of Lib13 primer pairs. Mixtures of type 1 (isolate TU502) and type 2 (isolate MD) DNA containing the amounts of DNA shown (inoocyst equivalents) above each gel were amplified with primer pairs Lib13SF/Lib13SRT-1 (upper gel) and Lib13SF/Lib13SRT-2 (lower gel). Horizontalarrows indicate a constant number (5× 104) of oocysts of isolate MDm (top) and TU502 (bottom). Note that the Lib13 assays detect type 1 and type 2DNA in the presence of a 10,000-fold excess of the other genotype.

were detected when present at ratios as low as 1:10,000.Lower dilutions were not attempted, but the relativelow level of amplification observed at 1:10,000 (fivetype 1 oocysts in 50,000 type 2 oocysts and vice versa)suggests that the limit of the assay had been reached. Melt-ing curves showed that the Lib13 amplicon generated amelting peak at 77–78◦C. Results from melting curve andgel analysis showed a complete agreement; the presenceof a melting peak at the diagnostic temperature invariablycorrelated to the presence of an electrophoretic band, andelectrophoretic bands of the expected size (156/160 bp)were only seen in samples which were positive by melt-ing curve analysis. Consistent with the 4-bp deletion inthe type 2 Lib13 sequence and a c–t mutation (Fig. 1),Tm values for the type 2 amplicon were lower than thosefor type 1. The mean temperature difference was 0.35◦C(77.30◦C for type 2 versus 77.65◦C for type 1). Thedifference was statistically highly significant (t-test, P <

0.001, n = 66). Importantly, attempts to amplify type1 DNA from TU502 with the heterologous primer pairLib13SF/Lib13SRT-2 and mouse-propagated MDm withLib13SF/Lib13SRT-1 were unsuccessful (Figs. 2 and 3),confirming the complete absence of cross-amplificationeven in the presence of high concentrations of heterologousDNA.

When the crossing point from the serial dilution exper-iments was plotted against the log of the relative DNAconcentration of the subpopulation, a linear relationshipwas obtained (R = 0.994; n = 4). The strong correlationbetween these variables indicates that the amplification of

the subpopulation is not inhibited by excess DNA of theother genotype. This finding also shows that in this assaycrossing point values are a valid measure of the relativeabundance of a subpopulation.

3.2. Analysis of various C. parvum isolates using theLib13 assay

The Lib13 PCR assay was used to genotype DNA samplesfromC. parvumoocysts originating from humans and calves.A total of 38 isolates from humans, 19 isolates from calves,1 from a pig, and 3 from mice were analyzed (Table 1). Forall human isolates previously classified as type 1 by RFLP,the genotype was confirmed. However, 2 isolates previouslyclassified as type 1, GCH2[20] and 2064D[22], displayed amixed type 1/type 2 Lib13 genotype (Fig. 4). The genotypeof GCH3, a rare case of mixed human infection, was con-firmed by Lib13 assay. Of the 19 isolates from calves, allpreviously classified as type 2 by RFLP, 3 isolates showedtraces of type 1 (Fig. 4), including GCH1, UCP[23], andMDc [16]. Free of type 1 were two type 2 isolates derivedfrom single oocysts, and maintained in mice (MDm and7C [17]). The absence of type 1 DNA in type 2 lines de-rived from single oocysts is a further confirmation that thetype 1 amplicon did not originate from cross-amplificationwith the type 2-specific assay. Moreover, DNA extractedfrom C. muris (isolate RN66) andC. meleagridis(isolateTU1867) failed to amplify with both Lib13 primer pairs,indicating that this PCR assay is specific forC. parvum(not shown).

18 S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22

Fig. 4. Genotypic analysis of human-, mouse-, pig-, and calf-propagated isolates reveals mixed infections. The type 1- and type 2-specific Lib13 PCRassays were applied in parallel to the isolates shown (seeTable 1 for details). These results were interpreted as follows: TU502 (pig), type 1; 2458L(human, HIV positive), type 1; GCH2 and GCH3 (human, HIV positive) type 1/type 2 mixed; 9897 (human), type 1; 2064D (human, HIV positive) type1/type 2 mixed; MDm (mouse), type 2; 7C (mouse), type 2; UCP (calf), type 1/type 2 mixed; GCH1 1/14/98, type 2; 1/6/95, 3/24/95, type 1/type 2mixed. Codes shown in italics are for isolates previously classified as type 2 using RFLP assays, regular font indicates type 1.

3.3. Fractionation of a type 1/type 2 mixed oocystpopulation from calf-propagated isolate GCH1

To confirm that type 1 and type 2 DNA detected by theLib13 PCR assay originates from different oocyst popula-tions, 6×107 purified oocysts of isolate GCH1 propagated ina calf during February, 2003, were fractionated on a 15–25%(w/v) linear Nycodenz gradient. Oocyst DNA was extractedfrom each of 14 fractions spanning the entire gradient. Formost fractions between 1 and 3× 104 oocysts, equivalentto about 1% of each fraction, were extracted. In fractions1–4, where relatively few oocysts sedimented, between 2400and 5000 oocysts were used for DNA extraction. Extractedoocyst DNA was then analyzed using both Lib13 PCR as-says (Fig. 5). With the exception of fractions 2, 6 and 7, the8 fractions from the bottom of the gradient were positive forboth type 1 and type 2. In fractions 9–13 no type 1 DNAwas detected. The topmost fraction (fraction 14) also ampli-fied with the type 1-specific assay. As expected for a bovineisolate, type 2 DNA was detected in all fractions, althoughamplification from fraction 2 did require over 36 cycles togenerate a visible melting curve, consistent with the low

oocyst concentration in this fraction. Confirming prior con-trol reactions with isolates TU502, MDm, and 7C, the ab-sence of type 1 DNA from the upper half of the gradientconfirmed the specificity of the PCR assay, and indicates thatGCH1 includes both type 1 and type 2 oocysts which sed-iment at slightly different densities. Crossing points of thetype 1 amplification reactions were higher than those of type2, the difference being typically 6–10 cycles, confirming thattype 2 oocysts constitute the vast majority of the parasitepopulation in this isolate. The presence of a type 1 signalwas also unrelated to the abundance of oocysts in each frac-tion or oocyst equivalents of DNA used in a reaction. Type1 positive and negative fractions included those with high(5–13) and low (1–4, 14) oocyst concentrations (Fig. 5A).

The differential sedimentation of type 1 and type 2oocysts on isopycnic gradients was confirmed using threeRFLP assays specific for the polyT[20], COWP[21] and�-tubulin locus[22]. In the present configuration, the polyTand COWP RFLP markers each contained one polymorphicRsaI site, and lacked a conserved restriction site. By usingboth markers to type oocysts from each gradient fraction,the possibility that inhibition ofRsaI activity could have

S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22 19

Fig. 5. Fractionation of oocysts of bovine isolate GCH1 reveals type 1 and type 2 oocysts. (A) Distribution of oocysts in a 15–25% continuous densitygradient. Genotypically mixed fractions (black bars) were predominant in the bottom half of the gradient. Grey bars show fractions in which only type2was detected. (B) Melting curve analysis of type 1-specific Lib13 assay. Continuous lines, fractions; dashed line, negative control MDm DNA. (C) Meltingcurve analysis of type 2-specific Lib13 assay. Lines as in panel B; negative control, TU502 DNA. Notice the complete absence of cross-amplification inMD and TU502 control samples. Genotypically mixed fraction are shown with black regular font; grey italicized numbers indicate fraction with type 2 only.

generated the restriction profiles observed in different re-gions of the gradient was excluded. Whereas the polyTamplicon of type 1 lacks theRsaI site, with the COWPmarker the pattern is reverse; type 1 is cleaved byRsaI, andtype 2 lacks the restriction site. Using both assays in paral-lel, results similar to those observed with the Lib13 assay(Fig. 6A and B) were obtained. As found with the Lib13method, type 1 oocysts were detected in the lower half ofthe gradient, as shown by full-length polyT amplicon andRsaI digested COWP amplicon.

To confirm the differential sedimentation of type 1 andtype 2 oocysts, 107 oocysts each of isolates GCH1 (type 2)

and 2458L (type 1) were mixed and fractionated on a con-tinuous Nycodenz gradient (15–30% in water). Oocyst DNAwas extracted from each of 12 fractions and typed using thepolyT RFLP assay. As illustrated inFig. 6C, type 1 and type2 oocysts sedimented in different regions of the gradient,although the distribution was overlapping. Whereas type 2oocysts were found primarily in the upper half of the gradi-ent, type 1 oocysts were more abundant towards the bottom.As seen with the Lib13 gradient experiment, the uppermostfraction contained a large proportion of type 1 oocysts. Toexclude that this observation may be isolate specific, an anal-ogous experiment was performed with a mixture of GCH1

20 S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22

Fig. 6. RFLP analysis of fractionated GCH1 oocysts. DNA extracted from the fractions shown above each gel image were typed using (A) the COWP[21] and (B) the polyT RFLP assay[20]. Both assays are based on a polymorphicRsaI site; the type 1 polyT amplicon remains uncleaved, whereas thetype 2 COWP amplicon lacks the restriction site. (C) Differential sedimentation of type 1 and type 2 oocysts from an experimental 1:1 mixture of type1 and type 2 oocysts revealed using the polyT RFLP assay. As seen with the Lib13 assay, type 1 oocysts sedimented towards the bottom of the gradientand in the top fraction (fraction 11). Unrestricted control polyT amplicon (−); RsaI restricted control polyT amplicon (+).

and HS6 (human, type 1) oocysts and each fraction typedusing aTsp509I polymorphism located in the�-tubulin gene[22]. Consistent with the previous experiment, type 1 andtype 2 oocysts sedimented preferentially to the bottom andtop half of the gradient, respectively (not shown). Togetherthese results confirm the presence of type 1 DNA in iso-late GCH1 and indicate the presence of type 1 and type 2oocysts in this isolate.

4. Discussion

The discovery that human cryptosporidiosis is caused bytwo genetically distinctC. parvumtypes[4,13,20,24]led tothe model of a zoonotic cycle ofC. parvumtype 2, and ananthroponotic cycle of type 1. This view was consistent withepidemiological studies showing that infected individualsexcrete type 1 or type 2 oocysts[6,13], whereas animalsexcrete only type 2.

Experimental infections of pigs, calves, and lambs withtype 1C. parvum[7,9,11]demonstrate that, given the rightconditions, type 1 is capable of infecting animals. This find-ing suggests that host specificity is not simply a function ofcompatible receptor–ligand interaction between host cell andparasite, but should be viewed as a spectrum in host–parasitecompatibility. According to this view, a low degree of com-patibility between type 1 and animal hosts can be overcomeby increasing the dose. Thus, in experimental infectionswhich typically use large numbers of oocysts, poor infec-tivity can be overcome by increasing the oocyst dose. Forinstance, the infection of a lamb with type 1 was achievedwith a dose of 5× 105 oocysts[11], whereas in this fa-cility, calves are typically infected with 108 type 1 oocysts(Akiyoshi, personal communication).

Although the relative differences in infectivity of type1 and type 2 may explain the absence of exclusively type1 infections in livestock, this model does not address thequestion of type 1 survival in nature. If type 1 was solely

S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22 21

dependent on human-to-human transmission, it is difficultto envision how this infection is maintained, particularly indeveloped nations where waterborne transmission is rela-tively infrequent. In contrast, where direct or waterbornetransmission is facilitated by poor sanitary conditions andcrowding, pure type 1 infections are to be expected, consis-tent with the analysis of human samples from Uganda[15].In areas of low transmission, the presence of type 1 subpop-ulations may indicate an alternative route of maintenance, azoonotic cycle where type 1 and type 2 are co-transmitted.An assumption underlying this hypothesis is that type 1transmission is facilitated by the co-infection with type 2since, given the low proportion of type 1 in mixed popu-lations, the number of ingested type 1 oocysts would bemuch smaller than the expected minimal infectious dose.A recent report of mixed type 1/type 2 human infections isconsistent with the assumption of relatively frequent type1/type 2 co-infections[14].

Because of the exponential nature of PCR, detection ofgenotypically distinct subpopulation is technically difficult.This is particularly the case with RFLP assays which rely on1 pair of primers to amplify two or more alleles. Althoughan RFLP assay targeted at the TRAP-C1 locus was reportedto detect a 10,000:1 ratio of type 2:type 1[7], sensitivities ofother RFLP methods for detecting mixed populations are inthe 10–25% range[25] (Widmer, unpublished observations).The sensitivity of commonly usedC. parvumRFLP assaysfor detecting subpopulations has not been reported. It canthus be assumed that most genotypic surveys ofC. parvumusing RFLP would not have detected subpopulations con-stituting less than 10% of the parasite population. Togetherwith the tendency of type 2 to outgrow type 1[7], this mayexplain the low number of mixedC. parvuminfections re-ported in the literature.

The origin of type 1 parasites in GCH1 is uncertain. Thisisolate was obtained in 1991 from an HIV positive indi-vidual [23] and has since been maintained by experimentaltransmission in calves. The genotype present in the originalhuman inoculum is unknown. Also unknown is whethertype 1C. parvumpresent in this isolate originates from theoriginal human infection, or was accidentally introducedduring propagation. Because calves used for propagationare brought into the facility from neighboring farms, thepossibility that exogenous oocysts were introduced intothis isolate is quite high. Genetic heterogeneity within thisisolate has been observed previously using a RFLP assayspecific for the�-tubulin locus[22]. The intermittent pres-ence of type 1 in GCH1, could indicate that the proportionof type 1 is near the limit of detection, or that type 1 isperiodically lost and re-acquired. The former possibility issupported by the fact that replicate Lib13 analyses of indi-vidual GCH1 samples did not consistently detect type 1. Forinstance in five replicate analyses of a GCH1 sample from1995, three were positive for type 1, but two failed to detectthis subpopulation. The rates of type 1 detection for otherGCH1 samples were 2/3, 4/5 and 3/4. In contrast replicate

analyses of human type 1 samples found to harbor type 2subpopulation were consistent, 3/3, 2/2 and 2/2 positivesfor GCH2, GCH3 and 2064D, respectively. The observationthat type 1 was also present in a second calf-propagatedisolate (UCP), indicates that mixed isolates are not lim-ited to those originating from humans, and that type 1found in GCH1 may be unrelated to the original humanisolate.

What are the possibilities that the mixed populationsresulted from laboratory contaminations? Contaminationcould occur at three levels: DNA contamination, contami-nation of oocyst stocks with exogenous oocysts, and iso-late contamination during propagation. The possibility ofDNA contamination can be excluded based on three ob-servations: (1) negative control reactions (no DNA) wereincluded throughout, (2) mouse-propagated isolates origi-nating from single type 2 oocysts were consistently free oftype 1 DNA, (3) the distribution of type 1 DNA in densitygradients demonstrates that type 1 DNA was enclosed inoocysts and excludes the possibility of naked DNA havingaccidentally been introduced. The first two arguments alsoargue against contamination of oocysts stocks. With respectto contamination of isolates during propagation, we noticethat several isolates which were never propagated experi-mentally (2064D, GCH2, GCH3) were mixed. The fact thattype 1 and type 2 isolates are processed in different facili-ties and using dedicated equipment also argues against thepossibility of accidental cross-contamination of isolates. Itshould also be emphasized that mixed GCH1 populationswere identified in samples collected over a period of 10years (seeTable 1), indicating that the presence of type 1in this isolate is not restricted to a specific experimentalinfection.

The biochemical and structural differences causing type1 and type 2 oocysts to sediment at different densities areunknown. Because of the large overlap in the distributionof both oocyst types, the fractionation on Nycodenz willnot allow the recovery from the same isolate of homoge-neous populations for biochemical studies. Measurementsof light scatter properties in individual fractions were per-formed by flow cytometry, but no consistent trends wereobserved. Because type 1 and type 2 oocysts collected fromthe same host display different sedimentation properties,sedimentation property appears to be an intrinsic propertyof the oocysts, and is not a consequence of different pu-rification procedures, storage conditions, or host species.Interesting is the presence of a large proportion of type 1 inthe top layer of certain gradients. Microscopic examinationrevealed large numbers of empty oocysts or oocysts whichappeared to be damaged in these fractions. Consistent withthe fast rate of type 1 oocysts decay observed with certainisolates [9], this indicates that type 1 oocysts are morerapidly inactivated during storage. More detailed studiesof the genotypic composition of mixed isolates will shedlight on the population dynamics ofC. parvum, and helpelucidate the role of zoonotic transmission ofC. parvum.

22 S. Tanriverdi et al. / Molecular & Biochemical Parasitology 130 (2003) 13–22

Acknowledgements

We thank Glenda Batzer and Kristina Ciociola for tech-nical assistance, Hal Stibbs forC. muris oocysts, andXiaochuan Feng for advice with the Lib13 assay. Finan-cial support from the NIH (Grant AI052781 and ContractN01AI25466) and the USDA (Grant 98351026757) isgratefully acknowledged.

References

[1] Widmer G. Genetic heterogeneity and PCR detection ofCryp-tosporidium parvum. Adv Parasitol 1998;40:223–39.

[2] Morgan-Ryan UM, Fall A, Ward LA, et al.Cryptosporidium homi-nis n. sp. (Apicomplexa: Cryptosporidiidae) fromHomo sapiens. JEukaryot Microbiol 2002;49:433–40.

[3] Awad-El-Kariem FM, Robinson HA, Petry F, McDonald V, EvansD, Casemore D. Differentiation between human and animal isolatesof Cryptosporidium parvumusing molecular and biological markers.Parasitol Res 1998;84:297–301.

[4] Peng MM, Xiao L, Freeman AR, et al. Genetic polymorphism amongCryptosporidium parvumisolates: evidence of two distinct humantransmission cycles. Emerg Infect Dis 1997;3:567–73.

[5] Morgan-Ryan UM, Xiao L, Hill BD, et al. Detection of theCryp-tosporidium parvum“human” genotype in a dugong (Dugong dugon).J Parasitol 2000;86:1352–4.

[6] Spano F, Putignani L, Crisanti A, et al. Multilocus genotypic anal-ysis of Cryptosporidium parvumisolates from different hosts andgeographical origins. J Clin Microbiol 1998;36:3255–9.

[7] Akiyoshi DE, Feng X, Buckholt MA, Widmer G, Tzipori S. Geneticanalysis of aCryptosporidium parvumhuman genotype 1 isolatepassaged through different host species. Infect Immunol 2002;70:5670–5.

[8] Pereira SJ, Ramirez NE, Xiao L, Ward LA. Pathogenesis of humanand bovineCryptosporidium parvumin gnotobiotic pigs. J InfectDis 2002;186:715–8.

[9] Widmer G, Akiyoshi D, Buckholt MA, et al. Animal propagationand genomic survey of a genotype 1 isolate ofCryptosporidiumparvum. Mol Biochem Parasitol 2000;108:1878–97.

[10] Widmer G, Akiyoshi D, Buckholt MA, et al. Animal propagationand genomic survey of a genotype 1 isolate ofCryptosporidiumparvum. Mol Biochem Parasitol 2000;108:187–97.

[11] Giles M, Webster KA, Marshall JA, Catchpole J, Goddard TM.Experimental infection of a lamb withCryptosporidium parvumgenotype 1. Vet Rec 2001;149:523–5.

[12] McLauchlin J, Pedraza-Diaz S, Amar-Hoetzeneder C, Nichols GL.Genetic characterization ofCryptosporidiumstrains from 218 patients

with diarrhea diagnosed as having sporadic cryptosporidiosis. J ClinMicrobiol 1999;37:3153–8.

[13] Sulaiman IM, Xiao L, Yang C, et al. Differentiating human fromanimal isolates ofCryptosporidium parvum. Emerg Infect Dis1998;4:681–5.

[14] Mallon M, MacLeod A, Wastling J, Smith H, Reilly B, Tait A.Population structures and the role of genetic exchange in the zoonoticpathogenCryptosporidium parvum. J Mol Evol 2003;56:407–17.

[15] Tumwine JK, Nabukeera N, Akiyoshi DE, Rich SM, Widmer G,Feng X, et al.Cryptosporidium parvumin children with diarrhea inMulago Hospital, Uganda. Am J Trop Med Hygiene 2003, in press.

[16] Okhuysen PC, Rich SM, Chappell CL, et al. Infectivity of aCryp-tosporidium parvumisolate of cervine origin for healthy adults andinterferon-gamma knockout mice. J Infect Dis 2002;185:1320–5.

[17] Feng X, Rich SM, Tzipori S, Widmer G. Experimental evidence forgenetic recombination in the opportunistic pathogenCryptosporidiumparvum. Mol Biochem Parasitol 2002;119:55–62.

[18] Widmer G, Tzipori S, Fichtenbaum CJ, Griffiths JK. Genotypic andphenotypic characterization ofCryptosporidium parvumisolates frompeople with AIDS. J Infect Dis 1998;178:834–40.

[19] Tanriverdi S, Tanyeli A, Baslamisli F, et al. Detection and genotypingof oocysts ofCryptosporidium parvumby real-time PCR and meltingcurve analysis. J Clin Microbiol 2002;40:3237–44.

[20] Carraway M, Tzipori S, Widmer G. A new restriction fragmentlength polymorphism fromCryptosporidium parvumidentifies ge-netically heterogeneous parasite populations and genotypic changesfollowing transmission from bovine to human hosts. Infect Immunol1997;65:3958–60.

[21] Spano F, Putignani L, McLauchlin J, Casemore DP, Crisanti A.PCR-RFLP analysis of theCryptosporidium oocyst wall protein(COWP) gene discriminates betweenC. wrairi and C. parvum, andbetweenC. parvum isolates of human and animal origin. FEMSMicrobiol Lett 1997;150:209–17.

[22] Widmer G, Tchack L, Chappell CL, Tzipori S. Sequence polymor-phism in the beta-tubulin gene reveals heterogeneous and variablepopulation structures inCryptosporidium parvum. Appl Environ Mi-crobiol 1998;64:4477–81.

[23] Tzipori S, Rand W, Griffiths J, Widmer G, Crabb J. Evaluation of ananimal model system for cryptosporidiosis: therapeutic efficacy ofparomomycin and hyperimmune bovine colostrum-immunoglobulin.Clin Diagn Lab Immunol 1994;1:450–63.

[24] Carraway M, Tzipori S, Widmer G. Identification of genetic het-erogeneity in theCryptosporidium parvumribosomal repeat. ApplEnviron Microbiol 1996;62:712–6.

[25] Reed C, Sturbaum GD, Hoover PJ, Sterling CR.Cryptosporidiumparvummixed genotypes detected by PCR-restriction fragment lengthpolymorphism analysis. Appl Environ Microbiol 2002;68:427–9.

[26] Akiyoshi DE, Dilo J, Pearson C, Chapman S, Tumwine J, Tzi-pori S. Characterization ofCryptosporidium meleagridisof hu-man origin passaged through different host species. Infect Immunol2003;71:1828–32.