|Home | About | Journals | Submit | Contact Us | Français|
Background. Giardia lamblia is ubiquitous in multiple communities of nonindustrialized nations. Genotypes A1, A2, and B (Nash groups 1, 2, and 3, respectively) are found in humans, whereas genotypes C and D are typically found in dogs. However, genotypes A and B have occasionally been identified in dogs.
Methods. Fecal Giardia isolates from 22 families and their dogs, living in Pampas de San Juan, were collected over 7 weeks in 2002 and 6 weeks in 2003. Samples were genotyped, followed by sequencing and haplotyping of many of these isolates by using loci on chromosomes 3 and 5.
Results. Human infections were all caused by isolates of genotypes A2 and B. Human coinfections with genotypes A2 and B were common, and the reassortment pattern of different subtypes of A2 isolates supports prior observations that suggested recombination among genotype A2 isolates. All dogs had genotypes C and/or D, with one exception of a dog with a mixed B/D genotype infection.
Conclusions. In a region of high endemicity where infected dogs and humans constantly commingle, different genotypes of Giardia are almost always found in dogs and humans, suggesting that zoonotic transmission is very uncommon.
Giardia lamblia (also known as Giardia intestinalis, Giardia duodenalis) is an intestinal protozoan parasite that is ubiquitous in mammals. It has 2 stages, a trophozoite and cyst, the latter being responsible for transmission. Pampas de San Juan de Miraflores is a periurban shantytown outside of Lima, Peru, where G. lamblia is endemic. The majority of children in this shantytown are infected with this protozoan parasite by the age of 3 years .
Molecular analysis of G. lambliafield isolates has been critical to better trace this parasite within regions of endemicity and in outbreak situations. G. lamblia has been divided into genotypes A through G on the basis of DNA sequence differences that reflect differences in host specificity and biology. The “ human” genotypes were originally designated as groups 1, 2, and 3 [2, 3] (genotypes A1, A2, and B, respectively). Genotypes A1 and A2 are ~98%–99% identical in the regions that have been sequenced, whereas genotypes A and B are so different as to likely represent different species [2, 3]. Genotypes C and D are typically found in dogs, whereas genotypes E-G are found in other mammals.
Molecular typing of G. lamblia samples can answer questions for tracing the epidemiology of the parasite in humans and can also be useful in resolving the question of zoonotic transmission. Although genotypes A1, A2, and B are found primarily in humans, they have occasionally been found in other mammals, including beavers, cats, and dogs. Specifically, genotype A [4–7] and genotype B  have been reported in dogs. However, in other studies, predominantly genotypes C or D [9, 10] (one dog in this study had genotype E by small subunit-recombinant DNA [SSU-rDNA]) or only genotypes C or D  have been found in dogs. For this reason, controversy continues regarding the possibility and frequency of zoonotic transmission of G. lamblia, particularly among dogs [12, 13]. In particular, epidemiologic studies of giardiasis in dogs have yielded conflicting results regarding the presence of isolates in dogs that are capable of infecting humans.
Most previous molecular epidemiologic studies have been performed with the assumption of asexual clonal reproduction [14, 15]. If propagation is clonal, then comparison of any 2 loci should yield similar results. However, recent data from this area of Peru with high endemicity have provided evidence for recombination among genotype A2 isolates . Other recent studies have also yielded different inheritance patterns for different loci [4, 17], consistent with the possibility of recombination among genotype A2 isolates.
We have used genotyping of G. lamblia from fecal isolates obtained from humans and dogs that are in close proximity to each other to evaluate the potential for transmission between humans and dogs. In addition, we have used sequence analysis at 2 loci on different chromosomes to refine our investigation of household transmission patterns and to detect recombinant types among the genotype A2 isolates.
Study site. Pampas de San Juan de Miraflores is a desert shantytown 25 km to the south of Lima, Peru. The last census in April 1997 recorded the population at 38,721 [18–20]. Inhabitants are mainly mestizo migrants from the Peruvian Andes Mountains who earn a living by performing unregulated day labor . Details about the demographics and environment of this shantytown have been described elsewhere [18–20].
Subjects. Families were previously enrolled in a longitudinal cohort study, and 1 child per family was designated as the index case. The inclusion criteria were age 10 years and residence in Pampas de San Juan de Miraflores. Exclusion criteria included diagnosis of moderate or severe malnutrition according to the World Health Organization guidelines (2 Z -score) and participation in any other study involving treatment of or vaccination against parasitic disease. Children from the longitudinal cohort who tested positive for G. lamblia by stool microscopy during June-August 2002 and May-July 2003 and all other people living in the household were invited to participate in the current study. All adult patients enrolled in the study signed written consent forms, and parents signed written consent forms for children and infants. The Institutional Review Board of PRISMA (a Peruvian nongovernment organization) and the Human Subjects Approval Committee at the University of Arizona approved the protocol used for the study. Samples for this study were collected at biweekly intervals, with a total of 1–15 samples per participant, with follow-up on each person for the duration of the outlined study periods. Stool samples were also collected from all dogs that belonged to the household.
Canine fecal samples were obtained from families enrolled in the study. In year one, samples were collected from the household premises. Due to low sampling rates of dogs in the first year, dogs were given suppositories and visibly monitored for defecation during the second year (Table 1).
Identification of Giardia- positive samples for DNA purification and polymerase chain reaction. Stool samples (human and canine) were concentrated by sedimentation as described elsewhere [22, 23] with the exception that purified, double-distilled water was used in lieu of formalin. Concentrated stool samples were screened microscopically for G. lamblia cysts, and positive samples were prepared for DNA analysis. G. lamblia - positive fecal concentrates were then purified by the QIAmp DNA Stool Kit (Qiagen); polymerase chain reaction (PCR) and sequencing were performed on the samples (Figure 1).
Genotyping strategy. A nested PCR that amplifies the triose phosphate isomerase locus (tpi)  was used to genotype isolates obtained from humans; this method can detect genotypes A and B, as well as genotypes C and D. A second nested tpi PCR approach  was used to detect mixed A and B genotype samples, and to distinguish A1 and A2 isolates. Samples were visualized by agarose gel electrophoresis of nested products for the B genotype and restriction fragment length polymorphism using enzyme RsaI to distinguish A1 from A2 as described elsewhere .
Molecular typing of canine samples using SSU-rDNA. Canine samples were genotyped by amplifying and sequencing a portion of the SSU-rDNA gene , using a nested PCR to increase the yield and sensitivity .
Genotype A2 sequencing and/or subtyping at 2 chromosomal loci. A nested PCR yielding a 526 base pair (bp) product was designed for A2 molecular typing at chromosome 3 within open reading frames (ORFs) GL_50803_113553 and GL_50803_3095 in the G. lamblia genome database (GiardiaDB, http://www .giardiadb.org) because of discriminatory single-nucleotide polymorphisms (SNPs) within this locus . External primer pair, F-5 ‘TGGAGGCGGTCAAGATACTC-3 ’, R-5 ‘-CTCGACGATTATGCTCCACGACG-3 ’was used to generate a PCR product that was then used as the template for a nested PCR using F‘-CGGTCAAGATACTCTACGATCG-3’, R-5‘-CTCGACGATTATGCTCCACGACG-3’.
A nested PCR generating a 732 bp product was designed on chromosome 5 within the translation initiation factor ORF GL_50803_39587 (http://www.giardiadb.org). This ORF was determined to be the most heterogeneous after sequencing 9.5 kb near the tpi gene in several A2 field isolates . Primer pair F-5 ‘GGCGAGTGCAGTCCTGAGTGG-3 ’, R-5 ‘CCTGGCTTGTTAACTACATCC3 ’was used for amplification of the external product, and the primer pair F-5‘-CAGTTTGGAAGAGCAGGACTCG-3’, R-5‘-GTCATCTTTCTATGCCTTCTTCG-3’was used for nested PCR.
A previous study described 4 different subtypes of A2 isolates (55, 246, 303, and 335) in addition to the reference isolate JH, based on sequence differences at loci on chromosomes 3, 4, and 5 . In the current study, we have used the sequences of these 5 isolates to designate the subtype at each of the 2 loci. Thus, a 55/JH subtype is identical to 55 at the chromosome 3 locus and identical to JH at the chromosome 5 locus.
Genotype B sequencing at 2 loci. A nested PCR reaction generating a 772 bp product was designed for molecular typing of B genotype samples at the γ-giardin gene on chromosome 3. Primer pair F-5 ‘-CTTCTCGTTCATGCCACTGATGA-3 ’,R ‘-GTTGAGACAGAGAAATGTATCCAACG-3 ’was used for amplification of the external product, and the primer pair F ‘-CTTGTACTTATGTGTGGAAAGTTCG-3 ’, R-5 ‘-GTTGAGACAGAGAAATGTATCCAACG-3 ’was used for the nested PCR using the external amplification product as template. The nested PCR protocol amplifying the tpi gene (chromosome 5) was used as described elsewhere .
PCR and sequencing. Concentrations of PCR master mix ingredients were as follows: 1.5 mmol/L magnesium chloride, 200 nmol/L of each primer, and 200 nmol/L of each deoxynucleotide triphosphate unless otherwise indicated. To improve yield, all PCR master mixes included nonacetylated bovine serum albumin and dimethyl sulfoxide. Five microliters of template were used for external PCRs, whereas 2 μL of external PCR product were used as template for nested PCRs, unless otherwise indicated. Thermocycler conditions were as follows, unless otherwise indicated: 2 min initial denaturing (94°C), 40 cycles of 30 s denaturing (94°C), 30 s of annealing (51°C), and 1 min of extension (72°C), followed by 10 min of final extension (72°C).
Nested PCR products were purified using the microcentrifuge-based gel extraction protocol of the QIAquick PCR purification kit (Qiagen). Sequencing was performed on nested PCR products using BigDyeTerminator reagents and on nested amplification primers in a 3730 DNA Analyzer (Applied Biosystems) with the addition of 3.3% dimethyl sulfoxide.
Description of genotyping. A total of 104 families were enrolled in the study—43 families in 2002 and 61 families in 2003 (see Table 1). Altogether, they contributed 1531 human and 605 canine samples. Giardia cysts or trophozoites were detected by microscopy in 313 human samples, and 178 of the 313 positive samples were selected for genotyping (Figure 1). The 178 were chosen by giving priority to isolates from families with dogs and from those with >1 infected person. They were typed using a nested PCR assay described by Amar et al  that distinguishes genotypes A1, A2, and B. Among the 178 samples chosen for genotyping, there were no A1 isolates, 64 A2 isolates, 78 B isolates, 22 mixed A2/B isolates, and 12 samples that were not successfully amplified (Figure 1).
In addition, a second approach described by Sulaiman et al  that distinguishes genotypes A, B, C, and D was used for 114 of the 178 isolates, primarily to determine whether any humans were infected with genotypes C or D  (Figure 1). None of these 114 samples were from genotypes C or D. Although our study was not specifically designed to compare the results of the 2 typing methods, a few comments regarding the comparison are in order. For the isolates that were analyzed by both methods, the concordance was reasonably good (Figure 1). Of the 32 that were typed as genotype A2 by the Amar approach and tested by the Sulaiman approach, 2 (6%) were classified as genotype B. One of these 2 genotypes was sequenced and confirmed as genotype A2. Of the 63 isolates classified as genotype B by the Amar method and tested by the Sulaiman approach, 1 (2%) was classified as genotype A. The majority of those classified by the Amar approach were successfully sequenced at 1 or both of the chromosome 3 and 5 loci. Of the 66 that were classified as genotype A2 by the Amar method, 59 were sequenced and all were confirmed as genotype A2. Of the 80 that were classified as genotype B, 62 were sequenced and 59 were confirmed as genotype B, whereas 3 were confirmed as genotype A2. Thus, the accuracy of the PCR-based method described by Amar et al  was 100% for the classification of genotype A2 isolates, whereas the accuracy for classification of genotype B isolates was 95%.
Subtyping and evidence for recombination among isolates. The A2 isolates were also subtyped at 2 sites: (1) adjacent to tpi (chromosome 5) and (2) adjacent to γ-giardin (chromosome 3) by sequence analysis of uncloned PCR products. The tpi and γ-giardin genes themselves showed no polymorphism, but these adjacent regions had numerous polymorphisms that allowed accurate sequence-based haplotyping within the genotype A2 isolates (Table 2) . There were 5 informative sites at the chromosome 3 locus and 7 informative sites at the chromosome 5 locus. We were able to type a total of 80 samples from 33 individuals at both loci and have named them according to previously reported haplotypes . When identical samples from individuals were omitted from analysis, there were 27 samples, and when duplicated types within a family were included, there were 19 samples. The 4 subtypes at the chromosome 3 locus and the 5 subtypes at the chromosome 5 locus allow a possible 20 recombination types, and 12 of those 20 were represented in this data set (Table 3).
For the genotype B isolates, there were numerous polymorphisms within the tpi and g-giardin reading frames, so these regions were sequenced. There were 106 polymorphic sites, and in all but one, these were transition substitutions (A/G or C/ T). However, at all but 3 of these polymorphic sites, there were some sequences with double peaks on the chromatogram. These double peaks could result from mixed genotype B infections or from allelic sequence heterozygosity (ASH). We have sequenced multiple cloned PCR products from some samples (data not shown), and in these cases, the data are more consistent with ASH than with mixed infection.
In some cases, different samples from a single individual or a single family varied among each other (eg, A, G, or double-peak purine). For this reason, we considered these sites ambiguous and were unable to assign subtypes for the genotype B isolates. In contrast, there were only 5 potential ASH sites among the genotype A2 isolates, all at the chromosome 5 locus.
The genotype A2 isolates are distinct from genotype B isolates within the tpi and γ-giardin reading frames and demonstrated no polymorphism among themselves. Thus, there is no evidence for recombination between genotype A2 and B isolates at these loci.
Potential for transmission from dogs to humans. Sixty-seven of the 88 canine fecal samples from 46 dogs that were positive for Giardia by microscopy were successfully amplified and genotyped by a previously described method that uses sequencing of a portion of the SSU-rDNA  (Figure 1). Thirty-two dog samples had the D genotype, and 9 samples had the C genotype. Twenty-six had mixed C/D genotype infections that were detected by a double peak in the chromatogram. For 13 of the 19 dogs that were sampled more than once, the genotype changed (C, D, or mixed). Each of the 88 dog samples was also tested with primers designed to amplify and distinguish DNA from A1, A2, B, or mixed A/B infections . This approach had the advantage of being able to detect coinfections in which the A or B genotypes were present in relatively small quantities in comparison to the C and D genotypes. Using this approach, a single dog sample had a genotype B isolate (see Figure 2, family 95). This dog had samples collected in 3 successive weeks, all with genotype D, and in the middle week had mixed genotypes B and D. The rDNA typing had identified only genotype D, suggesting that D was present in greater quantity. These findings are more indicative of anthropozoonotic transmission with transient genotype B colonization of the dog or of mechanical passage of cysts that had been ingested from human feces than they are of zoonotic transmission from dog to human.
Transmission patterns within families. Twenty-two families representing 120 individuals were chosen for more detailed analysis because they had multiple G. lamblia -infected family members, and 8 of these families had 1 or more dogs (Table 4). The infection patterns for 8 of these families are shown in greater detail in Figure 2.
Different members of a family harbor different genotypes. Individual B from family 106 was infected with a genotype B isolate on multiple occasions over both years, whereas individuals C and E were infected with the same subtype of genotype A2. Likewise, the 2 members of family 61 carried distinct subtypes of A2.
Mixed infections or changing genotypes. Family 100 is particular instructive in that 4 individuals were initially infected with genotype B, and then genotype A2 was introduced by individual c. By year 2, the 2 individuals had genotype A2 only, whereas another 2 had genotypes A2+B. Also of note is that 2 subtypes of A2 were found in this family with 335/246 in individual D and 335/335 in individuals E and H. (Individual F could not be typed at locus 2.) The presence of multiple genotypes and subtypes within single households creates an optimal setting for mixed infections to occur, which in turn could facilitate recombination among isolates.
Spatial analysis of G. lamblia infections. A map showing the spatial distribution of families and genotype/molecular type information was generated (Figure 3). These data indicate that genotypes A2 and B were widely distributed within the population, an observation that is consistent with the frequent observation of mixed A2 and B infections.
The question of zoonotic transmission of Giardia infections remains controversial. Numerous mammals can be infected with 1 or more of a complex of organisms that are all morphologically identical and are all given the species label G. lamblia . However, these organisms differ at the molecular level and have been divided into at least 7 genotypes or assemblages with varying degrees of host specificity. Genotypes A and B are the only ones found in humans, but these organisms have also been found in multiple other mammals. Because dogs are typically the mammals with the greatest proximity to humans, a better understanding of the potential of transmission between dogs and humans will help to determine whether dogs play a role in human Giardia infections. Genotypes C and D have most commonly been identified in dogs, but some reports have identified genotypes A or B in dogs. Dogs in close proximity to humans under compromised hygienic conditions in a rural Aboriginal community in Australia always harbored genotypes C or D . In contrast, dogs in an urban setting in Perth, Australia  and in tea plantation communities in India  often carried genotypes A and B. Hopkins et al  proposed that (1) there were 2 cycles of Giardia transmission in the rural Aboriginal community, one in humans of genotypes A and B and one in dogs with genotypes C and D, (2) genotypes C and D may be better adapted to dogs and outcompete genotypes A and B, and (3) contact between dogs in the urban community was rare, with the result that most dog infections in this setting were due to genotypes A and B from humans . The current study included 8 families that had multiple persons infected with Giardia and had at least 1 dog. Despite this, the only dog with a “ human genotype” had genotype D infection with transient genotype B infection that was probably present in lower quantities than the genotype D isolate. We suggest that most likely this dog was transiently infected with Giardia from a human source, but that the genotype D organism was better adapted to the dog, allowing it to persist. These data certainly do not rule out occasional transmission from dogs to humans but suggest that it is not frequent in this Giardia -endemic setting.
Molecular epidemiologic studies of the epidemiology giardiasis have mostly assumed clonal rather than sexual or par-asexual reproduction, because Giardia has been assumed to be an asexual organism until recently. However, the finding of the meiotic gene repertoire in the Giardia genome  and population genetic evidence for recombination among the A2 isolates  means that recombination may explain certain molecular findings. One of the predictions of sexual reproduction is independent reassortment of chromosomes, whereas clonal reproduction predicts a lack of reassortment. The current study identified a total of 12 of the 20 possible reassortment types in a set of only 19 samples, providing evidence that chromosome reassortment is frequent in this setting of highly endemic transmission. Mixed infections with genotypes A2 and B and even 2 different subtypes of A2 were frequent in this setting, creating an ideal environment for recombination to occur. Thus, it is notable that, despite the abundant evidence for recombination among A2 isolates, there was no evidence for recombination between genotype A2 and B isolates. The lack of recombination between A2 and B would suggest that these 2 genotypes are reproductively isolated from each other and would provide additional support to prior studies suggesting that these organisms belong to different species [2, 3]. The likelihood that genotypes A and B represent different species is also supported by a recent report of the genome of the genotype B isolate, GS  and by our own sequence of GS (unpublished data).
Even though genotype B isolates appear to have greater heterogeneity than A2 isolates, all the informative SNPs at the loci chosen for the current study also showed ASH in all of the isolates, so we could not confidently use the sequences to type these isolates. Other studies have also shown frequent substitutions in genotype B isolates in settings where single isolates might have been suspected [30–32].
These results are consistent with previous reports of high prevalence rates in this area with poor water quality and suggest that human-to-human transmission is much more important than animal-to-human transmission, and that dog giardiasis is more likely dog to dog and not human to dog. Genotypes A2 and B frequently coexisted within families and coinfected individuals, and in this setting, recombination among A2 isolates was prevalent, but there was no evidence of recombination between A2 and B isolates, which supports prior suggestions that these represent different Giardia species.
We appreciate the helpful comments of B. H. Jost, D. Wolk, and L. Joens. L. Cabrera was critical for subject recruitment and sample collection at the study site. C. Taquiri performed the microscopy identification of Giardia from human and dog fecal samples. M. Verastegui provided technical laboratory assistance. We appreciate the assistance of C. Morrison in finalizing the figures. This study was possible because of the assistance of PRISMA.
Potential conflicts of interest: none reported.
Presented in part: Third International Giardia and Cryptosporidium Conference, Orvieto, Italy, October 2009 (abstract O69).
Financial support: National Institutes of Health (NIH) Minority International Research Training program (grant 5 T37 TW00036-09 to M.A.C.) and Achievement Rewards for College Scientists (to M.A.C). The project was also funded in part by grant UDSA-ARZT-136034-H-02-124 (to C.R.S.).