Giardia duodenalis, the causative agent of giardiasis in humans, is prevalent worldwide with a broad clinical spectrum ranging from asymptomatic infections to acute or chronic diarrhoea. Giardiasis in children in developing countries results in malnutrition, stunting and deficits in cognitive function [Reference Celiksoz1]. Eight assemblages (A–H) have been described of which assemblages A and B are described as being human associated but have also been detected in a wide range of animal hosts. The other assemblages infect animals, although occasionally occur in human samples and are often linked with suppression in immunity [Reference Gelanew2, Reference Feng and Xiao3]. The distribution of the human-associated assemblages A and B varies geographically. There are conflicting reports on their association with symptomatic disease with some authors, including our group, finding an association with assemblage A [Reference Ajjampur4, Reference Haque5] and others with assemblage B [Reference Gelanew2]. Several loci have been described for determining assemblage of Giardia isolates including triose phosphate isomerase (tpi), glutamate dehydrogenase (gdh), β-giardin, small subunit ribosomal RNA (SSU rRNA) and elongation factor genes. Discrepancy between the loci for both human and animal isolates has been reported, possibly due to recombination or multiple infections. Some studies have suggested applying a multi-locus genotyping approach to assign assemblage type to a particular isolate (e.g. [Reference Caccio6]). In this study, we characterized Giardia isolates from children and adults in South India to determine the predominant circulating assemblages using a multi-locus genotyping approach.

Faecal samples from 25 children with and without diarrhoea collected between January 2002 and July 2008 from an urban slum in Vellore district of Tamil Nadu, India with Giardia cysts or trophozoites identified by microscopic examination were included in this study. Samples from children with diarrhoea are referred to as ‘CRI’ as they were collected as part of a study on childhood rotavirus infections while samples from children without diarrhoea are referred to as Giardia study (GS) or Kaniyambadi area (KB). In addition, Giardia-positive samples from 24 adults (referred to as adult Giardia or ‘AG’) with gastrointestinal symptoms visiting Christian Medical College Hospital, Vellore collected between September and October 2008 were included. Diarrhoea in children was defined as at least three loose stools in a 24-h period and gastrointestinal symptoms for the adult subjects were defined as having any one or combination of the following symptoms or clinical diagnosis of dyspepsia or functional bowel disorder: diarrhoea, malaise, flatulence, foul smelling greasy stool, abdominal cramps, bloating, anorexia, nausea, weight loss, vomiting, fever and constipation. For samples from children enrolled in this study, informed consent was obtained from parents or guardians. Consent for samples from adults was waived as samples were taken from routine diagnostic work without patient identifiers. This study was approved by the Institutional Review Board of the Christian Medical College, Vellore.

DNA was extracted from the faecal samples using the Qiamp DNA stool minikit (Qiagen, USA) according to the manufacturers' protocol. The tpi, gdh and β-giardin PCR–RFLP was performed using previously published protocols with the minor modification of adding 1% DMSO to the PCR master mix [Reference Ajjampur4, Reference Lalle7, Reference Bertrand, Albertini and Schwartzbrod8]. The three markers were chosen to resolve sub-assemblages AI and AII (tpi), BII and BIV (gdh) and to differentiate human-associated assemblages (A and B) from the other, mostly zoonotic, assemblages (C–G, β-giardin). For confirmation of RFLP results, DNA extracted from representative samples of each assemblage determined by polymerase chain reaction–restriction fragment length polymorphism (PCR–RFLP) at the three loci were sequenced (MWG Biotech Pvt Ltd, India) and uploaded to Genbank (see Table 1 for accession numbers). Samples that did not have mixed infections/recombinations with more than one assemblage or sub-assemblage and for which PCR at all three loci had worked were selected. Raw sequences were then compared with reference sequences [Reference Feng and Xiao3] in a pairwise identity matrix using Genetool version 1.0 [Reference Layon9]. The reference sequences used were EF688040 for tpi assemblage AI [Reference Lasek-Nesselquist10], U57897 for assemblage AII [Reference Baruch, Isaac-Renton and Adam11] and EU041754 for assemblage AIII [Reference Lalle12]. For the gdh locus, the reference sequences used were EF685701 for assemblage AI [Reference Lasek-Nesselquist10], EF507677 for assemblage AII [Reference Souza13], EU637582 for assemblage AIII [Reference Caccio6], AF069059 for BIII [Reference Monis14] and AY178738 and L40508 for assemblage BIV [Reference Monis15]. The β-giardin reference sequences used were X14185 for assemblage AI [Reference Holberton, Baker and Marshall16], DQ116610 for AII [Reference Di Giovanni17], and EU621373 for assemblage AIII.

Table 1. Pairwise identity matrix of assemblage A and assemblage B sequences

AG, Adult with giardiasis; CRI, child with giardial diarrhoea; GS and KB, child with asymptomatic giardiasis; RFLP, restriction fragment length polymorphism.

Of the three typing methods, tpi PCR–RFLP was the most sensitive with amplification of DNA extracted from all 74 samples, while gdh and β-giardin PCRs amplified DNA extracted from 46 (62·2%) and 42 (56·8%) of the samples. In a previous study, Bertrand et al. also found tpi PCR to be more sensitive than gdh PCR [Reference Bertrand, Albertini and Schwartzbrod8]. The gdh PCR was positive in 15 children with diarrhoea (60%), 13 (52%) without diarrhoea and in 18 adults (75%) while the β-giardin PCR was positive in 14 children with and without diarrhoea (56%) and 14 (58%) adults. The assemblage and sub-assemblage distribution for each locus is given in Table 2. In both adults and children, assemblage B was predominant (by tpi PCR–RFLP). Previous studies by us as well as other authors in the Indian subcontinent have also demonstrated a predominance of this assemblage [Reference Ajjampur4, Reference Haque5, Reference Singh18]. The tpi PCR–RFLP detected a few sub-assemblage AII infections but no sub-assemblage AI. When the gdh PCR–RFLP was performed, both sub-assemblage BIII and BIV were detected. PCR–RFLP techniques (gdh and β-giardin) did not detect any other non-human assemblages in this population. This could indicate that zoonotic transmission is not common in this setting. However, since we did not use assemblage-specific primers, amplification of the more predominant assemblages could have occurred.

Table 2. Giardial assemblages determined by polymerase chain reaction–restriction fragment length polymorphism at the tpi, gdh and β-giardin loci

* No mixed assemblage A and assemblage B infections detected.

The tpi PCR–RFLP was better able to resolve mixed infections/recombinations with assemblages A and B since both the gdh and β-giardin PCR–RFLP typed these samples as either assemblage A or assemblage B infection alone. Interestingly, these mixed infections/recombinations with two assemblages were seen only in children (6/50), four of whom had diarrhoea indicating that they are probably mixed infections occurring in younger and possibly more susceptible individuals. Among these samples, β-giardin PCR identified three samples as assemblage B and one as assemblage A while gdh PCR identified one as an assemblage BIII and BIV co-infection, one as assemblage BIII and one as assemblage AII. Mixed-infections/recombinations with sub-assemblages BIII and BIV identified by gdh PCR–RFLP were also seen more frequently in children (8/15 in children with diarrhoea and 5/13 in children without diarrhoea) than adults (3/18) (Fisher's exact test, P=0·058).

When a pairwise identity matrix of a few isolates identified as assemblage AII by RFLP was constructed (Table 1), tpi sequences were better able to resolve the sub-assemblages (>98% identity with sub-assemblage AII and 92–95% with sub-assemblage AI) while gdh and β-giardin loci showed a poorer resolution. For assemblage B, gdh sequences of a few isolates identified as sub-assemblages BIII and BIV were analysed but were of limited use in resolving sub-assemblages. Limitations in the current sub-assemblage classification based on sequencing due to heterogeneity at these loci and the need for refinement of these techniques has been described elsewhere [Reference Bonhomme19].

In our study, the tpi PCR was more sensitive in detecting Giardia infection and mixed assemblage infections/recombinations than the gdh or β-giardin PCRs. Comparison of the assemblage A sequences at the tpi locus in the pairwise identity matrix was more discriminatory in identifying sub-assemblages than the other loci. The predominant circulating Giardia assemblage in adults and children in this geographical location was found to be assemblage B and no non-human assemblages were detected suggesting a mostly anthroponotic transmission. However, more detailed studies using assemblage-specific primers and sequencing are required to corroborate findings of the absence of assemblage C–H as well as sub-assemblage AI and AIII in human samples. Studies on animal samples are also required to determine the true risk of zoonotic transmission. We also found that mixed assemblage and mixed sub-assemblage infections occur more commonly in children and the significance of this novel finding needs to be evaluated in future longitudinal studies.