Dientamoeba fragilis is a protozoan parasite found in the mucosal crypts of the large intestines of humans. Originally
D. fragilis was considered an ameba, but based on ultrastructural characteristics (
2), antibody data (
8), and phylogenetic data originating from 16S-like rRNA gene sequences, it has been established that it is a trichomonad (
20), with no identified cyst stage. Most recent literature accepts that
D. fragilis is an important enteric pathogen (
7,
10,
18), with an estimated incidence of symptomatic infection of between 4 and 91% (
11,
21,
25,
26). Symptoms include abdominal pain, bloating, and diarrhea.
Because of the lack of a cyst stage, diagnosis can be performed only on freshly passed stool or by the use of fixatives and permanent stains. In addition, day-to-day shedding is highly variable, which imposes the need for multiple sampling (
24). These features have likely led to an underestimation of the prevalence of
D. fragilis, which is reported to vary between 0.2% and more than 19% depending upon the population studied (
3,
9,
15,
16,
26).
Despite the relatively high prevalence of
D. fragilis and its apparent role in patients presenting with gastrointestinal complaints, surprisingly little is known about its pathogenicity, route of transmission, epidemiology, and genetics. Because only some infected persons experience symptoms, it is possible that
D. fragilis is a heterogeneous species with nonpathogenic and pathogenic variants with similar morphology but different pathogenicities. This has been suggested for other lumen-dwelling protozoa such as
Giardia duodenalis (
13) and demonstrated for
Entamoeba histolytica/Entamoeba dispar (
4,
6). In a first report on variability in
D. fragilis, the 16S-like ribosomal subunit DNA sequence of cultured
D. fragilis from a small number of patients was analyzed by restriction fragment length polymorphism (RFLP) (
14). Two of the 11 isolates gave a different restriction fragment pattern, indicating that there was genetic diversity in these
D. fragilis cultures.
RESULTS
DNA isolated from stool samples of an initial group of 11 patients containing microscopy-detected
Dientamoeba fragilis was amplified by PCR with the primers and conditions described by Johnson and Clark (
14). This PCR amplifies the complete coding region of the ssu rRNA gene and produces an amplicon of approximately 1.7 kbp. Only 3 of the 11 samples tested gave a product of the expected size, while many additional nonspecific bands were observed for all patients' samples. Amplification of DNA extracted from cultured
D. fragilis (kindly provided by C. G. Clark) or cultured
Trichomonas vaginalis showed only the 1.7-kbp product. This indicated that the only PCR method published for the detection of
D. fragilis (
14) was not species specific and inefficient for direct amplification from stool specimens, while subsequent RFLP analysis of the amplicon would be severely hampered by the presence of the nonspecific bands. No inhibitory effect of fecal material could be detected, as DNA extracted from two
D. fragilis-negative stool samples spiked with
D. fragilis chromosomal DNA gave bands of the same intensity as without DNA from
D. fragilis-negative feces.
To obtain a more sensitive and species-specific PCR for the detection of
D. fragilis, primers DF1 and DF4 were designed, amplifying the region from positions 100 to 761 of the ssu rRNA gene. Both primers contain several consecutive nucleotides at their 3′ ends that match only the
D. fragilis sequence but not the other known ssu rRNA genes of closely related
Trichomonas species. PCR analysis of DNA extracted from cultured
D. fragilis or
T. vaginalis with primers DF1 and DF4 showed amplification of
D. fragilis DNA only. This 662-bp amplicon contained two
RsaI restriction sites; the
RsaI site at position 313 is species specific and not present in other trichomonads, while one of the four
DdeI restriction sites present in the amplicon (position 644) could be used to discriminate between the two known genetic variants in
D. fragilis (
14).
To determine the sensitivity of the PCR, the 662-bp amplicon was cloned and a known number of copies were then amplified with the same conditions as for the patients' samples. This indicated that the detection limit was 10 plasmid copies or an equivalent of approximately 0.1 D. fragilis trophozoite. D. fragilis does not have a cyst stage, and trophozoites are known to degenerate within a few hours of stool passage. The PCR assay developed was therefore further evaluated with four microscopically D. fragilis-positive stool samples stored at room temperature for various time intervals. In all four samples D. fragilis DNA could be detected up to 1 week, after which the signal could no longer be observed or became very weak.
PCR amplification with primer set DF1/DF4 of the DNA samples from the 11 patients mentioned above yielded a single product of the expected size in 9 samples. Restriction enzyme digestion of these PCR products with
RsaI produced identical patterns (Fig.
1, top panel) that were as predicted from the gene sequence. Digestion with
DdeI revealed a pattern that was distinct for the two known RFLP haplotypes of
D. fragilis (Fig.
1, bottom panel; G2 and G1, lanes 10 and 11, respectively) and could be used to analyze the variation in the patients' stool samples.
D. fragilis amplified directly from the first nine patient stool samples were all of genotype 1.
To further study the genetic diversity of D. fragilis in the human population, stool samples from additional patient groups and controls, microscopically positive for D. fragilis, were analyzed. These included children (≤18 years) and adults with gastrointestinal complaints, travelers returning from tropical countries, and asymptomatic carriers. The sample taken on the second day of the TFT test (TFT2) does not contain a fixative and can be used for DNA isolation and subsequent PCR analyses. TFT2 samples subjected to PCR originated from TFTs were either TFT1 or TFT3 positive for D. fragilis, or TFT1 and TFT3 positive for D. fragilis by microscopic examination.
Interestingly, intermittent shedding of D. fragilis observed in the TFT was also observed by PCR analyses. In cases where both TFT1 and TFT3 were positive for D. fragilis, the PCR detected D. fragilis in TFT2 in a high percentage (74%), while in the cases where only TFT1 or TFT3 were positive, the percentage of positive TFT2 was much lower (18%). In total, 59 samples were found positive by PCR for D. fragilis (20 of 28 children [71%], 24 of 50 adults [48%], 9 of 15 travelers [60%, all adults], and 6 of 6 asymptomatic carriers [100%, all adults]) and subjected to RFLP analysis by digestion with DdeI. Again all samples were of genotype 1.
Of the 59 PCR-positive samples, 16 patients (seven children and nine adults) and four asymptomatic carriers were selected for sequence analysis of the PCR product. In parallel, the sequence of the coding region of the ssu rRNA gene of
D. fragilis genotypes 1 and 2 was determined. The alignment of these sequences indicated that there was no variation in the sequences obtained from the patients or asymptomatic carriers and that all were identical to genotype 1 (Fig.
2). In addition to several nucleotide changes, the sequence of
D. fragilis genotype 2 contained a
DdeI site at position 644, as expected from the RFLP haplotyping. The sequence divergence between the ssu RNA genes of genotypes 1 and 2 was approximately 2%, based on the 10 nucleotide differences in the 558 bp of the ssu rRNA genes.
DISCUSSION
In the present study we analyzed the DNA of
Dientamoeba fragilis directly from feces of various groups of patients and asymptomatic carriers without the need for prior culturing of the organism. Culturing of
D. fragilis has been reported to be notoriously difficult, and the species has never been cultivated axenically (
5). The absence of a cyst stage and the rapid degeneration of
D. fragilis after it leaves its host are probably also important factors that have impaired progress in research of this globally occurring parasite. Our knowledge of
D. fragilis is deficient in many aspects; nothing is known about the pathogenicity and route of transmission, while the genetic data are confined to a single partially sequenced ssu rRNA gene. Our data indicate that it is now possible to study
D. fragilis sequences directly from stool samples and that intact DNA could still be recovered from feces that have been at room temperature for 1 week.
Because
D. fragilis is reported to give symptoms in only some infected persons (4 to 91%), it is possible that the species contains genetic variants with different pathogenic potential. Genetic variation of intestinal protozoa in humans is not without precedent;
Blastocystis hominis displays a marked heterogeneity in ssu rRNA gene sequences, but these have not been linked to clinical symptoms (
1,
22).
Giardia lamblia also displays genetic variability at various loci (
23), which are also possibly associated with different gastrointestinal complaints (
13). The most-studied and best-known example of genetic variation of a intestinal protozoan “species” which was linked to clinical presentation is that of
Entamoeba histolytica/E. dispar. Infection with
E. histolytica may lead to severe disease, whereas carriers of
E. dispar remain asymptomatic. Although microscopic techniques are unable to discriminate between these two morphologically identical variants, DNA-based detection techniques have now firmly established that amoebiasis is caused by
Entamoeba histolytica, while
Entamoeba dispar is not pathogenic (
4,
6).
As a first step towards the identification of genetic variants in
D. fragilis, Johnson and Clark (
14) analyzed nine short-term cultures obtained from stool samples of patients positive for
D. fragilis (no asymptomatic carriers of
D. fragilis were analyzed). After amplification of the ssu rRNA gene and RFLP analysis, it appeared that two genetically different variants could be identified in this small number of isolates of patients with gastrointestinal complaints. These two variants differed by three restriction enzyme sites in their ssu rRNA gene. Although no sequence data were presented, this suggested a sequence divergence of approximately 2% between the two
D. fragilis variants (
14). In the current study, part of the ssu rRNA gene of
D. fragilis was directly analyzed from stool samples of several different patient groups, including children and adults with gastrointestinal symptoms, travelers, and asymptomatic carriers.
Although the PCR fragment used in the current study is smaller (661 bp) than the PCR product of 1,674 bp described by Johnson and Clark (
14), it appeared to be far more efficient and specific for direct analysis of
D. fragilis from stool specimens, and it could be used to differentiate between the two described genotypes, as it contained one of the polymorphic
DdeI sites. The genotype could be established from all PCR-positive samples (
n = 59) through RFLP analysis. Remarkably, we were unable to find any variation in RFLP pattern in these groups, and all corresponded to genotype 1. RFLP of PCR products is an easy-to-use technique that is commonly employed to identify genetic variation between closely related organisms. The obvious disadvantage is that it will detect sequence variation only within the recognition sites of the restriction enzymes that mostly cover only 4 or 6 bp.
D. fragilis contains three stretches of almost exclusively adenine/uracil in its ssu RNA sequence at positions 564 to 616, 685 to 733, and 1385 to 1411 that are absent from ssu RNA sequences from closely related trichomonads and are responsible for the reduced G/C content and increased length of the ssu RNA of
D. fragilis (
20). These adenine/uracil expansion fragments are known to be hypervariable in eukaryotic rRNA but are difficult to access for variability by RFLP. Two of these adenine/uracil expansion segments are present within the DF1/DF4 PCR fragment (positions 564 to 618 and 685 to 733), and to further study the possible sequence variation in
D. fragilis, the DF1/DF4 PCR fragment was sequenced for 16 patients and four asymptomatic carriers. The adenine/uracil segment at positions 564 to 618 could be sequenced from both strands and did not show any variation among patient- and carrier-derived sequences. Importantly, this sequence was identical for genotype 1 and genotype 2 except for a single transversion from A to T at position 659. This again indicated that there was no variation in the sequences obtained by direct amplification of
D. fragilis from fecal samples. The estimated sequence divergence between the two ssu rRNA genotypes observed by Johnson and Clark (
14) was 2%, which correlates well with our sequencing data, in which 10 of the 558 bp analyzed were different between genotypes 1 and 2.
Our findings that D. fragilis displays only a single genotype in fecal samples of various patient groups, including travelers, suggest that genotype 2 is either very rare or that culturing of D. fragilis leads to a strong bias in favor of the more rare genotype 2.
Variations in rRNA gene sequences are used only as markers to analyze speciation and search for phylogenetic associations in closely related species. The variation itself is unlikely to have any influence on the pathogenic potential of parasites. It could therefore be worth studying variation in genes in
D. fragilis which are potentially involved in pathogenicity. Similar studies have been performed on
Entamoeba histolytica and resulted in the identification of polymorphic gene products that are correlated with virulence in amebiasis (
17,
19) and gene products that could be used to determine the geographic origin of isolates and routes of transmission (
12,
27,
28). Similar studies on
D. fragilis should begin with identifying such protein coding genes, as to date only the coding region of the ssu rRNA gene has been published (
20). The findings presented in the current paper indicate that variation in such newly identified genes could be directly analyzed from large numbers of human fecal samples without the need for prior culturing of
D. fragilis.