Entamoeba Species in South Africa: Correlations With the Host Microbiome, Parasite Burdens, and First Description of Entamoeba bangladeshi Outside of Asia

Background
Diarrhea is frequent in communities without clean water, which include low-income South African populations in Giyani and Pretoria. In these populations, the amount of diarrhea caused by Entamoeba histolytica, inclusive of all ages, sexes, and human immunodeficiency virus status, is uncertain. Infection with E. histolytica can modulate the host microbiota, and a key species indicative of this is the Prevotella copri pathobiont.


Methods
A cross-sectional study of patients attending gastroenterology clinics was conducted to determine the frequency and burden of 4 Entamoeba species and P. copri.


Results
Entamoeba species were present in 27% of patients (129/484), with E. histolytica detected in 8.5% (41), E. dispar in 8% (38), E. bangladeshi in 4.75% (23), and E. moshkovskii in 0%. This is the first description of E. bangladeshi outside Bangladesh. In E. histolytica-positive samples, the levels of both the parasite and P. copri were lower in nondiarrheal samples, validating the results of a study in Bangladesh (P = .0034). By contrast, in E. histolytica-negative samples positive for either of the nonpathogenic species E. dispar or E. bangladeshi, neither P. copri nor Entamoeba levels were linked to gastrointestinal status.


Conclusions
Nonmorphologic identification of this parasite is essential. In South Africa, 3 morphologically identical Entamoeba were common, but only E. histolytica was linked to both disease and changes in the microbiota.

To test these hypotheses, DNA was extracted from the Giyani and Pretoria samples, and a multiplex quantitative PCR (qPCR) assay was used to detect both E. histolytica and the other morphologically identical Entamoeba species, such as E. dispar, E. moshkovskii, and E. bangladeshi. This assay was also modified to include a general Entamoeba probe to capture data on the presence of novel Entamoeba species genetically similar to the pathogenic species E. histolytica that may be present in the South African population.
The qPCR assay data captured quantitative information, and this permitted us to examine the correlation between the parasite burden in these samples and the outcome of infection. A link between parasite burden and symptomatic disease has been found in previous studies [4,8,9]. Recent studies have also highlighted the relationship between Entamoeba species and the bacterial communities of the gut. The presence of Entamoeba organisms was associated with a decrease in the abundance of Prevotella copri in farmers and fishermen from Southwest Cameroon [10], and the abundance of P. copri increased in diarrheal E. histolytica cases [9]. Hence, we also sought to quantify this bacterium in the microbiome of Entamoeba-positive samples in our study population.

Ethics Statement
The research and ethics committee of the University of Venda granted institutional approval. The study received ethical clearance from the Department of Health and Welfare, Polokwane, Limpopo Province, South Africa. We also obtained permission from the ethics committee of participating hospitals and clinics to collect samples. The objectives and concepts of the study were clearly explained in the language understood by the potential participants (ie, English, Sepedi, Xitsonga, and Tshivenda). A written, informed consent form was signed prior to study enrollment. In cases where the participant was either a non-English speaker or illiterate, a witness also signed the consent form.

Study Area and Population
The tested stool samples were predominantly from urban and rural populations of moderate-to-low socioeconomic status [11,12]. They were collected between November 2013 and June 2015 from diarrheal and nondiarrheal patients in the rural Nkomo clinic (Giyani) and the urban clinic within the Dr George Mukhari Hospital (Soshanguve District).
The catchment area for the Dr Georges Mukhari Hospital includes Soshanguve, Ga-Rankuwa, Mabopane, and parts of Madibeng District. The Nkomo clinic serves households within the Greater Giyani Local Municipality, Mopani District. Household water and sewage access is summarized in Table 1 [11,12].
Both adults and children of all ages were eligible for participation. A questionnaire was used to collect sociodemographic information, such as the age, sex, and origin of the study participants.

Sample Collection
After the patients were given a clear explanation of the stool sample collection process, they received screw-cap bottles into which they placed their samples. Stool samples were classified as diarrheal or nondiarrheal on the basis of the physical presentation of the sample, as defined by the Bristol stool form scale (diarrheal specimens, types 6 and 7; nondiarrheal specimens, types 1-5) [13]. The bottles were labelled with unique participant identifiers and then placed in a cooler box and transported to the University of Venda microbiology laboratory for further processing. Upon arrival to the laboratory, samples were aliquoted in 2-mL tubes and stored frozen at −20 o C. The aliquoted samples were shipped to the University of Virginia Infectious Diseases Research laboratory for analysis.

Genomic DNA Purification
Genomic DNA from each patient's sample was extracted using a QIAamp DNA Stool Mini Kit (Qiagen) according to the manufacturer's recommended procedures, using approximately 200 mg of stool samples with the modifications described by Liu et al [14]. One stool sample from a healthy US child whose stool had previously been tested and found to be negative for all Entamoeba species was included in each batch, to monitor for the occurrence of contamination during extraction. The DNA was eluted in 200 μL of elution buffer (Qiagen) and stored at −80°C until further analysis.

Multiplex qPCR Assay for Detection of Entamoeba Species
A multiplex qPCR assay was used for the amplification and detection of all Entamoeba species. Genus-specific primers were used in combination with a 42-nucleotide probe that should hybridize to E. bangladeshi, E. dispar, E. histolytica, and E. moshkovskii amplicons. Owing to the length required to generate this probe in these A/T-rich genomes, a double quencher was included in the design of the probe (Biosearch Technologies; Figure 1). This probe recognizes E. histolytica, E. moshkovskii, E. dispar, E. bangladeshi, and E. hartmanni amplicons but was not similar to the ribosomal RNA (rRNA) region in  Table 2.

Species Level by the qPCR Assay
Any sample that gave a positive signal with the Entamoeba probe but was negative for all 4 Entamoeba species of interest (ie, E. histolytica, E. moshkovskii, E. dispar, and E. bangladeshi) was characterized further by amplifying additional 18S regions and determining their sequences [3,15]. This DNA was amplified by primers Ehd-88R and EM-RT-F2 (  [3]. A 2% agarose gel stained with 3 µL of ethidium bromide was used to separate the amplified DNA. The PCR products were extracted from the agarose gel by using the Qiagen QIAquick Gel Extraction Kit, and the purified amplicons were sequenced using the Sanger method (Genewiz).

Sequence and Phylogenetic Analysis
After sequence results were obtained, the ABI files were downloaded and trimmed using Geneious (version 7.0.6). A phylogenetic tree was created to examine the phylogenetic relationship between novel Entamoeba species and the known Entamoeba parasites by use of the neighbor-joining algorithm included in Geneious (Biomatters) [16].

P. copri qPCR Assay
To improve the specificity of qPCR detection of P. copri in clinical fecal samples, a new TaqMan assay was designed, using the National Center for Biotechnology Information (NCBI) reference sequence NR_113411.1. Optimal qPCR assay conditions were initially determined using DNA amplified from cultured P. copri (CB7, DSMZ; a gift from D. Littman). Assay specificity was determined by purification and sequencing of the amplicons from selected fecal samples and by comparison with both the NCBI reference sequence NR_113411.1 and the sequence obtained from the cultured P. copri DNA. The PCR was performed with a 25-µL reaction mixture containing Bio-Rad iQ powermix, 0.4 µM of primers, and 0.2 µM for each probe. Probes, primers, and reaction conditions are shown in Table 2.
The previously described Enterobacteriaceae assay was used to normalize both the P. copri levels and as a measure of extracted bacterial DNA quality [9]. Enterobacteriaceae-and P. coprinegative samples were omitted from the quantitative analysis of P. copri (Supplementary Table 1).

Statistical Analysis
The Fisher exact test was used to analyze contingency tables. The D' Agostino and Pearson omnibus normality test and the  [41]. The target of the Entamoeba genus probe is highlighted in gray, and the sequences of the species-specific probes and the genus-specific primers are underlined.
nonparametric Mann-Whitney comparisons test were used to analyze and compare qualitative data. Tests were performed using GraphPad Prism, version 6. The differences were considered significant if the P value was <.05.

Demographic and Clinical Features
A total of 484 participants were recruited in this study, of whom 227 (47%) were from Giyani (a rural setting) and 257 (53%) were from Pretoria (an urban setting). Table 3 summarizes the demographic data of the study population.

Prevalence and Distribution of Entamoeba Species, by qPCR Analysis
Of the 484 samples tested by qPCR analysis, 29% (140) were positive for Entamoeba species. The total frequency of E. dispar detected by qPCR in the study population was 8% (38), with E. histolytica detected in 6.4% (31), E. bangladeshi detected in 4.5% (22), and unknown Entamoeba species detected in 10% (49). In 11 samples, coinfections with different Entamoeba species were observed (7 were coinfected with E. histolytica and E. bangladeshi, 1 was coinfected with E. bangladeshi and E. dispar, and 3 were coinfected with E. histolytica and E. dispar). In line with previous results, E. moshkovskii was not identified in the South African populations studied ( Figure 2).

Confirmatory Testing
To confirm the E. bangladeshi qPCR assay results, we sequenced the amplicon from 4 positive samples and compared the sequences to the E. bangladeshi sequence deposited in the NCBI GenBank (accession number KR025412.1). The South African sequences were identical to that of E. bangladeshi.

Species-Specific Probes
Entamoeba primers (Ehd-88R; EM-RT-F2) were used to amplify DNA fragments from the 49 samples that were qPCR  positive for the broad range Entamoeba but negative for all the species-specific probes. The amplified DNA was separated by electrophoresis, and in the 35 cases where the bands of the size predicted for the Entamoeba species were identified, it was purified from the agarose by using the QIAquick Gel Extraction Kit (Qiagen). The SSU rRNA gene amplicon was detected in 35 samples. Sequencing of the purified amplicon identified 10 additional E. histolytica-positive samples (n = 41), with an adjusted frequency Entamoeba qPCR-positive samples of 29.3%, and 1 additional E. bangladeshi sample (n = 23), with an adjusted positivity frequency of 16.4% ( Figure 2). This result suggested that in these samples the parasite level had simply fallen below the detection limit of the species-specific qPCR assay. These samples were not included in the later analysis. In 13 cases, the 18S rRNA amplicon sequences were similar to those of the nonpathogenic species E. hartmanni (all sequences were deposited in GenBank under accession numbers MF471201-MF471217), and in the remaining 11 cases either no useful sequence data were obtained or findings were similar to sequences from bacteria and had no significant similarity to any Entamoeba reference sequence in the NCBI database.

Parasite Burden in South African Samples
Other enteropathogens are common in this South African population, and in diarrheal samples coinfections can make it challenging to identify the causal organism [4]. Entamoeba were no more frequent in diarrheal samples than in controls (data not shown). No differences in Entamoeba frequency was observed between the rural and urban populations. The cycle value at which the (baseline-corrected) amplification curve exceeds the background fluorescence (Cq) is closely related to the amount of input DNA. The Cq data provided by the qPCR assay can therefore be used to determine whether the Entamoeba species burden was different in diarrheal and control fecal samples [4,13,17]. As the distribution of Cq values were non-Gaussian, significance was determined using the Mann-Whitney test. As expected, a significant difference in Entamoeba levels was observed in cases of E. histolytica-associated diarrhea (P = .0072; Figure 3A), but the level of the nonpathogenic species E. dispar was unchanged in control and diarrheal samples ( Figure 3C). Interestingly, the level of E. bangladeshi was also unchanged ( Figure 3B). Again, no significant differences were observed in the parasite burden in rural and urban samples.

Quantity of P. copri in Entamoeba-Positive Samples.
In a Bangladesh study, elevated levels of the pathobiont P. copri were associated with E. histolytica-associated diarrhea [9,18]. The level of P. copri in Entamoeba-positive diarrheal and control samples was measured and, to control for variations in fecal bacterial numbers, was normalized using an Enterobacteriaceae bacterial reference [9,19]. A fecal DNA standard was used to control for any differences in amplification efficiency in the P. copri and Enterobacteriaceae qPCR assays. Samples negative for either P. copri or Enterobacteriaceae were omitted from the quantitative analysis towing to concerns about sample quality (Supplementary Table 1). To convert the qPCR results to bacteria concentrations, DNA was extracted from a known amount of E. coli (ATCC 25922) and assayed. The relative level of P. copri was 1 log lower in E. histolytica-colonized samples as compared to the level in E. histolytica diarrheal samples ( Figure 4A) but was unchanged in E. dispar or E. bangladeshi infections when diarrheal and nondiarrheal cases were compared ( Figure 4B and 4C).

DISCUSSION
The present study reports an overall frequency of Entamoeba species in our samples collected from gastrointestinal clinics as 27% (129/484), with E. histolytica being present in 6.4% of the cases (31/484). Differences in the assay used, as well as in age, geographic location, and the fact that these samples were collected from gastrointestinal clinics, make it difficult to compare these results to those obtained from previous population-based studies [4,8]. A weakness in the current study was that information on human immunodeficiency virus status (expected to increase with age) was not available. In addition, the study was not adequately powered to analyze the susceptibility to Entamoeba among participants stratified by age [20,21]. The Cq of the majority of our Entamoeba-positive asymptomatic samples was ≥35 and would have been missed by a less sensitive assay. Assay specificity at high Cq values was confirmed by amplicon sequencing of select samples (data not shown). In the work reported here, the assay included an Entamoeba general probe that acted as an independent control to identify any potential closely related novel South African Entamoeba species present in these samples (an in-depth surveillance of the Entamoeba species in the Mopani district of South Africa had not previously been done). All assay results were analyzed to be certain that the species-specific signal remained at a constant ratio to the result obtained from the broad range probe. The probe would have recognized any Entamoeba species similar to the pathogenic species E. histolytica, E. moshkovskii, and E. bangladeshi or the nonpathogenic species E. dispar ( Figure 1A). The sequences of the closely related species also blocked nonspecific hybridization, allowing the higher assay Cq cutoff of ≤40 and increased assay sensitivity ( Figure 1A) [22].
A higher E. histolytica parasite burden in samples increases the probability that the E. histolytica strain detected is responsible for diarrheal symptoms [9,17]. In agreement with the previous studies, our results showed a statistically significant increase in the E. histolytica parasite load in South African diarrheal samples. The level of the nonpathogenic species E. dispar did not significantly change. This suggested that diarrhea coincident with E. dispar infections was due to other pathogens. The  pathogenicity of the recently identified E. bangladeshi is still uncertain, but the level of E. bangladeshi was also the same in both diarrheal and nondiarrheal South African samples. Additional work is planned to identify whether other coinfecting enteric pathogens are present in these samples.
Novel Entamoeba species have been identified in different geographical contexts. Therefore, samples that were positive with the Entamoeba general probe but negative with the species-specific probes were characterized by amplicon sequencing ( Figure 3) [1,3,23,24]. While novel South African Entamoeba species were not identified, to our knowledge this study is the first to describe the presence of E. bangladeshi in samples collected outside Bangladesh. This species was first described in Bangladesh in 2011, but our results suggest that E. bangladeshi may actually have a broad geographical range and is prevalent in both Asian and African continents [3]. This finding also suggests that other members of the Entamoeba genus not identified in previous surveys may also be common in South Africa [5,[25][26][27][28] In addition to the parasite burden, predisposition to diarrheal disease is thought to be influenced by the parasite environment [29][30][31][32]. Moreover, it has been suggested that specific components of the microbiota might be associated with symptomatic or asymptomatic E. histolytica colonization [33][34][35]. We examined the level of P. copri in the South African samples positive for Entamoeba species. Consistent with previous studies, the level of this bacterium was lower in asymptomatic E. histolytica-positive samples when compared to the level in E. histolytica-associated diarrheal samples [9]. Future work is needed to determine the significance of E. histolytica-associated changes in the microbiota. The gut Prevotella species are anaerobic bacilli predominant in the lumen of the colon [36]. Recent studies, however, suggest that disruption of the host mucosa can result in an increase in Prevotella species at mucosal sites and a subsequent increase in host inflammatory responses [18,37]. Additional studies are needed to determine whether low P. copri levels could mitigate the host immune response occurring during amoebic colitis. It is possible that E. histolytica, unlike nonpathogenic Entamoeba species, disrupts the protective mucosal layer and exposes the host epithelium to the luminal microorganisms. This could expose the epithelium to high P. copri levels, as well as to E. histolytica, and result in an excessive inflammatory response with subsequent diarrhea [38][39][40]. In samples positive for the commensal E. dispar (which is not known to induce an inflammatory response or diarrhea), P. copri levels were not significantly different in either diarrheal or nondiarrheal samples [39].
In summary, an increase in the commensal bacterium P. copri was associated with diarrhea due to E. histolytica. The interplay between the pathogen, host, and host microbiota may be of importance in the development of symptomatic disease.

Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.