This work has been carried out as part of the "SchistoVAN" project (clinicaltrials.gov ID NCT01553552), sponsored by the Biomedical Research Centre EPLS (Saint-Louis, Senegal) ( http://www.espoir-sante.org). The present study was approved by the National Ethics Committee of the Ministry of Health of Senegal (September 2011; protocol number SEN11/43, clinicaltrials.gov: NCT01553552). Oral and written informed consents were obtained from the parents or the legal guardians of the children in accordance with the Code of Ethics of the World Medical association (Declaration of Helsinki). Children enrolled in the present study formed a sub-cohort of the "AnoPalAnoVac" project (clinicaltrials.gov ID NCT01545115) and were recruited on the basis of their age (6 to 10 years of age in October 2011) through village nurses, healthcare workers in the community and school directors. Date of birth was ascertained from vaccination cards or school register. A standardized questionnaire was completed for each child (see Additional file 1). In particular, children were asked about each gastrointestinal symptom individually including abdominal pain, diarrhea, and vomiting. This cross-sectional study was conducted in October 2011, in 3 villages of the Podor district located in northern Senegal: Agniam Towguel (16°32′N-14°48′W; total population (TP): 989; several temporary ponds, traditional housing, irrigated crops), Fanaye Diery (16°32′N-15°13′W; TP: 6781; animal husbandry, irrigated crops, some urbanized habitat), and Niandane (16°35′N-14°59′W; TP: 5100; rice and banana farming, irrigated crops, some urbanized habitat, no access to running water). In this region, the climate is Sahelian, with annual rainfall between July and September. Mean temperature ranges between 20°C and 30°C during the cool season (November to February) and between 25°C and 38°C during the warm season (March to October). In this study, stool samples were collected from 93 symptomatic and asymptomatic children (Agniam = 17; Fanaye = 38; Niandane = 38) presenting or not gastrointestinal disorders. These boys (n = 46) and girls (n = 47) were aged 6 to 10 years (median age 8.4; quartiles: 6.6; 9.5) (Table  1). Children presented no symptoms of severe illness and/or fever and were subject to a questionnaire that was designed to collect clinical data especially regarding recent diarrhea, vomiting, and abdominal pain. For storage and transport at -20°C to Lille, stool samples were added in Stool Transport and Recovery (S.T.A.R.) buffer (Roche) in a ratio of 1:3 according to the manufacturer’s recommendations then homogenized by shaking.

^aAccording to the standard terminology including 17 STs 2 ; (Sc) and (P): non qPCRs using the primer pair described by Scicluna et al. 40 and Poirier et al. 20 , respectively.

^bDetermined in the common region of clones belonging to the same ST and sequenced for each sample.

Samples stored in S.T.A.R. buffer were stirred and then centrifuged for 1 min at 1,000 g. Total genomic DNA was extracted from 200 μl of the cleared stool supernatant using the QiaAMP DNA Stool Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. For each sample, 5 μl of extracted DNA was submitted to non-quantitative Polymerase Chain Reactions (non-qPCRs) as previously described using two independent pairs of Blastocystis sp.-specific primers designed by Scicluna et al. 40 and Poirier et al. 20 , both targeting the small subunit (SSU) rDNA coding region. The respective 600 bp- and 320 bp-amplified domains have been shown to contain sufficient information for accurate subtyping of Blastocystis sp. isolates 20 40 . Non-qPCR amplifications were carried out in 50 μl according to standard conditions for Platinum Taq High-Fidelity DNA polymerase (Invitrogen). For each DNA sample, the non-qPCR product with the highest intensity on agarose gel was purified and sequenced. Direct sequencing of few non-qPCR products generated mixed signals that could reflect infections by different STs or was of poor quality probably due to a too small amount of parasite DNA matrix. Because of the difficulty in interpreting the corresponding DNA sequencing chromatograms, same samples were re-analyzed by non-qPCRs and positive bands of the expected sizes were purified and cloned as previously described 41 . Three positive clones containing inserts were arbitrarily selected for each sample and sequenced on both strands. The DNA samples negative by non-qPCRs were subsequently amplified using the real time qPCR assay developed by Poirier et al. 20 . Due to the high sensitivity of this detection method, various controls were performed to prevent artifacts related to contamination through different sources: DNA extraction controls (isolation of DNAs without stool and from a Blastocystis sp.-negative stool) subsequently used in qPCR assays and negative (DNA matrix replaced by water) and positive (DNA obtained from a Blastocystis sp. ST4 culture) qPCR controls. The expected 320 bp-amplified variable region of the SSU rRNA gene was directly sequenced for subtyping. The SSU rRNA gene sequences obtained in this study have been deposited in GenBank under accession numbers KF848509 to KF848624. Obtained sequences were compared with all Blastocystis sp. SSU rRNA gene sequences available from the National Centre for Biotechnology Information (NCBI) using the BLAST program. STs were identified by determining the exact match or closest similarity against all known Blastocystis sp. STs according to the updated classification by Alfellani et al. 2 .

In this study, fecal samples were collected from 93 children, 46 boys and 47 girls, aged 6 to 10 years and living in 3 villages situated in the north of Senegal. This cohort consisted of two groups: one symptomatic composed of 54 children presenting variously with abdominal pain, vomiting, and diarrhea and one asymptomatic including 39 other children (Table  1). All 93 samples were screened for Blastocystis sp. by non-qPCR and qPCR methods using specific primer pairs amplifying distinct domains of the SSU rDNA coding region 20 40 . Non-qPCR products were directly sequenced or subsequently cloned whether the resulting sequence chromatogram showed double or poor quality signals. In the latter case that involved 32 samples, three positive clones containing inserts were arbitrarily selected and sequenced. Overall, 71 samples were identified as positive for Blastocystis sp. by non-qPCR methods. The remaining 22 samples negative by non-qPCR nevertheless proved to be positive for the parasite by qPCR. Thereby, by means of these different molecular approaches, Blastocystis sp. was recognized in 100% of stool samples of this Senegalese population. To our knowledge this represents the highest prevalence ever recovered worldwide for this parasite.

In almost all molecular epidemiological studies conducted to date, only one nucleic acid-based diagnostics methodology was used to identify and subtype Blastocystis sp. isolates in stool samples. Briefly, PCR screening of fecal samples can be achieved using ST-specific sequence-tagged-site (STS) primers or genus-specific primers with subsequent sequencing for ST identification. Several regions in the SSU rDNA gene have been proposed by different authors for sequencing 24 but the "barcode" region designed by Scicluna et al. 40 has been extensively used as in the present study. Advantages and limitations of these two PCR approaches were recently largely discussed 24 42 and a comparative study using both techniques recommended barcoding as the relevant method for Blastocystis sp. diagnosis and subtyping 43 . However, as confirmed from our data, direct sequencing of non-qPCR products may be unsuccessful in the case of mixed infections with more than one Blastocystis sp. ST in the same sample. Therefore subtyping of isolates in mixed infections requires cloning of the non-qPCR product and sequencing of several clones as previously suggested 44 . In addition, about 24% of the samples have not been diagnosed as positive for Blastocystis sp. by non-qPCR methods in our study. Hence, at this stage of our screening, the prevalence of the parasite was 76%, which greatly underestimated its true prevalence as determined in a second step by qPCR (100%). Consequently, our study confirmed the impact of detection methods to ascertain the actual prevalence of Blastocystis sp. in human populations and the interest and higher sensitivity of qPCR assays in comparison to non-qPCR methods for epidemiological surveys.

Although the prevalence of Blastocystis sp. never reached 100% in any cohort studied so far, it was described as very high in some developing countries specially, in children. By using non-molecular standard methods, the prevalence of the parasite already exceeded 40% in communities in Venezuela 45 , Peru 46 , Colombia 7 , and in school children or orphanage in Indonesia 6 , Thailand 47 , and Lebanon (El Safadi unpubl.). Moreover, the prevalence of Blastocystis sp. was around 60% in immigrants in the city of Naples (southern Italy) who came for a large majority from northern and western African countries 48 . In Libya, the prevalence of the parasite has been previously reported close to 30% 49 . In Senegal, although the prevalence of Blastocystis sp. had never been determined before, intestinal parasites (protozoa and helminths) represent for a long time a major public health problem. Indeed, the global prevalence of intestinal parasites in children ranged from 32% to 84% depending on the study areas 50 51 52 53 . Such impact of intestinal parasites previously reported in the Senegalese population is clearly in accordance with the pattern found for Blastocystis sp. in the current study.

In all the children, the expected SSU rDNA domains targeted by non-qPCR and qPCR methods were amplified and successfully sequenced. Each of the new SSU rDNA gene sequences obtained in this study showed very high similarity (98% to 100%) to homologous sequences of the other Blastocystis sp. isolates reported so far. The comparison with representative sequences of all known Blastocystis sp. STs allowed the direct subtyping of the new isolates (Table  1). As stated above, for 32 of the 93 samples, cloning of the non-qPCR product was necessary for subtyping and 3 positive clones were arbitrarily selected and sequenced. For 14 of these 32 samples, the three sequenced clones were identical (Table  1). For 10 other samples presenting infection by a single ST, clones showed 1 to 14 nucleotide differences between them. As previously suggested 40 41 , these nucleotide differences may reflect either coinfection with two isolates of the same ST or sequence variations between SSU rDNA gene copies within the same isolate 23 . Interestingly, the higher number of nucleotide difference between the clones (11 to 14) was observed in children infected by ST3. Recently, by analyzing and comparing respectively SSU rDNA sequences or data from multilocus sequence typing, Meloni et al. 44 and Stensvold et al. 54 revealed substantial intra-ST diversity in ST3 and more limited intra-ST2 and intra-ST4 variations. In either case, we have considered in the following analysis of our data that two different clones of the same ST identified in the same patient derived from the same isolate of Blastocystis sp. For the last 8 samples, mixed infections with two (6 samples) or three STs (2 samples) were identified (Table  2). The majority of the subtyped samples included in the present study (85/93) represented single infections. With the addition of 8 mixed infections consisting of two or three different isolates, we analyzed a total of 103 isolates. As summarized in Table  3, ST3 was the most common Blastocystis sp. ST in our Senegalese population (49.5%) followed by ST1 (28.2%), ST2 (20.4%) and ST4 (1.9%). The distribution of STs was almost similar between the three villages with a majority of infections attributable to ST3. The only significant difference was that ST1 followed at position 2 in Niandiane and Agniam Towguel and at position 3 in Fanaye Diery.

The distribution of Blastocystis sp. STs in Senegal is quite similar to that observed in a majority of countries all over the world with a predominance of ST3 followed by ST1 then ST2 or ST4 5 . If we focused on the studies previously conducted in Africa, the only available reports concerned the ST distribution in Egypt 21 37 38 , Libya, Nigeria, Liberia 2 , and Tanzania 39 , the latter survey only including 6 isolates (Figure  1). Globally, ST1, ST2, and ST3 represent around 82% of all 369 African isolates subtyped so far including those of the present study. ST3 is highly prevalent in African countries with an average of 46% of the 369 isolates. Consequently, because ST3 is rarely found in animals, a large-scale human-to-human transmission might likely explain the predominance of this ST in African countries as also in the rest of the world. Three studies conducted in different cities in Egypt and that in Liberia have identified ST3 as the predominant ST, as was true for our study population. However ST1 was detected with the highest frequency in Libya and Nigeria, and ST2 in Tanzania. Since ST1 and ST2 are frequently found in various animal groups including livestock 2 4 55 , it can be hypothesized that a not yet evaluated proportion of human infections by both STs are of zoonotic origin. Interestingly ST4 was not detected in the studies reported in Egypt, Libya and Tanzania and its prevalence is quite low in Senegal (1.9%). Although ST4 was relatively frequent in Liberia (12%) and Nigeria (14%), these data confirmed that ST4 is much less frequently detected or absent in Africa while it is commonly found in Europe 5 34 . However, the reasons for the heterogeneous geographical distribution of ST4 remain unknown and its reservoir hosts have to be clarified. Additional STs including ST6 and ST7 were commonly found only in two studies carried out in Egypt 21 38 . Indeed, these STs were not detected in Senegal, Liberia, Nigeria, and Tanzania as in a third survey in Egypt 37 and only one isolate belonging to ST7 was identified in Libya. Both STs are usually hosted by birds (avian STs) 56 and rarely found in mammals. This was confirmed in the present study. Given their apparent host specificity, it is highly likely that human infections by avian STs are of zoonotic origin.

The prevalence of mixed ST infections was 8.6% (8/93 samples) in our Senegalese cohort. This percentage is probably underestimated but roughly similar to that calculated (6%) by taking into account subtyped isolates from all epidemiological studies published to date (191 of 3171 samples) and listed by Alfellani et al. 5 . Mixed ST infections were found in all three villages of our study but with different proportions. In Agniam Towguel, only one mixed infection by 3 STs was found (ST1/ST2/ST3) (5.9%) while 4 mixed infections by 2 or 3 STs (10.5%) (ST2/ST3, ST3/ST4 and ST1/ST2/ST3) were recorded in Fanaye Diery. Regarding Niandane, 3 mixed infections by 2 STs (7.9%) were identified (ST1/ST2, ST1/ST4, and ST2/ST3) (Table  2). Mixed infections mostly corresponded to ST1, ST2, and ST3 isolates and were more likely to occur considering the high prevalence of each of these STs in our cohort. Curiously, the only two ST4 isolates of our study were identified in the context of mixed infections and not as part of single infections.

According to recent epidemiological reports, links between ST and pathogenicity are still matter of debate 8 29 33 34 35 36 41 . To highlight a possible relationship between ST and major intestinal symptoms associated with Blastocystis sp. infection (diarrhea, vomiting, and abdominal pain), the ST distribution was evaluated in our cohort of 93 symptomatic or asymptomatic children. Interestingly, the majority of children infected by Blastocystis sp. (54/93 or 58%) presented various gastrointestinal disorders as previously observed in a pediatric unit in Turkey 57 . In the following analysis, symptomatic (6) and asymptomatic (2) children presenting mixed infections by two or three STs were removed. Thereby the remaining 48 isolates from symptomatic children and 37 isolates from asymptomatic children were considered (Table  4). The most dominant ST in both groups was ST3, and 63% of the ST3 isolates were found in symptomatic individuals. Predominance or high prevalence of ST3 in symptomatic groups of patients was also shown for instance in France 41 , Australia 8 , Turkey 57 58 , and Iran 35 . Regarding ST1 and ST2 isolates, both STs were detected with equal frequency in the symptomatic and asymptomatic groups of our cohort (Chi square test, p = 0.845). However, ST1 was over-represented in groups of symptomatic individuals in previous epidemiological studies in China 59 , Turkey 60 , Iran 35 , Egypt 21 , and Lebanon 36 . Moreover, ST1 was the most prevalent ST in some cohorts presenting with IBS 38 61 and was shown to be pathogenic in rat models of blastocystosis 21 . In contrast, ST2 was not usually overrepresented in groups of symptomatic patients from different countries and Dogruman-Al et al. 57 suggested that ST2 was a non-pathogenic ST. However, in our study, the proportion of ST2 isolates in symptomatic patients was far from negligible. Moreover, all ST2 isolates identified in French and Australian sampling were from symptomatic individuals 8 41 . To complete our observations, it can also be underlined that the only two ST4 isolates found in our study were associated with symptoms but in the context of mixed ST infections (associated to ST1 or ST3). Therefore, the pathogenic potential of ST4 could not be assessed from our epidemiological data. However, a high prevalence of ST4 was recently found in Danish patients presenting with acute diarrhea 33 and Spanish symptomatic individuals 62 . Globally, no association between ST and symptom status was statistically significant from our present data.

The fecal-oral route is the main mode of transmission of Blastocystis sp. like the other common gastrointestinal parasites 1 . To reach a prevalence of 100% as in the present study, it is reasonable to argue that infection of Senegalese children by the parasite occurred through multiple transmission sources. Although we did not had the opportunity to collect and analyze environmental (mainly water samples) and animal samples, it appears clear that individuals living in small communities in a rural area with modest socio-economic status (average income of 0.5 $ per person per day in these Senegalese villages), poor hygiene sanitation, close contact with domestic animals and livestock, and water supply directly from well and river as in our study greatly increase their risks of infection by Blastocystis sp. Indeed, recent studies have revealed a human-to-human transmission coupled with a high prevalence of the parasite in a close community of girls in Thailand 13 as well as in a rural village in Nepal 14 whose description and environment (animals reared in or next to villager dwellings and rivers as water sources of the community) are quite similar to those of Senegalese villages. In this context, zoonotic transmission of Blastocystis sp. would be another significant source of infection due to the wide range of animals potentially infected with the parasite and in close contact with humans 2 4 55 56 . However the impact of this mode of transmission through contact with contaminated animal fecal materials coupled with poor hand hygiene of villagers remains difficult to quantify therefore that it involves STs frequently found in humans (ST1 to ST4) as in the present study. Finally, there is a concern that water should be a main source of infection by Blastocystis sp. although large-scale waterborne outbreaks involving the parasite have not yet been documented. Ability of Blastocystis sp. cysts, the only transmissible stage of this parasite to survive for long periods of time in the environment together with their low size 1 15 represent some factors that favor waterborne transmission of the parasite. In 2006, Blastocystis sp. has also been added by the World Health Organization to the list of waterborne parasites 63 . Blastocystis sp. was previously reported in drinking water, rivers, tanks, and sewage samples 11 12 14 15 64 . Interestingly, in these latter studies, the same STs were identified in both the water samples and the stools of consumers providing evidence for water transmission. In the Senegalese villages of the present study, river and wells represent the main water sources for drinking and cooking purposes as well as for bathing of the population. Consequently, water sources of this community potentially contaminated with human and animal feces, may largely facilitate transmission of Blastocystis sp. and should be considered as serious potential threats to consumers. Further, the analysis of fecal samples from animals living in or near the Senegalese villages and of water samples would help to confirm their epidemiological importance as sources of infection and contamination for inhabitants.