Contrasting microbiota profiles observed in children carrying either Blastocystis spp. or the commensal amoebas Entamoeba coli or Endolimax nana

Recent studies have shown how intestinal parasites can modulate gut microbiota. This observation is not surprising since the human intestinal lumen, like any other niche, is a battlefield of microbial competition, and Eukaryotes can affect bacterial populations. Intestinal pathogenic protist has been associated with reshaping the microbial community structure; however, the interactions between the colonic bacterial communities and parasites like Blastocystis spp., Entamoeba coli, and Endolimax nana have been poorly studied. In this work, we studied the distal intestinal bacterial microbiota of 49 children attending 7 public daycare centers in Medellin, Colombia, and compared the bacterial microbiota structure in the presence or absence of the protists Blastocystis spp., E. coli, and E. nana. Parasite colonization was associated with an increase in bacterial richness. Moreover, Blastocystis spp. presented a positive relationship with Prevotella, since this bacterium was selectively enriched in children carrying it. Remarkably, the E. coli colonized children showed a microbial profile that was closer to uninfected controls, although some bacterial taxa displayed to be enriched. This is the case for Akkermansia, which showed to be favored in E. coli colonized individuals, while notably reduced in the Blastocystis spp. parasitized group.


Results
The 16S metataxonomic experiment started with 50 k reads per library. After MOTHUR processing, the read number was rarefied with the totalgroup function to an average of 12,249 clean reads per sample, and they ranged between 12,326 and 12,526. The coverage analysis showed that at least 97.4% of the expected OTUs were observed within our analysis individuals and went up to 99% in several samples. The observed OTU (3%) count varied between 193 and 551 within all tested individuals (Supplementary Table 1).
Effect of harboring protists. Initially, we tested the differences between children with no detectable parasites on the stool microscopical analysis (NPDM group) and those confirmed to harbor either Blastocystis spp. (Blasto group), Entamoeba coli (E_coli group) or Endolimax nana (E_nana); these last three were grouped in the PROTIST-infected category. As shown in Fig. 1, the Observed OTU median values rise from 281 in NPDM controls (negative) to 426 in the protist-infected ones (positive), with a p-value = 3.785e-05, using the Kruskal-Wallis rank-sum test. Similarly, the Chao1 and ACE indices were significantly higher, using the same test, in infected individuals compared to controls, rising from 439 to 919 (p-value = 0.0001335), and from 453 to 821 (p-value = 0.0003182), respectively. The Shannon and Simpson indices for both groups were similar and showed Scientific RepoRtS | (2020) 10:15354 | https://doi.org/10.1038/s41598-020-72286-y www.nature.com/scientificreports/ no statistically significant differences with the Kruskal-Wallis test. Furthermore, sex did not show any significant difference with the richness/diversity indices studied, when the Kruskal-Wallis test was applied.
Regarding age, there was a weak significative correlation with the Shannon index (Spearman correlation p-value = 0.02515, rho 0.3196735), which was not the case for the Simpson index, that showed no significant results. The 'lm' model of Age and Shannon gives shannon = 0.1Age + 2.8 with (p-value = < 2e−16, for the intercept and p-value = 0.0596 for the slope). Our group of selected children, most are in the range of 2 to 5yo, life period in which the intestinal microbiota tends to stabilize.
Finally, Fig. 2 shows the ordination plot using NMDS distance, which mostly separates the individuals of both groups. We determined which variables most strongly affected the structure of the children gut microbiota using a permutational multivariate analysis of variance (PERMANOVA) test of the Bray-Curtis dissimilarities. Again,  Bacterial microbiota profiles associated with Blastocystis spp., Entamoeba. coli or Endolimax nana. The next step in our analysis was oriented to discriminate if there was any differential relationship between the specific protist harbored by each individual and the microbiota profiles. To do so, we separated the infected children into three groups regarding the detected parasite: Blastocystis spp., E. coli, or E. nana; and compare them with the NPDM controls. Following the previous observations, the median values of the richness indices observed OTUs, Chao1, and ACE were augmented in all the infected groups (Fig. 3). The species richness (Observed OTUs) showed statistical significance, applying the Kruskal-Wallis rank-sum test, among all four tested groups (p-value = 0.0004175) and proved to be significantly higher in pairwise comparisons with the Wilcoxon rank-sum test between the Blastocystis spp. (median 413) and E. coli (median 473) parasitized groups (Blasto and E_coli, p = 0.0043 and p = 0.0010, respectively) compared to the control group (NPDM, median = 281). Additionally, the Kruskal-Wallis test showed statistical differences within the groups in all measured indices (Simpson, p-value = 0.004808; ACE, p-value = 0.002159; Chao1, p-value = 0.001095; Shannon, p-value = 0.03467 (Age-adjusted)). The Pairwise comparisons using the Wilcoxon rank-sum test indicated that the Chao1 index was only significantly different between the controls (median = 439) and the Blasto (median = 1,162, p = 0.0016) groups. The ACE index showed similar results in these two groups (NMPD, median 453; vs. Blasto, median 951; p = 0.0019). Additionally, this index showed significant differences among the NMPD controls vs. E_coli group (p = 0.0313). The Shannon and Simpson indices median values showed slight variations but were not statistically significant with the Wilcoxon rank-sum test when the controls were compared with the parasitized groups ( Fig. 3). Again, the ordination analysis plot shows a segregation pattern of most of the non-parasites controls versus the Blastocystis spp., E. coli, or E. nana groups. Among the parasitized groups, it is not possible to observe an apparent clustering of the individual based on each protist species (Fig. 2).
In general, the most copious Phyla were Bacteroidetes and Firmicutes, followed by Proteobacteria, Actinobacteria, and Verrucomicrobia. This last Phylum was observed significatively more abundant in the E_coli (median = 470) group compare to the Blasto group (median = 0) (Wilcoxon rank-sum test, p = 0.019). Remarkably, two individuals of the NMPD control group presented a significant proportion of Fusobacteria (Supplemen-taryFigure 1). At the family taxonomic category, we observed that the taxa Prevotellaceae, Ruminococcaceae, Bacteroidaceae, and Lachnospiraceae were the most prevalent across all samples. It is noteworthy that the bars of Prevotellaceae are more prominent in the Blasto group samples compare to NPDM controls. Conversely, the control group seems to have a higher relative proportion of Ruminococcaceae compared to Blastocystis spp. parasitized individuals (SupplementaryFigure 2).
Scientific RepoRtS | (2020) 10:15354 | https://doi.org/10.1038/s41598-020-72286-y www.nature.com/scientificreports/ Prevotellaceae and Verrucomicrobiaceae also showed the best statistical significance in the pairwise comparisons of the relative abundances using the Wilcoxon rank-sum test. For Prevotellaceae, the significative difference (p-value = 0.01638) was observed between the Blasto (median = 5,912) and the NMPD controls (1696). In the case of Verrucomicrobiaceae, median values drop from 470 in the E_coli group to 0 in the NPDM controls (p-value = 0.019). Porphyromonadaceae also showed significant results between the Blasto and E_coli groups (p-value = 0.021), while in Veillonellaceae the significative differences were observed between the NPDM controls and the E_coli group (p-value = 0.031).
At the genus category, Prevotella, Bacteroides, and Faecalibacterium were, in order, the dominant microoganisms over all individuals (SupplementaryFigure 3). These results indicate changes in Prevotella proportions, which seem to be enriched in the Blasto group. One more detailed and quantitative analysis on Prevotella showed an increase in the median values of normalized counts of this anaerobe in the Blasto (5,367) group compared to the controls (1696) that were statistically significative (p = 0.011) (Fig. 5A).
Prevotella is enriched within the individuals harboring Blastocystis while Akkermansia seems to be favored with Entamoeba coli colonization. To further confirm the previous findings on the changes in the relative abundance of the mentioned microbe families, a LEfSe analysis was performed, aiming to identify marker microorganisms within the tested groups. After filtering the results excluding the unclassified organisms and LDA score above 3, 15 OTUs were kept with significant p values (Table 1). Prevotella (OTU00001) showed the highest abundance shift with an LDA score of 5.32 towards the Blasto group. This group was also enriched with the bacteria Haemophilus, Holdemanella, and Butyricicoccus.
The second highest score was obtained for Akkermansia (LDA 4.57) in the E_coli group, which was also enriched with Coprococcus and Alistipes. The control group had two enriched biomarker organisms, both from the Firmicutes phylum, Blautia, and Streptococcus. The case of Akkermansia is quite impressive since it is not included in the top ten most abundant organisms, although it was the second most abundant biased OTU in the tested groups. To gain insights into this particular group, the normalized counts of Akkermansia were extracted and plotted (Fig. 5B). These results display a reduced number of counts in the Blasto group with a median value of zero, while the control and the E_coli group, 23 and 470, respectively. The Kruskal-Wallis test showed statistical

Discussion
Children distal intestinal microbiota tends to stabilize and begin to be more adult-like around 3-yo 8 . In our selected children group only two were younger than 2-yo, most are in the range of 3 to 5-yo, when intestinal microbiota starts to stabilize. This can explain why, albeit the described association of gut microbiota diversity and age in children, a weak correlation with the Shannon diversity index was observed. Additionally, since these children receive most of their meals in the daycare center, allowing them to have a more similar intestinal microbiota due to the effect of a normalized diet. Nonetheless, despite the influence of age within the intestinal  www.nature.com/scientificreports/ microbiota in the studied children, our results showed that Protist colonization have a relevant impact on the intestinal microbial community in children.
Eukaryotic parasites are major competitors in the microbial world due to bacterivorous activity or direct competition for nutrients 27,53,54 . The effect of intestinal nematodes and protists on the human gut bacterial microbiota has been proved in several studies, and some of them show, as a common trend, that bacterial richness is increased in individuals that carry intestinal parasites. However, diversity is not always augmented as well [20][21][22][23][28][29][30]52 .
In the present study, we observed that all the studied protist, Blastocystis spp., Entamoeba coli, and Endolimax nana; were associated with a significant increase in bacterial richness, with median values that almost doubles the control group. We cannot conclude if this is cause or effect, but these findings are in concordance with previous reports for similar studies on the relation of Blastocystis spp. with the intestinal microbiota. For Entamoeba coli and Endolimax nana it was not possible to find previous scientific publications addressing this topic. To the best of our knowledge, this is the first report that observes the relation of the intestinal gut microbiota and these two amoebae using a 16S metataxonomic approach. It is essential to highlight that no definitive conclusion can be drawn from the individuals infected with Endolimax nana due to the low number of children included in this group, only 4. Endolimax nana is an intestinal amoeba of humans that has a cosmopolitan distribution, most likely as a commensal or nonpathogenic protozoon, with an estimated global prevalence in healthy individuals of 13.4%. The scientific evidence to date is inconclusive in terms of its host specificity, epidemiology, morphology, taxonomy and genetic diversity 55 . Although some authors suggest that E. nana feed exclusively on bacteria and that it could have a pathogenic potential, with case reports of parasitized patients suffering arthritis 56 , intestinal symptoms [57][58][59] , and urticaria 60,61 , there is not enough evidence that supports this statements. Therefore, studies focused on the parasite genetic variability and crosstalk interaction with the microbiota and the immune system are needed to provide data that clarify the effects of this protozoan in the human intestine. Furthermore, the fecal-oral transmission suggests that Endolimax can be used as a biological marker suitable for the hygiene measures of the population and fecal contamination of food or water 62 .
Blastocystis spp. has been associated in several studies with increased bacterial diversity in western European adults and Mexicans 30,46 . In our study in Colombian children, although we found an increased richness in the colonized individuals, no significant differences were observed in the diversity indices. This controversial finding might be attributed to a differential response in the child or adult intestinal microbiota to the Blastocystis challenge. More studies in children need to be performed in order to understand in detail this phenomenon.
Since its first observation in 1849, Blastocystis spp. was initially described as a pathogenic parasite being formally termed in early 20th Century 63,64 . After several decades of debate about its classification and host preferences, in the second half of this century, it was generally accepted as a pathogenic protist for humans that can cause diarrhea and abdominal pain 65 . However, several researchers have raised concerns about the evidence that supports the pathogenicity of Blastocystis spp. in humans, and its clinical significance remains controversial [66][67][68] . In vitro and in vivo studies demonstrate pathogenic potential but also show considerable inter and intra subtype variation, which provides a possible explanation on the conflicting reports on clinical significance. Blastocystis spp. have intestinal immunomodulatory effects and can release proteases that affect the integrity of the epithelium. This situation might facilitate colonization by other enteric pathogens either directly or by the resultant changes in the gut microbiota 69 .
Our findings unveil that children carrying Blastocystis spp. display a different microbial community compared to uninfected controls with a tendency to a Prevotella-driven enterotype. This is in concordance with similar studies carried out in adults around the world [28][29][30] . Blastocystis spp. showed a significant shift in the Prevotella proportion enriching it, favoring an enterotype switch. Prevotella strains are associated with plant-rich diets, including fibers, simple sugars, and carbohydrates, suggesting that it is a beneficial microbe 70 . However, Prevotella in the gut has also been linked with inflammatory diseases, which made it difficult to predict its behavior in any given gut ecosystem 70 . Our results are similar to those found in healthy children from several developing countries, who had a gut microbiome dominated at the genus rank by Prevotella [71][72][73] . High species diversity at this genus could be related to its different effects on host health.
Andersen et al. found that Blastocystis spp. colonization was positively associated with species richness, and this parameter was negatively correlated with the Bacteroides-driven enterotype. The authors concluded that Blastocystis spp. establishment in the intestine probably depends on the activity of certain types of bacteria that are generally not present in individuals with low richness colonic microbiota 74 . Since Blastocystis spp. is an obligated anaerobe, in order to survive, it should favor the predominance of bacterial taxa that maintains a strict anaerobic environment in the gut lumen 75 .
An interesting finding was that Akkermansia, a bacterium effective in increasing mucus thickness and gut barrier function, was reduced in the Blastocystis spp. infected group, suggesting that this protist poses an unfavorable condition to this beneficial bacterium. This reduction has been previously described, and it was related to specific Blastocystis subtypes, with an inverse correlation of subtypes 3 and 4 with Akkermansia, suggesting differential associations between subtypes and host health 29 . Other authors have also shown that Blastocystis spp. infection also leads to changes in the abundance of other groups of bacteria, reducing Bacteroides, and increasing Prevotella 28,29,75 .
Entamoeba coli colonized children showed a bacterial community that closely resembles the control group without a Prevotella-driven enterotype. We also found an increase in the relative abundance of the beneficial bacterium Akkermansia in this group, contrary to the pattern observed in the Blastocystis spp. infected children. This commensal amoeba probably contributes to maintaining gut favorable conditions to beneficial bacteria, like Akkermansia. Although it is challenging to fully interpret the role of any microorganism in a complex community such as the gut microbiota, our results indicate that commensal protists like Entamoeba coli could be related to a healthy status.
Scientific RepoRtS | (2020) 10:15354 | https://doi.org/10.1038/s41598-020-72286-y www.nature.com/scientificreports/ Nowadays, the definition of the pathogenicity of intestinal parasites should not only be restricted to its capacity to alter or invade the intestinal mucosa, but the alteration of the healthy gut microbiota might also be a cause of disease 1,14,76,77 . The alteration profile of the distal microbiota observed in the individuals colonized by Blastocystis spp. have been associated with intestinal bowel disease, and a reduced abundance of Akkermansia is associated with diseases like Atherosclerosis 78 , ulcerative colitis 79 , appendicitis 80 , overweight and obesity 81 . From this point of view, changes in the intestinal gut microbiota seem to correlate or exacerbate several diseases, so it should be considered at the moment of defining the pathogenic capacity of a parasite. It seems clear with the actual scientific evidence that Blastocystis spp. has the power to promote the displacement of the "healthy intestinal microbiota", rendering the children more susceptible to other diseases thanks to the increase in Prevotella and the reduction of Akkermansia. By definition, a commensal parasite, like Entamoeba coli, should not affect the normal physiology of the host. However, the evidence shown in this paper add arguments in favor of the pathogenic behavior of Blastocystis spp. in children.

Methods
Population and sample selection. Children attending seven public daycare centers in the three nearby neighborhoods in Medellin, Colombia, were selected for this study. Feces samples were collected in screwcapped containers without any preservatives and then transported to the lab within a maximum period of 3 h. Samples for DNA extraction were frozen at − 70 °C for a period that did not exceed 7 days, and then DNA was extracted. The microscopical parasitological analysis was performed on the same day of the collection by observation of direct (fresh and iodine solution) and modified Ritchie concentrated stool samples. Modified Ziehl-Neelsen slides of the feces samples were also prepared and observed microscopically to detect intestinal apicomplexan parasites. Specific PCR for Cryptosporidium was performed using the protocol described by Xiao et al. 82,83 . All selected samples were negative with the test mentioned above for intestinal apicomplexan parasites.
Subjects were selected based on being positive for any of the following protists: Blastocystis spp., Entamoeba coli, or Endolimax nana. A control group negative for intestinal parasites was also included. These children received the same food (breakfast and lunch) while assisting the daycare centers and the age ranged from one to five years old (1yo, n = 2; 2yo, n = 11; 3yo, n = 16, 4yo, n = 9, 5yo, n = 11). By sex, the distribution was 17 females and 32 males. Enrolled children were classified into four groups: NPDM: Control group of children with no positive results for parasites (n = 25). Blasto: Children parasitized only by Blastocystis spp., no other parasites observed (n = 11). E_coli: Children parasitized only by Entamoeba coli, no other parasites observed (n = 9). E_nana: Children parasitized only by Endolimax nana, no other parasites observed (n = 4). The Protist-infect group was set adding all the children of the Blasto, E_coli and E_nana groups (n = 24). DNA extraction and 16S metataxonomic experiment. DNA extraction was carried with the STOOL DNA kit NORGEN (Canada). Extracted DNA was quantified using UV absorption and the Picogreen fluorescent method. The DNA quality was assessed by gel electrophoresis and control PCR amplifying the complete eDNA 16S gene with universal primers 27F and 1492R. For the metataxonomic experiment, the primers Bakt_341F CCT ACG GGNGGC WGC AG and Bakt_805R GAC TAC HVGGG TAT CTA ATC C, that amplify the V3/V4 region, were used. The 16S metataxonomic experiment was hired to MACROGEN, Korea. An average of 100.000 reads per library was generated in the MiSeq platform with PE reads of 300 bases. For further bioinformatic analysis 50,000 reads were randomly selected for each sample with the program SEQTK (https ://githu b.com/lh3/seqtk ). Bioinformatic analysis. Amplicon processing was carried out with MOTHUR V1.42 84 . Following the MiSeq standard operating protocol provided by the authors (https ://www.mothu r.org/wiki/MiSeq _SOP). Briefly, Amplicon forward and reverse read were merged and those containing Ns or homopolymers larger than 6 bases were removed. Then, the reads were mapped to the SILVA database, and only those that mapped to the V3/V4 region were kept. Chimera removal was performed with VSEARCH 85 . Reads per library were rarified with the totalgroup strategy to an average of 12,249 clean sequences. OTUs supported by less than 4 reads were excluded. A BIOM file was prepared with the MOTHUR function make.biom, for further statistical analysis in R language. Statistics analysis. The statistical analysis was performed with package PHYLOSEQ in the R environment.
The data was imported as a BIOM file generated with MOTHUR. Alpha diversity statistics were calculated and plotted in boxplots. The ordination plot was generated with NMDS and BRAY distances. Read counts per library were normalized with median sequencing depth. With these normalized counts, we compared taxa abundance at Phylum, family, and genus categories and generated the bar plots of the top ten most abundant taxa, the remaining groups were gathered in the others category. Statistical significance of the differences in the median values was performed with Kruskal-Wallis and pairwise Wilcoxon rank-sum test. The age effect on the Shannon index was assessed using the 'lm' function. It was adjusted by subtracting the age contribution to the index.
Permutational multivariate analysis of variance (PERMANOVA) was performed using the Adonis function in the vegan library of R with the Bray-Curtis dissimilarity matrix.
Ethics statement. The ethical clearance of this study was followed by the ethics of the Helsinki declaration and resolution No. 008430 of 1993 from the Ministry of Health from Colombia. The study was approved by the Ethics Committee from Sede de Investigación Universitaria, Universidad de Antioquia, under the official document No. 14-06-564. Parents or legal guardians of all the enrolled individuals signed informed consent.