Study site, population, questionnaire, and fecal sample collection

The study was conducted on family and industrial pig farms located in multiple municipalities across the states of Rio de Janeiro and Minas Gerais, Brazil. A total of 15 farms were included, of which 11 were family-ownedand 4 were industrial.

In the metropolitan region of Rio de Janeiro, two farms were included in each of the municipalities of Rio Bonito and Maricá, and one farm each in Itaboraí, Tanguá, and Silva Jardim. Three farms were included in the mountain region (two in Cachoeiras de Macacu and one in Nova Friburgo). Two farms were included in the coastal lowlands (Casimiro de Abreu and Saquarema), and one in the Médio Paraíba region (Pinheiral). Due to the limited number of industrial farms in Rio de Janeiro, two additional farms from neighboring municipalities in Minas Gerais (Rio Pomba and Barbacena) were included.

All participants either resided on or worked at pig farms. The study included pig handlers (n = 36) and their family members (n = 11), recruited through convenience sampling between July 3, 2023, and July 30, 2024. Written informed consent was obtained from all adult participants. For minors, assent forms were completed with authorization from their parents or legal guardians. Individuals who did not provide formal consent or who did not submit fecal samples were excluded from the study.

Participants who signed the informed consent completed a semi-structured questionnaire designed to collect general and sanitary -related information. The questionnaire comprised 23 open- and closed-ended questions, which were answered verbally and individually by each participant. For minors (children and adolescents), assent was provided by the participants, while parents and/or legal guardians signed the informed consent form and were responsible for completing the questionnaire. Responses were recorded in an online form (Jotform®) by the research team to minimize misinterpretation and issues related to visual impairment or illiteracy. Each participant was assigned a unique identification number to preserve anonymity.

After completing the questionnaire, study participants received kits for fecal sample collection. Each kit contained two collection containers: one without chemical preservative, which was stored under refrigeration (4°C), and another containing 70% ethanol, which could be kept at room temperature. Additionally, the kit included two wooden spatulas and a technical guide with detailed instructions on sample collection and storage. Verbal instructions on collection procedures were also provided at the time of kit delivery. Collections were scheduled in advance to ensure samples were refrigerated for no longer than two days.

Pig fecal samples were collected either directly from the rectal ampulla (using a glycerin-lubricated palpation glove) or immediately after spontaneous defecation. Following collection, all samples were individually labeled with identification numbers.

All human and pig fecal samples were placed in insulated containers and transported to the Parasitology Laboratories, Biomedical Institute, Federal Fluminense University, Niterói, Rio de Janeiro, Brazil.

Laboratory and microscopic parasitological analyses

Human samples were processed separately from pig samples to avoid cross-contamination. Accordingly, human samples were processed in dedicated rooms, using commercial kits and other reagents used exclusively for these samples. Details on the collection and processing of pig fecal samples have been published previously [21].

In the laboratory, human fecal samples were subjected to on the same day to direct examination, followed by qualitative coproparasitological methods, including Lutz’s sedimentation [22] and Sheather’s centrifugal flotation [23] modified by Huber et al. [24], and the quantitative FLOTAC technique [25]. Slides from each method and the FLOTAC chambers were examined under an Olympus® BX41 light microscope at 100 × magnification, with confirmation at 400× when necessary.

DNA extraction, polymerase chain reaction and phylogeny analyses

Fecal material from one pig per farm showing ciliated eukaryotic stages (cysts and trophozoites), along with all fecal samples from pig farmers and their family members, were subjected to DNA extraction. For DNA extraction from fecal samples, 200 mg of each sample were processed using the QIAamp Fast DNA Stool Mini Kit (Qiagen®). All extraction steps followed the manufacturer’s instructions, with two modifications suggested by a previously published protocol [26].

Extracted DNA was amplified by polymerase chain reaction (PCR) using the Hot Start Platinum Master Mix (Invitrogen®) and primers B58D (5’ GCTCCTACCGATACCGGGT 3’) and B58RC (5’ GCGGGTCATCTTACTTGATTTC 3’), which amplify the ITS1–5.8S–ITS2 region of ciliates, generating fragments of approximately 500 bp. Amplification followed the protocol described by Ponce-Gordo et al. [27]. A total of 5 µL of extracted DNA from each sample were used in the reaction. PCR products were visualized on a 1.5% agarose gel stained with GelRed® (Biotium®) and bromophenol blue (LGC®). PCR products were purified using ExoSAP-IT (Invitrogen®) and sequenced on an ABI 3730xL automated sequencer at the IOC/Fiocruz platform. Negative controls (ultrapure water) were included in all DNA extraction and PCR steps.

The ITS1–5.8S–ITS2 sequences obtained by Sanger sequencing were edited using ChromasPro® v.1.7.5 and incorporated into a sequence database retrieved from GenBank, comprising 255 sequences, including 243 additional Balantioides coli sequences and a set of 12 sequences from Balantidium entozoon (JQ408693.1, JQ408694.1), Buxtonella sulcata (KP016715.1, KP016716.1, JQ073387.1, JQ073386.1), Buxtonella-like (JQ073341.1, JQ073369.1, JQ073381.1, JQ073384.1, JQ073385.1), and Troglodytella abrassarti (EU680314.1), which were included as outgroups.

For phylogenetic analyses, we selected sequences identified as B. coli corresponding to the ITS1–5.8S–ITS2 region, with a length greater than 500 bp and containing variant descriptions. Sequences with ambiguous variant annotations were excluded from the dataset.

The resulting dataset was aligned using MAFFT [28] and subsequently processed with Gblocks [29] to remove poorly aligned regions, yielding a final alignment of 401 bp. The final dataset was visually inspected in SeaView (v.5.0.5) to ensure quality. Maximum likelihood phylogenetic reconstruction was performed using the IQ-TREE Web Server [30] with 1,000 bootstrap pseudoreplicates, employing the TN + F + G4 substitution model, which was automatically selected by the software.

Library preparation and metabarcoding analyses

DNA extracted from pigs (n = 15) and humans (n = 13) that showed molecular evidence of Ciliophora by PCR was sent to Genone® for sequencing on the Illumina HiSeq 2500 platform. Amplicon libraries were prepared by PCR amplification of the V3–V4 region of the bacterial and archaeal 16S rRNA gene using primers 341F (CCTAYGGGRBGCASCAG) and 806Rb (GGACTACNNGGGTATCTAAT) [31], and by amplification of the V4 region of the eukaryotic 18S rRNA gene using primers TAReuk454FWD1 (CCAGCASCYGCGGTAATTCC) and TAReukREV3 (ACTTTCGTTCTTGATYRA) [32].

Initial quality control of Illumina sequencing results was performed by the sequencing provider. Subsequent analyses were then performed in QIIME 2 software [33]. Raw paired-end sequences had adapters removed and were demultiplexed using the q2-cutadapt [34] and q2-demux plugins, respectively. Reads were merged, denoised, and dereplicated using the q2-dada2 plugin [35]. Reads with a forward–reverse overlap of <20 bp and reads shorter than 225 bp were discarded. Chimeric sequences were also identified and removed using the “consensus” method implemented in q2-dada2. Amplicon sequence variants (ASVs) recovered and separated by q2-dada2 were taxonomically assigned using the QIIME2 feature-classifier (Sklearn) trained on the PR2 database [36]. ASVs identified as potentially pathogenic were retained for downstream diversity analyses.

Data analyses

Results for Balantioides coli were interpreted through the integration of microscopic and molecular techniques. Molecular detection was performed by PCR targeting the ITS1–5.8S–ITS2 fragment, followed by Sanger sequencing, to confirm the parasite’s presence and to characterize its genetic variants.

The frequency of the parasite was calculated by dividing the number of positive samples by the total number of samples collected, and the results were expressed as percentages. All data, including questionnaire results, were stored in Microsoft Excel. Qualitative information collected from the questionnaire was tabulated and presented descriptively as percentages, with the most frequently reported responses highlighted in tables.

The nucleotide sequences of B. coli obtained by Sanger sequencing were compared phylogenetically with other sequences deposited in GenBank, considering country of origin, host species, and reported genetic variant.

Relative abundance analyses and Venn diagrams were used to explore the taxonomic composition of the overall eukaryotic and prokaryotic microbiota, as well as potentially pathogenic parasites present in the fecal samples of rural pig farmers and their family members raising pigs infected with B. coli. These plots were generated using the Python libraries matplotlib [37], plotnine [38], and pandas [39].

Microbiota dissimilarity and diversity were assessed using principal coordinates analysis (PCoA) based on Bray–Curtis distances and Weighted UniFrac metrics. Bray–Curtis considers taxon richness and abundance, whereas Weighted UniFrac also incorporates phylogenetic relatedness. PERMANOVA (999 permutations) was performed, considering p ≤ 0.05 as statistically significant. All analyses were performed using the q2-diversity plugin in QIIME 2 [33].

Ethical considerations

This study was approved by the Human Research Ethics Committees of the participating institutions: Federal Fluminense University (UFF) under CAAE number 63709022.0.0000.5243, protocol no. 5.966.227; Federal Institute of Education, Science and Technology of Rio de Janeiro (IFRJ) under CAAE number 63709022.0.3003.5268, protocol no. 6.045.672; and Federal Institute of Education, Science and Technology of Minas Gerais (IFMG) under CAAE number 63709022.0.3002.5588, protocol no. 6,072,109.

In addition, the study was approved by the Animal Use Ethics Committee of Fluminense Federal University (protocol no. 8394191222), as well as by the Animal Use Ethics Committees of the participating Federal Institute of Education, Science, and Technology Rio de Janeiro (protocol no. 03/2022) and Minas Gerais (protocol no. 02/2023).

Participant profile and microscopic parasitological results

A total of 47 individuals participated in this study, including farm owners, employees, and their family members. Of these, 36 were pig handlers and 11 were family members residing on the farm premises. Twenty participants worked on industrial farms, and 27 on family farms. Most participants (82.9%) were adult males who directly handled pigs as primary caretakers. In addition, 61.7% reported living on the farm premises and, 80.8% had previously undergone the stool examinations, with 29.7% reporting annual testing. Self-reported symptoms compatible with parasitic infections—such as abdominal pain, diarrhea, visible worms, and blood in stool—ranged from 4.2% to 23.4%. More than 80% reported previous use of antiparasitic drugs; however, over half could not recall the frequency of use (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Positivity for gastrointestinal parasites, with emphasis on Balantioides coli, based on molecular and microscopic parasitological techniques, along with general sanitary information from fecal samples of pig farmers on family and industrial farms in the states of Rio de Janeiro and Minas Gerais, Brazil.

https://doi.org/10.1371/journal.pone.0338487.t001

Untreated water sources, including wells and springs, were the most frequently reported (61.7%). Regarding vegetables, 72.3% of participants consumed them after washing with water or with solutions containing chlorine, vinegar, or lemon, and 57.4% maintained vegetable gardens on their properties. Some individuals reported using animal manure, including pig manure, to fertilize crops. More than 65% of participants reported using personal protective equipment when handling animals, and changing clothes after farm work (Table 1).

Three individuals tested positive by microscopic coproparasitological techniques: two harbored hookworm eggs (8–16 eggs/g feces), and one showed vacuolar forms of Blastocystis spp. All positive individuals were family members residing on a farm in Cachoeiras de Macacu, Rio de Janeiro. No forms compatible with the phylum Ciliophora were observed in participant samples.

Nucleotide sequences and phylogenetic analysis of B. coli in rural workers in association with general and sanitary data

PCR amplification targeting the ITS1–5.8S–ITS2 region detected DNA of the phylum Ciliophora in fecal samples from 13 pig farmers and their family members. Of these, seven individuals yielded nucleotide sequences suitable for interpretation, all confirmed as B. coli, indicating that these individuals (7/47; 14.9%) carried parasite sequences in their feces. Four of these originated from caretakers on industrial farms and three from family-owned farms, all located in Rio de Janeiro.

Phylogenetic analysis of the sequences obtained in this study (Fig 1) revealed a distribution across distinct clades, indicating high genetic diversity. Three sequences from human samples (H53, H63, and H70) from an industrial farm in Nova Friburgo belonged to the B0 genetic variant, with identities of 97.2%–100% compared with GenBank references. Despite sharing the same variant, these sequences did not cluster within the same clade (Fig 1). In contrast, the A0 variant was detected in one human sample (H61) from the same farm and in a caretaker from a family-owned farm (H31), with identities of 98.6%–99.1%. Two additional sequences from pig handlers from family-owned farms (H46 and H51) were classified as atypical and clustered within a clade closely related to sequences from a neotropical primate (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Maximum likelihood phylogeny of ciliates based on ITS sequences.

Sequences of ciliates obtained in this study (highlighted in blue) and reference sequences from different countries, hosts, and genetic variants. The concentric rings indicate: (i) country of origin (outermost ring), (ii) host species (middle ring), and (iii) genetic variant (innermost ring). Scale bar: 0.1 substitutions per site.

https://doi.org/10.1371/journal.pone.0338487.g001

All pig samples analyzed—one per farm, totaling 15 animals—yielded PCR amplification products for the ITS1–5.8S–ITS2 region. The 15 sequences generated were identified as B. coli, with identities of 97.8%–100% compared with GenBank references. Among these, three were classified as the A0 variant, while the remaining 12 belonged to the B0 variant. Notably, this variant was the only one identified in pigs from farms where B. coli sequences were also detected in humans (Fig 1). All sequences obtained in this study were deposited in GenBank with accession numbers as described in S1 Table.

Among the seven individuals positive for B. coli by PCR and confirmed by Sanger sequencing, most were male (71.4%), adults over 18 years (85.7%), and directly involved in pig husbandry (85.7%). More than half (=40%) reported living on the farm premises, had undergone previous testing for intestinal parasites, and had prior use of antiparasitic drugs. However, only two participants reported undergoing annual stool examinations, and most of them could not recall the frequency of antiparasitic drug use (Table 1).

Clinical symptoms compatible with parasitic infections were reported by some participants: 28% experienced diarrhea, and one reported abdominal pain with blood in the stool. None observed worms in their feces. More than 70% of positive participants reported consuming untreated water obtained from springs or artesian wells. All reported consuming vegetables after washing with water and additives such as chlorine, vinegar, or lemon, although only 57% maintained home gardens. Regarding fertilizer use, five participants (70%) could not provide this information (Table 1).

More than 70% of positive participants reported using personal protective equipment during pig handling, including boots, closed shoes, and long pants. One participant, however, reported handling animals without any protective measures. Similarly, 70% reported changing work clothing after animal contact (Table 1).

Fecal samples from rural pig producers and pigs positive for B. coli: Comparison of unicellular eukaryote microbiota and other co-occurring taxa

Overall, the 13 human fecal samples that showed bands compatible with the phylum Ciliophora by polymerase chain reaction—originating from family farms in Saquarema and Silva Jardim, and from an industrial farm in Nova Friburgo (RJ)—as well as one positive pig sample from the same farms, were subjected to metabarcoding analysis of the V4 region of the 18S rRNA gene.

The initial processing generated a dataset comprising 1,170,330 reads. After primer removal and processing in QIIME2 software, the number of reads was reduced to 985,842. These reads were clustered into 628 amplicon sequence variants (ASVs). They were classified into 15 phyla, 62 classes, 139 families, 181 genera, and 189 species, with approximately 366 ASVs identified at the species level. Among these, all ASVs corresponding to potentially pathogenic organisms were filtered to create a secondary dataset, comprising 730,256 reads and 118 ASVs. This subset was distributed across five phyla, 10 classes, 14 families, 16 genera, and 22 species, from which 12 ASVs remained unclassified at the species level.

Analysis of the overall fecal microbiota community based on metabarcoding results revealed a wide diversity of taxa, with high abundances of eukaryotes from the family Blastocystidae, as well as members of the phylum Amoebozoa and the family Balantididae in both human and pig samples. Additionally, ASVs corresponding to other eukaryotic taxa and from other kingdoms were also detected (Fig 2A). Focusing specifically on ASVs from the family Balantididae, these were detected in the feces of two individuals, H31 and H70, both rural producers with direct contact with pigs—H31 from Silva Jardim and H70 from Nova Friburgo—as well as in the feces of pigs from their respective properties, pig 129F and pig 565F (Fig 2B and C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Comparison of eukaryotic and other taxonomic communities in humans and pigs via metabarcoding.

(A) Relative abundance of all ASVs classified into major taxonomic groups. (B–C) Relative abundance of major groups and Venn diagrams showing shared and unique ASVs between pigs (129F in B; 565F in C) and humans (H31 in B; H70 in C), all positive for B. coli ASVs. (D) Relative abundance of ASVs assigned to potentially pathogenic organisms grouped into major categories. (E–F) Equivalent comparisons to B–C, restricted to potentially pathogenic groups, between pig–human pairs positive for B. coli (129F and H31 in E; 565F and H70 in F).

https://doi.org/10.1371/journal.pone.0338487.g002

A total of 68 ASVs were identified in H31 and 55 in pig 129F from the same property. Two ASVs were shared between these hosts, one belonging to the family Balantididae and the other to the kingdom Fungi. Only one ASV of the family Balantididae was detected in H31, which is not shown in the general community abundance plot due to its low frequency (eight reads). H31 exhibited lower eukaryotic taxon diversity compared to H70, with no ASVs from the family Blastocystidae detected (Fig 2B). In H70, 178 ASVs were observed, whereas 38 ASVs were detected in pig 565F. Of these, 12 ASVs were shared between both host species, belonging to the family Balantididae, the phylum Amoebozoa, and the kingdom Fungi. Additionally, shared ASVs from the family Blastocystidae, the phylum Dinoflagellata, and the kingdom Metazoa were detected in both hosts, although they were not highlighted in the relative abundance plot of pig 565F due to their low frequency (Fig 2C).

The sampling depth used for analyses considering all identified ASVs was 43,773. Principal Coordinates Analysis (PCoA), based on Bray–Curtis (Fig 3A) and Weighted UniFrac (Fig 3B) metrics, revealed distinct patterns in the overall eukaryotic community structure among the evaluated groups: humans with B. coli ASVs (H70 and H31) (HWB), humans without ASVs of this parasite (HWTB), and pigs that also harbored B. coli ASVs. In both analyses, pig samples formed a cohesive cluster, whereas humans without B. coli ASVs remained separated from this cluster, showing greater dissimilarity relative to pigs. Conversely, human samples with B. coli ASVs displayed an intermediate distribution, with some points positioned closer to the pig group, suggesting higher similarity between these two groups.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Fecal eukaryotic and other taxonomic communities in humans and pigs, stratified by Balantioides coli.

(A–B) Principal Coordinates Analysis (PCoA) based on Bray–Curtis distances (A) and Weighted UniFrac distances (B), showing separation among pig samples, humans without B. coli ASVs (HWTB), and humans with B. coli ASVs (HWB, H31, and H70). (C–D) Boxplots of Bray–Curtis (C) and Weighted UniFrac (D) distances comparing human and pig samples.

https://doi.org/10.1371/journal.pone.0338487.g003

The distance boxplots for pigs (Fig 3C and D) reinforced this observation. For the Bray–Curtis metric (Fig 3C), human samples without B. coli ASVs showed higher distances relative to pig samples, whereas the feces from humans H70 and H31 exhibited lower values, indicating greater compositional similarity. The same pattern was observed for the Weighted UniFrac metric (Fig 3D), which accounts for phylogenetic relationships, confirming the higher structural similarity between the feces of pigs and humans harboring confirmed B. coli ASVs.

Statistical analyses confirmed these patterns. In the overall analysis, significant differences were observed among the groups for both metrics (Bray-Curtis: Pseudo-F = 1.34, p = 0.021; Weighted UniFrac: Pseudo-F = 3.25, p = 0.007). Pairwise comparisons revealed that humans lacking B. coli ASVs differed significantly from pig fecal samples (Bray-Curtis: p = 0.004; Weighted UniFrac: p = 0.004), whereas feces from producers harboring B. coli ASVs did not differ significantly from pig samples (Bray-Curtis: p = 0.483; Weighted UniFrac: p = 0.289). Moreover, the Weighted UniFrac results indicated a trend toward separation between humans with and without the parasite ASVs (p = 0.088), suggesting that the presence of B. coli influences the phylogenetic structure of the human eukaryotic gut community.

Regarding potentially pathogenic eukaryotic parasites, high abundances of Blastocystis spp. ASVs were detected. These were present in human feces (12/13) and pig feces (2/3), in addition to Entamoeba spp. (5/13 and 3/3) and B. coli (2/13 and 3/3) in humans and pigs, respectively (Fig 2D).

A total of ten B. coli ASVs were recovered, six of which were identified in human feces. Among these, five ASVs were also detected in the feces of pigs from Nova Friburgo, Silva Jardim, and Saquarema, i.e., in the properties where the presence of the parasite’s DNA had been confirmed in human feces. Focusing on the eukaryotic microbiota composition of potentially pathogenic parasites in the pig producers H31 and H70, it was observed that both shared only B. coli ASVs with their respective pig, with this genetic variant being the sole ASV detected between producer H31 and pig 129F (Fig 2E). In contrast, in the fecal material of producer H70 and pig 565F, in addition to two shared B. coli ASVs, Entamoeba spp., Blastocystis spp., and other eukaryotes were also detected between these hosts (Fig 2F). Overall, 9,919 B. coli reads were identified in the fecal material of producer H70.

Beta-diversity analysis was also conducted for the dataset consisting solely of potentially pathogenic parasite ASVs, using a sampling depth of 38,759. Due to this cutoff, samples H31, H53, H59, and H61 were excluded from diversity analyses because they had very low read counts. The Bray-Curtis metric (Fig 4A) revealed a clear separation between pig and human samples: humans without B. coli ASVs were more distant from the pig cluster, while the human sample with B. coli ASVs was positioned closer to the pig samples. This pattern was even more pronounced using the Weighted UniFrac metric (Fig 4B), where the human sample with the parasite ASVs grouped closely with the pig cluster.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Potentially pathogenic eukaryotic parasites in feces of pigs and humans, stratified by Balantioides coli presence.

(A–B) Principal Coordinates Analysis (PCoA) based on Bray–Curtis distances (A) and Weighted UniFrac distances (B), showing separation among pig samples, humans without B. coli ASVs (HWTB), and humans with B. coli ASVs (HWB, H31, and H70). (C–D) Boxplots of Bray–Curtis (C) and Weighted UniFrac (D) distances comparing human and pig samples.

https://doi.org/10.1371/journal.pone.0338487.g004

Boxplots (Fig 4C and D) supported these observations: humans without B. coli ASVs showed greater distances relative to pigs, whereas the human sample with the ciliate ASVs displayed lower distances, indicating higher similarity with the pig samples. Statistical tests for Bray-Curtis were not significant globally (p = 0.055), but pairwise comparisons revealed a significant difference between human samples without B. coli ASVs and pig feces (p = 0.012). Weighted UniFrac analysis was highly significant (p = 0.005), confirming that humans lacking B. coli ASVs harbored a markedly distinct community of potentially pathogenic eukaryotes compared to pigs (p = 0.005) (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. PERMANOVA results from dissimilarity analyses (Bray–Curtis and Weighted UniFrac) comparing eukaryotic community structure in fecal samples from pig producers and their household members, and pigs with Balantioides coli.

https://doi.org/10.1371/journal.pone.0338487.t002

Fecal samples from rural pig farmers and pigs positive for B. coli: Comparison of the associated unicellular prokaryotic microbiota

The same fecal samples analyzed for eukaryotes by metabarcoding were also assessed for their unicellular prokaryotic composition. Initial processing generated 1,278,967 reads. After primer removal and QIIME2 filtering, 1,273,379 reads remained and were clustered into 15,225 amplicon sequence variants (ASVs), classified into 57 phyla, 154 classes, 604 families, and more than 1,200 genera. ASVs corresponding to organisms associated with gastrointestinal eubiosis and potentially pathogenic taxa (grouped at the genus level) were filtered, resulting in a secondary dataset comprising 335,289 reads and 1,025 ASVs. This subset was distributed across four phyla, five classes, nine families, and approximately 19 genera.

Analysis of the relative abundance of the overall prokaryotic community revealed broad diversity. Firmicutes and Bacteroidota were the predominant phyla, while Proteobacteria showed notable abundance in some individuals (H71, H70, H61, H55, H51, 565F). Spirochaetota and Actinobacteriota appeared at lower frequencies and only in specific producers and pigs (Fig 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Comparison of prokaryotic fecal communities in humans and pigs recovered via metabarcoding analyses.

(A) Relative abundance of all ASVs classified at the phylum level. (B–C) Relative abundance at the phylum level and Venn diagrams showing shared and unique ASVs between pigs (129F in B; 565F in C) and humans (H31 in B; H70 in C), all positive for Balantioides coli. (D) Relative abundance of ASVs belonging to two prokaryotic groups: taxa associated with gastrointestinal eubiosis and potentially pathogenic taxa, grouped at the genus level. (E–F) Equivalent to panels B–C, but restricted to these prokaryotic groups, comparing pig–human pairs positive for B. coli (129F and H31 in E; 565F and H70 in F).

https://doi.org/10.1371/journal.pone.0338487.g005

A targeted analysis of the prokaryotic communities in the fecal samples of producers H31 and H70 and their pigs (129F and 565F), which also harbored Balantididae ASVs, identified 2,569 ASVs in pig 129F and 564 in its producer (H31). Of these, 58 ASVs were shared, mainly within Firmicutes, Actinobacteriota, Bacteroidota, and Proteobacteria. These phyla were the most abundant overall, while ASVs from other phyla were detected at lower proportions (Fig 5B). In pig 565F, 808 ASVs were detected compared with 2,845 in producer H70, of which 217 were unique to each host. The same dominant phyla shared by H31 and 129F were also present in H70 and 565F, although additional low-frequency taxa were observed among shared ASVs in the latter pair (Fig 5B and C). Analyses including all ASVs were conducted using a sampling depth of 43,904 sequences per sample to standardize effort across groups.

A focused analysis of prokaryotes associated with gastrointestinal eubiosis and potential pathogenicity showed that the most abundant ASVs in both producers and pigs belonged to the genera Bacteroides and Prevotella. Other genera, including Escherichia–Shigella, Ruminococcus, Treponema, Roseburia, and Klebsiella, were detected at lower proportions in both hosts. Clostridium and Streptococcus were more abundant in pigs, whereas Succinivibrio and Faecalibacterium were more prevalent in humans (Fig 5D).

Among producers positive for B. coli ASVs, H31 and H70 harbored 94 and 140 prokaryotic ASVs, respectively. Their pigs (129F and 565F) harbored 132 and 143 ASVs, respectively. The overlap between hosts included eight shared ASVs in the H31–129F pair and 17 in the H70–565F pair. Shared genera included Prevotella, Faecalibacterium, Clostridium, Escherichia–Shigella, Klebsiella, Roseburia, and Bacteroides. In H70 and 565F, Streptococcus was also detected (Fig 5E and F). For analyses restricted to eubiosis-associated and potentially pathogenic taxa, a sampling depth of 11,965 sequences per sample was applied. As a result of this cutoff, samples 565F and 129F were excluded because of low counts of ASVs of interest.

Principal coordinates analysis (PCoA) based on Bray–Curtis dissimilarities (Fig 6A) showed that humans without B. coli ASVs formed a compact cluster clearly separated from pigs harboring Ciliophora ASVs. In contrast, humans with B. coli ASVs were more dispersed, with H70 positioned closer to the pig microbiota. Average distance analysis for pigs (Fig 6C) confirmed this pattern: Bray–Curtis values indicated shorter distances between pigs and humans with B. coli ASVs compared with humans lacking these ASVs, who consistently showed higher dissimilarity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Prokaryotic fecal communities in pig farmers, household members, and pigs, according to Balantioides coli presence.

(A–B) Principal coordinates analysis (PCoA) based on Bray–Curtis (A) and Weighted UniFrac (B) distances, showing the separation among pig samples, humans without B. coli ASVs (HWTB), and humans with B. coli ASVs (HWB, H31, and H70). (C–D) Boxplots of Bray–Curtis (C) and Weighted UniFrac (D) distances comparing human and pig samples.

https://doi.org/10.1371/journal.pone.0338487.g006

Weighted UniFrac ordination (Fig 6B) further supported these results. Pigs (129F and 565F) grouped closer to producer H70, while humans without B. coli ASVs formed a cohesive cluster with pig 257F. Boxplots (Fig 6D) also reflected this trend: humans without B. coli ASVs exhibited greater distances from pigs, whereas humans with the parasite clustered closer to them.

PERMANOVA post hoc tests corroborated these observations. Significant differences were found among groups for both Bray–Curtis (Pseudo-F = 1.71, p = 0.001) and Weighted UniFrac (Pseudo-F = 2.06, p = 0.019). Pairwise comparisons showed significant differences between producers without B. coli ASVs and pigs (Bray–Curtis: p = 0.002; UniFrac: p = 0.004), but no significant differences between producers harboring B. coli ASVs and pigs (Bray–Curtis: p = 0.097; UniFrac: p = 0.613) (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of PERMANOVA tests applied to dissimilarity analyses (Bray–Curtis and Weighted UniFrac) comparing the prokaryotic community structure in fecal samples from pig farmers and their household members, and pigs with Balantioides coli.

https://doi.org/10.1371/journal.pone.0338487.t003

Scatterplots assessing eubiosis-associated and potentially pathogenic prokaryotes (Fig 7A and B) showed no clearly defined clusters among humans with B. coli ASVs, humans without ASVs, and pigs. Nonetheless, a trend toward separation between pig and human microbiota was observed, particularly relative to humans lacking B. coli. Distance boxplots (Fig 7C and D) confirmed this trend: individuals without B. coli ASVs exhibited higher median distances from pigs compared with producers harboring B. coli ASVs. Post hoc tests revealed no statistically significant global differences. However, some p-values exhibited a marginal trend toward significance, particularly when comparing pigs with humans lacking B. coli ASVs (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Prokaryotic fecal communities, including potentially pathogenic taxa, in humans and pigs stratified by Balantioides coli.

(A–B) Principal coordinates analysis (PCoA) based on Bray–Curtis (A) and Weighted UniFrac (B) distances, showing the separation among the positive pig sample, humans without B. coli ASVs (HWTB), and humans with B. coli ASVs (HWB, H31, and H70). (C–D) Boxplots of Bray–Curtis (C) and Weighted UniFrac (D) distances comparing human samples with the positive pig sample.

https://doi.org/10.1371/journal.pone.0338487.g007