Maternal Mycobiome, but Not Antibiotics, Alters Fungal Community Structure in Neonatal Piglets

ABSTRACT

Early-life antibiotic exposure is associated with diverse long-term adverse health outcomes. Despite the immunomodulatory effects of gastrointestinal fungi, the impact of antibiotics on the fungal community (mycobiome) has received little attention. The objectives of this study were to determine the impact of commonly prescribed infant antibiotic treatments on the microbial loads and structures of bacterial and fungal communities in the gastrointestinal tract. Thirty-two piglets were divided into four treatment groups: amoxicillin (A), amoxicillin-clavulanic acid (AC), gentamicin-ampicillin (GA), and flavored placebo (P). Antibiotics were administered orally starting on postnatal day (PND) 1 until PND 8, except for GA, which was given on PNDs 5 and 6 intramuscularly. Fecal swabs were collected from piglets on PNDs 3 and 8, and sow feces were collected 1 day after farrowing. The impacts of antibiotics on bacterial and fungal communities were assessed by sequencing the 16S rRNA and the internal transcribed spacer 2 (ITS2) rRNA genes, respectively, and quantitative PCR was performed to determine total bacterial and fungal loads. Antibiotics did not alter the α-diversity (P = 0.834) or β-diversity (P = 0.565) of fungal communities on PND 8. AC increased the ratio of total fungal/total bacterial loads on PND 8 (P = 0.027). There was strong clustering of piglets by litter on PND 8 (P < 0.001), which corresponded to significant differences in the sow mycobiome, especially the presence of Kazachstania slooffiae. In summary, we observed a strong litter effect and showed that the maternal mycobiome is essential for shaping the piglet mycobiome in early life.

IMPORTANCE This work provides evidence that although the fungal community composition is not altered by antibiotics, the overall fungal load increases with the administration of amoxicillin-clavulanic acid. Additionally, we show that the maternal fungal community is important in establishing the fungal community in piglets.

INTRODUCTION

During early life, microbes in the gastrointestinal tract (GIT) play essential roles in host health and development. During this critical period of microbial exposure, microbes aid in the development of the host immune system (1). Aberrant microbial exposure in early life has been associated with long-term adverse outcomes such as autoimmune disease, obesity, and asthma (2). The GIT harbors a vast array of microbes, including bacteria, fungi, archaea, protozoa, and viruses (3). While the impact of bacteria on immune development has been well characterized and studied, fungi (the mycobiome) have received less attention.

Fungal colonization of both the human and pig GIT occurs shortly after birth (4, 5), and fungi can be acquired vertically, horizontally, and environmentally (6, 7). In humans, the mycobiome changes in both composition and abundance during the first 2 years of life and resembles the adult mycobiota at approximately 2 years of age (8). In pigs, the mycobiome changes most dramatically during the weaning transition and looks similar to the mycobiome of adult pigs in the postweaning period by as early as 24 to 35 days of age (9). Human bacterial ecosystems change in both richness and diversity most dramatically in the first year of life and stabilize at around 3 years of age (10). Likewise, pigs experience drastic changes in bacterial richness and diversity around the weaning transition, with bacterial stability occurring 2 to 3 weeks after weaning (11). However, fungal populations show more interindividual and intraindividual variation than bacterial populations (11).

Early-life antibiotic exposure is associated with the development of asthma (12), type 1 diabetes (13), and inflammatory bowel disease (14) later in life. It is thought that disruptions in bacterial colonization during the critical early-life period alter the immune response, leading to the development of disease (15). In a piglet model, we have identified that early-life amoxicillin (A) results in altered pancreatic development and divergent responses to immune and metabolic challenges compared to placebo-treated piglets (16–18). However, the role of antibiotics in fungal communities has not been investigated, with the exception of Candida albicans, where antibiotics have been shown to alter the intestinal metabolite pool in a way that promotes C. albicans growth (19). It should be noted that fungi are immunomodulatory (20), and changes in the mycobiome may be a factor contributing to disease development.

Pigs serve as a good human model as they are similar to humans in both intestinal structure and immune system function (21). In the present study, we used a piglet model to investigate the role of antibiotics in fungal community dynamics in early life, with potential implications for both human and pig health. The objectives of this study were to characterize the mycobiome in response to 3 different antibiotic regimens that are commonly prescribed in early childhood and to quantify changes in the mycobiome in response to antibiotic administration in conjunction with changes in the bacterial community structure. The 3 antibiotic regimens used for this study included amoxicillin, amoxicillin-clavulanic acid (AC), and gentamicin-ampicillin (GA). Amoxicillin with or without clavulanic acid was chosen because it is commonly prescribed to children for respiratory tract and ear infections (22, 23), and amoxicillin treatment has been shown to result in altered phenotypes in our piglet model (16–18). Both regimens have been shown to increase the incidence of infection with Candida, a common yeast of the GIT (24). The combination of gentamicin and ampicillin was chosen because this combination is recommended for neonates with bacterial sepsis (25), and exposure to aminoglycoside antibiotics, such as gentamicin, is a risk factor for invasive Candida infections (26). In the context of this paper, the term antibiotic refers only to those compounds that target bacteria and not fungi. It is worth noting that antibiotics themselves do not kill fungi, but by altering the bacterial community, they decrease competition and allow fungi to flourish and persist in the gut (27). Here, we show the impact of these antibiotics on the bacterial and fungal communities and discuss the role of the maternal mycobiome in piglet mycobiome development.

RESULTS

Piglet performance was not different among treatment groups or litters.

The average daily weight gain did not differ between piglets based on treatment (P = 0.198) or litter (P = 0.848) at day 8 (see Table S1 in the supplemental material). All piglets remained healthy, with a health score of zero (Table S2), and no incidence of diarrhea was observed throughout the study.

Litter but not antibiotics drove the mycobiome composition.

Antibiotic treatment did not alter fungal β-diversity (R² = 0.08; P = 0.565; β-dispersion P = 0.116) (Fig. 1a) or α-diversity (P = 0.834) (Fig. 1b) on postnatal day (PND) 8. It was noted, however, that fungal community compositions differed by litter. On PND 3, α-diversity differed between litters (P = 0.015) (Fig. 2b), but β-diversity did not change between litters (R² = 0.112; P = 0.352; β-dispersion P = 0.054) (Fig. 2a). By PND 8, both α-diversity (P < 0.001) (Fig. 2d) and β-diversity (R² = 0.246; P < 0.001; β-dispersion P = 0.705) (Fig. 2c) differed between litters. Differences in fungal community structures between litters may be explained by differences in the relative abundances of Kazachstania (P < 0.001) (Fig. 3a) and Nakaseomyces (P < 0.001) (Fig. 3b) on PND 8. Kazachstania made up to 99.9% of the fungal community in some piglets by PND 8, whereas other piglets had no Kazachstania (Fig. 4). Kazachstania was more abundant in litters A and B than in litters C and D on PND 8 (Fig. 4). On PND 3, the Kazachstania abundance was lower, ranging from 0.4 to 11.2% (Fig. 5). Piglets without Kazachstania had more unclassified fungi and a more diverse mycobiome on both PNDs 3 and 8. The Kazachstania abundance also differed drastically in sows, with a range from 0 to 98.0% (Fig. 4 and 5). Piglets from sows with high Kazachstania abundances had increased Kazachstania abundances by PND 8 compared to PND 3 (Fig. 4). On PND 8, the fungal community structure of piglets was more similar to that of their dams than to that of the other dams based on Bray-Curtis dissimilarity (P < 0.05) (Fig. 6).

FIG 1.

Fungal diversity following antibiotic treatment. Shown are values for fungal β-diversity using Bray-Curtis metrics (PERMANOVA R² = 0.08; P = 0.565; β-dispersion P = 0.116) (a) and the Shannon diversity index (b) on PND 8 after treatment with amoxicillin (A) (n = 6), amoxicillin-clavulanic acid (AC) (n = 8), gentamicin-ampicillin (GA) (n = 7), or flavored placebo (P) (n = 8) (P = 0.834 by a Kruskal-Wallis test). Significance was defined as a P value of ≤0.05.

FIG 2.

Fungal diversity by litter. (a) PND 3 fungal β-diversity by litter as measured by Bray-Curtis dissimilarity (PERMANOVA R² = 0.112; P = 0.352; β-dispersion P = 0.054). (b) PND 3 α-diversity by litter as measured by the Shannon diversity index (P = 0.015 by a Kruskal-Wallis test). (c) PND 8 fungal β-diversity by litter as measured by Bray-Curtis dissimilarity (PERMANOVA R² = 0.246; P < 0.001; β-dispersion P = 0.705). (d) PND 8 α-diversity by litter as measured by the Shannon diversity index (P < 0.001 by a Kruskal-Wallis test). Litters are represented by A to D (n = 6 to 8 per litter). Significance was defined as a P value of ≤0.05.

FIG 3.

Differential fungal taxa between litters. Shown are data for Kazachstania (P < 0.001 by a Kruskal-Wallis test) (a) and Nakaseomyces (P < 0.001 by a Kruskal-Wallis test) (b) by litter on PND 8 (n = 8 per litter). Litters that do not share a letter indicate significance. Significance was defined as a P value of ≤0.05. Error bars represent means with standard errors of the means.

FIG 4.

Relative abundances of the top 11 fungal genera on PND 8 in piglets by litter (n = 8 per litter). Bars representing sows show fungal relative abundances 1 day after farrowing. A to D indicate sows and their corresponding litters.

FIG 5.

Relative abundances of the top 11 fungal genera on PND 3 in piglets by litter (n = 8 per litter). Bars representing sows show fungal relative abundances 1 day after farrowing. A to D indicate sows and their corresponding litters.

FIG 6.

Similarity of the piglet fungal community to the maternal community. The distance from the maternal sow versus the distance from a sow of a different litter was based on the Bray-Curtis dissimilarity of fungal communities on PND 8. Statistical analyses were performed by a Mann-Whitney U test. Significance was defined as a P value of ≤0.05. Error bars represent means with standard errors of the means. All sows had 6 to 8 piglets representing all treatment groups.

Antibiotic treatment impacted bacterial community structure.

Antibiotic treatment altered bacterial β-diversity on PND 8 (R² = 0.228; P < 0.001; β-dispersion P = 0.706) (Fig. 7a). However, bacterial α-diversity did not differ following antibiotic treatment on PND 8 (P = 0.313) (Fig. 7b). Animals treated with AC had a decreased abundance of Lactobacillus (P < 0.001) (Fig. 8). On PND 3, litter influenced bacterial β-diversity (R² = 0.201; P < 0.001; β-dispersion P = 0.187) (Fig. 9a) but did not impact α-diversity (P = 0.287) (Fig. 9b). On PND 8, litter again impacted β-diversity (R² = 0.159; P = 0.015; β-dispersion P = 0.521) (Fig. 9c) but not α-diversity (P = 0.789) (Fig. 9d). On PND 8, differences in community structures between litters may be explained by the relative abundance of Akkermansia (P < 0.001) (Fig. 10). In litters A and B, the piglet microbiomes were no closer to that of their maternal sow than to those of the other sows in the study (P > 0.05) (Fig. 11a and b). In litter D, piglet microbiomes were closer to that of their maternal sow than to those of the sows of other litters (P = 0.032) (Fig. 11c). Sequencing results from sow C did not pass quality control and therefore were not included. Unlike the mycobiome composition, the bacterial community composition was strongly influenced by antibiotic treatment, although litter effects were still observed.

FIG 7.

Bacterial diversity following antibiotic treatment. Bacterial β-diversity (PERMANOVA R² = 0.228; P < 0.001; β-dispersion P = 0.706) (a) and α-diversity (b) were measured by the Shannon index on PND 8 after treatment with antibiotics or placebo, including amoxicillin (A) (n = 8), amoxicillin-clavulanic acid (AC) (n = 8), gentamicin-ampicillin (GA) (n = 8), or flavored placebo (P) (n = 8) (P = 0.313 by a Kruskal-Wallis test). Significance was defined as a P value of ≤0.05.

FIG 8.

Relative abundance of Lactobacillus on PND 8 after treatment with amoxicillin (A), amoxicillin-clavulanic acid (AC), gentamicin-ampicillin (GA), or flavored placebo (P) (P < 0.001 by a Kruskal-Wallis test). Significance was defined as a P value of ≤0.05. Error bars represent means with standard errors of the means. Litters not sharing a letter were significantly different.

FIG 9.

Bacterial diversity by litter. (a) PND 3 bacterial β-diversity by litter as measured by Bray-Curtis dissimilarity (PERMANOVA R² = 0.227; P < 0.001; β-dispersion P = 0.501). (b) PND 3 bacterial α-diversity by litter as measured by the Shannon index (P = 0.287 by a Kruskal-Wallis test). (c) PND 8 bacterial β-diversity by litter as measured by Bray-Curtis dissimilarity (PERMANOVA R² = 0.159; P = 0.015; β-dispersion P = 0.521). (d) PND 8 α-diversity by litter as measured by the Shannon index (P = 0.789 by a Kruskal-Wallis test). Significance was defined as a P value of ≤0.05. Litters are represented as A to D, and all litters had 7 to 8 piglets.

FIG 10.

Differential bacterial genera by litter. Shown is the relative abundance of Akkermansia on PND 8 by litter (n = 8 per litter) (P < 0.001 by a Kruskal-Wallis test). Significance was defined as a P value of ≤0.05. Error bars represent means with standard errors of the means. Litters that do not share a letter were significantly different.

FIG 11.

Similarity of the piglet bacterial community to the fungal community. The distance from the maternal sow versus the distance from a sow of a different litter was based on Bray-Curtis dissimilarity and measured using a Mann-Whitney U test of bacterial communities on PND 8. Significance was defined as a P value of ≤0.05. Error bars represent means with standard errors of the means. All sows had 6 to 8 piglets representing all treatment groups.

Antibiotic treatment increased the ratio of total fungi to total bacteria.

The total bacterial load did not change with antibiotic treatment (P = 0.325) (Fig. 12a); however, total fungi tended to be impacted by antibiotic treatment (P = 0.083) (Fig. 12b). As a result, antibiotic treatments caused a significant change in the ratio of total fungi to total bacteria (P = 0.001) (Fig. 12c). Post hoc tests revealed that treatment with AC increased the ratio of total fungi to total bacteria compared to placebo (P = 0.027) (Fig. 12c).

FIG 12.

Fungal and bacterial loads. (a) Total bacteria by treatment determined by qPCR (PROC MIXED P = 0.325). (b) Total fungi by treatment determined by qPCR (PROC MIXED P = 0.083). (c) Ratio of total fungal to total bacterial loads (PROC MIXED P = 0.001). A significant reduction in the ratio of total fungi to total bacteria was detected between the P and AC groups (Bonferroni-corrected P = 0.027). Significance is indicated by * and is defined as a P value of ≤0.05 (P, n = 5; A, n = 7; AC, n = 8; GA, n = 8). Error bars represent means with standard errors of the means.

DISCUSSION

In this study, we sought to determine the impact of common early-life antibiotic treatments on both the fungal community composition and fungal load. Although antibiotic treatments altered the bacterial community composition on PND 8, we did not observe any changes in the fungal community composition (Fig. 1a). While it has been well documented that antibiotic treatments can cause Candida overgrowth, thus leading to an altered fungal community composition, there appears to be a great deal of individual variation in susceptibility to Candida overgrowth (28). This is due to a variety of factors, including genetics (29), the intestinal metabolite profile (19), and the host’s immune status (30). One recent study noted that the human mycobiome was altered by antibiotic administration to the greatest degree 1 month after antibiotic treatment, which suggests that there may be a delayed response following antibiotic treatment (31). Therefore, it is possible that either the pigs used in the present study were not susceptible to fungal dysbiosis following antibiotic treatments or there was a delayed change in the fungal community composition that was outside the study window. While pigs share some similarities in microbes, such as Candida, it may be possible that strain-level differences between humans and pigs may account for the lack of community changes in response to antibiotic treatment.

The numerical reduction of total bacteria and the trend of increasing total fungi following treatment with AC resulted in an increase in the ratio of total fungi to total bacteria in AC-treated animals (Fig. 12c). Amoxicillin-clavulanic acid is a commonly used antibiotic to treat respiratory tract infections during the first year of life (32). Amoxicillin is a β-lactam antibiotic that is effective against both Gram-positive and Gram-negative bacteria, including Escherichia coli and Salmonella species (33). However, over time, many strains of bacteria have developed resistance to β-lactams through the production of β-lactamase. To overcome this resistance, clavulanic acid is frequently added to amoxicillin as it acts as a β-lactamase inhibitor (34). While we did not see a statistically significant difference in the abundance of total bacteria following treatments with any of the antibiotics in the present study, there appears to be a slight decrease in the abundance of total bacteria in pigs treated with AC (Fig. 12a). This suggests that β-lactam resistance was present. The combination of gentamicin-ampicillin is commonly used to treat neonatal sepsis and is commonly prescribed for a 2-day period (25). However, the short treatment duration may be responsible for the lack of a reduction in the bacterial load in this treatment group. In the present study, the lack of a significant reduction in total bacteria may be the reason why we did not see an increase in total fungi as expected. Additionally, we saw a decrease in the abundance of Lactobacillus following treatment with AC. The presence of C. albicans has been shown to result in a long-term reduction of Lactobacillus in the gut (35). Since only a few of the piglets in the present study had Candida present, we are not able to say whether Candida increased following antibiotic treatment; however, it is possible that the increased ratio of total fungi to total bacteria promotes an environment in which Lactobacillus is suppressed.

Interestingly, the litter effect was the main driver of fungal community composition in the current study rather than antibiotic exposure. Using Bray-Curtis dissimilarity, it was noted that on PND 8, piglets had mycobiomes that more closely resembled the mycobiome of their maternal sow than the mycobiomes of other sows. This suggests that maternal fungal colonization is the driving factor in piglet mycobiome development. These results are contrary to what is seen in human infants, with one study observing that infants in their first month of life were no more similar to their mothers than to a randomly selected mother, suggesting that the external environment plays a larger role than the maternal mycobiome in mycobiome composition (35). Compared to human infants, piglets have more exposure to maternal feces due to the environment in which they live. Piglets engage in coprophagy of these feces, with one study showing a rate of consumption of sow feces of approximately 20 g/day (36). Coprophagy has been shown to increase piglet feed intake, weight gain, and white blood cell counts compared to piglets deprived of maternal feces in the first 7 postnatal days (37). These differences may be due to the acquisition of microbes from the sow via coprophagy. Therefore, it is possible that compared to humans, piglets obtain more of their mycobiome from their mother’s feces. However, maternal bacterial colonization did not drive piglet bacterial colonization in the same way in the present study.

On PND 8, Nakaseomyces was present in litter B in all 8 piglets, ranging from 0.009% to 19.4% of the fungal community, but was absent from the other litters (Fig. 3b). Sow B did not have any Nakaseomyces present. Nakaseomyces is the name given to a clade of pathogenic Candida species, which includes Candida glabrata, Candida nivariensis, and Candida bracarensis (38). The drivers behind Nakaseomyces detection in litter B remain unclear, and no clinical illness was observed in these piglets. The other fungal species responsible for the differences in community structures on PND 8 was Kazachstania (Fig. 3a). Kazachstania slooffiae is one of the most abundant fungi in pigs and is present in pigs reared under various conditions and in different locations (39, 40). K. slooffiae is thought to be commensal; several benefits to the host have been noted, including increased short-chain fatty acid production and symbiotic relationships with beneficial bacteria such as Lactobacillus, and it possesses a favorable amino acid profile for pig growth (5, 41–43). Piglets’ levels of Kazachstania colonization reflected the levels of Kazachstania in their respective sows, which were highly variable. Given the potential benefits of Kazachstania colonization, future work should focus on the long-term health outcomes in animals with low colonization levels during the early-life period as well as the impact of maternal Kazachstania colonization on piglet health outcomes.

A litter effect was also present in bacterial communities. In this case, on PND 8, it appeared to be driven by Akkermansia. It is worth noting that Akkermansia was the most abundant organism in litter D, which was also the litter with very low levels of Kazachstania colonization. A previous study noted a potential interaction between Akkermansia and yeast fermentate (44). Ducray et al. found that supplementing rats with a yeast fermentate prebiotic from Saccharomyces cerevisiae prevented a heat-stress-associated rise in Akkermansia (44). However, it is unclear what effect live yeast would have on Akkermansia colonization, and this merits further exploration.

Conclusion.

In conclusion, antibiotic treatment altered the bacterial, but not the fungal, community composition. However, AC was found to increase the ratio of total fungal to total bacterial loads in fecal contents. It was found that the maternal mycobiome played a major role in shaping the piglet mycobiome, and a strong litter effect on fungal communities was observed on PND 8. This research indicates that the mycobiome of piglets is highly variable and dependent on the litter of origin, and future research into the piglet mycobiome should account for variations between litters.

MATERIALS AND METHODS

Animals and housing.

This animal study was approved by the Animal Care and Use Committee of the University of Alberta and conducted in accordance with the guidelines of the Canadian Council on Animal Care at the Swine Research and Technology Centre (Edmonton, AB, Canada) under animal use protocol number AUP00000922. A total of 32 crossbred Duroc × (Large White/Landrace) piglets and 4 Large White/Landrace sows were used in this study. On postnatal day (PND) 1, 4 litters of piglets were weighed, and 8 piglets from each litter were selected for the study based on sex and weight (4 piglets above and 4 piglets below the median litter weight [litters labeled A to D]). Piglets were balanced for sex and weight and remained with their mother for the duration of the study. Two piglets from each litter were then assigned to 1 of 4 treatment groups (n = 8): A (amoxicillin) (30 mg/kg of body weight/day orally every 12 h on PNDs 1 to 8), AC (amoxicillin-clavulanic acid) (30 mg/kg/day orally every 12 h on PNDs 1 to 8), GA (gentamicin-ampicillin) (gentamicin at 5 mg/kg/day once daily by intramuscular injection and ampicillin at 100 mg/kg/day twice daily by intramuscular injection on PNDs 5 to 6), or P (flavored placebo) (30 mg/kg orally every 12 h on PNDs 1 to 8). All oral preparations included an artificial maple flavoring and were identical except for the addition of antibiotics. Piglets in the GA group also received the placebo treatment on PNDs 1 to 4, 7, and 8 to account for handling stress. An injection control was not used as all of the piglets in the other groups continued to receive oral treatments on PNDs 5 and 6 and therefore experienced similar amounts of handling stress. Fecal swabs and samples were collected on PND 3 and PND 8 from each piglet. Fecal samples were collected from defecating sows 1 day after farrowing. Neither piglets nor sows received any antibiotics or other medications outside the study treatment groups. Creep feed was not provided to piglets, and all sow diets were the same. PND 3 was chosen as the first time point as obtaining enough feces to get good-quality reads from a younger piglet was not possible. A study duration of 7 days was chosen because we have previously shown that amoxicillin can cause changes in the bacterial component of the microbiome by as early as PND 3 and that these changes begin normalizing by PND 7 (17). Samples were stored at −80°C until further processing. Piglets were weighed on PND 1 and PND 8 and were scored for health (see Table S1 in the supplemental material) and diarrhea daily throughout the study, as previously described (45).

DNA extraction.

Total genomic DNA was extracted from fecal swabs using the DNeasy PowerSoil Pro kit (Qiagen, CA, USA) according to the manufacturer’s instructions, with no modifications. Bead beating was performed on a FastPrep-24 homogenizer (MP Biomedicals, OH, USA) at 5 m/s for 45 s. The DNA concentration was quantified using a Quant-iT PicoGreen dsDNA (double-stranded DNA) kit (Invitrogen, CA, USA).

Fungal sequencing.

Internal transcribed spacer 2 (ITS2) sequencing was performed at Microbiome Insights (University of British Columbia, BC, Canada). The following primers were used: forward primer 5′-CCTCCGCTTATTGATATGC-3′ (ITSF) and reverse primer 5′-CCGTGARTCATCGAATCTTTG-3′ (ITSR). A paired-end sequencing run was performed on the Illumina MiSeq platform using 2× 300-bp cycles.

Sequencing analysis was performed using Quantitative Insight into Microbial Ecology 2 (QIIME 2) (v2021.4) (46). Only forward reads were utilized to account for the variation in the length of the ITS2 region, and reads were not truncated. The Divisive Amplicon Denoising Algorithm version 2 plug-in (47) was used to perform demultiplexing, quality filtering, denoising, and filtering out chimeras. Amplicon sequence variants (ASVs) were aligned using mafft (48). Taxonomy was assigned to the resulting ASVs using the classify-sklearn naive Bayes taxonomic classifier (via the q2-feature-classifier plug-in) (49) against the UNITE database version 8.3 (50). The R package phyloseq (v1.34.0) was used to analyze microbial community structure and diversity (51). α-Diversity was measured using the Shannon index at a sampling depth of 1,000 reads and analyzed using a Kruskal-Wallis test. Alterations in the overall fungal community composition were measured using Bray-Curtis dissimilarity and permutational multivariate analysis of variance (PERMANOVA), which were visualized using principal-coordinate analysis (PCoA) (R v4.0.5). The homogeneity of dispersion was measured using the Betadisper function in phyloseq (51).

Bacterial sequencing.

Amplicon libraries of the V3-V4 region of the 16S rRNA gene were constructed in-house according to the Illumina 16S metagenomic sequencing library preparation protocol. The DNA concentration was determined using the Quanti-iT PicoGreen dsDNA assay kit (Invitrogen, CA, USA). The following amplicon primers were used: forward primer 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3′ and reverse primer 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3′. Sequencing was performed on an Illumina MiSeq platform using 2× 300-bp cycles (Illumina Inc., San Diego, CA). Due to the declining quality of the reverse reads, only forward reads truncated at 208 bp in length were utilized. Sequence assembly was performed as described above for fungi, except that taxonomy was assigned using the SILVA database version 138.1 (52) and the Shannon index was measured at a depth of 11,000 reads.

Quantitative PCR.

The total fungal load was measured using quantitative PCR (qPCR), which was performed on a StepOnePlus real-time PCR system (Applied Biosystems, CA, USA). Each reaction was performed in duplicate, and the reaction mixture consisted of 5 μL of PerfeCTa SYBR green supermix (Quantabio, MD, USA), 0.8 μL (10 μM/L) of forward primer NL1 (5′-GCATATCAATAAGCGGAGGAAAAG-3′) (53) and reverse primer LS2 (5′-ATTCCCAAACAACTCGACTC-3′) (54), 1.4 μL of nuclease-free water, and 2 μL of template DNA. The following cycling parameters were used: 10 min at 95°C and 40 cycles of 95°C for 15 s, 59°C for 15 s, and 72°C for 15 s (55). A standard curve was generated using DNA extracted from Kazachstania slooffiae in the manner outlined above for DNA extraction. K. slooffiae was isolated from a pig in the same barn where the present study took place. K. slooffiae DNA was quantified using the Quanti-iT PicoGreen dsDNA assay kit (Invitrogen, CA, USA).

The total bacterial load was measured via qPCR as described above. Forward primer SRV3-1 (5′-CGGYCCAGACTCCTACGGG-3′) and reverse primer SRV3-2 (5′-TTACCGCGGCTGCTGGCAC-3′) (56) were used. The following cycling parameters were used: 95°C for 3 min and 40 cycles of 95°C for 10 s and 60°C for 30 s. A standard curve was generated from the PCR amplicon of pooled genomic DNA, which was quantified using the Quanti-iT PicoGreen dsDNA assay kit (Invitrogen, CA, USA).

Statistical analyses.

Unless otherwise stated, all statistical analyses were done using GraphPad Prism 9.2.0. Differentially abundant taxa between treatments and litters were determined using ANCOM in QIIME 2021.4. The identified taxa were subsequently compared between litters using a Kruskal-Wallis test. Piglet average daily gain was analyzed using one-way analysis of variance (ANOVA). The distance from the maternal sow to her piglets versus that of a sow of another litter was determined based on Bray-Curtis dissimilarity using a Mann-Whitney U test. qPCR data were analyzed using PROC MIXED with blocking by litter, followed by a Bonferroni post hoc test, using SAS software (SAS OnDemand; SAS Institute Inc., NC, USA).

Data availability.

ITS2 and 16S rRNA gene reads were deposited in the National Center for Biotechnology Information Sequence Read Archive and are available under BioProject accession numbers PRJNA877644 and PRJNA878465, respectively.