ABSTRACT
The gut-lymph node axis is a critical player in the symbiotic relationship between gut microbiota and the host. However, little is known about the impact of diet-related bacterial shifts in the gut lumen on bacterial translocation into lymph nodes. Here, we (i) characterized changes in the viable microbiota composition along the ileal digesta-mucosa-lymph node axis and (ii) examined the effect of dietary phytase supplementation and lactic acid (LA) soaking of cereals on the bacterial taxonomy along this axis, together with their effect on the mucosal expression of innate immune and barrier function genes in pigs (n = 8/diet). After 18 days on diets, ileal digesta, mucosa, and ileocecal lymph nodes (ICLNs) were collected for RNA isolation and 16S rRNA-based high-resolution community profiling. Bacterial communities were dominated by Lactobacillaceae and Clostridiaceae, with clearly distinguishable profiles at the three sampling sites. Specific bacterial subsampling was indicated by enrichment of the ICLNs with Lactobacillaceae, Lachnospiraceae, Veillonellaceae, and Methanobacteriaceae and less Clostridiaceae, Pasteurellaceae, Helicobacteraceae, and Enterobacteriaceae compared to that of the mucosa. LA treatment of cereals reduced proteolytic taxa in the lumen, including pathobionts like Helicobacteraceae, Campylobacteraceae, and Fusobacteriaceae. When combined, phytase- and LA-treated cereals largely increased species richness, while the single treatments reduced Actinobacteria and Bacteroidetes in ICLNs and increased mucosal MUC2 expression. In contrast, phytase reduced mucosal CDH1 expression, indicating altered barrier function with potential effects on bacterial translocation. Overall, both treatments, although often differently, changed the viable microbiome along the digesta-mucosa-lymph node axis in the ileum, probably due to altered substrate availability and microbial-host interactions.
IMPORTANCE A host’s diet largely determines the gut microbial composition and therefore may influence bacterial translocation into ICLNs. Due to its importance for cell metabolism, the intestinal phosphorus availability, which was modified here by phytase and LA treatment of cereals, affects the intestinal microbiota. Previous studies mainly focused on bacteria in the lumen. The novelty of this work resides mainly in that we report diet-microbe effects along the digesta-mucosa-ICLN axis and linked those effects to mucosal expression of barrier function genes as crucial components for host health. Lymph nodes can serve as reservoir of pathobionts; therefore, present diet-microbiome-host interactions have implications for food safety.
INTRODUCTION
Commensal microbes that translocate from the intestinal lumen to lymphatic tissues challenge and train the mammalian immune system, thereby contributing to establishment of host-microbe tolerance (1). The diet is thereby one of the most vital players shaping the taxonomic and metabolic composition of the gut microbiome (2). Both nutrient deficiency and nutrient excess can cause large shifts in the bacterial taxonomic and functional composition, altering the mucosal cross talk with the host animal and potentially affecting gut barrier function (3, 4). For this reason, alterations in the expression of proinflammatory cytokines (interleukin-1 [IL-1], IL-6, IL-8, and IL-10) as a response to dietary changes may be related to diet-associated shifts in the gut microbial composition. If the gut mucosa responds with an increased cytokine expression to dietary changes, this may impair intestinal integrity and epithelial function (5), thereby allowing more bacteria to translocate. Bacterial translocation to lymphoid organs occurs via sampling by dendritic cells (6) or via passing through epithelial microfold cells (M cells) (7) and tight junctions of the epithelium (8). Recently, we found large numbers of Proteobacteria, including both commensals and pathogens, in granulomatous ileocecal lymph nodes (ICLNs), indicating increased bacterial translocation from the lumen of the gastrointestinal tract compared to that of asymptomatic and enlarged ICLN (9). Overall, little is known about how diet-related shifts in the luminal microbiota composition affect bacterial translocation into ICLN. Proteobacteria may benefit from higher intestinal P availability, as we could show for ileal digesta- and mucosa-associated bacterial communities (10–12). Consequently, if opportunistic pathogens are promoted, higher intestinal P availability may compromise gut and systemic health.
Over recent decades, efforts were made to decrease the (inorganic) P content in pig diets due to environmental and economic reasons (13). Phosphorus is an essential nutrient for intestinal bacteria; if deficient, bacteria will downregulate cell replication and metabolism (14). Low intestinal P availability, for instance, compromises fiber fermentation (14, 15). While mammals cannot utilize phytate-P, the storage form of P in plants, intestinal bacteria have the enzymatic capacities to hydrolyze phytate-P; however, they express phytases only when the available P in their surroundings becomes low (16). Any dietary strategy used to increase the P availability for the host animal will also enhance the luminal P availability for the microbes in the upper gastrointestinal tract (10–12). One common dietary strategy to increase the phytate-P availability for pigs is to add microbial-derived phytase to pig diets. Soaking and fermentation of dietary cereals and legumes, as traditional strategies to reduce phytate-P in human foods, are mostly used in liquid feeding systems in pig production (17). While dietary phytase mainly enhances the mineral availability in the upper digestive tract, soaking of cereals additionally modifies the other nutrient fractions, such as complex carbohydrates (18). Although the capability of dietary minerals and carbohydrates to alter the bacterial community and host inflammatory response has been described (11, 19, 20), relatively little is known about their potential to alter intestinal bacterial translocation. Therefore, the aims of the present study were (i) to characterize the changes in the microbiome composition along the ileal digesta-mucosa-lymph node axis and (ii) to investigate the effect of phytase supplementation and/or lactic acid (LA) treatment of dietary cereals on viable bacterial microbiome in ileal digesta, at the mucosa, and in ICLNs, as well as on the mucosal expression of innate immune and barrier function genes in pigs. We hypothesized that an altered P availability by both dietary treatments and a modified complex carbohydrate composition in the upper small intestine due to the LA treatment of cereals would modify the ileal abundances of Proteobacteria and fibrolytic taxa, which, in turn, would change the mucosal Toll-like receptor (TLR)-mediated signaling and barrier function, thereby allowing a distinct bacterial community to assemble in the ICLN.
RESULTS
Diets and animals.Four wheat- and corn-based diets were formulated (see Table S1 in the supplemental material for detailed dietary composition) and fed to the pigs in a 2-by-2 completely randomized factorial design. Diets differed in the treatment of cereals, which were either soaked in 2.5% LA for 48 h or not, and dietary phytase supplementation (0 versus 500 phytase units [FTU]/kg complete feed). Soaking wheat and corn in 2.5% LA increased the dry matter content but leached a small amount of ash, including P, and protein into the soaking solution. With respect to the complex carbohydrate fraction, the LA treatment of cereals decreased the content of neutral detergent fiber (NDF) and modified the resistant starch (RS) to nonresistant starch (NRS) proportion. Pigs were clinically healthy throughout the study, and the inner organs as well as the ileal mucosa did not show signs of disease or inflammation when examined at sample collection at the end of the experiment.
Microbial taxonomic composition.In order to study the effects of dietary phytase supplementation and LA treatment of cereal grains on the metabolically active bacterial community in ileal digesta, at the ileal mucosa, and in ICLNs, RNA was isolated and transcribed into cDNA, and 16S rRNA sequencing was employed. Total bacterial numbers were higher in digesta than at the mucosa and in the ICLNs (P < 0.001) (Table S2), amounting to 10.3 ± 0.19, 9.2 ± 0.19, and 7.3 ± 0.19 log10 gene copies in ileal digesta, at the mucosa, and in ICLNs, respectively (P < 0.001) (Table S2). This was reflected in species richness and α-diversity being highest in digesta and lowest in ICLNs. After quality and chimera filtering, a total of 2,221,409 reads, with a mean of 24,682 sequences per sample and a mean read length of 415 bp, were obtained. The two most abundant phyla were Firmicutes and Proteobacteria in ileal digesta (85.8% and 9.7% of reads, respectively), at the ileal mucosa (69.4% and 28.2% of reads, respectively), and in ICLNs (72.3% and 15.3% of reads, respectively) (Table S3). At the family level, reads were classified into 33 families in ileal digesta, 26 families at the ileal mucosa, and 12 families in ICLNs, where Clostridiaceae and Lactobacillaceae were the two most abundant families across the three sampling sites (Fig. 1 and Table S4). Accordingly, Lactobacillus and unclassified Clostridiaceae were the two most dominant genera across the three sampling sites (Table S5). Despite the similar predominant taxa, the three sampling sites were comprised of distinct bacterial communities (P = 0.001) (Table S6), as visualized in the nonmetric multidimensional scaling (NMDS) plot (Fig. 2A).
Relative abundance of bacterial families (>0.1% relative abundance) in ileal digesta at the mucosa and in ileocecal lymph nodes of pigs. Values are presented as least square means, n = 32 per site.
Nonmetric multidimensional scaling (NMDS) plot of pairwise Bray-Curtis dissimilarities between bacterial communities in ileal digesta, mucosa, and ileocecal lymph nodes of pigs (>0.01% relative abundance). (A) Ileal compartment (gray, ileocecal lymph nodes [ICLNs]; green, ileal mucosa [ILMU]; red, ileal digesta [ILDI]); (B) treatment effect per compartment (gray, control diet; red, diet supplemented with phytase; blue, diet containing LA-treated cereals; green, diet containing phytase and LA-treated cereals). Ellipses represent the standard deviations. Red, ileal digesta, n = 32 per sampling site.
Diet-related changes of bacterial community in ileal digesta, ileal mucosa, and ileocecal lymph nodes.The LA treatment of cereals increased the total bacterial 16S rRNA copy numbers in ileal digesta by 0.6 log units (P = 0.046) (Table S2), whereas the total number of bacteria was not changed at the mucosa and in the ICLNs. Accordingly, the LA-treated cereals tended to decrease the species richness in ileal digesta (P < 0.10) (Table 1). Moreover, the LA-treated cereals tended to increase species richness but reduced the evenness (Shannon index) at the mucosa (P < 0.10) (Table 1). In contrast, in the ICLNs, phytase supplementation and LA treatment of cereals tended to interact (P < 0.10), resulting in enhanced species richness when LA-treated cereals were combined with dietary phytase (Table 1). In contrast, diversity (Shannon) tended (P < 0.10) to be reduced with the single treatments but not when LA-treated cereals and phytase were combined. Visualization of the ileal communities in digesta, at the mucosa, and in ICLNs using NMDS plots indicated bacterial communities at the mucosa and in ICLNs of pigs fed the diet containing LA-treated cereals and phytase that were distinct from those of the other dietary treatment groups (Fig. 2B).
Species richness and α-diversity indices of bacterial communities in ileal digesta, mucosa, and ileocecal lymph nodes of pigs fed diets with or without phytase and lactic acid-treated cereals
Animal-to-animal variation was high with respect to Lactobacillaceae and Clostridiaceae abundances; therefore, only numerical differences for treatment effects were observed (Table 2). Regarding dietary effects on the taxonomic composition in ileal digesta, the LA-treated cereals decreased levels of the phylum Fusobacteria and eight families (Helicobacteraceae [−0.81-fold], Streptococcaceae [−0.94-fold], Fusobacteriaceae [−0.92-fold], Micrococcaceae [−0.85-fold], Moraxellaceae [−0.88-fold], Neisseriaceae [−0.92-fold], unclassified Rickettsiales [−1.00-fold], and unclassified Streptophyta [−0.92-fold]), whereas it promoted unclassified Clostridiales2 (+0.50-fold) (P < 0.05) (Table 2). At the ileal mucosa, LA-treated cereals decreased the abundance of the phylum Euryarchaeota and of two families, including Methanobacteriaceae and Campylobacteraceae, whereas phytase supplementation affected only Methanosphaera (P < 0.05) (Table 2). The phytase × LA-treated cereal interaction for Peptostreptococcaceae (P = 0.007) at the ileal mucosa showed that both LA-treated cereals and phytase supplementation decreased the relative abundance of this family, whereby the depression was less pronounced with the combination of phytase and LA-treated cereals. In ICLN, the phytase × LA-treated cereal interactions indicated different responses of Actinobacteria and Bacteroidetes bacteria depending on the effect, whether phytase and LA-treated cereals were fed alone or in combination (Table 2).
Relative abundance of bacterial taxa (>0.01% of all reads) in ileal digesta, mucosa, and ileocecal lymph nodes of pigs fed diets with or without phytase and lactic acid-treated cereals
Diet-related expression of innate immune genes at the ileal mucosa.Results for the expression of genes related to TLR signaling and barrier function showed that cytokines, most suppressors of cytokine signaling, and tight-junction proteins were similarly expressed among diets at the ileal mucosa (Table 3). In contrast, phytase supplementation decreased the relative expression of CDH1 and IAP by 42.5 and 33.0%, respectively (P < 0.05). The phytase × LA interaction for the relative expression of MUC2 indicated that the LA treatment of cereals increased the expression level of MUC2 at the ileal mucosa but only with the diet without the phytase supplementation (P = 0.049) (Table 3). Likewise, LA-treated cereals as single treatment tended (P = 0.086) to cause higher IAP expression than the other three diets.
Relative expression of genes related to innate immunity and barrier function at the ileal mucosa of pigs fed diets with or without phytase and lactic acid treatment of cereals
Relationships between most influential OTUs at different sampling sites and expression of innate immune genes.Supervised sparse partial least-squares-discriminant analysis (sPLS-DA) was used to identify positive and negative relationships between the most influential operational taxonomic units (OTUs) at the different sampling sites (ileal mucosa and ICLNs) and mucosal expression of innate immune genes. The results for component 1 and 2 of the sPLS-DA are illustrated in Fig. 3. Lactobacillus mucosae-OTU7 at the mucosa and in ICLNs were positively correlated. Moreover, mucosa-associated SMB53-OTU61 and unclassified Clostridiaceae-OTU100 were positively associated with the abundances of L. mucosae-OTU7 and Lactobacillus-OTU10, whereas they were negatively correlated with Lactobacillus-OTU6. Mucosa-associated OTUs that were identified to influence mucosal MUC2 and SOC5 were L. mucosae-OTU7, SMB53-OTU61, and unclassified Clostridiaceae-OTU75 (only MUC2 expression) and -OTU100 (Fig. 3A and Table S7). For the community in ICLNs, L. mucosae-OTU7 and Lactobacillus-OTU10 were negatively associated with mucosal expression levels of MUC2 and SOC5, whereas Lactobacillus-OTU6 in ICLNs was positively linked to the expression of these two genes. For component 2 (Fig. 3B and Table S7), Actinobacillus-OTU8 and -OTU11 and unclassified Clostridiaceae-OTU63 at the ileal mucosa and Lactobacillus-OTU6 in ICLNs were negatively correlated with the expression level of CDH1, whereas unclassified Clostridiaceae-OTU1 at the ileal mucosa and Clostridiaceae-OTU144 in ICLNs were positively linked to the expression of this gene. Moreover, sPLS-DA identified relationships between the most influential OTUs in ileal digesta and at the mucosa, showing, for instance, for component 2 that Lactobacillus agilis-OTU225 and Veillonella dispar-OTU118 in digesta were positively correlated with Methylobacterium-OTU69 and Sphingomonas-OTU157 at the mucosa (Fig. S1 and Table S7).
Circos plots of horizontal sparse partial least-squares-discriminant analysis displaying correlations between the identified best discriminant bacterial operational taxonomic units (OTU; relative OTU abundance, >0.01%) at the ileal mucosa (MUC) and in ileocecal lymph nodes (ICLNs) and gene expression levels (Gene) at the ileal mucosa for component 1 (A) and component 2 (B). Positive and negative correlations (|r| > 0.5) are displayed by red and blue links, respectively. OTU1, unclassified Clostridiaceae; OTU2, OTU2.1 Lactobacillus; OTU3, OTU3.1 unclassified Clostridiaceae; OTU6 Lactobacillus; OTU7, OTU7.1 Lactobacillus mucosae; OTU8, Actinobacillus; OTU9, unclassified Enterobacteriaceae; OTU10, OTU10.1 Lactobacillus; OTU11, Actinobacillus; OTU14, Clostridiaceae; OTU29, Actinobacillus porcinus; OTU35, Lactobacillus; OTU44, Clostridium; OTU60, unclassified Clostridiaceae; OTU61, SMB53; OTU63, unclassified Clostridiaceae; OTU74, unclassified Clostridiaceae; OTU75, unclassified Clostridiaceae; OTU83, unclassified Neisseriaceae; OTU100, unclassified Clostridiaceae; OTU144, Clostridiaceae; OTU197, Flexispira; OTU349, Rhodobacteraceae paracoccus.
DISCUSSION
Since the ileum comprises organized lymphoid tissues (Peyers’s patches) (21), characterization of dietary effects on the microbial communities, including microbiome in digesta, at the mucosa, and in ICLNs, as well as components of the innate immune response and gut barrier, is highly informative to our understanding of diet-gut microbiome-host interactions. Targeting only viable bacteria by using an RNA-based approach (22), this study provides valid data on the metabolically active microbiome along the ileal digesta-mucosa-lymph node axis, which is able to grow and actively influence the host animal. Not surprisingly, present results demonstrate that the dietary treatments (phytase supplementation versus LA treatment of cereals) affected more taxa in digesta than the bacterial communities at the mucosa and in ICLNs. Nevertheless, the great rise in species richness in ICLNs when both dietary treatments were combined indicated changes in bacterial translocation due to altered mucosal barrier function and/or subsampling of bacteria by immune cells. In spite of this finding, dietary treatments affected the ileal expression levels of only a few barrier function genes (e.g., MUC2, IAP, and CDH1), indicating the low inflammatory potential of the present dietary treatment. From the diet-related alterations in the bacterial communities in digesta and at the mucosa (e.g., reduced abundance of certain Proteobacteria taxa), we would have assumed changes in the TLR–NF-κB proinflammatory signaling pathways. However, after pigs were on the diets for 18 days, a new host-microbial tolerance (23) may have been established, which may have modified the mucosal proinflammatory response. Despite not finding changes in TLR2 expression, an important receptor for building tolerance toward the commensal microbiota, TLR2-triggered expression of anti-inflammatory cytokines and negative regulators, such as Tollip and A20, may have established new tolerance toward an altered bacterial composition (23).
As illustrated in the NMDS plot, results show distinct bacterial populations in ileal digesta, at the host mucosa, and in ICLNs. In taking into account that mucosa-associated bacteria utilize host mucin as the substrate (24), greater luminal substrate availability may explain the higher bacterial richness and total bacterial gene copy number in ileal digesta than in the mucosa. In contrasting previous studies (25–27), Clostridiaceae and Lactobacillaceae were the predominant taxa at all three ileal sites, implying the importance of the luminal bacteria for colonization of the mucosa and translocation into ICLNs in the present pigs. Previously, Kelly et al. (25) reported a predominance of Helicobacteraceae and Campylobacteraceae at the ileal mucosa, which were less abundant in the current study. Likewise, Mann et al. (12) and Zwirzitz et al. (27) found that the microbiome of ICLNs was mainly composed of Proteobacteria. Moreover, high abundances of Lactobacillus species in porcine ICLNs have been reported before (9). Due to their mucosal predominance, Lactobacillaceae may have been frequently sampled by immune cells and transported into the ICLNs. However, Clostridiaceae, which formed the second most abundant family at the ileal mucosa and are comprised of some porcine opportunistic pathogens (e.g., Clostridium perfringens), were less abundant in ICLNs, supporting targeted bacterial subsampling (9, 12, 28) or inhibition by other bacteria, such as Lactobacillaceae. In this regard, selective subsampling of Lactobacillus species, such as L. murinus and L. taiwanensis, by phagocytes in the small intestinal lamina propria and Peyer’s patches has been reported for a mouse model (28). Besides Lactobacillaceae, the ICLNs were enriched in Lachnospiraceae, Veillonellaceae, and Methanobacteriaceae compared to levels in the mucosa, whereas families that comprised pathobionts (e.g., Pasteurellaceae, Helicobacteraceae, and Enterobacteriaceae) were less frequent in ICLNs. The lower frequency of pathobionts may have been an effect of a reduced subsampling by dendritic cells or the controlling action of other bacteria, such as members of the family Lactobacillaceae, e.g., via the production of antimicrobial peptides and organic acids (e.g., lactic acid and acetic acid). Lactobacilli have been reported to hinder pathogens to adhere to the mucosa and to proliferate, with benefits for host gut health (29). Although their presence in ICLNs has been repeatedly described (9, 27), little is known about whether Lactobacillus species exert similar effects in the ICLNs. Interestingly, bacterial taxa that are known to interact with lactobacilli in the gut lumen via cross-feeding, such as Lachnospiraceae and Veillonellaceae (30), were also enriched in ICLNs. However, far less is known about their role in ICLNs and their potential as pork contaminants or disease agents. Like Lactobacillaceae, Lachnospiraceae and Veillonellaceae contribute to gut health via the production of a range of bioactive molecules, such as propionate, butyrate, and valerate (30). These short-chain fatty acids, for instance, support gut barrier function and health via modulation of gene expression and suppression of proinflammatory signaling pathways in immune cells and as an energy source (31). Specific bacterial subsampling or translocation was also obvious for Methanobacteriaceae, whose abundance largely increased from digesta to the mucosa and into the ICLNs. Although this is in line with our recent findings (27), like for many other bacteria found in ICLNs, little is known about their metabolic role and impact on the host. Theoretically, Methanobacteriaceae may have benefited from cross-feeding relationships with eubacteria, as they obtain their energy for growth mostly from the reduction of CO2 and formate with H2, or, in the case of Methanosphaera, from the reduction of methanol with H2 (32).
Bacterial metabolism and replication largely rely on a sufficient P availability in their surrounding medium (33). However, the present findings of a greater effect of the LA-treated cereals on viable bacteria in ileal digesta and at the mucosa than of the phytase supplementation demonstrate that the intestinal P level was not the most important factor for bacterial growth. During the soaking process, hydrolysis and modification of the molecular structures of the various nutrient fractions (e.g., carbohydrates, proteins, and minerals) in the cereal grains occur (34), modifying their amount and degradability for microbes. Both dietary treatments improved the intestinal P digestibility and retention (35). With the jejunum being the main absorptive gut segment (33), it is valid to assume that less P was ileally available for microbes and host mucosal metabolism with the single and combined dietary treatments. However, alterations in the bacterial community indicated less favorable conditions for proteolytic bacteria, including Proteobacteria (e.g., Pasteurellaceae, Helicobacteraceae, and Campylobacteraceae), proteolytic Firmicutes (e.g., Streptococcaceae and Micrococcaceae in digesta), and Fusobacteria in digesta and/or at the mucosa with the LA-treated cereals, many of them comprising important porcine pathobionts. According to this, the reduced dietary protein availability with the LA-treated cereals may have been more important for the bacterial abundances in digesta and at the mucosa than changes in P and carbohydrate fractions (10, 17). Other microbe-to-microbe interactions, such as cross-feeding of primary fermentation metabolites (e.g., H2 in the case of Helicobacteraceae) (36) or inhibition by other bacterial species, may have played a role as well. Large animal-to-animal variation for the dominant Lactobacillaceae, however, made it difficult to distinguish specific inhibitory relationships with proteolytic taxa. Nevertheless, the lower abundance of digesta- and mucosa-associated pathobionts probably supported ileal health, causing less upregulation of the mucosal immune response via activation of pathogen recognition receptors and production of gut mucosa-irritating biogenic amines and ammonia (21).
Expression of mucin genes, such as of MUC2, may help explain mucosal abundances of mucus-colonizing species. MUC2 expression was upregulated with the LA-treated cereals but only when fed as single treatment. Helicobacter and Campylobacter, capable of binding to the glycosylated epitopes in the glycolipid fraction of the outer mucus layer (37), however, were depressed with the LA-treated cereals, indicating the interference of other factors related to the mucosal defense or microbial nature of their mucosal abundances. Moreover, in spite of being a mucin utilizer (38), Peptostreptococcaceae were also less frequent when more MUC2 was expressed with the diet containing the LA-treated cereal diet. This was also the diet with the highest expression of IAP, which has potential to modulate the gut microbiota and downregulates proinflammatory signaling at the mucosa (39). Although the present sequencing approach did not cover all methanogens (40), changes in cross-feeding relationships with carbohydrate-fermenting bacteria and lower availability of H2 (32, 41) at the ileal mucosa may explain the lower abundance of Methanobrevibacter and Methanosphaera with the LA-treated cereals. Moreover, the lower abundance of Methanosphaera with the phytase supplementation suggests an additional effect of lower ileal P availability in the bacterial-archaeal relationship.
Effects of the dietary treatments on the bacterial communities in the ICLNs may have been mostly indirect via changes in bacterial translocation and consequently to mucosal barrier. This may explain why there were mostly interactive effects between phytase supplementation and the LA-treated cereals on the bacterial composition in the ICLNs. The present gene expression results related to barrier function supported changes in the first line of defense. The higher MUC2 expression with the LA-treated cereals as single dietary treatment may be an explanation for the reduced abundance of Pasteurellaceae and Helicobacteraceae in ICLNs of pigs fed this diet. Conversely, the decreased expression of CDH1 and IAP with the phytase supplementation suggested an influence of the luminal P availability on expression of barrier function genes (19). Nevertheless, expression results for single genes should not be overinterpreted, as the mucosal barrier function is a consolidated action of several components, preventing or allowing the translocation of bacteria (21). Against this background, the identified associations of bacterial OTUs at the mucosa and in ICLNs with mucosal gene expression for component 1 would support an enhanced translocation and enrichment of certain bacterial species, e.g., L. mucosae, which may have been promoted by lower mucin 2 secretion or thinner mucus layer. Likewise, the negative correlation between Lactobacillus-OTU6 and CDH1 expression for component 2 may associate an increased translocation of this bacterium with the lower CDH1 expression.
In conclusion, present results demonstrate dietary treatment (phytase supplementation and LA-treated cereals)-related alterations in the viable microbiome along the digesta-mucosa-lymph node axis in the ileum, probably due to changes in ileal substrate availability and altered microbial-host interactions. Bacterial taxa in digesta and at the mucosa were more influenced by the LA-treated cereals than by the added phytase. In particular, proteolytic bacteria were depressed when pig’s diet contained the LA-treated cereals, whereas the combination of both dietary treatments largely increased the species richness in the ICLNs, indicating alterations in mucosal barrier function and subsampling of bacteria by immune cells. This was supported by the relationships of mucosal MUC2, IAP, and CDH1 expression levels and Lactobacillus, Clostridiaceae, and Actinobacillus OTUs.
MATERIALS AND METHODS
Animals and diets.All procedures involving animal handling and treatment were approved by the institutional ethics committee of the University of Veterinary Medicine and the national authority according to paragraph 8 of the Law for Animal Experiments, Tierversuchsgesetz (TVG) (BMWFW-68.205/0158-W F/V/3b/2016). A total of 32 castrated male pigs (Large White, 13.1 ± 2.3 kg; age, 6 to 8 weeks) were obtained from the university research farm (University of Veterinary Medicine Vienna) and randomly assigned to one of four dietary treatments in a 2 (phytase supplementation, 0 versus 500 FTU/kg complete feed)-by-2 (LA-treated versus nontreated cereals) factorial design with four replicate batches (n = 8 per replicate batch). Pigs were fed one of four diets: control diet, diet containing LA-treated cereals, control diet with phytase, and diet with phytase and LA-treated cereals. The experimental diets were based on wheat, corn, and soybean meal and were formulated to meet or exceed the current recommendations for nutrient requirements (42, 43) (see Table S1 in the supplemental material). The diets were mixed with one of two premixes, either the premix without phytase or the premix with phytase (500 FTU/kg complete feed). For the preparation of the LA-treated cereals, the corn and wheat were soaked in 2.5% LA for 48 h (34). Soaking conditions were selected based on the results of our previous in vitro experiment (34). After soaking for 48 h, the LA-treated cereals were dried in an oven at 70°C for 1 h and at 60°C for 23 h and subsequently ground to pass a 5-mm sieve before preparing the diets. At mealtime, the feed was mixed with water in a ratio of 1.15:1. Feed was offered 3 times daily at 8 a.m., 12 a.m., and 4 p.m. Feed allowances corresponded to 3-times maintenance requirement, [(body weight0.6 × 197)/238.68] × 3 (42). Two pigs were fed the same diet per replicate batch. This resulted in eight observations per diet across all four replicate batches. Pigs were individually housed and fed in metabolism pens (1.20 m by 1.00 m) with Plexiglas walls to allow visual contact. Each pen was equipped with one nipple drinker with free access to demineralized water, one feeder, and one heating lamp. Before the start of the experiment, the pigs were allowed 2 days to acclimatize to the room, pen, and feeding schedule. Pigs were weighed at the beginning and end of the experimental period. Only pigs that were clinically healthy were used in this experiment. Each replicate batch lasted 19 days, with sampling of intestinal digesta and tissues on days 18 and 19.
Slaughtering and sampling.On the last two experimental days, pigs were anesthetized (ketamine hydrochloride [Narketan], 100 mg/ml, 1 ml/10 kg body weight; Vétoquinol GmbH, Germany; and azaperon [Stresnil], 40 mg/ml, 0.5 ml/10 kg body weight; Elanco Deutschland GmbH, Germany) and euthanized (intracardiac injection of 10 ml/kg embutramide; MSD Animal Health, Vienna, Austria) 2 h after their last feeding. On both days, the order of pigs was balanced for treatments. The abdominal cavity was opened, the inner organs were examined for signs of disease, and the entire gastrointestinal tract was removed. The small and large intestines were carefully dissected from the mesentery, and clamps were used to prevent mixing of digesta between intestinal segments. After separation of the individual segments, ileal digesta was aseptically collected from the 30-cm cranial to ileocecal valve, thoroughly homogenized, aliquoted, and snap-frozen in liquid nitrogen. The ileal tube piece was opened at the mesentery, cleaned with phosphate-buffered saline, and blotted dry with paper towels. The ileal mucosa was then scraped using a glass slide and snap-frozen in liquid nitrogen. Before sampling, the ileal mucosa was examined for signs of inflammation. The ICLNs were collected near the ileocolic artery caudal to the cecum and placed into ice-cold phosphate-buffered saline. ICLNs were disinfected by flaming. Surrounding tissues (i.e., fat and connective tissues) were removed before cutting the lymph nodes in tiny pieces and snap-freezing them in liquid nitrogen. All samples for RNA isolation were collected within 20 to 30 min after the death of the animal. All digesta and tissue samples were stored at − 80°C before analysis.
Chemical analysis.Prior to chemical analysis, diets were homogenized and ground through a 1-mm screen (GRINDOMIX GM200 and Ultra-Zentrifugalmühle ZM200; Retsch, Haan, Germany). Proximate nutrient analysis, including the determination of dietary Ca and P, was performed according to VDLUFA (44). Analysis of alpha amylase-stable NDF exclusive of residual ash and acid detergent fiber exclusive of residual ash was done using Fiber Therm FT 12 (Gerhardt GmbH & Co. KG, Königswinter, Germany) with heat-stable α-amylase (18). Resistant starch and nonresistant starch were analyzed spectrophotometrically using a commercial enzymatic assay (K-RSTAR; Megayzme International Ireland, Ltd., Braz, Ireland).
Mucosal gene expression.Total RNA was extracted from 20 mg ileal mucosal scrapings using the RNeasy Mini Qiacube kit (Qiagen, Hilden, Germany). Each sample was combined with 350 μl lysis buffer (Qiagen) and 0.6 g ceramic beads (diameter [Ø], 1.4 mm). Samples were homogenized for 30 s (6.5 m/s) using the FastPrep-24 instrument (MP Biomedicals). Afterwards the remaining procedures were performed on the Qiacube robotic workstation (Qiagen, Hilden, Germany). The isolated RNA was treated with DNase I (Turbo DNA kit; Life Technologies, Ltd., Vienna, Austria) to remove genomic DNA. The Qubit 2.0 fluorometer (Life Technologies Corporation, CA, USA) and the Qubit RNA assay kit were used to quantify the total RNA concentration. The quality of total RNA was checked using an Agilent Bioanalyzer 2100 (Agilent Technologies, Wagehaeusel-Wiesental, Germany) and an Agilent RNA 6000 Nano assay. The obtained RNA integrity numbers ranged between 7 and 10. The cDNA was synthesized from 1 μg RNA using a high-capacity reverse transcription kit (Life Technologies, Foster City, CA) by following the manufacturer’s instructions modified by adding 0.5 μl RNase inhibitor (Qiagen, Hilden, Germany) to each reaction mixture.
Primers were designed and, together with the previously published ones (Table 4), verified using PrimerBLAST (www.ncbi.nlm.nih.gov/tools/primer-blast/). The genes selected were related to the innate immune response and included genes for barrier function, proinflammatory, anti-inflammatory, and regulatory cytokines and genes within the Toll-like-receptor 2 and 4–NF-κB signaling pathways. The expression of target innate immune genes (i.e., tight junction protein 1 [ZO1], occludin [OCLN], cadherin 1 [CDH1], claudin 1 [CLDN1] and CLDN4, mucin 2 [MUC2] and MUC4, Toll-like receptor 2 [TLR2] and TLR4, interleukin-1B [IL-1B], IL-6, IL-8, and IL-10, interferon gamma [INFG], tumor necrosis factor alpha [TNFA], transforming growth factor beta 1 [TGFB1], suppressor of cytokine signaling 1 [SOC1], SOC2, SOC3, SOC4, SOC5, and SOC6, intestinal alkaline phosphatases [IAP], and nuclear factor kappa B [NFKB]) and of housekeeping genes were amplified on a Stratagene Mx3000P quantitative PCR system (Agilent Technologies) using previously published or newly designed primer sets (Table 4). All reactions were performed in duplicates in a total volume of 20 μl containing 50 ng cDNA, 10 μl Fast Plus Eva green master mix with low ROX (Biotium, Hayward, CA, USA), 100 nM (each) forward and reverse primers, and 10 μl diethyl pyrocarbonate-treated water (Bioscience, Missouri, MO, USA) on a 96-well plate. Duplicates of no template controls and minus reverse transcription controls were included on each plate. The amplification program used the following cycle conditions: denaturation of 95°C for 5 min followed by 40 cycles of 95°C for 15 s, primer annealing at 60°C for 30 s, and elongation at 72°C for 30 s, whereby the fluorescence was measured in the last step of each cycle. Melting curve analysis was used to determine the specificity of the amplification. Of the 5 housekeeping genes (HKG), B2M, GAPDH, and HPRT were most stably expressed and used to normalize the expression levels of the target genes (3). The relative expression of the target genes was calculated according to the 2−ΔΔCT method. The averaged quantification cycle (Cq) values of the most stably expressed HKGs were subtracted from the Cq value of the target gene to calculate the ΔCq value. The ΔΔCq value was calculated from the difference between ΔCq value of each sample and a reference sample that had the lowest ΔCq value and, hence, highest expression for the respective gene.
Primers used for gene expression related to innate immunity and barrier function
Analysis of the metabolically active bacteria.Total RNA was isolated from ileal digesta, mucosa, and ICLN using the RNeasy PowerMicrobiome kit by following the manufacturer’s instructions (Qiagen, Hilden, Germany), with some modifications with respect to the bead-beating procedure. Approximately 250 mg of frozen samples was weighed into 2-ml tubes containing 0.6 g glass beads (Ø, 0.1 mm), 0.4 g sterile ceramic beads (Ø, 1.4 mm), 0.55 g sterile ceramic beads (Ø, 2.8 mm), and 650 μl guanidinium thiocyanate buffer. The samples were beaten 3 times for 1 min (6.5 m/s) on the FastPrep-24 instrument (MP Biomedicals, Heidelberg, Germany) with cooling on ice between the individual bead-beating steps. The remaining steps of the RNA isolation were according to the protocol of the RNeasy PowerMicrobiome kit. Reverse transcription PCR as well as quantification and quality measurements were done as described for the RNA isolation from mucosal tissue.
Amplicon sequencing of cDNA was performed on an Illumina MiSeq sequencing platform by a commercial provider (Microsynth, Balgach, Switzerland) and included the 16S rRNA PCRs, library preparation, and sequencing. The primers 341F-ill (5′-CCTACGGGNGGCWGCAG-3′) and 802R-ill (5′-GACTACHVGGGTATCTAATCC-3′) were used to target the V3-V4 hypervariable regions of the bacterial 16S rRNA gene, which generate an amplicon of approximately 460 bp. The 16S rRNA PCRs were performed using the KAPA HiFi HotStart PCR kit (Roche, Baden, Switzerland). The Nextera XT DNA sample preparation kit (Illumina) was used for preparation of libraries by ligating sequencing adapters and indices onto the purified PCR products. After library normalization, the equimolar quantities of each library were pooled and sequenced on an Illumina MiSeq sequencing v2 platform using a paired-end protocol. Subsequently, reads were demultiplexed and adapter sequences were removed using cutadapt (https://cutadapt.readthedocs.org/). The overlapping paired-end reads were stitched using USEARCH (drive5/com).
The stitched reads obtained from Microsynth were processed using the software package Quantitative Insights into Microbial Ecology (QIIME, v1.9.4) (45). Quality filtering of fastq files was performed using the split_libraries_fastq.py command for nonmultiplexed Illumina fastq data with Phred offset of 33. Detection and removal of chimeric sequences were performed with the UCHIME method using the 64-bit version of USEARCH (46, 47) and the GOLD database (www.drive5.com). Open-reference OTU picking was performed to cluster sequences at 97% similarity level using UCLUST (46). Taxonomy was assigned using the Ribosomal Database Project (RPD) classifier (naive Bayesian classifier) with the GreenGenes database (http://qiime.org/home_static/dataFiles.html) (version 13_8) (48), whereby rare OTUs with fewer than 10 sequences were removed. At all taxonomic levels, the raw read counts from the various bacterial taxa were collapsed and compositionally normalized such that each sample sums to 1.
Alpha-diversity measurements were performed using the vegan R package (version 2.5.2) (49) in R studio (version 1.0.136). For β-diversity analysis, dissimilarity matrices (Bray-Curtis) derived from OTU data were calculated with permutational multivariate analysis of variance (PERMANOVA) using the adonis2 function and visualized in two-dimensional nonmetric multidimensional scaling (NMDS) ordination plots obtained with the metaMDS function in the vegan package (49). The identification of the most discriminant OTUs and expressed innate immune genes was performed by multigroup supervised DIABLO N-integration in the R package mixOmics (version 6.3.2) (50). Two horizontal sparse partial least-squares-discriminant analyses (sPLS-DA) using the block.splsda function were applied to identify key features and their relationships among data sets. The first sPLS-DA was applied to integrate the data sets of relative abundances of OTUs at the ileal mucosa (data set MUC) and ileocecal lymph nodes (data set ICLN) as well as the expression levels of innate immune genes (data set Gene) at the ileal mucosa to classify and select key parameters from each data set. In order to determine the main OTUs at the mucosa and in ICLN and mucosal expression levels of genes that allow discrimination of treatments groups with the lowest possible error rate in the sPLS-DA, we tuned the number of retained variables. In doing so, we retained one-fourth of all OTUs at the ileal mucosa (n = 12), the 6 major OTUs (relative abundance, >0.01%) in ICLN, and the most influenced innate immune genes (n = 3) for components 1 and 2. In the second sPLS-DA, the data sets for OTUs in ileal digesta (data set DIG) and OTUs at the ileal mucosa (data set MUC) were tuned, retaining one-third and one-fifth of the variables from the DIG and MUC data sets, respectively, for components 1 and 2. The sPLS-DA results were visualized as Circos plots showing the strongest positive and negative Pearson’s correlations between most discriminant variables for each subset of data.
Statistical analysis.Bacterial taxa (>0.01% of all reads) appearing in >50% of samples were considered. The ranked relative abundances were analyzed in SAS. The Shapiro-Wilk test with the UNIVARIATE procedure in SAS (version 9.4; SAS Institute, Inc., Cary, NC) was used to check for normal distribution of all variables. Nonnormally distributed data were log transformed. Repeated measures were used to assess differences in the total bacterial 16S rRNA gene copies, species richness, and α-diversity indices among ileal digesta, mucosa, and ICLN samples using the MIXED procedure in SAS. To compare differences between dietary treatments, data for species richness, α-diversity, and bacterial taxa, as well as for relative gene expression, were subjected to ANOVA using the MIXED procedure in SAS. The model accounted for the fixed effects of phytase supplementation, LA treatment of grains, their two-way interaction, separately per sampling site, and replicate batch as random effect. Pig was the experimental unit. Degrees of freedom were approximated by the Kenward-Rogers method (ddfm = kr). The pairwise comparisons among least-square means were performed using the pdiff statement. The least-square means ± standard errors of the means (SEM) are reported. A significant difference was declared at a P value of ≤0.05 and trends at 0.05 < P ≤ 0.10.
Sequence data accession number.Raw sequence data are available at the NCBI BioProject databank (PRJNA565652).
ACKNOWLEDGMENTS
We thank T. Enzinger, A. Sener, M. Hollmann, A. Dockner, M. Wild, and S. Sharma of the Institute of Animal Nutrition and Functional Plant Compounds for their excellent assistance with animal work and laboratory analysis.
J.K. acknowledges funding under the project of the ASEAN-European Academic University Network (ASEA UNINET) from the Austrian Agency for International Cooperation (Österreichischer Austauschdienst, OEAD) for the financial support in the Ph.D. program in Austria.
FOOTNOTES
- Received 17 September 2019.
- Accepted 20 November 2019.
- Accepted manuscript posted online 22 November 2019.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.