Initial clinical course
Among the 75 cesarean-delivered preterm piglets, nine were excluded before randomization (e.g. failed resuscitation, stillbirth), whereas the remaining 66 animals were group allocated. An additional seven animals were euthanized preschedule for reasons not related to the interventions (respiratory failure, iatrogenic complications). Two animals were euthanized preschedule with clinical NEC signs (1 CON, 1 FFTr), whereas the remaining 57 animals survived until day 5. During the course of the experiment, we observed rectal bleeding in 31% (5/16) of CON and 19% (3/16) of FMT animals relative to 0% (0/13) in both FFT groups (p < 0.05 vs. CON).
Gut pathological evaluation
We performed gross examination of the gastrointestinal tract supported by histopathological assessment of ileum and colon to evaluate the severity and extent of NEC-like lesions. The NEC-like pathological phenotype in formula fed preterm pigs consisted of extensive hemorrhage with or without patchy necrosis of the mucosa and pneumatosis intestinalis, mostly affecting the ascending and transverse colon, and to a lesser extent ileum and stomach (Fig. 1B). At the microscopic level, the pathological observations included subtle changes to the mucosal architecture, progressing from epithelial sloughing and hemorrhage to complete destruction of mucosal integrity among the most severe cases (Fig. 1C).
In this setting, oro-gastric FFT administration markedly reduced macroscopic NEC severity (p < 0.01, Fig. 1B) and consequently reduced NEC incidence to 0% (p < 0.01 vs. CON). Rectally administered FFT also reduced NEC severity (p < 0.01) but not incidence relative to CON. However, rectally administered FMT, which we previously found to be clearly NEC protective [8], failed to reduce NEC severity and incidence relative to CON. No bleeding occurred after rectal fluid administration, and no rectal lesions were observed at necropsy in rectally administered animals. The microscopic evaluation supported the macroscopic effects of oro-gastric FFT, which reduced the histopathological NEC severity and incidence relative to both CON and FMT (all p > 0.05, Fig. 1C), whereas no significant effects were found for rectally administered FFT on microscopic level.
Safety assessment
We next investigated a series of safety parameters. Initially, we found that the FMT group had a significantly higher proportion of animals with a negative body growth rate compared with CON and FFTo (both p < 0.05, Fig. 2A, B). Furthermore, the relative weight of the small intestine but not the colon was lower in FMT animals relative to FFTo (p < 0.05, Fig. 2C). Additionally, the urinary lactulose-mannitol ratio, an in vivo marker of small intestinal permeability, was robustly decreased by FFTr (p < 0.05 vs. CON, Fig. 2D), whereas the FMT group had a significantly higher permeability than both FFT groups (both p < 0.05). Finally, the lactase enzymatic activity, a measure of mucosal integrity, was significantly decreased in the FMT group relative to FFTr (Fig. 2E).
A Relative body growth rate from birth to day 5. B Proportion of animals with negative body growth rate from birth to day 5. C Relative small intestinal weight. D Small intestinal permeability. E Brush-border lactase enzyme activity in three segments of small intestine. SI small intestine; Asterisk and double asterisks denote statistical probability levels below 0.05 or 0.01, respectively.
Bacterial composition of gut mucosa and lumen
In general, the composition of mucosa-associated bacteria differed from luminal bacteria (R2 = 0.11, p < 0.001, Supplementary Fig. S2). Intervention effects were seen in bacterial composition of both the mucosa (R2 = 0.17, p < 0.001, Fig. 3A) and luminal compartments (R2 = 0.27, p < 0.001). The greatest effect was observed in the FMT group, but both FFT groups were also significantly different from CON in both mucosa and gut lumen. However, FFT route of administration (FFTr vs FFTo) did not affect the bacterial composition. The Shannon index of mucosa-associated bacteria was similar among groups, but interestingly a positive correlation between Shannon index and small intestinal NEC severity was observed (p < 0.001, Supplementary Fig. S3). The luminal bacterial Shannon index increased only in response to FMT treatment (Fig. 3B).
A Principal component analysis plots visualizing beta diversity based on unweighted UniFrac metrics. FDR-adjusted probability levels of pairwise comparisons are reported in adjacent tables. B Shannon index as a measure of alpha diversity. C Relative bacterial abundance summarized at genus level. Genera with more than 1% relative abundance across groups were included. Bacterial composition of donor fecal material is included as reference. Asterisk and double asterisks denote statistical probability levels below 0.05 or 0.01, respectively.
Relative to CON, most relative differentially abundant bacterial OTUs were observed in the luminal compartment of FMT animals, whereas the number of differentially abundant mucosa-associated bacteria were comparable between treatments (Supplementary Fig. S4). FFT treatment substantially increased Streptococcus relative abundance but not Lactobacillus relative to CON, whereas FMT conversely increased Lactobacillus relative abundance but not Streptococcus, although the donor stool contained high proportions of both these genera (Fig. 3C). Interestingly, while the relative abundance of Proteobacteria (~10%) was similar among groups in the gut lumen, <1% Proteobacteria were detected from the mucosa of oro-gastrically administered FFT animals, whereas ~10% of the mucosal microbiota of CON consisted of Proteobacteria (e.g. Enterobacteriaceae, Klebsiella) and ~5% for FMT (Fig. 3C).
Viral and phage composition of gut mucosa and lumen
The luminal virome of FMT animals differed from all remaining groups, and both FFT groups differed from CON, whereas the effect of FFT administration route was only borderline significant (Fig. 4A). Interestingly, as opposed to bacteria, the Shannon index of mucosa-associated viruses increased following FMT and FFTo, but not FFTr relative to CON (both p < 0.05, Fig. 4B). The same pattern was observed for viral Shannon index in the luminal compartment. No correlation between mucosal viral Shannon index and small intestinal NEC severity was observed (data not shown). The majority of identifiable viruses across the groups were prokaryotic viruses (phages) primarily belonging to the order Caudovirales (Fig. 4C), although eukaryotic viruses were also identified but in low relative abundance (less than 1%). Notably, while FMT led to mucosal and luminal enrichment of eukaryotic virus families (e.g. Herpesviridae) relative to CON, this was not the case for FFT treatment (Supplementary Fig. S5). The number of relative differentially abundant viral OTUs was again highest in the gut lumen of the FMT recipients, but for the mucosa-associated virome, oral but not rectal FFT administration changed as many viral OTU relative abundances as FMT. Remarkably, several members of the phage family Microviridae, which naturally prey on enterobacteria, were increased in relative abundance only in the mucosa of the orally administered FFT group (Supplementary Fig. S5).
A Principal component analysis plots visualizing beta diversity based on constrained Bray–Curtis dissimilarity. FDR-adjusted probability levels of pairwise comparisons are reported in adjacent tables. B Shannon index as a measure of alpha diversity. C Mean relative viral abundance in gut mucosa and lumen summarized at family level. Taxa with more than 0.5% relative abundance across groups were included. Viral composition of donor fecal material is included as reference. Viral taxa marked with an asterisk are bacteriophages. Asterisk and double asterisks denote statistical probability levels below 0.05 or 0.001, respectively.
Host gut mucosal transcriptome profiling
To gain insight into the host mucosal response to treatment, we then performed RNA-Seq on ileal mucosa samples. Initially, a plotting of principal components showed that the FMT group separated from the remaining groups along the second principal component (Fig. 5A). Notably, for the two FFT groups we found no statistically significant differentially expressed genes relative to CON (FDR adjusted p < 0.10), whereas FMT increased the expression of 86 genes and decreased the expression of 41 genes compared with CON (Fig. 5B). When applying fold-change criteria (log2 > 1), 29 and 16 known genes were up- and downregulated by FMT relative to CON, respectively (Supplementary Table S2). A network analysis of differentially expressed genes identified lipopolysaccharide (LPS) response genes (e.g. TLR4, CD14, THEMIS2, TNIP3) as key genes in the FMT-enriched network, whereas several interferon-induced genes such as IFIT1 and OASL were downregulated by FMT (Fig. 5C). Indeed, functional annotation of differentially expressed genes in FMT vs. CON mucosa showed that the most significantly affected pathways were related to immune activation and host defense mechanisms. Interestingly, FMT upregulated genes involved in bacterial response pathways and downregulated genes related with viral response (Supplementary Table S3).
A Principal component analysis plot of global gene expression differences based on RNA Seq. data. B Volcano plots of pairwise comparisons with the CON group, highlighting differentially expressed genes based on false discovery rate-adjusted statistics and fold-change criteria. C Gene interaction network showing FMT-upregulated genes in blue and FMT-downregulated genes in red. The genes with the highest number of interactions are highlighted with larger nodes.
Systemic immune cell characterization
Finally, we measured the levels of basic immune cell subtypes in the bloodstream both shortly after intervention and at euthanasia to investigate any induction of systemic immunity. On day 3, i.e. shortly after the final treatment administration, we found marginally increased neutrophil counts in all intervention groups relative to CON (all p < 0.05, Supplementary Fig. S6), whereas monocyte and total lymphocyte levels were not affected. However, the helper T cell (CD4+CD8−) and naive T cell fraction (CD4−CD8−) were increased and decreased, respectively in FMT relative to FFTr (both p < 0.05). Two days later, these differences had all disappeared.
Source: Ecology - nature.com
