Selenium yeast modulated ileal transcriptome and microbiota to ameliorate egg production in aged laying hens

Background: The declines in both laying performance and egg shell quality during late production period have adverse effects on long production cycle. Improving nutrition of laying hens is a crucial measure to reverse the declination. Selenium (Se) plays important roles in antioxidant defense, redox balance, immune response, and modulation of gut microbiota. However, the mechanism underlying selenium yeast regulating the interaction between transcriptome and gut microbiota to inuence laying performance, is still unclear. Here, we use the transcriptome and 16S rRNA analysis to investigate how selenium yeast alters the gene expression and microbiota composition of ileum in aged laying hens. Results: In this study, selenium yeast ameliorated the depression in aged laying performance with a signicant increase of laying rate in 0.30 mg/kg group. Furthermore, functional enrichment and STEM analysis were performed using RNA-Seq, which indicated selenium yeast activated metabolic progresses (e.g. Glycerolipid metabolism, Glycerophospholipid metabolism, and fatty acid metabolism), immune response and oxidative stress response. Four hub genes (TXNRD1, DLD, ILK and LZTS2) were involved in intestinal metabolism which was closely associated with Se deposition/status. Additionally, Se increased the abundance of benecial bacteria including Veillonella, Turicibacter, and Lactobacillus whiledecreasing the abundance of pathogenic bacteria Stenotrophomonas by 16S rRNA-Seq. The Integrated analysis of omics revealed that several microbiotas (Maritalea, Alteromonas, Geobacter, etc.) were positively associated with both Se content and laying rate, and there was a markable correlation between several specic microbiotas (Aliivibrio, Anaerobacillus, Shewanella, etc.) and the immune response pathways (regulation of acute inammatory response, positive regulation of lymphocyte activation and IFN gamma response). Meanwhile, the “switched on” gene PSCA had a positive relationship with Veillonella, and a negative relationship with the opportunistic pathogens Stenotrophomonas. CCA analysis indicated that both the Se content and the laying rate these indicated that Se supplementation can modulate the microbiota hens by not only increasing the abundance of some benecial bacteria such as Veillonella, Turicibacter and Lactobacillus, but also by decreasing the abundance of the opportunistic pathogen Stenotrophomonas. correlate with obesity-associated metabolic parameters. represents the relationship is positive, rectangle towards innermost represents the relationship is negative. (c) A heatmap of signicant correlations between GO pathways in specic clusters drawn from GSEA and genera abundance for aged laying hens determined by the HALLA method. (d) Heatmaps of Pearson correlation analysis between abundance of bacterial genera and ON as well as OFF gene. (e) A heatmap of Pearson correlation analysis between abundance of bacterial genera and selenoprotien gene expression in aged laying hens. (f) Constrained correspondence analysis reveals the correlations among the relative abundance of the specic microbes, the selenium content in ileum, laying rate, and the pathways in specic clusters in the SY groups. represents the relationship is positive, rectangle towards innermost represents the relationship is negative. (c) A heatmap of signicant correlations between GO pathways in specic clusters drawn from GSEA and genera abundance for aged laying hens determined by the HALLA method. (d) Heatmaps of Pearson correlation analysis between abundance of bacterial genera and ON as well as OFF gene. (e) A heatmap of Pearson correlation analysis between abundance of bacterial genera and selenoprotien gene expression in aged laying hens. (f) Constrained correspondence analysis reveals the correlations among the relative abundance of the specic microbes, the selenium content in ileum, laying rate, and the pathways in specic clusters in the SY groups.

Se-limiting conditions, the gut microbiota might compete with the host for the limited Se supply [29], this could also occur in the small intestine, where Se is believed to be primarily absorbed [30].
Currently, the potential mechanisms regarding the correlation between dietary selenium yeast, laying performance, microbiota, and the host's gene expression are still ambiguous. On the basis of the interactions between Se intake and gut microbiota, we hypothesized that the bene ts of selenium yeast supplementation on gut health were associated with alterations in gene expression and gut microbiota, contributing to a better egg laying performance in poultry. To clarify how dietary selenium yeast affects intestinal tract in aged laying hens, we performed transcriptomic and bioinformatics analysis in order to explore the relationships among laying performance, selenium yeast intakes and gene expression of the host. Additionally, 16S rRNA sequencing was used to analyze the composition of gut microbiota in aged laying hens maintained on selenium-de cient, sodium-selenite, and selenium-yeast diets. Moreover, the correlation analysis of the host's transcriptome, microbiota, egg production and Se intake revealed their interaction effects on egg production in aged laying hens.

Results
A proper dose of selenium yeast supplementation ameliorates the depression in laying performance during the late laying period To explore the effect of selenium yeast supplementation on production performance in aged laying hens, we administered a selenium-de cient diet for 6 weeks to obtain the selenium de cient hens. Immediately following selenium-de cient treatment, the Se content in plasma has dramatically dropped and then remained at a lower status for another 4 weeks (Fig. 1a), suggesting that the Se in aged laying hens has depleted. Then, a total number of 375 selenium-exhausted aged laying hens were fed selenium-de cient (SD), selenium yeast (SY), and sodium selenite (SS) diet for 12 weeks. Resulting in the gradual decline of the laying performance during the supplementation period (Fig. 1b). Within 4 weeks, no signi cant differences in laying performance were found, while being signi cantly higher in the SY0.30 group than the SS0.45 group and SY0.15 group after 8-week and 12-week supplementation, respectively (Fig. 1b). These results revealed that 0.30 mg/kg selenium yeast supplementation alleviate the depression in laying performance during the late laying period. We further investigated Se status in ileum after Se supplementation. The Se content in Se supplementation of higher dose had dramatically increased compared to the SD group (Fig. 1c). Meanwhile, the Se content in the SY group was higher than the SS group at the same dose (Fig. 1c). These data highlighted that the Se deposition in ileum is dosedependent on the dietary Se intake, and compared with inorganic selenium, selenium yeast has a wellabsorbed characteristic.
In general, the peaking production of commercial laying hens usually remains at the age of 60 weeks, and after that period the laying rate and egg quality begin to decline [31]. Thus, age may be one of the leading factors causing egg performance reduction. Additionally, we investigated the effect of Se supplementation on ileum aging by testing β-galactosidase activity. There were no signi cant differences between the SD group and the SY groups (Fig. 1d, e). Interestingly, a signi cant difference was observed between the SS group and the SY group (Fig. 1f), suggesting that different forms of Se supplementation may have reversed effects on ileum aging.
Transcriptome analysis revealed selenium yeast supplementation may affect metabolic pathways and immune response in the ileum in aged laying hens To gure out the effects of different doses and forms of Se, we analyzed the gene expression pro le of ileum in aged laying hens under Se supplementation using RNA-sEq. Compared with the SD group, 1857(538 upregulated; 1319 downregulated), 627(207 upregulated; 420 downregulated) and 3305(1379 upregulated; 1926 downregulated) differentially expressed genes (DEGs) were obtained in SY0.15 group, SY0.30 group and SY0.45 group, respectively ( Fig. 2a and Table S1). Simultaneously, 366 DEGs were shared among three groups. To de ne the biological functions of all DEGs in the SY groups, GO term and KEGG pathway analysis were carried out. The common enriched pathways in three groups were Glycerophospholipid metabolism, Glycerolipid metabolism and DNA-binding transcription factor activity, suggesting selenium yeast supplementation (discarding dose effect) primarily in uenced lipid metabolism (Glycerolipid and Glycerophospholipid metabolism) on ileum in aged laying hens (Fig. 2b, Fig. S1a and Table S2). Due to 0.30 mg/kg selenium yeast alleviating the depression in laying performance, we focus on the unique pathways in SY0.30 group. The ion transmembrane transport, calcium ion transport into cytosol, and MAPK signaling pathway were unique in the SY0.30 group ( Fig. 2b and Fig. S1a). These results suggested that 0.30 mg/kg selenium yeast probably regulated both the redox status and cell proliferation, and alleviated the depression in laying performance through MAPK signaling pathway and calcium ion transport.
Next, the relationship between Se content and gene expression pro le in ileum was analyzed by the weighted gene co-expression network analysis (WGCNA). WGCNA analysis showed that all the expressed genes were segmented into 17 modules and the correlation of each mcodule with Se content was estimated (Fig. 2c, Fig. S1b and Table S3). The expression of genes in the MEbrown and MEdarkgrey module was positively correlated with the Se content of ileum, while it was negatively correlated in MEgreen, MEblue, MEturquoise, and MEdarkred. To understand the features of the correlative gene expression module, we performed GSEA analysis. Co-expressed host genes in the MEbrown module were enriched in functions related to fatty acid metabolism, metabolism of steroids, bile acid metabolic process, and glycerolipid catabolic process, while those in the MEgreen module were related to organs or tissues speci c immune response and regulation of T/B cell proliferation (Fig. 2d). Meanwhile, coexpressed host genes in the MEblue module were involved in the carbohydrate transmembrane transporter activity and the activation of matrix metalloproteinases (Fig. 2d). To screen the core genes affected by selenium yeast treatment, we constructed the network structure of genes within modules and assessed the connectivity of host genes to identify hub genes. A total of 68, 5, 634, and 164 candidate hub genes exhibited high intramodular connectivity and signi cant correlation with Se deposition in ileum were identi ed from the brown, green, blue, and turquoise module, respectively ( Fig. 2e and Fig. S1c). Based on the GO annotation and genes closely related to the metabolism pathway, we further selected the candidate hub genes, such as TXNRD1 (enriched in Selenocompound metabolism pathway), DLD (enriched in Citrate cycle/TCA cycle pathway, and Pyruvate metabolism pathway), ILK and LZTS2 (enriched in regulation of canonical Wnt signaling pathway (Fig. S1d, e). Overall, these data demonstrated that the genes modulated by dietary selenium yeast intake were mainly involved in fatty acid metabolism, immune response, and carbohydrate transmembrane transporter activity, which in uenced deeply intestinal digestion and metabolism.
Furthermore, to investigate the alterations of these DEGs patterns with different dose selenium yeast treatment, trend analysis was carried out. The total DEGs were clustered into 10 pro les on the basis of the colored block signi cant enrichment (P 0.05) and the same color block with similarity greater than 0.88 (Fig. 2f). The clusters strongly associated with selenium yeast treatment further investigated the biological functions by GSEA. The expression of 199 genes (cluster 3) displayed a decrease at 0.15 mg/kg selenium yeast and an increase at 0.30 mg/kg selenium yeast and then again a decrease at 0.45 mg/kg selenium yeast, which were mainly enriched in the metabolic process (including carbohydrate metabolic process, fatty acid metabolic process, triglyceride metabolic process, and cellular ketone metabolic process), in ammatory response, reactive oxygen species biosynthetic process and developmental cell growth ( Fig. 2g and Table S4). The expression of 74 genes (cluster 8) increased during low dose, peaked at 0.15 mg/kg and subsequently decreased at 0.45 mg/kg, returning to the presupplementation levels (basal lines). The GSEA results showed that cluster 8 was major enriched in regulation of digestive system process, aging and in ammatory response (Fig. 2f, g). The expression of 44 genes (cluster 9) had a decrease at a dose of 0.15 mg/kg and 0.30 mg/kg and was restored at 0.45 mg/kg, which were predominantly enriched in innate immune response and Wnt signaling pathway (Fig. 2f, g). Additionally, the expression of 89 genes (cluster 10) exhibited an opposite trend compared with the cluster 9, which was majorly enriched in intestinal absorption and digestion, metabolic process (including lipid metabolic process, glycerolipid metabolic process, glycerophospholipid metabolic process, sterol and steroid metabolic process), oxidative stress response and in ammatory response ( Fig. 2f, g). These results indicated that the dose of selenium yeast supplementation affected various biological functions.
Generally, selenium yeast is a high-quality organic Se source for animals which is incorporated into proteins structures to improve bioavailability compared with inorganic sources, such as sodium selenite [28]. Differences in the sources of Se may have diverse effects on gene expression of hosts. Thus, we investigated the effect of different Se sources under the adequate Se status in aged laying hens. 586 DEGs were obtained in the SY0.45 group relative to the SS0.45 group (Fig. S1f and Table S1). GO enrichment analysis indicated that the differences between selenium yeast and sodium selenite supplementation were enriched in fatty acid metabolic process, cellular lipid metabolic process, and oxidoreductase activity ( Fig. S1g and Table S2). Meanwhile, KEGG analysis suggested that the major differences were in Glycerophospholipid metabolism, Glycerolipid metabolism, PPAR signaling pathway, Arachidonic acid metabolism, mTOR signaling pathway, and Fatty acid degradation in aged laying hens ( Fig. S1h and Table S2).
Selenium yeast supplementation may affect the ileal metabolic process by changing the expression of redox and aging genes On the basis of the above results, we further investigated the mechanism of regulating ileal metabolism and immune function after selenium yeast supplementation. The gut could turn 'on' or 'off' collections of the genes via epigenetics and receptor-driven transcription factors in response to the perceived environment [32]. Thus, we aimed to detect the genes switched-on or off under different doses of selenium yeast supplementation. NCAN was switched on in 0.15 mg/kg selenium yeast, while ST8SIA6 and LOC101747588 were switched off. BCAN was switched on in 0.30 mg/kg selenium yeast, while CSRP3 was switched off. CDRT1 was switched on in 0.45 mg/kg selenium yeast, while MTNR1B and SLCO1A2 were switched off ( Fig. 3a and Table S5). Furthermore, Protein-Protein Interaction (PPI) analysis showed that switched-on genes including NCAN, BCAN, CSF3 and CDRT1 were associated with carbohydrate metabolism or immune response. Switched-off genes including ST8SIA6, SLCO1A2 were associated with carbohydrate biosynthesis and lipid metabolism, respectively (Fig. 3b). These ndings suggested that selenium yeast may affect metabolism and immune system of aged laying hens through switching on or off the expression of speci c genes.
Different doses of selenium yeast supplementation can determine tissue Se deposition. Thus, the patterns of selenoprotein gene expression in ileum were examined. Twenty-one selenoprotein genes were divided into three expression patterns. The genes of pattern 1, including TXNRD1, TXNRD2, TXNRD3, DIO1, SELENOO, and SELENOI, decreased in groups with Se supplementation compared with that in SD group. Six selenoprotein genes, GPX4, MSRB1, SELENOF, SELENOT, SELENOK, SELENOS, in the pattern 2 exhibited the highest expression in the SY0.15 group, and decreased when increasing the dose. the expression of nine selenoproteins in pattern 3, namely DIO2, GPX1, GPX3, GPX7, GPX8, MSRB3, SELENOP2, SELENOM, SELENOW, were increased with dose increasing of Se (Fig. 3c). The GSEA analysis showed that the effects of selenium yeast supplementation were associated with the expression of selenoproteins enriched in adipose tissue development, lipid localization, selenoamino acid metabolism and Glycerophospholipid biosynthesis ( Fig. 3d and Table S6).
In general, dietary Se supplementation plays an important role in numerous antioxidative and in ammatory processes [35]. Thus, the expression of genes associated with the redox process had to be evaluated (Table S7). The expression of these genes was classed into three patterns and enriched in intestinal absorption, glycerolipid catabolic process, transport of bile salts and organic acids, and regulatory T cell differentiation (Fig. 3e, f and Table S6). In order to explore the effect of Se intake on aging, we analyzed the expression of genes associated with aging under selenium yeast supplementation of different doses ( Fig. 3g and Table S7). The functions of these genes were mainly enriched in intestinal absorption, digestive system process, organ or tissue speci c immune response, and gluconeogenesis ( Fig. 3h and Table S6). These ndings provided the evidence that supplementation of selenium yeast may affect the pattern of ileal selenoprotein expression, including the intestinal absorption, the digestive system process and the immune response, by regulating the genes associated to redox and aging process.
Se supplementation modulation of gut microbiota composition and structure in aged laying hens It is known that dietary Se supplementation can affect the composition of the gut microbiota, while the gut microbiota also in uences Se bioavailability and selenoprotein expression in mice [24,29]. Therefore, in order to assess the relationship between the microbiota and Se supplementation, the 16S rRNA sequencing based on V3-V4 regions was performed. In the comparison between the SD group and SY groups, 1214231 high-quality sequence reads were generated and the length was distributed on 420-440 bp ( Figure S2a). The remaining clean tags were clustered into OTUs based on 97% similarity. 217, 237, 212 and 234 OTUs were generated in the SD, SY0.15, SY0.30 as well as SY0.45 group, respectively. 137 OTUs were shared among 4 groups, and 18, 18, 8 and 12 OTUs were unique in the SD, SY0.15, SY0.30 as well as SY0.45 group, respectively (Fig. 4a). Compared with the SD group, supplementation of selenium yeast had no signi cant effect on the α-diversity of the bacteria community including richness by Chao1 estimation and diversity re ected by the Shannon index ( Figure S2b). PLS-DA analysis showed there were signi cant differences in the composition of the microbiota (Anosim, P = 0.002) after selenium yeast supplementation (Fig. 4b) compared to the SD group, and especially between the latter and SY0.  (Table  S8). The relative abundances of the intestinal bacteria at the genus level were further analyzed. The abundances of several bacteria such as Veillonella (P 0.05) and Campylobacter (P 0.05) increased in selenium yeast supplementation groups, and selenium yeast supplementation markedly reduced the abundances of both Stenotrophomonas (P 0.01) and Faecalicoccus (P 0.05) (Fig. 4c).
Additionally, we examined the effect of inorganic selenium supplementation on composition of the ileum microbiota by 16S rRNA-sEq. The distribution of clean tags length was mainly 420-440 bp ( Figure S2d). 155 and 152 OTUs were generated in the SD and SS0.45group, respectively, and 33 and 30 OTUs were both unique in two groups, while 122 OTUs were shared (Fig. 4d). There were no signi cant differences in the richness and diversity of gut microbiota between the SD group and SS0.45 group ( Figure S2e), but microbial community was considerably different (Anosim, P = 0.049, R = 0.2216) (Fig. 4e). Firmicutes was dominated in two groups ( Figure S2f). Compared with the SD group, The abundances of Lactobacillus (P 0.01) and Stenotrophomonas (P 0.05) were noticeably reduced by sodium selenite supplementation, while Romboutsia (P 0.01), Turicibacter (P 0.05), Veillonella (P 0.05), and Corynebacterium_1 (P 0.05) were dramatically increased (Fig. 4f). Sodium selenite supplementation also suppressed Actinobacteria phylum and increased Proteobacteria and Cyanobacteria phylum (not signi cant) ( Table S8).
The absorption e ciency of sodium selenite and selenium yeast in the intestine are different. Thus, the effects on intestine microbiota may also be different. As shown in Fig. S2g, the distribution of clean tags length was 420-440 bp. There were 131 shared OTUs between SS0.45 and SY0.45group, while 23 and 36 unique OTUs were gained, respectively (Fig. 4g). Relatively to the SS0.45 group, in the SY0.45 group there was no signi cant effect on the Chao1, Good coverage, observed species, and PD index of the bacteria community. However, the Shannon and Simpson index were signi cantly higher in the SS0.45 group ( Figure S2h). There were statistical differences in the compositions of gut microbiota between the two groups (Anosim, P = 0.001, R = 0.5296) analyzed by PLS-DA (Fig. 4h). The level of the phylum Cyanobacteria (P 0.05) has signi cantly decreased in the SY0.45 group (Table S8). The relative abundances of the intestinal bacteria at the genus level between two groups were further analyzed. The relative abundances of Romboutsia (P 0.01) and Methylobacterium (P 0.05) were markedly reduced in the SY0.45 group, while Lactobacillus (P 0.01) and Escherichia-Shigella (P 0.05) were dramatically elevated (Fig. 4i). These results suggested that, to some extent, different forms and doses of Se may have distinctive effects on the composition of the respective gut microbiota.
To further understand the function of the altered gut microbiota by Se supplementation in aged laying hens, we carried out PICRUSt analysis based on the microbiota composition. Some enriched pathways, including nitrogen metabolism, MAPK signaling pathway-yeastas well as translation proteins of gut microbiota, were activated by both selenium yeast and sodium selenite supplementation, while the Pentose phosphate pathway was suppressed, compared to the SD group (Fig. 4j, k and Fig. S3). In addition, selenium yeast supplementation may restrain Steroid biosynthesis and metabolism of bothterpenoids and polyketides (Carotenoid biosynthesis), while sodium selenite supplementation may signi cantly suppress the carbohydrate metabolism (including Glycolysis / Gluconeogenesis, Citrate cycle and Carbohydrate digestion and absorption), lipid metabolism (including Biosynthesis of unsaturated fatty acids, Glycerophospholipid metabolism and Glycerolipid metabolism) as well as metabolism of other amino acids (Selenocompound metabolism and Glutathione metabolism), it also may increase energy metabolism (Oxidative phosphorylation), amino acid metabolism (Cysteine and methionine metabolism) and bacterial movement (Bacterial motility proteins and Bacterial chemotaxis) (Fig. 4j, k and Fig. S3). Moreover, the differences of effect between selenium yeast and sodium selenite were majorly enriched in carbohydrate metabolism (Carbohydrate digestion and absorption, Inositol phosphate metabolism, and Starch and sucrose metabolism), lipid metabolism (Glycerophospholipid metabolism, Glycerolipid metabolism, Sphingolipid metabolism, Biosynthesis of unsaturated fatty acids, and Fatty acid biosynthesis), nucleotide metabolism (Purine metabolism), amino acid metabolism (Glycine, serine and threonine metabolism), other amino acids metabolism (Glutathione metabolism and Selenocompound metabolism), cell cycle (Apoptosis, Cell cycle-Caulobacter, and DNA replication) as well as other biological processes (ABC transporters, Phosphatidylinositol signaling system, and Peroxisome)( Fig. 4l and Fig. S4). These results indicated that Se supplementation may suppress the carbohydrate metabolism (Pentose phosphate pathway) of gut microbiota and increase energy metabolism (Nitrogen metabolism), and that the different forms of Se have distinctive effects on gut microbiota in carbohydrate metabolism, lipid metabolism, energy metabolism, amino acid metabolism, Glutathione metabolism and Selenocompound metabolism.

Correlation of transcriptome and microbiota reveals the effects of selenium yeast supplementation on laying rate and Se deposition in the ileum
To investigate the effect of the interaction between transcriptome and microbiota on the laying rate and ileum Se content of aged laying hens after selenium yeast supplementation, we evaluated the relationship among the ileal transcriptomes, the laying rate, the Se content of ileum and the microbiota. Pearson correlation tests showed there were signi cant associations between various microbial species and Se content, as well as between some microbial species and the laying rate. The Eubacterium_hallii_group, Mesorhizobium, and Anaerobacillus were positively correlated with Se content in the ileum, while Brevibacterium, Aquabacterium, Faecalicoccus, and Stenotrophomonas were negatively correlated with Se content (Fig. 5a). Moreover, there was a negative correlation between the laying rate and the abundance of Romboutsia, Intestinibacter, Turicibacter, Fusobacyerium, Aeromonas, and Alteromonas, respectively, whilea positive correlation was found between the laying rate and the abundance of Lactobacillus, Anaerovorax, Maritalea, Alteromonas and Geobacter (Fig. 5b). These data indicated that the changes of certain species of microbial in ileum were actually associated to selenium yeast intake and egg production in the aged laying hens.
To further assess the relationship between microbiota and transcriptome, Pearson correlation analysis was performed to investigate the correlations between various pathways form GSEA of four clusters and the abundance of microbial species. The abundance of Mesorhizobium and Bacillus was highly correlated with several pathways (regulation of acute in ammatory response, metallopeptidase activity and regulation of protein oligomerization, etc.) in cluster 3. Meanwhile, interferon-gamma response pathway of cluster 8 was highly associated with the abundance of various microbiota, such as Aliivibrio, Anaerobacillus, Shewanella, Moritella, Psychrilyobacter, Rosei exus, Anaerovorax, Mycobacterium. The abundance of Paucibacter was hugely correlated with the pathways in cluster 9. The abundance of Methylobacterium, Aeriscardovia, Rhodoplanes and Aerococcus was greatly associated with the pathways of cluster 10. Additionally, the abundance of Psychrobacter was extremely correlated with peptidyl lysine trimethylation of cluster 8 and cellular protein complex disassembly of cluster 10, respectively (Fig. 5c).
To elucidate the relationship between bacterial abundance and switch-on or switch-off genes, the Pearson correlation coe cient was calculated (Fig. 5d). The abundance of Stenotrophomonas, Comamonas, Actinomyces, Acinetobacter, Methylobacterium and Psychrobacter was negatively correlated with the expression of ON genes. However, the abundance of Campylobacter, Ruminococcaceae_UCG.014, RB41 and Eubacterium_hallii_group was positively correlated with the expression of ON genes. Furthermore, the abundance of Acinetobacter, Psychrobacter, Delftia, Paucibacter, Stenotrophomonas, Faecalicoccus, Comamonas, Actinomyces and Methylobacterium had positive correlations with the expression of OFF genes. Alternatively, the abundance of Massilia, RB41, Campylobacter, and Eubacterium_hallii_group had negative correlations with the expression of OFF genes. In brief, selenium yeast supplementation directly in uenced the biosynthesis of the selenoprotein. To gain further insight into the relationship between selenoprotein expression and microbiota, we conducted the Pearson correlation coe cient analysis (P 0.01; R 0.6) (Fig. 5e). SELENOP2 had positive correlations with almost the whole microbiota, but SELENO had negative correlations with the abundance of Mesorhizobium, Blautia, Campylobacte. The expression of GPX3, MSRB3, GPX8, and SELENOW was positively associated with the abundance of Ruminococcaceae_UCG.013. On the other hand, the Campylobacter was negatively correlated with TXNRD3 gene. These ndings indicated that, to some extent, there are interactions between the ON/OFF genes as well as the selenoprotein genes and the speci ed microbiota, respectively.
In order to provide an initial visualization of the relationships among Se content in the ileum, laying rate, transcriptome pathways of interest and microbiota, we generated a biplot using the R package vegan ( Fig. 5f). CCA analysis revealed that the content of Se in the ileum was highly positively correlated with the positive regulation of lymphocyte activation (cluster 9), regulation of digestive system progress (cluster 8) as well as the abundance of Anaerobacillus, Alteromonas, and Loktanella. In contrast, aging (cluster 8), fatty acid metabolic process (cluster 3), and the abundance of Streptococcus, Brevibacterium, Aerococcus, and Devosia were inversely correlated with Se content. Moreover, the laying rate had a positive correlation with positive regulation of the lymphocyte activation (cluster 9), the response to carbohydrate (cluster 9), and the abundance of Mesorhizobium and Paracoccus, while having a negative relationship with fatty acid metabolic progress (cluster 3), the abundance of Streptococcus, Devosia, Aerococcus, and Intestinibacter.

Discussion
Selenium plays bene cial roles in laying production and health of hens, especially when it comes to antioxidants, immune function, catalytic function [37], and modulation of gut microbiota [24]. Interestingly, many studies indicated that gut health and the intestinal microbiota status are crucial for the productivity of hens. Gut microbiota can in uence the genomic stability of host cells by regulating various signaling pathways to maintain gut health [29]. However, the functions of microbiota are governed via the interactions between the host and diet (macro and micronutrients) [38]. The certain metabolites and the outcomes of the diet are essential co-factors for epigenetic enzymes impinging upon the epigenetic regulation of gene expression [39]. However, the speci c link among gut microbiota, host gene expression, Se state and productivity has not been established, and likely, multiple mechanisms may be involved in the complex interactions among microbiome, diet and host.
This study evaluated the effect on laying rate of different concentration of dietary selenium yeast in aged laying hens. Generally, the peak production of commercial laying hens usually lasts 60 weeks of age, and the laying rate and egg quality decline with the age increasing [40,41]. Se is a crucial microelement for poultry nutrition and plays a essential role in the production [42]. In this study, 0.30 mg/kg selenium yeast supplementation for 12 weeks was able to signi cantly alleviate the decline of laying rate. This is consistent with the results of previous studies showing that selenium yeast can improve the egg production in aged broiler breeder hens [43]. Therefore, supplementing selenium yeast to improve the laying production in aged laying hens may depend on the species of poultry.
In this study, transcriptome analysis was performed to detect the effect of different level of selenium yeast supplementation on ileal gene expression patterns in aged laying hens. The transcriptome results, with a notable difference in transcript pro les between SY and SD group, revealed that selenium yeast supplementation in ileum tissue changed the expression of genes associated with Glycerophospholipid and Glycerolipid metabolism. This result is consistent with the previous studies that showed DEGs were mainly enriched in Glycerolipid metabolism after 200 µM Se treatment in silkworm [44], suggested that Se supplementation has a potential effect on Glycerolipid metabolism in livestock and poultry. The links between Se deposition and genes expression analyzed by WGCNA revealed that six modules had signi cant correlations with Se content, while the enriched biological processes and pathways were not only associated with fatty acid metabolism, but also organ or tissue speci c immune response as well as carbohydrate transmembrane transporter activity. Coherently with previous studies, selenium supplementation tends to increase the gene transcripts behind the regulation of fatty acid metabolism pathways in mice [45]. Moreover, fatty acid composition of egg and muscle in poultry may be in uenced by dietary Se [46,47], demonstrating that Se can modulate the metabolic process by altering the fatty acid metabolism. In addition, the hub genes closely related to Se deposition were identi ed, namely TXNRD1, DLD, ILK and LZTS2. TXNRD1 is one of the TXNRDs system, a key antioxidant system modulating redox signaling and maintaining intestinal health [48,49], involved in Selenocompound metabolism pathway, and also associated with the cellular defense against oxidative stress [50]. DLD is a key enzyme involved in energy metabolism [51], Citrate cycle/TCA cycle and Pyruvate metabolism pathway, and also providing the intermediates for the metabolism of amino acids, nucleic acids, carbohydrates and lipids. ILK has a key role in the regulation of cell migration which is thought to be associated with the canonical Wnt signaling pathway [52]. Knockdown ILK in human intestinal cells severely inhibited itsspreading and migration [53,54]. Therefore, we speculate that the selenium yeast may affect various cellular behaviors in aged laying hens such as cell migration, key enzyme production and defense against oxidative stress, thereby affecting the function of fatty acid metabolism, intestinal speci c immune response as well as carbohydrate transmembrane transporter activity. It also provided clues for identifying novel molecular markers related to selenium yeast supplementation in aged laying hens.
To investigate the dose effect of selenium yeast on aged laying hens, we evaluated the dynamic changes of selenoprotein genes expression in the ileum after selenium yeast supplementation. In different doses of the latter, the expression patterns of several selenoprotein genes, including GPX4, MSRB1, SELENOF, SELENOT, SELENOK, SELENOS, were similar. It was proved that GPX4 plays an important role in oxidative phosphorylation as well as mitochondrial dysfunction pathways to protect mitochondria from oxidative damage [55]. MSRB1 [56], SELENOK [57] and SELENOS [58] are involved in in ammatory responses and intestinal health. the expression of DIO2, GPX1, GPX3, GPX7, GPX8, MSRB3, SELENOP2, SELENOM, and SELENOW increased accordingly to the rise of selenium dose. GPX1 and GPX3 are the major components in the Glutathione peroxidase family, they are responsible reduce hydrogen peroxide, organic hydroperoxides and/or phospholipid hydroperoxides with a vital role in the amelioration of peroxidemediated deleterious effects [33]. Moreover, SELENOW and SELENOT have antioxidant functions [59], meanwhile, SELENOM had the capacity of encoding oxidoreductase [60]. These selenoproteins were regarded as direct or indirect potential regulators of oxidative/redox balance [61] modulating intestinal health. Indeed, most of these selenoproteins (GPX1, GPX3, GPX4, SELENOM, SELENOU) were responsive to alterations of Se status. Thus, we speculated that selenium yeast supplementation can affect the health and functions of intestines by regulating selenoprotein expression [61]. Other two selenoprotein, namely TXNRD2 and TXNRD3, can maintain proliferation and differentiation processes by regulating Wnt pathway [62]. Consistently with other studies, the expressions of TXNRD2 and TXNRD3 were the highest in the selenium de ciency group [62]. In summary, the selenium yeast can regulate the expressions of selenoprotein genes to affect the health and function of intestines through regulating oxidative/redox balance, maintaining proliferation and differentiation processes as well as metabolism balance.
In addition, the bene ts of different chemical forms of Se, such as inorganic sodium selenite (widely used as dietary supplement) and organic selenium yeast, are controversial. Previous studies showed that Sodium selenite and selenium yeast had different effects on overall gene expression pattern in the small intestine of mice [64]. We also proved that there were differences between selenium yeast and sodium selenite supplementation in several metabolic pathways, including fatty acid metabolic process, cellular lipid metabolic process, oxidoreductase activity, Glycerolipid metabolism, Glycerophospholipid metabolism, PPAR signaling pathway as well as Arachidonic acid metabolism and mTOR signaling pathway. Previous studies have demonstrated that Se status impacted two transcription factors, namely nuclear factor-κB (NF-κB) and peroxisome proliferator-activated receptor (PPAR)γ, which are involved in the activation of immune cells [57]. We suggest that the deposition effect of different Se forms may in uence the Se status, which will have a profound impact on the PPARγ signaling pathway in activating the immune cells.
Se supplementation affects the intestinal barrier by modulating gut microbiota [26]. In this study, we investigated the effect of selenium yeast supplementation on microbiota abundance in aged laying hens.
In general, different microbiotas were observed at the genus level, including Veillonella, Stenotrophomonas and Faecalicoccu s. A number trials on both animals and humans have con rmed that Veillonella is a producer of short-chain fatty acid acetate and propionate [65]. The result indicated that increasing the abundance of this family may bene t poultry health by increasing the fatty acids in the chicken's intestine.Stenotrophomonas is considered as an opportunistic pathogen [66] and can consequently be a potential risk for animal health. Faecalicoccus can produce not only butyric acids, but also lactic and formic acids as major metabolic end products that play important roles in gut physiological activity [67]. Selenium yeast supplementation could signi cantly increase the abundance of Veillonella, while decreasing the abundance of Stenotrophomonas and Faecalicoccus, which may bene t chicken's intestinal health. Furthermore, PICRUSt analysis demonstrated that the altered microbiota was enriched in energy metabolism (nitrogen metabolism) and translation proteins with selenium yeast supplementation. Selenium yeast can interfere with the diversity of gut microbiota and reduce the abundance of harmful bacteria caused by Ochratoxin-A [28]. The increase in the abundance of Veillonella may produce more propionate and maintain energy homeostasis [68][69][70]. Thus, selenium yeast may enhance the energy metabolism of gut microbiota by increasing the abundance of Veillonella.
The protective effect of selenium on the intestine may be chemical form dependent, although the results were not consistent [22]. We found that sodium selenite supplementation of 0.45 mg/kg elevated substantially the abundance of Turicibacter compared to the SD group, which is consistent with the results suggesting that that the number of Bi dobacterium, Turicibacter and Akkermansia has increased in rodents fed with Se-supplemented diets [26,71]. Meanwhile, Turicibacter had a potential antiin ammatory activity in the gut, and high levels of these bacteria were also observed in mutant mice that were resistant to colitis [72,73]. Thus, inorganic selenium may change the abundance of bene cial bacteria in order to affect gut immune responses. Different sources of Selenium may have different effects on individual health and performance via modulating the intestinal microbiota. Differences of genera microbiota between the selenium yeast supplementation and sodium selenite supplementation were mainly in Lactobacillus, Escherichia-Shigella, Romboutsia, and Methylobacterium. The increased abundance of the Lactobacillus demonstrated that selenium yeast supplementation in aged laying hens may contribute in improving gut health compared with sodium selenite supplementation. another study had proved 0.9 mg/kg that selenium nanoparticles could improve gut health in poultry by increasing the abundance of bene cial bacteria, such as Lactobacillus and Facealibacterium. Based on PICRUSt analysis, the greatest difference of gut microbiota functions in different forms of Se supplementation were enriched in both lipid metabolism (Glycerolipid and Glycerophospholipid metabolism) and carbohydrate metabolism. thus, demonstrating that there is a correlation between the abundance of Lactobacillus and lipid and carbohydrate metabolism [70]. Interestingly, the common pathways obtained by PICRUSt and the analysis of the transcriptome include Glycerolipid metabolism, Glycerophospholipid metabolism, ABC transporters, as well as MAPK signaling pathway, suggesting that the differential effects between selenium yeast and sodium selenite on eukaryotic and prokaryotic organisms are conservative. these results indicated that Se supplementation can modulate the composition of gut microbiota in aged laying hens by not only increasing the abundance of some bene cial bacteria such as Veillonella, Turicibacter and Lactobacillus, but also by decreasing the abundance of the opportunistic pathogen Stenotrophomonas.
Interestingly, recent studies have demonstrated that dietary Se affects both the composition of the intestinal microbiota and the colonization of the gastrointestinal tract, which in its turn in uences the host Se status and selenoproteome expression [24]. However, little informations are available for de ning the interactions among laying production, Se status, host gene expression and microbiota. Our study investigated the interaction of gut microbiota, phenotype (laying rate and Se content) and ileal transcriptome. It has been estimated that there is a signi cant positive correlation between the abundance of Lactobacillus and the laying rate, which is consistent with other results. For instance, the abundance of Lactobacillus, Bi dobacterium, Acinetobacter, Flavobacteriaceae, Lachnoclostridum and Rhodococcus was higher in the High egg-laying performance group in comparison with Low egg-laying performance group [9]. A 0.6% Lactobacillus supplement in the diet increased the egg production of feeding laying hens [9]. Therefore, the laying rate was signi cantly higher in the SY0.30 group than in the SS0.45 group, which may be due to the abundance Lactobacillus in the ileum.
Detailed informations indicated that gut microbiota is actually related to plasma levels of Se in mice [29]. In this study, Eubacterium_hallii_group had a positive relationship with Se content in the ileum, and it was a member of the butyrate-producers [75][76][77] allowing it to metabolizes glycerol into reuterin [75][76][77]. Stenotrophomonas, an opportunistic pathogen, had a negative relationship with Se content in ileum. These two results suggested that Se status in ileum of aged laying hens may modulate the balance of the bene cial bacterium and pathogens in order to rise egg laying performance. Moreover, we identi ed several microbiotas including Maritalea, Alteromonas, Geobacter, Mesorhizobium, Anaerobacillus, Anaerovorax, Kribbella, which were all positively correlated with both the Se content and the laying rate, making them potential novel markers related to egg production affected by selenium yeast. Dietary Se supplementation may change the expression of selenoprotein genes through modulating the Se status in the ileum to in uence microbiota. Our results showed that SELENOP2, as a biomarker to assess the Se status, had positive correlations with some microbiota such as Butyrivibrio_2, Ferrimonas and Faecalibacterium. While Campylobacter, a zoonotic foodborne pathogen causing acute gastroenteritis, was negatively correlated with TXNRD3. TXNRD3 is a member of Txnrds family that plays major roles in both antioxidant defense and redox regulation. Our further results proved that SELENOP may be used as microbiota biomarkers,andthe expression of TXNRD3 can re ect the abundance of Campylobacter Dietary Se regulated gene expression by effecting the Se status in ileum of laying hens. Many genes were switched on or off by Se supplementation. NCAN and AMBP, as "ON genes", and TECTB, as "OFF genes", were identi ed in all three doses of selenium yeast supplementation. Moreover, NCAN and AMBP were negatively associated with the abundance of Psychrobacter in the 0.15 mg/kg selenium yeast supplementation as well as the abundance of Psychrobacter, Stenotrophomonas and Faecalicoccus in the 0.30 mg/kg selenium yeast supplementation. Psychrobacter was associated with the spoilage of meat products [80], while Stenotrophomonas was reported as an opportunistic pathogen [66]. These data suggested that selenium yeast supplementation may switch on both NCAN and AMBP genes in order to reduce the abundance of harmful bacteria including Psychrobacter and Stenotrophomonas. In addition, PSCA was identi ed as "ON gene" in 0.30 mg/kg Se supplementation, possibly playing a special role in follicle selection which was an important process affecting laying performance [81] [82]. In the present study, the results showed that PSCA had a positive relationship with the abundance of propionate producer Veillonella while a negative relationship with the abundance of opportunistic pathogens Stenotrophomonas was obtained. Thus, 0.30 mg/kg selenium yeast supplementation may alleviate the decline of laying rate through switching on PSCA to alter the abundance of both bene cial and harmful bacteria.
Correlational analyses between the pathways and the abundance of microbiota showed that the pathways in cluster 3 were signi cantly related with Bacillus, and Bacillus. Bacillus as probiotics which could enhance gut health [83] and produce a variety of enzymes such as protease, amylase and lipase [84,85]. Meanwhile, IFN gamma response in cluster 8 was signi cantly correlated with most of the microbiota, highlighting that selenium yeast may modulate the interactions of immune response and microbiota in the ileum. Interestingly, four microbiotas including Mesorhizobium, Ruminococcaceae_UCG.014, Eubacterium_hallii_group, and Psychrobacter were remarkably associated with the pathways in four clusters. Ruminococcaceae UCG-014, a common genus, was associated with the maintenance of gut health and had the enzymatic ability to degrade cellulose and hemicellulose [86,87]. These results indicated that selenium yeast supplementation can affect the interaction of the transcriptome pathways and special gut microbiotas in order to improve gut health.
CCA analysis results showed that Se content was positively correlated with the positive regulation of lymphocyte activation (cluster 9) and the regulation of digestive system progress (cluster 8). It was also associated with the abundance of Anaerobacillus, Alteromonas, and Loktanella. Anaerobacillus grown under alkaliphilic or halophilic conditions through fermentative or anaerobic respiration [88], it might also take part in the regulation of digestive system progress and intestinal environment. It has been reported that Loktanella was enriched in diseased tissues, suggesting it could be a candidate opportunistic pathogen [89] involved in immune status regulation. These results suggested that selenium yeast supplementation might balance the number of special bacteria by activating immune response and digestive progress in order to improve the ileum environment. In addition, the laying rate was also positively correlated with the positive regulation of lymphocyte activation and the response to carbohydrate in cluster 9, while these two pathways were positively related to Mesorhizobium and Paracoccus, suggesting that selenium yeast elevated the egg production of aged laying hens by in uencing immune reactions, carbohydrate metabolism and speci c microbiota. Meantime, both the Se content and the laying rate were negatively correlated to fatty acid metabolic progress in cluster 3. More ndings demonstrated that the microbiota was negatively correlated with Streptococcus, Intestinibacter, Devosia and Aerococcus which were involved in fatty acid metabolic progress. Streptococcus was reported to produce conjugated linoleic acid from linoleic acid [89], which was related to the fatty acid metabolic process. Interestingly, Devosia may serve as a promising biomarker for the early detection of colorectal cancer (CRC) [91]. The different dietary ber sources altered the microbial community such as Intestinibacter in a pigs [92], the latter was found to be associated with type 2 diabetes disease in human [93]. Taken together, these results suggested that selenium yeast supplementation improve the Se content and laying rate in aged laying hens by activating immune response pathways, fatty acid and carbohydrate metabolic progress, as well as modulating digestion and the speci c microbial community.

Conclusion
Our results revealed that selenium yeast supplementation can slow down the deterioration in egg production in aged laying hens. In addition, the transcriptome functional enrichment analysis showed that selenium yeast supplementation may affect the metabolic processes (Glycerolipid and Glycerophospholipid metabolism), the immune response as well as the intestinal absorption in aged laying hens. Furthermore, 16S rRNA analysis suggestedd that Se supplementation can modulate the composition of gut microbiota in aged laying hens by increasing the abundance of speci c bene cial bacteria including Veillonella, Turicibacter, Lactobacillus, while decreasing the abundance of pathogenic bacteria Stenotrophomonas. Further, integrated analysis indicated that several microbiotas (Maritalea, Alteromonas, Geobacter, Mesorhizobium, etc.) were positively associated with both Se content and laying rate. Furthermore,there was a marked correlation between some speci c microbiotas (Aliivibrio, Anaerobacillus, Shewanella, Moritella, etc.) and the immune response pathways (regulation of acute in ammatory response, positive regulation of lymphocyte activation and IFN gamma response).
Meanwhile, the identi ed switched on gene PSCA had a positive relationship with the Veillonella, a propionate producer, and a negative relationship with the opportunistic pathogen Stenotrophomonas. CCA analysis revealed that both the Se content in the ileum and the laying rate were highly positively correlated with Anaerobacillus, Alteromonas, Loktanella and the positive regulation of lymphocyte activation, but were negatively correlated with Streptococcus, Devosia, Aerococcus, Intestinibacter and fatty acid metabolic progress. Taken together, selenium yeast supplementation can affect immune response, metabolic processes, and speci c microbiota composition to ameliorate the deterioration in egg production in aged laying hens.

Animals, experimental design, diets, and husbandry
Body Se of Three hundred and seventy-ve (375) aged laying hens (Jing Hong) were consumed for 6 weeks (from 76 to 82 weeks of age) in order to gain the Se-de ciency hens. After Se-consumption Jing Hong laying hens with similar laying rate were randomly allocated to 5 treatment groups with 5 replicates (15 chickens per replicate and 3 chickens per cage) and were supplemented with different doses  Table S9), which had an average basal Se content of 0.056 mg/kg. In the supplementation period, one group was fed the basal diet only (SD), the remaining four groups were supplemented with 0.15, 0.30, 0.45 mg/kg selenium yeast (SY), and 0.45 mg/kg sodium selenite (SS). The Se contents of different diets are shown in Table S10. Batches of the experimental diets were produced every 4 weeks to prevent the feed from mildewing. The hens were housed in an environmentally controlled room maintained at 25℃ and had a daily lighting schedule of 16 h (from 5 am to 9 pm) and 8 h of darkness (from 9 pm to 5 am).

Data and sample collection
Blood samples (8 mL per laying hen) were taken from the main wing vein and collected into an anticoagulant tube every 2 weeks during the Se depletion period. Plasma was separated by centrifugation at 4℃, 3000 rpm for 10 min and stored at -30℃ for further analysis. After Se supplementation for 12 weeks, 10 randomly chosen laying hens from each dietary treatment were slaughtered, the ileum and its chyme were sampled and frozen in liquid nitrogen immediately. All samples were stored at -80℃ prior to analysis.

Production performance
The laying rate of aged laying hens was measured from 83 to 95 weeks of age. Daily egg production was determined per replicate unit at 2:30 pm. Monthly laying rate was measured by daily egg production in four weeks.

Determination of Se content
To determine the Se content in feed, plasma, and ileum. All samples (0.5 g-1 g for feed and ileum, and 0.5 ml-1 ml for plasma) were digested in a mixture of HNO 3 and HClO 4 (2:1) for about 2 hours and measured by uorescence method using Hitachi 850 uorescence spectrophotometer (Tokyo, Japan) [94].
Measurement of β-galactosidase assay and βgalactosidase staining

Transcriptome Bioinformatic and statistical analysis
After the quality raw RNA-seq data was controlled, we obtained clean data for further analysis. Read count matrices were obtained using FeatureCounts package, the data was aligned to reference chicken genome (GRCg6a) downloaded from Ensembl (https://asia.ensembl.org/index.html)and was performed using HISAT2. DEGs were accessed using R package DESeq2 [95]. Unless otherwise speci ed, the signi cance thresholds were FDR < 0.05 and |log2foldchange| > 1.0, while heatmaps were drew using R package ComplexHeatmap [96]. Functional enrichment analysis was performed to identify signi cant biological activities and required genes involved on PANTHER [97] (http://pantherdb.org/) and KOBAS [98] (http://kobas.cbi.pku.edu.cn/kobas3/genelist/) website, for GO and KEGG analysis respectively. The signi cance threshold P value was no more than 0.05 and DESeq2 normalized results were used as background gene in GO analysis. To nd more meaningful results, pathways considered only when gene number in reference list was between 30 and 500 for GO analysis or between 10 and 500 for KEGG analysis. The Top 80 signi cant pathways (P < 0.05) were selected for further visualization and was performed with R package ggplot2 [99].
The expression data of ileum tissues in the SD, SY0.15, SY0.30, and SY0.45 groups were subjected to weighted gene co-expression network construction using the WGCNA [100,101] package. First, Pearson's correlations were calculated between each gene pair and their matrices were performed for all pairwise genes. Then we constructed a weighted adjacency matrix by raising the co-expression similarity to a power of β = 4. Next, the adjacency was transformed into a topological overlap matrix (TOM), which was used as an input for hierarchical clustering analysis. Finally, average linkage hierarchical clustering was conducted on the basis of a TOM-based dissimilarity measurement and gene modules were identi ed using a dynamic tree-cutting algorithm with a minimum module size of 80 genes. The co-expression modules were automatically color-coded by WGCNA and their structure was visualized by heatmap plots of topological overlaps. In addition, the principal component analysis was completed and the MEs were generated from the rst principal component. The relationships among modules were summarized by a hierarchical clustering dendrogram of the eigengenes and by a heatmap plot of the corresponding to its network. To identify modules that were signi cantly associated with the Se content in ileum, we carried out Pearson correlation between the MEs and Se content in ileum (P < 0.05 and |R| > 0.4). Then, in order to explore the potential mechanism by which module genes impact the correlated Se content, GO and KEGG enrichment analysis were also performed on PANTHER and KOBAS websites. Multiple tests were used to provide corrected P-values. We identi ed the hub genes with high intramodular connectivity in modules through the gene signi cance (GS) and the module membership (MM) to explore more signi cant genes associated with Se deposition. Genes with a GS > 0.7 and MM > 0.9 were considered to be candidate intramodular hub genes. Ultimately, the real hub genes were top 30 in each module chosen by soft Connectivity function. Gene-gene interaction network was constructed and visualized by Cytoscape software, and DEGs in any selenium yeast supplementation groups were marked as V if founded in modules.
In order to investigate the dynamic changes in gene expression supplemented by different doses of selenium yeast, Short Time-series Expression Miner [102] (STEM) was performed. DEGs among selenium yeast supplementation groups were strati ed along with supplementation gradients and log transformation. We expected to capture nicety tendencies as well as a high consistency withinclusters. Therefore, Maximum pro les were set relatively high at 100 and any cluster required intra-pro les correlation that were higher than 0.88. Then we performed Gene set enrichment analysis (GSEA) to get more functional information in the clusters of interest. Single sample GSEA algorithm, ssgsea, was used combined with other results for further analysis and was implemented in R package GSVA [103]. A matrix of enrichment scores for each gene set and sample was obtained, with DESeq2 being normalized results as input data. When it comes to visualization, the only results with standard deviation among samples were no more than 0.1.
Next, the switched on or off genes (ON and OFF genes) regulated by selenium yeast were screened. First, genes whose variance was more than 1.5-fold of quartile deviation were ltered out. Then, if the median of a gene's expression within a given group equals zero, the genes were de ned as OFF-genes, making ON-genes were an opposite state. Meanwhile, standard deviation of genes was no more than 1.0 counts and the mean of gene expression and each gene should show a statistical signi cance (P < 0.05) through Wilcoxon signed-rank test. Two states of control group were respectively intersected with the opposite states of treatments. Then genes that switched between ON and OFF states were derived. Then PPI was performed on STRING [104] (https://string-db.org/) website to exhibit the relationship between ON and OFF genes. Only the connected nodes with con dence more than 0.4 were displaced.
Moreover, we investigated the changes of redox-related and aging-related genes to get more insights about Se yeast functions, the gene sets were retrieved on the Molecular Signatures Database [105] website (https://www.gsea-msigdb.org/gsea/msigdb/genesets.jsp) (Table S7).
16S rRNA microbial community analysis using Illumina Analysis Pipeline Version 2.6. The raw data were rst screened and sequences shorter than 230 bp, had a low quality score (≤ 20), contained ambiguous bases or did not exactly match to primer sequences and barcode tags, were removed from consideration. Quali ed reads were separated using the sample-speci c barcode sequences and trimmed with Illumina Analysis Pipeline Version 2.6. And then the dataset was analyzed using QIIME. All sequences were used for the comparison of relative abundance of bacterial taxa and were clustered into operational taxonomic units (OTUs) according to a 97% similarity, in order to generate rarefaction curves and to calculate the richness and the diversity indices. The Ribosomal Database Project (RDP) Classi er tool was used to classify all sequences into different taxonomic groups. Finally, the relative abundance of each bacterial taxa was analyzed by QIIME and predicted functional genes were analyzed by PICRUSt based on the KEGG pathway [106].

Host transcriptome-microbiota correlation analysis
The associations between the host transcriptome and the ileum bacteria were further explored using the Se content in ileum, the laying rate, the interesting clusters (cluster3, 8, 9, and 10), the ON and OFF genes, the selenoprotein genes, and the relative abundance of the identi ed ileum bacterial genera. The associations between OTUs and phenotypes (laying rate and Se content in ileum) were explored using built-in function cor of R package. Correlation between GSVA output and OTUs was performed using the dmic for similarity measurement implemented in Hierarchical All-against All (HAllA) software developed for multiomic data sets, which refers to this study [107]. Signi cant correlated pairs were retained with default parameters. Further visualization was performed using R package ComplexHeatmap. The associations between ON/OFF and selenoprotein genes and ileum bacteria OTUs were explored by Pearson correlation, R package Hmisc was used for calculating the Pearson correlation and the asymptotic P-values between two high-dimensional data sets, and the signi cant standards were set as |R| > 0.6, P < 0.05. Further visualization was performed using R packages ComplexHeatmap as well as ggplot2. Lastly, CCA was perform using R package vegan. Along with two phenotype traits, the Se content in ileum and laying rate, GSEA results for signi cant clusters in STEM were used as explanatory variables with the collinearity removed.

Statistical analysis
Results were presented as mean ± standard error of the mean (SEM), and differences between two groups were analyzed by an unpaired Student's t-test, data that involved more than two groups was analyzed with one-way ANOVA, along with Duncan test (SPSS for Windows, version 25; IBM). Statistical differences were considered signi cant at P 0.05. Phenotypic data presentation was carried out using GraphPad Prism (version 7.0, GraphPad Software Inc, San Diego, CA, USA). Every experiment was repeated at least three times. aging associated genes. Table S8. The abundance of bacteria in different comparisons. Table S9. The composition and nutrient of the basal diet.