Gut microbiota and transcriptome analysis reveals a genetic component to dropping moisture in chickens

Abstract

High dropping moisture (DM) in poultry production has deleterious effects on the environment, feeding cost, and public health of people and animals. To explore the contributing genetic components, we classified DM of 67-wk-old Rhode Island Red (RIR) hens at 4 different levels and evaluated the underlying genetic heritability. We found the heritability of DM to be 0.219, indicating a moderately heritable trait. We then selected chickens with the highest and lowest DM levels. Using transcriptome, we only detected 12 differentially expressed genes (DEGs) between these 2 groups from the spleen, and 1,507 DEGs from intestinal tissues (jejunum and cecum). The low number of DEGs observed in the spleen suggests that differing moisture levels are not attributed to pathogenic infection. Fourteen of the intestinal high expressed genes are associated with water-salt metabolism (WSM). We also investigated the gut microbial composition by 16S rRNA gene amplicon sequencing. Six different microbial operational taxonomic units (OTUs) (Cetobacterium, Sterolibacterium, Elusimicrobium, Roseburia, Faecalicoccus, and Megamonas) between the 2 groups from jejunum and cecum are potentially biomarkers related to DM levels. Our results identify a genetic component to chicken DM, and can guide breeding strategies.

INTRODUCTION

Dropping moisture (DM) refers to the water content in chicken feces (Kavous and Mccormick, 1991; Ouachem and Kaboul, 2012), and it is influenced by several factors, such as genetics, nutrition, environment, and disease. A high DM, particularly in the summer, can negatively affect egg production, manure management, pathogen control, the immediate farm environment, and the global environment. At present, the comprehensive utilization rate of excrement from livestock in China is less than 60%. Unprocessed excrement (containing ${\text{SO}}_4^{2-}$, ${\text{NO}}_3^{-}$, and ${\text{NH}}_4^{+}$) can substantially pollute the atmosphere and water resources, resulting in particulate matter emissions (Cambra-López et al., 2010), water body eutrophication, and decreased crop production (Gu et al., 2008). Moreover, pollution from unprocessed excrement can be detrimental to the health of livestock and humans (Avery et al., 2004; Morken and Zehra, 2013; Smit et al., 2014).

A high DM stimulates fermentation by fecal microbes, which produce toxic gases such as hydrogen sulfide, ammonia, and methyl amine. These gases affect the health of the chicken house (Achiano and Giliomee, 2005). In flatland chicken houses, fermentation in bedding can rot and contaminate feathers, leading to the growth of parasites and pathogenic bacteria (Li et al., 2012; Thanasarasakulpong et al., 2015). In cage chicken houses, high DM feces adhere to scraping boards, making it more difficult and costly to keep them clean.

Increased DM can be attributed to physiological diuresis (Saengchan et al., 2008a) or pathological diarrhea (Ruiz-Palacios et al., 1981). High DM levels are sometimes a normal physiological response to nutritional factors such as drinking water (Saengchan et al., 2008b), electrolytes, pH, osmotic pressure imbalances, or nutritional imbalances. If persistent, these external factors can upset the balance of the gut microbiota and even cause intestinal mucosal damage (Gaucher et al., 2015). In addition, pathological changes caused by Salmonella can also lead to kidney and intestinal endothelium inflammation, which impairs water reabsorption and results in diarrhea (Tie et al., 2018).

Genetic factors also play a role in DM levels. In our previous study, we conducted a genome-wide association study on a Rhode Island Red (RIR) chicken population and found that DM is a moderately heritable trait that is strongly associated with some candidate genes (Zhu et al., 2020). With the development of high-throughput sequencing technology, 16S rRNA gene amplicon sequencing technology provides a convenient way for researchers to investigate the gut microbiota. About 1,014 microorganisms, including bacteria, fungi, and viruses colonize the intestinal tract of human and animals (Zhao and Elson, 2018). In addition, the gut microbiota directly and indirectly produces various short chain fatty acids (SCFAs) and bactericidal bacteriocins to reduce pH or pathogenic microbes, therefore, playing an irreplaceable role in nutrient digestion and absorption (Semova et al., 2012; Martinez-Guryn et al., 2018) and promoting immune defense (Wrzosek et al., 2013; Wang et al., 2019). At present, studies on the gut microbiota in poultry are systematic and comprehensive (Yadav and Jha, 2019). However, most studies focus on gut microbial colonization (Ding et al., 2017; Lee et al., 2019), fat deposition (Wen et al., 2019), and feed efficiency (Singh et al., 2012; Stanley et al., 2012). However, studies on DM have yet to be performed. Thus, it is interesting to further investigate how host-inherited genes regulate DM formation at the transcriptome level by affecting water-salt metabolism (WSM) in RIR chickens.

In the present study, DM does not affect the health of chickens but causes the environmental pollution, which indirectly affects the health of our experimental chickens. To explore the relationship between DM and chickens, we systematically evaluated the influence of the gut microbiota in different intestinal segments to DM. We further explored differentially expressed genes (DEGs) in the spleen and intestinal tissue to understand high DM in chickens. We anticipate that our results will provide effective guidance to solve the problem of environmental pollution in the poultry industry. Further, our results provide a theoretical basis for the development of probiotic feed and breeding, ultimately promoting the sustainable development of animal husbandry.

MATERIALS AND METHODS

Animal Experimental Design

A total of 3,103 age-matched RIR hens from the same henhouse were used in this study. All experiments were approved by the Committee for Animal Care and Use of China Agricultural University (Approval ID: XXCB-20090209). The experimental procedures using chickens were performed according to the Guidelines for Experimental Animals established by the Ministry of Science and Technology (Beijing, China). All the hens were raised in the individual cages with the same feed (Supplementary Table S1) and water supply. We classified DM at 4 levels by the appearance of the feces using the method described in our previous study (Zhu et al., 2020). Briefly, level 1 is very dry, and moisture is invisible to the naked eye. Level 2 is moist and contains a small amount of water. Feces in level 3 have high moisture content and are almost shapeless. Level 4 is liquid and behaves similarly to water on the scraper (Zhu et al., 2020).

When the hens were 67 wk old (November 2018), the DM levels were recorded for each 3 d lasting 12 d (4 consecutive observations; November 10th, November 13th, November 16th, and November 19th). DM from level 1 and level 4 individuals were collected for further analysis. Since Salmonella pullorum (SP) and avian leukosis virus (ALV) are very common in poultry farms, we tested all chickens for SP antibodies and ALV p27 antigen titers by plate agglutination tests (Guo et al., 2010). None of our experimental chickens were infected with either pathogen.

Then, a total of 8 hens (4 per group) from DM level 1 and DM level 4 were euthanized at 72 wk of age (January 2019). Blood, intestinal content, and tissue samples were collected.

Serological Indices

Blood samples from the 8 chickens were incubated at 4°C overnight. Serum was separated by centrifugation at 20,000 g for 5 min in anticoagulant tubes. Immunoglobulin G (IgG), immunoglobulin M (IgM), immunoglobulin A (IgA), endotoxin (LPS), gamma interferon (IFN-γ), tumor necrosis factor alpha (TNF-α), interleukin 4 (IL-4), interleukin 6 (IL-6), and interleukin 10 (IL-10) were measured by ELISA (Lequin, 2005; Table 1) at the Huaying Biotechnology Research Institute (Beijing, China). Then, Student's t tests were used to analyze the difference in serological indices between 2 groups.

Table 1.

Results of serum indices of RIR hens.

Serum index	Detection method	DM group	Concentration range	P value
IgG (g/L)	ELISA	lowest	4.13 ± 0.08	0.13
		highest	4.38 ± 0.12
IgM (g/L)	ELISA	lowest	1.63 ± 0.03	0.53
		highest	1.68 ± 0.06
IgA (g/L)	ELISA	lowest	2.18 ± 0.04	0.11
		highest	2.33 ± 0.06
LPS (EU/mL)	colorimetry	lowest	0.79 ± 0.07	0.57
		highest	0.85 ± 0.08
IFN-γ (pg/mL)	ELISA	lowest	50.81 ± 12.17	0.05
		highest	18.53 ± 4.87
IL-4 (pg/mL)	ELISA	lowest	9.83 ± 1.59	0.22
		highest	12.88 ± 1.56
IL-6 (pg/mL)	Ultraviolet colorimetry	lowest	148.80 ± 13.52	0.02
		highest	94.65 ± 10.29
IL-10 (pg/mL)	colorimetry	lowest	15.99 ± 0.82	0.75
		highest	16.77 ± 2.15
TNF-α (pg/mL)	ELISA	lowest	71.48 ± 15.90	0.04
		highest	27.60 ± 6.03

Abbreviation: RIR, Rhode Island Red.

Estimation of Quantitative Genetic Parameters

In our study, we selected 3103 RIR hens (67 wk old) to assess their DM at 4 levels. We systematically collated the DM observation data recorded for 4 different periods of each chicken, and then eliminated the individual data with incomplete records in the 4 times. We calculated the average DM observation data of 4 times for each chicken. Finally, we evaluated the DM variance according to the statistical model below:

$$\mathrm{Y}=\mathrm{X}\beta +\mathrm{Z}\alpha +\mathrm{e}$$

where Y is the vector of the daily average DM (phenotypic values); β and α are the fixed effects (cage height) vector; X and Z are incidence matrices for the fixed and random additive effects, respectively; and e is the vector of random residuals (Zhu et al., 2020). The DM component estimation was performed using the DMU software package (Madsen, 2006). In our study, the DM observation data was performed using the average information restricted maximum likelihood (AI-REML) algorithm based on the DMUAI module. Heritability was evaluated using single-trait analysis, then a further multitraits analysis was performed to estimate genetic correlations between 2 periods.

In our previous study, we examined phenotypic DM values when the chickens were 45 wk old (July 2018) (Zhu et al., 2020). In this study, we observed the phenotypic DM values for the same chickens at 67 wk of age. Then, the genetic correlation between these 2 periods was calculated using DMU software to evaluate whether the DM trait persisted in a specific individual.

Sample Collection

After the 8 experimental chickens were humanely euthanized, we immediately removed the entire intestine and spleen. Then, the duodenum, jejunum, cecum, and rectum content and mucosal surfaces were collected after dissection. To ensure sample consistency among individuals, we took a 3-cm-long fixed section of each intestinal segment to collect at least 0.5 g content from each chicken. The selected intestinal segment was cut off and the content was squeezed into a storage tube, even the content of the mucosa was scraped down with forceps and a small spoon if necessary. The intestinal content and mucosa were mixed uniformly using a small sterilized wire in the storage tube. Then, all storage tubes were immediately placed in liquid nitrogen and stored at −80°C.

At the same time, after the jejunum and the left cecum segment contents were collected, 100 to 200 mg tissue samples were cut out of the middle of each section. The spleen was diced into 100 to 200 mg samples from middle of the open end. The tissue samples were washed in 0.9% normal saline, placed in RNAse-free storage tubes, frozen in liquid nitrogen, and stored at −80°C.

16S rRNA Gene Amplicon Sequencing of Gut Microbiota

Microbial genomic DNA was extracted with the Qiagen QIAamp DNA Stool Mini Kit (Qiagen, Germany) method and amplified by PCR (Zhang et al., 2018). The 16S rRNA gene V4 hypervariable region was amplified from the microbial DNA using 515F (5′-515F-GTGCCAGCMGCCGCGGTAA) and 806R (5′-GGACTACHVGGGTWTCTAAT) primers (Chen et al., 2019) under the following thermal cycling conditions: 95°C for 2 min, followed by 35 cycles of 95°C for 20 s, 50°C for 20 s, and 72°C for 30 s, with a final extension at 72°C for 10 min. All psyllid DNA samples were amplified in triplicates and subsequently pooled to minimize PCR bias. And the PCR products were purified using 2% 1 × TAE agar gel electrophoresis. In the end, the target bands were collected. PCR amplicons were purified using a GeneJET gel recovery kit (Thermo Scientific). An Ion Plus Fragment Library Kit 48 rxns library kit (Thermo Scientific, MA) was used to construct the library. An Ion S5 XL sequencing platform (Thermo Scientific) was used to construct a small fragment library for 300-bp single-end sequencing at Novogene Company (Tianjin, China).

After removing the adaptor and primer sequences, the raw data of each sample were obtained according to a unique barcode. According to the QIIME2 (Rai et al., 2019) quality control process, chimeric sequences in the raw data were removed to obtain high-quality clean tags (clean data). Low-quality sequences were filtered based on the following criteria (Chen and Jiang, 2014): average Phred score < 19, mismatch ratio < 0.1, and mononucleotide repeats > 10 bp. 16S rRNA gene amplicon sequencing from the gut microbiota were analyzed according to QIIME. Sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs). Annotation of the OTU sequences was performed using the Silva (SSU132) 16S rRNA database (https://www.arb-silva.de/) and a confidence threshold of 0.8 (Edgar, 2013; Quast et al., 2013; Yilmaz et al., 2014; Balvočiūtė and Huson, 2017).

The community composition of each sample was calculated at the kingdom, phylum, class, order, family, and genus levels. Principal coordinate analysis (PCoA) was performed using the weighted UniFrac distance (Lozupone and Knight, 2005). QIIME software (https://docs.qiime2.org/2019.10/) was used to calculate the Shannon index to evaluate the richness and diversity of microbial community in samples. R software (https://www.r-project.org/) was used to show alpha diversity index of differences between the 2 groups. LEfSe software (https://huttenhower.sph.harvard.edu/galaxy/) was used to identify OTUs of gut microbiota. Linear discriminant analysis (LDA) score > 2 and the P-value was corrected to identify significant differences between groups (Segata et al., 2011).

RNA-Seq for Spleen, Jejunum, and Cecum

Total RNA from chicken spleen, jejunum, and cecum was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA). The RNA concentration was measured using a NanoDrop ND2000 spectrophotometer (Nano-drop products, Wilmington, NC). The integrity of the extracted RNA was determined using agarose gel electrophoresis, and the RNA integrity number (RIN) was determined with an Agilent Bioanalyzer 2100 system (Agilent Technologies, CA).

Then, high-quality RNA samples (concentration >50 ng/μL, OD260/280 = 1.8 ∼2.2, OD260/230 = 1.8∼ 2.2, RIN > 8, 28S:18S ≥0.5) were used to construct sequencing libraries. Transcriptome libraries were prepared using an Illumina TruSeq RNA Sample Prep Kit using 3 μg total RNA. Briefly, poly(A) mRNA was isolated from the total RNA samples with Oligo-dT magnetic beads (Invitrogen) (Xu et al., 2013). Then, second-strand cDNA was synthesized using RNase H and DNA polymerase I. Libraries were size selected for cDNA target fragments of 200 to 300 bp using 2% Low Range Ultra Agarose gels, followed by PCR amplification. PCR amplification was used T7 and SP6 primers under the following conditions: 94°C for 2 min followed by 35 cycles at 95°C for 15 s, 56°C for 30 s, 72°C for 2 min, and a final cycle at 72°C for 5 min. PCR products were then analyzed on 2% agarose gels and sequenced. Finally, the 24 libraries were sequenced using an Illumina NovaSeq 6000 platform (150 bp paired-end reads) at Neogen Biotechnology Co., LTD (Shanghai, China).

Raw FASTQ data were first checked for quality using fastp software with default parameters (Chen et al., 2018) to remove joint, blank read, and low-quality sequences (sequences with N ratio > 10% and Q-value < 20%). Then clean reads were mapped to the chicken reference genome Gallus_gallus-5.0 (https://www.ncbi.nlm.nih.gov/assembly/GCF_000002315.6) using HISAT2 software (Wen, 2017). Gene expression was normalized using the fragments per kilobase million (FPKM) method (Trapnell et al., 2010). Then, we performed differential expression analysis of the cecum, jejunum, and spleen in the high and low DM groups using the DESeq package (Anders and Huber, 2010) in R software. And the DEGs were determined using the cutoff at P-value < 0.05, and |log2FoldChange|≥1.

Gene ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) (Kanehisa et al., 2014) functional enrichment were carried out for DEGs in different tissues. The pathway and annotation information were derived from GO and KEGG using clusterProfiler software (Yu et al., 2012) with P < 0.05, indicating significantly enriched pathways and annotated GO terms.

RESULTS

The Character of DM From Our Experimental Chickens

Through long-term observation of DM for several times, we found that DM of most laying hens was at the level 1. The higher the DM was, the smaller the number of chickens was, and the DM 4 was less than 10. The DM of some hens would change, but most of them would remain the same. The former observation also showed that there was no significant difference in important economic traits (egg-laying rate, egg weight, mortality, etc) of chickens with different levels of DM, indicating that DM may be caused by physiological factors.

Serological Indices Between the Highest and Lowest DM Chickens

In this study, serological indices were analyzed in 8 chickens using ELISA. Only TNF-α (P = 0.04) and IL-6 (P = 0.02) showed significant differences between the high and low DM groups. There were no significant differences in the other serological indices between the 2 groups (P > 0.05; Table 1).

DM Heritability and Genetic Correlation Between Two Age Groups

The heritability of DM in 67-wk-old chickens was 0.219, indicating a medium heritability. This is similar to our previous calculation in 45-wk-old hens (Zhu et al., 2020). Additionally, the genetic correlation of DM between 45-wk-old and 67-wk-old chickens was 0.435, indicating a positive correlation between the 2 age groups.

Gut Microbiome Composition in Different DM Groups

The rarefaction curve of Shannon index (Supplementary Figure S1) showed the amount of sequencing data in the sample is sufficient. The sequence mapping rate of each sample was qualified with mapping ratio >99% (Supplementary Table S2). To explore the differences in the gut microbiota from the duodenum, jejunum, cecum, and rectum, intragroup difference analysis was conducted based on Shannon diversity indices (Figure 1). The results showed that cecum microbial richness was higher than in the other 3 intestinal segments, and the abundance of cecal microorganism in lowest DM group (DM = 1) was significantly higher than that in highest DM group (DM = 4). While the other 3 groups showed no significant intragroup differences.

Figure 1.

High and low DM level of RIR hens showed different Shannon indices in 4 different intestinal segments. Abbreviations: DM, dropping moisture; RIR, Rhode Island Red.

The weighted UniFrac PCoA showed that the microbes in the cecum were distinct from the other 3 intestinal segments and that the microbial communities were more stable. There are largely similar microbial communities between the duodenum, jejunum, and rectum, but the microbial communities of the rectum are more diverse (Figure 2). So the PCoA shows the similar results to Shannon diversity indices.

Figure 2.

PCoA on weighted UniFrac distance by Bray-Curtis for high and low DM in 4 different intestinal segments. Abbreviation: DM, dropping moisture.

To investigate the predominant microbes of 4 intestinal segments, we compared the composition and abundance of the gut microbiota at the phylum (Figure 3) and the genus levels (Figure 4). Eight RIR hens were divided into the highest and lowest DM groups. In this study, the top 10 microbes were selected to draw the relative abundance histogram, so as to demonstrate the proportion of microbial composition among different samples or groups. Overall, the dominant phyla in the 4 different intestinal segments were Firmicutes, Proteobacteria, and Bacteroidetes.

Figure 3.

The top 10 intestinal microphylum composition in 4 intestinal segments. The first 4 numbers (1–4) represent the low DM group and the latter 4 numbers (6–8) refer to the high DM group. Abbreviation: DM, dropping moisture.

Figure 4.

The top 10 microbial composition at the genus level in the duodenum (A), jejunum (B), cecum (C), and rectum (D). The first 4 numbers represent the low DM group and the latter 4 numbers refer to the high DM group. Abbreviation: DM, dropping moisture.

To further characterize differences at the genus level, 4 different intestinal segments were individually analyzed (Figure 4). Then, LDA Effect Size (LEfSe) analysis was performed based on microbial taxonomy to check for microbes as differential bacteria that had a significant effect to distinguish the samples (Segata et al., 2011; Figure 5).

Figure 5.

LEfSe analysis of the gut microbiota in the duodenum (A), jejunum (B), cecum (C), and rectum (D).

The dominant bacterial genera in all the duodenum, jejunum, and rectum samples were Lactobacillus and Stenotrophomonas. The dominant bacterial genera in the cecum are Bacteroidetes, Faecalibacterium, and Lactobacillus. In addition, the cecum microbiome is more diverse than the others. There was a large microbiome difference between the duodenum samples from different chickens, and the individual differences were also obvious.

Lactobacillus was the dominant bacteria, the second is Stenotrophomonas, in the highest DM group. While the Lactobacillus and Stenotrophomonas had the similar relative abundance in the lowest DM group (Figure 4A). LEfSe analysis showed that Cetobacterium was significantly enriched in the highest DM group, while Sterolibacterium was more significantly enriched in the lowest DM group (Figure 5A).

The jejunum and duodenum are both small intestine tissues, and their gut microbial diversity is similar (Figure 2). Similar to the microbial compositions in duodenum, Lactobacillus was still the most dominant bacteria while Stenotrophomonas ranked second in jejunum (Figure 4B). Each group was identified as viable biomarker by LEfSe analysis, with Caulobacterales in the lowest DM group and Terrisporobacter in the highest DM group (Figure 5B).

Similar to the phylum level, the cecum has a unique microbial composition at the genus level. Bacteroides, Faecalibacterium, and Lactobacillus were the dominant bacteria in the cecum. In addition, there was a significant difference between the dominant bacteria in the cecum and the other 3 intestinal segments. As a whole, the total relative abundance of the top 10 bacteria (35.25%) in the cecum were lower than the abundance of the top 10 bacteria in the other 3 intestinal segments, with relative abundances of 87.68% (duodenum), 84.54% (jejunum), and 75.25% (rectum), respectively. Moreover, there was no significant difference in the abundance of cecal microbiota among different individuals (Figure 4C). In addition, 8 bacteria were detected as potential biomarkers to distinguish different microbes in the cecum between the 2 groups, among which Elusimicrobia (phylum), Elusimicrobia (class), Elusimicrobiales (order), Elusimicrobiaceae (family), Elusimicrobium (genus) may all due to the effect of Elusimicrobium (genus) (Figure 5C).

There were large differences in bacterial abundance among different individuals in the rectum, which was more pronounced in the lowest DM group (Figure 4D). LEfSe analysis revealed that only 4 rectal bacteria in the 2 groups could be used as potential biomarkers to distinguish the different DM groups as Lactobacillales in highest DM group while Terrisporobacter, Blautia and Megamonas in lowest DM group (Figure 5D).

Then, we further display the relative abundance of differential bacteria excavated in the 4 intestinal segments through LEfSe analysis (Additional file 1). The results indicate that the relative abundance of Bacilli (class), Clostridia (class), Lactobacillales (order) were generally high, averaging over 1%. And Megamonas and Faecalicoccus were followed, while the other microbes are less abundant. Finally, 6 different bacteria identified by LEfSe analysis in the rectum and cecum were reported to involved in WSM (Supplementary Table S3).

Differential Expression Analysis and Functional Enrichment

The clean reads of the 23 analyzed samples are all high-quality with Q20 >96% and Q30 >91%, and the GC content was approximately 50% (Supplementary Table S4). To further explore the genetic mechanism of DM in RIR hens, we performed RNA-seq analysis of tissue samples from the spleen, jejunum, and cecum (Figure 6). DEGs functions were further revealed through GO gene enrichment analysis (Figure 7). Only 12 DEGs were found in the spleen between the highest and lowest DM of RIR hens (Figure 6A), so the GO and KEGG pathways could not be analyzed using the DEGs in the spleen because of the small number. A total of 1,342 DEGs (731 downregulated genes and 611 upregulated genes) were detected in the jejunum (Additional file 2, Figure 6B). These 1,342 genes were significantly enriched in only 3 GO terms, which all indicated extracellular products (Figure 7A). No KEGG pathway was enriched for these genes. A total of 165 DEGs (45 downregulated genes and 120 upregulated genes) were observed in the cecum (Additional file 3, Figure 6C). Numerous GO terms were annotated using DEGs from the cecum (Figure 7B).

Figure 6.

Overall analysis of gene expression by DM of RIR hens in different tissues. Volcano plots of DEGs from the spleen (A), jejunum (B), and cecum (C) in the highest and lowest DM of chickens. Abbreviations: DEGs, differentially expressed genes; DM, dropping moisture; RIR, Rhode Island Red.

Figure 7.

Annotation of differentially expressed genes from RIR hens' intestinal tissues. The bar graph shows the top 10 GO terms of DEGs from the jejunum (A) and cecum (B). Abbreviations: DEGs, differentially expressed genes; GO, gene ontology; RIR, Rhode Island Red.

Subsequently, the upregulated and downregulated genes in jejunum and cecum were analyzed by KEGG based on R packages clusterProfiler. The upregulated genes in jejunum were mainly enriched in metabolism of nutrients and the synthesis of biomacromolecules, such as carbon metabolism, glucagon signaling pathway, pyruvate metabolism, galactose metabolism, biosynthesis of cofactors, biosynthesis of amino acids, fructose, and mannose metabolism, etc (Supplementary Figure S2). While the downregulated genes in jejunum were mainly enriched in Inhibit the occurrence of disease, such as Cytokine-cytokine receptor interaction, Viral protein interaction with cytokine and cytokine receptor, T cell receptor signaling pathway, PD-L1 expression and PD-1 checkpoint pathway in cancer, chemokine signaling pathway, calcium signaling pathway, etc (Supplementary Figure S3). However, there were fewer DEGs in cecum as the downregulated genes were not annotated to KEGG pathway, while the upregulated genes were annotated to 6 KEGG pathways related to metabolism and transport, such as PPAR signaling pathway, Taste transduction, Proximal tubule bicarbonate reclamation, folate biosynthesis, glyoxylate and dicarboxylate metabolism, ABC transporters (Supplementary Figure S4).

Finally, to accurately select the DEGs that were highly related to WSM, we used |log2FoldChange|≥3 to screen out 256 DEGs from jejunum and cecum. Among them, 8 downregulated genes and 13 upregulated genes were detected in the cecum, while 153 downregulated genes and 82 upregulated genes were observed in the jejunum. Of these genes, 14 DEGs participated in WSM (Table 2) will be the focus of subsequent functional verification.

Table 2.

The selected differential genes (|log2FoldChange|≥3) related to WSM in the cecum and jejunum.

Intestinal segment	Gene	Log2FoldChange	DM group	P-value	Description
Cecum	DMBT1L1	−5.3468	lowest	0.011217	deleted in malignant brain tumors 1 like 1 (pseudogene)
Cecum	PCK1	2.90258	highest	0.014889	phosphoenolpyruvate carboxykinase 1
Cecum	USH1G	3.276039	highest	0.004827	USH1 protein network component sans
Jejunum	SLC5A12	−11.523	lowest	0.000135	solute carrier family 5 member 12
Jejunum	FABP6	−10.528	lowest	7.47E-11	fatty acid binding protein 6
Jejunum	SLC10A2	−10.4631	lowest	4.1E-09	solute carrier family 10 member 2
Jejunum	SLC26A3	−8.47042	lowest	7E-08	solute carrier family 26 member 3
Jejunum	TMIGD1	−7.7718	lowest	5.43E-07	transmembrane and immunoglobulin domain containing 1
Jejunum	SLC6A20	−5.29769	lowest	1.6E-05	solute carrier family 6 member 20
Jejunum	VPREB3	−4.26711	lowest	1.83E-05	V-set pre-B cell surrogate light chain 3
Jejunum	CD72	−4.21279	lowest	7.98E-06	CD72 molecule
Jejunum	KCNJ15	−3.42447	lowest	0.000236	potassium inwardly rectifying channel subfamily J member 15
Jejunum	SLC28A3	3.370265	highest	2.24E-05	solute carrier family 28 member 3
Jejunum	STEAP4	4.164456	highest	0.000125	STEAP4 metalloreductase

Note: Compared to lowest DM group, Log2FoldChange means highest DM group are down (-) or up (+) regulated.

DISCUSSION

In this study, DM was influenced by multiple factors. To explore the genetic mechanism of DM, we assessed its heribility and analyzed the gut microbiota and DEGs of hens.

DM is a Heritable Physiological Phenomenon

First, the serological indicators for both DM groups were measured. Only TNF-α and IL-6 showed significant differences between the highest and lowest DM of our experimental chickens. TNF-α and IL-6 are primarily associated with immunity in poultry (Sevimli et al., 2008; Steiner et al., 2014). Thus, we might infer that the high DM of RIR hens was not a pathogenic trait such as diarrhea, but a physiological phenomenon. The high DM trait might be caused by WSM related to host genetics and the gut microbiota.

Then, we calculated the DM heritability. The heritability of DM calculated at 45 and 67 wk was similar. Moreover, there was a positive genetic correlation between the 2 data sets. This result indicates that DM traits persisted in the chicken population and were heritable. Therefore, in addition to improving the feeding environment and feed conditions, it is also of practical significance to explore the genetic mechanism of DM.

DM Has an Effect on Gut Microbiota of Chicken

Considering the close relationship between DM and intestinal health, we tested the gut microbiota of the 2 chicken groups. Gut bacteria are dynamic, developing, complex, and play an important role in host nutrition, WSM, and immunity (Gomaa, 2020; Pan et al., 2021; Wastyk et al., 2021). There were significant differences in the gut microbiota in different intestinal segments (Yan et al., 2019) and different developmental stages (Ding et al., 2017).

Just like intestinal tract of many mammals, the chicken gut is an important tissue for the absorption and metabolism of many nutrients. The gut microbiota in the process of metabolism will produce a lot of inorganic salts or small molecules, such as SCFAs (Clavijo and Flórez, 2018), this process will affect the osmotic pressure or filtration barrier of the intestinal tract, resulting in the reabsorption or drainage of water. Both the urination and defecation organs of chickens are cloaca, and the mixture of feces and urine will cause a large difference in DM. The gut microbiota of the large intestine is richer and can absorb moisture and inorganic salt (Gasaway et al., 1976). Because cecum of chicken is a relatively enclosed space, which making the microbial metabolism in it more stable and enduring than other intestinal tracts (Lan et al., 2002; Saxena et al., 2016). In this study, PCoA analysis and the Shannon index demonstrated similar results. Fortunately, we have also found some bacteria involved in water and salt metabolism in cecum of chickens.

In our study, we focused on the difference of gut microbiota raleted to WSM in high and low DM chickens based on LEfSe analysis. Among them, 6 gut microbiota were reported to physiologically contribute to WSM, so that forms DM in chickens. Cetobacterium, the dominant bacterium in most fish intestines, produces vitamin B12 that is absorbed into the blood and liver by the gastrointestinal mucosa (Sugita et al., 1991; Tsuchiya et al., 2008). Sterolibacterium metabolism in the gut produces cholesterol, which can also be synthesized in the liver bile acid (Chiang et al., 2008). Additionally, bile acids are involved in gut microbial metabolism, and can change intestinal permeability to affect stool consistency (Odunsi–Shiyanbade et al., 2010). E. coli bacteriophages can make Elusimicrobium significantly reduced in the intestines of broiler hens, thus relieving E. coli-induced diarrhea to some extent (Lu et al., 2017). Roseburia belongs to Firmicutes, and can metabolize polysaccharides to produce butyrate in the human colon, which affects the environmental osmotic pressure in the intestinal tract (Sheridan et al., 2016). Faecalicoccus adapts in chicken cecum and can effectively regulate intestinal pH and water and salt balance (De Maesschalck et al., 2014). We also observed that Faecalicoccus was enriched in the cecum samples from this study. Megamonas produce acetate and propionate in the duck cecum, and can also affect the physiological balance of the intestine (Chevrot et al., 2008). Moreover, the relative abundances of Faecalicoccus and Megamonas were higher in the lowest DM group. In summary, the pivotal bacteria are directly and indirectly involved in the transport and metabolism of water and inorganic salts in the animal gut. Thus, the formation of high DM may be the result of a combination of microorganisms and other factors.

Host Genes Related to WSM Are Involved in DM Formation

In the transcriptome analysis of the spleen, we found only 6 of the 12 DEGs participate in disease resistance traits, and none of them are involved in WSM. The 6 genes of LECT2 (L’Hermitte et al., 2019), CCL20 (Ranasinghe and Eri, 2018), TGM3 (Feng et al., 2020), AvBD2 (Terada et al., 2018), AvBD7 (Bailleul et al., 2016), and PDIA2 (Zou et al., 2013) are related to antibacterial immunity. This result indicates that there is almost no significant difference in systemic immunity between high and low groups. While the DEGs in the cecum numbered up to 164, which was less than that in jejunum, but the associated GO terms were more diverse. We identified 2 important candidate genes, PCK1 and LOC395381, which are related to intestinal absorption in the top 10 GO terms. PCK1 is a rate-limiting enzyme that regulates gluconeogenesis and is involved in maintaining blood glucose levels (Proft et al., 1995).

In addition, KEGG enrichment analysis was performed on the upregulated and downregulated genes in cecum and jejunum, respectively, and it was found that the upregulated genes in cecum and jejunum mainly regulate the metabolism of nutrients. While the downregulated genes in jejunum mainly regulate the occurrence of diseases. The cecum in chicken can absorb large amounts of water and inorganic salts, and upregulated DEGs in cecum were enriched in PPAR signaling pathway and ABC transporters. While PPAR signaling pathway can regulate the absorption of carbohydrates and fatty acids in the intestines of animals, and even affect the permeability of the intestines of animals, thus affecting the abundance of intestinal microflora (Zong et al., 2019; Tomas et al., 2016). Therefore, PPAR signaling pathway also plays an important role in maintaining intestinal immune barrier function. And ABC transporters, transmembrane proteins expressed in kidneys and intestine, could regulate absorption of intestinal nutrients and oral drugs, playing a vital role in maintaining intestinal barrier function (Brand et al., 2006; Vagnerová et al., 2019).

To further screen the DEGs, we tightened our screening standards, which identified 256 highly expressed genes. Among them, 14 genes participate in WSM, especially in the animal gastrointestinal tract (Table 2). DMBT1 is associated with gastrointestinal epithelial cell proliferation and differentiation. This gene is also involved in gastrointestinal mucosa protection and maintaining homeostasis in the mucosal environment (Kang and Reid, 2003). USH1C, forms a complex that stabilizes intestinal microvilli, and can form dense condensates via liquid-liquid phase separation in cells and in vitro (He et al., 2019). The jejunum, in contrast, had significantly more DEGs. Five of these genes (KCNJ15, FABP6, SLC10A2, SLC26A3, and SLC6A20) are involved in WSM in the gastrointestinal tract. KCNJ15 regulates the gastric mucosal cell permeability to K⁺, thereby affecting gastric juice secretion (He et al., 2011). The ileal fatty acid binding protein (FABP6) is involved in enterohepatic bile acid metabolism and intracellular transport in the small intestine (Bernlohr et al., 1997; Landrier et al., 2002). SLC10A2 is highly expressed in the ileum, and is an efficient transporter of conjugated bile acids (Love et al., 2001; Jung et al., 2004). Mutations of SLC26A3 Cl/HCO exchangers in the colon could cause congenital chloride diarrhea (Mäkelä et al., 2002; Talbot and Lytle, 2010). SLC6A20 mediates most neutral l-amino acid transport to proximal tubular epithelial cells in the small intestine and kidney via a Na⁺ co-transport system (Ristic et al., 2006). Three genes (SLC5A12, TMIGD1, and SLC28A3) are expressed in kidney. SLC5A12 is expressed predominantly in the kidney cortex and gastrointestinal tract, and can induce Na⁺-dependent lactate and nicotinate transport (Srinivas et al., 2005). TMIGD1 regulates and participates in the transmembrane transport of some macromolecular substances in renal epithelial cells (Arafa et al., 2015; Meyer et al., 2017). Human concentrated nucleoside transporter 3 (hCNT3) protein is a SLC28A3 variant that plays a crucial role in absorption along the nephron (Errasti-Murugarren et al., 2009). The other 3 genes are mainly expressed in the liver (STEAP4) and immune B cells (VPREB3 and CD72). A small amount of hepatic STEAP4 is involved in vesicle secretion and endocytosis, and can regulate blood glucose balance (Ramadoss et al., 2010). VPREB3 may regulate the maturation and secretion of immune factor in chicken B cells (Rosnet et al., 2004). CD72 regulates IgM levels in the blood, thereby affecting the B cell immune response (Hitomi et al., 2012). Therefore, we speculated that chickens with high DM have poor cecal and jejunal permeability, which affects the absorption and osmotic function of nutrients, destroying the intestinal barrier function and indirectly leading to the destruction of the intestinal environment. it could then break the balance of abundance of gut microbiota. And this is a normal physiological phenomenon, so it would not affect the health of chickens. The specific genetic mechanism still needs further functional verification to be revealed.

In conclusion, DM is heritable in RIR hens and is not a pathological response, but is a combined action of host genetics and the gut microbiota that influences intestinal WSM. A limitation of this study is the small sample size, and the main reason is that the grade of DM was not invariable in the previous observation. So it was hard to pick out more experimental chickens with DM of level 4. Moreover, 16S rRNA gene sequencing analyses and transcriptome analyses in animal experiments generally require at least 3 biological replicates (Ng et al., 2015; Cui et al., 2017; Liu et al., 2019; Huang et al., 2020). Though there are existing individual differences which may affect the accuracy of the results to a certain extent, this study can still well reveal the main mechanism of DM formation in RIR hens.

SUPPLEMENTARY DATA

All the bacterial 16S rRNA amplicon sequencing data are deposited in the NCBI Short Read Archive under the accession number PRJNA662163. The RNA-seq datasets are available with accession number PRJNA661615.