All animal studies were conducted according to the guidelines for the care and use of experimental animals established by the Ministry and Rural Affairs of the People’s Republic of China. The project used for this animal experiment was approved by the Animal Care and Use Committee at Guizhou University, China (approval number: EAE-GZU-2022-T050).

The chickens used in these experiments were raised at the Scientific Research Chicken Farm of Guizhou University from June 2022 to October 2022. A total of 106 chickens were collected including 51 Guizhou yellow chickens (25 male chickens and 26 female chickens) and 55 Wumeng Black-Bone chickens (29 male chickens and 26 female chickens). Guizhou Yellow broiler is a Chinese hybrid line (Weining ♀ × New Hampshire × Plymouth Rock ♂) that possesses high market weights and tender meat, while Wumeng Black-Bone chicken possesses black tissues and bones. They are both meat-type chicken and are popular with consumers. The hatching batches, feeds, feeding methods, and management conditions were consistent. Specifically, chickens were fed in the same house, and cages were constructed of three-layer iron. The stocking density was 16 chickens from 0 to 4 weeks (male and female chickens mixed, not gendered); 8 chickens from 4 to 10 weeks (4 male and 4 female chickens mixed); and one chicken from 10 to 18 weeks. The chicken house temperature control was regulated using roller shades. The daily lighting time was 16 h, and the temperature ranged from 15 to 35°C. The chickens were fed with commercial feed twice a day at 9 AM and 5 PM, with access to food and watered ad libitum. Chickens were vaccinated according to routine immunization procedures for Marek’s and Newcastle disease, infectious bronchitis, bursal virus, and avian influenza. The chickens were weighed every 2 weeks with an electronic scale. No antibiotics were added within 1 month prior to cecal content sample collection.

Since local chickens in these areas are generally 18 weeks old, these chickens were slaughtered at 18 weeks of age in this study. Chickens were euthanized by CO₂ asphyxiation referred to previously reported methods (Haetinger et al., 2021). Whole blood was collected from wing veins and then was left undisturbed at room temperature to clot. After that, the clot was removed by centrifuging at 1000 × g for 5 min in a refrigerated centrifuge to obtain serum. Approximately 2 g cecal contents were collected into a clean sterile plastic 2 mL Eppendorf tube. Samples were immersed in liquid nitrogen immediately and were stored at −80°C. Due to five samples that showed hemolysis, only 101 serum samples were obtained for metabolome analysis including serum samples from 50 Guizhou yellow (25 male chickens and 25 female chickens) and 51 Wumeng black-bone (26 male chickens and 25 female chickens) chickens. All 106 cecal content samples for 16S rDNA gene sequencing and 101 serum samples for non-targeted metabolomics were processed at Shanghai Majorbio Bio-pharm Technology (Majorbio, Shanghai, China) and Shanghai Applied Protein Technology (Aptbio, Shanghai, China), respectively. The experimental flow chart is shown in Figure 1.

DNA was extracted from cecal content samples using a Magnetic Soil and Stool DNA Kit (Tiangen, Beijing, China) following kit instructions. The concentration and purity of DNA were determined using UV spectroscopy with a NanoDrop 1,000 instrument (Thermo Fisher, Pittsburg, United States) and 0.8% agarose gel electrophoresis. Primers 338F (5’-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5’-GGACTACHVGGGTWTCTAAT-3′) were used to amplify the V3–V4 region of the 16S rRNA gene as previously described (Huse et al., 2008; Caporaso et al., 2011). The annealing temperature was set at 55°C for 27 cycles. The amplicons were purified from agarose gels using an AxyPrep DNA gel extraction kit (Corning, Glendale, United States). Purified amplicons were pooled in equimolar amounts, and paired-end sequencing was performed on an Illumina MiSeq platform (Illumina, San Diego, USA) according to standard protocols. A two-step tailed PCR approach was used to construct the paired-end libraries (Miya et al., 2015). The first-round PCR amplified the target region using a region of interest-specific primer; the overhang adapter sequence was used in the second-round tailed PCR to add indices and adapter sequences. The constructed libraries were sequenced with NovaSeq 6,000 SP 500 Cycle Reagent Kit (Illumina, San Diego, USA) at Majorbio Bio-pharm Technology (Shanghai) Co., Ltd. Barcodes, primers, and low-quality and ambiguous sequences were filtered out with Trimmomatic software (version 0.33) (Bolger et al., 2014). Cut adapt (version 1.9.1) was applied to identify and remove primer sequences (Martin, 2011). Paired-end reads from the clean data sets were clustered into tags by FLASH (version 1.2.11) (Magoc and Salzberg, 2011). Tags were then clustered into amplicon sequence variants (ASVs) using DADA2 (Callahan et al., 2016). ASV taxonomic assignments were conducted by the RDP classifier (version 2.2) (Wang et al., 2007). ASVs were annotated in the database Silva¹ (Quast et al., 2012).

Relative abundance of ASVs >0.05% was selected to construct clustering co-occurrence networks. The SparCC algorithm (Friedman and Alm, 2012) was used to construct a bacterial co-occurrence network. Correlations between ASVs (nodes) were calculated based on relative abundance using the PCIT algorithm (Reverter and Chan, 2008), and the paired taxa with absolute sparse correlation coefficient > 0.35 were selected for the next network construction. Cytoscape (version 3.7.1) (Lopes et al., 2010) was employed to evaluate the topological characteristics of co-occurrence networks and visualize the network. The stability of the network was represented by the percentage of negative interactions (competition) (Coyte et al., 2015; Hernandez et al., 2021). The complexity was calculated using the average number of lines connecting each node (Bader and Hogue, 2003).

QIIME 2 platform (Bolyen et al., 2019) ² was used to calculate the alpha diversity of cecal microbiota with the Chao1, Shannon, and phylogenetic diversity (PD) indices (Shannon, 1948; Chao, 1984; Faith, 1992). Other statistical analyses and result visualization were performed on the R V4.4.0 environment (R Core Team, 2022). The performance data were analyzed using the two-way ANOVA on SPSS (version 26.0). Bray–Curtis distances were calculated to compare the beta diversity of the cecal microbial community between male and female chickens using principal coordinate analysis (PCoA). Alpha-diversity differences between the two groups were analyzed using the Wilcoxon rank-sum tests. Permutational multivariate analysis of variance (PERMANOVA) was used to explore the effects of genders on the composition of cecal microbiota (McArdle and Anderson, 2001). The effect size of each factor (R2) of PERMANOVA was used to determine the contribution and significance of genders on cecal microbial composition, and the p-values were calculated based on 9,999 permutations. Linear discriminate analysis (LDA) coupled with effect size measurements (LEfSe) was performed online platform³ (Segata et al., 2011) to identify bacterial taxa that differed significantly between male and female chickens, and the threshold was set to LDA > 2.0 and p-value <0.05. Spearman’s correlation analysis was conducted on R with ‘cor.test’ function to compute Pearson’s correlation coefficients between serum metabolites and cecal microbiota.

The metabolome data were processed on the MetaAnalyst 5.0 online platform (Pang et al., 2021).⁴ The dataset was normalized by log₁₀ transformation of the m/z values. After sum normalization, the processed data of the two groups were compared using orthogonal partial least squares discriminant analysis (OPLS-DA). The robustness of the model was evaluated with 7-fold cross-validation and response permutation testing. The variable importance in the projection (VIP) value of each variable in the OPLS-DA model was calculated to assess its contribution to the classification. The importance threshold for the impact values for metabolic pathway topology analysis was set at 0.10 (Wang et al., 2012).

We utilized 106 chickens to examine whether breed factors significantly influenced production performance indicators and initially compared growth performance differences between Guizhou yellow chickens (GHC, n = 51) and Wumeng Black-Bone (WMC, n = 55) chicken breeds. We found that except for day 0 (t-test, p = 0.47) and week 4 (p = 0.077), GHC and WMC chickens significantly differed in their body weights (p < 0.05; Supplementary Table S1; Figures 2A C). The results of two-way ANOVA showed that there was no significant interaction between breed and sex for 18-week-age weights (p = 0.681; Supplementary Table S2). We then evaluated the role of gender on chicken growth performance and found that gender differences were significant for body weights (p < 0.01; Figure 2B). In particular, the body weights for the GHC male chickens (2205.92 ± 225.65 g, n = 25) were higher than body weights of GHC female chickens (1801.81 ± 191.62 g, n = 26) at the age of 18 weeks (Figure 2A), and the weights of WMC male chickens (1891.24 ± 220.36 g, n = 29) were also higher than WMC female chickens (1453.85 ± 189.00 g, n = 26).

A total of 5,087,788 high-quality reads were generated from 106 samples (47,998 reads per sample), and 1955 ASVs were clustered. Taxonomic assignments revealed that these ASVs could be classified into 16 bacterial and archaeal phyla, and 3 phyla were present in all samples. The mean relative abundance of 5 phyla was >1%, including Bacteroidetes (46.88%), Firmicutes (45.52%), Desulfobacterota (2.41%), Actinobacteriota (2.38%), and Synergistota (1.46%) (Supplementary Table S3). We identified 63 genera, and the predominant microbes in the ceca were Bacteroides (35.95%), Megamonas (8.48%), Phascolarctobacterium (7.87%), Prevotellaceae_UCG-001 (6.09%), and Ruminococcus_torques_group (5.83%) (Figure 3; Supplementary Table S4). A permutation analysis for male chickens and female chickens showed that gender (r = 0.039, p = 0.002; Supplementary Figure S1) significantly affected the composition of chicken cecal microbiota. We further examined the structural characteristics of cecal microbiota interaction networks according to gender. A total of 252 and 273 ASVs were selected from cecal microbiota of the male and female groups, respectively, to construct microbial co-occurrence networks. The results showed that the network stability index for male and female chickens was 37.50 and 46.26%, respectively, and the complexity for female chickens was 7.01 and was 3.35-fold larger than that of the male microbial network (2.09) (Table 1).

In our taxonomic groupings, the Wilcoxon rank-sum tests and PCoA indicated the absence of gender differences for alpha and beta diversity of the cecal microbiome in these chickens (Figures 4A,B). We used LEfSe analysis at different taxonomic levels to determine whether we could identify specific taxa that displayed a gender bias. In the total cohort, no significant gender differences in bacterial phyla were detected (Supplementary Figure S2) although genera more abundant in male chickens included Lactobacillus and Family_XIII_UCG_001 (LDA > 2.0, p < 0.05) while Eubacterium_nodatum_ group, Blautia, unclassified_Anaerovoraceae, Romboutsia, Lachnoclostridium, and norank_Muribaculaceae were more abundant in female chickens (Figure 4C).

A non-targeted metabolomics analysis was performed on 101 chicken serum samples (male chickens n = 51, female chickens n = 50), 7,086 and 4,910 metabolite features were obtained in positive and negative modes, and 494 and 272 metabolites were annotated, respectively (Supplementary Tables S5, S6). An OPLS-DA model was constructed to identify differential serum metabolites between male and female chickens, and these plots displayed a clear separation within both positive and negative modes (Figures 5A,B).

To evaluate the differences in metabolic profiles between male and female chickens, we examined the patterns of specific metabolites. In the positive mode, 37 differential metabolites (VIP > 2, p < 0.05) were identified (Figure 5C; Table 2). In particular, 22 metabolites were enriched for male chickens and they included myristoyl-l-carnitine, prostaglandin F2α alcohol methyl ether, and Lys-Ile-Lys. Conversely, 15 metabolites were enriched in female chickens including riboflavin, taurochenodeoxycholic acid, and 3-hydroxy-3′,4′-dimethoxyflavone. In negative mode, a total of 13 metabolites showed significant differences between the male and female groups in which 7 metabolites were increased in the male chickens including 1β-hydroxytestosterone, prostaglandin F2β, and linoleic acid (Figure 5D; Table 3). Concentrations of 6 metabolites were significantly higher in female chickens and included phosphatidylethanolamine 32:1, phosphatidylglycerol 34:2, and N-(2-furoyl)glycine. It was worth noting that sphingomyelin metabolites contained N-tetracosenoyl-4-sphingenine (M649T31, positive mode), sphingomyelin (M792T139, positive mode), and N-[1,3-dihydroxyoctadec-4-en-2-yl]tetracos-15-enamide (M647T30, negative mode), were increased in female serum, and were significantly correlated with each other (Figure 6A). Among them, N-[1,3-dihydroxyoctadec-4-en-2-yl]tetracos-15-enamide and N-tetracosenoyl-4-sphingenine were most highly and significantly (r = 0.922, P < 0.001) correlated with female chickens. The correlation between sphingomyelin with N-tetracosenoyl-4-sphingenine and N-[1,3-dihydroxyoctadec-4-en-2-yl]tetracos-15-enamide was 0.515 and 0.491, respectively (P < 0.001). Three hormone-related metabolites (prostaglandin F2α alcohol methyl ether, prostaglandin F2β, and 11β-hydroxytestosterone) were enriched in male chicken serum. However, no hormone-related metabolites were identified in female chickens. We then performed a metabolic pathway analysis for the differential metabolites we identified above. The pathway enriched in male chickens was linoleic acid metabolism, while the sphingolipid metabolism pathway was enriched in female chickens (Figures 5E,F).

We further established relationships between differential serum metabolites and differential cecal microbes based on gender using Spearman’s rank correlation analysis. The correlation coefficients were calculated for each pair of relative abundance of bacterial species and metabolites (Figures 6B,C; Supplementary Tables S7, S8). For sex hormone metabolites, 11β-hydroxytestosterone was enriched in male chickens and was significantly negatively correlated with Blautia (r = −0.217, p = 0.03). All three sphingomyelin metabolites were significantly positively correlated with a variety of cecal microbes that included Blautia, unclassified_Anaerovoraceae, Romboutsia, and Lachnoclostridium. Specifically, N-[1,3-dihydroxyoctadec-4-en-2-yl]tetracos-15-enamide was positively correlated with Blautia (r = 0.291, p = 0.003), unclassified_Anaerovoraceae (r = 0.267, p = 0.007), Romboutsia (r = 0.426, p < 0.001), and Lachnoclostridium (r = 0.347, p < 0.001). N-tetracosenoyl-4-sphingenine was positively correlated with Blautia (r = 0.266, p = 0.007), unclassified_Anaerovoraceae (r = 0.251, p = 0.011), Romboutsia (r = 0.360, p < 0.001), and Lachnoclostridium (r = 0.304, p = 0.002). Sphingomyelin was positively correlated with unclassified_Anaerovoraceae (r = 0.209, p = 0.036) and Romboutsia (r = 0.227, p < 0.022).

Poultry production led by empirical evidence has traditionally set differing production systems based on gender. Sexual dimorphism was also associated with a variety of traits in chickens including growth rate, slaughter performance (increased muscle and decreased abdominal fat), feed conversion efficiency, and mineral utilization ability (Rose et al., 1996; Lumpkins et al., 2008; Cui et al., 2021; Tumova et al., 2021). Feed additives have demonstrated differential effects on animal health or production performance by gender (Han and Baker, 1993; Zhang D. et al., 2021). Therefore, gender, production performance, and nutrient utilization are linked and have important practical significance. However, the mechanisms that mediate these associations were poorly understood. A large body of data indicated clear contributions of gut microbiota or metabolites to these processes. We, therefore, conducted the current study to provide a research basis for sexual dimorphism in chicken production performance from the perspective of gut microbiota and metabolomics. We explored the differences in cecal microbiota and serum metabolome between male and female chickens and initially established the relationships between sex, the cecal microbiota, and serum metabolites.

Gender is an important factor affecting animal gut microbiota diversity (Zhang X. et al., 2021), and higher microbial diversity in female chickens has been demonstrated in studies of humans (Falony et al., 2016), pigs (Zhang D. et al., 2021), mice (Kozik et al., 2017), and chickens (Cui et al., 2021). In contrast, we found that cecal microbiota diversity did not differ significantly between male and female chickens. However, our network analysis indicated that sex could have a significant selection effect on cecal microbial interactions, and female chickens possessed a more complex and stable microbiome (Wang et al., 2021). Greater complexity and stability translate into greater resistance to the external environment (Coyte et al., 2015), and this implicates sexual traits with a significant selection effect on the microbiome. One possible reason for these effects is due to longer colon transit times in female chickens (Degen and Phillips, 1996; Graff et al., 2001), and longer colon transit times were associated with higher gut microbial complexity (Roager et al., 2016). Interestingly, previous studies in humans and other mammals have not consistently shown this (Haro et al., 2016; Gao et al., 2018; Takagi et al., 2019). However, reports of sex bias in the chicken gut microbiome have been inconsistent and may have been due to noise introduced by factors such as breed, diet, and feeding mode (Lee et al., 2017).

Prior chicken studies have reported the presence of characteristic microbiota for male and female chickens. Unlike these results of previous research results (Lee et al., 2017; Cui et al., 2021), we found that the abundance of Lactobacillus and Family _XIII_UCG-001 was significantly higher in the cecal microbiota of male chickens, while Eubacterium_nodatum_group, Blautia, unclassified_Anaerovoraceae, Romboutsia, Lachnoclostridium, and norank_Muribaculaceae were more abundant in female chickens. Multiple previous studies have also confirmed that Lactobacillus is an important gender-differentiated microbe and its abundance in male chickens was significantly higher than for female chickens (Sha et al., 2013; Poutahidis et al., 2014; Sherman et al., 2018; He et al., 2019) and enrichment for Lactobacillus salivarius, Lactobacillus crispatus, and Lactobacillus aviaries was identified in male chickens (Torok et al., 2013). These studies indicated that Lactobacillus may be an important gut microbe relevant to male chickens. The presence of Lactobacillus in human or animal males has also been linked to inhibition of inflammation and increased testosterone levels (Poutahidis et al., 2014). Dietary supplementation with probiotic Lactobacillus reuteri could prevent age-related testicular atrophy in mice and male hypogonadism in humans (Cross et al., 2018). Conversely, studies in mice have reported a different pattern where females exhibited a more abundant presence of Lactobacillus in their bacterial community, which was linked to their stronger immunological response to pathogenic bacteria infection than males (Ge et al., 2006). We found that Blautia was enriched in the ceca of female chickens and was consistent with human studies that linked Blautia with adult women (Markle et al., 2013). Eubacterium in pigs has also been reported to be sex-differentiated (He et al., 2019). However, unclassified_Anaerovoraceae, Romboutsia, Lachnoclostridium, and norank_Muribaculaceae (this study) have not been previously reported.

Metabolome analysis was carried out in the current study, revealing clear differences in serum metabolites between male and female chickens. Three sex hormones were enriched in male chicken serum (prostaglandin F2α and its alcohol methyl ether isomer and 11β-hydroxytestosterone), while no characteristic steroid hormone metabolites were identified in female chickens. Male chickens in this study were 18 weeks old and had reached sexual maturity, and this is most likely the reason for these results. In addition, we identified other serum metabolites showing significant differences by gender and included three sphingomyelin metabolites (N-tetracosenoyl-4-sphingenine, sphingomyelin, and N-[1, 3-dihydroxyoctadec-4-en-2-yl] tetracos-15-enamide) that were enriched in female chickens. Metabolite pathway enrichment analysis also verified that the sphingolipid metabolism pathway was the primary metabolic pathway enriched in female chicken, sera and sphingomyelin is essential for egg yolk formation (Yang et al., 2018). Sex hormones are the primary causes of sexual dimorphism in animals (Degen and Phillips, 1996; Org et al., 2016; Lee et al., 2017). Chickens in our study were already mature sexually, and some began to lay eggs. Female chickens began to shift from growth and development to reproduction and gradually prepared for the formation of follicles (Kim et al., 2020), and this might be the reason for the elevated levels of sphingomyelin metabolites we found. In addition, linoleic acid was enriched in male sera, and this result was also verified in metabolic pathway analysis. Linoleic acid could change the structure of cumulus granulosa cell membranes and interfere with gonadotropins thereby inhibiting oocyte maturation (Prades et al., 2003). Our female chickens were physiologically prepared for laying eggs with associated elevated estrogen levels, and this could exert a negative (inhibitory) effect on linoleic acid production.

Since the gut microbiota was involved in the excretion and circulation of sex hormones, researchers have proposed the concept of a microgenderome to characterize the microbiome associated with sex hormone metabolism (Yoon and Kim, 2021). In our study, we identified numerous significant pairs and intensity of correlation, and some serum metabolites displayed strong correlations with the cecal microbiomes. A correlation analysis was conducted between gender-differential sex hormone metabolites in serum and gender-differential microbiota, and Blautia and 11β-hydroxytestosterone displayed a significant negative correlation. Blautia has been reported as an important characteristic genus in women (Markle et al., 2013). Notably, we also found that three sphingomyelin metabolites were positively correlated with Blautia, unclassified_Anaerovoraceae, Romboutsia, and Lachnoclostridium. These associations suggested that the cecal microbiota may be involved in the synthesis or metabolism of these sphingolipid metabolites or that these metabolites had a role in regulating the composition of the chicken cecal microbiome. Humans possess numerous sphingolipid-producing bacteria including Bacteroides and Parabacteroides (Gupta and Lorenzini, 2007; Heaver et al., 2018). Sphingomyelin was also linked to the regulation of the abundance of specific microbes in the intestine. For example, 1% purified glycosylceramide (N-[1,3-dihydroxyoctadec-4-en-2-yl]tetracos-15-enamide, this study) fed continuously to mice for 1 week increased the abundance of Blautia coccoides, and this species could degrade glycosylceramide into ceramide that female chickens can metabolize into fatty acids and sphingosine that could be absorbed by the intestine and exert beneficial effects on the host (Hamajima et al., 2016). In addition, unclassified_Anaerovoraceae, Romboutsia, and Lachnoclostridium have not been previously reported to be linked with host serum sphingomyelin metabolism (Hannun and Obeid, 2018; Heaver et al., 2018). Together with our study, these reports point to an important causal relationship between cecal microbes and sphingomyelin metabolites, and this extends to the female chicken microbiome.