Background

Myostatin (MSTN), also known as growth differentiation factor 8 (GDF-8), is highly expressed in skeletal muscle tissue and negatively regulates the development of muscles [1]. Inactivation of MSTN expression causes muscle hyperplasia and hypertrophy, which has been observed in mice [1], cattle [2,3,4], dogs [5, 6], sheep [7], goats [8, 9], pigs [18]. A related study revealed oral administration of fecal samples from healthy calves increased the abundance of Porphyromonadaceae, relieved diarrhea, and improved the overall growth of those with diarrhea [19]. These findings suggest that the gut microbiome has a synergistic effect on host growth and health. Previous studies have revealed a positive change of microbiota in pigs and cattle following mutation of the MSTN [20, 21]. However, the effect of MSTN gene editing on the composition and function of gut microbiota in sheep is lacking.

In the present study, we used metagenomic sequencing to investigate the MSTN editing effects on gut microbiota composition and function of sheep, and found that MSTN inactivation improved the abundance of beneficial gut microbiota in sheep. The findings in this study provide new insights into the relationship between MSTN inactivation and the changes of gut microbiota in sheep.

Results

Genotype identification

We sequenced the second and third exons of MSTN-edited female sheep (GEF), MSTN-edited male sheep (GEM), wild-type female sheep (WTF) and wild-type male sheep (WTM). The GEF and GEM groups contained 3 bp (c.469_471delTGG) deletions in the second exon and 11 bp (c.879_889delTGGATGGGATT) deletions in the third exon, but the WTF and WTM groups harbored complete base sequence (Fig. 1; Supplementary Fig. 1).

Fig. 1
figure 1

A MSTN edited sites in the second and third exons. B The sequences of the second and third exons of the MSTN based on deep sequencing

Full size image

Slaughter parameters and meat quality analysis

Three MSTN-edited male sheep and three wild-type male sheep were slaughtered and slaughter parameters and mutton quality were tested. The results showed that the body weight (P = 0.016), carcass weight (P = 0.005) and eye muscle area (P = 0.003) of MSTN-edited sheep were significantly higher than wild-type sheep (Table 1). However, there were no significant differences in the meat quality indexes (Table 2).

Table 1 Effect of MSTN editing on slaughter parameters of gene-edited and wild-type sheep
Full size table
Table 2 Effect of MSTN editing on meat quality as assessed on the longissimus dorsi of gene-edited and wild-type sheep
Full size table

Alpha and beta diversity of the gene-edited and wild-type sheep

The gut microbiota of the four groups (GEF, WTF, GEM and WTM) were analysed. Compared to the WTF, the Simpson index of the GEF was significantly higher (P = 0.039), and the Chao and Shannon index showed no significant difference (P = 0.383, P = 0.148). Also, compared to the WTM, the Simpson index of the GEM was significantly higher (P = 0.008), the Shannon index was significantly lower (P = 0.016), and the Chao index showed no significant difference (P = 0.999) (Fig. 2A). Principal coordinates analysis (PCoA) revealed the GEF and WTF were distinct. Analysis of Similarities (ANOSIM) indicated the differences between the GEF and WTF were greater than that within their own (Fig. 2B, R = 0.283, P = 0.005). Comparable findings were observed between the GEM and WTM (Supplementary Fig. 2A, R = 0.254, P = 0.011).

Fig. 2
figure 2

The composition of gut microbiota in the MSTN-edited and wild-type groups. A The statistical graphs of alpha diversity of gut microbiota among the GEF, WTF, GEM and WTM. B Bray–Curtis-based PCoA of the gut microbiota in the GEF and WTF. The P-value was based on ANOSIM. The boxplot shows the discrete distribution of samples along the PC1 and PC2 axes. C The significantly different phyla between the GEF and WTF. D The histogram of the dominant genera in the GEF, WTF, GEM and WTM. E The significantly different genera between the GEF and WTF. F The significantly different species between the GEF and WTF. (0.01 < P ≤ 0.05 *, 0.001 < P ≤ 0.01 **). G LEfSe analysis of the GEF and WTF. The histogram shows the microbes that can best illustrate the difference between the GEF and WTF. The larger the LDA score, the greater the contribution of the corresponding microbe to the difference

Full size image

Composition of gut microbiota of gene-edited and wild-type sheep

We compared the difference of the abundance of gut microbiota between the GEF and WTF, as well as the GEM and WTM. We observed the MSTN editing did not change the dominant phyla of sheep, the top 7 phyla were Firmicutes, Bacteroidota, Spirochaetes, Verrucomicrobia, Proteobacteria, Euryarchaeota and Lentisphaerae (Supplementary Fig. 2B). However, the Wilcoxon rank-sum test revealed that compared to the WTF, the abundance of Firmicutes (P = 0.005 in female sheep, P = 0.041 in male sheep), Chloroflexi (P = 0.04 in female sheep, P = 0.007 in male sheep), Synergistetes (P = 0.014 in female sheep, P = 0.01 in male sheep) and Thaumarchaeota (P = 0.01 in female sheep, P = 0.007 in male sheep) was significantly higher but that of Bacteroidota (P = 0.01 in female sheep, P = 0.041 in male sheep) was significantly lower in the GEF (Fig. 2C). The comparison between the GEM and WTM revealed similar findings (Supplementary Fig. 2C). Further analyses revealed that Firmicutes/Bacteroidota (F/B) ratio was significantly higher in the GEF (P = 0.016), but there was no statistical difference in the F/B ratio between the GEM and WTM groups (P = 0.054). At the genus level, the top 9 dominant genera across the 4 groups were Bacteroides, Alistipes, Clostridium, Ruminococcus, Prevotella, Treponema, Phocaeicola, Oscillibacter and Akkermansia (Fig. 2D). Compared to the WTF, the abundance of Flavonifractor (P = 0.014 in female sheep, P = 0.007 in male sheep), Pseudoflavonifractor (P = 0.014 in female sheep, P = 0.004 in male sheep), Intestinimonas (P = 0.031 in female sheep, P = 0.003 in male sheep), Subdoligranulum (P = 0.031 in female sheep, P = 0.01 in male sheep), Anaerotignum (P = 0.041 in female sheep, P = 0.031 in male sheep), Evtepia (P = 0.007 in female sheep, P = 0.014 in male sheep), Hungatella (P = 0.004 in female sheep, P = 0.041 in male sheep), Streptococcus (P = 0.01 in female sheep, P = 0.004 in male sheep), Ruthenibacterium (P = 0.041 in female sheep, P = 0.007 in male sheep), Agathobaculum (P = 0.018 in female sheep, P = 0.024 in male sheep), Oribacterium (P = 0.004 in female sheep, P = 0.014 in male sheep), Lactobacillus (P = 0.018 in female sheep, P = 0.01 in male sheep), Aminipila (P = 0.004 in female sheep, P = 0.005 in male sheep) and Colidextribacter (P = 0.018 in female sheep, P = 0.005 in male sheep) was significantly higher, whereas that of Alloprevotella (P = 0.007 in female sheep, P = 0.031 in male sheep) was significantly lower in the GEF (Fig. 2E). Comparable results were observed between the GEM and WTM (Supplementary Fig. 2D). At the species level, the top 10 most abundant species across the four groups included Clostridiales_bacterium, Ruminococcaceae_bacterium, Bacteroidales_bacterium, Lachnospiraceae_bacterium, Clostridia_bacterium, Clostridiaceae_bacterium, Rikenellaceae_bacterium, Firmicutes_bacterium_CAG:110, Alistipes_sp._Z76 and Treponema_porcinum (Supplementary Fig. 2E). The abundance of Clostridiales_bacterium (P = 0.031 in female sheep, P = 0.014 in male sheep), Oscillibacter_sp. (P = 0.007 in female sheep, P = 0.041 in male sheep), Clostridiales_bacterium_Marseille-P2846 (P = 0.024 in female sheep, P = 0.031 in male sheep), Eubacteriaceae_bacterium (P = 0.003 in female sheep, P = 0.01 in male sheep), Firmicutes_bacterium_CAG:170 (P = 0.014 in female sheep, P = 0.004 in male sheep), Firmicutes_bacterium_Adur. Bin467 (P = 0.01 in female sheep, P = 0.005 in male sheep), Firmicutes_bacterium_CAG:124 (P = 0.018 in female sheep, P = 0.003 in male sheep), Evtepia_gabavorous (P = 0.007 in female sheep, P = 0.014 in male sheep), Flavonifractor_plautii (P = 0.024 in female sheep, P = 0.003 in male sheep) and Hungatella_hathewayi (P = 0.007 in female sheep, P = 0.024 in male sheep) were higher both in the GEF and GEM groups, whereas those of Phocaeicola_vulgatus (P = 0.031 in female sheep, P = 0.041 in male sheep) and Bacteroidia_bacterium (P = 0.024 in female sheep, P = 0.041 in male sheep) were lower both in the GEF and GEM groups (Fig. 2F; Supplementary Fig. 2F).

The LEfSe at LDA > 3 revealed that 8 bacterial taxa including Firmicutes, Clostridia, Eubacteriales and Clostirdiales_bacterium were significantly more abundant in the GEF, whereas Bacteroidota, Bacteroidia, Bacteroidales and Prevotellaceae were more abundant in the WTF (Fig. 2G). A comparison of the GEM and WTM revealed comparable findings (Supplementary Fig. 2G).

In addition, we considered the top 30 species of the four groups for correlation network analysis. In general, MSTN mutation had no significant effect on the interaction of gut microbes (Fig. 3).

Fig. 3
figure 3

The microbial correlation networks of the (A) MSTN-edited groups (including GEF and GEM) and (B) wild-type groups (including WTF and WTM). The nodes colored differently represent different phyla; the size of nodes corresponds with the species abundance, the thickness of lines corresponds with the strength of association, the number of lines on the nodes corresponds with the relatedness of the species, the red color represents positive correlations whereas the green color represents negative correlations

Full size image

MSTN mutation altered potential function of hindgut microbiota based on metagenomic sequencing

Given that MSTN mutation influences the abundance of the gut microbiota, we performed KEGG analysis to investigate whether the mutation had an effect on the function of the related biological pathways. PcoA revealed a clear distinction between the GEF and WTF. ANOSIM analysis showed that the differences between the two groups were greater than that within their own (Fig. 4A, R = 0.180, P = 0.032). The comparison between the GEM and WTM revealed similar results (Supplementary Fig. 3A, R = 0.199, P = 0.026). Besides, we found a significant raising of methane metabolism (P = 0.024 in female sheep, P = 0.024 in male sheep) and a falling of fatty acid degradation (P = 0.014 in female sheep, P = 0.041 in male sheep) in GEF and GEM (Fig. 4-B, Supplementary Fig. 3-B). At level 3, the top 5 dominant pathways of the 4 groups were metabolic pathways, biosynthesis of secondary metabolites, microbial metabolism in diverse environments, biosynthesis of amino acids and carbon metabolism, indicating that the MSTN mutation did not influence the dominance of main pathways. Besides, we dug a total of 24 significantly affected pathways, 11 of them were upregulated in the GEF and GEM, including ABC transporters (P = 0.005 in female sheep, P = 0.041 in male sheep), sulfur relay system (P = 0.001 in female sheep, P = 0.01 in male sheep), benzoate degradation (P = 0.005 in female sheep, P = 0.01 in male sheep), synthesis and degradation of ketone bodies (P = 0.024 in female sheep, P = 0.01 in male sheep), styrene degradation (P = 0.003 in female sheep, P = 0.014 in male sheep), chlorocyclohexane and chlorobenzene degradation (P = 0.024 in female sheep, P = 0.002 in male sheep), D-arginine and D-ornithine metabolism (P = 0.01 in female sheep, P = 0.018 in male sheep), autophagy-yeast (P = 0.003 in female sheep, P = 0.01 in male sheep), carotenoid biosynthesis (P = 0.014 in female sheep, P = 0.018 in male sheep), biosynthesis of various secondary metabolites-part3 (P = 0.031 in female sheep, P = 0.004 in male sheep) and fluorobenzoate degradation (P = 0.031 in female sheep, P = 0.041 in male sheep). Another 13 pathways including RNA degradation (P = 0.031 in female sheep, P = 0.004 in male sheep), nicotinate and nicotinamide metabolism (P = 0.018 in female sheep, P = 0.014 in male sheep), peroxisome (P = 0.004 in female sheep, P = 0.031 in male sheep), adipocytokine signaling pathway (P = 0.003 in female sheep, P = 0.031 in male sheep), ferroptosis (P = 0.007 in female sheep, P = 0.031 in male sheep), thermogenesis (P = 0.005 in female sheep, P = 0.005 in male sheep), polyketide sugar unit biosynthesis (P = 0.041 in female sheep, P = 0.041 in male sheep), ubiquinone and terpenoid-quinone biosynthesis (P = 0.018 in female sheep, P = 0.018 in male sheep), apoptosis (P = 0.003 in female sheep, P = 0.018 in male sheep), longevity regulating pathway (P = 0.018 in female sheep, P = 0.005 in male sheep), N-glycan biosynthesis (P = 0.018 in female sheep, P = 0.018 in male sheep), MAPK signaling pathway–fly (P = 0.014 in female sheep, P = 0.024 in male sheep) and biosynthesis of siderophore group nonribosomal peptides (P = 0.024 in female sheep, P = 0.041 in male sheep) were downregulated in the GEF and GEM (Fig. 4C, Supplementary Fig. 3C).

Fig. 4
figure 4

Upregulated and downregulated signaling pathways and CAZymes in the MSTN-edited and wild-type groups. A Bray–Curtis-based PCoA of the functional pathways of the GEF and WTF. The P-value was based on ANOSIM. The boxplot for the discrete distribution of samples along the PC1 and PC2 axes. B Significantly changed pathways of the GEF and WTF at level 1 (metabolism pathways) and (C) level 3. The vertical axis represents different pathways, and the horizontal axis represents the proportions of corresponding pathways. (0.01 < P ≤ 0.05 *, 0.001 < P ≤ 0.01 **, P ≤ 0.001 ***). D Bray–Curtis-based PCoA of CAZymes of the GEF and WTF. The P-value is based on the ANOSIM. The boxplot shows the discrete distribution of samples along the PC1 and PC2 axes. E The significantly different CAZymes of gut microbiome between the GEF/GEM and WTF/WTM. The vertical axis represents different CAZymes, and the horizontal axis represents the proportions of corresponding enzymes. (0.01 < P ≤ 0.05 *, 0.001 < P ≤ 0.01 **)

Full size image

MSTN mutation altered the metabolic activities of CAZymes

Changes of the abundance of CAZymes can indirectly reflect the alterations of gut microbiota. According to PcoA, we observed a clear distinction of CAZymes between the GEF and WTF. ANOSIM analysis revealed that the differences between these two groups were greater than that within their own (Fig. 4D, R = 0.190, P = 0.035). Similar results were certified in the GEM and WTM. (Supplementary Fig. 3D, R = 0.197, P = 0.034). We found a total of 7 significant changed enzymes, and the abundance of the GH13_39 (P = 0.01 in female sheep, P = 0.007 in male sheep), GH4 (P = 0.014 in female sheep, P = 0.041 in male sheep), GH137 (P = 0.041 in female sheep, P = 0.041 in male sheep), GH71 (P = 0.018 in female sheep, P = 0.007 in male sheep) and PL17 (P = 0.018 in female sheep, P = 0.004 in male sheep) were upregulated, the GT41 (P = 0.004 in female sheep, P = 0.031 in male sheep) and CBM20 (P = 0.024 in female sheep, P = 0.01 in male sheep) were downregulated (Fig. 4E).

Discussion

Whether MSTN-edited sheep will degrade their meat quality and taste is an important question. In this study, we analyzed the meat color, pH, water loss rate and cooking loss and proved that there was no difference in meat quality indices between the MSTN-edited sheep and the wild-type sheep.

Gender affects the composition of microbes, and sex hormones are one of the important reasons for this difference. A study analyzed the 16S rRNA dataset of the human microbiome project (HMP) and found that male stool tended to have a higher abundance of Prevotella and Bacteroides [22]. Another study with a larger individual showed significantly different of gut microbiome composition of female and male, and female group harbored higher α diversity [23]. In animal studies, α diversity of cecal microbiota in 10-week-old female mice was significantly higher than that in 10-week-old male mice. PCA analysis showed that female and castrated male microbiota were closer to each other than to male microbiota [24]. To eliminate the potential influence of gender on the results, we sequenced the gut microbiome of MSTN-edited male sheep, wild-type male sheep, MSTN-edited female sheep and wild-type female sheep, and selected the microbes with the same change trend in different genders for further analysis. The results revealed that MSTN inactivation had no significant effect on the key dominant gut microbiota in sheep, consistent with previous research in pigs [20, Genotype identification

These experimental sheep were the offspring of MSTN± male sheep carrying 3 bp (c.469_471delTGG) deletions in the second exon and 11 bp (c.879_889delTGGATGGGATT) deletions in the third exon of the MSTN and wild-type females. In order to ensure the accuracy of the sheep grou**, we verified the genotypes by sequencing. Total DNA was extracted using the Blood Genomic DNA Mini Kit (CoWin Biosciences, Shanghai, China). The MSTN gene was amplified using the 2 × Rapid Taq Master MIX (Vazyme Biotech Co., Ltd., Nan**g, China). The PCR products were separated and assessed on 1.0% agarose gel. Sanger sequencing was performed at Sangon Biotech Co., Ltd. (Shanghai, China). Deep sequencing was performed at Novogene Co., Ltd. (Tian**, China).

Determination of slaughter parameters and meat quality

The experimental sheep were slaughtered on the same day by following the Halal slaughtering procedure, which involved exsanguination without electrical stunning. After skinning and removing heads, hooves and internal organs, the carcasses were weighted and the slaughter rate (carcass weight/body weight) were calculated. Finally, longissimus dorsi muscles were isolated for meat color, pH, water loss rate, cooking loss, intramuscular fat and eye muscle area measuring. Meat quality analysis was performed as previously described [54]. Briefly, the meat color (L*, a* and b*) and pH values at 45 min and 24 h in longissimus dorsi were measured by using a Minolta CM-600D spectrophotometer (Konica Minolta Sensing Inc., Osaka, Japan) and a pH–STAT meter (SFK-Technology, Denmark), respectively. For the water loss rate, a circular sampler was used to cut the longissimus dorsi about 20 g and an analytical balance MS1602S (Thermo Fisher Scientific, Waltham, MA, USA) was used to weight the pre-pressure weight. Then the sample was sandwiched into the multi-layer filter papers, pressed 35 kg with water loss rate tester Bulader-M10 (Bulader, Bei**g, China) for 5 min and reweighted the sample. Water loss rate was expressed as a percentage of weight change before and after pressure. For the cooking loss analysis, 100 g of muscle was weighed and cooked in a thermostat water bath DK-S22 (Shanghai **g Hong Laboratory Instrument Co., Ltd., Shanghai, China) at 85 ℃ for 50 min. After that, the sample was removed, blotted dry on filter paper, and reweighted. Cooking loss was expressed as a percentage of weight change before and after cooking. In addition, intramuscular fat content was determined by the soxhlet extraction method. Longissimus dorsi were lyophilized to a constant weight (no evaporable moisture remaining) using vacuum freeze dryer (SCIENTZ-18ND, Ningbo, Zhejiang, China), and then subjected to ether extraction using the Soxhlet apparatus and diethyl ether. After 8 h of extraction, the samples were removed, air-dried and reweighed to determine fat loss. Moreover, the eye muscle area was measured by pasting a sulfate paper on the cross section of the muscle and measuring the area of the outline of the muscle. The data are expressed as the mean ± SEM and analyzed using Student’s t-test with significant differences considered at P < 0.05.

Metagenomic sequencing, assembly, and construction of the gene library

To construct the paired-end (PE) library, DNA extracted from the GEF, WTF, GEM and WTM groups were fragmented into small 300 bp segments using the Covaris M220 machine (Gene, Hong Kong, China). The PE library was then constructed using the TruSeq DNA Sample Prep Kit (Illumina, San Diego, CA, USA), following the manufacturer’s instructions. Sequencing of the PE reads was performed using the Illumina HiSeq 4000 platform (Illumina, San Diego, CA, USA). A total of 2,269,248, 212 raw reads were obtained. The minimum number of raw reads per group was 61,035,344 but averaged 70,914,007. Adapter sequences were removed from the 3’ and 5’ ends of the raw reads. Low-quality raw reads (length < 50 bp, quality values < 20, or containing N bases) were also removed using the FastP software (Version 0.20.0) [55]. A total of 2,235,699,610 clean reads were retained (98.52% of raw reads). The minimum number of clean reads in a given sample was 60,093,748 but averaged 69,865,613. After removing sequences for the Ovis aries genome (assembly ARS-UI_Ramb_v2.0) sequences with BWA (version 0.7.9a) [56]. A total of 1,841,359,674 high quality reads, representing 81.17% of total raw reads, were obtained. A minimum of 40,225,450, reads, but an averagely of 57,542,490 per sample, were cleaned. Contigs were generated using the Megahit software (Version 1.1.2) [57]. Only 25,233,472 contigs larger than 300 bp were retained. The minimum number of contigs per sample was 561,569 but averaged 788,546. The average size of the contigs was 208,436 bp. The maximum N50 and N90 were 705 bp and 357 bp, the minimum N50, and N90 were 547 bp and 338 bp, and the average N50 and N90 were 622 bp and 347 bp. The open reading frames (ORFs) in contigs were predicted using the MetaGene software [58]. A total of 57,239,602 ORFs larger than 100 bp, which were translated into amino acid sequences at the National Center for Biotechnology Information (NCBI), were retained. The maximum ORFs ranged between 100 bp and 27,414 bp but averaged 440 bp. All predicted genes with a 90% sequence identity (90% coverage) were clustered into groups using the CD-HIT software (version 4.6.1) [59]. The final non-redundant gene sets contained 25,836,001 ORFs, which was 45.14% of the original ORFs. The average length of ORFs was 497 bp. To evaluate gene abundances in each sample, quality ORFs within each sample were mapped to the corresponding sequence with 95% identity using SOAP aligner (Version 2.21) [60]. For taxonomic annotations, selected non-redundant gene sets were blasted in the NCBI-NR database using the DIAMOND software (parameter: BLASTP; E-value ≤ 1e−5) [61, 62]. A total of 5 domains, 13 kingdoms, 235 phyla, 449 classes, 895 orders, 1,779 families, 5,369 genera and 35,877 species were annotated. Similarly, for KEGG pathway analysis [63], non-redundant gene sets were blasted in the GENES using the DIAMOND software (parameter: BLASTP; E-value ≤ 1e−5), and for CAZymes, non-redundant gene sets were blasted in the CAZy database (E-value ≤ 1e−5) using the HMMSCAN [64].

Statistical analysis

The difference in gene expression between the GEF and WTF groups, GEM and WTM groups were analyzed using the Wilcoxon rank-sum test at two-tailed P-values. The gene expression data were presented using box-and-whisker plots. Alpha and beta diversity indices (Bray–Curtis) were calculated using the Mothur software (Version 1.30.2) and R (Version 3.3.1). The differences in the bacterial taxa between different groups were assessed using Analysis of similarity (ANOSIM) of Bray–Curtis distances. Data analysis and FDR correction of the p-values were performed using the R package (Version 3.3.3). The operational taxonomic units most likely to explain differences between classes by coupling standard tests for statistical significance with additional tests encoding biological consistency and effect relevance were identified based on Linear discriminant analysis (LDA) effect size (LEfSe). The correlation network analysis was performed using the Networks software (Version 2.1).