Bone morphogenetic protein 2 (BMP2) has been reported to regulate adipogenesis, but its role in porcine beige adipocyte formation remains unclear. Our data reveal that BMP2 is significantly induced at the early stages of porcine beige adipocyte differentiation. Additionally, supplementing rhBMP2 during the early stages, but not the late stages of differentiation, significantly enhances porcine SVF adipogenesis, thermogenesis, and proliferation. Furthermore, compared to the empty plasmid-transfected-SVFs, BMP2-overexpressed SVFs had the enhanced lipid accumulation and thermogenesis, while knockdown of BMP2 in SVFs exhibited the opposite effect. The RNA-seq of the above three types of cells revealed the enrichment of the annotation of thermogenesis, brown cell differentiation, etc. In addition, the analysis also highlights the significant enrichment of cell adhesion, the MAPK cascade, and PPARγ signaling. Mechanistically, BMP2 positively regulates the adipogenic and thermogenic capacities of porcine beige adipocytes by activating PPARγ expression through AKT/mTOR and MAPK signaling pathways.

Genome-wide association study (GWAS) analyses have identified numerous loci associated with economic traits in cattle. Many of these loci reside in noncoding regions, and the regulatory mechanisms through which they influence complex traits remain poorly understood. Here, we integrated 657 RNA-seq libraries from 275 Huaxi cattle across three tissues (longissimus dorsi muscle, liver, and subcutaneous backfat) with ∼ 10 million imputed SNP genotypes to systematically map cis-molecular quantitative trait loci (cis-molQTLs) across four transcriptomic regulatory layers: gene expression (eQTLs), splicing (sQTLs), alternative polyadenylation (aQTLs), and RNA editing (edQTLs). These cis-molQTL classes display distinct genomic distributions and functional enrichments, yet operate in a coordinated manner within complex trait regulatory networks and are significantly enriched near GWAS- and QTLdb-reported loci for growth, carcass, and meat quality traits. Using 1788 genotyped and phenotyped Huaxi cattle, a GREML framework showed that these multi-layer cis-molQTL SNPs collectively explain 61.9% of total SNP-based heritability across 19 complex traits. Incorporating cis-molQTL annotations into genomic prediction models, including MultiBLUP, BayesRC, and molGBLUP, improved prediction accuracy for most traits relative to the baseline GBLUP model (mean increase of 0.05), highlighting the value of multi-layer regulatory variation for functionally informed genomic prediction and precision breeding.