**Background:** Producing omega-3-enriched milk by feeding linseed to dairy cows often causes milk fat depression (MFD), leading to economic penalties. Previous work showed wide variation in individual cow sensitivity to MFD on linseed-rich diets. The authors hypothesized that some cows may be genetically resistant to MFD. This study aimed to identify differentially expressed genes, metabolic pathways, transcription factors, and SNPs that distinguish resistant (R-MFD) from sensitive (S-MFD) cows.
**Methods:** The experiment was conducted on a commercial dairy farm in Catalonia, Spain, with 800 Holstein cows. Cows were initially fed a diet containing 6.1% extruded linseed (LIN), then switched to a control diet (CTR) with no linseed. Milk samples were collected monthly for four months before and after the diet change. Two R-MFD cows (high milk fat on both CTR [4.06%] and LIN [3.90%]) and two S-MFD cows (high fat on CTR [3.87%] but low on LIN [2.52%]) were selected. mRNA was extracted from milk somatic cells, which are representative of the mammary gland transcriptome. RNA-sequencing was performed on an Illumina HiSeq2000, yielding 100 bp paired-end reads. Reads were mapped to the Bos taurus genome (version 4.6.1). Differential expression analysis used Cuffdiff with criteria: RPKM ≥ 0.2, fold change >2 or <−2, and p-value < 0.01. Ingenuity Pathway Analysis identified metabolic pathways and key gene regulators. SNP detection required minimum average quality of 15, central base quality of 20, coverage ≥10 reads, and variant frequency ≥20%.
**Key Results:** An average of 72 million sequence reads per sample were obtained, with 80–85% mapping to the bovine reference genome. Approximately 90% of annotated Bos taurus genes (24,881 of 27,368) were detected. Differential expression analysis between R-MFD and S-MFD cows identified 1,316 differentially expressed genes (DEGs) on the CTR diet and 1,888 DEGs on the LIN diet. Comparing LIN vs. CTR diets within groups, 816 DEGs were found in R-MFD cows but only 43 in S-MFD cows (all over-expressed on LIN). DEGs were linked to 13–117 metabolic pathways and 27–294 key gene regulators, involving immune/inflammatory systems, development/growth, and lipid metabolism/FA synthesis. Between 25,000 and 34,000 polymorphic SNPs were detected per cow. In R-MFD cows, 6,700–7,300 polymorphic SNPs were associated with DEGs and key gene regulators; in S-MFD cows, 6,900–8,900. Of these, 641 SNPs were unique to R-MFD cows and 1,024 unique to S-MFD cows. In R-MFD cows, 63% of SNPs were intron-variant, 12% synonymous codon, 7% utr-variant-3-prime, and 15% other types. In S-MFD cows, 65% were intron-variant, 13% synonymous codon, 9% utr-variant-3-prime, and 14% other types (including splice-acceptor-variant and stop-gained). Fifteen genes met all three criteria (differentially expressed, key gene regulator, and harboring SNP) in the comparison of R-MFD vs. S-MFD cows on the LIN diet: APBB1, CD38, EREG, FLT1, ITGB4, MTOR, NFATC2, NOTCH1, PDPK1, PROM1, RICTOR, TGFBR3, WWC1, and ZNF217. Most of these genes were also differentially expressed in the CTR diet comparison. MTOR was uniquely down-expressed in R-MFD cows on LIN but over-expressed on CTR compared to S-MFD cows. MTOR, NFATC2, and PDPK1 were down-expressed in R-MFD vs. S-MFD on LIN and also down-regulated in LIN vs. CTR within R-MFD cows, suggesting both a constitutive difference and a linseed response.
**Clinical Implications:** The 15 identified genes, particularly those in the PI3K/Akt/MTOR/SREBP1 signaling axis (MTOR, PDPK1, RICTOR, EREG, NOTCH1, ZNF217, TGFB3), represent novel candidate markers for genetic selection of cows resistant to MFD. Selecting for MFD resistance could enable production of omega-3-enriched milk without the economic penalties of reduced milk fat content. The finding that R-MFD cows showed many more DEGs (816) in response to linseed compared to S-MFD cows (43) suggests resistant cows activate compensatory mechanisms to maintain fat synthesis. The small sample size (n=4) is a limitation, but the authors argue that high fold-change thresholds (>2) and the use of each cow as its own control provide adequate statistical power for detecting large-effect genes. Validation in larger populations is needed before these markers can be used in breeding programs.