Stage-Specific Genes Shape Milk Urea Nitrogen Across Lactation in Iranian Holsteins

Milk urea nitrogen, a routine byproduct measurement that reflects how efficiently a dairy cow converts dietary protein into milk rather than urinary waste, does not behave as a single, static genetic trait. A new study of Iranian Holstein cattle, published in Dairy Science and Management, shows that the genomic regions influencing this trait change substantially as lactation progresses, with only three single nucleotide polymorphisms overlapping among the top-ranked markers identified for early, middle and late lactation. The finding comes from the first application of a single-step genome-wide association study combined with random regression models to milk urea nitrogen in this population, an approach designed to track genetic effects day by day across the 305-day lactation rather than averaging them into one number.

The researchers drew on an unusually deep dataset: 347,639 test-day records from 52,219 first-parity Holstein cows raised in 297 herds, collected between 2018 and 2023 through the Iranian National Animal Breeding Center. Records were filtered to cows with at least three test days, days in milk between 5 and 305, milk urea nitrogen values between 7 and 26 milligrams per deciliter, and ages at first calving between 18 and 41 months, with outliers beyond three standard deviations excluded. On the genomic side, 2,189 bulls genotyped on panels ranging from 20K to 140K markers were processed; lower-density genotypes were imputed to the 50K panel using FImpute, and after quality control removing markers with minor allele frequency below 0.05, Hardy-Weinberg deviations, sex chromosome markers and excess missingness, 2,187 bulls and 41,102 SNPs remained for analysis.

The statistical engine of the study was the single-step genomic best linear unbiased prediction framework, which integrates phenotype, pedigree and genotype data in one model. Rather than replacing the pedigree relationship matrix outright, the method blends pedigree and genomic information through a combined matrix, with a blending parameter of 0.05 applied to keep the genomic matrix invertible. The pedigree extended up to 22 generations, anchoring relationship estimates in deep ancestral information. Fixed effects included herd-test-day and the age-year-season of calving, modeled with Legendre polynomials, while additive genetic and permanent environmental effects were each fitted with third-order Legendre polynomials, the order previously found optimal for milk urea nitrogen in this population.

What distinguishes this work from a conventional genome-wide association study is the treatment of time. Because milk urea nitrogen is expressed repeatedly throughout lactation, a variant with a strong effect in early lactation can appear diluted if its influence is averaged over the full 305 days. Random regression models allow SNP effects to be estimated as continuous functions of days in milk. The team converted genomic estimated breeding value solutions into SNP effects using POSTGSF90, then computed daily SNP effects from day 5 to day 305 through the Legendre polynomial covariates. The top 50 SNPs by absolute effect were selected independently within three non-overlapping stages: early lactation from days 5 to 105, middle lactation from days 106 to 205, and late lactation from days 206 to 305.

The temporal architecture that emerged was striking. Of the 150 stage-specific SNPs selected across the three windows, only three appeared in more than one stage, underscoring how thoroughly the genetic control of this trait shifts with the cow’s physiological state. One marker, on bovine chromosome 3 at position 17,160,057, was repeatedly significant and sits near PGLYRP3 and S100A9, genes implicated in immune regulation and nitrogen metabolism. By contrast, SNPs near CYP1A1 and ABCB11 were associated exclusively with late lactation, consistent with roles in detoxification and bile acid excretion that may become more prominent as the lactation cycle winds down.

Quantifying how much signal the top markers carry reinforced the picture of a highly polygenic trait. Using a sum-of-squares procedure, the researchers calculated that the top 150 SNPs explained on average 1.72 percent of the total SNP-effect sum of squares across days in milk, with a median of 1.67 percent. That modest figure is exactly what would be expected for a trait shaped by many variants of small effect, and it argues against any simple marker-based shortcut: improving nitrogen efficiency through genetics will require genome-wide information rather than a handful of diagnostic loci.

Functional annotation of the significant regions identified 17 positional candidate genes within 100 kilobases of the top SNPs, grouped into three functional categories. Immune regulation was represented by PGLYRP3, PGLYRP4 and S100A9, genes involved in mucosal immunity, inflammatory signaling and microbial defense; inflammatory cytokines can disrupt protein and amino acid turnover, potentially raising nitrogenous waste in milk. Xenobiotic metabolism and bile transport included ABCB11, which encodes the bile salt export pump, along with the cytochrome P450 genes CYP1A1 and CYP1A2, which catalyze oxidative conversion of nitrogenous substrates. Nutrient transport and energy metabolism contributed SLC5A11, a sodium-coupled glucose cotransporter whose activity may reduce reliance on protein catabolism for energy, plus PHKG1 and ABCA13.

Enrichment analysis of quantitative trait loci near the significant SNPs added a broader physiological context. The majority of enriched QTLs related to milk composition traits, including milk protein percentage, casein and kappa-casein content, and milk fat percentage, traits tightly connected to nitrogen partitioning since casein alone accounts for roughly 77 to 81 percent of total milk nitrogen. Additional QTLs overlapped health traits such as somatic cell count and susceptibility to bovine respiratory disease, reproductive traits including heifer pregnancy and interval to first estrus, and metabolic traits such as feed intake and lactation persistency. This pleiotropic overlap suggests that selection aimed solely at lowering milk urea nitrogen could carry unintended consequences for fertility, health or production, and the authors argue for multiple-trait genomic selection frameworks that weigh these correlated traits jointly.

The implications for breeding programs are concrete. Because milk urea nitrogen is moderately heritable, with reported estimates ranging from 0.15 to 0.30, and because lower concentrations are associated with reduced urinary nitrogen excretion, the trait is a plausible selection target for shrinking the nitrogen footprint of dairy farming, a sector whose manure contributes substantially to ammonia emissions; livestock are estimated to account for nearly 36 percent of ammonia emissions in Europe. Stage-specific SNP effects suggest that time-dependent genetic evaluations, rather than lactation averages, could target nitrogen efficiency at the physiological moments when intervention matters most. In Iranian Holsteins, where first-parity milk urea nitrogen typically ranges from 13 to 16 milligrams per deciliter, such evaluations could eventually be folded into existing genomic pipelines.

The authors are careful to flag limitations. The reliance on 50K SNP arrays may have missed rare alleles or regulatory variants, and future work with whole-genome sequencing or higher-density arrays, ideally with sliding-window analyses, could sharpen QTL mapping. Functional validation of the candidate genes is still required to confirm their roles in nitrogen metabolism, and integrating transcriptomic and metabolomic data across lactation stages would clarify mechanisms. The study also covered only first-lactation animals, which minimizes confounding but may not capture the genetic architecture of later lactations or of herds under different management. Whether the temporal patterns and candidate genes identified here hold across multiple parities, environments and populations remains an open question, and the authors note that findings should be validated against the newer ARS-UCD2.0 genome assembly, since the analysis used ARS-UCD1.2.

Subject of Research: Genome-wide association analysis of milk urea nitrogen across lactation stages in Iranian Holstein cattle

Article Title: Genome-wide association analysis based on random regression models for milk urea nitrogen in Iranian Holstein cattle

Article References: Mortazavi, M., Zandi, M. B., Pahlavan, R., Nasab, M. E., Mulim, H. A., & de Oliveira, H. R. (2025). Genome-wide association analysis based on random regression models for milk urea nitrogen in Iranian Holstein cattle. Dairy Science and Management, 2(1), Article 12. https://doi.org/10.1186/s44363-025-00015-9

Image Credits: AI Generated

DOI: 10.1186/s44363-025-00015-9

Keywords: milk urea nitrogen, Iranian Holstein cattle, single-step GWAS, random regression models, candidate genes, genomic selection, nitrogen use efficiency, lactation stages, QTL enrichment, dairy cattle genetics, Genome-wide, association