Association of a lysine-232/alanine polymorphism in a bovine gene encoding acyl-CoA:diacylglycerol acyltransferase (DGAT1) with variation at a quantitative trait locus for milk fat content

Abstract

DGAT1 encodes diacylglycerol O-acyltransferase (EC ), a microsomal enzyme that catalyzes the final step of triglyceride synthesis. It became a functional candidate gene for lactation traits after studies indicated that mice lacking both copies of DGAT1 are completely devoid of milk secretion, most likely because of deficient triglyceride synthesis in the mammary gland. Our mapping studies placed DGAT1 close to the region of a quantitative trait locus (QTL) on bovine chromosome 14 for variation in fat content of milk. Sequencing of DGAT1 from pooled DNA revealed significant frequency shifts at several variable positions between groups of animals with high and low breeding values for milk fat content in different breeds (Holstein-Friesian, Fleckvieh, and Braunvieh). Among the variants was a nonconservative substitution of lysine by alanine (K232A), with the lysine-encoding allele being associated with higher milk fat content. Haplotype analysis indicated the lysine variant to be ancestral. Two animals that were typed heterozygous (Qq) at the QTL based on marker-assisted QTL-genotyping were heterozygous for the K232A substitution, whereas 14 animals that are most likely qq at the QTL were homozygous for the alanine-encoding allele. An independent association study in Fleckvieh animals confirmed the positive effect of the lysine variant on milk fat content. We consider the nonconservative K232A substitution to be directly responsible for the QTL variation, although our genetic studies cannot provide formal proof.

Figure 1

Structure of bovine DGAT1. Exons are shown as rectangles. The sequence of intron 10 (≈65 bp) could not be completely resolved by sequencing. K232A indicates the only nonsynonymous mutation.

Figure 2

Alignment of the DGAT amino acid sequences of Arabidopsis thaliana (Ath), Brassica napus (Bna), Perilla fructescens (Pfr), Caenorhabditis elegans (Cel), Mus musculus (Mmu), Rattus norvegicus (Rno), Ceropithecus aethiops (Cae). Homo sapiens (Hsa) and two alleles of Bos taurus (Bta_ 232K, Bta_232A) by using PILEUP of the GCG package (23). Sequences were assembled by using BOXSHADE (http://www.isrec.isb-sib.ch:8080/software/BOX_form.html). Numbers on the left indicate amino acid positions. Red and blue letters indicate identical and conserved amino acids, respectively. Red arrows indicate identical lysine residues that might play a role in acyl-CoA binding. The blue arrow indicates conserved amino acids in animal species and in the bovine allele associated with high milk fat content. The lysine-to-alanine mutation at this position is not conservative and could have a negative effect on the acyl-CoA-binding capacity of DGAT.

Figure 3

(A) Haplotypes of DGAT1 based on nucleotide positions 3343, 10433, 10434, 11030, 11048, and 11993 determined by direct sequencing and (B) preliminary frequency estimates for the lysine (red) and alanine (yellow) encoding alleles in different species and breeds determined by RFLP assay (n, number of animals).

Figure 4

(A) Distributions of breeding values for milk fat content of Holstein–Friesian, Fleckvieh, and Braunvieh artificial insemination bulls. Colored areas indicate the range of the breeding values, from which bulls were chosen for the extreme positive (+, red) and negative (−, yellow) pools. (B and E) Sequence trace views of sequencing traces for positions 10430–10437 within DGAT1 for DNA pools (B) and individual animals (E). The vertical line indicates nucleotide position 10433. Positions 10433 and 10434 are responsible for the K232A substitution. (C) Estimates of allelic frequencies in the +pool (red) and −pool (yellow) for each breed based on sequencing traces. (D) Frequencies of alleles with 3, 4, 5, 6, and 7 repeat units in the 5′ region of DGAT1 in the +pool (red) and −pool (yellow) for each breed determined by fragment analysis.

Figure 5

(A) Across family test statistic curve for QTL analyses of milk fat content on chromosome 14 for a Fleckvieh granddaughter design. F ratios testing for the presence of a segregating QTL are plotted for given positions along the chromosome. The marker map with distances in centimorgans (cM) between markers is shown on the x axis. Empirical chromosome-wide and genome-wide 1% significance levels achieved by 10,000 permutations are indicated as horizontal lines. (B) Bars show transformed significance levels [log (1/p)] of the test statistic for a segregating QTL present at 0 cM within each family (x axis). The horizontal line indicates the transformed 1% significance level for a single family after correcting for multiple testing of 20 families. QTL effects for milk fat content and their respective standard errors are shown on top of the bars for significantly segregating sires. (C) Detection of allelic variation at nucleotide positions 10433 and 10434 (K232A) of the DGAT1 gene by CfrI cleavage in a 411-bp PCR product from bovine genomic DNA of sires 1–16. Cleavage by CfrI is diagnostic for the allele encoding alanine (GC). No DNA samples were available for sires 17–20.

Figure 6

Haplotypes of two segregating (Qq) Fleckvieh bulls. The haplotype on the black background is found in Fleckvieh, Anatolian Black, and Sahival; the haplotype on the gray background is the only lysine-encoding haplotype that we found in Holstein–Friesian (Fig. 3A). The arrows indicate the homozygous sites, implicating that these variants are not causal.

Figure 7

Distribution of breeding values for sons of nonsegregating sires grouped according to whether or not they have received the lysine allele from their dams. The numbers of bulls (n), the corrected means (μ), and standard deviations (σ) are indicated for the two groups.