A PLAG1 mutation contributed to stature recovery in modern cattle

Abstract

The recent evolution of cattle is marked by fluctuations in body size. Height in the Bos taurus lineage was reduced by a factor of ~1.5 from the Neolithic to the Middle Ages, and increased again only during the Early Modern Ages. Using haplotype analysis, we found evidence that the bovine PLAG1 mutation (Q) with major effects on body size, weight and reproduction is a >1,000 years old derived allele that increased rapidly in frequency in Northwestern European B. taurus between the 16^th and 18^th centuries. Towards the 19^th and 20^th centuries, Q was introgressed into non-European B. taurus and Bos indicus breeds. These data implicate a major role of Q in recent changes in body size in modern cattle, and represent one of the first examples of a genomic sweep in livestock that was driven by selection on a complex trait.

Figure 1

Schematic of clinal and temporal variation in cattle stature. The line art was produced in R v.3.3.2 and enhanced in Inkscape v0.48.4-r9939.

Figure 2

Identification of a haplotype tagging the PLAG1 mutation (Q) in B. indicus. (a) Scatterplot showing the birth weight (dEBV) haplotype association mapping on chromosome 14 in Nellore cattle. Maximum association (p = 4.83 × 10⁻¹⁷) was detected in a ~39.5 kbp segment spanning positions 24973324 to 25012733, where MOS and the 3′ end of PLAG1 are located. (b) The distribution of birth weight dEBVs according to number of copies of the tag haplotype indicates an additive effect of 0.311 ± 0.037 kg. (c) Orthology analysis suggesting Q to be a derived mutation.

Figure 3

B. taurus introgression as a source for Q in B. indicus. (a) Ancestry analysis showing that both birth weight dEBVs and occurrence of Q were positively associated with percentage of B. taurus ancestry on chromosome 14. (b) A principal components analysis revealed two breeding subgroups of Nellore cattle differing in B. taurus ancestry, frequency of Q, incidence of polledness (presumably also resulting from B. taurus introgression) and mean of birth weight dEBV. (c) Next-generation sequencing data of 24 Nellore bulls confirming a B. taurus origin of Q in B. indicus. Each row represents a bull, and colored vertical bars represent genotypes at different variant sites.

Figure 4

Haplotype diversity at the PLAG1 locus in the Bovine HapMap data^,. Each node represents a haplotype and edges connect nodes sequentially differing in one or two nucleotides. Node size is proportional to haplotype frequency. The Q-tagging haplotype is shown to be highly frequent in breeds originated from Northwestern Europe. A bifurcation diagram (rooted at rs109815800) is also shown, portraying the long-range linkage disequilibrium (LD) and low haplotype diversity around Q.

Figure 5

Atlantic Europe as the most likely centre of recent selection for Q. The heatmap was generated with ggplot2 v2.1.0 using inverse weighted distance exact interpolation of GGGTTCG frequency from breed origin (crosses).

Figure 6

Time to coalescence for the PLAG1 haplotype. (a) Extended haplotype homozygosity (EHH) analysis (rooted at rs109815800). In Northwestern European breeds, the signature dates back to the period of stature recovery in the 16^th - 18^th century. The haplotype is much more recent in B. indicus and Japanese B. taurus breeds, indicating introgression. (b) Simulation of increase in haplotype frequency according to different selection coefficients (s) and frequency prior to selection (p ₀). (c) Mean log-ratio from bone fragments recovered from archaeological sites in Iceland dating to the 14^th – 15^th century and the 17^th – 18^th century. The later period corresponds to early selection for Q and presents bone fragments on average 1.2 times larger than those observed in the earlier period. Error bars represent the standard errors of the means.

Figure 7

Insights on the age of Q from rs109231213 genotypes in ancient DNA. (a) Alignments of next-generation sequencing data reported by Park et al. from a >6,000 years old bovine humerus found in England. This specimen carried a qq genotype, as evidenced by all reads (horizontal grey bars) presenting allele G at rs109231213 (dashed rectangle). (b) A polled cranium (specimen #159) from ~1,000 yBP was recovered from a ritual gathering site in Hofstadir, Iceland. Target-sequencing of its petrous bone revealed a QQ genotype (i.e., all clones exhibited allele C at rs109231213). (c) A second polled cranium (specimen #2439) recovered from Skalholt and dating >300 yBP returned a heterozygous genotype. (d) Two molar teeth from a Medieval site in Northern Italy (US123 and US124) and a molar tooth (MU15) and a long bone (MU18) belonging to the late Roman age (>1,700 yBP) all presented qq genotypes.

Figure 8

Functional candidate variants underlying the PLAG1 chromosomal domain. Variant rs109231213 is predicted to change a conserved miRNA binding site, whereas variants ss319607405 and ss319607406 modify the transcription activity of the bi-directional promoter of PLAG1 and CHCHD7.