Multiple Origins and Genomic Basis of Complex Traits in Sighthounds

Abstract

Sighthounds, a distinctive group of hounds comprising numerous breeds, have their origins rooted in ancient artificial selection of dogs. In this study, we performed genome sequencing for 123 sighthounds, including one breed from Africa, six breeds from Europe, two breeds from Russia, and four breeds and 12 village dogs from the Middle East. We gathered public genome data of five sighthounds and 98 other dogs as well as 31 gray wolves to pinpoint the origin and genes influencing the morphology of the sighthound genome. Population genomic analysis suggested that sighthounds originated from native dogs independently and were comprehensively admixed among breeds, supporting the multiple origins hypothesis of sighthounds. An additional 67 published ancient wolf genomes were added for gene flow detection. Results showed dramatic admixture of ancient wolves in African sighthounds, even more than with modern wolves. Whole-genome scan analysis identified 17 positively selected genes (PSGs) in the African population, 27 PSGs in the European population, and 54 PSGs in the Middle Eastern population. None of the PSGs overlapped in the three populations. Pooled PSGs of the three populations were significantly enriched in "regulation of release of sequestered calcium ion into cytosol" (gene ontology: 0051279), which is related to blood circulation and heart contraction. In addition, ESR1, JAK2, ADRB1, PRKCE, and CAMK2D were under positive selection in all three selected groups. This suggests that different PSGs in the same pathway contributed to the similar phenotype of sighthounds. We identified an ESR1 mutation (chr1: g.42,177,149 T > C) in the transcription factor (TF) binding site of Stat5a and a JAK2 mutation (chr1: g.93,277,007 T > A) in the TF binding site of Sox5. Functional experiments confirmed that the ESR1 and JAK2 mutation reduced their expression. Our results provide new insights into the domestication history and genomic basis of sighthounds.

Keywords: artificial selection; origin; population genomics; sighthounds.

Fig. 1.

Geographic and genetic grouping of samples. (a) Locations of sample collection, with photos of dogs sourced from AKC and Wiki. (b) PCA plot showing the distribution of all dogs, excluding East Asian village dogs (left); neighbor-joining phylogenetic tree highlighting relationships between samples (middle); and admixture analysis displaying population structure (right). Sighthound individuals are represented by shaded branches in the phylogenetic tree, while the framed structure bar indicates sighthound individuals. Notably, sighthound samples from the Middle East, Africa, and Europe closely aligned with their respective local samples. Shaanxi sighthounds (indicated by arrowhead pointing to clades) merged with the European sighthound population in both the phylogenetic tree and admixture analysis. The Sloughi individual (marked by asterisks) was excluded from selection and population analysis due to genetic admixture and its outlier position in PCA, neighbor-joining tree, and structure analysis. (c) RFMix result illustrating genetic ancestry of sighthound populations.

Fig. 2.

Gene flow between sighthounds and ancient wolves. (a) TreeMix simulation diagram showing gene flow between sighthounds and each ancient wolf. Lines with arrows indicate gene flow and direction, and numbers on the lines represent the frequency of these events in the TreeMix analysis between the 67 ancient wolves and sighthounds. (b) Outgroup D-statistics depicting phylogeny analysis; red bars indicate significant differences from nonadmixed part of sighthounds and modern wolves. (c) D-statistics comparing African dogs to Eurasia ancient wolves.

Fig. 3.

Positive selection in sighthounds from Africa, Europe, and Middle East. (a–c) Genome-wide original Fst and empirical P values of XP-EHH and XP-nSL for African native sighthound. We highlighted candidate genes that above the cutoff in all three methods, and genes mentioned in the main text are labeled. (d–f) Genome-wide original Fst and empirical P values of XP-EHH and XP-nSL for European sighthound. (g–i) Genome-wide original Fst and empirical P values of XP-EHH and XP-nSL for Middle Eastern sighthounds.

Fig. 4.

Luciferase reporter on chr1: g.42,177,149 T > C of ESR1. (a) Fisher's exact test of site allele frequency in ESR1 region of each comparison, with chr1: g.42,177,149 T > C (red dot) is beyond the significance level of 0.01 (−log P > 1.97) in all comparisons. (b) Sequence logo of the Stat5a-binding site on ESR1 provided by JASPAR (https://jaspar.genereg.net/matrix/MA0519.1/); arrowhead points to chr1: g.42,177,149 T > C mutation position. (c) Luciferase reporter was applied to H9C2 cells, and ESR1 expression in wild type (T) was significantly higher (t-test P = 2.56 × 10⁻²) than that in mutation type (C). (d) Luciferase reporter was applied to A673 cells, and ESR1 expression in wild type (T) was significantly higher (t-test P = 2.22 × 10⁻⁶) than that in mutation type (C).

Fig. 5.

Luciferase reporter on chr1 93,277,007 T > A of JAK2. (a) Fisher's exact test on-site allele frequency in the JAK2 region of each comparison, with chr1 93,277,007 T > A (red dot) is beyond the significance level of 0.001 (−log P > 9.88) in all comparisons. (b) Sequence logo of the Sox5-binding site on JAK2 provided by JASPAR (https://jaspar.genereg.net/matrix/MA0087.1/). (c) Luciferase reporter was applied to H9C2 cells, and JAK2 expression in wild type (T) was significantly higher (t-test P = 2.94 × 10⁻²) than that in mutation type (A). (d) Luciferase reporter was applied to A673 cells, and JAK2 expression in wild type (T) was significantly higher (t-test P = 4.86 × 10⁻³) than that in mutation type (A).