Copy number variation underlies complex phenotypes in domestic dog breeds and other canids

Extreme phenotypic diversity, a history of artificial selection, and socioeconomic value make domestic dog breeds a compelling subject for genomic research. Copy number variation (CNV) is known to account for a significant part of inter-individual genomic diversity in other systems. However, a comprehensive genome-wide study of structural variation as it relates to breed-specific phenotypes is lacking. We have generated whole genome CNV maps for more than 300 canids. Our data set extends the canine structural variation landscape to more than 100 dog breeds, including novel variants that cannot be assessed using microarray technologies. We have taken advantage of this data set to perform the first CNV-based genome-wide association study (GWAS) in canids. We identify 96 loci that display copy number differences across breeds, which are statistically associated with a previously compiled set of breed-specific morphometrics and disease susceptibilities. Among these, we highlight the discovery of a long-range interaction involving a CNV near MED13L and TBX3, which could influence breed standard height. Integration of the CNVs with chromatin interactions, long noncoding RNA expression, and single nucleotide variation highlights a subset of specific loci and genes with potential functional relevance and the prospect to explain trait variation between dog breeds.

Dogs have been the subject of intense study over many decades (Vilà et al. 1999; Ostrander and Wayne 2005; Freedman et al. 2014; Ostrander et al. 2019), providing valuable insight into human history, disease, and evolution (Coelho et al. 2018; Ní Leathlobhair et al. 2018; Wang et al. 2019). Much has been learned about canines through traditional approaches, including genotype studies with microsatellites (Irion 2003), single nucleotide polymorphisms (SNPs) (Gundry et al. 2007; Boyko et al. 2010; Vaysse et al. 2011), and, finally, whole genome sequencing (WGS) (Lindblad-Toh et al. 2005; Freedman et al. 2014; Plassais et al. 2019).As a result of the extensive history of genetic studies in dogs, remarkable advances have been made toward the resolution of the canine phylogeny (vonHoldt et al. 2010; Parker et al. 2017) and the temporal, geographic, and demographic history of dog domestication (Freedman et al. 2014; Shannon et al. 2015; Skoglund et al. 2015). Studies suggest that dogs were initially domesticated from gray wolves 15,000 to 40,000 yr ago (Freedman et al. 2014; Skoglund et al. 2015; Freedman and Wayne 2017; Ostrander et al. 2019), with a rapid diversification of breeds occurring within the past few hundred years. Currently, about 400 dog breeds exist worldwide, 193 recognized by the American Kennel Club and 360 by the Fédération Cynologique Internationale. Breed classification schemes have been proposed based on occupation, morphology, and geographic origin (American Kennel Club 2007; Wucher et al. 2017). The most recent genetic analysis, encompassing nearly 200 breeds and populations, suggests a monophyletic origin for most modern breeds and provides data regarding their origins and timing (Parker et al. 2017). Clusters of genetically similar breeds were identified and assigned to clades, which often reflected occupational and geographical origins.Targeted and genome-wide genotyping approaches have led to the discovery of nearly 400 variants associated with more than 270 traits, over 220 of which correspond to possible models for human diseases (Online Mendelian Inheritance in Animals [OMIA], Sydney School of Veterinary Science, https://omia.org/). Particularly, genome-wide association studies (GWASs) involving modest size cohorts of dogs have led to the identification of variants controlling a variety of morphological, behavioral, and disease traits (Akey et al. 2010; Vaysse et al. 2011; Rimbault et al. 2013; Hayward et al. 2016; MacLean et al. 2019; Plassais et al. 2019).The recent and intense artificial selective pressure exerted on dogs has induced pronounced inter-breed phenotypic differences while preserving intra-breed homogeneity. This process makes dogs of the same breed more likely to share not only morphometric traits but also disease susceptibilities (Karlsson and Lindblad-Toh 2008; Chase et al. 2009; Akey et al. 2010; Boyko et al. 2010; Marchant et al. 2017; Mansour et al. 2018; Ostrander et al. 2019). The level of anatomic similarity among dogs of any one breed is sufficiently strong that genetic studies have been successfully executed using breed standards as phenotypes, thus unraveling the genetic bases of some complex traits such as body size or behavior (Akey et al. 2010; Boyko et al. 2010; Vaysse et al. 2011; Hayward et al. 2016; MacLean et al. 2019; Plassais et al. 2019), which remain elusive, even in humans.However, all these analyses have been performed using a subset of indicative SNPs and, more recently, SNPs from WGSs (Jagannathan et al. 2019; Plassais et al. 2019), but other forms of genomic variation have rarely been studied systematically. In fact, there is still a lack of fine-scale, genome-wide analyses of any variants other than SNPs across dog breeds, a notable exception when compared to humans and other model organisms (Yalcin et al. 2011; Brown et al. 2012; Sudmant et al. 2015). Copy number variation (CNV) has been previously studied in canines to elucidate specific phenotypes (Karyadi et al. 2013; Arendt et al. 2014; Waldo and Diaz 2015; Deane-Coe et al. 2018). However, most studies have focused on the comparison of dogs and wolves using array-based technologies, rather than undertaking a comprehensive and unbiased examination of all CNVs across the genome of distinct breeds (Berglund et al. 2012; Schoenebeck et al. 2012). Most CNV-related studies published to date only aimed to identify segmentally duplicated regions and did not aim to produce quantitative copy-number (CN) genotypes (Quilez et al. 2012; Molin et al. 2014). Knowing the exact number of copies at a locus is crucial for an accurate comparison of closely related organisms, such as distinct dog breeds and wild canids.Here, we present a fine-scale CNV map of over 300 canid samples using WGS to produce the most extensive, high-resolution CNV panel in dogs to date. We examine more than 145 individual breeds, as well as nonbreed dogs, including village dogs, dingoes, captive New Guinea singing dogs, and wild canids such as wolves. We employ this data set to determine the ability of CNVs to recreate a current dog phylogeny. Moreover, we test for breed-phenotype associations using an extensive data set of breed standards as individual phenotypes in the first CNV-based GWAS performed in dogs to date.     

Complex mosaic structural variations in human fetal brains

Somatic mosaicism, manifesting as single nucleotide variants (SNVs), mobile element insertions, and structural changes in the DNA, is a common phenomenon in human brain cells, with potential functional consequences. Using a clonal approach, we previously detected 200–400 mosaic SNVs per cell in three human fetal brains (15–21 wk postconception). However, structural variation in the human fetal brain has not yet been investigated. Here, we discover and validate four mosaic structural variants (SVs) in the same brains and resolve their precise breakpoints. The SVs were of kilobase scale and complex, consisting of deletion(s) and rearranged genomic fragments, which sometimes originated from different chromosomes. Sequences at the breakpoints of these rearrangements had microhomologies, suggesting their origin from replication errors. One SV was found in two clones, and we timed its origin to ∼14 wk postconception. No large scale mosaic copy number variants (CNVs) were detectable in normal fetal human brains, suggesting that previously reported megabase-scale CNVs in neurons arise at later stages of development. By reanalysis of public single nuclei data from adult brain neurons, we detected an extrachromosomal circular DNA event. Our study reveals the existence of mosaic SVs in the developing human brain, likely arising from cell proliferation during mid-neurogenesis. Although relatively rare compared to SNVs and present in ∼10% of neurons, SVs in developing human brain affect a comparable number of bases in the genome (∼6200 vs. ∼4000 bp), implying that they may have similar functional consequences.

Somatic mosaicism, the presence of more than one genotype in the somatic cells of an individual, is a prominent phenomenon in the human central nervous system. Forms of mosaicism include aneuploidies and smaller copy number variants (CNVs), structural variants (SVs), mobile element insertions, indels, and single nucleotide variants (SNVs). The developing human brain exhibits high levels of aneuploidy compared to other tissues, generating genetic diversity in neurons (Pack et al. 2005; Yurov et al. 2007; Bushman and Chun 2013). Such aneuploidy was suggested to be a natural feature of neurons, rather than a distinctive feature of neurodegeneration. However, the frequency of aneuploidy in neurons has been debated, with a separate study suggesting that aneuploidies occur in only about 2.2% of mature adult neurons (Knouse et al. 2014). They hence infer that such aneuploidy could have adverse effects at the cellular and organismal levels. Additionally, analysis of single cells from normal and pathological human brains identified large, private, and likely clonal somatic CNVs in both normal and diseased brains (Gole et al. 2013; McConnell et al. 2013; Cai et al. 2014; Knouse et al. 2016; Chronister et al. 2019; Perez-Rodriguez et al. 2019), with 3%–25% of human cerebral cortical nuclei carrying megabase-scale CNVs (Chronister et al. 2019) and deletions being twice as common as duplications (McConnell et al. 2013). Given that CNVs often arise from nonhomologous recombination and replication errors, their likely time of origin is during brain development. However, when CNVs first arise in human brain development has not yet been investigated. The present work is the first to examine this question using clonal populations of neuronal progenitor cells (NPCs) obtained from fetal human brains.Detection of CNVs in single neurons is challenging, given the need to amplify DNA. Such amplification may introduce artifacts that could, in turn, be misinterpreted as CNVs. In order to address this technical limitation, Hazen et al. reprogrammed adult postmitotic neurons using somatic cell nuclear transfer (SCNT) of neuronal nuclei into enucleated oocytes (Hazen et al. 2016). These oocytes then made sufficient copies of the neuronal genome allowing for whole-genome sequencing (WGS), thus eliminating the need for amplification in vitro. Using this method, they identified a total of nine structural variants in six neurons from mice, three of which were complex rearrangements. However, it is not possible to extend such studies to humans, given the ethical issues involved, besides the technical challenges in obtaining and cloning adult neurons. To circumvent the need of single-cell DNA amplification or nuclear cloning, we examined clonal cell populations obtained from neural progenitor cells from the frontal region of the cerebral cortex (FR), parietal cortex (PA) and basal ganglia (BG) and describe here the discovery and analysis of mosaic SVs in these NPCs (Bae et al. 2018). These clones were sequenced at 30× coverage (much higher than most previous single-cell studies), allowing identification of SVs other than large deletions and duplications as well as precise breakpoint resolution.