Extensive genetic diversity and substructuring among zebrafish strains revealed through copy number variant analysis

Abstract

Copy number variants (CNVs) represent a substantial source of genomic variation in vertebrates and have been associated with numerous human diseases. Despite this, the extent of CNVs in the zebrafish, an important model for human disease, remains unknown. Using 80 zebrafish genomes, representing three commonly used laboratory strains and one native population, we constructed a genome-wide, high-resolution CNV map for the zebrafish comprising 6,080 CNV elements and encompassing 14.6% of the zebrafish reference genome. This amount of copy number variation is four times that previously observed in other vertebrates, including humans. Moreover, 69% of the CNV elements exhibited strain specificity, with the highest number observed for Tubingen. This variation likely arose, in part, from Tubingen's large founding size and composite population origin. Additional population genetic studies also provided important insight into the origins and substructure of these commonly used laboratory strains. This extensive variation among and within zebrafish strains may have functional effects that impact phenotype and, if not properly addressed, such extensive levels of germ-line variation and population substructure in this commonly used model organism can potentially confound studies intended for translation to human diseases.

Fig. 1.

Experimental design overview. Twenty zebrafish individuals from each of AB, WIK, and Tu and a native strain from Bangladesh were analyzed (10 individuals per laboratory site). A single individual from each strain was randomly selected as the strain reference to compare against remaining individuals in the strain using a custom Agilent 1 Million feature aCGH array platform. Analyses produced 31,749 CNVs that were combined into 6,080 CNVEs using a 50% reciprocal overlap criterion.

Fig. 2.

CNV size variation. (A) Cumulative distribution function plots of CNV sizes for each strain. Faster-rising lines indicate an increased frequency of smaller CNVs, and slower-rising lines indicate an increased frequency of larger CNVs. Differences between strains were significant, with Tu having significantly larger CNVs and the native fish having significantly smaller CNVs (ANOVA; P < 0.001). (B) CNV frequency histogram of the percentage of total CNV calls within 10-kb size bins indicating a significantly higher percentage of small CNVs in native fish and a significantly higher percentage of larger CNVs in Tu (ANOVA; P < 0.001).

Fig. 3.

CNV map. A combined zebrafish CNVE map with copy number gains (green) and losses (red) distributed along chromosome lengths. Length of green and red lines reflect relative CNVE frequencies at respective chromosomal locations.

Fig. 4.

Strain-specific differences. (A) A Venn diagram indicating strain-specific CNVEs and the numbers of overlapping CNVEs between strains. CNVEs observed in three or more strains are represented as raised shallow plateaus and three-way overlaps, with the highest plateau representing CNVEs occurring in all strains. (B) Structure plot of CNVE data analyzed by FRAPPE for population substructure found an optimal value of K = 4. Analyses were also performed for K = 5–8 with no significant changes in the structure.

Fig. 5.

Composite stocks are created by combining individuals from multiple locally adapted populations. Combined stocks then hybridized in a random intraspecific manner, increasing genetic variability through extensive recombination between divergent populations. In nature, such hybrids are frequently selected against, but in a commercial fish farm lacking selective pressure, hybridized individuals survive and reproduce randomly within the composite population, further increasing genetic variation. Tu, having been created from such a composite population and without significant selective measures, exhibits increased CNV numbers and sizes.