Genomic and phenotypic characterization of a wild medaka population: towards the establishment of an isogenic population genetic resource in fish

Abstract

Oryzias latipes (medaka) has been established as a vertebrate genetic model for more than a century and recently has been rediscovered outside its native Japan. The power of new sequencing methods now makes it possible to reinvigorate medaka genetics, in particular by establishing a near-isogenic panel derived from a single wild population. Here we characterize the genomes of wild medaka catches obtained from a single Southern Japanese population in Kiyosu as a precursor for the establishment of a near-isogenic panel of wild lines. The population is free of significant detrimental population structure and has advantageous linkage disequilibrium properties suitable for the establishment of the proposed panel. Analysis of morphometric traits in five representative inbred strains suggests phenotypic mapping will be feasible in the panel. In addition, high-throughput genome sequencing of these medaka strains confirms their evolutionary relationships on lines of geographic separation and provides further evidence that there has been little significant interbreeding between the Southern and Northern medaka population since the Southern/Northern population split. The sequence data suggest that the Southern Japanese medaka existed as a larger older population that went through a relatively recent bottleneck approximately 10,000 years ago. In addition, we detect patterns of recent positive selection in the Southern population. These data indicate that the genetic structure of the Kiyosu medaka samples is suitable for the establishment of a vertebrate near-isogenic panel and therefore inbreeding of 200 lines based on this population has commenced. Progress of this project can be tracked at http://www.ebi.ac.uk/birney-srv/medaka-ref-panel.

Keywords: Medaka; inbreeding; population genomics; strain specific features.

Figure 1

Geographical localization of the sampling sites for inbred strains and the Kiyosu population. (A) Medaka consists of four major populations in East Asia. The Nilan strain comes from Ilan-City, Taiwan, and is a representative of the Chinese/West Korean population; the HSOK strain is from Sokcho-City, Gangweon-do, Korea, and is a member of the East Korean population. The Japanese populations are divided by the Japanese Alps into Northern and Southern strains. Whereas Kaga and HNI belong to the Northern populations, HdrR is the Southern reference strain that was used for the Medaka genome sequencing project (Kasahara et al., 2007). In red is the site in Kiyosu where the founders for the inbreeding panel were sampled (see Materials and Methods for the GPS location). (B) Representative pictures of lateral and dorsal views of 10 days’ postfertilization-old larvae of the different Japanese inbred strains. Icab, HNCMH2, and HO5 are further Southern Japanese inbred strains whose original sampling site is not shown in (A).

Figure 2

Levels of heterozygosity in medaka inbred lines. Four plots showing regions of heterozygosity consistent with a wild origin in the HNI, HSOK, Kaga, and Nilan inbred strains. Black blocks show windows of heterozygosity >7.1 × 10⁻⁴. At this threshold, 90% of the genome in the wild catches is classified as heterozygous. Gray blocks are assembly gaps. Chromosome 1 in medaka is the known location of the sex determination locus as indicated by the black arrow heads (Katsumura et al. 2009).

Figure 3

Phylogenetic relationships of medaka inbred strains and the Kiyosu population. (A) Phylogenetic relationships between the wild Southern population, five inbred strains, and stickleback based on high-throughput genome sequencing. PHYLIP clustering of inbred strains using a neighbor-joining algorithm based on Kimura two-parameter distances is shown. Wild Southern is a synthetic sample based on the sequencing of the Kiyosu population and determining bases using majority vote across 48 haplotypes. Differences between medaka samples and stickleback were too large to estimate distances. (B) Hierarchical clustering with Gower distances of genotypes of Kiyosu population founder samples, and representative Southern and Northern strains.

Figure 4

Linkage disequilibrium (LD) in the Kiyosu medaka population. (A) Decay of LD for a sample of medaka SNPs. LD (r²) was calculated for all SNP pairs in 1-Mb windows overlapping by 500 kb across medaka chromosome 1, and r2 was plotted against distance for a sample of 0.03% (21976) of pairs. The main plot shows a blow-up of distances between 0 and 25 kb, with data for pairs of SNPs up to 1 Mb apart shown in the inset region. Multiple sampling and analysis of other chromosomes gave similar results (not shown). (B) Pairwise LD plot for a representative region of chromosome 11. Pairwise LD (D′) for the 25-kb region 11:375000−399999 as displayed in Haploview is shown. The plot shows the LD of pairs of SNPs by squares joining the two SNPs via the diagonal and is color coded as described for the standard LD color scheme at http://www.broadinstitute.org/science/programs/medical-and-population-genetics/haploview/ld-display such that pairs of SNPs in LD at D′ = 1 are colored blue at LOD <2 and red at LOD ≥ 2, whereas those with D′ < 1 are white at LOD < 2 and shades of red/pink at LOD ≥ 2.

Figure 5

Distribution of population size estimates from the Kiyosu individuals. Population size estimates over time were calculated using the psmc package for the founder individuals in the Kiyosu wild catch as described in Materials and Methods. The graph shows the distribution of estimates after combining by binning.

Figure 6

Morphometric analysis of medaka inbred strains. (A) Four lateral (L1-L4) and five dorsal (D1-D5) morphometric distances were extracted and analyzed for each inbred strain. (B) A bar chart showing the proportion of variance explained by the difference between strains as a fraction of the total variance. The variables are the measurements corrected by the appropriate body length measurement (L4 and D5, respectively). (C) Example pictures showing substantial differences in lateral length and eye diameter between fish from the two Southern inbred strains HdrR and Icab. (D) Box plots showing the distribution of eye diameter, lateral length, and the ratio between eye diameter and lateral length for HdrR and Icab. Eye diameter and lateral length: y-axis = value in pixel. HdrR (n = 70), Icab (n = 78).