Centromere localization in medaka fish based on half-tetrad analysis

Gene-centromere (G-C) mapping provides insight into vertebrate genome composition, structure and evolution. Although medaka fish are important experimen-tal animals, no genome-wide G-C map of medaka has been constructed. In this study, we used 112 interspecific triploid hybrids and 152 DNA markers to make G-C maps of all 24 linkage groups (LGs). Under the assumption of 50% interference, 24 centromeres were localized onto all corresponding medaka LGs. Comparison with 21 centromere positions deduced from putative centromeric repeats revealed that 19 were localized inside the centromeric regions of the G-C maps, whereas two were not. Based on the centromere positions indicated in the G-C maps and those of centromeric repeats on each LG, we classified chromosomes as either biarmed or monoarmed; n = 24 = 10 metacentrics/submetacentrics + 14 subtelocentrics/acrocentrics, which is consistent with the results of previous karyological reports. This study helps to elucidate genome evolution mechanisms, and integrates physical and genetic maps with karyological information of medaka.


INTRODUCTION
The identification of centromere positions is essential for the integration of cytogenetic and genetic linkage maps and is also an initial step toward understanding the composition and structure of the centromeric region as well as the whole genome. Mainly due to the lack of well-defined genetic linkage maps using co-dominant markers, centromeres have been located only in very limited fish species so far. Most previous gene-centromere (G-C) map studies on fish species were based on allozyme markers and conducted decades ago. With the advantages and popularity of co-dominant DNA markers, G-C maps have recently been reported for many fish species. Zebrafish, Danio rerio, was the first fish for which all 25 centromeres were localized on genetic linkage maps (Johnson et al., 1996), followed by rainbow trout, Oncorhynchus mykiss (Sakamoto et al., 2000), loach, Misgurnus anguillicaudatus (Morishima et al., 2001), Japanese eel, Anguilla japonica (Nomura et al., 2006), turbot, Scophthalmus maximus (Martínez et al., 2008), large yellow croaker, Pseudosciaena crocea (Li et al., 2008), half-smooth tongue sole, Cynoglossus semilaevis (Ji et al., 2009), walking catfish, Clarias macrocephalus (Poompuang and Sukkorntong, 2011) and bighead carp, Hypophthalmichthys nobilis (Zhu et al., 2013).
Medaka fish (Oryzias latipes, O. sakaizumii and O. sinensis) represent one of the most important experimental animals worldwide, and have been used in fields such as genetics and developmental biology. The first medaka G-C map was made for linkage group (LG) 1, which is the sex LG (Sato et al., 2001). The distance between the marker and centromere is very important for estimating centromeric regions: the shorter the distance, the more accurate the centromeric regions. In this study, we aimed to localize centromeres onto all 24 LGs. The information obtained from this G-C map will be useful for understanding the genome structure and chromosome evolution of this species group.

MATERIALS AND METHODS
Fish All fish strains used in this study were supplied by the Faculty of Science, Niigata University, a subcenter of the National BioResource Project (medaka) supported by the Ministry of Education, Culture, Sports, Science and Technology of Japan, and were reared in 150 × 250 × 105mm plastic vessels with 2 l of water at 25 °C. Oryzias luzonensis was originally obtained from Luzon (Formacion and Uwa, 1985), and two inbred strains of medaka, HNI-II and Hd-rR (derived from O. sakaizumii and O. latipes, respectively), were described by Hyodo-Taguchi and Sakaizumi (1993).
Crosses Hd-rR strain females were mated with HNI-II males, which resulted in F1 progeny (DNF1). DNF1 females and O. luzonensis males were isolated the day before they were mated. Fertilized eggs were immediately collected after spawning; after 2 min and 45 sec, they were exposed to 42 °C water for 2 min to block the second meiotic division (Naruse et al., 1985). Allodiploid embryos of O. latipes/sakaizumii and O. luzonensis almost all develop abnormally or die before hatching (Formacion and Uwa, 1985). The hatching rate of the hybrid embryos is improved by high-temperature treatment, and this improvement has been verified to be the result of triploidization (Sato et al., 2001). Thus, all or most of the hatched fry are expected to be allotriploids.
Genotyping DNA was extracted from the hatched fry by proteinase K digestion, phenol-chloroform extraction and isopropanol precipitation (Shinoyama et al., 1999). DNA samples were dissolved in TE buffer. To find DNA markers to detect polymorphisms between Hd-rR, HNI-II and O. luzonensis, we selected expressed sequence tag (EST) markers from the medaka EST database (http://mbase.nig. ac.jp/mbase/medaka_top.html), and made new markers as necessary (Supplementary Table S1). PCR amplification of genomic DNA was performed as follows: 35 cycles at 95 °C for 30 sec, 55 °C for 30 sec, and 72 °C for 2 min. All PCR products were electrophoresed on polyacrylamide gels as described by Kimura et al. (2004). In this study, we developed an SNP genotyping system between the HNI-II and Hd-rR inbred strains. SNP genotyping was performed using the MassARRAY iPLEX Gold Assay (Sequenom) (Takada et al., 2013). The information used for genotyping is described in Supplementary Table S2.
Centromere localization The second meiotic division segregation frequency (y), which is a function of the recombination rate between the marker and centromere, was estimated by counting the proportion of progeny heterozygous for Hd-rR and HNI-II alleles in allotriploid hybrids, which allowed us to estimate G-C distance (x) (Johnson et al., 1996). G-C distance may be estimated under different models of chiasma interference (Danzmann and Gharbi, 2001). Assuming 50% chiasma interference, G-C distance was calculated as follows: x = [ln (1 + y) − ln (1 − y)] × 100 / 4. Chromosome type classification Chromosomes were classified according to the criterion of the centromeric index (CI; long arm to short arm ratio) introduced by Levan et al. (1964). Chromosome type and corresponding CI values were as follows: metacentric, 1.0-1.7; sub-metacentric, 1.7-3.0; subtelocentric, 3.0-7.0; and acrocentric, more than 7.0.

RESULTS AND DISCUSSION
Construction of centromere-linkage map In this study, we used 112 triploid hybrids and 152 DNA markers (Supplementary Table S1). Hatched fry were all allotriploid hybrids, as determined by genotyping using DNA markers. The map length of each chromosome was 44.3-58.7 cM (average, 50.2 cM). The linkage map spanned a genetic length of 1,204 cM (Table 1), which is 4% shorter than that proposed in a previous study (1,257 cM in female, Kimura et al., 2005); however, we used terminal markers for the physical map of each chromosome in this study. This difference may also be partially due to long interval distances between loci (average interval distance, 9.3 vs. 5.4 cM in Kimura et al., 2005). Centro- *Centromeric and other regions are depicted as gray and white boxes, respectively, in G-C maps (Fig. 1). Gene-centromere mapping of medaka
A comparison between the positions of putative centromeric repeats on 23 LGs and the centromeric regions in the G-C maps revealed that repeat positions on 19 LGs (1-5, 7-12, 14, 16-20, 22 and 23) were inside their respective centromeric regions in the G-C maps, whereas those of LGs 6 and 13 were not. The putative centromeric repeats on LG6 were at 0 Mbp in Hd-rR, but at the opposite end in HNI-II, and those on LG13 may not be functional. For LG 21, two positions (2.5 and 14.5 Mbp) of centromeric repeats were found in Hd-rR, whereas only one (2.5 Mbp) was found in HNI-II; additionally, the repeats at 14.5 Mbp were located inside the centromeric region in the G-C map, which indicates that these repeats have centromeric function. The repeats at 2.5 Mbp may be silenced in both Hd-rR and HNI-II. The centromere of HNI-II is assumed to be located at the same position as that of Hd-rR. Although centromeric repeats were not identified on LG24, the centromere was mapped at one end (19-23 Mbp) of the chromosome.
The results of this study integrate the centromere map and genetic linkage map in O. latipes/O. sakaizumii, which provides valuable information that will help to consolidate the genetic and physical maps in the near future. These findings will also be useful for studies on genome structure, chromosome evolution and positional cloning of genes in this species complex.