Definition of the zebrafish genome using flow cytometry and cytogenetic mapping

The zebrafish (Danio rerio) has long been appreciated as a unique model system for vertebrate genetics and developmental biology. The zebrafish genome is about half the size of most mammalian genomes [1] containing some 4.6 pg of DNA [2] distributed across 25 pairs of chromosomes (2n = 50) [3]. Initial comparisons of zebrafish and mammalian gene maps have revealed extensive conservation of syntenic chromosome regions among vertebrates [4–10]. Hence, through the use of zebrafish mutants and a well annotated zebrafish reference genome, genetic mechanisms that are conserved among all vertebrates, can be investigated.

Since the launch of the Trans-NIH Genome Initiative in 1997 [11], excellent genetic and genome resources have been established to facilitate the genetic analysis of zebrafish. The first genetic linkage map was constructed using haploid genetics with random amplified polymorphic DNA (RAPD) markers [12, 13]. Several meiotic panels (e.g., the Boston MGH Cross, GAT, Heat Shock, and Mother of Pearl) and radiation hybrid (RH) panels (e.g., T51 and LN54) have been generated [9, 14–25]. Most recently, the Wellcome Trust Sanger Institute (Hinxton, United Kingdom) released the sixth zebrafish genome assembly (Zv6), comprising of 1.6 Gb [26]. This assembly was generated from 7,615 BAC clones, which were fingerprinted and then arranged into contigs by overlap analysis to generate a physical map. DNA sequence data from whole genome shotgun contigs were then used to fill gaps. While more high-throughput methods for DNA sequencing are now emerging (e.g., the 454 sequencing platform – which can sequence 25 million bases in about four hours [27]), a major limitation of many of these technologies is the provision of relatively short DNA sequences (about 100–200 bases per read), making such sequencing technologies problematic with the assembly of repeated DNA elements [28].

Cytogenetic methodologies provide a complementary mapping approach for assembly of reference genomes. In this study, we have employed two cytogenetic approaches to define the zebrafish genome: flow karyotype analysis and chromosomal mapping by fluorescence in situ hybridization (FISH). In flow karyotyping, each chromosome is labeled directly with two DNA-binding fluorochromes (i.e., Hoechst 33258 and Chromomycin A3) and then analyzed on a flow cytometer. The fluorescence from the two DNA-binding dyes enables different chromosome types to be identified by both DNA content and base-pair ratio, providing a means to estimate the relative size of each zebrafish chromosome. In FISH, cloned DNA sequences are hybridized to their complementary sequences and visually mapped onto metaphase chromosomes. In this study, we have unequivocally mapped 575 large-insert DNA clones onto zebrafish metaphase chromosomes using multi-color FISH methods. These clones can be used as anchors for annotation of the zebrafish reference genome and provide independent validation of the sequence map, in a fashion similar to how cytogenetic data from the human genome was integrated into the human DNA sequence map [29]. These data reinforce the use of cytogenetic methodologies for the assembly of complex genomes.

Following analysis of the initial zebrafish gene maps, it has been suggested that during the course of evolution, the zebrafish and human lineages shared two rounds of whole genome duplication with a third whole genome duplication event occurring before the last teleost radiation [10, 35]. The duplicated genome of the zebrafish has contributed to the increased difficulty in constructing an accurate genetic map for this organism, in comparison to the mouse and human. In addition, similarity in the size, morphology, and banding pattern of the zebrafish chromosomes has resulted in at least 12 different published zebrafish karyotypes [reviewed in [36]]. Phillips et al. [37] used a combination of relative size and chromosome arm ratios to estimate the sizes of each zebrafish chromosome. When comparing the chromosome size estimates from our flow cytometry data to the size estimates reported by Phillips et al. [37], the four largest LG chromosomes (i.e., LG chromosomes 5, 7, 4, and 3) and the three smallest LG chromosomes (i.e., LG chromosomes 22, 24, and 25) are in agreement. However, the ranking of sizes of most other LG chromosomes are discrepant and should be investigated further by other independent studies.

Cytogenetic mapping is a complementary approach to other mapping strategies (e.g., genetic and radiation hybrid mapping) and can be used to confirm existing framework maps. Using fluorescence in situ hybridization (FISH), 510 BAC clones were assigned to unique LG chromosome locations in this study. This suggests that if there is widespread duplication in the zebrafish genome, a large proportion of these appear to be syntenic and localized (e.g., tandem duplications). FISH mapping data was over 85% concordant for three of the four zebrafish genome databases (i.e., Ensembl, Vega, and WebFPC) with 75.9% concordance for the UCSC Genome Browser, when evaluating the BAC clones present in these databases (Table 2). Discrepancies between the FISH mapping data and the LG chromosome assignment by the zebrafish genome databases may be attributed to cross-contamination of the clones, clone tracking inconsistencies, incorrect assignment by previous mapping efforts, or inconsistencies among the four zebrafish genome databases. From the comparisons, it became evident that at the time of analysis, there is some degree of discrepancy among all four of the zebrafish genome databases. In addition, the Ensembl Zv6 (version6-release 40, August 2006) used in this analysis did not contain any of the BAC clones of the zK library. Of the 510 BAC clones that were assigned to specific LG chromosomes by FISH, only 3 clones did not have mapping information in any of the databases. In addition for 53 of the 507 clones with database mapping information, none of the databases had the same chromosomal assignment as was assigned by FISH (Additional file 1).

A small percentage of the clones were observed to hybridize to multiple locations within the zebrafish genome (Additional file 3). One clone contained duplicated regions within the same chromosome, while the remaining thirty clones hybridized to more than one chromosome. Many of the BAC clones that hybridized to multiple chromosomal locations localized near the centromeres, providing further evidence for the lack of chromosome-specific repetitive DNA sequences near these regions. These clones provide clues to duplications and exchanges that have occurred within and between chromosomes. Further investigation is needed to determine the functional significance of these duplications.

Some BAC clones yielded weak or no fluorescence signals. The lack of signals for these clones may be due to poor quality of probe DNA, contamination with non-zebrafish DNA, loss of insert DNA or poor hybridization due to factors such as increased repetitive DNA elements within a clone.

One particularly fascinating observation was a peri-centromeric inversion on LG chromosome 3 (Figure 3). Interestingly, this heteromorphism was seen only in the AB strain and not in Tu embryos. This finding suggests that there are strain-specific chromosomal differences in the zebrafish. Large chromosomal rearrangements (e.g., inversions) are important for evolution and are suspected to play a role in speciation. Studies have suggested that inversions can influence patterns of nucleotide diversity between diverging taxa [38] and encourage speciation by suppressing recombination in heterozygotes [39]. Chromosomal inversions can also play a role in fertility. For example, peri-centromeric inversion carriers may have a reduction in fertility due to the production of chromosomally unbalanced gametes. In a recent study, it has also been suggested that chromosomal inversions may be beneficial. Carriers for an inversion on human chromosome 17 were reported to have increased fertility rates in comparison to non-carriers [40].

Chromosomal mapping of zebrafish BAC clones by FISH in this study resulted in a second generation near-centromere and near-telomere BAC clone probe panel, which includes 41.3% new clones. These new BAC clones are visually mapped closer to the centromere or telomeres of each LG chromosome than those in the original probe panel. Furthermore, a near-telomeric BAC clone was identified at the long arm terminus of LG chromosome 4 (Figure 4). The long arm of LG chromosome 4 (i.e., chromosome 3 as denoted by the karyotype of Gornung et al. [41] and Phillips et al. [37]) has been previously shown to contain a substantial amount of constitutive heterochromatin [36]. Regions of constitutive heterochromatin often consist of highly repetitive DNA and are largely transcriptionally silent. The heterochromatic region of LG chromosome 4 is late replicating [42], stains darker than the short arm by C-banding [41], and is reported to also harbor 5S rDNA [43, 44]. In the current study, 25 BAC clones mapped uniquely to the heterochromatic region of LG chromsome 4 (e.g., Figure 4). Moreover, five BAC clones that mapped to multiple chromosome locations included this heterochromatic region on LG chromosome 4 (Additional file 3; Figure 2C). This would suggest a complex genome structure of multiple repetitive DNA families in this region on LG chromosome 4, some of which are shared sequences at other chromosomal regions.

We have used flow cytometry and chromosomal mapping to develop a cytogenetic map of the zebrafish genome. These data can be integrated with genetic and radiation hybrid mapping data to produce a refined sequence map and to confirm the existing genomic framework of the zebrafish. A refined zebrafish genome serves as a valuable resource for genetic research that utilizes this model organism for understanding of the molecular basis of human diseases. For example, the mapped clones can be used to narrow the search for candidate genes, which may be altered in zebrafish mutant models using FISH assays. This strategy was recently used to identify a loss of one allele of the separase gene, when a probe containing the separase gene locus was hybridized upon epithelial tumor specimen, indicating that a loss of one copy of separase is sufficient to promote tumorigenesis in the zebrafish [45]. The cytogenetically mapped clones can also be used in the development of a zebrafish array comparative genomic hybridization platform for use in identifying genomic imbalances in developmental and disease zebrafish models. Integrating these BAC clones with the zebrafish reference sequence will allow investigators to directly determine the genomic regions that are gained or lost. Overall these data contribute towards the ultimate defintion of a reference zebrafish genome and provide insight into the evolution of vertebrate chromosomes.