Coupled Mutagenesis Screens and Genetic Mapping in Zebrafish

Corresponding author: Stephen L. Johnson, 4566 Scott Ave., St. Louis, MO 63110., sjohnson{at}genetics.wustl.edu (E-mail)

Communicating editor: D. J. GRUNWALD

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Forward genetic analysis is one of the principal advantages of the zebrafish model system. However, managing zebrafish mutant lines derived from mutagenesis screens and mapping the corresponding mutations and integrating them into the larger collection of mutations remain arduous tasks. To simplify and focus these endeavors, we developed an approach that facilitates the rapid mapping of new zebrafish mutations as they are generated through mutagenesis screens. We selected a minimal panel of 149 simple sequence length polymorphism markers for a first-pass genome scan in crosses involving C32 and SJD inbred lines. We also conducted a small chemical mutagenesis screen that identified several new mutations affecting zebrafish embryonic melanocyte development. Using our first-pass marker panel in bulked-segregant analysis, we were able to identify the genetic map positions of these mutations as they were isolated in our screen. Rapid mapping of the mutations facilitated stock management, helped direct allelism tests, and should accelerate identification of the affected genes. These results demonstrate the efficacy of coupling mutagenesis screens with genetic mapping.

GENETIC screens provide an invaluable tool in the identification of genes required for specific cellular and developmental processes. In animals, this approach has been most successful in the Caenorhabditus elegans and Drosophila melanogaster model systems, where large-scale genetic screens have revealed mechanisms underlying many aspects of animal development and physiology. In contrast, forward genetic analysis of biological processes in vertebrates has traditionally been an exceptionally strenuous task. This problem has been addressed recently through the successful use of zebrafish (Danio rerio) in a growing number of genetic screens (DRIEVER et al. 1996 ; HAFFTER et al. 1996A ; HENION et al. 1996 ; for review, see PATTON and ZON 2001 ). These screens have resulted in an impressive collection of mutants defective for a variety of cellular and developmental processes, and the identification of genes affected in these mutants is providing new insights into vertebrate biology. Additionally, the pursuit of saturation mutagenesis and the creation of new techniques to assay biological processes are now compelling new generations of zebrafish mutant screens (JAGADEESWARAN et al. 2000 ; FARBER et al. 2001 ; PATTON and ZON 2001 ).

This wealth of mutations is accompanied by the challenges of mutant stock management and identification of affected genes. These tasks are greatly facilitated by identifying the map position of individual mutations. Mapping of new mutations facilitates stock management through genotyping of heterozygous carriers, candidate gene identification, and testing for allelism to previously identified mutant loci. Therefore, the development of techniques that accelerate the mapping process should have widespread benefits for zebrafish genetics.

The need for more rapid mapping techniques is exemplified in the field of zebrafish pigment pattern biology for two principal reasons. First, the salient nature of pigment pattern phenotypes has resulted in the identification of several hundred zebrafish pigment pattern mutations (JOHNSON et al. 1995B ; HAFFTER et al. 1996B ; HENION et al. 1996 ; KELSH et al. 1996 ; ODENTHAL et al. 1996 ; RAWLS and JOHNSON 2001 ; this work). Second, our incomplete understanding of zebrafish pigment pattern biology compels further genetic analysis. Among the multiple cell types that comprise the zebrafish pigment pattern, the melanocyte (also called melanophore in poikilotherms) is the only pigment cell type common to both fish and mammals. Similar to other vertebrates, the melanocytes that comprise the zebrafish pigment pattern are derived from the neural crest, and the melanocytes of the pigmented retinal epithelium (PRE) are derived from the optic cup. Like other neural crest derivatives, melanocytes must be properly specified from neural crest progenitor cells, migrate to appropriate positions in the body to compose the pigment pattern, and persist through subsequent stages of development. Melanocytes therefore provide a good model for the analysis of cell fate specification, migration, survival, and organogenesis (for review, see RAWLS et al. 2001 ). Furthermore, an understanding of melanocyte development in zebrafish may also provide insights into human melanocyte disorders such as vitiligo, Waarderburg syndrome, and melanoma. As our understanding of the mechanisms underlying zebrafish pigment pattern biology remains incomplete, new genetic screens are needed. The efficiency of new screens and the subsequent analysis of mutants will be greatly increased if researchers can rapidly map the resulting mutations.

Inbred lines provide a valuable tool in the mapping of mutant loci. Good mapping strains require extensive polymorphism between lines and reproducible alleles (e.g., PCR product sizes) within lines. The inbred lines C32 and SJD generally meet these criteria. Each of these lines is >93% homozygous for simple sequence length polymorphism (SSLP) markers (NECHIPORUK et al. 1999 ; S. L. JOHNSON, unpublished results), and more polymorphisms are evident between these lines than between either of them and WIK, another commonly used mapping strain (NECHIPORUK et al. 1999 ). Furthermore, we find extensive single nucleotide polymorphisms (SNPs) between C32 and SJD lines. Initial analysis of expressed sequence tags (ESTs) and cDNAs suggests 1 SNP every 100 bp in noncoding DNA and 1 SNP every 300 bp in coding sequence between C32 and SJD (J. F. RAWLS and A. R. MCADOW, unpublished results). These SNPs will further facilitate the rapid mapping and refinement of meiotic maps in pursuit of positional cloning projects.

In mouse, the availability of SSLP markers that distinguish between haplotypes of different inbred strains has facilitated the management of F₃ mutagenesis screens and initial mapping of mutations derived from them (KASARSKIS et al. 1998 ; HERRON et al. 2002 ). Here, we describe a zebrafish mutagenesis screen that similarly incorporates inbred lines and genomewide mapping using SSLP markers. We improve upon the approaches of KASARSKIS et al. 1998 and HERRON et al. 2002 by performing our initial genetic mapping in the same F₂ haploid embryos in which the mutant phenotype is first observed. This is made possible by the relatively large clutch sizes usually obtained from zebrafish. Thus, by coupling the initial phenotypic screening with genetic mapping, we avoid the need to perform additional crosses for the initial mapping of the mutation, thereby substantially accelerating the mapping process and facilitating mutagenesis screen management.

Several methods exist for initial mapping of zebrafish mutations (POSTLETHWAIT et al. 1994 ; JOHNSON et al. 1995A ; SINGER et al. 2002 ), and in this study we describe a method to improve upon one of them. We first describe a panel of SSLP markers polymorphic between SJD and C32 at 20-cM intervals throughout the zebrafish genome. We conducted a small-scale F₂ haploid screen for new mutants defective for embryonic melanocyte development (WALKER 1999 ) and then used the marker panel in a bulked-segregant genome scan using DNA from F₂ haploid individuals to rapidly identify the candidate map positions of these new mutants (MICHELMORE et al. 1991 ). These candidate map positions were then confirmed by recombination analyses on individual haploids. Further, the identification of SSLP markers closely linked to mutant loci (on SJD haplotypes) provided surrogate markers for identification of carriers following introgression and stock management in the C32 background. The map positions indicated by this work also allowed us to test for and identify allelism between several of our new mutations and known loci. We show that by using the C32 and SJD inbred lines and this panel of polymorphic markers, new zebrafish mutations can be rapidly and accurately mapped as they are generated through mutagenesis screens. The incorporation of this technique into future zebrafish mutagenesis screens should accelerate the rate of initial mutant mapping, simplify stock management, and help focus subsequent investigations.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Inbred lines:
The original SJD inbred line was derived from a natural population collected in 1989, passed through two generations of full sibling matings, through two sequential generations of early pressure parthenogenesis (JOHNSON et al. 1995A ), and finally through seven additional generations of full sibling matings, resulting in >98% homozygosity across the female meiotic map (see http://zfish.wustl.edu). Because original SJD stocks in our colony are predominantly male, we are currently introgressing genes from the C32 genetic background that dominantly promote female development into the SJD genetic background. We are carrying this introgression through eight sequential generations of backcross to the original SJD strain. We have several parallel lines that produce 25–50% females at the fifth or sixth backcross generation, breed well by natural crosses, and also easily yield eggs and sperm following squeezing for in vitro fertilization. We can currently provide fish from our SJD female introgression lines (six generations of sequential backcross, estimated <5% C32). A similar introgression of genes from SJD has been performed to enhance vigor in C32 stocks, and we can provide fish from our ninth sequential backcross generation (requests to S. Johnson).

Mutant screening:
Adult SJD male zebrafish were mutagenized with N-ethyl-N-nitrosourea (ENU) according to SOLNICA-KREZEL et al. 1994 . Mutagenized SJD males (10) were bred with C32 females, and the resulting F₁ progeny were reared at 28.5°. F₁ stocks derived from different mutagenized males were sometimes mixed together, and thus we cannot always rule out the possibility that multiple isolates at the same loci are not the same allele. Clutches of eggs from individual F₁ females were collected, and each clutch was split in half. Half of each clutch was fertilized with sperm of the C32 haplotype to produce F₂ outcross lines, and half of the clutch was fertilized with UV-inactivated sperm to produce F₂ haploid embryos for phenotypic screening (STREISINGER et al. 1981 ; see Fig 1). When >20 haploid embryos were available in a clutch, we split them between a group allowed to develop at 25° and a second group allowed to develop at 33° (505/604 clutches tested by these means).

View larger version (33K): In this window In a new window Download PPT slide   Figure 1. Design of F2 haploid screen for melanocyte mutants and subsequent bulked-segregant mapping.

Using a dissecting light microscope, we evaluated F₂ haploid embryos at multiple time points between 1 and 4 days postfertilization (dpf) for defects in melanocyte development using wild-type haploid siblings as controls. We considered an F₁ founder successfully screened if at least eight F₂ haploid embryos were produced and evaluated. Haploid clutches displaying defects in melanocyte development in 50% of individuals were identified, and the corresponding clutch of F₂ outcross fish was maintained. All staging of zebrafish embryos was performed according to KIMMEL et al. 1995 . Morphological descriptions of mutant phenotypes in the text are for embryos reared at 25°, except where noted.

Bulked-segregant PCR:
Genomic DNA was isolated from individual mutant haploid embryos and their wild-type haploid siblings into 100 µl TE using standard protocols. Equivalent aliquots from individual embryo DNA preparations were pooled (7–14 individuals per pool) and then diluted 30-fold to generate mutant and wild-type DNA pools for each mutation. Three microliters from each diluted DNA pool was then used as template in 25-µl PCR reactions using primer concentrations of 1 µM. The reactions were amplified using the following protocol: initial denaturation at 92° for 3 min, followed by 36 cycles of 92° for 30 sec, 94° for 5 sec, 55° for 30 sec, and 72° for 90 sec, with a postamplification extension at 72° for 4 min. The resulting amplicons were resolved on 2% agarose gels and visualized with ethidium bromide. Error estimates are reported as standard deviation.

Nomenclature:
We follow the general rules of zebrafish nomenclature for designating locus and allele names. Thus, locus names and abbreviations are registered with ZFIN, and each allele is assigned a unique allele name. Locus names are provisional, because they may correspond to previously identified but still unmapped mutations. Allele designations in this screen begin with j (for Johnson lab), followed by a number to indicate our locus designation (for instance, 1 represents kit and 121 represents chloe), a letter to indicate mutagenesis method (for instance, e represents ENU), and a final number to indicate order of isolation in our allelic series. Reported map positions refer to the MOP panel (JOHNSON et al. 1996 ; see Meiotic Maps at http://zfish.wustl.edu).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of first-pass SSLP markers for bulked-segregant analysis:
To identify SSLP markers for our first-pass panel for bulked-segregant analysis, we took advantage of the 2000 previously published SSLP markers (SHIMODA et al. 1999 ), as well as markers not yet published but available on the Massachusetts General Hospital website (http://zebrafish.mgh.harvard.edu). We first tested the first-pass markers suggested at the above web site, as appropriate. Additionally, we developed 40 SSLP markers derived from repeats identified in our EST project (CLARK et al. 2001 ; see Table 1), which we mapped on the LN54 radiation hybrid panel (HUKRIEDE et al. 2001 ), the MOP haploid panel (JOHNSON et al. 1996 ), and the HS meiotic panel (not shown; KELLY et al. 2000 ). We screened for markers that were 20 cM apart, gave reliable amplification from C32 and SJD DNA, had significant size polymorphism between the C32 and SJD lines (generally >10 bp), and amplified both alleles in heterozygotes (to facilitate bulked-segregant analysis). We screened >500 SSLP markers for meeting the above criteria and selected a minimal panel of 149 markers spaced on average 22 ± 8 cM apart on the female meiotic map or 5 ± 7 cM from the apparent ends of the meiotic linkage groups. In a few cases, we had to accept markers >30 cM apart (intervals 39–77 cM on LG3, 55–95 cM on LG4, 0–32 cM on LG6, 98–141 cM on LG7, 75–112 cM on LG8, 2–43 cM on LG9, 37–68 cM on LG12, 5–39 cM and 55–87 cM on LG14, 6–44 cM on LG20, and 57–100 cM on LG 22). In these intervals, few SSLP markers were available, indicating that these regions have exaggerated rates of meiotic recombination. This suggests that these are relatively small physical regions, a notion that is generally supported by the low number of ESTs mapped to these regions (HUKRIEDE et al. 1999 ; S. L. JOHNSON, unpublished results). Our preferred SSLP markers for first-pass screening by bulked-segregant analysis are listed in Table 2 and are also posted at http://zfish.wustl.edu.

Production and mapping of new melanocyte mutants:
We performed a genetic screen for mutations that affect the development of the embryonic melanocyte pigment pattern (Fig 1). While melanocytes are not directly required for survival, genes required for melanocyte development may also be required for development of other essential neural crest derivatives (see RAWLS et al. 2001 ). To facilitate the identification of melanocyte mutations that are lethal due to defects in neural crest stem cells or other essential neural-crest-derived cell types, we focused our screen on the phenotype of the embryonic melanocyte pattern. Since wild-type haploid zebrafish embryos are viable through 4 dpf and develop a relatively normal melanocyte pattern (WALKER 1999 ), we screened for abnormal melanocyte phenotypes in haploid embryos (see MATERIALS AND METHODS).

We screened F₂ haploid clutches representing 604 mutagenized haploid genomes and found 31 possible melanocyte mutants. Additionally, we added to our analysis two melanocyte mutants (j108e1 and j121e1) identified in a previous early pressure parthenogenesis screen conducted in our lab. Of these mutants, we attempted to map 20 and were able to identify map positions for 19 (95%). Although 2 of our mapped mutants were subsequently lost (j120e1 and j130e1), all others are currently maintained in our lab as breeding stocks or frozen sperm.

Mapping of new mutations was accomplished using bulked-segregant analysis (MICHELMORE et al. 1991 ; see Fig 1). Using selected polymorphic marker primers (Table 2), bulked-segregant PCR was performed on mutant and wild-type DNA pools from each mutant clutch (see MATERIALS AND METHODS). Candidate-linked markers were identified as those in which preferential amplification of the SJD and C32 haplotypes was observed in the mutant and wild-type pools, respectively. Markers that suggested linkage in bulked-segregant PCR were then tested on DNA from the individual mutant and wild-type embryos that comprised the respective DNA pools. If linkage of candidate markers was confirmed on DNA from individual haploid embryos, then markers flanking that candidate marker were similarly tested for polymorphism and linkage to refine the mutant map position. After identifying the map position of an individual mutant, corresponding F₂ diploid outcross fish were genotyped to identify individuals carrying the SJD haplotype at the linked markers. These heterozygous carriers were then intercrossed to produce diploid mutants for phenotypic analysis and to confirm the map position. In some cases, map positions were also confirmed by complementation analysis (see below) or half-tetrad analysis. Map positions of mutations identified in this screen and several other embryonic pigment pattern mutants identified in our lab and mapped in this project are shown in Table 3.

In some cases, mutations with similar phenotypes mapped to the same location. For example, alleles j121e1, j121e2, j121e3, and j121e4 all mapped near gof1 on LG25 and failed to complement each other when carriers were interbred. Similarly, alleles j126e1 and j126e2 mapped near z4421 on LG23 and failed to complement each other. Because these alleles may have been derived from the same mutagenized male, it is not yet clear whether these are indeed independent alleles or multiple isolates of the same allele (except for j121e1, which was derived from a different mutagenized male than j121e2–4). In other instances, new mutants mapped to the same locations as previously identified pigment pattern loci, directing us to test for allelism by complementation analysis. For example, allele j1e249 mapped to z10756 on LG20, the map position of the kit receptor tyrosine kinase gene (also called sparse; PARICHY et al. 1999 ). We found that j1e249 mutants failed to complement the kit^b5 null allele when carriers were interbred, confirming that j1e249 is an allele of kit. Similarly, allele j124e1 was mapped to near z11685 and z7654 on LG18, the recently identified map position of touchtone (B. ARDUINI and P. HENION, personal communication). Interbreeding of j124e1 and touchtone^os1 carriers showed a failure to complement, demonstrating that j124e1 is an allele of touchtone. Finally, j125e1 mapped to z4999 on LG7, near the recently identified map position of endzone (B. ARDUINI and P. HENION, personal communication). In contrast to kit and touchtone, interbreeding of endzone^b431 (HENION et al. 1996 ) and j125e1 carriers revealed complementation, indicating that j125e1 and endzone are different genes.

Melanocyte mutant phenotypes:
Zebrafish melanocytes first become visible beginning around 1 dpf, as expression of melanin pigment begins in the PRE and in neural-crest-derived melanocytes (KIMMEL et al. 1995 ). Some neural-crest-derived melanocytes remain at the dorsal aspect of the neural tube, while others migrate to positions throughout the body to comprise the embryonic melanocyte pattern (Fig 2A and Fig B). This pattern largely persists through the larval stages (MILOS and DINGLE 1978 ) until the onset of larval-adult metamorphosis beginning 14 dpf (JOHNSON et al. 1995B ). The embryonic melanocyte mutants identified and mapped in this work are defective for distinct aspects of melanocyte development, allowing them to be grouped into several phenotypic classes. These phenotypic classes are discussed below, with reference to previously identified zebrafish melanocyte mutants. As most of the previously identified melanocyte mutations remain unmapped, it is unclear whether they are candidate alleles of the new mutations reported here (see DISCUSSION). In naming these new melanocyte mutant loci, we have followed our custom of naming lethal melanocyte mutations after deceased or extinct cats and viable melanocyte mutations after extant cats.

View larger version (57K): In this window In a new window Download PPT slide   Figure 2. Mutations affect distinct aspects of embryonic melanocyte development. Wild-type (WT) embryos at 2 dpf (A) have darkly pigmented neural-crest-derived melanocytes along the dorsal stripe and over the yolk (solid arrowheads). The PRE is also darkly pigmented by 2 dpf (open arrowhead) and continues to develop to contribute to the functional eye organ by 5 dpf (B). Mutations at perruquej126e1 (prq) cause punctate neural-crest-derived melanocytes at 2 dpf (C, solid arrowheads), while the PRE remains faint (open arrowhead) and begins to degenerate by 5 dpf (D). In sourmashj118e1 (srm) mutants, the PRE is largely absent at 2 dpf (E, open arrowhead) and does not recover through 5 dpf (F). chloej121e1 (chl, G) mutants have normal distribution and morphology of neural-crest-derived (solid arrowheads) and PRE melanocytes (open arrowhead), yet they fail to become fully pigmented. The selimaj108e1 (sel, H) mutation causes a severe deficit of neural-crest-derived melanocytes (solid arrowhead) and defective PRE development (open arrowhead). All mutants shown are homozygous. Bars, 400 µm (A, C, E, G, and H) and 200 µm (B, D, and F).

Mutations causing degeneration of all melanocytes: Mutations at perruque (Fig 2C and Fig D) and blatherskite (not shown) cause degeneration of both neural-crest-derived and PRE melanocytes. In these phenotypes, neural-crest-derived melanocytes initially appear smaller than wild type but in a relatively normal distribution, while the PRE initially appears pale (Fig 2C). Both types of melanocytes then degenerate and disappear over subsequent days of development (Fig 2D). The blatherskite^j122e1 allele also shows brain degeneration beginning 3 dpf (not shown). Together, these defects resemble those observed in the previously identified mutants delayed fade, fade out, fading vision, blurred, ivory (KELSH et al. 1996 ), piegus, punktata, and mizerny (MALICKI et al. 1996 ).

Mutations causing deficiencies in PRE melanocytes: Mutant alleles of sourmash (Fig 2E and Fig F) are deficient for PRE melanocytes, but develop a normal pattern of neural-crest-derived melanocytes. This mutant phenotype resembles the previously identified retina mutants oko meduzy, glass onion, nagie oko, and heart and soul (MALICKI et al. 1996 ).

Mutations causing defects in melanin synthesis: Mutations at rubis, chloe, and racan cause deficiencies in melanin synthesis. The rubis^j127e1 allele results in the complete absence of melanin in both neural-crest-derived and PRE melanocytes (not shown). Mutations in chloe (Fig 2G) and racan (not shown) result in substantially delayed and incomplete melanization in both neural-crest-derived and PRE melanocytes. However, the individual morphology of embryonic melanocytes and their resulting pattern are largely normal in these two mutant phenotypes, suggesting that these animals are deficient for melanin synthesis. Since chloe, racan, and rubis map to different linkage groups, they represent distinct loci. The embryonic phenotypes of chloe, racan, and rubis are similar to those caused by previously isolated mutations at albino (CHAKRABARTI et al. 1983 ), golden, brass (STREISINGER et al. 1986 ), casper (HENION et al. 1996 ), lead, mustard, nickel, pewter, sandy, and brassy (KELSH et al. 1996 ). Interestingly, all alleles of chloe isolated in this study are viable and result in adults displaying severe melanocyte and iridophore deficiencies.

Mutations causing reduction of neural-crest-derived melanocytes: Mutations at pinky cause a substantial reduction in the number of neural-crest-derived melanocytes, in addition to severe anterior-posterior body axis defects (not shown). We identified one mutant, selima, that lacked virtually all neural-crest-derived melanocytes. Homozygous selima mutants develop pale PRE and display a number of other pleiotropic defects after 2 dpf (Fig 2H). These melanocyte deficiency phenotypes resemble those of previously identified mutations at endzone (HENION et al. 1996 ), colorless, white tail (KELSH et al. 1996 ), nacre (LISTER et al. 1999 ), and alyron (CRETEKOS and GRUNWALD 1999 ).

Mutations affecting melanocyte morphology: Neural-crest-derived melanocytes in wild-type embryos usually appear stellate with uniform pigmentation (Fig 3A and Fig B). We identified two mutants, cheshire^j129e1 (Fig 3C and Fig D) and touchtone^j124e1 (not shown), that develop melanocytes with varying morphologies and pigmentation. In these phenotypes, neural-crest-derived melanocytes appear either pale with normal stellate morphology or punctate with apparently normal melanization. Through the first week of development, some of the punctate melanocytes appear to be extruded through the epidermis. These phenotypes resemble those seen in previously identified mutant alleles at touchtone (B. ARDUINI and P. HENION, unpublished results), cold-light, polished, freckles, and punkt (KELSH et al. 1996 ). As mentioned above, allele j124e1 was identified as an allele of touchtone first by our mapping efforts and subsequently by complementation testing.

View larger version (85K): In this window In a new window Download PPT slide   Figure 3. Mutation at cheshire causes variable melanocyte morphologies. Wild-type (WT) neural-crest-derived melanocytes on the dorsal aspect of the head at 2.5 dpf (A) and in the dorsal stripe at 6 dpf (B) are all stellate and darkly pigmented. Homozygous cheshirej129e1 (chs) mutants at 2.5 dpf (C) and 6 dpf (D) show both punctate darkly pigmented melanocytes (solid arrowheads) and faintly pigmented stellate melanocytes (open arrowheads) within the dorsal stripe. Bars, 200 µm.

Mutations affecting melanocyte migration: We identified three mutant alleles, serpolet^j123e1 (not shown), tammany^j120e1 (not shown), and kit^j1e249 (Fig 4), which were defective for melanocyte migration. In these phenotypes, neural-crest-derived melanocytes largely failed to migrate properly away from the dorsal aspect of the neural tube. The kit^j1e249 allele was found to be temperature sensitive, with homozygotes reared at 25° developing a wild-type melanocyte pattern (Fig 4A and Fig B) and those reared at 33° displaying melanocyte migration defects at 2.5 dpf (Fig 4C). In contrast to the kit null allele phenotype, where neural-crest-derived melanocytes begin to die by 4 dpf and are largely absent by 7.5 dpf (see PARICHY et al. 1999 ), melanocytes in kit^j1e249 homozygotes reared at 33° persist through 7.5 dpf (Fig 4D). These melanocyte migration phenotypes resemble those of previously isolated mutant alleles at kit (PARICHY et al. 1999 ) and sparse-like (KELSH et al. 1996 ).

View larger version (48K): In this window In a new window Download PPT slide   Figure 4. Temperature-sensitive mutation at kit affects melanocyte migration. When reared at 25°, neural-crest-derived melanocytes in kitj1e249 (kit) mutants migrate normally into ventral positions by 2.5 dpf (A, open arrowheads) and persist in these ventral locations through 7.5 dpf (B, open arrowheads). In contrast, neural-crest-derived melanocytes in kitj1e249 mutants reared at 33° fail to migrate to these ventral positions by 2.5 dpf (C), yet persist through 7.5 dpf (D). All mutants shown are homozygous. Bars, 200 µm.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Powerful forward genetics is one of the principal advantages of the zebrafish model system. However, managing mutant lines, mapping mutations, and positionally cloning the affected genes remain laborious tasks. Therefore, an important challenge facing zebrafish workers is to develop techniques to simplify and to accelerate these processes. Among the approaches that might help meet these challenges are the development of highly polymorphic inbred lines, the derivation of those polymorphisms into reliable markers, and the exploitation of those polymorphic markers to expedite and enhance the mapping process. Here we have taken advantage of the polymorphism between the C32 and SJD inbred lines by developing a minimal first-pass panel of SSLP markers for these strains and then using that panel in bulked-segregant analysis to rapidly map a set of new melanocyte mutants.

The ability to distinguish between alleles from different strains is essential to all common mutant mapping approaches in zebrafish. DNA sequence polymorphisms present between mapping strains and the markers derived from them generally provide this ability. For example, bulked-segregant analysis depends on the ability to assess linkage of mutations to polymorphic markers located along each linkage group (MICHELMORE et al. 1991 ). Our work identifies a minimal panel of SSLP markers between SJD and C32 that can be used in bulked-segregant analysis to rapidly identify map positions of new mutations. Importantly, the mapping approach described here is relatively simple and inexpensive, allowing it to be performed even in small laboratories with modest budgets.

The intention of this work was to develop a rapid and reliable technique for determining the map position of new zebrafish mutations as they are identified in mutagenesis screens. By coupling mutagenesis screens to mapping of the resulting mutants, we hoped to facilitate several subsequent tasks. First, the rapid mapping of new mutations simplifies the task of stock management. By identifying SSLP markers closely linked to a mutant locus, recognition of heterozygous carriers can be accomplished through PCR-based genotyping, instead of through more laborious progeny testing. In our study, this method of carrier identification allowed us to quickly identify F₂ outcross siblings heterozygous at markers linked to the respective mutation.

Second, the rapid mapping of new mutations helps to direct allelism testing against previously identified mutant loci. Previous mutagenesis screens in zebrafish have produced thousands of mutant lines, many with similar phenotypes. After isolating new mutations, zebrafish workers are then challenged to distinguish between new mutations at previously identified loci and mutations at novel loci. Since these mutant lines are often maintained in separate laboratories, often on different continents, complementation testing of all possible permutations is a nontrivial task. Identification of map locations first can be used to unambiguously exclude the possibility of allelism for many mutations (see below). Because initial mapping in zebrafish has become relatively inexpensive (especially compared to costs of exchanging stocks between labs), this provides a useful alternative to the complementation test in the management of mutant screens. In those instances in which mutations with similar phenotypes do map to the same genomic region, the complementation test is then compelled to determine whether the mutations affect the same gene. Because few mutations in zebrafish are mapped, the identification of map positions for mutations identified here will provide candidate positions for those pigmentation mutations with similar phenotypes that remain unmapped. We also imagine that coupled mutagenesis and mapping, as described here, can be used in new mutagenesis screens to rapidly select for mutations at previously uncharacterized loci.

To demonstrate the efficacy of this approach, we subjected the new mutations described here to the analysis discussed above. After initial mapping, we found that several mutations mapped to the same genomic regions as known pigment pattern loci. We then performed complementation analyses to test for allelism. These tests revealed that j124e1 is a new allele of touchtone and that j1e249 is a new allele of kit. Complementation in crosses between carriers for pinky^j125e1 and endzone, which map to a similar region on LG7, tends to exclude allelism of these mutations. Similarly, identification of mutations with similar map positions within this screen compelled complementation tests that revealed that j121e1, j121e2, j121e3, and j121e4 are all alleles of the chloe locus and that j126e1 and j126e2 are both alleles of the perruque locus. (Note that we assign allele names after mapping and complementation tests are completed; see MATERIALS AND METHODS.)

By comparing the map positions of new mutations to previously identified mutations with similar phenotypes, we were also able to exclude a substantial fraction of these known mutations as potentially allelic to the new mutations described here. These results are discussed below.

We identified several mutations that affect melanin synthesis, including rubis (LG17), chloe (LG25), and racan (LG21 at 22 cM). While several melanin synthesis mutations have been previously reported, the majority of these can be excluded as possibly allelic to our new mutations on the basis of their map positions. Specifically, the melanin synthesis mutations albino (LG21 at 52cM), brass (LG13; POSTLETHWAIT et al. 1994 ), golden (LG18; see http://zfin.org), and sandy (LG15; W. STAUB and H. BAIER, personal communication) do not map to the same genomic regions as the new mutations described here, while the map positions of casper (HENION et al. 1996 ), lead, mustard, nickel, pewter, and brassy (KELSH et al. 1996 ) remain unknown. The map positions of rubis, chloe, and racan therefore provide candidate positions for the remaining unmapped melanin synthesis mutations listed above or for additional melanin synthesis mutations isolated in the future.

Mutations at sourmash (LG21) were found to cause PRE defects similar to several previously described mutations. That PRE mutations nagie oko and heart and soul map to different linkage groups (LG17 and LG2, respectively; HORNE-BADOVINAC et al. 2001 ; WEI and MALICKI 2002 ) excludes the possibility that sourmash is an allele of either of those genes. It remains possible that other, yet unmapped, PRE mutations, such as glass onion or oko meduzy (MALICKI et al. 1996 ), correspond to sourmash.

In addition to the new kit allele (j1e249) identified in this screen (see above), we identified and mapped mutations at two additional loci that affected migration of neural-crest-derived melanocytes: serpolet (LG5) and tammany (LG22). Since only one other melanocyte migration mutation has been identified (sparse-like; KELSH et al. 1996 ), our map positions for serpolet and tammany provide two candidate map locations for sparse-like and identify at least one novel locus required for melanocyte migration.

Two mutations identified in this screen, pinky (LG7 at 55 cM) and selima (LG7 at 125–141 cM), caused reduction in the number of neural-crest-derived melanocytes. Several mutations that cause reduction in neural-crest-derived melanocytes have been identified and mapped, including colorless (LG3; DUTTON et al. 2001 ), nacre (LG6; LISTER et al. 1999 ), alyron (LG15; CRETEKOS and GRUNWALD 1999 ), and endzone (LG7 at 55 cM; B. ARDUINI and P. HENION, personal communication; see above for discussion of nonallelism between endzone and pinky). Therefore, selima and pinky provide candidate positions for the remaining unmapped mutation described in this class, white tail (KELSH et al. 1996 ).

In addition to the new touchtone (LG18) allele (j124e1) identified in this screen, we identified a mutation at one other locus, cheshire (LG9), that results in mixed melanocyte morphologies. This is similar to mutations at cold light, polished, freckles, and punkt (KELSH et al. 1996 ). Since map positions for these mutations have not been reported, it remains unclear whether cheshire identifies a novel locus or instead provides a candidate map position for these genes.

We also identified and mapped mutations at two loci, blatherskite (LG22) and perruque (LG23), that cause melanocyte degeneration phenotypes similar to previously identified mutations at delayed fade, fade out, fading vision, blurred, ivory (KELSH et al. 1996 ), piegus, punktata, and mizerny (MALICKI et al. 1996 ). Since map positions have not been reported for these mutations, it remains unclear whether blatherskite and perruque identify new loci or may provide candidate positions for previously identified members of this class.

Finally, the rapid initial mapping of new mutations will accelerate the identification of candidate genes. With the availability of dense gene maps (GEISLER et al. 1999 ; BARBAZUK et al. 2000 ) and the imminent completion of the zebrafish genome, the map position of a new mutation could quickly yield a list of relevant candidate genes in the genomic region. This technique could also be used to screen for mutations linked to specific candidate genes of interest.

Although our efforts described here should help accelerate the initial mapping of mutations, improvements are still needed in this and subsequent steps. For example, increasing the density of reliable SSLPs and SNPs between C32 and SJD should accelerate further resolution of the mutant map positions suggested by initial mapping efforts. A potential problem in mapping may arise from our reciprocal introgression schemes to remediate sex ratios in SJD or lack of vigor in C32 inbred strains (see MATERIALS AND METHODS). This scheme may result in residual blocks of introgressed donor DNA in the host chromosomes unrelated to the selected trait. Thus, in crosses between the new C32 and SJD lines, there may be some regions that are not informative. The number and extent of these regions will likely be quite small, as the effect of sequential backcrosses is to serially dilute out the unselected donor DNA. If there is no effect of selection on the maintenance of donor DNA, we would expect the level of donor contamination summed between these two lines to be <1% (<0.2% SJD DNA in the invigorated C32 strain following nine sequential backcrosses and <0.4% donor C32 DNA in the feminized SJD lines following eight sequential backcrosses). In practice, the effect of selection for vigor or for enhanced sex ratios may be to retain more heterozygosity, resulting in a somewhat higher amount of donor DNA in the introgression lines. Such heterozygosity can be detected by PCR-based screens, using the first-pass panel described here, or ultimately, by resequencing the genomes of each strain (for instance, see MITRA and CHURCH 1999 ). The problem of residual donor DNA may also be remediated by using crosses to a third strain, such as WIK, in subsequent mapping experiments. WIK, albeit not inbred, is highly polymorphic with both C32 and SJD (NECHIPORUK et al. 1999 ).

We expect that the success of this approach with respect to SJD and C32 inbred lines will prompt development of additional inbred lines. Additional first-pass marker panels specific to different pairs of genetic backgrounds may also extend the usefulness of this approach, further facilitating the rapid mapping of mutations and management of mutagenesis screens. Importantly, such reagents will be useful in managing not only haploid screens as described here, but also early pressure parthenogenesis screens (JOHNSON and WESTON 1995 ; HENION et al. 1996 ) or F₃ intercross screens (DRIEVER et al. 1996 ; HAFFTER et al. 1996A ).

	FOOTNOTES

¹ Present address: Department of Molecular Biology and Pharmacology, Washington University School of Medicine, 660 So. Euclid Ave., St. Louis, MO 63110.

	ACKNOWLEDGMENTS

The authors thank Colleen Boggs, Sarah Callier, and Steven Jacobs for fish husbandry. The authors are especially grateful to Wendy Straub and Herwig Baier for sharing unpublished map data, and to Brigitte Arduini and Paul Henion for sharing unpublished map data and performing complementation tests. This work was supported by National Institutes of Health grant R01-GM56988.

Manuscript received October 18, 2002; Accepted for publication November 26, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BARBAZUK, W. B., I. KORF, C. KADAVI, J. HEYEN, and S. TATE et al., 2000  The syntenic relationship of the zebrafish and human genomes. Genome Res. 10:1351-1358.[Abstract/Free Full Text]

CHAKRABARTI, S., G. STREISINGER, F. SINGER, and C. WALKER, 1983  Frequency of gamma-ray induced specific locus and recessive lethal mutations in mature germ cells of the zebrafish, Brachydanio rerio.. Genetics 103:109-124.[Abstract/Free Full Text]

CLARK, M. D., S. HENNIG, R. HERWIG, S. W. CLIFTON, and M. A. MARRA et al., 2001  An oligonucleotide fingerprint normalized and expressed sequence tag characterized zebrafish cDNA library. Genome Res. 11:1594-1602.[Abstract/Free Full Text]

CRETEKOS, C. J. and D. J. GRUNWALD, 1999  alyron, an insertional mutation affecting early neural crest development in zebrafish. Dev. Biol. 210:322-338.[Medline]

DRIEVER, W., L. SOLNICA-KREZEL, A. F. SCHIER, S. C. F. NEUHAUSS, and J. MALICKI et al., 1996  A genetic screen for mutation affecting embryogenesis in zebrafish. Development 123:37-46.[Abstract]

DUTTON, K. A., A. PAULINY, S. S. LOPES, S. ELWORTHY, and T. J. CARNEY et al., 2001  Zebrafish colourless encodes sox10 and specifies non-ectomesenchymal neural crest fates. Development 128:4113-4125.[Abstract/Free Full Text]

FARBER, S. A., M. PACK, S.-Y. HO, I. D. JOHNSON, and D. S. WAGNER et al., 2001  Genetic analysis of digestive physiology using fluorescent phospholipid reporters. Science 292:1385-1388.[Abstract/Free Full Text]

GEISLER, R., G. J. RAUCH, H. BAIER, F. VAN BEBBER, and L. BROBETA et al., 1999  A radiation hybrid map of the zebrafish genome. Nat. Genet. 23:86-89.[Medline]

HAFFTER, P., M. GRANATO, M. BRAND, M. C. MULLINS, and M. HAMMERSCHMIDT et al., 1996a  The identification of genes with unique and essential functions in the development of the zebrafish, Danio rerio.. Development 123:1-36.[Abstract]

HAFFTER, P., J. ODENTHAL, M. C. MULLINS, S. LIN, and M. J. FARRELL et al., 1996b  Mutations affecting pigmentation and shape of the adult zebrafish. Dev. Genes Evol. 206:260-276.

HENION, P. D., D. W. RAIBLE, C. E. BEATTIE, K. L. STOESSER, and J. A. WESTON et al., 1996  Screen for mutations affecting development of zebrafish neural crest. Dev. Genet. 18:11-17.[Medline]

HERRON, B. J., W. LU, C. RAO, S. LIU, and H. PETERS et al., 2002  Efficient generation and mapping of recessive developmental mutations using ENU mutagenesis. Nat. Genet. 30:185-189.[Medline]

HORNE-BADOVINAC, S., D. LIN, S. WALDRON, M. SCHWARZ, and G. MBAMALU et al., 2001  Positional cloning of heart and soul reveals multiple roles for PKC lambda in zebrafish organogenesis. Curr. Biol. 11:1492-1502.[Medline]

HUKRIEDE, N. A., L. JOLY, M. TSANG, J. MILES, and P. TELLIS et al., 1999  Radiation hybrid mapping of the zebrafish genome. Proc. Natl. Acad. Sci. USA 96:9745-9750.[Abstract/Free Full Text]

HUKRIEDE, N., D. FISHER, J. EPSTEIN, L. JOLY, and P. TELLIS et al., 2001  The LN54 radiation hybrid map of zebrafish expressed sequences. Genome Res. 11:2127-2132.[Abstract/Free Full Text]

JAGADEESWARAN, P., M. GREGORY, S. JOHNSON, and B. THANKAVEL, 2000  Haemostatic screening and identification of zebrafish mutants with coagulation pathway defects: an approach to identifying novel haemostatic genes in man. Brit. J. Haematol. 110:946-956.[Medline]

JOHNSON, S. L. and J. A. WESTON, 1995  Temperature-sensitive mutations that cause stage-specific defects in zebrafish fin regeneration. Genetics 141:1583-1595.[Abstract]

JOHNSON, S. L., D. AFRICA, S. HORNE, and J. H. POSTLETHWAIT, 1995a  Half-tetrad analysis in zebrafish: mapping the ros mutation and the centromere of linkage group I. Genetics 139:1727-1735.[Abstract]

JOHNSON, S. L., D. AFRICA, C. WALKER, and J. A. WESTON, 1995b  Genetic control of adult pigment stripe development in zebrafish. Dev. Biol. 167:27-33.[Medline]

JOHNSON, S. L., M. A. GATES, M. JOHNSON, W. S. TALBOT, and S. HORNE et al., 1996  Centromere-linkage analysis and consolidation of the zebrafish genetic map. Genetics 142:1277-1288.[Abstract]

KASARSKIS, A., K. MANOVA, and K. V. ANDERSON, 1998  A phenotype-based screen for embryonic lethal mutations in the mouse. Proc. Natl. Acad. Sci. USA 95:7485-7490.[Abstract/Free Full Text]

KELLY, P. D., F. CHU, I. G. WOODS, P. NGO-HAZELETT, and T. CARDOZO et al., 2000  Genetic linkage mapping of zebrafish genes and ESTs. Genome Res. 10:558-567.[Abstract/Free Full Text]

KELSH, R. N., M. BRAND, Y.-J. JIANG, C.-P. HEISENBERG, and S. LIN et al., 1996  Zebrafish pigmentation mutations and the processes of neural crest development. Development 123:369-389.[Abstract]

KIMMEL, C. B., W. W. BALLARD, S. R. KIMMEL, B. ULLMANN, and T. F. SCHILLING, 1995  Stages of embryonic development of the zebrafish. Dev. Dyn. 203:253-310.[Medline]

LISTER, J. A., C. P. ROBERTSON, T. LEPAGE, S. L. JOHNSON, and D. W. RAIBLE, 1999  nacre encodes a zebrafish microphthalmia-related protein that regulates neural-crest-derived pigment cell fate. Development 126:3757-3767.[Abstract]

MALICKI, J., S. C. F. NEUHAUSS, A. F. SCHIER, L. SOLNICA-KREZEL, and D. L. STEMPLE et al., 1996  Mutations affecting development of the zebrafish retina. Development 123:263-273.[Abstract]

MICHELMORE, R. W., I. PARAN, and R. V. KESSELI, 1991  Identification of markers linked to disease-resistance genes by bulked segregant analysis: a rapid method to detect markers in specific genomic regions by using segregating populations. Proc. Natl. Acad. Sci. USA 88:9828-9832.[Abstract/Free Full Text]

MILOS, N. and A. D. DINGLE, 1978  Dynamics of pigment pattern formation in the zebrafish, Brachydanio rerio. I. Establishment and regulation of the lateral line melanophore stripe during the first eight days of development. J. Exp. Zool. 205:205-216.

MITRA, R. D. and G. M. CHURCH, 1999  In situ localized amplification and contact replication of many individual DNA molecules. Nucleic Acids Res. 27:e34.[Abstract/Free Full Text]

NECHIPORUK, A., J. E. FINNEY, M. T. KEATING, and S. L. JOHNSON, 1999  Assessment of polymorphism in zebrafish mapping strains. Genome Res. 9:1231-1238.[Abstract/Free Full Text]

ODENTHAL, J., K. ROSSNAGEL, P. HAFFTER, R. N. KELSH, and E. VOGELSANG et al., 1996  Mutations affecting xanthophore pigmentation in the zebrafish, Danio rerio.. Development 123:391-398.[Abstract]

PARICHY, D. M., J. F. RAWLS, S. J. PRATT, T. T. WHITFIELD, and S. L. JOHNSON, 1999  Zebrafish sparse corresponds to an orthologue of c-kit and is required for the morphogenesis of a subpopulation of melanocytes, but is not essential for hematopoiesis or primordial germ cell development. Development 126:3425-3436.[Abstract]

PATTON, E. E. and L. I. ZON, 2001  The art and design of genetic screens: zebrafish. Nat. Rev. Genet. 2:956-966.[Medline]

POSTLETHWAIT, J. H., S. L. JOHNSON, C. N. MIDSON, W. S. TALBOT, and M. GATES et al., 1994  A genetic linkage map for the zebrafish. Science 264:699-703.[Abstract/Free Full Text]

RAWLS, J. F. and S. L. JOHNSON, 2001  Requirements for the kit receptor tyrosine kinase during regeneration of zebrafish fin melanocytes. Development 128:1943-1949.[Abstract/Free Full Text]

RAWLS, J. F., E. M. MELLGREN, and S. L. JOHNSON, 2001  How the zebrafish gets its stripes. Dev. Biol. 240:301-314.[Medline]

SHIMODA, N., E. W. KNAPIK, J. ZINITI, C. SIM, and E. YAMADA et al., 1999  Zebrafish genetic map with 2000 microsatellite markers. Genomics 58:219-232.[Medline]

SINGER, A., H. PERLMAN, Y. YAN, C. WALKER, and G. CORLEY-SMITH et al., 2002  Sex-specific recombination rates in zebrafish (Danio rerio). Genetics 160:649-657.[Abstract/Free Full Text]

SOLNICA-KREZEL, L., A. F. SCHIER, and W. DRIEVER, 1994  Efficient recovery of ENU-induced mutations from the zebrafish germline. Genetics 136:1401-1420.[Abstract]

STREISINGER, G., C. WALKER, N. DOWER, D. KNAUBER, and F. SINGER, 1981  Production of clones of homozygous diploid zebra fish (Brachydanio rerio). Nature 291:293-296.[Medline]

STREISINGER, G., F. SINGER, C. WALKER, D. KNAUBER, and N. DOWER, 1986  Segregation analysis and gene-centromere distances in zebrafish. Genetics 112:311-319.[Abstract/Free Full Text]

WALKER, C., 1999  Haploid screens and gamma-ray mutagenesis. Methods Cell Biol. 60:43-70.[Medline]

WEI, X. and J. MALICKI, 2002  nagie oko, encoding a MAGUK-family protein, is essential for cellular patterning of the retina. Nat. Genet. 31:150-157.[Medline]

This article has been cited by other articles:

				V. Guryev, M. J. Koudijs, E. Berezikov, S. L. Johnson, R. H.A. Plasterk, F. J.M. van Eeden, and E. Cuppen Genetic variation in the zebrafish Genome Res., April 1, 2006; 16(4): 491 - 497. [Abstract] [Full Text] [PDF]

				J. A. Yoder, R. T. Litman, M. G. Mueller, S. Desai, K. P. Dobrinski, J. S. Montgomery, M. P. Buzzeo, T. Ota, C. T. Amemiya, N. S. Trede, et al. Resolution of the novel immune-type receptor gene cluster in zebrafish PNAS, November 2, 2004; 101(44): 15706 - 15711. [Abstract] [Full Text] [PDF]