Zebrafish mutants with disrupted early T-cell and thymus development identified in early pressure screen


  • Nikolaus S. Trede,

    Corresponding author
    1. Children's Hospital Boston, Harvard Medical School, Boston, Massachusetts
    2. Huntsman Cancer Institute, Department of Pediatrics, University of Utah, Salt Lake City, Utah
    • Huntsman Cancer Institute, Department of Pediatrics, University of Utah, Salt Lake City, UT 84112
    Search for more papers by this author
  • Tatsuya Ota,

    1. Boston University School of Medicine, Boston, Massachusetts
    Current affiliation:
    1. Department of Evolutionary Studies of Biosystems, School of Advanced Studies and Hayama Center for Advanced Studies, The Graduate University for Advanced Studies, SOKENDAI, Hayama, Kanagawa, 240-0193 Japan
    Search for more papers by this author
  • Hirohide Kawasaki,

    1. Boston University School of Medicine, Boston, Massachusetts
    Current affiliation:
    1. Department of Evolutionary Studies of Biosystems, School of Advanced Studies and Hayama Center for Advanced Studies, The Graduate University for Advanced Studies, SOKENDAI, Hayama, Kanagawa, 240-0193 Japan
    Search for more papers by this author
  • Barry H. Paw,

    1. Children's Hospital Boston, Harvard Medical School, Boston, Massachusetts
    2. Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Tammisty Katz,

    1. Huntsman Cancer Institute, Department of Pediatrics, University of Utah, Salt Lake City, Utah
    Search for more papers by this author
  • Bradley Demarest,

    1. Huntsman Cancer Institute, Department of Pediatrics, University of Utah, Salt Lake City, Utah
    Search for more papers by this author
  • Sarah Hutchinson,

    1. Huntsman Cancer Institute, Department of Pediatrics, University of Utah, Salt Lake City, Utah
    Search for more papers by this author
  • Yi Zhou,

    1. Children's Hospital Boston, Harvard Medical School, Boston, Massachusetts
    2. Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Candace Hersey,

    1. Children's Hospital Boston, Harvard Medical School, Boston, Massachusetts
    2. Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Agustin Zapata,

    1. Department of Cell Biology, Faculty of Biology, Complutense University, Madrid, Spain
    Search for more papers by this author
  • Chris T. Amemiya,

    1. Boston University School of Medicine, Boston, Massachusetts
    Current affiliation:
    1. Benaroya Institute at Virginia Mason, Seattle, WA 98101
    Search for more papers by this author
  • Leonard I. Zon

    1. Children's Hospital Boston, Harvard Medical School, Boston, Massachusetts
    2. Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author


Generation of mature T lymphocytes requires an intact hematopoietic stem cell compartment and functional thymic epithelium. We used the zebrafish (Danio rerio) to isolate mutations that affect the earliest steps in T lymphopoiesis and thymic organogenesis. Here we describe the results of a genetic screen in which gynogenetic diploid offspring from heterozygous females were analyzed by whole-mount in situ hybridization for the expression of rag-1. To assess immediately if a global defect in hematopoiesis resulted in the mutant phenotype, α-embryonic globin expression was simultaneously assayed for multilineage defects. In this report, we present the results obtained with this strategy and show representative mutant phenotypes affecting early steps in T-cell development and/or thymic epithelial cell development. We discuss the advantage of this strategy and the general usefulness of the zebrafish as a model system for vertebrate lymphopoiesis and thymic organogenesis. Developmental Dynamics 237:2575–2584, 2008. © 2008 Wiley-Liss, Inc.


The development of the immune system in vertebrates involves thymic organogenesis and the differentiation of lymphoid progenitor populations during embryogenesis. The interaction of T cells and thymic epithelial cells during embryogenesis and early postnatal life is of pivotal importance for both T-cell development and thymic epithelial differentiation. The factors involved in T-cell ontogeny have been identified through engineered (Rothenberg et al.,2008) or natural mutations that disrupt T-cell development (Rodewald and Fehling,1998; Fischer,2001; Rothenberg et al.,2008). Such “experiments of nature” are rare, but highly informative and lead to the discovery of novel genes or genes with previously unknown functions in the immune system (Fischer,2002). To gain a more complete picture of genes that participate in the ontogeny of the immune system, particularly if inactivation causes embryonic lethality, model systems are required which allow an unbiased, phenotype-driven approach through forward genetics. The zebrafish (Danio rerio) has emerged as an attractive vertebrate model system. Large- and small-scale zebrafish screens have identified mutants that disrupt virtually all biologic pathways and organ systems. Of particular relevance for the current study are the similarities between zebrafish and higher vertebrates in development of the thymus anlage and T-cell ontogeny as judged by histology and gene expression patterns (reviewed in Trede and Zon,1998; Schorpp et al.,2000; Boehm et al.,2003; Traver et al.,2003; Trede et al.,2004).

A large-scale screen in medaka has recently been published, and in this setting several mutants with aberrant T-cell development have been described (Iwanami et al.,2004). The majority of mutants were characterized by defects in eye and pharyngeal arch development. Similarly, a zebrafish pilot screen by the group of T. Boehm (Schorpp et al.,2000) and a large-scale screen by the same group in collaboration with the European Zebrafish Consortium in Tübingen described mutants with T-cell deficiency, including a mutation in the ikaros transcription factor (Schorpp et al.,2006).

Here, we report the first early pressure screen in zebrafish to identify genes required for early steps in T-cell and/or thymic epithelial cell development. This screen is time- and space-efficient and does not require a large-scale approach. We produced gynogenetic diploid embryos from mutagenized zebrafish to screen for mutations of the lymphoid program directly in the F2 generation. Using a combination of in situ hybridization probes for rag-1 and globin RNA we were able to identify six complementation groups of mutants that affect T-cell and/or thymic epithelial cell differentiation, but not hematopoiesis globally. These mutants fall into two categories based on normal or abnormal pharyngeal arch development. Analysis of gene expression patterns and thymic architecture in these mutants suggests that mutants with abnormal pharyngeal arches have a defect in the formation of the thymic rudiment. Mutants with normal pharyngeal arches have defects either in early events of thymic development or in homing to and/or expansion of pro-T cells in the thymic anlage. The study of these mutated genes will yield insights into early T-cell differentiation and/or thymic organogenesis.


αe1 alpha embryonic globin1 WISH whole-mount in situ hybridization hpf hours postfertilization ENU ethylnitrosourea EP early pressure ICM intermediate cell mass of Oellacher


Design of Screen

Male zebrafish of the AB background were mutagenized and crossed to wild-type AB females. Based on pigment noncomplementation tests, mutagenesis frequency was estimated at 1:1,000, which is in keeping with mutagenesis efficiency observed in other screens (Solnica-Krezel et al.,1994; Haffter et al.,1996). Eggs from 330 heterozygous F1 females were in vitro fertilized with UV inactivated sperm (Table 1) and subjected to Early Pressure (EP). The resulting gynogenetic diploid offspring (see the Experimental Procedures section) were grown to 5 days postfertilization (dpf) and subjected to whole-mount in situ hybridization (WISH) with the rag-1 and αe1 probes (Fig. 1). Mutant phenotypes observed were of the three expected categories. An example of a lymphoid mutant phenotype, which was observed in offspring from 23 females, is given in Figure 1B (I). Erythroid-specific mutants showed up as αe1 negative and rag-1 positive in offspring from 10 females (Fig. 1B II). Putative global hematopoiesis mutants were negative for expression of both, αe1 and rag-1, and were found in three independent females (Fig. 1B III). The frequency of the observed mutant phenotypes in the three categories is given in Table 1. To verify that mutant phenotypes were based on recessive mutations in the germline of F1 females and not secondary to artifacts from the EP or the WISH procedure, F2 incrossing was performed. If the mutant phenotype could not be reproduced in 10 F2 incrosses, the EP result for this potential mutant was considered false positive and the F1 female and its progeny were discarded (see the Experimental Procedures section). To date, we have identified eight recessive rag-1 negative mutants in the F2 generation that fall into six complementation groups. All eight mutations are lethal, with edema starting at 6 dpf and death ensuing between 7 and 9 dpf. Of the remaining 15 potential mutants, we were unable to outcross five F1 females that gave rise to EP progeny with absent rag-1 expression, and could not confirm the rag-1 negative phenotype in the F2 progeny of ten potential mutants. One of the 10 globin defective mutants was confirmed by F2 incrossing, whereas none of the three F1 females giving rise to the potential global hematopoiesis mutants was successfully outcrossed to confirm the phenotype in the F2 generation.

Table 1. Results from Screen
F1 females screenedSib cohorts with >50 larvaePositive in F2 generationaPositive in F3 generationComplementation groups
  • a

    Number of gynogenetic diploid sib cohorts from F1 females with defective gene expression.

  • b

    Number of rag-I negative sib cohorts.

  • c

    Number of αeInegative sib cohorts.

33016523 (rag-Ib)86
  10 (αeIc)1N.A.
  3 (both)0N.A.
Figure 1.

Design of screen and classes of mutants obtained. A:Schematic diagram of screen design. For details, refer to the Results and Experimental Procedure sections. Whole-mount in situ hybridization (WISH) at 5 days postfertilization (dpf) wild-type (wt) larvae with rag-1 and αe1 probes shows expression of rag-1 in bilateral thymi (arrows) and αe1 in the heart and tail (arrowhead). B:Three classes of mutants were observed in the screen. In each panel the wild-type (wt) larvae are on top, the mutant (mut) larvae are on the bottom. I. Lymphoid mutants showed normal expression of αe1 (arrowheads) but absence of rag-1 expression (arrows) compared with wt controls (top panel). II. Erythroid mutants showed normal expression of rag-1, but defective expression of αe1 (middle panel). III. Mutants showing defects in both rag-1 and αe1 expression point to a possible defect in the early hematopoietic compartment (bottom panel).

We named the rag-1 negative complementation groups after teas (for T cell and Thymic mutants, and in agreement with the medaka mutants; Iwanami et al.,2004): assam (asm), chamomile (cam), ceylon (cey), earl grey (egy), jasmine (jas) and oolong (oln). oln mutant larvae are characterized by severe growth delay, microcephaly, and onset of edema at 4 dpf, and were, therefore, excluded from more detailed analysis. We have recently published the phenotype and positional cloning of egy (Trede et al.,2007). Representative staining with rag-1 for the remaining four complementation groups at 5 dpf is shown in Figure 2. rag-1 expression was greatly diminished or absent in cey and asm mutant larvae and completely absent in the remaining mutants.

Figure 2.

Lymphoid mutants recovered from screen. rag-1 whole-mount in situ hybridization (WISH) in 5 days postfertilization (dpf) wild-type (arrowhead, top panel) and mutant (lower panels) larvae assam (asm), chamomile (cam), ceylon (cey), and jasmine (jas). Allele designations are given below mutant names in parentheses.

Phenotypic Analysis of the Mutants Identified in the Screen

We analyzed the steps in T-cell and/or thymic differentiation that could be affected in the mutants. Visible circulating erythrocytes and positive staining with the αe1 probe were present in all six mutant complementation groups, indicating that there was no global defect in erythropoiesis. Markers of early hematopoietic stem/progenitor cells, scl and c-myb, were expressed at comparable levels in mutants and wild-types at 24 and 36 hr (data not shown). We, therefore, conclude that hematopoiesis is globally intact in all mutants.

To define the timing of the block in lymphoid development, expression of the early lymphoid gene ikaros was assayed at 5 dpf. Absence of ikaros expression suggests a very early block in T-cell or thymic development (Rothenberg et al.,2008), while later blocks in T-cell development and potentially also in thymic development leaves ikaros expression intact. Similarly to rag-1, ikaros expression is markedly reduced in cey and asm and absent in the remaining mutants, placing the block in lymphoid development upstream of ikaros, or indicating that the defects affect early steps in thymic organogenesis. In asm, where a small but reproducible signal was observed with the ikaros probe at 5 dpf (Fig. 3, top panel), ikaros expression in the thymus had disappeared by 8 dpf (lower panel).

Figure 3.

Ikaros expression in wild-type and asm mutants. Ventral view of ikaros expression in wild-type (right) and asm mutant (left) lavae at 5 dpf (top panel) and 8 days postfertilization (dpf; lower panel). ikaros expression is seen in wt larvae in the thymic area (arrowhead). In asm mutants, ikaros expression is seen at 5 dpf, but absent at 8 dpf.

Next, we assayed expression of foxn1, currently the only available marker of thymic epithelial cells (TECs) in zebrafish (Schorpp et al.,2002). Expression of foxn1 is first detectable by WISH at 3 dpf as a concentrated signal bilaterally just ventral to the otic vesicle (Schorpp et al.,2002). Subsequently, the area of the signal increases while its intensity decreases, reflecting the dispersion of TECs by infiltration with and proliferation of immature T cells. Staining of a wild-type larva with a foxn1 anti-sense probe is shown at 5 dpf, when dispersion of TECs has already begun (Fig. 4). Representative examples of mutant egy, cey, and asm larvae are shown (Fig. 4). Compared with wild-type, asm mutant larvae maintain the concentrated pattern of foxn1 expression. Interestingly, foxn1 expression was reproducibly discontinuous in egy, a pattern that was not reflected in abnormal endodermal patterning (Trede et al.,2007). Finally, cey appeared to have a more generous foxn1 signal. While the ikaros signal in asm mutants was undetectable at 8 dpf (Fig. 4), foxn1 expression was unchanged at 8 dpf (Fig. 4). Depending on the stage of development, the distinction between a primary thymic and a primary T-cell defect can be difficult because of the interdependence of T cells and TECs for normal development. To further explore possible TEC abnormalities suggested by abnormal foxn1 expression, we examined the thymus of mutant larvae histologically. Histological sections and electron microscopy were carried out at 7 dpf when thymus tissue is sufficiently developed to optimize analysis. Figure 5A shows that while pharyngeal cavity, otic capsule and blood vessels are normally formed in the different mutants, a gradation of thymic abnormalities was observed. The most severely affected TEC compartment was found in egy mutant larvae, where the thymic tissue consists merely of a thin epithelial rim (Trede et al.,2007). cam and cey mutant larvae showed more generous thymic epithelium, albeit with reduced number of cell layers compared with wild-type, with occasional empty spaces possibly evidence of TEC degeneration. The jas mutant larvae showed some organization into a thymic rudiment, but thymic tissue is markedly reduced compared with wild-type. Thymic architecture was closest to normal in asm mutants. Ultrastructural analysis revealed absence of lymphocytes in egy (data not shown), cam, cey, and jas, while in asm occasional immature lymphocytes were observed (Fig. 5B). Architecture of the pharyngeal arches was assessed by alcian blue staining (Fig. 6). Wild-type number of arches were present in asm and jas, albeit in shortened form in the latter. In cam the caudal arches were markedly underdeveloped, similarly to egy (Trede et al.,2007). The most severely affected arches were found in cey, where in addition to shortening of rostral and absence of caudal arches the orientation of arches 3–5 was horizontal. As the thymus arises between pharyngeal arches three and four from the third pharyngeal pouch, we classified the mutants into two groups, in accordance with previously published classifications (Iwanami et al.,2004; Schorpp et al.,2006): Class I mutants have normal pharyngeal arches and in class II mutants pharyngeal arches are abnormal. Table 2 summarizes phenotypic characteristics of mutants in classes I and II.

Figure 4.

foxn1 expression in selected mutants. Lateral view of 5 days postfertilization (dpf; top four panels) and 8 dpf (bottom panels) foxn1 expression in wild-type and mutant larvae at ×20 magnification. Insert shows ×40 magnification of thymic signal. Scale bar = 200 μm. For better mounting and imaging, the left eye was removed from 8 dpf larvae. Note the wide dispersion of the foxn1 signal in 8 dpf wild-type larva (bottom left panel).

Figure 5.

Abnormal thymic architecture in mutant larvae. A: Light microscopic examination of wild-type 7 dpf thymus shows a heterogeneous, cellular thymus (dark arrowhead). Epithelium of the otic capsule is indicated by grey arrowhead. Thymus of cey and cam mutant larvae shows reduction in thymic epithelial cell layers and empty spaces (arrows). Magnification was ×100 for all panels. B:Electron microscopy of 7 dpf wt thymus (top panel) shows a heterogeneous population of cells. Arrowheads indicate basement membrane. Insert (×54,000) in right lower corner shows desmosomes (D, arrow) between thymic epithelial cells. Occasional lymphoblasts were only found in asm mutants. Empty spaces suggesting degeneration in the thymic epithelial cells were encountered in cey and cam mutants (asterisks). Magnification was ×4,000 for all panels. Ch, chloride cell, a salt-producing cell occurring in the epithelium of pharyngeal cavity of many fish species; Fb, fibroblast; Lb, lymphoblast; Mu, mucine producing cell; Ph, pharyngeal epithelial cell; TEC, thymic epithelial cell.

Figure 6.

Abnormalities in pharyngeal arch architecture in mutant larvae. Architecture of pharyngeal arches by alcian blue staining shows presence of seven fully chondrified arches in 7 days postfertilization (dpf) wild-type larvae (indicated by arrows, p1 to p7). Various defects are seen in cey and cam mutants, while asm resembles wt morphology and jas has a complete set of shortened arches.

Table 2. Phenotypic Characteristics of rag-I Deficient Mutants
Complement GroupAllelesGene expressionPharyngeal ArchesAdditional FeaturesaMap position linkage group
  • a

    mc, microcephaly; mo, microphthalmia

Class I        
assamCZ-20posnegneg7normalswimbladder inflated mild mc and mo18
jasmineCZ-18negnegneg7slightly shortenedmild mc and mo13
Class II        
ceylonCZ-26negnegneg5deformedmoderate mc and mo3
chamomileCZ-32negnegnegneg5reducedmoderate mc and mo1
earl greyCZ-3, 5, 11negnegneg5reducedmoderate mc and mo5
oolongCZ-22negnegneg4markedly reducedgrowth delay, severe mc and mo19

To determine whether endodermal patterning is affected in the mutants, we examined expression of nkx2.3, which stains endoderm in the pharyngeal pouches. At 2 dpf, nkx2.3 expression in wild-type embryos is detected in the five pharyngeal pouches (between pharyngeal arches 2 and 7) and in the gut tube. Expression in selected class I (asm) and class II (cey) mutant embryos was equivalent to wild-type in both expression domains, indicating normal endodermal patterning in mutant embryos (Fig. 7).

Figure 7.

nkx2.3 expression in wt and selected mutant embryos. In dorsal view of 2 days postfertilization (dpf) wild-type (wt; top panel) and mutant (lower panels) two nkx2.3 expression domains are observed: pharyngeal pouches (red arrow) and gut tube (arrowhead).

Chromosomal Mapping of the Identified Mutants

All eight mutants were crossed to each other and six complementation groups were established. The egy complementation group contains three independent alleles, while all other complementation groups contain a single mutant. The mutated genes were assigned to a chromosomal location for all complementation groups by half tetrad analysis or genome-wide scanning. asm was mapped to linkage group (LG) 18, cey to LG 3, cam to LG 1, jas to LG 13 and oln to linkage group 19 (Table 2). All three mutants in the egy complementation group map to the same locus on LG 5. The egy mutation in CZ-3, CZ-5, and CZ-11 disrupts recycling of the U4/U6 splicing complex and is caused by a 50-kb insertion into the p110/sart3 gene, suggesting a founder effect that predated mutagenesis.


Thymic organogenesis and T-cell commitment, hallmarks of adaptive immunity, are initiated at the end of embryogenesis and are unique to vertebrates. In this study, we describe the first early pressure forward genetic screen for lymphoid defects in zebrafish. This type of screening provides a new tool to the field of immunology. Our results characterize mutations that disrupt normal T-cell and/or thymic epithelial cell development.

Advantages and Challenges Using Early Pressure Genetic Screening in Zebrafish

In conventional saturating forward genetic screens, 5,000 F2 families of heterozygous mutants (five times coverage of the genome) are raised and mutant phenotypes are detected by F2 incrossing in the F3 generation (Mullins et al.,1994; Solnica-Krezel et al.,1994). Our strategy involved the use of gynogenetic diploid embryos using EP (Streisinger et al.,1981; Beattie et al.,1999). This procedure has the advantage of allowing immediate visualization of mutant phenotypes in the F2 generation. As this approach obviates the need to raise large numbers of families, it is particularly well suited for addressing specific aspects of a developmental process, where the number of genes involved is restricted. This type of screen can also be carried out in laboratories where the logistics do not allow for large-scale screens. The disadvantages of this procedure reside in the suboptimal recovery of viable embryos and the nonmendelian distribution of mutant phenotypes after the EP procedure (Beattie et al.,1999). As shown in Table 1, only 50% of the clutches had more than 50 viable embryos, the number required for detecting at least two mutants in a clutch with 95% confidence (Beattie et al.,1999). Small clutch size may be caused by early lethal mutations, reduced egg viability in F1 females, and epigenetic effects from the EP procedure (Streisinger et al.,1981). It is possible that mutations were missed in clutches with low number of individuals, because the frequency of mutation detection in telomeric regions is only approximately 5% (Streisinger et al.,1986). This suggests that we have confidently reached one-thirtieth saturation (165 F1 females with >50 offspring analyzed) with our current screen. Given that we have identified 6 complementation groups of mutants, a total of 150–180 genes are expected to affect T-cell and/or thymic organogenesis to a degree that their inactivation leads to absent rag-1 expression. This number may be an underestimate of all genes that lead to an immune defect because we may have missed mutations that lead to reduced, but not absent rag-1 expression. Such phenotypes are observed in the mouse by disrupting several genes, including Pax-1 (Wallin et al.,1996), IL-7 (von Freeden-Jeffry et al.,1995), IL-7 receptor (Peschon et al.,1994), p56lck (Molina et al.,1992), and pTα (Fehling et al.,1995). Furthermore, we may have missed incompletely penetrant or hypomorphic mutations of genes that lead to a rag-1 negative phenotype if completely disrupted. In a recent screen in medaka, an F2 screen for nonexpression of rag-1, 22 mutations were identified in 538 F2-families screened (Iwanami et al.,2004). As this screen reached approximately 1/9 saturation, the estimate for the total number of genes affecting T-cell and/or thymic development is 200. In addition, our estimate closely matches the observed frequency of rag-1 deficiency in a saturation screen carried out by the group of T. Boehm in association with the 2000 European Consortium, where 141 rag-1 deficient mutants were identified (Schorpp et al.,2006). Recovery of mutants in our screen was approximately 35% (8 of 23 potential mutants confirmed in the F2 generation). Failure to outcross F1 females and false positive results after the EP procedure are the most plausible explanation for this finding. Despite nonsaturating conditions in the present screen, the egy mutation was detected in offspring of three independent F1 females. This suggests that this mutation was likely present in the germline of the males before mutagenesis, and was confirmed by positional cloning of the affected p110/sart3 gene (Trede et al.,2007).

Characterization of Mutants

Mutants with absent rag-1 expression identified in this screen fall into six complementation groups that can be further subdivided into two broad categories: Mutants with normal pharyngeal arches and mutants with abnormal pharyngeal arches (Table 2). These mutants likely affect the early steps in the homing to and/or expansion of pro-T cells in the thymus (class I) and early steps in the formation of the thymic anlage (class II). Mutants with normal pharyngeal arches could have a leaky phenotype, where small numbers of pro-T cells are attracted to the thymic anlage, but cannot expand. These mutants could have been missed in our screen because a positive rag-1 signal, albeit weak, is expected to be present.

In class I mutants pharyngeal arch architecture is grossly normal, so that neural crest contribution to thymic organogenesis is expected to be normal. Normal expression of nkx2.3 also suggests an intact endodermal compartment. While thymic stroma in these mutants is mildly (asm) to markedly decreased (jas), lymphoblasts are undetectable by electronmicroscopy in jas. The defect in this group of mutants could, therefore, primarily affect thymic stromal development. In this scenario thymic epithelial cells are unable to differentiate to a point where they are capable of attracting pro-T cells efficiently, or to allow for expansion of attracted pro-T cells. Alternatively, defects in the T-cell compartment can also severely affect the competence of pro-T cells to home to the thymus or their capacity to expand once having reached the thymic rudiment. For example, mutations in the GATA-3 gene (Ting et al.,1996), the Ikaros gene (Georgopoulos et al.,1994), or the c-kit−/−γc−/− mice (Rodewald et al.,1997) lead to complete absence or severe reduction of pro-T cells in the thymic rudiment. In all these cases, the thymic anlage remains in a rudimentary state, characterized by cyst formation and lack of corticomedullary differentiation (Rodewald and Fehling,1998; van Ewijk et al.,2000). If class I mutants have a T-cell autonomous defects, the affected gene is expected to act at an early stage in thymocyte maturation. The asm phenotype most closely resembles that of the FOXN1 deficient nude mouse. In the latter, Ikaros-positive cells are observed in the mesenchyme surrounding the thymic anlage, but they do not enter into the anlage that fails to proliferate and grow (Itoi et al.,2001). However, the murine FOXN1 mutation does not cause an eye phenotype or gestational lethality. The disappearance of ikaros positive cells in asm mutants at 8 dpf also points to a difference to the nude mouse phenotype. In support of this conclusion, the thymic rudiment forms normally in asm, as evidenced by normal initial foxn1 expression. In conclusion, the rudimentary state of the thymic anlage in class I mutants may be primary or secondary to absent T-cell/thymic epithelial cell interaction.

Embryologically class II defects could arise from a defect in neurectoderm or a lack of neural crest/endoderm interaction, as described for example in the Hoxa-3 −/− mice (Manley and Capecchi,1995; Su and Manley,2000) or in sucker zebrafish. In the latter example postmigratory neural crest response to endodermal signals is missing secondary to a mutation in the endodermal endothelin-1 gene (Miller et al.,2000). In each of these models, the integrity of the pharyngeal arch architecture is compromised to varying degrees. This mirrors the phenotypes of the class II mutants described in our screen. The fact that the caudal pharyngeal arches, which critically depend on neural crest cell (NCC) contribution, are abnormal in class II mutants, suggests that the primary defect in this group may reside in abnormal migration or function of NCCs. This conclusion is further supported by the observed abnormalities in craniofacial skeleton of class II mutants. Despite these abnormalities, the thymic rudiment forms in all of the mutants analyzed, as evidenced by expression of the TEC marker foxn1.

Recent work has demonstrated that in zebrafish normal eye development is a prerequisite for normal anterior NCC migration (Langenberg et al.,2008). Hence, a primary eye defect could explain craniofacial abnormalities observed in our microphthalmia mutants, even if NCCs were not primarily affected. The association of abnormalities in thymic development with microphthalmia is an unexpected finding in our screen, as no such correlation has been described in mammals. Interestingly, in a recent screen for T-cell and/or thymic abnormalities in medaka, a similar correlation was described, where 21 out of 22 mutants with absence or reduction of rag-1 had microphthalmia (Iwanami et al.,2004). As both, retina and thymic epithelium are highly proliferative tissues, an underlying mutation could be a rate-limiting factor in rapidly cycling cells. Given that production of erythrocytes, which relies on high cell-turnover, is normal in all mutants, such a mutation is not expected to be a general cell cycle factor. This conclusion is bolstered by the recent cloning of the egy mutation (Trede et al.,2007), where disruption of the p110/sart3, a U4/U6 recycling factor essential in rapidly proliferating cells (Medenbach et al.,2004), leads to tissue-specific defects, including eye and thymus epithelium. Additionally, genes that influence patterning of both thymus and eye in fish could be affected.

Isolation of the Mutant Genes

All six complementation groups of mutants have already been mapped to a chromosomal region on the zebrafish genome (Table 2). Closely linked markers (between 0.5 and 4 cM) have been identified for these mutations. No obvious candidate genes have so far been identified. As the cloning of the egy mutation demonstrates, the isolation of the mutant genes may reveal novel genes or novel functions of known genes that disrupt normal thymic and/or T-cell development. Taken together, the above data establish zebrafish screens for lymphoid mutants as complementary to murine genetic approaches, and as a valuable additional tool in the analysis of critical developmental steps in the vertebrate adaptive immune system.


Zebrafish Embryos

Zebrafish were maintained as described in (Mullins et al.,1994). Developmental stages at 28.5°C were determined by embryo morphology (Kimmel et al.,1995). The studies described in this study have been reviewed and approved by the Children's Hospital Institutional Review Board. For in situ hybridization experiments beyond 24 hours postfertilization (hpf), embryos were incubated in 0.003% 1-phenyl-2-thiourea (PTU, Sigma Chemical Co, St. Louis, MO) at 19 to 36 hpf and maintained in PTU thereafter until they were fixed in 4% paraformaldehyde. The wild-type AB strain was obtained from C. Kimmel (Eugene, OR). The WIK strain was obtained from C. Nüsslein-Volhard (Tübingen, Germany).


Males of the AB strain were treated with the chemical mutagen ethylnitrosourea (ENU) as described (Solnica-Krezel et al.,1994). Efficiency of ENU mutagenesis was estimated at 1:1,000 by noncomplementation of the pigment deficient golden indicator locus (data not shown). Mutagenized males were then crossed to wild-type females of the AB strain and F1 heterozygous females were raised. Progeny of 330 F1 females were analyzed for mutant phenotypes.

Early Pressure

Eggs were extruded by mechanical pressure from F1 females and fertilized with ultraviolet-inactivated sperm. Second meiotic division was inhibited by application of 8,000 lbs/sq.in. 80 sec after fertilization as described (Streisinger et al.,1981). Pressure was maintained for 280 sec and then slowly released. Resulting gynogenetic diploid embryos were then grown for 5 days at 28°C, fixed in 4% paraformaldehyde and subjected to whole-mount in situ hybridization.

Recovery of Mutant Lines

To verify the mutant phenotype in their F2 gynogenetic diploid offspring, F1 females were outcrossed to wild-type AB individuals. Their putatively heterozygous F2 offspring were then incrossed and the F3 generation was fixed and analyzed for a mutant phenotype at 5 dpf as described above. If 10 random F2 incrosses failed to result in F3 offspring with the mutant phenotype observed in the F2 gynogenetic diploid offspring, the putative mutation was scored as “false positive.” This is based on the fact that the chance of a false negative result (missing a mutation that is actually present in the F2 family) after performing 10 independent crosses is calculated to be 5.6 percent using the binomial distribution (where n = 10, k = 0, and P = 0.25).

Isolation of cDNA Clones

The isolation of the ikaros (Willett et al.,2001) cDNAs was described elsewhere. The rag-1 and rag-2 probes were amplified from zebrafish genomic DNA based on published sequences (Willett et al.,1997). The foxn1 probe (Schorpp et al.,2002) was a gift from T. Boehm.

WISH, Histology, and Alcian Blue Staining

Digoxigenin-labeled antisense and sense RNA probes were synthesized as described (Harland,1991). Whole-mount in situ hybridizations were performed according as described (Thompson et al.,1998) with the following modifications: First, permeabilization of embryos 30 hr or older and larvae (2 days and older) was carried out for 20 min in 20 mg/ml proteinase K (Roche Molecular Biochemicals, Indianapolis, IN). Younger embryos were not subjected to proteinase K treatment. Second, embryos were blocked in 10% heat-treated lamb serum, and 2% BMB block (Boehringer Mannheim, Indianapolis, IN) in MABT (100 mM maleic acid/150 mM NaCl/ 0.01% Tween-20/ pH 7.5). Third, signal was detected with BMB Purple substrate (Roche Molecular Biochemicals). Embryos were mounted in 95% glycerol in phosphate buffered saline (PBS) for microscopy. Larvae for histology were fixed, sectioned and stained as described (Willett et al.,1997).

For alcian blue staining, 7 dpf larvae were fixed overnight in 4% paraformaldehyde. Three washes in PBST (PBS with 0.1% Tween) were followed by overnight staining of larvae in 0.1% alcian blue/1% concentrated HCl/70% EtOH at room temperature.

Mapping of Mutated Genes

Mapping strains were generated by mating F1 heterozygous AB females with polymorphic WIK males. Linkage analysis for asm, egy, cam, and jas was performed by half tetrad analysis (Johnson et al.,1996) using gynogenetic diploid larvae from AB/WIK heterozygous females. DNA from phenotypic wild-type and mutant larvae was extracted as described (Zhang et al.,1998). Linkage to a zebrafish chromosome was established using centromeric CA markers that are polymorphic between the AB and WIK strains. Linkage was confirmed in each case by testing several polymorphic markers on the chromosome that was initially identified on the DNA of diploid mutant and wild-type larvae. The cey and oln mutations were mapped using diploid larvae derived from heterozygous parents and assaying 239 polymorphic SSLP markers (9 to 10 per chromosome) for differences between mutant and wild-type larvae.


N.S.T. was funded by NIH (K08 HL004233) and the Huntsman Cancer Foundation, and L.I.Z. was funded by the NIH.