During the development of the zebrafish nervous system both noi, a zebrafish pax2 homolog, and ace, a zebrafish fgf8 homolog, are required for development of the midbrain and cerebellum. Here we describe a dominant mutation, aussicht (aus), in which the expression of noi and ace is upregulated. In aus mutant embryos, ace is upregulated at many sites in the embryo, while noi expression is only upregulated in regions of the forebrain and midbrain which also express ace. Subsequent to the alterations in noi and ace expression, aus mutants exhibit defects in the differentiation of the forebrain, midbrain and eyes. Within the forebrain, the formation of the anterior and postoptic commissures is delayed and the expression of markers within the pretectal area is reduced. Within the midbrain, En and wnt1 expression is expanded. In heterozygous aus embryos, there is ectopic outgrowth of neural retina in the temporal half of the eyes, whereas in putative homozygous aus embryos, the ventral retina is reduced and the pigmented retinal epithelium is expanded towards the midline.
The observation that aus mutant embryos exhibit widespread upregulation of ace raised the possibility that aus might represent an allele of the ace gene itself. However, by crossing carriers for both aus and ace, we were able to generate homozygous ace mutant embryos that also exhibited the aus phenotype. This indicated that aus is not tightly linked to ace and is unlikely to be a mutation directly affecting the ace locus. However, increased Ace activity may underly many aspects of the aus phenotype and we show that the upregulation of noi in the forebrain of aus mutants is partially dependent upon functional Ace activity. Conversely, increased ace expression in the forebrain of aus mutants is not dependent upon functional Noi activity. We conclude that aus represents a mutation involving a locus normally required for the regulation of ace expression during embryogenesis.
During early vertebrate development, the nervous system becomes subdivided into discrete regions along its anterior- posterior and dorsoventral axes and within the anterior neural plate, domains of cells give rise to the telencephalon, diencephalon, eyes and midbrain. Many genes are known to be involved in early regional patterning of the anterior central nervous system (CNS) and, in a few cases, the genetic pathways underlying regional patterning are beginning to be unravelled. For instance, transplantation and ablation studies have demonstrated that a group of cells located at the midbrain/hindbrain boundary (MHB) are required for patterning the midbrain and cerebellum (Bally-Cuif and Wassef, 1995). The secreted proteins Fgf8 and Wnt1, and the transcription factors Engrailed (En), Pax2/5/8, Gbx2 and Isl3 are all known or strongly suspected to be involved in the production, reception or propagation of signals from the MHB (Crossley and Martin, 1995; Joyner, 1996; Crossley et al., 1996; Brand et al., 1996; Favor et al., 1996; Kikuchi et al., 1997; Reifers et al., 1998; Lun and Brand, 1998; Pfeffer et al., 1998). However, our understanding of how these genes interact and are themselves regulated remains superficial. In this study, we identify a novel mutation in zebrafish which leads to misregulated expression of ace/fgf8 and noi/pax2.1 both at the MHB region and within the rostral forebrain.
Pax2 belongs to the Paired-box family of transcription factor encoding genes related to the Drosophila segmentation gene paired. Loss-of-function mutations in the zebrafish pax2 homolog, noisthmus (noi), result in loss of the midbrain tectum and cerebellum (Brand et al., 1996; Lun and Brand, 1998). eng and wnt1 expression is reduced or absent in noi mutants (Brand et al., 1996; Lun and Brand, 1998) and, indeed, studies in mice have shown Pax protein binding sites within the promoter of the eng2 gene (Song et al., 1996). In addition, noi mutants show phenotypic defects in the rostral forebrain. In the developing eye, the choroid fissure fails to close resulting in coloboma, a phenotype that is also observed in mice and humans carrying mutations in the Pax2 gene (Sanyanusin et al., 1995; Favor et al., 1996; Torres et al., 1996; Macdonald et al., 1997). Phenotypic defects are also observed in the differentiation of the optic stalks and in guidance of axons across the midline in the postoptic commissure and optic chiasm (Torres et al., 1996; Macdonald et al., 1997).
The phenotype of noi embryos is reminiscent of another zebrafish mutant, acerebellar (ace). Recent studies have shown that ace is a probable loss-of-function mutation in zebrafish fgf8 (Reifers et al., 1998 and see Fürthauer et al., 1997). Fgf8 is a member of the large family of fibroblast growth factors that have been implicated in many aspects of early developmental patterning. In ace mutant embryos, the anlage of the cerebellum does not form and midbrain-like tissue is continuous with the medullary region of the hindbrain (Brand et al., 1996; Reifers et al., 1998). A role for Fgf8 in patterning the midbrain and cerebellum is supported by studies in chick that have shown that exogenously applied Fgf8 can mimic many of the effects of the MHB organiser (Crossley et al., 1996). Similar to noi, ace is also expressed in the rostral forebrain and phenotypic analysis indicates that Ace is also important for patterning this region of the CNS (R. Macdonald, M. Brand, and S. W. W., unpublished observations).
Although ace and noi are both required for correct development of the MHB region, neither gene is required for the initial induction of the other, as ace expression is initially present in noi embryos and noi expression is initially present in ace embryos (Brand et al., 1996; Reifers et al., 1998). However, this changes over time and, within a few hours, noi expression is no longer detectable in ace mutant embryos and ace expression is lost in noi mutants. These results suggest that, within the MHB region, ace and noi are initially activated independently but may later depend upon each other for maintained expression. This conclusion is supported by the observation that both genes are initially expressed in adjacent domains of the neural plate and only later are they co-expressed in cells around the MHB (Reifers et al., 1998). The relationship between Noi and Ace activity in other regions of the embryo has yet to be determined. One further regulator of noi expression within the rostral forebrain is the secreted signalling protein Sonic hedgehog (Shh). In mice lacking Hedgehog signalling in the rostral forebrain, pax2 expression is severely reduced (Chiang et al., 1996) and overexpression of Hedgehog proteins in fish leads to widespread ectopic induction of noi expression (Macdonald et al., 1995; Ekker et al., 1995).
In this study, we analyze a mutation named aussicht (aus) which causes overexpression of both noi and ace. In aus mutant embryos, ace is overexpressed at many sites in the embryo while noi expression is only upregulated within the rostral forebrain and around the MHB, the only sites in the CNS where the expression domains of noi and ace overlap. Furthermore, by analyzing embryos double mutant for aus and ace, we find that the upregulation of noi within aus mutant embryos is at least in part dependent upon functional Fgf8 activity. In contrast, the analysis of aus/noi double mutant embryos shows that loss of Noi activity has little consequence upon the upregulation of ace in aus mutant embryos. Although noi expression is responsive to Shh, the overexpression of noi in aus mutant embryos is unlikely to be due to ectopic Shh activity as we observe no ectopic shh expression in aus mutants.
Fish that give rise to aus mutant embryos also give rise to embryos exhibiting a cell death phenotype suggesting that the aus mutation may represent a balanced translocation. Although the aus mutation is dominant, the phenotype of aus heterozygous and putative homozygous mutants is indistinguishable at early stages of development, with embryos exhibiting a variety of phenotypic alterations including reduced pax6 expression in the ventral retina, delayed formation of the anterior and postoptic commissures and reduced expression of genes within the pretectal area. During subsequent development, some aus heterozygous embryos show ectopic outgrowth of temporal retina while, in suspected homozygous embryos, there is a reduction of ventral retinal tissue and an expansion of the retina towards the midline. In the midbrain of aus mutant embryos, the expression of wnt1 and eng is increased. Based upon the known functions of noi and ace, the majority of the defects in aus mutants are consistent with what one might predict from increased activity of these two genes.
Analysis of embryos mutant for both aus and noi and for aus and ace indicates that aus is unlikely to constitute a gain-of- function allele of either noi nor ace. We therefore conclude that aus is likely to represent a mutation affecting a novel locus involved in the regulation of noi and ace expression during embryonic development of the zebrafish nervous system.
MATERIALS AND METHODS
Maintenance of fish
Breeding fish were maintained at 28.5°C on a 14 hour light/10 hour dark cycle. Embryos were collected by natural spawning and were staged according to Kimmel et al. (1995).
Fish lines and genetics
The aust294 allele described in this paper was found in a fish whose male parent had been mutagenized with ENU. aus represents a mutation with a dominant but not fully penetrant phenotype and we were able to identify adult aus heterozygous carriers that appear phenotypically wild type and were fertile in both sexes (for a more detailed description and discussion of the genetics of the aus mutation see Results and Discussion). To generate double mutant aus/noi and aus/ace embryos, we identified fish carrying both mutations in the progeny of a cross between aus/+ and noitu29a/+ and aus/+ and aceti282a/+, respectively. Since embryos heterozygous for aus and homozygous for noi or ace showed characteristics of both single mutant phenotypes, we were able to phenotypically identify them from early pharyngula stage onwards.
Where percentages are cited in the results, they are always based upon analysis of more than 200 embryos.
Whole-mount antibody labeling and in situ hybridization
Standard procedures were used for both antibody and in situ labeling as described by Hammerschmidt and Nüsslein-Volhard (1993). For antibody staining, the Vectastain detection kit was used. The anti-Pax6 antibody (Macdonald et al., 1994) was diluted 1:400, the anti-En antibody (Developmental Hybridoma Bank) 1:25 and the anti- acetylated alpha tubulin antibody (Sigma) 1:1000. Antisense digoxigenin-labelled RNA probes were synthesised using the digoxigenin RNA labeling kit (Boehringer Mannheim). As templates, full-length pax2.1, fgf8, islet-1, zash-1b, ephrin-A-l2, ephrin-A-rtk2 and ephrin-A-rtk7 cDNAs, and 0.8 kb netrin-1a, 1.6 kb shh and 0.58 kb wnt-1 cDNA fragments were used. The in situ hybridisation staining was detected using BM-purple substrate (Boehringer Mannheim), embryos were then fixed in 4% paraformaldehyde in phosphate-buffered saline for 1 hour, washed in phosphate-buffered saline, cleared in 70% glycerol and mounted on a glass slide.
Embryos were dehydrated in 100% ethanol, embedded in JB4 resin (Agar Scientific Ltd), and sectioned (15 μm) using a tungsten knife on a Jung 2055 Autocut.
Genetics of the aussicht (aus) mutation
aust294 is a dominant mutation with a partially penetrant phenotype in which transcriptional regulation of noi (pax2.1) and ace (fgf8) is disturbed (see below). The aus phenotype was originally observed in a cross between a fish whose male parent had been mutagenised with ethylnitrosurea and a heterozygous carrier for an unrelated mutation. Subsequent matings of the fish carrying the aus mutation to wild-type fish showed that the aus phenotype is dominant. In crosses between a carrier of aus and a wild-type fish, approximately 12% of embryos show the aus phenotype and 18% exhibit extensive cell death in the CNS (Table 1). The cell death phenotype (degeneration, degt294a) represents a second dominant mutation invariably carried by fish that possess the aus mutation. We were able to raise both female and male fish that carried aus to adulthood and, in subsequent generations, these fish also turned out to be carriers of the deg mutation. The expression of two separate lethal phenotypes, each in less than 25% of embryos, from viable fertile adult fish is characteristic of a balanced translocation (Morgan et al., 1925; Talbot et al., 1998 – see Discussion).
In crosses between two heterozygous carriers of aus, there were four different phenotypes: the aus and deg phenotypes were each visible in approximately 20% of embryos (Table 1). During subsequent development, aus and deg mutant embryos could be subdivided into about half showing a phenotype equivalent to embryos originating from crosses between a heterozygous carrier and a wild-type fish, while the other half exhibited a related but more severe phenotype (see below and data not shown). The two additional phenotypes may represent the homozygous conditions for aus and deg mutations since both phenotypes were only detectable in crosses between two heterozygous carriers but not in crosses between a heterozygous carrier and a wild-type fish. In the following analysis of the aus phenotype, we name both putative aus heterozygous and homozygous mutant embryos as ‘aus mutant embryos’ up until the pharyngula stage since we are not able to phenotypically distinguish between putative heterozygous and homozygous mutants at these stages. At older stages, when the phenotypes are distinct, we refer to the likely homozygous mutants as ‘putative homozygous aus mutant embryos’.
noi (pax2.1) and ace (fgf8) are mis-regulated in aus mutant embryos
The most obvious aspects of the aus phenotype in living embryos and sectioned tissue were enlarged optic stalks and expanded optic recesses (Fig. 1C,D). We have previously shown that the zebrafish Pax2 ortholog, Noi, is required for correct differentiation of the optic stalks (Macdonald et al., 1997) and so we examined expression of this gene in aus mutant embryos. noi expression in the optic stalks and at the midbrain-hindbrain boundary (MHB) of aus mutants was expanded into the retina and the posterior midbrain respectively (Fig. 2 and see Fig. 8). Moreover, in some aus mutants, there was an ectopic patch of noi expression in the anterior telencephalon (Fig. 2F). Other domains of noi expression (spinal cord and pronephric duct) appeared normal in aus embryos.
The ace mutation has some similar phenotypic consequences to mutations in noi (Brand et al., 1996) and recent analysis has shown that ace is a mutation in the fgf8 gene (Reifers et al., 1998). We therefore examined the expression of ace in aus mutant embryos. Similar to noi, ace expression in the optic stalk area and around the MHB of aus embryos was expanded into the anlage of the retina and the posterior midbrain, respectively, from early somite stages (Fig. 3D,F). However, in contrast to noi, ace expression was elevated at other sites at which ace is normally expressed in aus mutants (Fig. 3D,F). The degree of upregulation varied substantially depending on the site of ace expression. While ace expression is expanded in the tailbud, there is little or no upregulation within the somites (Fig. 3D). The tailbud appeared broader while the somites were morphologically unaffected and there was no obvious change in her1, myoD and Myosin expression in the somites of aus mutant embryos (Fig. 3D and data not shown). On the basis of morphology alone, we were not able to identify aus mutant embryos until late somite stages, but from as early as gastrula stages, some embryos showed increased ace expression with respect to similarly labelled siblings (Fig. 3A,B). We presume that these embryos would be likely to exhibit the aus phenotype at later stages of development. Beyond the pharyngula stage, levels of ace expression gradually returned to normal in aus mutants (Fig. 3G,H) indicating that there is recovery in this aspect of the phenotype over time.
Since noi and ace are both upregulated in the optic stalk region and around the MHB of aus mutant embryos, we focussed our subsequent analysis upon the development of the eyes and MHB of aus embryos.
The aus mutation affects differentiation of the eye
In crosses between heterozygous aus carriers and wild-type fish, approximately 70% of mutant embryos that exhibited large optic stalks and an expanded optic recess at pharyngula stage, showed a temporal outgrowth of retinal tissue during subsequent development (Fig. 4B). Sections through these eyes showed that the retinal outgrowth is an outfolding of an otherwise normally layered neural retina (data not shown). The remaining ∼30% of embryos that had shown enlarged optic stalks and expanded optic recesses at earlier stages were later superficially indistinguishable from wild-type siblings, again indicating that the aus phenotype can recover over time.
Crosses between two heterozygous aus carriers gave rise to mutants exhibiting two differing eye phenotypes. Of the embryos that showed expanded optic recesses at pharyngula stage, about 45% later showed a temporal outgrowth of retinal tissue as described above (Fig. 4B). In a further 45% of the mutant embryos, the eyes and, in particular, the ventral/nasal part of the retina, were reduced (Fig. 4C) and at later stages pigmented retina extended from the back of the eye towards the midline (Fig. 4D,E). This was always accompanied by a failure of the choroid fissure to close (coloboma; Fig. 4G). Since this phenotype was only detectable in mutant embryos from crosses between two heterozygous aus carriers, we believe that it represents the homozygous aus phenotype whereas the temporal outgrowth of the retina also seen in crosses between heterozygous aus carriers and wild-type fish is likely to represent the heterozygous aus phenotype. The remaining 10% of the embryos that showed expanded optic recesses at pharyngula stage, recovered over time and were later indistinguishable from wild-type embryos.
Somewhat surprisingly, 10% of embryos from crosses between two heterozygous aus carriers, which were phenotypically wild type at pharyngula stage, showed slightly smaller eyes in which retinal tissue extended out of the back of the eye during subsequent development (data not shown). As this phenotype was also not observed in crosses between aus carriers and wild-type fish, it may represent a weaker expressivity of the homozygous phenotype that is only detectable at later stages of development.
Altered gene expression in the retinas of aus mutant embryos
To address the alterations in gene activity that might underlie the morphological abnormalities in the eyes of aus mutants, we examined the expression of various genes known or suspected to be involved in patterning of the eye. Islet1 is expressed in the first neurons that differentiate in the developing eyes (Dorsky et al., 1996). In aus mutant embryos, there was a reduction of islet1 expression within the ventral/nasal half of the retina suggesting that neuronal differentiation is delayed (Fig. 5A,G). Within the developing optic vesicles, noi and pax6 are expressed in complementary domains with noi expression restricted to the optic stalks and cells lining the choroid fissure and pax6 expression restricted to the retina (Macdonald et al., 1995, 1997). In aus mutant embryos, ectopic noi expression within the ventral/nasal part of the eyes (Fig. 5B,H) was accompanied by a downregulation of pax6 expression in a similar region (Fig. 5C,I), indicating an involvement of aus in the regulation of both of these Pax genes.
Eph receptors and their ligands, the ephrins, are expressed in discrete domains of the developing eye and have been shown to be involved in cell-to-cell interactions required for patterning various tissues in the embryo (Drescher et al., 1997). ephrin-A-l2, a GPI-linked ephrin, is expressed in nasal retina, whereas the Eph receptor eph-A-rtk2 is expressed in the temporal retina of wild-type embryos. In some of the aus mutant embryos showing much enlarged optic recesses, ephrin-A-l2 expression within the nasal retina was upregulated and expanded into the temporal half of the eye whereas eph- A-rtk2 expression within the temporal retina was reduced (Fig. 5D,E,J,K). These observations raise the possibility that aus may be involved in regulation of nasotemporal patterning within the developing retina.
The netrins are a family of secreted proteins that are involved in axonal guidance within the CNS. Within the eye, Netrin activity is required for guidance of axons out of the choroid fissure (Deiner et al., 1997) and we have previously shown that netrin expression in the retina is dependent upon functional Noi (Macdonald et al., 1997). In all aus mutant embryos, there was an increase in net1a expression around the choroid fissure and ectopic expression throughout much of the ventral retina (Fig. 5F,L).
As mentioned above, some of the aus mutant embryos that exhibit visible phenotypic defects at early stages later recover and are superficially indistinguishable from wild-type siblings. Supporting the conclusion that the aus phenotype can recover over time, we found that changes in the expression pattern of islet1, pax6, ephrin-A-l2, eph-A-rtk2 and net1a became less prominent over time (data not shown).
aus mutant embryos exhibit commissural pathway defects in the rostral forebrain
We have previously shown that commissural axonal guidance is perturbed in embryos lacking functional Noi protein (Macdonald et al., 1997). As noi expression is upregulated in aus mutant embryos, we examined axonal patterning and gene expression in the vicinity of the rostral commissures in mutant embryos. Staining embryos with an axonal differentiation marker shows that both the anterior and postoptic commissure have not formed in aus mutants by the early pharyngula stage (Fig. 6A,B). Other pathways in the forebrain and midbrain appeared relatively normal, although there was some axonal disorganization in the hindbrain (not shown). Over time, the rostral commissural defects show recovery such that by prim- 25 stage, axons have crossed the midline in both commissures (Fig. 6C,D). By 3 days postfertilization (protruding mouth stage), the anterior commissure appeared relatively normally positioned and fasciculated whereas the postoptic commissure/optic chiasm was defasciculated in some of the heterozygous and homozygous aus mutant embryos (Fig. 6E- H). In addition, retinal ganglion cell axons were less tightly fasciculated in the retinas of putative homozygous aus mutant embryos, a phenotype that is also observed in noi mutant embryos (Fig. 6G,H) (Macdonald et al., 1997).
To better understand the changes in midline patterning that might underlie the commissural defects, we examined the expression of noi, ace and several other genes potentially involved in midline patterning in pharyngula stage aus mutant embryos. noi is normally expressed in a group of cells ventral to the optic recess which are directly dorsal to the position at which the postoptic commissure forms (Macdonald et al., 1997). In aus mutant embryos, noi expression extended dorsal to the optic recess and into the ventral telencephalon (Fig. 7A,B). Furthermore, ace expression, normally present in a group of cells similar to those that express noi ventral to the optic recess, also crossed the optic recess and was expanded throughout much of the ventral telencephalon of aus mutants (Fig. 7C,D).
net1a encodes a secreted axon guidance protein that is strongly expressed dorsal to the optic recess within the ventral telencephalon. In aus mutant embryos, net1a is ectopically expressed in the optic stalks and levels of transcripts are increased throughout much of the rostral diencephalon (Fig. 7E,F).
Shh encodes a secreted signalling protein that has previously been shown to promote noi expression in the optic vesicles in fish (Macdonald et al., 1995; Ekker et al., 1995) and be required for pax2 expression in this location in mice (Chiang et al., 1996). However, there was little if any ectopic shh expression in aus mutant embryos suggesting that the aus mutation may lead to ectopic noi expression via a Shh- independent pathway. We also examined the expression of several other genes (eph-A4, eph-A-rtk2, eph-A-rtk7, ephrin-A- l2 and ephrin-B2) that all showed no major change in their expression pattern in this region of the embryonic brain (data not shown).
The failure to establish the anterior and postoptic commissures at pharyngula stage in aus mutants therefore coincided with an expansion of the expression domains of noi, ace and net1a at the commissure-forming region of the midline neuroepithelium.
aus mutants exhibit patterning defects within the midbrain and pretectal area of the forebrain
Both noi and ace mutations were originally identified on the basis of patterning defects around the MHB. As ace is upregulated in this region in aus embryos, we examined whether there are any alterations in patterning of this territory in mutant embryos.
In the midbrain, we examined the expression of noi, En and wnt1 in aus mutant embryos. Both noi and En expression is slightly expanded and levels of wnt1 transcripts appear to be increased in the midbrain, the MHB and the hindbrain of pharyngula stage aus mutants (Fig. 8A-F). One caveat is that En expression does normally change over time at the MHB and so if aus affects temporal aspects of midbrain patterning, this could also contribute to the alterations in En expression in aus mutants.
In noi mutant embryos, pretectal gene expression is altered (M. Brand and others, personal communication) and so we examined expression of four genes characteristic of this region of the diencephalon. In late pharyngula stage aus mutants, the pretectal expression domain of zash1b is absent and eph-A4 and eph-A-rtk7 expression in this same region is reduced (Fig. 8G-J and data not shown). In contrast, pretectal expression of pax6 appeared unchanged in pharyngula stage aus mutant embryos (data not shown). Despite these alterations in gene expression, the nucleus of the posterior commissure located in the pretectum appeared to be normal in mutant embryos (data not shown).
aus is unlikely to be a mutation in the noi or ace genes
The observation that aus is a dominant mutation and that noi and ace are overexpressed in aus mutants raised the possibility that aus represented a gain-of-function allele of one of these genes. To test this possibility, we determined if it is possible to generate embryos mutant for both aus and noi and for both aus and ace. If double mutant embryos for aus and noi or aus and ace can be generated this would indicate that aus is not tightly linked to noi or ace and therefore unlikely to be a gain-of- function allele of one of these genes.
In crosses of a double carrier for aus and ace with a single carrier for ace, approximately 7% of embryos showed a forebrain phenotype (expanded optic recesses and eye defects) similar but weaker to aus and a cerebellar defect similar to ace (Fig. 9E-H). Similarly, in crosses of a double carrier of aus and noi with a single carrier for noi, about 7% of embryos looked similar to aus mutants in the forebrain and lacked the MHB as seen in noi embryos (see below). These results suggest that aus is unlinked to noi and ace and is therefore unlikely to represent a gain-of-function allele of either noi or ace.
The increased expression of noi and ace in aus mutants raised the possibility that an increase in the activity of one gene product mediates the increase in the other. To test this possibility, we examined if there is still an increase of noi expression in aus;ace double mutants and an increase in ace expression in aus;noi double mutants. Expression of ace in aus;noi double mutant embryos was expanded in the eyes and forebrain to a degree similar to aus single mutants (Fig. 9A- D). In contrast, there was only a weak upregulation of noi expression in the forebrain of aus;ace double mutants as compared to aus single mutant embryos (Fig. 9I-L). This indicates that in the forebrain, aus functions independently of noi in upregulating ace expression but is partially dependent on the presence of functional Ace protein for the upregulation of noi. By pharyngula stage, in the midbrain of aus mutants, ace expression is lost in the absence of Noi and noi expression is lost in the absence of Ace, similar to the situation in ace and noi single mutant embryos (Reifers et al., 1998).
Genetics of the aus mutation
In this study, we show that aus is a partially penetrant dominant mutation affecting the regulation of ace and noi expression. All adult carriers of the aus mutation were also carriers of a second dominant mutation, deg, which gives rise to an extensive cell death phenotype. This indicates that fish that carry both mutations are preferentially viable compared to carriers of either mutation alone. This pattern of inheritance of the aus and deg mutations is characteristic of a reciprocal translocation between two chromosomes (Talbot et al., 1998). In such a scenario, the adult carrier of the aus and deg mutations would be a balanced translocation heterozygote that possessed both translocation chromosomes. Gametes from this adult could either inherit both translocation chromosomes (and be a viable carrier of aus and deg), inherit neither and be wild type, inherit one translocation chromosome (leading to the aus phenotype) or the other (leading to the deg phenotype). It has recently been shown that the cycb213 allele represents a reciprocal translocation between LG2 and LG12. Similar to aus, the second phenotype resulting from the translocation is widespread cell death (Talbot et al., 1998). Chromosomal inversions can also sometimes lead to a pattern of inheritance similar to a reciprocal translocation (Klug and Cummings, 1991) and detailed mapping of the aus/deg mutations will be necessary to resolve the genetic defects in carriers of the mutation. Although the aus mutation may involve a chromosomal segment encompassing more than one gene, the specificity of the phenotype suggests that a single affected locus may be responsible for much of the phenotype as appears to be the case for several cyclops alleles and other zebrafish mutations generated by gamma ray mutagenesis (Fritz et al., 1996; Fisher et al., 1997). Conversely, the severity of the deg phenotype suggests severe genetic deficiencies in deg mutant embryos.
aus function in eye development
The aus mutation leads to abnormal development of the eyes in both heterozygous and putative homozygous conditions. At early stages of development, the optic stalks are enlarged while the prospective retina appears morphologically almost normal. However, in putative homozygous aus mutants, the ventral- nasal retina subsequently fails to develop and the choroid fissure remains open causing coloboma. Defects in the ventral eye are less apparent in heterozygous aus mutants in which the most noticeable morphological defect is outgrowth in the temporal retina. We believe that these alterations are at least in part due to expanded Ace and Noi activity within the developing eye as both of these genes are upregulated and ectopically expressed in the developing retina of aus mutants. Furthermore, morphological defects are reduced in the eyes of aus mutants that are homozygous for the ace mutation indicating a partial dependence of the phenotype upon functional Ace protein.
The reduction of Pax6 expression within the ventral/nasal half of the retina in aus mutants may underlie the reduced ventral retinal development observed in some mutant embryos. Pax6 is essential for retinal development, as mice lacking Pax6 activity form an optic vesicle but this vesicle subsequently fails to differentiate to form the retina (Hill et al., 1991). The reduction in Pax6 expression in the ventral retina of aus mutants may be due to expanded Noi activity in this same region of the retina. These two Pax proteins are normally expressed in mutually exclusive domains of the optic vesicle with Noi expression restricted to the optic stalks and Pax6 restricted to the retina (Macdonald et al., 1995). We have previously suggested that Noi might be involved in the suppression of pax6 expression allowing medial optic vesicle tissue to differentiate as optic stalk instead of retina (Macdonald et al., 1995). In support of this possibility, overexpression of shh leads to upregulation of noi throughout much of the optic vesicle, and this is accompanied by suppression of pax6 expression and inhibition of retinal development (Macdonald et al., 1995, Ekker et al., 1995). Furthermore, in mice lacking Pax2 function, retinal tissue appears to spread into the optic stalks, again raising the possibility that Pax2 suppresses pax6 expression (Torres et al., 1996). Finally, regulation of pax6 by Noi/Pax2 could be direct as Pax protein binding sites are present in the promoter of the mouse pax6 gene (Plaza et al., 1993).
Analysis of Eph receptor and ligand expression in aus mutant embryos suggests that aus may interfere with assignment of nasotemporal identity in the retina. In some aus mutants, expression of ephrin-A-l2 is expanded from the nasal half of the retina into the temporal retina whereas, conversely, the expression domain of eph-A-rtk2 within the temporal retina is reduced. As Eph family proteins have been implicated in the assignment of nasotemporal retinal identity (Drescher et al., 1997), these alterations in gene expression suggest that there is an expansion of nasal identity at the expense of temporal identity in aus mutants. The upregulation of fgf8 expression in the rostral CNS of aus mutant embryos may be responsible for the altered character of nasal and temporal retinal tissue as some ace mutant embryos also have some disruption to naso- temporal patterning (A. Picker, C. B., N. Holder and M. Brand, unpublished data).
One aspect of the aus phenotype that we do not yet understand is the expansion of retinal pigment epithelium (RPE) out of the back of the eye. We have previously observed that expansion of noi expression following overexpression of shh is correlated with reduced development of both RPE and neural retina fates (Macdonald et al., 1995). This does not seem to be the case in aus, although we do not know whether the expanded RPE of aus mutants represents cells that have adopted abnormal fates or alternatively whether RPE cells have overproliferated or migrated inappropriately. An expanded RPE phenotype is also seen in embryos in which signalling via the Eph-A-rtk1 receptor is compromised (Xu et al., 1996). In this case, it has been suggested that the phenotype might arise through an abnormal contribution of diencephalic cells to the optic vesicle. A similar phenotype is also observed in zebrafish embryos homozygous for the blowout mutation though once again, the phenotype is not well understood (Karlstrom et al., 1996).
aus function in axonal guidance in the forebrain
In aus mutant embryos, few or no axons cross the midline in the postoptic and anterior commissures at the stages when noi and ace show the greatest degree of overexpression in the forebrain. At later stages, there is substantial recovery of this phenotype such that by 2 days, the anterior commissure appears relatively normal and both the postoptic commissure and optic chiasm are present, although axons remain somewhat disorganised.
Within the forebrain, commissures appear to be established at the boundaries between domains of neuroepithelial cells that express different combinations of regulatory genes. We have previously proposed that one reason that commissures are established at such locations is because domains of cells on each side of the boundary express different combinations of proteins that repulse growth cones (Wilson et al., 1993, 1997). In this way, axons may preferentially extend along each other and form a tightly fasciculated commissure at the interface between the two domains. Analysis of embryos lacking Noi function supports a role for this transcription factor in regulating repulsive properties of a narrow domain of cells dorsal to the postoptic commissure (Macdonald et al., 1997). In noi mutant embryos, growth cones extend among the midline cells that lack Noi activity with the result that axons fail to form a tightly fasciculated commissure and cross the midline in aberrant locations. If Noi does regulate the expression of proteins that confer growth cone repulsive properties to expressing cells, then this could provide an explanation for the delay in commissure formation observed in aus mutants. In wild-type embryos, noi expression is restricted to a narrow band of cells dorsal to the postoptic commissure whereas, in aus mutants, noi is widely ectopically expressed throughout the midline territory within which the anterior and postoptic commissures would normally form. We suggest that this ectopic noi expression may lead to ectopic expression of growth cone repulsive molecules, which render the entire midline tissue of aus mutants impassable to the early commissural axons. At later stages, ectopic expression of ace and noi decreases and more normal midline domains of gene activity may be re-established. Unfortunately, to date there are no identified growth cone repulsive proteins that are known to be downstream of Noi in midline forebrain tissue. Noi does appear to regulate netrin expression within the eye, but perhaps not within the forebrain as midline net1a and net1b expression appear to be unaffected in the absence of Noi activity (Macdonald et al., 1997). While the changes in Noi activity are likely to contribute to the commissural defects of aus mutants, it is also likely that other regulatory pathways will be disturbed. For instance, ace mutant embryos have commissural defects more severe than noi mutants (R. Macdonald, M. Brand and S. W. W., unpublished observations) indicating that Ace may also influence midline commissure formation via Noi-independent pathways.
Putative homozygous aus mutants also exhibit defects in axon guidance out of the eye. Retinal axons still coalesced towards the choroid fissure in the ventral retina but remained defasciculated as they exited the eye. This phenotype is similar to that observed in mouse and fish embryos lacking Pax2/Noi activity (Torres et al., 1996; Favor et al., 1996; Macdonald et al., 1997). The optic nerve head at which retinal axons coalesce as they exit the eye is lined by Pax2/Noi-expressing glial cells and it is likely that defects in these glial cells underlie both the coloboma and axon guidance defects observed in the eyes of noi/pax2 mutant fish and mice. net1a and net1b are expressed in similar cells to Noi around the choroid fissure and expression of these genes is severely reduced in noi mutants suggesting that guidance of axons out of the eye is disrupted due to compromised Netrin signalling. This possibility gains support from analysis of mice in which Netrin activity is compromised and axons do indeed have problems exiting the eye (Deiner et al., 1997). It is surprising that aus mutants, in which both noi expression and net expression are expanded throughout the ventral retina, exhibit phenotypes resembling the loss of function of these genes. However, it may be essential to correctly localise Netrin guidance cues at the optic nerve head and the likely disorganisation of such cues in aus mutants may underlie the fasciculation defects.
aus function in the midbrain
The aus mutation leads to upregulation of ace and, to a lesser extent, noi in the midbrain and from analysis of embryos carrying mutations in noi and ace, it is known that both these genes are required for correct development of the midbrain and cerebellum (Brand et al., 1996). En genes are also important regulators of midbrain and cerebellar development, and it is believed that their graded activity contributes to the polarity of this region of the CNS (Itasaki and Nakamura, 1996). In the absence of Noi function, En expression is reduced or absent (Brand et al., 1996; Lun and Brand, 1998), a result consistent with the presence of two essential Pax protein binding sites in the mouse En-2 gene (Song et al., 1996). En expression is also eventually lost in ace mutant embryos supporting ectopic expression studies in mice and chicks which have shown that ectopic Fgf8 can induce En expression (Crossley et al., 1996; Brand et al., 1996; Lee et al., 1997; Reifers et al., 1998). From these results, it appears that Ace/Fgf8 and Noi/Pax2/5/8 are upstream of En genes in the midbrain and this provides a likely explanation of why En expression is enhanced in aus mutants.
Wnt1 is also required for development of this region of the CNS and this gene is also upregulated in some aus mutants. Wnt1 is required for survival of En-expressing cells in the midbrain (McMahon et al., 1992) and it is believed that it may also be involved in the regulation of fgf8 expression in the rostral metencephalon (Lee et al., 1997). Although these studies suggest that Wnt1 is upstream of fgf8, it is also true that ectopic Fgf8 can induce wnt1 (Crossley et al., 1996). Therefore, it is again possible that the increased wnt1 expression of aus mutants could be due to increased Ace activity. One surprising observation was that wnt1 expression also appeared to be increased in the rostral hindbrain of aus mutants. While this could be an independent effect of the aus mutation, Reifers et al. (1998) have recently shown that the rostral hindbrain is the initial site of ace expression during gastrulation.
In addition to the midbrain alterations in aus mutants, we also observed a reduction or absence of transcripts of several genes expressed in the pretectal region of the diencephalon. As yet we do not know if this is due to aus having activity in the pretectum or alternatively whether it is secondary to the effects in the midbrain. However, changes in pretectal gene expression are observed in noi mutant embryos (M. Brand and others, personal communication), raising the possibility that alterations in midbrain patterning could affect the caudal diencephalon.
aus is unlikely to be a mutation in the ace gene
The observation that aus is a dominant mutation in which ace is overexpressed at many of its sites of expression raised the possibility that aus could be a gain-of-function allele of ace. However, crosses between a carrier of ace and aus and a carrier of ace alone generate embryos that exhibit the mutant phenotypes for both aus and ace suggesting that aus is not tightly linked to ace. For a single embryo from such a cross to exhibit both phenotypes, one could propose that transheterozygous aus/ace embryos exhibit the double mutant phenotype. This is not the case, however, as crosses between a heterozygous carrier of ace and a heterozygous carrier of aus never generate the double mutant phenotype. Thus one can conclude that embryos exhibiting both phenotypes must carry two aceti282a alleles and at least one aus allele. If aus is a translocation allele of the ace gene, then it is indeed possible that single embryos could be aceti282a/aceti282a and aceaus/+. For this genotype to generate the observed double mutant phenotype, one would have to propose that expression of ace from the translocated aus allele is unable to rescue the midbrain phenotype but is sufficient to cause the aus forebrain phenotype. Furthermore, if aus represented a balanced translocation of the ace gene, then some embryos from a cross between two carriers of aus should inherit both deletion chromosomes, lack both copies of the ace gene and have no expression of ace RNA. A loss of ace expression was never observed in such crosses. While we cannot completely exclude the possibility of a translocation of the ace gene itself, a more parsimonious explanation of our results is simply that aus is not an allele of ace. In this scenario, double mutant embryos exhibit the aus phenotype in combination with the loss of function of Ace. The weaker expressivity of the aus phenotype in such double mutants is entirely consistent with the aus phenotype being partly dependent upon functional Ace protein.
Given the arguments stated above, we believe that aus is most likely to involve a locus that is normally required for transcriptional regulation of the ace gene. One of the simplest ways in which one might envisage aus affecting ace expression at all of its sites of expression is if Ace activity normally modulates its own expression and that the aus locus is normally involved in this autoregulation. A mutation involving the aus locus would interfere with this feedback and might lead to deregulation of ace expression. Analysis of ace mutant embryos supports the idea that Ace is indeed involved in a feedback loop to regulate its own transcription (Reifers et al., 1998; R. Macdonald, M. Brand and S. W. W., unpublished observations). Although clearly important, ace is not the only gene upon which aus must act, as embryos lacking Ace activity still exhibit aspects of the aus phenotype indicating that aus must affect some genes via an Ace-independent route.
noi expression in the forebrain is less severely affected in aus mutants that lack functional Ace indicating that much of the upregulation of noi is dependent upon Ace activity. This dependence upon Ace may also explain why noi is only upregulated at sites where its expression overlaps with ace and not in regions of the embryo where the two genes are likely to be independent of each other’s activity (such as spinal neurons and pronephros). Conversely, the increase in ace expression in the forebrain does not appear to be dependent upon functional Noi suggesting that ace is upstream to noi in the forebrain of aus mutants. However, noi expression is not lost altogether in the forebrain of ace mutants or ace;aus double mutants indicating that, while Ace is required for expansion of noi expression in the forebrain of aus mutants, it is not required for induction of noi expression.
The relationship between noi and ace in the midbrain is likely to be different to that in the forebrain. In this region, ace expression is lost in the absence of Noi activity and noi expression is lost in the absence of Ace activity (Reifers et al., 1998) suggesting mutual dependence. However, Reifers et al. (1998) have shown that the two genes are initially activated independent of each other and the late loss of expression could be due to cell fate alterations in the mutant embryos.
We thank Corinne Houart, Michael Brand and the late Nigel Holder for comments and advice on this study, many colleagues for providing probes used in this analysis, other members of our laboratories for suggestions throughout the course of the work and Michael Brand, Jörg Rauch and Pascal Haffter for providing data prior to publication. We also would like to thank Christiane Nüsslein-Volhard in whose laboratory the mutant described in this study was initially isolated. This study was supported by grants from The Wellcome Trust and BBSRC. C. P. H. was supported by Fellowships from EMBO and the EC, and S. W. W. is a Wellcome Trust Senior Research Fellow.