The hypothesis suggesting that Six3 was the functional ortholog of the Drosophila so gene has been challenged recently by the finding that at the sequence level, Six3 is more closely related to Drosophila optix, a newly identified member of the Six/so gene family (Toy et al., 1998), than to so (Seo et al., 1999). The counterpart of optix has been identified in vertebrates (Toy et al., 1998); thus, the number of murine Six gene family members has increased to six (Six1-6) (Jean et al., 1999). In addition, the importance of optix during eye development in Drosophila was recently demonstrated by the eyeless (ey)-independent induction of ectopic eyes in imaginal discs (Seimiya and Gehring, 2000). In contrast, so induces ectopic eye formation only in cooperation with eyes absent (eya), and this phenomenon is dependent on ey activity (Pignoni et al., 1997). Furthermore, an increase in eye size and expansion of the retina territory have also been observed after ectopic optix2/Six6 expression in Xenopus embryos (Zuber et al., 1999; Bernier et al., 2000). All these data clearly suggest that some members of the Six/so family are pivotal in visual system development in metazoans.
Here, we show that in mouse embryos, Six3 expression starts much earlier than that reported for Six6, being detected as early as E7.0–7.5 in the prospective anterior neuroectodermal region. By the analysis of Six3 expression in Small eye mutant mice we were able to determine that in the region corresponding to the developing lens placodal ectoderm, Six3 expression is controlled by Pax6. We also found that ectopic Six3 expression in the midbrain/hindbrain region is sufficient to induce the formation of ectopic vesicles in developing mouse embryos. Analysis of these vesicles by whole-mount in situ hybridization revealed that they are Pax6 positive. This result, together with the reported induction of retina tissue by ectopic Six3 expression in fish embryos (Loosli et al., 1999) supports the proposal that these induced structures probably correspond to ectopic optic vesicles. This article is one of the first reports showing direct evidence that, in mammals, different regions of the embryonic head ectoderm are competent to respond to specific inducing signals during development.
RESULTS AND DISCUSSION
In the mouse, low levels of Six6 expression are first detected at approximately E8.0 in the floor of the diencephalon (Jean et al., 1999; our own unpublished data). Later, its expression is localized in the ventral optic stalk and ventral neural retina (Jean et al., 1999). It has been previously shown that in mice, at approximately E8–8.5, Six3 is strongly expressed in the anterior neuroectoderm and developing retina field (Oliver et al., 1995b). Expression is seen in the lens, retina, and optic stalks at later stages (Oliver et al., 1995b). We have now determined that in mouse embryos, Six3 expression appears as early as E7.0–7.5 (Fig. 1a) in the anterior neurectodermal region, the site from which the forebrain eventually develops. This early expression is similar to that reported for Otx2, another homeobox gene important for anterior head development and, like Six3, contains a lysine at position 50 of the homeodomain (Simeone et al., 1993). This early expression of Six3 indicates that besides its putative role during visual system development, this gene may play a role in the determination and development of the anterior central nervous system. Six3 expression in the prospective retina field starts to be detected at around E8–8.5 (Oliver et al., 1995b). At approximately E9.0, Six3 expression was detected in a region of the head ectoderm that included the area that later on and after thickening will give rise to the lens placode (Fig. 1b,c). The set of regulatory molecules that controls this expression may be different from the set that regulates expression in the neuroectoderm and retina field (see below). Support for this hypothesis was provided by the partial characterization of different Six3 regulatory elements. We analyzed the expression pattern of a reporter gene fused to different Six3 genomic fragments in transient transgenic mouse embryos. The elements responsible for Six3 expression in the ventral forebrain and neuroretina (Fig. 2a,b,d) are located in regions that are separated from those containing elements that promote expression in the lens placode (Fig. 2c,e). Interestingly, when the element responsible for the expression in the lens placode (III in Fig. 2c,e) is included as part of the larger construct (II in Fig. 2b,d), only ventral forebrain and neuroretina expression is detected in the majority of the analyzed transient transgenic embryos. In addition, stable transgenic lines containing this regulatory element and the Cre recombinase (see Furuta et al., 2000) or this element and a green fluorescent protein (GFP) reporter (Lagutin and Oliver, this work) have been generated and in both cases retina- and ventral forebrain-specific expression is detected. This finding suggests that likely, some lens placode inhibitory regulatory elements are present in the larger fragment II. Sequence analysis revealed the presence of a highly conserved putative PRD binding site in the genomic fragment III (data not shown) according to the consensus PRD binding site previously reported (Czerny and Busslinger, 1995). In few cases, additional ectopic reporter expression is detected in branchial arches and cranial ganglia (Fig. 2c) of the transient transgenic embryos generated by the use of the element III (Fig. 2a), suggesting that it is likely due to the integration position of the transgene.
Analysis of Six3 expression in homozygous Small eye (Pax6sey1-neu) mutant (Sey/Sey) embryos provided further evidence in support of the hypothesis that Six3 expression is controlled by different regulatory pathways in the neuroectoderm and lens ectoderm. Homozygous Sey/Sey mice lack functional Pax6 activity and are characterized by the complete absence of eyes at midgestation; however, morphologically abnormal optic vesicles are initially formed. Early in development, the presumptive lens ectodermal region is present in the homozygous Sey/Sey embryos, but during later stages, it fails to thicken and to form the lens placode. Six3 expression in the pituitary gland and brain is not affected in these mutant embryos (Oliver et al., 1995b), nor in the mutant optic vesicle at approximately E9.5 (Fig. 3b,d). Similar findings have also been recently reported for the Six6 gene (Jean et al., 1999). However, Six3 mRNA (Fig. 3b) or protein (Fig. 3d) expression in the lens ectoderm is not detected in the mutant mice at this stage. These findings indicate either that at this stage in the ectodermal region of the head that will give rise to the lens placode, Six3 expression is dependent on Pax6 activity, or that differentiation of the lens placode is required to maintain Six3 activity. This result shows that Six3 is downstream of Pax6 in the pathway regulating the development of this specific tissue; however, Six3 expression seems to be independent of Pax6 in the rest of the embryonic tissues, including the optic vesicles. It is interesting to mention that in Drosophila, ey interacts directly with an eye-specific enhancer present in the so gene (Niimi et al., 1999). It seems likely that in mouse, Six3 expression in the lens placode is also under the control of the Pax6 protein. In contrast, Six6 is not normally expressed in the developing lens placode, suggesting that its transcription is not regulated by Pax6. Similarly, optix function in Drosophila is also independent of ey activity (Seimiya and Gehring, 2000).
Recently, overexpression experiments involving Pax6 (Chow et al., 1999), Six3 (Oliver et al., 1996; Loosli et al., 1999), and Six6 (Zuber et al., 1999; Bernier et al., 2000), have shown that each of these genes can promote the generation of ectopic lenses, optic vesicles, or relatively normal eyes in restricted regions of the vertebrate embryonic head. In all cases, these experiments were performed in fish or frog embryos, but not in mammals. The confirmation of these results in mammalian embryos is important, because species-specific differences in the expression patterns of some of these genes have been observed. Therefore, we wished to determine the functional role of these genes during mammalian embryonic development.
Previous experiments in medaka fish have demonstrated that some specific areas of the head territory, including the otic placodes, were competent to respond to ectopically expressed Six3 protein (Oliver et al., 1996). Other investigators performing similar experiments in fish and frog embryos have obtained similar results (Loosli et al., 1999; Bernier et al., 2000). Together, these findings suggest that the induction of the ectopic optic structures occur during early developmental stages, just before the commitment process of the ectodermal placodes of the head. Therefore, we decided to ectopicaly express Six3 under the control of the Pax2 promoter element in mouse embryos. As early as E7.5, the Pax2 regulatory elements can promote transgene expression in a region that will become midbrain, hindbrain, and otic vesicles (Rowitch et al., 1999). Analysis of approximately 200 E10.0 embryos resulting from the Pax2-Six3 injection experiments determined that around 8% (16 embryos) were transgenic. As predicted by the previous studies performed in medaka fish and Xenopus in which misexpression of Six3 lead to the formation of ectopic optic vesicles, we found that overexpression of Six3 in the hindbrain-midbrain area also led to the formation of ectopic vesicle-like structures in nine of the transgenic mice (Fig. 4b–e). These induced ectopic vesicles were exclusively found in the midbrain-hindbrain region. To confirm the identity of these ectopic structures whole-mount in situ hibridization experiments were performed. Normally, in addition to its expression in the forebrain and midbrain, Pax6 is also expressed in the developing optic vesicles (see Fig. 4). As shown in Figure 4, whole-mount in situ hybridization showed that the induced ectopic vesicles expressed Pax6 (Fig. 4b–d), suggesting that these induced structures may indeed correspond to ectopic optic vesicles. These results not only confirmed the proposed importance of Six3 during vertebrate eye development, but also supported the proposal that specific areas of the head territory of fish, frogs, and mouse embryos are equally competent to respond to ectopicaly provided eye-inducing signals. However, no ectopic lenses were observed.
In very few cases, additional vesicle-like structures located exclusively adjacent to the otic vesicles were also observed (Fig. 5a). Morphologically, they resembled smaller otic vesicles (Fig. 5b), but because of their rare occurrence, we were not able to analyze these structures for expression of otic vesicle-specific markers to further confirm their identity. However, we did observe that these structures were negative for Pax6 expression (data not shown) indicating that they were not optic vesicles. It is possible that in those cases, and upon targeting of a different competent region of the head ectoderm (i.e., otic field), Six3 function may replace that of Six1, another family member that is normally expressed in the otic vesicle (Oliver et al., 1995a). Also, an enlargement of the forebrain-midbrain territory was occasionally observed (data not shown), a result similar to that seen upon Six3 overexpression in zebrafish embryos and which suggested a role for this gene during normal forebrain development (Kobayashi et al., 1998).
In conclusion, the induction of ectopic optic vesicle-like structures in mouse embryos further validates the original hypothesis that Six3 is a crucial regulator of vertebrate eye development. This finding, together with those previously reported by other groups (Oliver et al., 1996; Kobayashi et al., 1998; Loosli et al., 1999; Zuber et al., 1999; Bernier et al., 2000), also indicates a direct correlation between the type of ectopic structure (i.e., optic vesicle, lens, retina tissue, and forebrain enlargement) and the experimental conditions. The final outcome of the structure strictly depends on the embryonic stage at which the gene is overexpressed, the type of targeted tissue (i.e., the specific regions of the head), and the type and dose of overexpression (whether derived from transgenic constructs or injected RNA). In our study, the results indicated that the targeted midbrain-hindbrain territory is competent to respond to an optic vesicle-inducing signal promoted by Six3.
Although expression of Six3 in the lens placode depends on Pax6, its expression in the optic vesicle is unaffected in Pax6 mutant mice. This finding suggests that, although some of the same genes may participate in both lens and optic vesicle development, the mechanisms leading to their development may be controlled by different gene pathways. At later stages, these two pathways might engage in complex cross-talk processes involving expression of different secreted molecules such as bone morphogenetic proteins (Furuta and Hogan, 1998) and fibroblast growth factors.
In Situ Hybridization
Whole-mount or 8-μm-thick sections were hybridized with digoxigenin or 35S-labelled Six3 or Pax6 RNA antisense probes, respectively, and visualized with alkaline phosphatase-coupled anti-digoxigenin antibody and NBT/NCIP substrate (Boehringer Mannheim) as described in (Oliver et al., 1995b) or exposed for approximately 10 days in the second case.
Embryos were fixed overnight in 4% paraformaldehyde, cryoprotected with 30% sucrose in PBS overnight, embedded in tissue freezing medium, and cryostat sectioned (8 μM). A mouse Six3-specific polyclonal antibody was prepared by immunizing rabbits with the peptide RLQHQAIGPSGMRSLAEPGC included in the 3′ region of the Six3 gene. The specificity of the antibody was determined by its specific Six3-expression pattern at different developmental stages as well as by competition experiments using the Six-3–specific peptide (data not shown). Cy3-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch) was used as secondary antibody for fluorescent staining. Rabbit anti-GFP (Molecular Probes) was used in horseradish peroxidase staining with a donkey anti-rabbit (Amersham) secondary antibody. Sections were counterstained with 0.5% methyl green.
Six3 Reporter Constructs and Generation of Transgenic Mice
Different fragments of a Six3 genomic clone isolated from a 129/Ola mouse library were subcloned into pBluescript KSII. Generation of construct I was previously described (Furuta et al., 2000); it was generated by replacing most of the second exon of the Six3 gene (the first coding exon) with a lacZ reporter gene. For construct II, the gene encoding GFP was cloned in frame in the Six3 coding region. This construct included approximately 7 kb of the Six3 genomic region. Construct III was a phsplacZpA-derived plasmid into which we subcloned a 1.5-kb genomic fragment located at the 3′ region of the Six3 genomic clone.
These constructs were tested for expression in transient reporter transgenic embryos. Reporter constructs were injected into mouse zygotes, and embryos were isolated at the appropriate stage of development. After dissection and fixation, the embryos were stained for lacZ according to standard protocols or were analyzed for GFP expression by using an Olympus SZH10 microscope with fluorescent illumination.
Small Eye Animals and In Situ Hybridization
Mutant strain of the Pax6Sey1-neu has been maintained under ICR background. Embryos from crosses of heterozygous animals were isolated at E9.5, and homozygous mutant embryos were identified based on the eye morphology. In situ hybridization was performed as described (Furuta and Hogan, 1998) by using a [35S]-UTP–labeled antisense Six3 probe.
We thank M. Murray for excellent technical assistance, G. Grosveld for critical reading of this manuscript and J. Swift and J. Raucci for transgenic injections. This work was partially supported by G.O. and Y.F. received support from the National Institutes of Health.