The vertebrate eye is derived from neural and non-neural ectoderms that form the neural retina and the lens, respectively. These two tissues are not independent entities in the developmental process; rather, they are strongly dependent on each other. The induction of lens in the non-neural ectoderm is thought to occur when the developing neural retina or optic vesicle evaginates to contact the overlying non-neural ectoderm, or the presumptive lens ectoderm (PLE). The induced ectoderm thickens to form the lens placode (LP), which further develops into the lens vesicle that pinches off from the surface ectoderm to mature as an ocular lens. In coordination with the development of lens lineage, the optic vesicle forms an optic cup and the optic cup further matures into the layered neural retina that consists of a ganglion cell layer, a nuclear layer, and a photoreceptor cell layer, with variations depending on species.
Classic embryological experiments have shown that heat damage to the optic vesicle primordium resulted in the absence of lens, demonstrating for the first time that the induction of lens requires an optic vesicle as the signal source (Spemann, 1901, cited in Hamburger,1988). However, subsequent studies of various vertebrate species including Xenopus laevis revealed that the optic vesicle is not always necessary for the induction of lens. There are many examples of lens formation in the absence of the optic vesicle (e.g., Henry and Grainger,1987; Mizuno et al.,1998; reviewed in Tahara,1962; Jacobson and Sater,1988). In this case, a “lentoid” or lens-like structure, instead of an accurately patterned ocular lens, is formed, strongly implicating the “formative effects” (Holtfreter and Hamburger,1955) of retina or developing retina on the normal morphogenesis of lens, which include regulation of form, size, maintenance, and differentiation of developing lens (Jacobson and Sater,1988). To illustrate, in newt embryos, the free lens derived from optic vesicle ablation consisted of only lens epithelium and differentiated lens fibers were rarely observed (Mizuno et al.,1998). Based on these findings, it can be summarized that lens development definitely requires the developing or presumptive retina in many respects, and in terms of lens induction, it is a matter of timing when inducing signal(s) emanate from the retina anlage or the anterior neural plate (Grainger,1996), at least in some vertebrate species. Recent molecular marker studies support this idea. In chick, for example, it was reported that ectodermal explants derived from a region surrounding the anterior neural plate autonomously expressed a series of lens marker genes when cultured alone, eventually differentiating into δ-crystallin-positive lentoids (Bailey et al.,2006). In zebrafish mutant chokh, a pair of lenses was formed although the optic vesicle failed to evaginate to contact the surface ectoderm (Loosli et al.,2003).
The current understanding of lens induction is that it does not occur in a single step. Through a series of tissue transplantation studies using Xenopus embryos, Grainger and his colleagues developed a stepwise determination model of lens induction (Grainger,1992) in which the process of lens induction consisted of competence, bias, specification, and differentiation. In Xenopus embryos, removal of the anterior neural plate at late neurula stage resulted in the formation of free lens, whereas removal at early neurula stage did not yield free lens (Henry and Grainger,1990). These findings indicate that around mid-neurula stage (st. 15/16), the neighboring non-neural ectoderm has already acquired lens-forming bias by receiving a signal(s) emanating from the anterior neural plate. Previous tissue transplantation studies showed that no lens was formed when ectoderm with no or low lens-forming competence was transplanted to an approximate region of the lens-forming field (or lens field: LF) at neurula stage (Henry and Grainger,1987; Servetnick and Grainger,1991). This indicates that the non-neural ectoderm requires appropriate competence to receive lens-inducing signals and to develop into lens (Henry and Grainger,1987). Collectively, these findings suggest that factors involved in the formation of lens bias and/or the establishment of lens competence are important for subsequent lens formation in Xenopus.
Genes that are activated in the lens lineage have been identified and molecular cascades that eventually lead to the expression of crystallin genes are well established (reviewed by Ogino and Yasuda,2000; Kondoh,2008). In Xenopus, otx2, six3, and pax6 are known to be expressed at open neural plate stage in a restricted region of the pre-placodal ectoderm (PPE, a.k.a. pre-placodal region, PPR), which surrounds the anterior neural plate and will form the cranial sensory placodes at later stages (reviewed in Schlosser and Ahrens,2004; Streit,2004). PPE expression of these genes is thought to mark LF and to represent the lens-biased ectoderm (Zygar et al.,1998). foxE3 (originally identified as lens1 in Xenopus) is a gene that shows lens-restricted expression from neurula stage (Kenyon et al.,1999), and this can also be a mark of lens-biased ectoderm. One way to understand the initial event of lens development would be to identify a gene regulating the expression of LF markers. We identified Xhairy2, a Xenopus hairy and enhancer-of-split (Hes) gene, as a factor that possesses such ability.
Hes family genes encode the basic-helix-loop-helix transcriptional repressor and are known to be involved in a variety of developmental processes among animal phyla (Fisher and Caudy,1998; Davis and Turner,2001). In vertebrate development, the regulation of neural development by Hes genes has been intensively studied, and lines of evidence have demonstrated that Hes genes block differentiation and promote cell proliferation (reviewed in Kageyama et al.,2008). Moreover, a recent report showed that Hes1 actively safeguarded against terminal differentiation and permanent cell cycle arrest in quiescent cells (Sang et al.,2008). Mouse Hes1 is known to be involved in eye development. Knockout studies have revealed that Hes1 regulates the differentiation of retinal neurons (Tomita et al.,1996; Lee et al.,2005) and the initial morphogenesis and outgrowth of the optic vesicle in coordination with Pax6 (Lee et al.,2005). In those cases, lens morphogenesis was also perturbed, mainly due to the lack of functional developing retina (Tomita et al.,1996). In terms of the requirement of Notch signaling in eye morphogenesis, which is known to be an upstream regulator of Hes genes in many, but not all, cases (reviewed in Kageyama et al.2008), a recent work that performed careful conditional knockout of Notch effector Rbpj showed that Notch signaling controlled the timing of primary lens fiber cell differentiation in the lens anterior epithelial layer (Rowan et al.,2008).
Xhairy2 is most similar to Hes1 among mammalian Hes genes in the nucleotide sequence, and is expressed in ectodermal tissues, such as the floor plate and the neural crest, as well as in mesodermal tissues, such as the anterior prechordal plate and somites (Tsuji et al.,2003). Our previous functional analyses suggested that Xhairy2 functions in the maintenance of tissue identity and of the proliferative as well as the undifferentiated state (Yamaguti et al.,2005; Nagatomo and Hashimoto,2007). Xhairy2 is also expressed in the PPE region from gastrula stage, but its possible contribution to cranial sensory organ formation was not elucidated.
In this study, we report that Xhairy2 is required for lens development as early as the stage of LF formation. Xhairy2 is expressed in the PPE region that includes future LF from gastrula stage. By means of morpholino (MO) injection, we show that Xhairy2 knockdown reduced the expression of lens marker genes at every step of lens determination, eventually resulting in ocular lens malformation. Interestingly, retina marker gene expression and retinal anlage morphology remained normal upon Xhairy2 knockdown. Xhairy2 overexpression, however, did not expand LF, indicating that Xhairy2 might function outside of the known cascade of transcription factors specific for lens development. Also, Xhairy2 knockdown did not affect the expression of genes that are components of the signaling pathway related to LF induction. Instead, loss of LF caused by Xhairy2 depletion was partially rescued by the simultaneous knockdown of p27xic1, a gene encoding the cell cycle inhibitor. However, pharmacological treatment denied the possibility that the decrease in cell number itself was a direct cause of LF loss. Based on these findings, we propose that Xhairy2 may maintain an intracellular environment in which lens-inducing signal(s) can be accepted.
Xhairy2 Knockdown Results in Loss of Ocular Lens Structure
Xhairy2 expression in the anterior neural border region (Fig. 1b,e,h) seems to correspond to the PPE region. We previously showed that the lateral part of border expression was important for neural crest formation (Nagatomo and Hashimoto, 2007), but could not clarify the function(s) of Xhairy2 in the anterior neural border region. In terms of expression layer, Xhairy2 is indeed expressed in the deep layer where PPE genes are expressed (Schlosser and Northcutt,2000; Fig. 1c,c',f,f',i,i'), and not in the surface layer where genes marking the future cement gland region are expressed (Drysdale and Elinson,1992; Sive and Bradley,1996). This is consistent with the expression pattern of the PPE gene dlx5 (Fig. 1j–r'), which is later expressed in all cranial sensory placodes except LP (for a review, see Schlosser and Ahrens,2004). The PPE-like expression of Xhairy2 is visible at mid-gastrula stage (st. 11.5–12), whereas dlx5 expression becomes clear only at late gastrula stage (st. 12.5: compare Fig. 1b,e,h with k,n,q), suggesting that the PPE expression of Xhairy2 is one of the earliest.
To examine the possible roles of the PPE expression of Xhairy2, we knocked down Xhairy2 in this region by injecting morpholino antisense oligo (Xhairy2 MOs; see Experimental Procedures section for the use of MOs). Xhairy2 (6.9 ng) or control MO (Co MO: 3.4 ng) was injected along with the lineage tracer EYFP mRNA (400 pg) into one dorsal animal blastomere of 8-cell-stage embryos. At stage 42, the injected tadpoles were sorted based on EYFP fluorescence. Only the tadpoles that showed EYFP fluorescence in the head region (Fig. 2a,e) were anesthetized, fixed, and subjected to morphological analyses. Other than the finding that the eye on the injected side of each embryo appeared to be slightly smaller than that on the non-injected side (84%, n = 62; Fig. 2e), no apparent abnormalities were observed. We thus proceeded with the histological analyses of Xhairy2 morphants. Although there were no remarkable morphological changes in the nasal pits and the inner ear (data not shown), interestingly, we found loss or an aggregate of cells in place of ocular lens in the Xhairy2-depleted embryos (6 of 6, Fig. 2f–h) but not in the Co MO–injected embryos (5 of 5, Fig. 2b–d). The injection of 6.8 ng of Co MO did not cause abnormal morphological development of the eye (5 out of 5, data not shown), indicating that a higher dose of MO did not cause lens malformation in the Xhairy2-depleted embryos (the dose of Co MO was set at 3.4 ng hereafter). To examine if the malformed lens or the lens-like aggregate of cells maintains lens character, we immunostained γ1-crystallin in MO-injected embryos. Interestingly, all the samples tested were immunopositive for γ1-crystallin, as shown in Figure 2f (5 out of 5; Co MO: 2 out of 2, Fig. 2b). This result suggests that Xhairy2 depletion appears to affect lens development.
Vertebrate eye morphogenesis is a stepwise process, and the developing retina sends signals to the adjacent (at open neural plate stage) or overlying ectoderm (after neural tube formation) to change the commitment level of the ectoderm of lens lineage. From this perspective, the presence of retina in terms of morphology and the absence of ocular lens (Fig. 2f,h,h') suggest that cells of lens lineage might be specifically affected. As eyes mature, signal communication between the developing lens and the developing retina becomes bidirectional so that the lens in some way is required for fine-tuning neural retina patterning (Ashery-Padan et al.,2000). Thus, the slight changes in the precise structure of retina (see Fig. 2c',d',g',h') would be due to the lack of a functional lens that promotes retina maturation in the late developmental stages. Eventually, we focused on the possible roles of Xhairy2 in the early stages of lens development or the formation of LP, as PPE contains LF that develops into PLE and LP.
Xhairy2 Depletion in Presumptive Lens Ectoderm and Lens Placode Reduces Lens Marker Gene Expression But Not Retina Marker Gene Expression
To assess the effects of Xhairy2 depletion more precisely, we investigated the expression of marker genes of both lens lineage and retinal lineage in tail-bud stage embryos. Xhairy2 MO or Co MO was injected along with lineage tracers, clacz mRNA (cytoplasmic lacZ, 400 pg) and EYFP mRNA (400 pg), into one dorsal animal blastomere of 8-cell-stage embryos. The injected embryos were sorted based on EYFP fluorescence, fixed at tail-bud stage (st. 25, 28, or 32), and stained with X-gal to visualize the distribution of injected MOs. Only the embryos showing X-gal-positive cells in the head ectoderm including LP were subjected to whole-mount in situ hybridization (WISH). Two important transcriptional regulator genes for lens development, foxE3 (or Lens1 in Xenopus; Kenyon et al.,1999; Ogino et al.,2008) and L-maf (Ogino and Yasuda,1998; Ishibashi and Yasuda,2001), are expressed in LP. Co MO injection did not have any significant effects on foxE3 (no change in 93%, n = 27; Fig. 3b,b') and L-maf (no change in 86%, n = 28; Fig. 3c,c') expression. Embryos injected with Xhairy2 MOs, however, showed loss or reduced expression of foxE3 (reduction in 73%, n = 30; Fig. 3h,h') and L-maf (reduction in 76%, n = 34; Fig. 3i,i') on the injected side. Interestingly, we never observed the complete loss of foxE3 expression, which was frequent in L-maf expression (compare Fig. 3h and i). This implied that Xhairy2 might selectively regulate LP marker expression. It was reported that notch2 gene was expressed in several areas of cranial ectoderm including lens anlage and that Notch signaling is a crucial component of foxE3 expression (Ogino et al.,2008). Embryos injected with Xhairy2 MOs showed reduced expression of notch2 in PLE and LP region (in 47%, n = 34; Fig. 3g,g'), while Co MO injection did not cause changes in notch2 expression at all (in 97%, n = 35; Fig. 3a,a'). The expression of the differentiation marker gene γ1-cry (Offield et al.,2000), a common crystallin among vertebrate species, was, as expected, strongly reduced in embryos injected with Xhairy2 MOs (reduction in 90%, n = 30; Fig. 3j,j') on the injected side but not in embryos injected with Co MO (no change in 89%, n = 18; Fig. 3d,d'). Importantly, embryos injected with Xhairy2 MOs or Co MO showed no significant defects in the expression of rx1 (Mathers et al.,1997), a definitive marker gene of developing retina, at least around the tail-bud stage (Xhairy2 MO: no change in 84%, n = 25, Fig. 3k,k'; Co MO: no change in 100%, n = 23, Fig. 3e,e'). Similar results were obtained in the expression of another retinal marker gene six6 (originally termed XOptx2, Zuber et al.,1999; Co MO: no change in 97%, n = 33, Fig. 3f,f'; Xhairy2 MO: no change in 84%, n = 32, Fig. 3l,l'). All these data are consistent with the morphological data presented above (Fig. 2) and suggest that Xhairy2 is necessary in lens lineage cells, at least from the stage of PLE formation.
Xhairy2 Is Required for Lens Field Formation in Neurulae But Not for Pre-Placodal Ectoderm Marker Gene Expression
The results so far indicate that Xhairy2 functions at the stages of PLE formation and LP formation. We next tried to investigate whether Xhairy2 is required for LF formation in neurulae. LF is lens-biased ectoderm within PPE (Fig. 4a) and is formed in mid-neurula stage in response to signals emitted from the adjacent anterior neural plate region (Zygar et al.,1998). LF is marked by pax6 (Hirsch and Harris,1997; Li et al.,1997; Zygar et al.,1998) and six3 (Zhou et al.,2000) expression outside the neural plate, which is distinct from future retina-forming expression within the neural plate. By precise two-color WISH with Xhairy2 expression that marks neural plate border at st. 15, we reconfirmed that a part of pax6 and six3 expression was indeed outside the neural plate and within the PPE region (Fig. 4b and c, respectively).
To see whether Xhairy2 knockdown affected only the LF expression of pax6 and six3, MO-injected embryos were analyzed at st. 15 or 17 by means of WISH. The results clearly showed that Xhairy2 knockdown reduced only the LF expression of pax6 (reduction in 66%, n = 83, Fig. 4e) and six3 (reduction in 73%, n = 51, Fig. 4g), while Co MO injection had virtually no effects (pax6: no change in 99%, n = 69, Fig. 4d; six3: no change in 98%, n = 54, Fig. 4f; pax6 in 6.8 ng Co MO embryos: no change in 94%, n = 17, not shown). Importantly, pax6 and six3 expression of the future retina field was not affected. Also, the LF phenotype of pax6 at st. 17 by means of Xhairy2 MO injection (reduction in 72%, n = 68, Fig. 4j) was rescued by co-injecting 20 pg of Xhairy2b mRNA that did not have MO annealing sequences (reduction in 35%, n = 81, Fig. 4k and l), indicating the specificity of Xhairy2 knockdown to the LF-loss phenotype. The LF-loss phenotype was also observed in st.-18 embryos, as shown by means of WISH with pitx-1 marking LF and future cement glands at these stages (Hollemann and Pieler,1999). Embryos injected with Xhairy2 MOs showed significant reduction, but not complete loss, of the LF expression of pitx-1 (reduction in 60%, n = 25, Fig. 4i), while injection of Co MO had no significant effects (no change in 100%, n = 24, Fig. 4h). Collectively, these results suggest that Xhairy2 is necessary for LF formation independent of retinal influence.
Xhairy2 Knockdown Shows Selectivity in PPE Marker Gene Regulation
Many genes that mark sensory and neurogenic placodes after neural tube closure are known to start expression as early as at the open neural plate stage (Schlosser and Ahrens,2004; Streit,2004; Schlosser,2006), and one shared feature of such genes as six1 (Pandur and Moody,2000) and dlx5 (originally named X-dll3; Papalopulu and Kintner,1993; Luo et al.,2001) is the expression pattern: a horseshoe-shaped expression surrounding neural plate and covering PPE is induced, similar to Xhairy2 expression, and the expression later sharpens into individual placodes after neural tube closure. We thus investigated the effects of Xhairy2 knockdown on the expression of dlx5 at st. 15 and of six1 at st. 18. Neither Co MO injection nor Xhairy2 MO injection resulted in changes in the expression of dlx5 (Co MO: n = 20, Xhairy2 MO: n = 25, Fig. 4q and r, respectively) and six1 (Co MO: n = 25, Xhairy2 MO: n = 24, Fig. 4s and t, respectively). These results would be consistent with the morphological phenotype shown in Figure 2 because both dlx5 and six1 are expressed in most of cranial placodes but not in LP after neural tube formation (Papalopulu and Kintner,1993; Pandur and Moody,2000). Based on this fact, we next checked two PPE genes, foxE3 and notch2, which we have examined in tail-bud stage embryos as markers covering PLE and LP. Consistent with the results shown in Figure 3, embryos injected with Xhairy2 MOs showed significant reduction in notch2 expression (reduction in 70%, n = 24, Fig. 4p), while embryos injected with Co MO showed no significant change (n = 20, Fig. 4o). However, neither Co MO nor Xhairy2 MO injection resulted in the reduction of foxE3 expression at st. 15 (Co MO: n = 29, Xhairy2 MO: n = 26, Fig. 4m and n, respectively), although the expression was down-regulated at st. 18/19 (Co MO: normal in 88%, n = 16, Xhairy2 MO: reduced in 70%, n = 20; see also Fig. 6). These results suggest that Xhairy2 might be specifically required for lens lineage or that Xhairy2 might function in parallel with other PPE genes for lens development.
Loss of pax6 Expression in Lens Region Is Partly Caused by Ectopic Expression of Cell Cycle Inhibitor p27xic1
The results so far suggest that in Xhairy2 morphants, lens lineage cells are affected from the LF stage, the earliest stage of lens development that can be recognized through specific marker gene expression. To explain the mechanism of Xhairy2 requirement, we quickly checked if a signaling pathway or a tissue that is known to be necessary for lens induction was affected at the level of gene expression. A previous study showed that FGF8 might be required for LF expression of pax6 (Ahrens and Schlosser 2005). However, Xhairy2 knockdown did not alter the expression of fgf8 (Christen and Slack,1997) or fgfr4c (Golub et al.,2000; data not shown). Likewise, head endomesoderm, which is shown to be necessary for lens induction (Henry and Grainger,1990), did not seem to be affected when checked with cer (Piccolo et al.,1999) expression (data not shown). Therefore, these results suggest that Xhairy2 might function outside of known LF-inducing mechanisms.
Since both Xhairy2a and Xhairy2b function as transcriptional repressors, similar to other Hes proteins (Murato et al.,2006,2007), we assumed that the up-regulation of genes that are not expressed in the future LF region at gastrula or early neurula stage might block LF formation at mid-neurula stage. In our previous study of the functions of Xhairy2 in neural crest formation (Nagatomo and Hashimoto,2007), we showed that Xhairy2 down-regulated subsets of proneural marker genes, such as X-delta-1 (Chitnis et al.,1995) and X-ngnr-1 (Ma et al.,1996). In the injection conditions of this study, these proneural marker genes were indeed up-regulated when Xhairy2 MOs were injected (data not shown). In addition, we further revealed that the expression of another proneural marker gene X-MyT1 (Bellefroid et al.,1996) was also up-regulated upon Xhairy2 knockdown (up-regulation in 59%, n = 22; Co MO: no change, n = 20; Fig. 5a and b, respectively). As LF and neural crest neighbor each other (Fig. 5g and h), which was visualized by WISH with foxD3 (Sasai et al.,2001) and pax6/six3, we hypothesized that LF formation might be regulated by Xhairy2 in a similar manner to neural crest formation. We therefore focused on p27xic1, which encodes a cell cycle inhibitor (Ohnuma et al.,1999).p27xic1 was ectopically up-regulated by Xhairy2 knockdown (Co MO: no change, n = 17; Xhairy2 MO: up-regulation in 90%, n = 20, Fig. 5c and d, respectively) causing loss of neural crest (Nagatomo and Hashimoto,2007). In our previous study, only the double knockdown of p27xic1 and Xhairy2 rescued the loss of neural crest phenotype caused by Xhairy2 single knockdown (Nagatomo and Hashimoto,2007). In other words, the up-regulation of other proneural marker genes, such as X-MyT1 (Fig. 5b), seems to be parallel to the loss of neural crest and LF. Furthermore, two-color WISH with p27xic1 and pax6 clearly showed that the expression of these two genes is in a complementary relationship at st. 12.5 (Fig. 5e) and st. 15 (Fig. 5f) in the anterior portion of embryos. In later stages, p27xic1 is expressed in the differentiating lens vesicle (Ohnuma et al.,1999), implying that Xhairy2 is part of a mechanism that reduces p27xic1 expression in lens lineage cells during early development.
To test the hypothesis that the ectopic p27xic1 up-regulation by means of Xhairy2 knockdown blocks LF formation, we first injected 20 pg of p27xic1 mRNA into one dorsal animal blastomere of 8-cell-stage embryos. WISH analyses with pax6 showed that p27xic1 mRNA injection indeed mimicked the LF loss phenotype caused by Xhairy2 knockdown (reduction in 41%, n = 61, Fig. 5i). Xhairy2 knockdown by co-injection of Xhairy2 MOs (6.8 ng) and Co MO (3.4 ng) reduced LF expression of pax6 (in 77%, n = 70, Fig. 5k), which was partly rescued by co-injecting p27 MO (3.5 ng) in place of Co MO (reduction in 53%, n = 75, Fig. 5l). As expected, p27 MO injection alone did not have any significant effects on the LF expression of pax6 (no change in 96%, n = 70, Fig. 5j). In the case of six3, another LF marker, the extent of rescue was very weak (data not shown), suggesting that factor(s) other than p27xic1 might be involved in the regulation of LF expression of six3. Nevertheless, the above results suggest that protection of the premature expression of p27xic1 by Xhairy2 is required for LF formation and subsequent lens development.
Xhairy2 Knockdown Affects Cell Proliferation Within LF in Late Neurula Embryos, But Cell Cycle Inhibition Does Not Cause Loss of LF and Lens
The results obtained so far suggest the partial requirement of p27xic1 regulation in LF formation. As p27xic1 is known to inhibit the cell cycle (Ohnuma et al.,1999), we next tried to determine whether cell cycle regulation via Xhairy2 was related to LF formation. To address this issue, we first checked if the loss of LF marker expression was coupled with reduced cell proliferation at mid-neurula and late-neurula stages by means of BrdU incorporation analyses. mRNAs of p27xic1, p27 MO, Xhairy2 MO with Co MO, or Xhairy2 MO with p27 MO were injected into one dorsal animal blastomere of 8-cell-stage embryos. At neurula stages, the embryos were further injected with BrdU. To visualize LF, samples were stained with pax6 (mid-neurula stage) or foxE3 (late-neurula stage). Co-injection of Xhairy2 MO and Co MO reduced the number of BrdU-positive cells within LF marked by foxE3 expression at late neurula stage (Fig. 6g,g',i), and this phenotype was not rescued by co-injecting p27 MO in place of Co MO (Fig. 6h,h',i). This is consistent with our previous study (Nagatomo and Hashimoto,2007), although it was not limited to LF. Interestingly, however, the results at the mid-neurula stage clearly showed that only the overexpression of p27xic1 significantly reduced the number of BrdU-positive cells within LF marked by pax6 expression (Fig. 6a–d,a'–d',i). The results suggest that ectopic p27xic1 expression via Xhairy2 depletion indeed occurred within LF lineage in terms of the reduced number of proliferating cells, although the reduced number of proliferating cells itself might not cause the loss of LF at least at mid-neurula stage.
Based on the results shown above, we further examined the possible relationship between LF formation and cell cycle regulation. For this purpose, we inhibited the cell cycle by treating embryos with hydroxyurea and aphidicolin (HUA), both of which are known to inhibit DNA replication. The efficiency of HUA treatment was confirmed by BrdU incorporation analyses (Fig. 7a–c). To test whether HUA treatment affects LF marker gene expression, embryos were treated with HUA or DMSO (as control) from early gastrula stage to mid-neurula stage, and were then subjected to WISH with pax6 or six3. The results showed that pax6 expression seemed normal upon HUA treatment (in 85%, n = 20, Fig. 7e; DMSO: normal in 95%, n = 20, Fig. 7d), although six3 expression was moderately reduced upon HUA treatment (in 50%, n = 20, Fig. 7g; DMSO: normal in 90%, n = 20, Fig. 7f). The level of mitotic marker phospho histone H3 (PHH3) was reduced in HUA-treated embryos compared with DMSO-treated embryos (n = 10, Fig. 7d,e). In support of the above results, lenses were formed in embryos that were treated with HUA from gastrula to neurula stages and were further raised in the absence of HUA (n = 10, Fig. 7i,i'; DMSO: all normal, n = 10, Fig. 7h,h'). Collectively, these results imply that cell cycle regulation might not underlie the function of Xhairy2 at the onset of LF formation.
In this study, we performed a series of loss-of-function studies on Xhairy2 with the aim of revealing its potential role(s) in cranial sensory organ formation. Xhairy2 knockdown resulted in the loss of LF in neurula that was marked by such marker genes as pax6 (Fig. 4). This continued to affect subsequent steps of lens development. The expression of determination marker L-maf and differentiation marker γ1-cry was strongly down-regulated by Xhairy2 knockdown (Fig. 3), eventually causing hypoplasia of lens in tadpoles (Fig. 2). Importantly, there were no significant changes in gene expression in the retinal lineage (Figs. 3 and 4) and layered pattern of retina in Xhairy2 morphants looked normal in terms of morphology (Fig. 2). The eye phenotype of Xhairy2 morphants is different from that of Hes1 knockout mice in which both retina and lens were malformed (Tomita et al.,1996). Lens malformation was thought to be due to the secondary effects of malformed retina (Tomita et al.,1996), and this might implicate inter-species variations between Xhairy2 and Hes1. As a molecular mechanism of Xhairy2 function, our data suggest the down-regulation of p27xic1 expression in PPE, including the future LF region (Fig. 5). As the inhibition of ectopic p27xic1 expression by p27 MO partially rescued the loss of LF phenotype of pax6 expression, the product of p27xic1 must cause the phenotype. This was further supported by the finding that the forced expression of p27xic1 phenocopied Xhairy2 knockdown in terms of the LF loss phenotype of pax6 expression (Fig. 5). Also, our transplantation experiments showed that LF expression of pax6 was not secondarily affected by the loss of neural crest via Xhairy2 knockdown (data not shown), suggesting that Xhairy2 functions in LF formation in a cell-autonomous fashion. Here, p27 MO rescue of Xhairy2 MO phenotype was not complete. Our preliminary studies showed that neither the expression of foxE3 (st. 28) or L-maf (st. 28) nor the formation of ocular lens (st. 42) was rescued by the double knockdown of Xhairy2 and p27xic1, although the overexpression of p27xic1 mRNA mimicked the late phenotypes of Xhairy2 knockdown (data not shown). This strongly suggests that a factor(s) other than p27xic1 might be involved in lens development downstream of Xhairy2. As p27xic1 overexpression produced results that are consistent with those of Xhairy2 knockdown in many cases (Figs. 5 and 6) and p27 MO did not rescue the reduced proliferation at late neurula stage upon Xhairy2 depletion (Fig. 6), one possibility is that other developmentally regulated Cip/Kip family members, such as p16Xic2 and p17Xic3 (Daniels et al.,2004), could be regulated by Xhairy2. We also presume that the relatively long incubation time would produce an unequal distribution of the three types of MOs. Based on these findings, we conclude that Xhairy2 is required for the formation of lens-biased LF within PPE or future PPE partly through the regulation of p27xic1 expression.
The complete loss of ocular lens in Xhairy2 morphants was rarely observed in this study. This might be related to the regulation of foxE3 expression by Xhairy2. FoxE3 is an important factor that regulates both proliferation and differentiation of lens cells in mice (Medina-Martinez et al.,2005). FoxE3 is also known to be important for lens morphogenesis in zebrafish (Shi et al.,2006). Examination of the changes in foxE3 expression at mid-neurula (Fig. 4), late neurula (Fig. 6), and late tail-bud stages (Fig. 3) revealed that the initial expression of foxE3 was not affected by Xhairy2 knockdown at mid-neurula stage (Fig. 4), and foxE3 expression persisted in late tail-bud-stage embryos (Fig. 3). The latter case appears to be consistent with the finding that the frequency of reduced notch2 expression upon Xhairy2 knockdown was around 50% in tail-bud-stage embryos, as Notch signaling was shown to be a crucial component of foxE3 expression in Xenopus (Ogino et al.,2008). In contrast, L-maf expression was strongly reduced upon Xhairy2 depletion (Fig. 3). This implies that Xhairy2 might function in lens formation in parallel with foxE3 and that the marked reduction of γ1-cry expression and the malformed lens might be mainly due to the reduction of L-maf expression. Also, our data implicate potential variations in the downstream regulation of pax6 between mammals and Xenopus. In mice, FoxE3 expression is highly sensitive to Pax6 dosage (for review, see Medina-Martinez and Jamrich,2007). This is apparently inconsistent with our data that the LF expression of pax6 is strongly reduced (Fig. 4). However, no Pax6 binding site was found in the regulatory core of foxE3 enhancer in Xenopus (Ogino et al.,2008). As L-maf is downstream of pax6 (for review, see Reza and Yasuda,2004), pax6 might be especially important for L-maf expression in Xenopus. In future studies, it would be interesting to examine the phenotypes of the double knockdown of foxE3 and Xhairy2.
From the expression pattern of Xhairy2, we at first assumed that it might be involved in the formation of other cranial sensory organs. However, Xhairy2 knockdown caused malformation of neither otic nor olfactory organs. Since the knockdown weakly reduced the early expression of pax8 and foxG1 at neurulae (data not shown), other Hes factors might be involved in the formation of these sensory organs in combination with Xhairy2. One candidate would be XHes2, which is expressed in the future otic region from early neurula stage (Solter et al.,2006). Our preliminary study showed that Xhairy2 knockdown did not affect XHes2 expression (data not shown). In addition, we showed that Xhairy2 knockdown did not affect pan-placodal marker six1 or dlx5 expression (Fig. 4). It was reported, for instance, that the development of both neurogenic and non-neurogenic placodes, except LP, was perturbed in mouse mutants of Six1 (e.g., Zou et al.,2004). As Xhairy2 expression starts prior to six1 expression, these findings suggest that Xhairy2 might be specific for lens formation in terms of cranial placode development. The identification of other Hes genes and the multiple knockdowns of such factors will help further elucidate this problem.
Many transcription factors are sequentially activated in the lens lineage (reviewed in Ogino and Yasuda,2000; Kondoh,2008). It is possible to position Xhairy2 in this kind of cascade to understand its importance in lens development. As Xhairy2 knockdown reduced LF expression of pax6 and six3 (Fig. 4), Xhairy2 is supposed to be epistatic to these two genes. This, in turn, means that Xhairy2 overexpression around the LF region should increase the LF expression of pax6 and six3. However, neither wild-type Xhairy2 overexpression nor engrailed-fused form of Xhairy2 overexpression (Murato et al.,2006) increased the LF expression of six3 (data not shown). Moreover, our data showed that Xhairy2 knockdown did not alter the expression of fgf8 or fgfr4c (data not shown). FGF signaling was reported to be necessary for the induction of placode-specific pax genes as well as six1 (Ahrens and Schlosser,2005). This may indicate that the signaling environment itself is normal in the absence of Xhairy2. Collectively, these suggest that Xhairy2 might actually function outside of the cascade of transcription factors specific for lens formation, which is initiated by signaling input.
Most of the cranial placodal cells, with the exception of trigeminal placodal cells, start differentiation after neural tube formation in terms of n-tubulin expression. Since specific marker gene expression that indicates initial fate decision starts from neurula stage, these facts suggest that PPE cells fated to many of the cranial placodes have to actively maintain an undifferentiated state against a potentially differentiation-inducing environment. This is, for example, partly supported by the observation in Xenopus that ngnr-1 expression is initially detected in the PPE region in gastrulae but is soon suppressed in the broad region except the future trigeminal placodal region (Schlosser and Ahrens,2004). Our data demonstrated that Xhairy2 knockdown transiently up-regulated the expression of proneuronal marker genes, such as X-ngnr-1, X-delta-1, and X-MyT1 (Nagatomo and Hashimoto,2007; this study). Therefore, one important function of Xhairy2, together with other factors, in PPE cells would be the active maintenance of the undifferentiated state.
The other clue to understanding Xhairy2 function(s) in relation to the state of PPE cells is the ectopic up-regulation of p27xic1 expression upon Xhairy2 knockdown (Fig. 5; Nagatomo and Hashimoto 2007). As the double inhibition of Xhairy2 and p27xic1 rescued the phenotype of neurogenic marker up-regulation (Nagatomo and Hashimoto,2007) as well as the loss of LF expression of pax6 (Fig. 5), p27xic1 is the most promising factor to better understand Xhairy2 function(s). p27xic1 belongs to the Cip/Kip family of CDK inhibitors that block kinase activity of CDK2 of the cyclin E/CDK2 complex to inhibit G1/S transition (Besson et al.,2008), and the up-regulation of these factors is a common feature of differentiating cells. Reflecting the molecular mechanism, p27xic1 is expressed in differentiating lens epithelial cells (Ohnuma et al.,1999). This is consistent with our data that foxE3 expression was reduced upon Xhairy2 knockdown (Fig. 3), because studies of frog, fish, and mouse embryos revealed that foxE3 positively regulates the undifferentiated state (Kenyon et al.,1999; Medina-Martinez et al.,2005; Shi et al.,2006). Our previous study showed that Xhairy2 knockdown decreased the number of mitotically active cells at late neurula stage (Nagatomo and Hashimoto,2007). It seems to coincide with the loss of PLE expression of foxE3 (Fig. 6 and data not shown). Since Xhairy2 appears to be outside of the cascade of lens-specific transcription factors as discussed above, the decrease in the number of mitotically active cells raises the possibility that Xhairy2 might merely influence the number of lens lineage cells without affecting gene expression. However, LF-loss embryos by means of Xhairy2 knockdown did not show any decrease in the number of mitotically active cells at mid-neurula stage (Fig. 6). Furthermore, forced inhibition of the cell cycle by means of HUA treatment from gastrula to mid-neurula stage affected neither LF expression of pax6 nor ocular lens formation (Fig. 7). These data suggest that Xhairy2 might function in the LF expression of pax6 without regulating the cell cycle. Interestingly, the function of p27xic1 is known to be divided into two: regulation of cell cycle and differentiation (Vernon et al.,2003; Vernon and Philpott,2003), suggesting that Xhairy2 may function to inhibit differentiation by reducing p27xic1 expression. For instance, overexpression studies using deletion variants of p27xic1 suggested that the N-terminal portion of p27xic1 appeared to promote differentiation independent of cell cycle regulation in Xenopus retina (Ohnuma et al.,1999) and differentiating primary neurons (Vernon et al.,2003). To further our understanding, it will be necessary to reveal the molecular nature of p27xic1. For this purpose, it will be necessary to investigate potential protein–protein interactions of p27xic1, particularly that mediated through the N-terminal structure of p27xic1.
All in all, our data suggested that Xhairy2 would not be a master regulator of the lens program, but the lens program did not work in the absence of Xhairy2. This suggests that Xhairy2 may maintain an intracellular environment in which inducing signal(s) can be accepted. In other words, Xhairy2 may keep lens lineage cells competent to incoming inducing signal(s) by maintaining undifferentiated, or pluripotent, states of those cells via regulation of p27xic1 expression. Interestingly, our previous report showed that Xhairy2 in the anterior prechordal plate mesoderm and the neural crest, which are thought to be a group of pluripotent cells, is necessary to maintain tissue identity or the undifferentiated state of such cells (Yamaguti et al.,2005; Nagatomo and Hashimoto,2007). Maintenance of these states of cells in the presence of Xhairy2 might be in part related to the still-unknown molecular nature of lens competence, which is conceptually the basal state of ectoderm leading to lens formation (Henry and Grainger,1987).
Embryonic Manipulations and Micro-Injection
Xenopus laevis embryos were in vitro fertilized, dejellied, and cultured as described (Hawley et al.,1995), and staged according to Nieuwkoop and Faber (1967). For overexpression studies, capped mRNAs of Xhairy2b and p27xic1 were synthesized with SP6 RNA polymerase (mMESSAGE mMACHINE Kit, Ambion) from Xhairy2b/pCS2AT+ (Yamaguti et al.,2005) and p27xic1/pCS2AT+ (Nagatomo and Hashimoto,2007), which were linearized with Not1. Morpholino antisense oligos (MOs, purchased from Gene Tools) used in this study were standard control morpholino (Co MO), Xhairy2 morpholinos (Xhairy2 MOs), and p27xic1 morpholino (p27 MO; Vernon et al.,2003). Xhairy2 MO is a mixture of Xhairy2a MO (Murato et al.,2007) and Xhairy2b MO (Yamaguti et al.,2005), prepared as described (Nagatomo and Hashimoto,2007), in order to knock down both Xhairy2 pseudoalleles. To confirm the injected region in in situ hybridization, β˜-gal activity of cytoplasmic lacZ was visualized with 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal, Wako).
Whole-Mount In Situ Hybridization
Whole-mount in situ hybridization (WISH) was performed as described (Harland,1991) with minor modifications. Embryos were bleached before hybridization as described (Hino et al.,2003). For chromogenic reaction, nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indoxyl phosphate (NBT/BCIP, Roche) was used as the substrate of alkaline phosphatase, but the concentration of NBT (Roche) was modified as described (Ma et al.,1996). For two-color WISH, chromogenic reaction of the first color was stopped by incubating embryos overnight with PBS-buffered 4% paraformaldehyde at 4°C with gentle shaking. Before the second blocking, embryos were washed with PBS, 5 min × 4. Alkaline phosphatase substrates were NBT/BCIP, BCIP, or BCIP-Red (Biotium). When embryos were stained with the combination of BCIP and NBT/BCIP, NBT/BCIP staining was chosen for second color development because trace carryover of NBT deteriorates the following BCIP staining.
Antisense ribo-probes for hybridization were prepared from linearized template encoding pax6 (Hirsch and Harris 1997), six1 (Pandur and Moody,2000), six3 (Zhou et al.,2000), foxE3 (Lens1, Kenyon et al.,1999), L-maf (Ishibashi and Yasuda,2001), γ1-crystallin (Offield et al.,2000), foxD3 (Sasai et al.,2001), pitx-1 (Hollemann and Pieler,1999), Xhairy2b (Tsuji et al.,2003), rx1 (Mathers et al.,1997), a partial fragment of notch2 (Ogino et al.,2008), dlx5 (Luo et al.,2001), X-MyT1 (Bellefroid et al.,1996), p27xic1 (Su et al.,1995), pax8 (Heller and Brandli,1999), foxG1 (Bourguignon et al.,1998), six6 (Zuber et al.,1999), and XHes2 (Solter et al.,2006). γ1-Crystallin-containing plasmid was a kind gift from Dr. H. Ogino. Synthesis of labeled RNA was carried out with MAXIscript Kit (Ambion).
Histology and Immunohistochemistry
Tadpoles were fixed with MEMFA for 6 hr and dehydrated with 100% methanol overnight. Later, samples were embedded in paraffin (Pathoprep 568: Wako, Japan) and sectioned to 10-μm thickness. The sections were then deparaffinized and stained with hematoxylin.
For immunohistochemistry, tadpoles were fixed with MEMFA overnight and dehydrated with 100% methanol. Samples were then embedded into paraffin and sectioned at 10-μm thickness. Deparaffinized samples were rehydrated with PBS. Samples were blocked with 3% BSA in PBS for 1 hr at room temperature. Mouse primary antibody for γ-crystallin (a kind gift from Dr. H. Ogino) was diluted with 1% BSA in PBST (1:30). Samples were incubated with the antibody solution overnight at 4°C. After rinsing with PBST, samples were washed with PBST, 5 min × 3. For the secondary antibody, anti-mouse IgG HRP conjugate (Promega, W402B) was used at the dilution of 1:2,000 with 1% BSA in PBST. Incubation was performed overnight at 4°C. After rinsing with PBST, samples were washed with PBST, 3 min × 5. Then, samples were incubated with 25 mM TBS for 5 min and reacted with diaminobenzidine (Sigma, D4293). Finally, samples were counter-stained with Vector Hematoxylin QS (Vector Laboratory, H-3404).
Anti-Phospho Histone H3 Staining
Anti-phospho histone H3 (PHH3) staining post-WISH was carried out based on L´eger and Brand (2002) with minor modifications. After NBT/BCIP staining of WISH, samples were washed with PBST, 5 min × 3, and blocked with 1% BSA in PBST for 1 hr at room temperature. Then, they were incubated overnight at 4°C with anti-PHH3 primary antibody (rabbit polyclonal IgG; Upstate Biotechnology; 1: 300 in 1% BSA in PBST) on a shaker. Samples were rinsed with PBST three times and washed with PBST, 30 min × 4. They were then incubated overnight with the secondary antibody (goat anti-rabbit IgG HRP conjugated, Promega; 1:200 in 1% BSA/PBST) on a shaker. Samples were rinsed with PBST three times and washed with PBST, 30 min × 4, and with Tris 0.1 M pH 7.4, 5 min × 2. For chromogenic reaction, samples were incubated with DAB solution containing H2O2 (Sigma, D-4293). The color developed rapidly. Samples were finally fixed with MEMFA for 30 min and dehydrated in 100% methanol for observation.
Pharmacological Treatment and BrdU Labeling/Detection
Embryos were incubated in HUA, 150 μM aphidicolin (Sigma, A0781) and 20 mM hydroxyurea (Sigma, H8627) in 0.3% DMSO, from gastrula stage (st. 10+) to mid-neurula stage (st. 15). Some embryos were fixed at st. 15 for WISH analyses and others were further incubated in the absence of the drugs until fixation at st. 38. To confirm the effects of drug treatment, embryos were incubated in HUA from gastrula stage (st. 10+) until fixation at early neurula stage (st. 13). For Figure 7, during the treatment, embryos were injected with 10 mM BrdU at st. 12 (2 injections within neural plate region, 1 injection into blastocoel, 8 nl each) and incubated until fixation (around 3 hr at 16°C). For Figure 6, MO-injected embryos were further injected with 10 mM BrdU (either side of the anterior neural plate region, 8 nl each) and incubated for 1 hr at 20°C. BrdU detection was performed with BrdU Labeling and Detection Kit II (Roche, 11299964001) as described previously (Hardcastle and Papalopulu,2000). Stained embryos were paraffin-sectioned at 10-μm thickness (Fig. 7) or 14-μm thickness (Fig. 6). The number of BrdU-positive cells in LF (Fig. 6) or the neural plate region (Fig. 7) was counted across five successive sections. Signals within the neural plate region were counted with ImageJ software. For Figure 6, signals were manually counted.
We thank Hajime Ogino (Nara Institute of Science and Technology) for γ1-crystallin plasmid and antibody for γ-crystallin; Tetsuya Oka (Toyohashi University of Technology) for sub-cloning X-MyT1 and dlx5; and Kan-ichiro Nagatomo (Shin Nippon Biomedical Laboratories) for discussion.