DEVELOPMENTAL DYNAMICS 238:2149–2162, 2009

SPECIAL ISSUE REVIEWS–A PEER REVIEWED FORUM

Signaling “Cross-Talk” Is Integrated by Transcription Factors in the Development of the Anterior Segment in the Eye

Philip J. Gage* and Amanda L. Zacharias

Extracellular signaling “cross-talk” between tissues is an important requirement for development of many organs yet the underlying mechanisms generally remain poorly understood. The anterior segment of the eye, which is constructed from four embryonic lineages, provides a unique opportunity to genetically dissect developmental processes such as signaling “cross-talk” without fear of inducing lethality. In the current review, we summarize recent data showing that PITX2, a homeodomain transcription factor, integrates retinoic acid and canonical Wnt/␤-catenin signaling during anterior segment development. Because the requirements for retinoic acid signaling, canonical Wnt/␤-catenin signaling, and PITX2 are not unique to the eye, this newly identified pathway may have relevance elsewhere during development and in tissue homeostasis. Developmental Dynamics 238:2149–2162, 2009. © 2009 Wiley-Liss, Inc.

Key words: retinoic acid signaling; canonical Wnt signaling; anterior segment; homeobox

Accepted 8 June 2009

INTRODUCTION primordia are required for correct pat- development within the anterior seg- terning and cell specification, which ment and many transcription factors In 1901, Hans Spemann reported that requires cells to integrate multiple are essential for the process as well. the optic cup had the ability to cause signaling inputs. A significant advan- However, only recently has the iden- overlying head surface ectoderm to tage of using the eye as a model sys- tification of an initial cohesive path- form a lens, the first step in the devel- opment of the ocular anterior segment tem is that it is not required for via- way that integrates cell signaling be- (Spemann, 1901). Spemann termed bility. Using appropriate genetic tween tissues with transcriptional this property “induction” and argu- techniques, one can test developmen- networks emerged from this multi- ably launched the modern era of ex- tal processes in great molecular detail tude of factors. Work from several lab- perimental developmental biology. without fear of inducing lethality. Ad- oratories has now provided evidence More than a century later, the ante- vances in understanding basic devel- linking retinoic acid signaling from rior segment of the eye remains an opmental mechanisms gained while the optic cup, the homeodomain tran- important model for investigating ba- studying the anterior segment are of- scription factor PITX2 within the neu- sic developmental mechanisms. As ten highly relevant when subse- ral crest and canonical Wnt/␤-catenin with other organs, the mature ante- quently applied to the development of signaling within the ocular surface ec- rior segment arises from multiple em- other organs or tissues that are re- toderm. Pitx2 is a critical integration bryonic primordia. Extensive induc- quired for viability. node that links the two signaling tive interactions and signaling “cross- All major signaling pathways have pathways. This network is required talk” between and within the distinct been implicated in various aspects of during early eye development to es-

Departments of Ophthalmology and Visual Sciences, and Cell and Developmental Biology, University of Michigan Medical School, Ann Arbor, Michigan Grant sponsor: NEI/NIH; Grant numbers: EY014126, EY007003. *Correspondence to: Philip J. Gage, Department of Ophthalmology and Visual Sciences, University of Michigan Medical School, 305 Kellogg Eye Center, 1000 Wall Street, Ann Arbor, MI 48105. E-mail: [email protected] DOI 10.1002/dvdy.22033 Published online 21 July 2009 in Wiley InterScience (www.interscience.wiley.com).

© 2009 Wiley-Liss, Inc. 2150 GAGE AND ZACHARIAS tablish the basic framework of general tion of a network of transcription will eventually develop from the ante- patterning and initial differentiation factor genes marking first a single and rior rim of the optic cup. on which future morphogenesis of the subsequently two “eye fields” within The days following formation of the anterior segment depends. How addi- the anterior neural ectoderm (Chow lens vesicle (e10–13.5) are critically tional cell signaling pathways and and Lang, 2001). The optic pits, rep- important for anterior segment devel- transcription factors may contribute resenting the first morphological ev- opment, because the general frame- to this pathway remains to be deter- idence of eye development, form as work of patterning and initial differ- mined. localized thickenings of neural ecto- entiation steps on which the future derm within the inner surface of the cornea, limbus, conjunctiva, and eye- OVERVIEW OF ANTERIOR anterior neural folds at the positions lids, will depend are laid down. This SEGMENT initially marked by the two eye fields period is marked by changes in both (Pei and Rhodin, 1970; Kaufman, morphology and gene expression, and MORPHOGENESIS 1992). By embryonic day 9 (e9), the coincides with the time at which the Major structures/tissues of the ma- optic pits deepen to form the optic newly identified RA/PITX2/Wnt regu- ture anterior segment include the vesicles, outgrowths of neural ecto- latory network is first required (see lens, ciliary body, iris, cornea, limbus, derm on either side of the dienceph- below). At the beginning of this phase bulbar conjunctiva, and the eyelid, alon (Fig. 2A) (Pei and Rhodin, 1970; (e10), morphogenesis of the cornea which is comprised of the palpebral Kaufman, 1992). The optic vesicles and other central anterior segment epidermis, palpebral conjunctiva, and continue to expand laterally until tissues starts with migration of eyelid mesenchyme (Fig. 1A). Four they meet the ocular surface ecto- largely neural crest mesenchyme into embryonic tissues contribute the cel- derm (Fig. 2A). the space between the newly formed lular building blocks for anterior seg- The ocular surface ectoderm (OSE) lens vesicle and the OSE (Fig. 2C) (Pei ment structures: neural ectoderm, is a multipotent patch of head surface and Rhodin, 1970; Kaufman, 1992; surface ectoderm, and neural crest ectoderm that, when appropriately Cvekl and Tamm, 2004). The result- and mesoderm mesenchyme (Fig. 1B). patterned, is fated to contribute the ing 7–8-cell-thick layer of mesen- Each of these primordia eventually lens, all epithelia of the eye surface chyme becomes increasingly more or- contributes multiple lineages within (cornea, limbus, conjunctiva), ecto- ganized and condensed, implying a the mature anterior segment (Fig. derm-derived components of the lacri- response to local cues (Fig. 2D). The 1B). In addition, most mature anterior mal and Harderian (in mice) glands, transcription factor genes Foxc1, segment structures are a synthesis of and the eyelid epidermis (Figs. 1, 2A) Foxc2, Lmx1b, and Pitx2 are all ini- both ectoderm and mesenchyme. For (Ashery-Padan et al., 2000). Anterior tially expressed in specific patterns example, in the cornea, the corneal segment development commences when within the mesenchyme at this time epithelium derives from surface ecto- the distal optic vesicle makes direct and are subsequently required for an- derm whereas the corneal stroma and physical contact with the overlying OSE terior segment development (Gage the corneal endothelium derive from on e9.5 in mice, resulting in induction of and Camper, 1997; Kidson et al., mesenchyme (Fig. 1A,B) (Gage et al., the lens placode within the OSE (Fig. 1999; Pressman et al., 2000; Kume et 2005). This is fundamentally different 2A) (Pei and Rhodin, 1970; Kaufman, al., 2001). Beginning shortly after im- from, e.g., retinal development, where 1992). Therefore, the optic vesicle acts migration of the neural crest mesen- all differentiated neurons and the early in anterior segment development chyme, distinct regions of the OSE Mueller glia differentiate from a sin- as an organizing center that is both nec- corresponding to the future corneal, gle, common pool of neural ectoderm essary and sufficient for specifying limbal, and conjunctival ectoderm, precursors. Work from many model where within the ocular surface ecto- and eyelid epidermis can be identified systems has contributed to our cur- derm the lens primordia will form. The based on differential expression and rent understanding of vertebrate eye lens placode invaginates towards the sub-cellular localization of Connexin development, including the anterior head midline, forming a lens pit (Fig. 43, a gap junction protein that is sub- segment (Jacobson and Sater, 1988; 2B), which then separates from the sequently essential for corneal func- Furukawa et al., 1999; Beebe and OSE to form the lens vesicle (e10.5), the tion (Wolosin et al., 2002). In addition, Coats, 2000; Fuhrmann et al., 2000). precursor to the mature lens (Fig. 2C) by e12.5 differential levels of canoni- Because some interspecies differences (Pei and Rhodin, 1970; Kaufman, 1992). cal Wnt/␤-catenin signaling activity exist, mouse will be used as the refer- The molecular cascade leading to lens distinguish future elements arising ence point for summarizing relevant development is known in considerable from the OSE (Smith et al., 2005). The stages of eye development and ante- detail and has been summarized else- appearance of the eyelid grooves rior segment morphogenesis in this where (Chow and Lang, 2001; Lang, within the OSE dorsal and ventral to review as a prelude to mouse genetics 2004). Coincident with formation of the the optic cup at e11.5 is the first mor- experiments that will be described lens vesicle, the optic vesicle is con- phological evidence of eyelid develop- later. verted into an optic cup, the inner layer ment (Fig. 2C). Several steps in eye development forming the retina, whereas the outer At e14.5, corneal development con- precede and are required for initiation layer will form the retinal pigmented tinues with the emergence of the pre- of normal anterior segment morpho- epithelium (Fig. 2B). The epithelium of sumptive corneal endothelium as a genesis. The earliest molecular evi- the ciliary body and iris, two important continuous sheet of cells from the mono- dence for the eye primordia is activa- structures within the anterior segment, layer of mesenchyme directly overlying SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2151

layer of surface ectoderm, which will form the future palpebral conjunc- tiva on the inner surface of the eyelid and the palpebral epidermis on the outer surface (Findlater et al., 1993). Regularly spaced eyelash follicles be- gin to emerge at the interface between the palpebral epidermis and conjunc- tiva towards the end of eyelid closure. Beginning approximately e15.5, peri- derm cells located at the leading edge of each eyelid rim begin to proliferate and migrate across the cornea until they meet and fuse by e16.5 (Fig. 2E,F). The eyelids will ultimately re- open by approximately postnatal day 10 (Findlater et al., 1993; Martin and Parkhurst, 2004; Xia and Karin, 2004). The increased organization and com- paction visible within the periocular mesenchyme (POM) of the anterior seg- ment shortly after migration between the lens vesicle and overlying OSE im- plies that the POM is responding to lo- cal cues. The concurrent activation of specific transcription factor genes within the POM required for anterior segment morphogenesis and corneal de- velopment further strengthens this view. Similarly, the differential expres- sion of Cx43 and the coincident emer- Fig. 1. Identities and embryonic origins of major anterior segment tissues. A: Adult eye with annota- tion of the major structures and tissue of the anterior segment. B: Adult eye illustrating embryonic gence of morphological changes demon- origins of major tissues from ocular surface ectoderm (blue), periocular mesenchyme (neural crest and strate that patterning within the OSE mesoderm, red/orange), and neural ectoderm (green). A, iridocorneal angle; BC, bulbar conjunctiva; CB, is established within a similar time ciliary body; CE, corneal epithelium; CEn, corneal endothelium; CS, corneal stroma; E, eyelid; EL, window. Disruption of these early pro- eyelash cilium; I, iris; L, lens; OSE, ocular surface ectoderm; PC, palpebral conjunctiva; PE, palpebral epidermis; POM, periocular mesenchyme (both neural crest and mesoderm); SC, Schlemm’s canal; cesses generally has a negative impact TM, trabecular meshwork. on subsequent steps in anterior seg- ment development. For example, muta- tions affecting differentiation of the cor- the anterior lens and the anterior rim of rior rim of the optic cup becomes pig- neal endothelium result in persistent the optic cup (Pei and Rhodin, 1970; mented and begins to grow into the cav- attachment of the posterior cornea to Kaufman, 1992; Cvekl and Tamm, ity between the lens and cornea, the lens (Blixt et al., 2000; Brownell et 2004). The maturing corneal endothe- generating the epithelia of the future al., 2000; Ring et al., 2000; Semina et lium physically dissociates from the an- ciliary body and iris. A new migration of al., 2001; Ormestad et al., 2002), and terior lens, forming a space that will mesenchyme into the angle formed be- disrupted patterning early during de- become the future anterior chamber tween the iris and the cornea (iridocor- velopment of the OSE can result in the once aqueous humor production begins neal angle) provides the precursors that subsequent emergence of ectopic eye- (Pei and Rhodin, 1970; Kaufman, 1992). will form the stroma of the ciliary body lash follicles within the conjunctiva Keratocytes of the corneal stroma dif- and iris, as well as components of the (distichiasis) (Fang et al., 2000; Kried- ferentiate somewhat later from mesen- future outflow tract within the iridocor- erman et al., 2003). Classic embryology chyme located between the newly neal angle. Development of these later experiments as well as mutant pheno- formed corneal endothelium and the emerging structures continues for sev- types in mice and humans implicate the overlying presumptive corneal epithe- eral weeks after birth (Gould et al., lens and anterior optic cup as likely lium, which arises from the OSE (Fig. 2004). sources of signaling molecules for orga- 2E) (Pei and Rhodin, 1970; Kaufman, After their patterning and specifi- nizing the anterior segment, but the 1992). By late gestation, keratocytes cation, eyelid primordia grow in size identities of the specific factors, as well adopt a lamellar arrangement and be- and extend toward each other across as the nature of the downstream re- gin to secrete a highly specialized extra- the cornea until about e15.5 (Fig. sponse, are unknown. Recent data now cellular matrix, properties that are re- 2E). During this period, the eyelid provide proof that the optic cup acts as a quired for corneal transparency. After consists of a loose array of mesen- signaling center that orchestrates ante- the cornea and lens separate, the ante- chyme sheathed within an outer rior segment morphogenesis through 2152 GAGE AND ZACHARIAS direct action on the adjacent neural crest, and, indirectly through action of the neural crest, on patterning and morphogenesis within the OSE.

RETINOIC ACID SIGNALING ORCHESTRATES ANTERIOR SEGMENT MORPHOGENESIS DURING THE CRITICAL TIME WINDOW FOLLOWING LENS VESICLE FORMATION Vitamin A is essential for eye develop- ment and exerts its biological effects through retinoic acid (RA), an enzy- matic metabolite. The primary activ- ity of RA during development is to act as an autocrine or paracrine signaling molecule (Fig. 3A). In this capacity, RA binds to nuclear receptors, activat- ing these ligand-gated transcription factors within competent cells (Fig. 3A). The nuclear receptors are consti- tuted by proteins from two families, the retinoic acid receptors (RAR) ␣, ␤, and ␥, and the retinoid X receptors (RXR) ␣, ␤, and ␥ (Chambon, 1996). The functional unit is a RAR/RXR het- erodimer and all RAR and RXR pro- teins are expressed in the developing eye (Ghyselinck et al., 1997; Mori et al., 2001). RAR proteins bind both all- trans- and 9-cis-RA, whereas RXR proteins bind only 9-cis-RA. However, 9-cis-RA is the active factor in cell sig- naling (Chambon, 1996). Evidence of a requirement for RA signaling in eye development first came from reports Fig. 2. Morphogenesis of the anterior segment. The anterior segment develops from ocular surface ectoderm (blue), neural ectoderm (green), and neural crest and mesoderm mesenchyme (orange). A: of the severe congenital eye defects Anterior segment morphogenesis is initiated when direct interaction of the distal optic vesicle induces present in vitamin A–deficient (VAD) formation of the lens placode within the overlying ocular surface ectoderm. B: The lens placode forms and/or RA-deficient human and live- by invagination of the lens pit while the optic vesicle is concurrently converted in the optic cup. The inner stock fetuses (Hale, 1933; Warkany layer of optic cup will form the retina, the outer layer will form the retinal pigmented epithelium, and the and Schraffenberger, 1946; Wilson et anterior rim of the optic cup will contribute the epithelial components of the ciliary body and iris. C: Loosely arranged mesenchyme migrates between the newly formed optic vesicle and anterior optic cup al., 1953; Dickman et al., 1997; Dupe and the overlying presumptive cornea. Eyelid folds emerge as the first morphological evidence of eyelid et al., 2003). Dysgenesis of the ante- development. D–F: Compaction of corneal stroma and eyelid growth continues. Extension of periderm rior segment is prominent in these in- from eyelid margins eventually results in eyelid fusion. C, cornea; E, eyelid; EF, eyelid folds; L, lens; LP, dividuals, and includes agenesis of the lens placode; LV, lens vesicle; M, mesoderm; OSE, ocular surface ectoderm; OV, optic vesicle; P, periderm; Pit, lens pit; R, retina; PRE, retinal pigmented epithelium. lens, iris and corneal stroma, the cor- neal endothelium, an absence of the anterior chamber, eyelids open at 1999). Genetic ablation of two or more types, vitamin A is converted to reti- birth, and (in rodents) no Harderian RAR-encoding genes in mice results in naldehyde by the enzyme, alcohol de- glands. Additional ocular features in- severe eye defects that mimic those hydrogenase type 3 (Fig. 3A) (Molot- clude retinal and optic disc coloboma, seen in severe VAD (Kastner et al., kov et al., 2002). Second, in selected a foreshortened ventral retina, and 1994, 1997; Lohnes et al., 1994; Ghy- cell types, retinaldehyde is converted to agenesis of the sclera. Humans with selinck et al., 1997; Mascrez et al., retinoic acid (either all-trans or 9-cis) mutations inhibiting the function of 1998). Therefore, the competence to by a retinaldehyde dehydrogenase serum retinol binding protein have respond to RA signaling is critical for (Fig. 3A) (Duester, 2000). Mammals eye defects, as do mice with a reduced normal eye development. have four retinaldehyde dehydrogenase capacity for RA receptor signaling Conversion of vitamin A to RA is a proteins (RALDH1-4) encoded by the (Lohnes et al., 1994; Seeliger et al., two-step process. First, in all cell genes Aldh1a1, Aldh1a2, Aldkh1a3, SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2153

specification and morphogenesis of the lens (Matt et al., 2005; Molotkov et al., 2006). Functional redundancy amongst the RALDH proteins does not account for this surprising result, since retinal patterning and lens ves- icle formation also appear unaffected in double gene knockout mice lacking any detectable RALDH activity in the eye (Matt et al., 2005; Molotkov et al., 2006). Therefore, RA signaling is not required for lens induction and mor- phogenesis, the earliest stages of an- terior segment development. In contrast, RA signaling is critically required for the next phase of anterior segment morphogenesis, which in- cludes initiation of all three layers of the cornea, patterning the remaining OSE into presumptive corneal, limbal, and conjunctival ectoderm and eyelid epidermis, and the emergence of the early eyelid primordia. RALDH1 is highly expressed in the corneal ecto- derm, lens, and dorsal retina through- out this time period (e10.5–13.5), and at lower levels in the ventral retina begin- ning by e11.5 (Fig. 3B) (Matt et al., 2005; Molotkov et al., 2006). Concur- rently, RALDH3 is strongly expressed in the corneal ectoderm, retinal pig- mented epithelium, and ventral retina, with weaker expression in the dorsal retina (Fig. 3B) (Matt et al., 2005; Mo- lotkov et al., 2006). In mice transgenic Fig. 3. Retinoic acid signaling during development of the anterior segment. A: Simplified overview for an RA-responsive lacZ transgene of retinoic acid production and function in intracrine or paracrine signaling. Note that additional (RARE-lacZ), ␤-gal is expressed in the factors involved in functions such as transport have been omitted for simplicity. B: Summary of tissues producing retinoic acid (blue) at e11.5 based on expression of RALDH1 and/or RALDH3. C: retina, optic nerve, corneal ectoderm, Summary of tissues responding to retinoic acid signaling at e11.5 based on expression pattern of and periocular mesenchyme at this RARE-lacZ reporter transgene. C, cornea; ADH, alcohol dehydrogenase; RAR, retinoic acid recep- stage, establishing the responsiveness tor; RALDH1-3, retinaldehyde dehydrogenase 1–3; RDH10, retinol dehydrogenase 10; M, mesen- of these tissues to RA (Fig. 3C) (Matt et chyme; OS, optic stalk; R, retina; PRE, retinal pigmented epithelium; RXR, retinoid X receptor. al., 2005; Molotkov et al., 2006). This same transgene is silent in the eyes of and Aldh8a1, respectively (Duester, icle at e8.5 (Li et al., 2000; Wagner et mutant mice lacking both RALDH1 and 2000; Lin et al., 2003). Temporal-spatial al., 2000; Mic et al., 2004). By e9.5, RALDH3, confirming that these two en- specificity of RA production, and there- RALDH1 is strongly expressed in the zymes account for all RA signaling ac- fore the capacity for RA signaling, is dorsal retina while RALDH3 is tivity in the eye during this critical pe- largely controlled by selective expres- strongly expressed in the ventral ret- riod of anterior segment development sion of the RALDH proteins during de- ina (McCaffery et al., 1999; Grun et (Matt et al., 2005; Molotkov et al., velopment. In addition, RA signaling al., 2000; Li et al., 2000; Mic et al., 2006). can be fine tuned by the activity of RA- 2000; Suzuki et al., 2000), prompting Genetic ablation of Aldh1a1 degrading enzymes such as CYP26A1 the hypothesis that RA signaling may (RALDH1) has little effect on expres- and CYP26C1 (Fujii et al., 1997; Sakai play an essential role in dorsal/ventral sion of the RARE-lacZ transgene or et al., 2004). patterning of the retina during devel- eye development (Matt et al., 2005; RALDH1–3 are expressed in dy- opment (Wagner et al., 2000; Drager Molotkov et al., 2006). Ablation of namic, overlapping patterns begin- et al., 2001; Peters and Cepko, 2002; Aldh1a3 (RALDH3) abolishes expres- ning early in eye development, coin- Sakai et al., 2004). However, func- sion of the RARE-lacZ reporter in the ciding with conversion of the optic tional studies including individual RPE, ventral corneal ectoderm, ven- vesicle into the optic cup and forma- gene ablation have not supported a tral retina, and ventral POM while tion of the lens vesicle. RALDH2 is role for RA signaling in early dorsal/ expression in the dorsal tissues is un- transiently expressed in the optic ves- ventral patterning in the retina or affected (Matt et al., 2005; Molotkov et 2154 GAGE AND ZACHARIAS al., 2006). Aldh1a3 mice exhibit mild primary target of RA signaling in the expresses lacZ in response to canoni- to moderate ventral defects, including eye during this critical period of ante- cal Wnt signaling has been used to shortening of the ventral retina, ven- rior segment patterning and morpho- evaluate pathway activity during eye tral rotation of the lens, retrolenticu- genesis. Loss of RA signaling does not development (Liu et al., 2003; Smith lar membrane, and thickening of the appear to affect neural crest cell sur- et al., 2005; Miller et al., 2006). The ventral POM (Matt et al., 2005; Molot- vival; on the contrary, apoptosis is re- levels of canonical Wnt/␤-catenin sig- kov et al., 2006). The latter appears to duced in neural crest of the mutant naling activity vary significantly in be due to a decrease in apoptosis that animals (Matt et al., 2005; Molotkov time and space within the OSE during normally occurs within the ventral et al., 2006). Absence of RA signaling the critical period for patterning and POM (Matt et al., 2005; Molotkov et results in the loss of expression within differentiation of the individual cell al., 2006). Similar apoptosis occurs the neural crest of transcription factor fates. Prior to and during the early within the dorsal POM of wild type genes (Eya2, Foxc1, and Pitx2) that phases of lens induction and morpho- mice but this is unaffected in Aldh1a3 are required for normal anterior seg- genesis (e8.5–9.5), canonical Wnt/␤- mutants. In contrast to the single mu- ment morphogenesis (Matt et al., catenin signaling activity is absent tants, ablation of both Aldh1a1 and 2005; Molotkov et al., 2006). There- from the OSE and remains undetect- Aldh1a3, and the accompanying ab- fore, RA signaling to the neural crest able during the lens vesicle stage sence of RA signaling, has profound orchestrates anterior segment mor- (e10.5) (Miller et al., 2006). Pathway effects on anterior segment patterning phogenesis at least in part through activity remains undetectable in cen- and morphogenesis (Matt et al., 2005; the activation of essential transcrip- tral tissues including the presumptive Molotkov et al., 2006). Anterior seg- tion factor genes. Whether or not the corneal ectoderm at e12.5, but is ment phenotypes include agenesis of effects on the target genes are direct markedly elevated peripherally, in- the corneal endothelium and stroma, or indirect and details regarding addi- cluding in the palpebral epidermis iris stroma, and formation of a thick tional downstream targets remain un- and conjunctiva (Fig. 4C) (Smith et layer of mesenchyme between the lens known. al., 2005). Local activity in the mesen- and cornea. Invagination of the dorsal chyme parallels that in the respective eyelid groove fails, leading to the ab- overlying ectoderm (Fig. 4C) (Smith et sence of a dorsal eyelid while only a CANONICAL WNT al., 2005). rudimentary eyelid forms ventrally. SIGNALING AND Manipulating canonical Wnt/␤-cate- This ventral eyelid remains small and PATTERNING OF THE nin signaling activity dramatically al- ultimately fuses with the presumptive OCULAR SURFACE ters patterning and development corneal ectoderm, forming a hypoplas- within the OSE. Genetically, the path- tic conjunctival sac. Additional pheno- ECTODERM way can be blocked by ablating the types include a persistent retrolentic- Canonical Wnt signaling, mediated by Catnb (␤-catenin) gene or activated by ular membrane, ventral rotation of the downstream effector protein rearranging Catnb to produce a highly the lens, shortening of the ventral ret- ␤-catenin, is widely required during stable form of ␤-catenin (Harada et ina, retinal coloboma, and agenesis of development, as well as stem cell biol- al., 1999; Brault et al., 2001). Gener- the sclera. Providing the pregnant ogy and adult tissue homeostasis (Lo- ating abnormally high levels of path- dam with an RA-supplemented diet gan and Nusse, 2004). Temporal-spa- way activity throughout the OSE largely restores normal eye develop- tial patterns of Wnt pathway activity blocks induction of the normal, cen- ment in the mutants (Molotkov et al., within a developing tissue or organ trally placed lens placode by the optic 2006). However, the competence of are tightly regulated (Logan and vesicle (Smith et al., 2005), whereas neural crest to respond to RA by acti- Nusse, 2004). Dys-regulation of ca- ablating pathway activity throughout vating Pitx2 expression and other nonical Wnt/␤-catenin signaling leads the OSE results in lentoid bodies steps required for normal anterior to abnormal development and plays a forming within the peripheral OSE segment development may be tran- prominent role in a number of dis- (Miller et al., 2006). Therefore, the sient (See and Clagett-Dame, 2009). eases, including cancer (Nusse, 2005). competence of cells within the OSE to The wild type expression of the Canonical Wnt/␤-catenin signaling ac- form lens correlates with low levels of RARE-lacZ reporter in multiple ocu- tivity is initiated by Wnt ligand bind- canonical Wnt/␤-catenin signaling ac- lar tissues combined with the severe ing to a heterodimeric cell surface re- tivity, suggesting that differences in Aldh1a1/3 phenotype raises the possi- ceptor, resulting in stabilization and the levels of signaling activity play an bility that normal anterior segment translocation of ␤-catenin to the nu- essential role in patterning the OSE. development could require intact RA cleus where it partners with TCF/LEF As eye development progresses, mu- signaling to the retina, surface ecto- transcription factors (Fig. 4A) (Logan tants with abnormally high canonical derm, and/or POM (Fig. 3C). However, and Nusse, 2004; Nusse, 2005). Ca- Wnt/␤-catenin signaling activity neural crest-specific ablation of two nonical Wnt/␤-catenin signaling can throughout the OSE fail to develop a (Rarb and Rarg) or all of the three promote gene transcription, cell prolif- cornea, conjunctiva, or eyelids, indi- genes encoding the RAR nuclear re- eration, differentiation, cell survival, cating that correct levels of pathway ceptors results in mice with ocular or developmentally programmed apo- activity remain essential for normal phenotypes identical to those in the ptosis (Logan and Nusse, 2004; Nusse, anterior segment development in the Aldh1a1/3 mice (Matt et al., 2005, 2005). period following lens formation 2008). Therefore, neural crest is the A TOPGAL reporter transgene that (Miller et al., 2006). Collectively, these SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2155 data indicate that normal patterning and differentiation of structures de- rived from the OSE correlate strongly with achieving the correct temporal/ spatial activation of canonical Wnt/␤- catenin signaling activity. Temporally and spatially restricted expression of Wnt ligands and other components required for signal trans- duction would be one mechanism that could account for the required local activation or suppression of pathway activity during anterior segment de- velopment. However, several Wnt li- gands that are active in the canonical signaling pathway, as well as multiple Frizzled receptor proteins required for Wnt signaling, are highly expressed in the presumptive corneal ectoderm during the period when pathway ac- tivity in the same cells and adjacent mesenchyme is low or undetectable (Fig. 4B) (Liu et al., 2003; Smith et al., 2005). Therefore, expression of the required signaling factors does not correlate with pathway activity levels in the developing OSE through this time period (Liu et al., 2003). This suggests that the pathway may be ac- tively inhibited. Canonical Wnt/␤-cate- nin signaling activity can be inhibited by secreted frizzled-related proteins (SFRP1-4) and Dickkopf proteins ␤ (DKK1–4), which antagonize active Fig. 4. Canonical Wnt/ -catenin signaling during development of the anterior segment. A: Left: Both the LRP5/6 and Fz proteins are present on the surface of a competent cell, providing available signaling by binding to and inactivating Wnt ligand with the full heterodimeric receptor required to transduce the signal. Activation of the full either the Wnt ligand (SFRPs) or the receptor by Wnt ligand induces biochemical changes within the cytoplasm that lead to accumu- Wnt receptor (DKKs) (Logan and lation of stabilized ␤-catenin, which is translocated to the nucleus where it activates transcription Nusse, 2004; Nusse, 2005; Niehrs, of target genes. Right: DKK proteins, when present, render a cell incompetent to respond to Wnt ligand because the LRP5/6 component required for receptor function is removed from the cell 2006). Although multiple Sfrp and membrane. DKK proteins bind LRP5/6, in combination with a third protein called Kremen, and this Dkk genes are developmentally ex- complex is cleared from the cell surface by endocytosis and subsequently degraded. This leaves pressed in the eye, data on the expres- the cell incompetent to transduce a signal to the nucleus, even when high quantities of Wnt ligand sion patterns within the relevant time are present, because a fully functional heterodimeric receptor cannot be formed. B: Summary of are incomplete and, in some cases, not Wnt ligand and Frizzled receptor expression sites (blue) in mouse eye tissues at e12.5. C: Summary of structures/cells exhibiting high (blue) or low (white) canonical Wnt signaling activity in mouse eye in agreement (Monaghan et al., 1999; tissues at e12.5. BC, bulbar conjunctiva; C, cornea; L, lens; Dkk, Dickkopf; Fz, any frizzled protein; Liu et al., 2003; Ang et al., 2004; Chen Krm, Kremen; LRP5/6, low-density lipoprotein receptor-related protein 5 or 6; M, mesenchyme; PC, palpebral conjunctiva; PE, palpebral epidermis; TCF, transcription factor 4; Wnt, Wnt ligand.

Fig. 5. Current model and important questions for future analysis. Established genetic and molecu- lar pathways are shown in blue. Known candi- dates that should be tested for potential rele- vance to the network are shown in orange. More open-ended unknown questions are shown in green. These include what other factors in addi- tion to RA may be required to activate Eya2, Foxc1, and/or Pitx2 (Q1), whether additional fac- tors also signal from the ocular surface ectoderm (Q2), what other targets lie immediately down- stream of RA signaling (Q3), and what other genes and classes of genes, in addition to genes for transcription factors Dkk2, are a part of the required response to RA signaling within the neu- ral crest (Q4). These questions are discussed in greater detail in the Perspectives section. Fig. 5. 2156 GAGE AND ZACHARIAS et al., 2004). In addition, until re- and Walter, 2000). The anterior seg- never expressed. These defects must cently, no functional analysis on the ment defects can affect multiple lin- arise out of non-cell-autonomous mech- effects of ablating SFRPs or DKKs in eages derived from the POM, includ- anisms, providing the first evidence the developing eye had been reported. ing the corneal and iris stroma, that critical downstream targets of Therefore, the mechanism(s) underly- corneal endothelium, and trabecular PITX2 might include genes encoding ing the essential temporal and re- meshwork and Schlemm’s canal factors involved in cell signaling path- gional activation and suppression of within the iridocorneal angle (Semina ways. canonical Wnt/␤-catenin signaling ac- et al., 1996). Axenfeld-Rieger patients Dkk2, which encodes a secreted ex- tivity within the developing OSE was also have elevated intraocular pres- tracellular antagonist of canonical unknown. sure, which likely contributes to their Wnt/␤-catenin signaling, was recently high risk for glaucoma. Ablation of identified as an important down- Pitx2 in mice results in agenesis of the stream effector of PITX2 during ocu- PITX2 INTEGRATES RA corneal endoderm and stroma, iris lar anterior segment morphogenesis AND CANONICAL WNT/␤- stroma, and the replacement of these (Gage et al., 2008). Normally, canoni- CATENIN SIGNALING structures by a thick layer of mesen- cal Wnt/␤-catenin signaling activity ACTIVITY DURING chyme lying between the lens and cor- requires the binding of Wnt ligands to ANTERIOR SEGMENT nea (Gage et al., 1999; Kitamura et heterodimeric cell surface receptors al., 1999; Lin et al., 1999; Lu et al., consisting of a Frizzled (FRZ) protein DEVELOPMENT 1999). In addition, eyelid development and a LRP5/6 protein (Logan and Pitx2 encodes a homeodomain tran- is severely affected. Elsewhere, there Nusse, 2004). DKK2 protein antago- scription factor (PITX2) whose expres- is shortening of the ventral retina, nizes canonical Wnt/␤-catenin signal- sion is first detectable within the neu- coloboma formation, ventral rotation ing by binding LRP5/6 together with a ral crest of the anterior segment of the lens, a severe dysmorphogen- third protein, Kremen, resulting in a beginning at e9.5–10, shortly after esis of the optic nerve, and agenesis of trimeric complex that is cleared from lens vesicle formation and early in the the sclera and extraocular muscles. the cell surface by endocytosis and de- critical period for subsequent anterior Mutants in which Pitx2 has been ab- graded within the cell by the proteo- segment morphogenesis (Hjalt et al., lated specifically in neural crest re- some (Niehrs, 2006; Gage et al., 2008). 2000; Gage et al., 2005). These initial capitulate many ocular features of The resulting inability to form a func- data suggest that Pitx2 expression is the global knockout mice, including tional heterodimeric receptor makes activated within ocular neural crest in within the anterior segment (Evans the cell incompetent to transduce a response to local cues, a fact confirmed and Gage, 2005). Therefore, Pitx2 ␤-catenin signal even in the presence by the recent demonstration that function in neural crest is required of high concentrations of Wnt ligand. Pitx2 expression is dependent on RA for the development of multiple ocu- Dkk2 expression is first detected in signaling from the optic cup (Matt et lar tissues, including the anterior anterior segment POM by e11 and al., 2005; Molotkov et al., 2006). Over segment. Anterior segment defects spreads to POM surrounding the optic the next 1–2 days, Pitx2 expression in global and neural crest–specific cup and extending into the emerging expands to include neural crest sur- Pitx2 knockout mice appear identi- eyelids over the next 1-2 days (Mon- rounding the optic cup and optic stalk, cal to those reported for the loss of aghan et al., 1999; Gage et al., 2008). within the hyaloid space, and extend- RA signaling mice, raising the possi- Unlike Pitx2, Dkk2 is not expressed in ing into the eyelid mesenchyme. Pitx2 bility that Pitx2 is potentially the the extraocular muscle primordia is also expressed in mesoderm within major downstream effector of RA sig- (Monaghan et al., 1999; Gage et al., the developing eye. Unlike neural naling in the neural crest. (Gage et 2008). Although the expression pat- crest, Pitx2 expression is high in me- al., 1999; Matt et al., 2005; Molotkov terns are nearly identical, the timing soderm, before these cells associate et al., 2006). of Dkk2 activation in the POM lags with the eye primordia, suggesting Although the importance of Pitx2 behind that of Pitx2, consistent with it that Pitx2 activation in mesoderm oc- for regulating eye development is well being a downstream effector (Gage et curs by a different mechanism than in established, little is known regarding al., 2008). Dkk2 is absent from eyes of neural crest. Ultimately, neural crest the underlying mechanism(s) or key global and neural crest–specific Pitx2 and mesoderm expressing Pitx2 will downstream regulatory genes/path- knockout mice and is significantly re- contribute to the structures through- ways. Many of the phenotypes in mice duced in Pitx2ϩ/Ϫ mice (Gage et al., out the anterior segment, as well as and humans affect cell types derived 2008). In addition, canonical Wnt/␤- the sclera, extraocular muscles, and from neural crest and mesoderm, lin- catenin signaling activity levels are el- all ocular blood vessels (Gage et al., eages in which Pitx2 is expressed, sug- evated throughout the anterior seg- 2005). gesting that cell-autonomous mecha- ment of Pitx2 mutant eyes (Gage et Heterozygous mutations in human nisms, such as regulation of additional al., 2008). PITX2 protein physically PITX2 are one cause of Axenfeld- transcription factor genes, account for associates with Dkk2 promoter se- Rieger Syndrome, a congenital condi- these defects. In contrast, the shortened quences in primary cultures of embry- tion resulting in a variable anterior ventral retina, coloboma, and dysmor- onic POM and a Dkk2 genomic frag- segment dysmorphogenesis and a phogenesis of the optic nerve all arise ment encompassing these sequences high risk for glaucoma (Semina et al., from developmental defects in neural is responsive to PITX2 in a dose-de- 1996; Kulak et al., 1998; Kozlowski ectoderm, a tissue in which Pitx2 is pendent manner when linked to a lu- SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2157 ciferase reporter cassette (Gage et al., in these mice leading this mesen- through production of RA, the optic 2008). Collectively, these data indi- chyme and surface ectoderm to at cup acts as a signaling center nucleat- cate that Dkk2 is a direct target of least partially adopt characteristics of ing anterior segment morphogenesis. PITX2 in neural crest during eye de- conjunctiva, a normally more periph- This function is strikingly analogous velopment. eral fate (Gage et al., 2008). Elevated to the role played by its predecessor, Ablation of Dkk2 results in dys-reg- canonical Wnt/␤-catenin signaling the optic vesicle at an earlier time ulation of canonical Wnt/␤-catenin within the mutant cornea mirrors that point in inducing the initiation of lens signaling levels, disrupted patterning, normally found in the conjunctiva, development at a specific site within and altered fate specification within providing evidence that patterning the overlying OSE. The transcrip- the OSE (Mukhopadhyay et al., 2006; achieved by activation/suppression of tional response within the neural Gage et al., 2008). Canonical Wnt/␤- elevated canonical Wnt/␤-catenin sig- crest includes activation of Eya2, catenin signaling activity is extremely naling plays a key role in specifying a Foxc1, and Pitx2, although it is highly low to undetectable in developing wild conjunctiva versus cornea fate. Post- likely that additional important tar- type bulbar conjunctiva, limbal, and natally in Dkk2 mutant eyes, ectopic gets are required (Fig. 5). corneal ectoderm, tissues that directly blood vessels extend from the iris col- The anterior segment phenotypes overlie the mesenchymal cells ex- larette across the anterior chamber to shared by RA-deficient and Pitx2 pressing DKK2 (Liu et al., 2003; the central corneal endothelium (Gage knockout mice establish that Pitx2 is a Smith et al., 2005; Miller et al., 2006; et al., 2008). The specific underlying critical target of RA signaling in the Gage et al., 2008). In contrast, canon- mechanism(s) leading to ectopic blood neural crest. One essential direct tar- ical Wnt/␤-catenin signaling activity vessel growth remain to be deter- get of PITX2 in the neural crest is the is elevated throughout the OSE of mined but it is likely that elevated gene encoding DKK2, a potent antag- Dkk2-deficient eyes, including the canonical Wnt/␤-catenin signaling ac- onist of canonical Wnt/␤-catenin sig- presumptive bulbar conjunctiva, lim- tivity is a general contributing mech- naling activity (Fig. 5). An important bus, and cornea (Mukhopadhyay et anism, since it is angiogenic in other property of DKK2 function is that the al., 2006; Gage et al., 2008). Specific systems (Duplaa et al., 1999; Wright target cells are rendered incompetent defects include mis-specification of et al., 1999; Longo et al., 2002; Mas- to mount a canonical Wnt/␤-catenin corneal ectoderm as conjunctival gob- ckauchan et al., 2005; Skurk et al., signaling response even in an environ- let cells, ectopic eruption of eyelash 2005; Goodwin et al., 2006). There- ment with significant concentrations follicles from surface ectoderm within fore, the abnormally high levels of ca- of Wnt ligand, such as the developing the palpebral and bulbar conjunctiva, nonical Wnt/␤-catenin signaling activ- cornea. DKK2 expressed downstream and limbus, and rudimentary growth ity may directly promote abnormal of PITX2 acts by both paracrine (act- of the eyelid (Gage et al., 2008). There- blood vessel growth within the mutant ing on the surface ectoderm) and au- fore, despite uniform high expression corneas. Alternatively, the normally tocrine (acting on the neural crest) of Wnt ligands and other components low levels of canonical Wnt/␤-catenin mechanisms to effectively block ca- of the pathway, the paracrine effects signaling activity may be required to nonical Wnt/␤-catenin signaling activ- of DKK2 represent at least one essen- make the cornea of wild type eyes non- ity within the presumptive cornea tial mechanism for negatively regulat- permissive to blood vessel growth. Fi- (Fig. 5). More peripherally, additional ing canonical Wnt/␤-catenin signaling nally, low levels of canonical Wnt/␤- phenotypes of Dkk2-deficient mice, in- activity that is required for normal catenin signaling activity levels may cluding distichiasis and gross defects patterning and differentiation within be required for expression of known in eyelid development, provide com- the OSE. inhibitors of corneal vascularization pelling functional evidence that DKK2 Ablation of Dkk2 also results in el- such as a soluble form of VEGF recep- must also act to moderate canonical evated canonical Wnt/␤-catenin sig- tor 1 (Ambati et al., 2006). Wnt/␤-catenin signaling activity within naling activity within the POM (Gage these tissues. Therefore, Pitx2 expres- et al., 2008). Similar to the OSE, the sion within the neural crest acts as a CURRENT MODEL greatest increase occurs in cells critical node integrating RA signaling within the presumptive bulbar con- Retinoic acid produced from the optic from the optic cup, required for morpho- junctiva, limbus, and cornea. The cup and lens serves as a critical mech- genesis of the anterior segment, with mesenchyme of the developing con- anism for orchestrating overall mor- activation and suppression of canonical junctiva and limbus of wild type mice phogenesis and development of spe- Wnt/␤-catenin signaling activity, re- is richly vascularized whereas the cor- cific structures within the anterior quired to correctly pattern the OSE and nea is avascular. In contrast, the cor- segment in the period following for- mesenchyme. neal stroma in Dkk2 mice is richly mation of the lens vesicle (Fig. 5). The This model prompts further hypoth- vascularized prior to birth, similar to adjacent neural crest within the nas- eses that should be tested in order to mesenchyme underlying the conjunc- cent anterior segment is a primary di- provide additional supportive evi- tiva (Gage et al., 2008). The vascular- rect target of RA signaling at this dence. For example, the model pre- ization of the mesenchyme, together time. The neural crest responds by ini- dicts that Dkk2 expression should be with the appearance of goblet cells in tiating a transcriptional program that lost and canonical Wnt/␤-catenin sig- the overlying corneal surface ecto- is required for subsequent patterning naling activity should be upregulated derm, provides further evidence that and development of the cornea, eye- in mutants where RA production by anterior segment patterning is altered lids, and other tissues. Therefore, ocular structures or the competence of 2158 GAGE AND ZACHARIAS neural crest to respond to RA signal- also results in anterior segment dys- 2000). FOXC1 and FOXC2 are highly ing is lost. These predictions are genesis within structures or tissues related forkhead transcription factors readily testable by assaying the RA derived from POM, including iridocor- and mutations affecting FOXC2 cause mutants for Dkk2 and Axin2 expres- neal adhesions, small or absent distichiasis in mice and humans and sion. Schlemm’s canal, and hypomorphic or anterior segment dysgenesis in mice absent trabecular meshwork (Chang (Mears et al., 1998; Kidson et al., et al., 2001). The downstream effec- 1999; Fang et al., 2000; Lehmann et PERSPECTIVES tors of both TGF-␤ and BMP signaling al., 2000; Kriederman et al., 2003). Demonstrating the interconnected are Smad proteins that translocate Mutations to the homeodomain tran- roles of retinoic acid, PITX2, and ca- from the cytoplasm to the nucleus and scription factor LMX1B cause anterior nonical Wnt/␤-catenin signaling rep- activate transcription of target genes segment defects in mice and are a sig- resents a significant breakthrough in after phosphorylation by ligand-acti- nificant risk for glaucoma in humans understanding anterior segment de- vated cell surface receptors (Wu and (Chen et al., 1998; McIntosh et al., velopment, because, for the first time, Hill, 2009). Importantly, RAR pro- 1998; Vollrath et al., 1998; Pressman it provides evidence for how multiple teins are able to physically interact et al., 2000). Future studies are signaling pathways are integrated at with and modify the activity of Smad needed to determine whether Foxc2 the level of transcriptional regulation. proteins through various mechanisms and Lmx1b expression in neural crest However, it is clear that the complete in other systems (Roberts and Sporn, is directly or indirectly dependent on network regulating anterior segment 1992; La et al., 2003; Pendaries et al., RA signaling. RA signaling may regu- development is significantly more 2003). Therefore, it will be important late expression of additional tran- complex. Some candidates that may to determine not only whether RA and scription factor genes in the neural participate at various levels of the net- TGF-␤ super-family signaling act in crest and it will be important to iden- work are already known, while others concert or independently within neu- tify these, as well as to determine very likely remain to be identified. ral crest of the anterior segment but whether they are direct or indirect The future pursuit of these factors and also the underlying mechanism(s). targets (Fig. 5, Q1). their integration into the existing Additional signaling molecules likely The repertoire of responses by neu- model is essential for advancing our act upon the neural crest and it will be ral crest to RA signaling is likely to understanding of this important pro- important to identify these molecules include genes encoding proteins with cess. and determine their potential rela- other functions, in addition to tran- Several potential candidates are tionship to RA signaling (Fig. 5, Q1, scription factors (Fig. 5, Q4). These members of the TGF-␤ super-family of Q2). additional genes will likely include signaling molecules, a third major es- Our understanding of the response both direct targets of RA signaling, as sential cell–cell signaling pathway by the neural crest is also in its in- well as targets of the transcription during development. This family is di- fancy. Pitx2 and Foxc1 expression are factors whose expression is activated vided into two branches: the TGF-␤/ dependent on RA signaling but in response to RA signaling. The visi- activin/nodal branch and the bone whether these effects are direct or in- ble changes in organization and com- morphogenetic protein (BMP)/growth- direct is unknown (Fig. 5). Although paction that occur within the mesen- derived factor (GDF) branch (Wu and Dkk2 is clearly an important target of chyme during its response to RA Hill, 2009). Ablation of TGF-␤ recep- Pitx2, the ocular phenotype in Dkk2 signaling suggest that one key target tor2(Tgfbr2) within the neural crest mutant mice is significantly less se- set will include genes for cell matrix results in agenesis of the corneal en- vere than the Pitx2 phenotype. There- and/or cell adhesion molecules. Genes dothelium with associated loss of the fore, Dkk2 likely represents one of encoding additional factors involved anterior chamber, pronounced thin- what are likely to be many important in different aspects of cell signaling in ning of the corneal stroma, and a se- downstream effectors of PITX2 during addition to Dkk2 may represent a sec- vere reduction or loss of Foxc1 and anterior segment development and ond key target set. These may encode Pitx2 expression, features that are identifying these additional factors re- factors that act like DKK2 on the over- reminiscent of the RA signaling mu- mains a high priority. Like Pitx2, lying surface or factors involved in sig- tants (Ittner et al., 2005). However, Foxc1 is certain to be a key target of naling back to the optic cup or lens, as cell death is significantly increased RA signaling that should be pursued well as autocrine factors. They may within the corneal mesenchyme of Tg- further because heterozygous and ho- also include expression of intrinsic fbr2 mice, a notable difference from mozygous null mutations cause ante- factors such as cell surface receptors RA signaling mutants (Ittner et al., rior segment dysgenesis in mice and or cytoplasmic regulators that change 2005). TGF-␤2 appears to be the ma- heterozygous mutations in FOXC1 are the competence of the neural crest to jor signaling ligand acting upon the a second cause of Axenfeld-Rieger respond to existing or new signaling neural crest as Tgfbr2 knockout mice Syndrome in humans (Mears et al., inputs. phenocopy the Tgfbr2 knockout mice 1998; Kidson et al., 1999; Lehmann et The expression pattern of the while anterior segment development al., 2000). Additional transcription RARE-lacZ reporter transgene identi- appears normal in the absence of factor genes to test as potential tar- fies structures derived from neural ec- TGF-␤1 and TGF-␤3 (Fig. 5) (Sanford gets of RA signaling in neural crest toderm (anterior optic cup, presump- et al., 1997; Saika et al., 2001). Reduc- include Foxc2 and Lmxb1 (Fig. 5, Q3) tive ciliary, and iris epithelium, RPE) tion of BMP4 levels in Bmp4ϩ/Ϫ mice (Pressman et al., 2000; Smith et al., and OSE (presumptive corneal, lim- SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2159 bal, conjunctival, and eyelid epithe- appears to play an essential role in for development of the branchial lia), in addition to POM, as competent relatively broad patterning events arches, brain, pituitary gland, and to respond to RA signaling in the pe- within the anterior segment, subse- heart, all processes that are highly de- riod following lens vesicle formation quent regulatory steps are likely to pendent on inductive patterning (Matt et al., 2005; Molotkov et al., function in more specific developmen- events, including by RA and canonical 2006). Although neural crest mesen- tal processes. As one example, canon- Wnt/␤-catenin signaling. It will be im- chyme is clearly a critical primary tar- ical Wnt/␤-catenin signaling levels are portant to determine whether the in- get of RA signaling, these expression known to regulate the degree of teractions between PITX2 and these data raise the question whether nor- branching during lacrimal gland de- signaling pathways are replicated mal anterior segment development velopment (Dean et al., 2005). FGF10 during the development or function of may also require a primary response signaling is required for induction of these additional organs. to RA signaling within the neural ec- lacrimal gland development from pe- toderm and/or OSE. Appropriate Cre ripheral OSE where canonical Wnt/␤- drivers for testing these possibilities catenin signaling activity levels are are available and should be used to apparently relatively high, suggesting ACKNOWLEDGMENTS test these possibilities. that perhaps FGF signaling together We thank Peter Hitchcock for helpful This signaling network has been de- with a particular level of canonical discussions and critical review of the fined in mouse but its relevance to Wnt/␤-catenin signaling activity may manuscript. This work was supported other organisms remains to be deter- be required to specify the lacrimal by NEI/NIH (EY014126, EY007003) mined. If essential in other experi- gland fate within the OSE (Makaren- to P.J.G. mental organisms such as fish or kova et al., 2000). However, trans- chickens, this will provide important genic expression of FGF10 from the additional avenues to further explore lens is able to induce ectopic lacrimal REFERENCES the pathway as these organisms offer glands to form within the presumptive certain experimental advantages that cornea (Govindarajan et al., 2000), Ambati BK, Nozaki M, Singh N, Takeda A, mice do not. Variation in PITX2 gene raising possibility that FGF signaling Jani PD, Suthar T, Albuquerque RJ, dose is a central underlying mecha- may be dominant. Richter E, Sakurai E, Newcomb MT, ␤ Kleinman ME, Caldwell RB, Lin Q, nism leading to anterior segment de- RA signaling, canonical Wnt/ -cate- Ogura Y, Orecchia A, Samuelson DA, fects in humans. An auto feedback nin signaling, and PITX2 are all im- Agnew DW, St Leger J, Green WR, Ma- loop between PITX2 and canonical portant for normal development of hasreshti PJ, Curiel DT, Kwan D, Marsh Wnt/␤-catenin signaling, if active in multiple organs. Therefore, aspects of H, Ikeda S, Leiper LJ, Collinson JM, humans, would provide an effective the model presented in Figure 5 may Bogdanovich S, Khurana TS, Shibuya M, Baldwin ME, Ferrara N, Gerber HP, De means of potentially permitting tis- not be limited to the anterior segment. Falco S, Witta J, Baffi JZ, Raisler BJ, sues within the anterior segment RA and canonical Wnt/␤-catenin sig- Ambati J. 2006. Corneal avascularity is some capability to respond to physio- naling interact elsewhere during de- due to soluble VEGF receptor-1. Nature logical change, such as environmental velopment and in disease. Induction of 443:993–997. Ang SJ, Stump RJ, Lovicu FJ, McAvoy JW. stressors, and return the system to a neuronal differentiation program in 2004. Spatial and temporal expression of homeostasis. However, the ability of mouse embryonic stem cells is medi- Wnt and Dickkopf genes during murine the system to respond in this fashion, ated by activation of Dkk1 down- lens development. Gene Expr Patterns if it exists at all, appears limited since stream of RA signaling and the ensu- 4:289–295. it cannot effectively compensate in pa- ing suppression of canonical Wnt/␤- Ashery-Padan R, Marquardt T, Zhou X, Gruss P. 2000. Pax6 activity in the lens tients who have even one hypomor- catenin signaling, a mechanism with primordium is required for lens forma- phic allele of PITX2. clear parallels to events in the ante- tion and for correct placement of a single This review has focused largely on rior segment (Verani et al., 2007). RA retina in the eye. Genes Dev 14:2701– the interconnected functions of RA and canonical Wnt/␤-catenin signal- 2711. ␤ Beebe DC, Coats JM. 2000. The lens orga- and canonical Wnt/ -catenin signal- ing function in vertebral patterning nizes the anterior segment: specification ing, together with the homeodomain along the anterior-posterior axis of the of neural crest cell differentiation in the transcription factor PITX2, in deter- embryo by synergistically activating avian eye. Dev Biol 220:424–431. mining overall patterning of the ante- expression of Cdx1 (Prinos et al., Blixt A, Mahlapuu M, Aitola M, Pelto- ␤ Huikko M, Enerback S, Carlsson P. rior segment in the period following 2001). RA and canonical Wnt/ -cate- 2000. A forkhead gene, FoxE3, is essen- formation of the lens vesicle. Although nin signaling also interact in breast tial for lens epithelial proliferation and these functions are essential at this and colon cancer (Szeto et al., 2001; closure of the lens vesicle. Genes Dev particular developmental stage as a Tice et al., 2002a,b). In general, ge- 14:245–254. prelude for events that occur later, it netic evidence for these interactions is Brault V, Moore R, Kutsch S, Ishibashi M, Rowitch DH, McMahon AP, Sommer L, is clear that execution of the later strong but the specific underlying Boussadia O, Kemler R. 2001. Inactiva- events requires additional regulatory mechanisms are unknown. Therefore, tion of the beta-catenin gene by Wnt1- steps and it will be important to iden- it may be that PITX2 or another tran- Cre-mediated deletion results in dra- tify these and determine how they re- scription factor functions as an inte- matic brain malformation and failure of craniofacial development. Development late to and/or build upon the earlier gration node in these systems, as oc- 128:1253–1264. steps that have been the focus of this curs in the anterior segment. In Brownell I, Dirksen M, Jamrich M. 2000. review. Unlike the early phase, which addition to the eye, Pitx2 is essential Forkhead Foxe3 maps to the dysgenetic 2160 GAGE AND ZACHARIAS

lens locus and is critical in lens develop- factor, are responsible for the hereditary Hjalt TA, Semina EV, Amendt BA, Murray ment and differentiation. Genesis 27:81– lymphedema-distichiasis syndrome. Am J JC. 2000. The Pitx2 protein in mouse 93. Hum Genet 67:1382–1388. development. Dev Dyn 218:195–200. Chambon P. 1996. A decade of molecular Findlater GS, McDougall RD, Kaufman Ittner LM, Wurdak H, Schwerdtfeger K, biology of retinoic acid receptors. Faseb J MH. 1993. Eyelid development, fusion Kunz T, Ille F, Leveen P, Hjalt TA, Suter 10:940–954. and subsequent reopening in the mouse. U, Karlsson S, Hafezi F, Born W, Som- Chang B, Smith RS, Peters M, Savinova J Anat 183:121–129. mer L. 2005. Compound developmental OV, Hawes NL, Zabaleta A, Nusinowitz Fuhrmann S, Levine EM, Reh TA. 2000. eye disorders following inactivation of S, Martin JE, Davisson ML, Cepko CL, Extraocular mesenchyme patterns the TGFbeta signaling in neural-crest stem Hogan BL, John SW. 2001. Haploinsuf- optic vesicle during early eye develop- cells. J Biol 4:11. ficient Bmp4 ocular phenotypes include ment in the embryonic chick. Develop- Jacobson AG, Sater AK. 1988. Features of anterior segment dysgenesis with ele- ment 127:4599–4609. embryonic induction. Development 104: vated intraocular pressure. BMC Genet Fujii H, Sato T, Kaneko S, Gotoh O, Fujii- 341–359. 2:18. Kuriyama Y, Osawa K, Kato S, Hamada Kastner P, Grondona JM, Mark M, Gans- Chen H, Lun Y, Ovchinnikov D, Kokubo H, H. 1997. Metabolic inactivation of reti- muller A, LeMeur M, Decimo D, Vonesch Oberg KC, Pepicelli CV, Gan L, Lee B, noic acid by a novel P450 differentially JL, Dolle P, Chambon P. 1994. Genetic Johnson RL. 1998. Limb and kidney de- expressed in developing mouse embryos. analysis of RXR alpha developmental fects in Lmx1b mutant mice suggest an Embo J 16:4163–4173. function: convergence of RXR and RAR involvement of LMX1B in human nail Furukawa T, Morrow EM, Li T, Davis FC, signaling pathways in heart and eye patella syndrome. Nat Genet 19:51–55. Cepko CL. 1999. Retinopathy and atten- morphogenesis. Cell 78:987–1003. Chen Y, Stump RJ, Lovicu FJ, McAvoy JW. uated circadian entrainment in Crx-defi- Kastner P, Mark M, Ghyselinck N, Krezel 2004. Expression of Frizzleds and se- cient mice. Nat Genet 23:466–470. W, Dupe V, Grondona JM, Chambon P. creted frizzled-related proteins (Sfrps) Gage PJ, Camper SA. 1997. Pituitary ho- 1997. Genetic evidence that the retinoid during mammalian lens development. meobox 2, a novel member of the bicoid- signal is transduced by heterodimeric Int J Dev Biol 48:867–877. related family of homeobox genes, is a RXR/RAR functional units during mouse Chow RL, Lang RA. 2001. Early eye devel- potential regulator of anterior structure development. Development 124:313–326. opment in vertebrates. Annu Rev Cell formation. Hum Mol Genet 6:457–464. Kaufman MH. 1992. The atlas of mouse Dev Biol 17:255–296. Gage PJ, Suh H, Camper SA. 1999. Dosage development. San Diego: Academy Cvekl A, Tamm ER. 2004. Anterior eye de- requirement of Pitx2 for development of Press. 346–347 pp. velopment and ocular mesenchyme: new multiple organs. Development 126:4643– Kidson SH, Kume T, Deng K, Winfrey V, insights from mouse models and human 4651. Hogan BL. 1999. The forkhead/winged- diseases. Bioessays 26:374–386. Gage PJ, Rhoades W, Prucka SK, Hjalt TA. helix gene, Mf1, is necessary for the nor- Dean CH, Miller LA, Smith AN, Dufort D, 2005. Fate maps of neural crest and me- mal development of the cornea and for- Lang RA, Niswander LA. 2005. Canoni- soderm in the mammalian eye. Invest mation of the anterior chamber in the cal Wnt signaling negatively regulates Ophthalmol Vis Sci 46:4200–4208. mouse eye. Dev Biol 211:306–322. branching morphogenesis of the lung Gage PJ, Qian M, Wu D, Rosenberg KI. and lacrimal gland. Dev Biol 286:270– Kitamura K, Miura H, Miyagawa-Tomita 2008. The canonical Wnt signaling an- 286. S, Yanazawa M, Katoh-Fukui Y, Suzuki tagonist DKK2 is an essential effector of Dickman ED, Thaller C, Smith SM. 1997. R, Ohuchi H, Suehiro A, Motegi Y, Na- PITX2 function during normal eye devel- Temporally-regulated retinoic acid de- kahara Y, Kondo S, Yokoyama M. 1999. opment. Dev Biol 317:310–324. pletion produces specific neural crest, oc- Mouse Pitx2 deficiency leads to anoma- ular and nervous system defects. Devel- Ghyselinck NB, Dupe V, Dierich A, Mes- lies of the ventral body wall, heart, ex- opment 124:3111–3121. saddeq N, Garnier JM, Rochette-Egly C, tra- and periocular mesoderm and right Drager UC, Li H, Wagner E, McCaffery P. Chambon P, Mark M. 1997. Role of the pulmonary isomerism. Development 126: 2001. Retinoic acid synthesis and break- retinoic acid receptor beta (RARbeta) 5749–5758. down in the developing mouse retina. during mouse development. Int J Dev Kozlowski K, Walter MA. 2000. Variation Prog Brain Res 131:579–587. Biol 41:425–447. in residual PITX2 activity underlies the Duester G. 2000. Families of retinoid dehy- Goodwin AM, Sullivan KM, D’Amore PA. phenotypic spectrum of anterior segment drogenases regulating vitamin A func- 2006. Cultured endothelial cells display developmental disorders. Hum Mol tion: production of visual pigment and endogenous activation of the canonical Genet 9:2131–2139. retinoic acid. Eur J Biochem 267:4315– Wnt signaling pathway and express mul- Kriederman BM, Myloyde TL, Witte MH, 4324. tiple ligands, receptors, and secreted Dagenais SL, Witte CL, Rennels M, Ber- Dupe V, Matt N, Garnier JM, Chambon P, modulators of Wnt signaling. Dev Dyn nas MJ, Lynch MT, Erickson RP, Caul- Mark M, Ghyselinck NB. 2003. A new- 235:3110–3120. der MS, Miura N, Jackson D, Brooks BP, born lethal defect due to inactivation of Gould DB, Smith RS, John SW. 2004. An- Glover TW. 2003. FOXC2 haploinsuffi- retinaldehyde dehydrogenase type 3 is terior segment development relevant to cient mice are a model for human auto- prevented by maternal retinoic acid glaucoma. Int J Dev Biol 48:1015–1029. somal dominant lymphedema-distichia- treatment. Proc Natl Acad Sci USA 100: Govindarajan V, Ito M, Makarenkova HP, sis syndrome. Hum Mol Genet 12:1179– 14036–14041. Lang RA, Overbeek PA. 2000. Endoge- 1185. Duplaa C, Jaspard B, Moreau C, D’Amore nous and ectopic gland induction by Kulak SC, Kozlowski K, Semina EV, PA. 1999. Identification and cloning of a FGF-10. Dev Biol 225:188–200. Pearce WG, Walter MA. 1998. Mutation secreted protein related to the cysteine- Grun F, Hirose Y, Kawauchi S, Ogura T, in the RIEG1 gene in patients with iri- rich domain of frizzled. Evidence for a Umesono K. 2000. Aldehyde dehydroge- dogoniodysgenesis syndrome. Hum Mol role in endothelial cell growth control. nase 6, a cytosolic retinaldehyde dehy- Genet 7:1113–1117. Circ Res 84:1433–1445. drogenase prominently expressed in sen- Kume T, Jiang H, Topczewska JM, Hogan Evans AL, Gage PJ. 2005. Expression of sory neuroepithelia during development. BL. 2001. The murine winged helix tran- the homeobox gene Pitx2 in neural crest J Biol Chem 275:41210–41218. scription factors, Foxc1 and Foxc2, are is required for optic stalk and ocular an- Hale F. 1933. Pigs born without eyeballs. both required for cardiovascular develop- terior segment development. Hum Mol J Hered 24:105–127. ment and somitogenesis. Genes Dev 15: Genet 14:3347–3359. Harada N, Tamai Y, Ishikawa T, Sauer B, 2470–2482. Fang J, Dagenais SL, Erickson RP, Arlt Takaku K, Oshima M, Taketo MM. 1999. La P, Morgan TA, Sykes SM, Mao H, MF, Glynn MW, Gorski JL, Seaver LH, Intestinal polyposis in mice with a dom- Schnepp RW, Petersen CD, Hua X. 2003. Glover TW. 2000. Mutations in FOXC2 inant stable mutation of the beta-catenin Fusion proteins of retinoid receptors an- (MFH-1), a forkhead family transcription gene. EMBO J 18:5931–5942. tagonize TGF-beta-induced growth inhi- SIGNALING “CROSS-TALK” IN EYE DEVELOPMENT 2161

bition of lung epithelial cells. Oncogene function 2 (AF-2) is important for mouse of retinoid receptors in developing and 22:198–210. development. Development 125:4691– adult mouse eyes. Invest Ophthalmol Vis Lang RA. 2004. Pathways regulating lens 4707. Sci 42:1312–1318. induction in the mouse. Int J Dev Biol Matt N, Dupe V, Garnier JM, Dennefeld C, Mukhopadhyay M, Gorivodsky M, Shtrom 48:783–791. Chambon P, Mark M, Ghyselinck NB. S, Grinberg A, Niehrs C, Morasso MI, Lehmann OJ, Ebenezer ND, Jordan T, Fox 2005. Retinoic acid-dependent eye mor- Westphal H. 2006. Dkk2 plays an essen- M, Ocaka L, Payne A, Leroy BP, Clark phogenesis is orchestrated by neural tial role in the corneal fate of the ocular BJ, Hitchings RA, Povey S, Khaw PT, crest cells. Development 132:4789–4800. surface epithelium. Development 133: Bhattacharya SS. 2000. Chromosomal Matt N, Ghyselinck NB, Pellerin I, Dupe V. 2149–2154. duplication involving the forkhead tran- 2008. Impairing retinoic acid signalling Niehrs C. 2006. Function and biological scription factor gene FOXC1 causes iris in the neural crest cells is sufficient to roles of the Dickkopf family of Wnt mod- hypoplasia and glaucoma. Am J Hum alter entire eye morphogenesis. Dev Biol ulators. Oncogene 25:7469–7481. Genet 67:1129–1135. 320:140–148. Nusse R. 2005. Wnt signaling in disease Li H, Wagner E, McCaffery P, Smith D, McCaffery P, Wagner E, O’Neil J, Petko- and in development. Cell Res 15:28–32. Andreadis A, Drager UC. 2000. A reti- vich M, Drager UC. 1999. Dorsal and Ormestad M, Blixt A, Churchill A, Martin- noic acid synthesizing enzyme in ventral ventral rentinoic territories defined by sson T, Enerback S, Carlsson P. 2002. retina and telencephalon of the embry- retinoic acid synthesis, break-down and Foxe3 haploinsufficiency in mice: a onic mouse. Mech Dev 95:283–289. nuclear receptor expression. Mech Dev model for Peters’ anomaly. Invest Oph- Lin CR, Kioussi C, O’Connell S, Briata P, 85:203–214. thalmol Vis Sci 43:1350–1357. Szeto D, Liu F, Izpisua-Belmonte JC, McIntosh I, Dreyer SD, Clough MV, Dun- Pei YF, Rhodin JA. 1970. The prenatal de- Rosenfeld MG. 1999. Pitx2 regulates ston JA, Eyaid W, Roig CM, Montgomery velopment of the mouse eye. Anat Rec lung asymmetry, cardiac positioning and T, Ala-Mello S, Kaitila I, Winterpacht A, 168:105–125. pituitary and tooth morphogenesis. Na- Zabel B, Frydman M, Cole WG, Franco- Pendaries V, Verrecchia F, Michel S, Mau- ture 401:279–282. mano CA, Lee B. 1998. Mutation analy- viel A. 2003. Retinoic acid receptors in- Lin M, Zhang M, Abraham M, Smith SM, sis of LMX1B gene in nail-patella syn- terfere with the TGF-beta/Smad signal- Napoli JL. 2003. Mouse retinal dehydro- drome patients. Am J Hum Genet 63: ing pathway in a ligand-specific manner. genase 4 (RALDH4), molecular cloning, 1651–1658. Oncogene 22:8212–8220. cellular expression, and activity in 9-cis- Mears AJ, Jordan T, Mirzayans F, Dubois Peters MA, Cepko CL. 2002. The dorsal- retinoic acid biosynthesis in intact cells. S, Kume T, Parlee M, Ritch R, Koop B, ventral axis of the neural retina is di- J Biol Chem 278:9856–9861. Kuo WL, Collins C, Marshall J, Gould vided into multiple domains of restricted Liu H, Mohamed O, Dufort D, Wallace VA. DB, Pearce W, Carlsson P, Enerback S, gene expression which exhibit features 2003. Characterization of Wnt signaling Morissette J, Bhattacharya S, Hogan B, of lineage compartments. Dev Biol 251: components and activation of the Wnt Raymond V, Walter MA. 1998. Muta- 59–73. canonical pathway in the murine retina. tions of the forkhead/winged-helix gene, Pressman CL, Chen H, Johnson RL. 2000. Dev Dyn 227:323–334. FKHL7, in patients with Axenfeld- LMX1B, a LIM homeodomain class tran- Logan CY, Nusse R. 2004. The Wnt signal- Rieger anomaly. Am J Hum Genet 63: scription factor, is necessary for normal ing pathway in development and dis- 1316–1328. development of multiple tissues in the ease. Annu Rev Cell Dev Biol 20:781– Mic FA, Molotkov A, Fan X, Cuenca AE, anterior segment of the murine eye. Gen- 810. Duester G. 2000. RALDH3, a retinalde- esis 26:15–25. Lohnes D, Mark M, Mendelsohn C, Dolle P, hyde dehydrogenase that generates reti- Prinos P, Joseph S, Oh K, Meyer BI, Gruss Dierich A, Gorry P, Gansmuller A, noic acid, is expressed in the ventral ret- P, Lohnes D. 2001. Multiple pathways Chambon P. 1994. Function of the reti- ina, otic vesicle and olfactory pit during governing Cdx1 expression during mu- noic acid receptors (RARs) during devel- mouse development. Mech Dev 97:227– rine development. Dev Biol 239:257–269. opment (I). Craniofacial and skeletal ab- 230. Ring BZ, Cordes SP, Overbeek PA, Barsh normalities in RAR double mutants. Mic FA, Molotkov A, Molotkova N, Duester GS. 2000. Regulation of mouse lens fiber Development 120:2723–2748. G. 2004. Raldh2 expression in optic ves- cell development and differentiation Longo KA, Kennell JA, Ochocinska MJ, icle generates a retinoic acid signal by the Maf gene. Development 127:307– Ross SE, Wright WS, MacDougald OA. needed for invagination of retina during 317. 2002. Wnt signaling protects 3T3–L1 optic cup formation. Dev Dyn 231:270– Roberts AB, Sporn MB. 1992. Mechanistic preadipocytes from apoptosis through in- 277. interrelationships between two super- duction of insulin-like growth factors. Miller LA, Smith AN, Taketo MM, Lang families: the steroid/retinoid receptors J Biol Chem 277:38239–38244. RA. 2006. Optic cup and facial pattern- and transforming growth factor-beta. Lu MF, Pressman C, Dyer R, Johnson RL, ing defects in ocular ectoderm beta-cate- Cancer Surv 14:205–220. Martin JF. 1999. Function of Rieger syn- nin gain-of-function mice. BMC Dev Biol Saika S, Liu CY, Azhar M, Sanford LP, drome gene in left-right asymmetry and 6:14. Doetschman T, Gendron RL, Kao CW, craniofacial development. Nature 401: Molotkov A, Fan X, Deltour L, Foglio MH, Kao WW. 2001. TGFbeta2 in corneal 276–278. Martras S, Farres J, Pares X, Duester G. morphogenesis during mouse embryonic Makarenkova HP, Ito M, Govindarajan V, 2002. Stimulation of retinoic acid pro- development. Dev Biol 240:419–432. Faber SC, Sun L, McMahon G, Overbeek duction and growth by ubiquitously ex- Sakai Y, Luo T, McCaffery P, Hamada H, PA, Lang RA. 2000. FGF10 is an inducer pressed alcohol dehydrogenase Adh3. Drager UC. 2004. CYP26A1 and and Pax6 a competence factor for lacri- Proc Natl Acad Sci USA 99:5337–5342. CYP26C1 cooperate in degrading reti- mal gland development. Development Molotkov A, Molotkova N, Duester G. noic acid within the equatorial retina 127:2563–2572. 2006. Retinoic acid guides eye morpho- during later eye development. Dev Biol Martin P, Parkhurst SM. 2004. Parallels genetic movements via paracrine signal- 276:143–157. between tissue repair and embryo mor- ing but is unnecessary for retinal dorso- Sanford LP, Ormsby I, Gittenberger-de phogenesis. Development 131:3021–3034. ventral patterning. Development 133: Groot AC, Sariola H, Friedman R, Boivin Masckauchan TN, Shawber CJ, Funahashi 1901–1910. GP, Cardell EL, Doetschman T. 1997. Y, Li CM, Kitajewski J. 2005. Wnt/beta- Monaghan AP, Kioschis P, Wu W, Zuniga TGFbeta2 knockout mice have multiple catenin signaling induces proliferation, A, Bock D, Poustka A, Delius H, Niehrs developmental defects that are non-over- survival and interleukin-8 in human en- C. 1999. Dickkopf genes are co-ordi- lapping with other TGFbeta knockout dothelial cells. Angiogenesis 8:43–51. nately expressed in mesodermal lin- phenotypes. Development 124:2659– Mascrez B, Mark M, Dierich A, Ghyselinck eages. Mech Dev 87:45–56. 2670. NB, Kastner P, Chambon P. 1998. The Mori M, Ghyselinck NB, Chambon P, Mark See AW, Clagett-Dame M. 2009. The tem- RXRalpha ligand-dependent activation M. 2001. Systematic immunolocalization poral requirement for vitamin A in the 2162 GAGE AND ZACHARIAS

developing eye: mechanism of action in 2000. Haploinsufficiency of the tran- pression of the Wnt inhibitor Dickkopf-1 optic fissure closure and new roles for scription factors FOXC1 and FOXC2 re- is required for the induction of neural the vitamin in regulating cell prolifera- sults in aberrant ocular development. markers in mouse embryonic stem cells tion and adhesion in the embryonic ret- Hum Mol Genet 9:1021–1032. differentiating in response to retinoic ina. Dev Biol 325:94–105. Spemann H. 1901. U¨ ber korrelationen in acid. J Neurochem 100:242–250. Seeliger MW, Biesalski HK, Wissinger B, die Entwickelung des Auges. [On corre- Vollrath D, Jaramillo-Babb VL, Clough Gollnick H, Gielen S, Frank J, Beck S, lations in eye development.] Verh Anat MV, McIntosh I, Scott KM, Lichter PR, Zrenner E. 1999. Phenotype in retinol Ges 15:61–79. Richards JE. 1998. Loss-of-function mu- deficiency due to a hereditary defect in Suzuki R, Shintani T, Sakuta H, Kato A, tations in the LIM-homeodomain gene, retinol binding protein synthesis. Invest Ohkawara T, Osumi N, Noda M. 2000. LMX1B, in nail-patella syndrome. Hum Ophthalmol Vis Sci 40:3–11. Identification of RALDH-3, a novel reti- Mol Genet 7:1091–1098. Semina EV, Brownell I, Mintz-Hittner HA, naldehyde dehydrogenase, expressed in Wagner E, McCaffery P, Drager UC. 2000. Murray JC, Jamrich M. 2001. Mutations the ventral region of the retina. Mech Retinoic acid in the formation of the dor- in the human forkhead transcription fac- Dev 98:37–50. soventral retina and its central projec- tor FOXE3 associated with anterior seg- Szeto W, Jiang W, Tice DA, Rubinfeld B, tions. Dev Biol 222:460–470. ment ocular dysgenesis and cataracts. Hollingshead PG, Fong SE, Dugger DL, Warkany J, Schraffenberger S. 1946. Con- Hum Mol Genet 10:231–236. Pham T, Yansura DG, Wong TA, genital malformations induced in rats by Semina EV, Reiter R, Leysens NJ, Alward Grimaldi JC, Corpuz RT, Singh JS, maternal vitamin A deficiency. I. Defects WL, Small KW, Datson NA, Siegel-Bar- Frantz GD, Devaux B, Crowley CW, of the eye. Arch Ophthalmol 35:150–169. telt J, Bierke-Nelson D, Bitoun P, Zabel Schwall RH, Eberhard DA, Rastelli L, Wilson JG, Roth CB, Warkany J. 1953. An BU, Carey JC, Murray JC. 1996. Cloning Polakis P, Pennica D. 2001. Overexpres- analysis of the syndrome of malforma- and characterization of a novel bicoid- sion of the retinoic acid-responsive gene tions induced by maternal vitamin A de- related homeobox transcription factor Stra6 in human cancers and its synergis- ficiency. Effects of restoration of vitamin gene, RIEG, involved in Rieger syn- tic induction by Wnt-1 and retinoic acid. A at various times during gestation. drome. Nat Genet 14:392–399. Cancer Res 61:4197–4205. Am J Anat 92:189–217. Skurk C, Maatz H, Rocnik E, Bialik A, Tice DA, Soloviev I, Polakis P. 2002a. Ac- Wolosin JM, Schutte M, Zieske JD, Budak Force T, Walsh K. 2005. Glycogen-Syn- tivation of the Wnt pathway interferes MT. 2002. Changes in connexin43 in thase Kinase3beta/beta-catenin axis pro- with serum response element-driven early ocular surface development. Curr motes angiogenesis through activation of transcription of immediate early genes. Eye Res 24:430–438. vascular endothelial growth factor sig- J Biol Chem 277:6118–6123. Wright M, Aikawa M, Szeto W, Papkoff J. naling in endothelial cells. Circ Res 96: Tice DA, Szeto W, Soloviev I, Rubinfeld B, 1999. Identification of a Wnt-responsive 308–318. Fong SE, Dugger DL, Winer J, Williams signal transduction pathway in primary Smith AN, Miller LA, Song N, Taketo MM, PM, Wieand D, Smith V, Schwall RH, endothelial cells. Biochem Biophys Res Lang RA. 2005. The duality of beta-cate- Pennica D, Polakis P. 2002b. Synergistic Commun 263:384–388. nin function: A requirement in lens mor- induction of tumor antigens by Wnt-1 Wu MY, Hill CS. 2009. Tgf-beta super- phogenesis and signaling suppression of signaling and retinoic acid revealed by family signaling in embryonic develop- lens fate in periocular ectoderm. Dev gene expression profiling. J Biol Chem ment and homeostasis. Dev Cell Biol 285:477–489. 277:14329–14335. 16:329–343. Smith RS, Zabaleta A, Kume T, Savinova Verani R, Cappuccio I, Spinsanti P, Gra- Xia Y, Karin M. 2004. The control of cell OV, Kidson SH, Martin JE, Nishimura dini R, Caruso A, Magnotti MC, Motolese motility and epithelial morphogenesis by DY, Alward WL, Hogan BL, John SW. M, Nicoletti F, Melchiorri D. 2007. Ex- Jun kinases. Trends Cell Biol 14:94–101.