An essential role for Rax in retina and neuroendocrine system development


Author to whom all correspondence should be addressed.



In vertebrates, the central nervous system (CNS) develops as a highly hierarchical, patterned organ with a vast diversity of neuronal and glial cell types. The vertebrate retina is developmentally a part of the CNS. Establishment of the vertebrate retina requires a series of developmental steps including specification of the anterior neural plate, evagination of the optic vesicles from the ventral forebrain, and differentiation of cells. The transcription factor RAX is a paired-type homeoprotein that plays a critical role in the eye and forebrain development of vertebrate species. Rax is initially expressed in the anterior neural region of developing mouse embryos, and later in the retina, pituitary gland, hypothalamus, and pineal gland. The targeted deletion of Rax in the mouse results in no eye formation and abnormal forebrain formation. In humans, mutations in the RAX gene lead to anophthalmia and microphthalmia. These observations indicate that RAX plays a pivotal role in the establishment of the retina. In addition, recent studies have reported that retina and pituitary gland tissues can be induced in a culture system from embryonic stem cells, using RAX expression as an indicator of neuronal progenitor cells in the induced tissue, and suggesting that the Rax gene is a key factor in neuronal regeneration. This review highlights the biological functions and molecular mechanisms of RAX in retina, pituitary, hypothalamus, and pineal gland development.

Identification and expression of the Rax homeobox gene

Homeobox genes, which encode highly conserved transcription factors, play essential roles in the development of various organs, including the central nervous system (CNS). Retina and anterior neural fold homeobox gene (Rax) was first identified as a homeobox gene expressed in the developing mouse retina (Furukawa et al. 1997a; Mathers et al. 1997). After this, the Rax gene was identified in many species, including Xenopus (Casarosa et al. 1997), Drosophila (Eggert et al. 1998), chicken (Ohuchi et al. 1999), zebrafish (Mathers et al. 1997; Chuang et al. 1999), and medaka (Deschet et al. 1999; Loosli et al. 2001). While Rax is also known as Rx (Mathers et al. 1997), in mammals “Rax” is the official gene nomenclature because mouse Rax cDNA was isolated and the Rax name was deposited in the GenBank for mammals before Rx came into use. The Rax gene is located on chromosome 18 in humans, mice, and rats.

Rax is a paired-type homeobox gene. Its coding protein contains two characteristic domains of homeodomain proteins: an octapeptide motif in the N-terminus and an OAR (otp, aristaless, and rax) domain in the C-terminus. The octapeptide motif exists in other homeodomain proteins such as PAX6 and CHX10. The OAR domain is a 15-amino-acid sequence that is conserved among otp, aristaless, and Rax genes (Furukawa et al. 1997a; Mathers et al. 1997). The Rax gene is evolutionarily well conserved from Drosophila melanogaster to humans (Fig. 1). Drosophila Rax is expressed in the fly brain, indicating that Rax expression in the CNS is also well conserved through evolution.

Figure 1.

Evolutionary conservation of RAX protein. Schematic presentation of putative Rax gene domains from mouse to Drosophila. RAX has octapeptide, homeodomain, and OAR (otp, aristaless, and rax) domains. The Drosophila RAX protein is more than twice as large as the vertebrate proteins.

In developing mouse embryos, the neuroepithelium of the retina and hypothalamus is derived from a region near the ventral diencephalon. At the initiation of neurogenesis, retinal and hypothalamic progenitor cells co-express some homeodomain transcription factors, including SIX3 and RAX, that are specific to the ventral diencephalic neuroepithelium, along with another ventral diencephalon-expressing gene such as the WNT receptor FRIZZLED5 (Ikeda et al. 2005; Byerly & Blackshaw 2009). At embryonic day 7.5 (E7.5) in mice, Rax expression is observed in the cephalic neural fold (head fold), which later develops into the forebrain and midbrain (Furukawa et al. 1997a). At E9.5, Rax mRNA is restricted to the optic vesicles, the optic stalk, and the ventral diencephalon. From E10.5 to E11.5, Rax expression is observed in the whole retinal region (Fig. 2). After birth, Rax expression is limited to a few regions in the retina, including the progenitor layer, the photoreceptor precursor layer, and the inner nuclear layer. On the other hand, during embryonic stages, Rax is also expressed in the hypothalamus, pituitary gland, and pineal gland in rodents (Asbreuk et al. 2002; Bailey et al. 2009; Medina-Martinez et al. 2009; Shimogori et al. 2010; Rohde et al. 2011). Between E10.5 and E12.5, Rax expression initially begins just in the hypothalamus and posterior pituitary of the ventral forebrain. In adult mice, Rax is expressed in the pineal gland as well (Asbreuk et al. 2002; Bailey et al. 2009; Rohde et al. 2011). From these reports, it is possible to follow mouse Rax expression from the developing diencephalon at embryonic stages to the retina and ventral hypothalamus after birth.

Figure 2.

Rax expression in the developing mouse retina. At embryonic day 14.5 (E14.5), Rax expression is restricted to the developing retina. At E16.5, Rax expression level reaches its highest level in retinal precursor cells, and then gradually decreases. The reduced Rax signal is observed at postnatal day 1 (P1) and P14 when retinal cell differentiation occurs. GCL, ganglion cell layer; INL, inner nuclear layer; NBL, neuroblastic layer (progenitor layer); ONL, outer nuclear layer (photoreceptor layer); RPE, retinal pigment epithelium. Scale bar, 100 μm.

Rax in retinal progenitor proliferation

RAX is a key transcription factor for the formation of eyes in vertebrates. Medaka harboring a mutation in the Rx3 locus does not develop optic vesicles, and loss of function mutations of Rx3 in zebrafish cause loss of the eyes (Loosli et al. 2001; Zilinski et al. 2002). Rax-null mutant mice show reduced brain size and absence of the optic vesicle (Fig. 3) (Mathers et al. 1997). Rax-null embryos do not develop any lens structure, despite the fact that Rax is not expressed in the lens (Swindell et al. 2008). In humans, mutations of the RAX gene are associated with anophthalmia and microphthalmia (Bailey et al. 2004). On the other hand, overexpression of Rax in Xenopus and zebrafish embryos leads to overproliferation of retinal cells (Terada et al. 2006). These reports indicate that Rax is essential for the development of eyes.

Figure 3.

Abnormal development of eyes in Rax conditional knockout (CKO) mice. Rax CKO mice were generated by mating Rax-flox mice with Dkk3-Cre mice (Sato et al. 2007) or Chx10-Cre mice (Muranishi et al. 2011). The Dkk3-Cre mouse predominantly expresses CRE in retinal progenitor cells. Developing eyes were observed at E15.5 in wild type (WT) mouse, while Rax CKO mice exhibited abnormally small eyes (Rax CKO by Chx10-Cre). At P30, WT mice have fully functional and mature eyes. Eyes are absent in Rax CKO mice at P30 (Rax CKO by Dkk3-Cre).

In the developing CNS of Xenopus laevis, Terada et al. (2006) identified Xhmgb3 (Xenopus high mobility group b3) as a downstream mediator of RAX. Forced Rax expression in the Xenopus animal cap induces Xhmgb3 expression. In contrast, knockdown of RAX by an antisense morpholino-oligo against Rax (Rax MO) reduces eye size in Xenopus embryos.

XOPTX2 (SIX6), a six-family homeodomain transcription factor, promotes cell proliferation in the eye field and developing retina in Xenopus (Zuber et al. 2004). Although the expression pattern of XOptx2 mRNA resembles that of Rax mRNA, XOptx2 expression begins later than Rax (Zuber et al. 2004, 1999). Moreover, the expression of XOptx2 is significantly reduced by using a dominant inhibitory mutant of Rax (Rax-EnR) (Andreazzoli et al. 1999, 2003), in which the engrailed repressor domain is fused to Rax. Finally, the blockade of Rax function by Rax MO resulted in small eye formation. Inhibition of Rax function reduces eye and brain sizes due mainly to reduced cell proliferation (Andreazzoli et al. 2003; Casarosa et al. 2003). Small eye and brain size from the blockade of RAX function cannot be rescued by Xhmgb3 overexpression only; however, it can be rescued by the simultaneous overexpression of both Xhmgb3 and XOptx2. Co-injection of Xhmgb3 and XOpt2 mRNAs showed a synergic effect, rescuing the knockdown phenotype of Rax MO. These results indicate that Xhmgb3 and XOptx2 are downstream molecules of RAX and synergistically regulate cell proliferation in the eye and brain (Fig. 4).

Figure 4.

A role of Rax function in Xenopus retinal development. A schema of the molecular mechanism underlying retinal development in the Xenopus embryo. Xenopus Rax (Xrax) orchestrates the expression of several key transcription factors, leading to normal retinal development, including XHMGB3, XOPTX2, XHAIRY2 and ZIC2. XHMGB3 mainly functions as a critical regulator of retinal development in the late embryonic stages, activating the expression of c-myc,N-myc, and XOptx2. XOptx2 plays an essential role in retinal development throughout the whole embryonic period. XOptx2 strongly promotes the expansion of the eye field, leading to optic vesicle formation in the early embryonic stages. Furthermore, XOptx2 regulates cell proliferation in the developing retina in the late embryonic stages. Xhairy2 and Zic2 act as inhibitors of neuronal differentiation and thus promote cell proliferation in the anterior neural plate during the early embryonic stages (Terada et al. 2006; Terada and Furukawa 2010a).

It should be noted that the XHMGB3 protein physically and functionally associates with SUMO E2 ligase UBC9 (Terada & Furukawa 2010). UBC9 did not sumolylate the XHMGB3 protein in cultured cells. Instead, UBC9 bound with XHMGB3 sumoylates and inactivates the SP1 transcription factor, which induces the expression of p27Xic1, a cell cycle-dependent kinase inhibitor, leading to the promotion of retinal progenitor cells (RPCs) and retinal stem cell proliferation.

Rax in optic vesicle development

Overexpression of eye field transcription factors, including Rax, Pax6, Otx2, XOptx2, Six3, Tll and ET, results in ectopic eye formation in Xenopus with a high frequency (Zuber et al. 1999). XOptx2 promotes cell proliferation specific to the eye field and developing retina (Zuber et al. 2004). Rax-null mice do not develop an optic vesicle, while homozygous Pax6 mutant mice initially develop an optic vesicle (Hill et al. 1991; Mathers et al. 1997). It should be noted that it is widely misunderstood that Pax6-null mice do not develop optic vesicles. Pax6-null mice initially form optic vesicles, but the vesicles do not significantly grow. They then degenerate, and eventually disappear. In the Xenopus animal cap assay system, Rax induces the transcription of Pax6 and XOptx2 (Zuber et al. 1999). Rax overexpression in Xenopus embryos induces an optic vesicle-derived tissue, RPE (Mathers et al. 1997). Furthermore, Otx2 and Sox2 interact directly with each other and synergistically activate Rax expression (Danno et al. 2008). Thus, previous studies revealed that eye field transcription factors play their roles in a complex context. These factors transform the neural plate into the eye field and promote cell proliferation to expand the eye field, leading to optic vesicle formation and eye development, either by activating or repressing, to modulate the activity of cells. However, the exact molecular mechanisms of RAX underlying optic vesicle formation are still poorly understood.

Rax in retinal cell fate determination

In the CNS, mechanisms precisely regulating cell proliferation and fate determination of neuronal progenitors are essential for the formation of layered structures, such as the retina and cortex. In these structures, the generation of definite cell types in distinct layers is also influenced by the time at which precursor cells exit the cell cycle (Livesey & Cepko 2001; Ohnuma et al. 2001). RPCs are multipotent and are therefore not limited to the generation of only one daughter cell type (Turner & Cepko 1987; Holt et al. 1988; Wetts & Fraser 2008; Turner et al. 1990). In the developing brain, many progenitor cells are able to give rise to several different cell types, including different types of neurons and glia, as in the developing retina (Walsh & Cepko 1988, 1993). During vertebrate retinal development, multipotent RPCs generate retinal ganglion cells, horizontal cells, rod and cone photoreceptor cells, amacrine cells, bipolar cells, and Müller glial cells in a conserved order during development. Rax is expressed in RPCs and its expression decreases as RPCs differentiate (Furukawa et al. 1997a). Xenopus neural retinal cells overexpressing Rax can develop into any retinal cell type, suggesting that RAX regulates cell proliferation by maintaining progenitor cells in a pluripotent state without affecting cell fate determination (Casarosa et al. 2003). Injection of a Rax-GFP virus into the retinas of rat pups at postnatal day 0 (P0) demonstrated that the GFP-labeled cells expressed Müller glial markers (Furukawa et al. 2000). These results suggest that RAX has the ability to drive RPC fate to Müller glial cells in the postnatal rat retina.

Rax in photoreceptor cell fate determination

Previously, OTX2 homeoprotein was reported to be a master control molecule of photoreceptor cell fate (Nishida et al. 2003). Otx2 conditional knockout (CKO) mice exhibit a total loss of retinal photoreceptors and a pronounced increase of amacrine-like cells. Forced expression of Otx2 in retinal progenitor cells using a retrovirus steers progenitor cell fate to photoreceptors in vivo. During retinal development, cone photoreceptors are born mainly in embryonic stages, while rod photoreceptors are generated mainly at postnatal stages. Otx2 transcripts are undetectable in retinal progenitor cells. At around E11.5, when cone photoreceptor precursors begin to differentiate, Otx2 transcription is activated in cone photoreceptor precursors. Since Otx2 is essential for cone photoreceptor cell fate determination, Otx2 transactivation is closely associated with photoreceptor cell fate determination. Thus, understanding the transcriptional regulatory mechanisms of Otx2 in photoreceptor precursors leads to elucidation of a molecular mechanism underlying photoreceptor cell fate determination.

We identified an approximately 500 bp cis-regulatory region, which is responsible for Otx2 expression in photoreceptor precursors with a LacZ reporter assay using transgenic mice in vivo. We named this enhancer region embryonic enhancer locus for photoreceptor Otx2 transcription (“EELPOT”). RAX binding to the EELPOT enhancer in vivo was demonstrated by ChIP assay using embryonic mouse retina extract, and it significantly transactivates an EELPOT-luciferase reporter in cultured cells. In addition, the expression of Otx2 remarkably decreased in the retina of photoreceptor precursor-specific Rax CKO mice (Muranishi et al. 2011), indicating that RAX directly regulates Otx2 transcription in the embryonic mouse retina.

The EELPOT enhancer contains several transcription factor-binding consensus sites, including N-boxes (HES family protein-binding sequence) and paired-type homeodomain-binding sites. RAX and HES family proteins have been implicated in the maintenance of RPCs (Levine & Green 2004; Ohsawa & Kageyama 2008). We showed that HES family transcription factors, HES1, HES5 and HEY1, downstream of the canonical Notch pathway, suppress RAX-induced EELPOT activation in a dose-dependent manner (Muranishi et al. 2011). In the last RPC cell cycle, the NOTCH-HES signal pathway becomes inactive, and RAX becomes active to cause EELPOT to activate Otx2 transcription (Fig. 5). After cell cycle exit, Rax expression decreases, but remains, and continuously transactivates Otx2 in postmitotic precursors. Thus, RAX is a key molecule for the cell fate determination of embryonic photoreceptors.

Figure 5.

An essential role for RAX in cell fate determination of mouse photoreceptor cells. A schema of the molecular mechanism underlying retinal development in the mouse embryo. RAX activates Otx2 expression through direct binding to the Otx2 enhancer “EELPOT” though NOTCH-HES signaling strongly suppresses the Rax function before the last cell cycle. RAX steers an undifferentiated progenitor cell to the photoreceptor cell fate by transactivating Otx2, which is a master control gene of photoreceptor cell fate. OTX (orthodenticle homeobox)2 activates its own and downstream genes' expressions including Crx, which is required for the maturation and maintenance of photoreceptors (Furukawa et al. 1997b, 1999). CRX (cone-rod homeobox) also transactivates its own transcription by an auto-positive feedback mechanism (Furukawa et al. 2002). On the other hand, Rax promotes the differentiation of Müller glial cells cooperatively by activating NOTCH-HES signaling. Red box indicates retinal progenitor cell proliferation.

Rax in in vitro induction of photoreceptor cells and the pituitary gland

Regenerative medical research is pursuing in vitro induction of the retina and pituitary gland from embryonic stem (ES) cells/iPS (induced pluripotent stem cells) cells. Previously, only infrequent expression of photoreceptor markers in ES cell-derived neural tissues had been reported (Zhao et al. 1988; Hirano et al. 2003). Recently an efficient method of in vitro generation of neural retinal precursors from mouse ES cells by combining the Serum-free Floating culture of Embryoid Bodies-like aggregates (SFEB) culture and extracellular inductive signals Dkk1, LeftyA, serum and activin (DLFA) was reported (Ikeda et al. 2005). Consistent with the properties of in vivo retinal precursors, the induced RAX-positive cells co-express Pax6 and the mitotic maker Ki-67 in vitro but not Nestin. In order to effectively collect RAX-positive retinal progenitor cells, a mouse reporter ES cell line in which green fluorescent protein (GFP) is knocked in at the Rax locus (Rax-KI) was generated (Osakada et al. 2008, 2009). Rax-KI ES cells were cultured under SFEB/DLFA conditions, and dissociated into single cells, which were then sorted by fluorescence-activated cell sorting. These culture conditions can induce the generation of putative photoreceptors from mouse, monkey and human ES cells. Similarly, the progenitor cells derived from Rax-KI ES cells can differentiate into early hypothalamic precursor neurons (Wataya et al. 2008) and generate a functional anterior pituitary gland in vitro from the ES cell-derived cells (Suga et al. 2011). In addition, by using this Rax-KI ES cell line as a retinal progenitor marker, induction of fully stratified retinal architecture from ES cells was reported (Eiraku et al. 2011; Eiraku & Sasai 2012). Therefore, Rax serves as a pivotal marker in the process of inducing retina and pituitary gland cells and tissue from ES cells in vitro.

Rax in pineal gland development

The pineal gland is a neuroendocrine organ located at the surface of the brain. In humans, an important function of the pineal gland is melatonin synthesis. Melatonin is a circadian signaling molecule in vertebrates and essential for the adequate integration of physiological functions with environmental lighting on a daily and seasonal basis (Klein 1985; Lincoln et al. 2006; Maronde & Stehle 2007; Bailey et al. 2009). The pineal gland and retina are related tissues in several ways. Functionally, both organs are circadian photoneuroendocrine systems that transfer information about environmental lighting conditions to the mammalian organism by an increase of pineal melatonin synthesis at night time (Klein et al. 2010). Evolutionarily, both neurons in the vertebrate pineal gland and photoreceptors in the retina are thought to have evolved from a common ancestral photoreceptive melatonin-producing cell (Klein 2004; Rohde et al. 2011). This idea is based on common structural features, and additionally, transcriptomes of both organs show significant similarity to each other (Bailey et al. 2009; Rohde et al. 2011). Several reports indicate that the transcription factor cascades that control the development of the pineal gland and the retina are similar (Ekstrom & Meissl 2003; Maronde & Stehle 2007). Microarray analysis between the pineal gland and the retina revealed that 55 common genes were highly expressed at similar levels. Rax is expressed in the pineal gland from embryonic stages to adult (Asbreuk et al. 2002; Bailey et al. 2009; Rohde et al. 2011). Similar to melatonin, Rax expression also shows a daily rhythm in the adult rat pineal gland (Rohde et al. 2011). The Rax expression level increases twofold during the light period and decreases gradually during the dark period. In addition, the developmental expression of Rax, Pax6, Otx2 and Crx, all of which are essential for retinal development, is also observed in the developing rat pineal gland (Fig. 6). Thus, elucidation of molecular mechanisms of Rax functions in the retina may contribute to our understanding not only of retinal development but also of the development and evolution of the pineal gland.

Figure 6.

Otx2 is expressed in the pineal gland. X-gal staining of the pineal gland of a transgenic mouse, which integrates the BAC#2 transgene containing the Otx2 enhancer connected to a LacZ reporter gene into the genome. LacZ expression was detected in the BAC#2 transgenic mice (Kurokawa et al. 2004a,b) but not in the control wild type (WT) mice at P60. Rax is responsible for transactivation of Otx2 in the developing retina, implying the involvement of Rax in Otx2 expression in the pineal gland. Scale bar, 200 μm.


The development of the CNS is strictly controlled by multiple factors including extracellular secreted molecules and spatiotemporal-specific transcription factors. Absence of the RAX transcription factor in mouse development leads to a total absence of the diencephalon and retina. Furthermore, RAX notably exerts multiple functions in the late development of the retina, including retinal progenitor proliferation, photoreceptor cell fate determination, and differentiation of Müller glial cells. The molecular mechanisms of how RAX regulates the proliferation of retinal progenitor cells has become clearer; however, how RAX-positive progenitor cells are led to the cell fate of either photoreceptor cells or Müller glial cells is still largely unknown. In addition, it would be interesting to know if Rax is possibly involved in the development of the postnatal retina. It would also be interesting to understand how RAX regulates cell fate determination together with other transcription factors in the development of the pituitary gland, retina, and pineal gland, all of which are derived from the ventral diencephalon. Finally, RAX serves as an important marker, which paves the road to realizing the in vitro induction of retina and pituitary gland tissues from ES/iPS cells in regenerative medicine.