In the last decade, a number of genes related to the induction, specification and regionalization of the brain were isolated and their functional properties currently are being dissected. Among these, Otx1 and Otx2 play a pivotal role in several processes of brain morphogenesis. Findings from several groups now confirm the importance of Otx2 in the early specification of neuroectoderm destined to become fore–midbrain, the existence of an Otx gene dosage‐dependent mechanism in patterning the developing brain, and the involvement of Otx1 in corticogenesis. Some of these properties appear particularly fascinating when considered in evolutionary terms and highlight the central role of Otx genes in the establishment of the genetic program defining the complexity of a vertebrate brain. This review deals with the major aspects related to the roles played by Otx1 and Otx2 in the development and evolution of the mammalian brain.
One of the most fascinating goals of molecular embryology is to understand the molecular mechanisms controlling induction, specification and regionalization of the brain. In vertebrates, a remarkable amount of data has been collected in recent years on the role of genes which are candidates for the control of developmental programs underlying brain morphogenesis. Most of these genes are the vertebrate homologs of Drosophila genes coding for signal molecules or transcription factors (Lemaire and Kodjabachian, 1996; Tam and Behringer, 1997; Rubenstein et al., 1998). Among these, the orthodenticle group is defined by the Drosophila orthodenticle (otd) and the vertebrate Otx1, Otx2 and Crx genes which contain a bicoid‐like homeodomain (Finkelstein and Boncinelli, 1994; Chen et al., 1997; Freud et al., 1997). The Drosophila otd gene is expressed at the anterior pole of the blastoderm embryo and later predominantly in the developing rostralmost brain neuromere (Finkelstein and Perrimon, 1990; Finkelstein et al., 1990; Cohen and Jürgens, 1990, 1991; Hirth et al., 1995; Younossi‐Hartenstein et al., 1997).
Together with two additional genes, the homeobox‐containing gene empty spiracles (ems) and the zinc‐finger gene buttonhead (btd), otd defines an ordered and partially overlapping expression pattern including adjacent head segments. Mutations in each of these three genes cause the loss of anterior head segments where they are expressed, suggesting that they might act as gap genes operating along the cephalic segments.
In the mouse, two cognates for otd (Otx1 and Otx2) and two for ems (Emx1 and Emx2) were identified by virtue of the high conservation in their homeobox sequences (Simeone et al., 1992a,b, 1993). Their expression patterns along the developing brain of 9.5 days post‐coitum (d.p.c.) embryos showed a remarkable similarity to the Drosophila counterpart genes and suggested that they might be part of a general control system operating in the brain and different from that coded by the HOX complexes controlling the hindbrain and spinal cord (Holland et al., 1992; Simeone et al., 1992a; Krumlauf, 1994). Further evidence deriving from expression data also suggested roles for Otx1 in corticogenesis, for Otx2 in the early specification of rostral neuroectoderm and for both genes in sense organ development (Simeone et al., 1993; Ang et al., 1994; Frantz et al., 1994). These potential roles have been the subject of intense study and are now being elucidated by genetic analyses.
Early Otx2 requirement for rostral CNS specification
Fate and patterning of tissues depend on the activity of organizer cells emanating signals to a responding tissue which undergoes morphogenetic changes resulting in a specific differentiated fate (Spemann and Mangold, 1924; Waddington, 1932; Gurdon, 1987). The first evidence of an organizer comes from transplantation experiments in amphibians, in which the dorsal lip of an early blastopore induces a new, ectopic secondary axis when transplanted to the ventral side of a host embryo (Spemann and Mangold, 1924). Functionally‐equivalent organizing regions named Hensen's node and the node have been identified in chick and mouse embryos, respectively.
It is also known that early patterning of the central nervous system (CNS) primordium is controlled by distinct mechanisms involving vertical signals directed from axial mesendoderm to the surrounding neural plate, and planar signals acting through the neuroectodermal plane (Doniach, 1993; Ruiz i Altaba, 1993, 1994; see Figure 1). In this context, it has been shown recently that in zebrafish, a small group of ectodermal cells located in the prospective head region is required for the patterning and survival of the anterior brain (Houart et al., 1998; Ruiz i Altaba, 1998). However, a large body of evidence indicates that the anterior region of the primitive visceral endoderm in mouse as well as the leading edge of the involuting endoderm in Xenopus also play a crucial role in head organizer activity (Bouwmeester et al., 1996; Thomas and Beddington, 1996; Varlet et al., 1997; Thomas et al., 1998). An increasing amount of data (see below) supports the importance of the role of Otx2 in the specification and patterning of the anterior neural plate. These findings lead to a tentative model for Otx2 action during gastrulation (Figure 1).
Otx2 is transcribed in the cells that are believed to emit signals in early specification and patterning of the neural plate (the anterior visceral endoderm and prechordal mesendoderm) as well as in those responding to these instructing signals (the epiblast and anterior neuroectoderm) (Simeone et al., 1993; Ang et al., 1994) (Figure 1).
At the onset of gastrulation, Otx2 is required in the visceral endoderm to maintain its transcription in the epiblast and to mediate Otx2‐dependent signals directed from the visceral endoderm to the epiblast. Embryos lacking Otx2 fail to generate this signal in the visceral endoderm, and display an abnormal mesoderm organization and the absence of the rostral neuroectoderm (see below). From early to late streak stage, this signal persists in the visceral endoderm to maintain Otx2 transcription in the surrounding ectoderm. It is not clear how the posterior repression of Otx2 is mediated during mid–late streak formation, though either the gradual anterior displacement of the visceral endoderm or primitive streak progression may be involved. At the headfold stage, a positive signal from the node‐derived anterior mesendoderm is required for Otx2 transcription in the surrounding neuroectoderm and possibly contributes to the maintenance of the anterior character, while a negative signal from the posterior mesendoderm represses Otx2, presumably contributing to the positioning of its posterior border (Ang et al., 1994; Foley et al., 1997). In this context, impaired axial mesendoderm of Otx2−/− embryos are likely to be a consequence of the Otx2 requirement at earlier stages in the visceral endoderm (see below). Furthermore, the Otx2 posterior border might also result from interaction with factors (e.g. retinoic acid) and/or other genes expressed through the neuroectoderm. Gbx2 might be a good candidate, since at the headfold stage its anterior border of expression is adjacent to that of Otx2 and, in Gbx2−/− mice, a posterior expansion of the Otx2‐expressing territory is evident (Wassarman et al., 1997). Finally, from the headfold stage onwards, Otx2 might be required autonomously in the neuroectoderm to specify and maintain fore–midbrain identity (Acampora et al., 1997; Rhinn et al., 1998; A.Simeone, unpublished results).
In vivo embryological and genetic manipulation experiments have contributed to the above model. In mouse, Otx2 null embryos die early in embryogenesis, lack the rostral neuroectoderm fated to become forebrain, midbrain and rostral hindbrain, and show major abnormalities in their body plan (Acampora et al., 1995; Matsuo et al., 1995; Ang et al., 1996). Heterozygous Otx2+/− embryos in an appropriate genetic background show defects of the head such as serious brain abnormalities and craniofacial malformations, which are reminiscent of otocephalic phenotypes (Matsuo et al., 1995).
The analysis of Otx2 null embryos reveals that at the late streak stage, the rostral neuroectoderm is not identified and the primitive streak as well as the node‐derived cells of the axial mesendoderm are severely impaired. Therefore, the resulting headless phenotype has been interpreted as the consequence of the abnormal development of prechordal axial mesendoderm which lacks head organizer activity. Indeed, a similar explanation has been argued for the headless phenotype of Lim1−/− mutants (Shawlot and Behringer, 1995). However, in embryos in which Otx2 is replaced with a LacZ reporter gene, the first abnormality is already detected at the early streak stage (Acampora et al., 1995). At this stage, LacZ staining and transcription are abolished in the epiblast while they remain high in all the visceral endoderm of Otx2−/− embryos. Furthermore, goosecoid (gsc) transcripts, which normally label early node precursor cells having inducing properties (Izpisùa‐Belmonte et al., 1993), are undetectable or confined to the proximal region of Otx2−/− embryos.
Thus, since Otx2 is already transcribed from the earliest stages (unfertilized egg in Xenopus and at least the morula in mouse), these data indicate that the maintenance of Otx2 transcription in the epiblast cells requires at least one normal allele expressed in the visceral endoderm, while Otx2 transcription in the visceral endoderm is independent of the presence of a normal allele. Therefore, in contrast to the former interpretation deduced from the late gastrula phenotype, these results support the possibility that abnormal primitive streak organization and the headless phenotype might be determined very early at the pre‐early streak stages by an impairment of visceral endoderm properties. These visceral endoderm properties could correspond to an Otx2‐dependent signal(s) with the epiblast cells as the target (Figure 1). In this context, it is noteworthy that the chick hypoblast is required for the correct formation of the primitive streak (Stern, 1992) and that the chick hypoblast and the murine visceral endoderm might share similar roles in primitive streak organization. Moreover, an increasing amount of data strongly supports a role for the anterior visceral endoderm in head organizer activity: (i) removal of a patch of anterior visceral endoderm cells expressing the Rpx/Hesx1 gene prevents the subsequent expression of the gene in the rostral headfolds which become reduced and abnormally patterned (Thomas and Beddington, 1996; Dattani et al., 1998); (ii) chimeric embryos composed of wild‐type epiblast and nodal−/− visceral endoderm are found to be heavily impaired in rostral CNS development (Varlet et al., 1997); (iii) transplantation of axial mesoderm in mouse induces a secondary axis lacking the most anterior neural tissues (Beddington, 1994); (iv) in Xenopus, the expression of the secreted molecule coded by the cerberus gene is restricted to the leading edge of the involuting endoderm and represents a potent head inducer (Bouwmeester et al., 1996; Bouwmeester and Leyns, 1997); and (v) most of the genes expressed in the node or in the axial mesendoderm cells at the mid–late streak stage are also expressed in the anterior visceral endoderm. Together, these findings reinforce the idea that in mouse the organizer might be split into at least two embryonic regions operating at different stages to specify head and trunk organizer signals (Thomas and Beddington, 1996; Belo et al., 1997; Ruiz i Altaba, 1998).
The relevance of Otx2 in the anterior visceral endoderm recently has been confirmed by generating murine chimeric embryos containing Otx2−/− epiblast cells and wild‐type visceral endoderm or vice versa (Rhinn et al., 1998). In these experiments, the rescue of the early neural plate by wild‐type visceral endoderm suggests both an Otx2‐mediated role of the visceral endoderm in early neural plate specification and an Otx2 cell‐autonomous requirement in the neuroepithelium. Conversely, when chimeric embryos consist of an Otx2−/− visceral endoderm and an Otx2+/+ epiblast, none of the phenotypic features of Otx2−/− embryos are rescued (Rhinn et al., 1998). This result also supports the argument, as previously suggested in Otx2−/− mice (Acampora et al., 1995), that an impaired axial mesendoderm in Otx2−/− embryos is a consequence of an Otx2 requirement at earlier stages in the visceral endoderm. It is worth noting that the Otx2, Lim1 and murine cerberus genes are all co‐expressed in the anterior visceral endoderm, thus suggesting that they may overlap in the earliest genetic pathway involved in organizing the head (Belo et al., 1997; Tam and Behringer, 1997).
Additional data indicating that Otx2 is responsive to inductive interactions between ectoderm and mesendoderm came from explant‐recombination experiments in gastrulating mouse embryos, showing that a positive signal from the anterior mesendoderm of headfold stage embryos is able to maintain Otx2 expression in the anterior ectoderm of early streak embryos, and that a negative signal from the posterior mesendoderm, mimicked by exogenous retinoic acid, represses Otx2 expression in the anterior ectoderm of late streak embryos (Ang et al., 1994). Similar interactions have also been demonstrated in Xenopus (Blitz and Cho, 1995).
The possibility that retinoic acid might contribute to distinguishing between fore–midbrain and hindbrain at an early stage by controlling Otx2 expression is supported by the finding that the administration of exogenous retinoic acid at the mid–late streak stage represses early Otx2 expression in both the axial mesendoderm and the posterior neural plate (Ang et al., 1994; Simeone et al., 1995; Avantaggiato et al., 1996). This repression correlates with the appearance of microcephalic embryos showing early anteriorization of Hoxb1 expression, hindbrain expansion (Sive and Cheng, 1991; Conlon and Rossant, 1992; Marshall et al., 1992; Krumlauf, 1994), loss of forebrain molecular and morphological landmarks and gain of midbrain molecular markers in the most anterior neuroectoderm (Simeone et al., 1995; Avantaggiato et al., 1996). Moreover, Otx2 responsiveness to retinoic acid application is a common feature in different species including Xenopus and chick (Bally‐Cuif et al., 1995; Pannese et al., 1995). Nevertheless, the question of whether the interaction between endogenous retinoic acid and Otx2 is a physiological event in rostral CNS demarcation still remains unsolved.
Finally, experiments performed in Xenopus embryos also highlight the Otx2 involvement in the early specification of rostral CNS and, to some extent, complement the results obtained in mouse. In fact, microinjection of synthetic Otx2 RNA results in an abnormal reduction in the size of tail and trunk structures, and in the appearance of a second cement gland (Blitz and Cho, 1995; Pannese et al., 1995). These phenotypes have been interpreted either as a possible Otx2‐mediated interference with movements of extension and convergence during gastrulation (for trunk and tail reduction) or as an Otx2 requirement in the specification of most anterior head structures (for the ectopic cement gland). Moreover, by using a dexamethasone‐inducible OTX2 protein, it has been shown that the Xenopus Otx2 activity is regulated by a regionally restricted factor(s), and that the cement gland‐specific gene Xcg is a direct target of the Otx2 gene product (Gammill and Sive, 1997).
Otx1 is required for corticogenesis and sense organ development
The cerebral cortex is one of the most complex and fascinating areas of the brain. Neurons within the neocortex are organized in a highly ordered and differentiated array and arise from dividing progenitors of a simple neuroepithelium in which cells appear morphologically indistinguishable (McConnell, 1995).
During murine embryogenesis, Otx1 expression is detected first at the 1–3 somite stage (8 d.p.c.) throughout the forebrain and midbrain neuroepithelium (Simeone et al., 1993). In particular, during cerebral cortex development, Otx1 initially is transcribed throughout the entire dorsal telencephalic neuroepithelium; subsequently, towards the stages corresponding to the generation of neurons belonging to the deep cortical layers, it is restricted to the ventricular zone and, later, at the end of gestation, it becomes prominent in the cortical plate consisting of post‐migratory neurons of layers 5 and 6 (Frantz et al., 1994). At the end of gestation, the Otx1 signal is weakened in the ventricular zone and, postnatally, it is expressed prevalently in a subset of neurons in layers 5 and 6 (Frantz et al., 1994).
Moreover, Otx1 is also expressed in restricted derivatives of olfactory, visual and acoustic sense organs (Simeone et al., 1993).
To gain an insight into its functional role, Otx1 null mice have been generated in two different laboratories. Due to the different genetic background, in one case Otx1−/− mice die at birth (Suda et al., 1996), while in the other only 30% of them die at the weaning stage (Acampora et al., 1996). Both cases attest to the involvement of Otx1 in specific brain areas and at specific developmental stages. In particular, the cortex of adult Otx1−/− brains is reduced, and the identification of the neuronal layers in the temporal and perirhinal areas is difficult. This suggests that Otx1 is required for the development of the entire dorsal telencephalic cortex, with a more specific effect in the temporal and perirhinal areas, where events specifying neuronal identity also might be affected. The cortical phenotype is well correlated with a perturbation of early proliferative potentialities of neuronal precursors (Acampora et al., 1998a). Abnormalities identified in sense organs indicate that Otx1 is also required to specify the ciliary process in the eye and the lateral semicircular duct in the inner ear (Acampora et al., 1996).
It is noteworthy that Otx1 gene disruption generates a clear epileptic phenotype and occasional movement disorders including high speed turning behavior (Acampora et al., 1996). Epilepsy, one of the most common conditions in human pathology affecting 0.5–2% of the population, includes a variety of disorders related to electrical activity abnormalities of the brain. It has been suggested that cortical dysgenesis might be common and responsible for the so‐called cryptogenic epilepsies (Meencke and Janz, 1984; Raymond et al., 1995). Although several etiological factors have been proposed to cause cortical dysgenesis, genes affecting events early in corticogenesis appear to be the major candidates (Noebels, 1996). Therefore, mutations in the Otx1 gene might be responsible for a cortical dysgenesis disorder leading to epilepsy even in humans, although no such mutations have been identified so far.
Otx genes in brain regionalization
Events underlying the antero‐posterior patterning of the CNS begin to be established during the early gastrulation stage and lead to the generation of distinct transverse domains along the antero‐posterior (A/P) body axis. These early events require interactions among different tissues (the anterior visceral endoderm, axial mesendoderm and ectoderm), and several genes contribute to the achievement of these (Tam and Behringer, 1997; Rubenstein et al., 1998). It has been proposed that organizing centers are generated at the boundary between juxtaposed differently specified territories where cooperative interactions result in the production of signaling molecules with inducing properties (Meinhardt, 1983).
An inductive signal may be generated either at the boundary between adjacent transverse domains or in a restricted longitudinal domain running all along the A/P axis. In both cases, target tissues activate specific differentiating programs depending on their ability to respond to the inductive signal. Territorial competence and inductive signals produced by organizing centers are the main contributors towards the establishment of the morphogenetic fate of distinct brain areas.
Elegant transplantation experiments indicate both the presence of an organizer at the isthmic constriction of the mesencephalic–metencephalic (mes–met) junction and the existence of a different territorial competence between the brain regions located rostrally (prosomeres 3–6) and posteriorly to the zona limitans intrathalamica (mesencephalon and prosomeres 1 and 2) (Martinez et al., 1991; Marin and Puelles, 1994) (Figure 2). Two relevant signal molecules coded by the fibroblast growth factor‐8 gene (Fgf‐8) (Crossley and Martin, 1995; Crossley et al., 1996; Lee et al., 1997; Meyers et al., 1998) and Sonic hedgehog (Shh) (Echelard et al., 1993; Martì et al., 1995; Roelink et al., 1995; Chiang et al., 1996; Ericson et al., 1996) are transcribed locally at the isthmic constriction and zona limitans intrathalamica, respectively. Fgf‐8 midbrain‐inducing properties have been demonstrated (Crossley et al., 1996) while it can be hypothesized that Shh has a similar role in organizing tissue and/or in conferring different regional competence between territories rostral and caudal to the zona limitans intrathalamica (Martinez et al., 1991; Figdor and Stern, 1993; Rubenstein et al., 1994, 1998; Bally‐Cuif and Wassef, 1995; Crossley et al., 1996).
It is crucial to determine the molecular mechanisms necessary to specify adjacent territories with a different identity (e.g. mesencephalon and metencephalon), and in turn to allow the correct positioning of an organizer (e.g. isthmic organizer). The mes–met junction is molecularly defined by the Wnt‐1 expression ring and the posterior borders of Otx1 and Otx2 on the mesencephalic side, and by the Fgf‐8 expression ring and the anterior borders of Gbx2 and Pax2 on the metencephalic side (Bally‐Cuif and Wassef, 1995; Ang, 1996; Joyner, 1996; Millet et al., 1996) (Figure 2). Findings collected from expression data (Simeone et al., 1992a, 1993), Otx2 null mice (Acampora et al., 1995; Matsuo et al., 1995; Ang et al., 1996), transplantation experiments (Millet et al., 1996) and retinoic acid‐induced phenocopies (Simeone et al., 1995; Avantaggiato et al., 1996) support the possibility that Otx genes might contribute to mes–met development and, indeed, two recent reports have demonstrated their involvement in this process (Acampora et al., 1997; Suda et al., 1997).
In the first report (Acampora et al., 1997), mice carrying only one functional copy of Otx2 (Otx1−/−; Otx2+/−) show molecular and morphological transformation of the caudal diencephalon (prosomeres 1 and 2) and mesencephalon into an enlarged metencephalon, and the acquisition of mesencephalic molecular features in the telencephalon. The observed repatterning was assessed by studying the expression pattern of genes functionally involved in the establishment of the mes–met region such as Wnt‐1, En‐1 and Fgf‐8 (Bally‐Cuif and Wassef, 1995; Joyner, 1996; Rubenstein et al., 1998). While in Otx1−/−; Otx2+/− embryos at 8.5 d.p.c., the expression domains of Wnt‐1 and En‐1 (two early markers of the mesencephalic and metencephalic regions) are unaffected, Fgf‐8 distribution, in contrast, is broader and invades adjacent rostral territory (Figure 2). A few hours later, both Wnt‐1 and Fgf‐8 fail to form their narrow stripes at the mes–met boundary, and Fgf‐8 transcripts are detected along all the presumptive rostral brain. At 10.5 d.p.c., the repatterning process begins to be evident, with the transformation of the mesencephalon into the metencephalon, the establishment of an isthmic‐like structure in the caudal diencephalon and, by 12.5 d.p.c., with the telencephalic acquisition of mesencephalic features such as the expression of En‐2 and Wnt‐1 genes (Figure 2). These findings indicate that the repatterning observed is triggered by the early Fgf‐8 misexpression in response to a critical low level of Otx gene products (Acampora et al., 1997).
In the second report (Suda et al., 1997), similar conclusions were drawn from the analysis of double heterozygous embryos (Otx1+/−; Otx2+/−) showing, in a different genetic background, defects similar to those reported in Otx1−/−; Otx2+/− embryos in a BL6/DBA2 genetic background. In this report, molecular and anatomical analyses also show a severe reduction of both the mesencephalon and posterior diencephalon with an expansion of rhombomere 1 (Suda et al., 1997). These studies, therefore, indicate that a crucial threshold of Otx gene products (possibly influenced by genetic background) is required to confer on the mesencephalic field a sufficient level of specification to allow the correct positioning of Fgf‐8‐inducing properties at the isthmic organizer.
Furthermore, the findings that in Otx1−/−; Otx2+/− embryos, the dorsal telencephalon acquires mesencephalic molecular features and the zona limitans intrathalamica is absent (as revealed by anatomical inspection and loss of Shh expression), suggest that the ability of the telencephalon to express mesencephalic genes could be related directly to the loss of the zona limitans intrathalamica and/or Shh‐mediated signaling (Acampora et al., 1997).
otd/Otx evolutionary conservation
Striking evolutionary conservation of regulatory genes that control vertebrate development is exhibited by HOM/HOX complexes (Lewis, 1978; Duboule and Dollé, 1989; Krumlauf, 1994; van der Hoeven et al., 1996) and ey/Pax6 genes (Callaerts et al., 1997).
otd/Otx genes are also likely to have a conserved functional role in brain morphogenesis. This assumption is argued from sequence homology, which is restricted mainly to the homeodomain, and from striking similarities in their expression patterns and mutant phenotypes in Drosophila and mouse. In otd mutants, most proterocerebral neuroblasts and some deuterocerebral neuroblasts do not form, giving rise to a dramatically reduced brain (Finkelstein and Perrimon, 1990; Cohen and Jürgens, 1991; Hirth et al., 1995; Thor, 1995; Reichert and Boyan, 1997; Younossi‐Hartenstein et al., 1997). otd mutants also have pattern deletions in cephalic structures. For example, in ocelliless, a viable otd allele expression in the vertex primordium is abolished, and the ocelli (light‐sensing organs) and associated sensory bristles are lost (Finkelstein et al., 1990). Moreover, different levels of OTD protein are required for the formation of specific subdomains of the adult head (Royet and Finkelstein, 1995). In mouse, Otx genes are required in the early specification and patterning of the rostral neuroectoderm, in corticogenesis and proliferation of early telencephalic neuroblasts, as well as the development of visual and acoustic sense organs (Acampora et al., 1995, 1996, 1997; Matsuo et al., 1995; Ang et al., 1996).
To gain insight into the possibility that a basic genetic program of cephalic development might be conserved between vertebrates and invertebrates, human Otx genes have been introduced and overexpressed in Drosophila otd mutants (Leuzinger et al., 1998; Nagao et al., 1998) and, conversely, the murine Otx1 has been replaced with the Drosophila otd gene (Acampora et al., 1998a). Human Otx1 and Otx2 genes complement the otd defects allowing the rescue of brain, ventral nerve cord and cephalic defects in Drosophila. Moreover, their ubiquitous overexpression in the fly is able to induce ectopic neural structures (Leuzinger et al., 1998). Similarly, the Drosophila otd gene is able to replace the mouse Otx1 gene and fully rescue corticogenesis impairment and epilepsy, and also partially to recover eye defects and brain patterning abnormalities detected in Otx1−/−; Otx2+/− embryos; in contrast, the defective lateral semicircular duct of the inner ear of Otx1−/− mice is never recovered (Acampora et al., 1998a), suggesting that the ability to specify this structure therefore represents an Otx1‐specific property.
Despite the functional rescue observed, several aspects are still unclear. For example, the finding that homeodomains of a specific type such as the otd type are highly conserved might imply that they are crucial in selecting the same target sequence(s) with a very high stringency. In this connection, rescues of Drosophila otd and mouse Otx1 mutant phenotypes support the possibility that they control genetic hierarchies which share, at least in part, common functional features, and that the homeodomain‐mediated ability to recognize the same target sequence(s) might have been retained in evolution. In contrast, the role of coding sequences outside the homeodomain is only poorly understood, and it is important to determine whether these regions code for new functions, are evolved versions of an old function or represent a combination of old and new functions.
The rostral architectural components of the vertebrate brain, the telencephalon, the diencephalon and the mesencephalon, are clearly recognizable in vertebrates, while their existence is less clear in lower chordates (Kuhlenbeck, 1973). otd‐related genes have been found in all chordates (Simeone et al., 1992a; Bally‐Cuif et al., 1995; Mercier et al., 1995; Pannese et al., 1995; Wada et al., 1996; Williams and Holland, 1996, 1998; Ueki et al., 1998), where their expression is always associated with the most anterior CNS independently of the complexity acquired by this area during evolution. We propose that the architecture of this rostral Otx‐expressing region of the CNS might have been greatly modified on the basis of new genetic instruction(s). A posterior displacement of the mes–met boundary as well as differential proliferative properties of the rostral neuroectoderm (forebrain and midbrain) versus the more posterior neuroectoderm (hindbrain and spinal cord) might have contributed to the vertebrate‐type brain respecification. In this context, it should be kept in mind that an anterior displacement of the mes–met boundary is seen in Otx1−/−; Otx2+/− embryos and that Otx1−/− mice have reduced proliferative activity in the rostral neuroepithelium (Acampora et al., 1997, 1998a). Nevertheless, the rescues of Otx1 abnormalities by otd and vice versa argue in favor of an evolutionary conservation of several common otd/Otx1 properties and support the idea that conserved genetic functions required in mammalian brain development evolved in a primitive ancestor of flies and mice >500 million years ago (Wray et al., 1996).
Conclusions and future directions
Published findings now indicate that Otx1 plays important roles in patterning and terminal differentiating events of brain morphogenesis and that Otx2 is required in early specification and patterning of the rostral CNS. Although additional aspects related to sense organ development and pituitary control of hormone production (Acampora et al., 1998b) have been reported, the involvement of Otx1 and Otx2 in the development and evolution of the brain seems to be the most exciting aspect. Future experiments should identify Otx1 and Otx2 regulatory controls as well as the functional domains of their gene products and, thereby, define properties that are required to control developmental pathways or that have been created and selected during evolution to specify the greater complexity of the mammalian brain. These studies, together with the identification of a molecular partner(s) and downstream target(s), will significantly contribute to our knowledge on the morphogenesis and evolution of the mammalian brain.
I thank D.Acampora, S.L.Ang, M.Gulisano, S.Martinez and H.Reichert for helpful discussion and comments on the manuscript. I also thank A.Secondulfo and T.Reynolds for preparation of the manuscript. This work was supported by the Italian Telethon Program, the Italian Association for Cancer Research and the Progetto Finalizzato Biotecnologie—CNR.
- Copyright © 1998 European Molecular Biology Organization