Iroquois proteins comprise a conserved family of homeodomain-containing transcription factors involved in patterning and regionalization of embryonic tissues in both vertebrates and invertebrates. Earlier studies identified four murine Iroquois (Irx) genes. The isolation of two additional members of the murine gene family, Irx5 and Irx6, is reported here. Phylogenetic analysis of the Irx gene family reveals distinct clades for fly and vertebrate genes, and vertebrate members themselves were classified into three pairs of cognate genes. Mapping of the murine Irx genes has identified two gene clusters located on mouse chromosomes 8 and 13, respectively. Each gene cluster is represented by three Irx genes whose relative positions within both clusters are strictly conserved. Combined results from phylogenetic, linkage, and physical mapping studies provide evidence for the evolution of two Irx gene clusters by duplication of a larger chromosomal region and dispersion to two chromosomal locations. The maintenance of two cognate Irx gene clusters during vertebrate evolution suggests that their genomic organization is important for the regulation, expression, and function of Irx genes during embryonic development (Peters, 2000).

In zebrafish, the organizer is thought to consist of two regions, the yolk syncytial layer (YSL) and the shield. The dorsal YSL appears to send signals that affect formation of the shield in the overlying mesendoderm. A domain of dorsal deep cells located between the YSL and the shield is marked by expression of the iro3 gene. As gastrulation proceeds, the iro3 positive domain involutes and migrates to the animal pole. Iro3 expression is regulated by Nodal and bone morphogenic protein antagonists. Overexpression of iro3 induces ectopic expression of shield-specific genes. This effect is mimicked by an Iro3-Engrailed transcriptional repressor domain fusion, whereas an Iro3-VP16 activator domain fusion behaves as a dominant negative or antimorphic form. These results suggest that Iro3 acts as a transcriptional repressor and further implicate the iro3 gene in regulating organizer formation. It is proposed that the iro3-expressing dorsal deep cells represent a distinct organizer domain that receives signals from the YSL and in turn sends signals to the forming shield, thereby influencing its expansion and differentiation (Kudoh, 2001).

Iro3 is expressed initially at the sphere stage on one side of the blastoderm margin, expanding throughout the margin by 30% epibody, when iro3 and mesodermal marker ntl domains largely coincide. At 40% epibody, ntl expands slightly toward the animal pole, whereas iro3 remains restricted to the vegetal side of the margin, similar to the pattern of the endodermal marker gata5, and squint (sqt). The similarity to gata5 implies that iro3 expression marks a presumptive endodermal domain. At the 50% epibody stage, when the marginal layer starts to involute, iro3 expression becomes restricted to the dorsal side. The iro3 domain involutes toward the animal pole, and at midgastrula a few iro3-positive cells surround the anterior edge of the prechordal plate that is marked by gsc. Iro3 RNA disappears from the deep layer around the bud stage. iro3 is also expressed in the notochord at the late gastrula stage and in the neural tube during somitogenesis (Kudoh, 2001).

It is suggested that iro3, activated by Nodal factors and restricted to the dorsal side by signals ultimately dependent on ß-catenin, has an important role in patterning the mesendoderm in the early gastrula. The iro3-expressing domain may define a region with a distinct role in organizer formation in the early embryo. A multiple-organizer model, in which distinct regions act in temporal and spatial succession, has been proposed in Xenopus. The Nieuwkoop center is thought to induce the late blastula organizer in the vegetal half of the Spemann organizer; the blastula organizer region gives rise to head mesoderm, anterior notochord, anterior somites, and pharyngeal endoderm. The late blastula organizer, in turn, induces immediately above it the gastrula organizer, which develops into the notochord (Kudoh, 2001).

This model in Xenopus has some similarities in temporal and spatial aspects to a model suggested for zebrafish on the basis of these studies on iro3. Here, the iro3-expressing domain corresponds to the Xenopus late blastula organizer and, like this structure, includes pharyngeal endoderm precursor cells. It is suggested that the iro3 domain is spatially and functionally interspersed between the YSL, which represents the Nieuwkoop center equivalent, and the gastrula organizer. In sum, it is proposeed that the early expression domain of iro3 marks a distinct region of the evolving organizer, and that Iro3 activity in this domain has an important role in organizer expansion and differentiation (Kudoh, 2001).

Two novel zebrafish iroquois genes, ziro1 and ziro5, homologs of mouse Irx1 and mouse Irx5, respectively, have been isolated. The expression of both genes is initiated in dorsal neuroectoderm and mesoderm during gastrulation. Later, their expression appears in the central nervous system (CNS), excluding the telencephalon and most of the diencephalon. ziro1 expression is complementary to that of ziro3 in the notochord and later in the gut. In contrast, ziro5 expression mostly overlaps with that of ziro3. Interestingly, all three iroquois zebrafish genes are expressed in the notochord while only Irx3 is active in the mouse notochord. Their expression in later stages of embryogenesis was also compared (Wang, 2001).

Three mouse Iroquois-related genes, Irx1, -2 and -3, have been identified that have a homeodomain very similar to that of the Drosophila Iro-C genes. The three Irx homeodomains have 92-95% identity with Drosophila Ara and Caup. Seven human sequences have also been identified. The sequence similarity implies that these three genes represent a separate homeobox family. All three genes are expressed with distinct spatio/temporal patterns during early mouse embryogenesis. Irx3 is the earliest of the Iroquois-family to be expressed. Transcripts are first found at E6.5 in the extraembryonic portion of the egg cylinder, at high levels in the ectoplacental cone and in the lining of the ectoplacental cavity. All the structures expressing Irx3 are trophectoderm derivatives. The embryonic expression of Irx3 starts at E7.5 as two symmetrical patches in the anterior region of the embryo. At the same time, the first Irx1 expression is seen in comparable regions of the embryo. Irx1 transcripts are restricted to the mesodermal wings overlying the presumptive head-fold region. In contrast, Irx3 signals are confined to the ecodermal layer where they represent two expression domains in the margins of the neural plate. At E8.5, Irx1 mRNA is found dorsolaterally in the neural folds of the prospective mesencephalon, within the rhombencephalon, mostly in the precursor regions of rhombomeres 1 and 3, and additionally in the rostral part of the closing neural plate. Irx1 and Irx2 transcript distributions within the developing nervous system appear more restricted in comparison to Irx3. The most prominent expression of Irx1 in the brain is confined to the dorsolateral walls of the mesencephalon. Ira2 expression appears for the first time in the rhombencephalon, specifically active in the presumptive region of the future rhombomere 4. During neurogenesis, from E9.5 to E10.5, the three genes are predominantly expressed along the anteroposterior axis of the CNS and are expressed in a complementary pattern in the early development of the inner ear. These patterns implicate them in a number of embryonic developmental processes: the A/P and D/V patterning of specific regions of the central nervous system (CNS), and regionalization of the otic vesicle, as well as in the limbs where they may contribute to the process of digit formation (Bosse, 1997).

The Drosophila homeoproteins Ara and Caup are members of a combination of factors (prepattern) that control the highly localized expression of the proneural genes achaete and scute. Two Xenopus homologs of ara and caup (Xiro1 and Xiro2) have been identified. Like their Drosophila counterparts, they control the expression of proneural genes and, probably as a consequence, the size of the neural plate. In Xenopus, ectopic expression of these genes expands the neural plate, similar to the effect of overexpressing XASH-3 and ATH-3. Xiro expression precedes expression of the proneural genes, and partially overlaps the domains of expression of XASH-3 and ATH-3 and those of X-ngnr-1, another proneural gene. When overexpressed, X-ngnr-1 causes the differentiation of ectopic neurons. Xiro1 and Xiro2 are themselves controlled by noggin and retinoic acid. Like ara and caup, they are overexpressed in Xenopus embryos as a result of the expression of Drosophila cubitus interruptus gene, suggesting that neurogenesis is induced by the hedgehog family of proteins. These and other findings suggest the conservation of at least part of the genetic cascade that regulates proneural genes, and the existence in vertebrates of a prepattern of factors important to control the differentiation of the neural plate (Gomez-Skarmeta, 1998).

In both Xenopus and the mouse, two highly related genes, Xiro3 and Irx3 respectively, have been identified that encode a Drosophila Iroquois-related homeobox transcription factor. Xiro3 in Xenopus and Irx3 in the mouse are expressed early in the prospective neural plate in a subset of neural precursor cells. In Xenopus, injection of Xiro3 mRNA expands the neural tube and induces ectopic neural tissue in the epidermis, based on the ectopic expression of early neural markers, such as Xsox3. In contrast, the differentiation of the early forming primary neurons, as revealed by the expression of the neuronal marker N-tubulin, is prevented by Xiro3 expression. Activation of Xiro3 expression itself requires the combination of a neural inducing (noggin) and a posteriorizing signal (basic fibroblast growth factor). Overexpression of Xiro3 outside the areas where XASH3 is normally express does not result in ectopic expression of XASH3. However, when Xiro3 is directed into the neural plate, where XASH3 is normally expressed, the domain of XASH3 is slightly expanded. This result suggests that additional factors might cooperate with Xiro3 to induce proneural gene expression. These factors might be downstream of Xiro3 but are not activated until the levels of Xiro3 decline. Ectopic expression of X-Delta-1 in Xiro3-injected embyos is compatible with the idea that Xiro3 induces proneural gene expression, which subsequently induces the expression of X-Delta-1. These results suggest that Xiro3 activation constitutes one of the earliest steps in the development of the neural plate and that it functions in the specification of a neural precursor state (Bellefroid, 1998).

The vertebrate heart consists of two types of chambers, the atria and the ventricles, which differ in their contractile and electrophysiological properties. Little is known of the molecular mechanisms by which these chambers are specified during embryogenesis. Here a chicken iroquois-related homeobox gene, Irx4, was identified that has a ventricle-restricted expression pattern at all stages of heart development. Irx4 protein has been shown to regulate the chamber-specific expression of myosin isoforms by activating the expression of the ventricle myosin heavy chain-1 (VMHC1) and suppressing the expression of the atrial myosin heavy chain-1 (AMHC1) in the ventricles. Thus, Irx4 may play a critical role in establishing chamber-specific gene expression in the developing heart (Bao, 1999).

cDNAs of mouse Iroquois-related homeobox genes Irx1, -2, -3, -4, and -5 and have been isolated and their patterns of expression in the developing heart have been characterized. Irx1 and Irx2 are expressed specifically in the ventricular septum from the onset of its formation onward. In fetal stages, the expression of both genes appear to gradually become confined to the myocardium of the atrioventricular bundle and bundle branches of the forming ventricular conduction system. Irx3 is expressed specifically in the trabeculated myocardium of the ventricles. Irx4 expression is observed in a segment of the linear heart tube and the atrioventricular canal and ventricular myocardium including the inner curvature after looping, resembling the pattern of MLC2V. Transcripts for Irx5 are detected specifically in the endocardium lining the ventricular and atrial working myocardium that also express von Willebrand factor, but are absent from the endocardium of the endocardial cushions, i.e., the atrioventricular canal, inner curvature, and outflow tract. The spatiodevelopmental pattern of Irx5 matchs that of ANF, a marker for the forming working myocardium of the chambers. Taken together, all members of the Irx gene family are expressed in highly specific patterns in the developing mouse heart, suggesting a critical role in the specification of the distinct components of the four-chambered heart (Christoffels, 2000).

Reported here is the isolation and characterization of murine and human cDNAs encoded for by Irx4 (Iroquois homeobox gene 4). Mouse and human Irx4 proteins are highly conserved (83%) and their 63-aa homeodomains are more than 93% identical to those of the Drosophila Iroquois patterning genes. Human IRX4 maps to chromosome 5p15.3, which is syntenic to murine chromosome 13. Irx4 transcripts are present in the developing central nervous system, skin, and vibrissae, but are predominantly expressed in the cardiac ventricles. In mice at embryonic day (E) 7.5, Irx4 transcripts are found in the chorion and at low levels in a discrete anterior domain of the cardiac primordia. During the formation of the linear heart tube and its subsequent looping (E8.0 -8.5), Irx4 expression is restricted to the ventricular segment and is absent from both the posterior (eventual atrial) and the anterior (eventual outflow tract) segments of the heart. Throughout all subsequent stages in which the chambers of the heart become morphologically distinct (E8.5-11) and into adulthood, cardiac Irx4 expression is found exclusively in the ventricular myocardium. Irx4 gene expression has also been assessed in embryos with aberrant cardiac development: mice lacking RXRalpha or MEF2c have normal Irx4 expression, but mice lacking the homeobox transcription factor Nkx2-5 (Csx) have markedly reduced levels of Irx4 transcripts. dHand-null embryos initiate Irx4 expression, but cannot maintain normal levels. These data indicate that the homeobox gene Irx4 is likely to be an important mediator of ventricular differentiation during cardiac development downstream of Nkx2-5 and dHand (Bruneau, 2000).

Mammalian homologs of the Drosophila Iroquois homeobox gene complex, which is involved in patterning and regionalization of differentiation, have been identified. The six members of the murine family are organized in two cognate clusters of three genes each: Irx1, -2, -4 and Irx3, -5, -6, respectively. As a basis for further study of their regulation and function a comparative analysis of the genomic organization and of the expression patterns of all six Irx genes was performed. The genes are expressed in highly specific and regionalized patterns of ectoderm, mesoderm and endoderm derived tissues. In most tissues the pattern of expression of the clustered genes, especially of Irx1 and -2 and of Irx3 and -5, respectively, closely resemble each other while those of Irx4 and -6 are very divergent. Interestingly, the expression of cognate genes is mutually exclusive in adjacent and interacting tissues of limb, heart and the laryncho-pharyncheal region. The results indicate that the Irx genes are coordinately regulated at the level of the cluster (Houweling, 2001).

Murine Irx1, -2 and -4 (IrxA cluster) are located on chromosome 13 (human chromosome 5) and Irx3, -5 and -6 (IrxB cluster) on chromosome 8 (human chromosome 16), respectively. For insight into the genomic structure of the murine clusters, an analysis was performed of the IrxB cluster. Irx3 and -5 are separated by approximately 550 kb of intergenic sequences whereas Irx5 and -6 are 350 kb apart. In addition, Irx5 and -6 were found to be oriented in the same direction, whereas Irx3 is transcribed in the opposite direction. Investigation of the human IrxB cluster on chromosome 16 reveals a strict conservation of this organization. Since the relative positions of the cognate genes within the two murine clusters are identical, the orientations of IrxA genes are also likely to be conserved. Indeed, initial analysis of the human IrxA gene cluster reveals that Irx2 and -4 are oriented in the same direction (the orientation of Irx1 is presently ambiguous. In the human genome, no other potential coding sequences are found in the intergenic regions of the Irx genes (Houweling, 2001).

To get an insight into the spatial distribution of the gene expression in relation to their genomic organization, serial sections of E12 mouse embryos were hybridized with probes for the six Irx genes. The overall patterns of Irx1 and -2 and of Irx3 and -5, respectively, were found to be highly similar, as can be seen in the brain, neural tube, ganglia, head and body mesenchyme, limbs, heart and rib cartilage. Irx4, in contrast, is mainly restricted to the heart, skin and hindbrain and Irx6 transcripts are detected at low levels in the endocardium of the heart, identical to Irx5, in skin, and are very restricted in the hindbrain and neural tube. A more detailed analysis of the developmental and spatial expression profile was performed (Houweling, 2001).

Irx3 is first expressed in the extra embryonic portion of the egg cylinder at E6.5. Irx1 and -3 are first observed in the embryo at E7.5, in the mesoderm of the head-fold and ectoderm of neural plate, respectively. Irx4 expression is first observed in the chorion at E7, in the cardiac crescent at E7.5 and in the hindbrain from E8.5 onwards. At E8.0, Irx5 expression is detected in the neural epithelium of the head-fold and in the lateral plate mesoderm lining the coelomic cavity, similar to Irx3. At E8.5, Irx2 expression appears for the first time in the rhombencephalon. Irx6 expression is first seen at E10.5 in the endocardium of the heart chambers. Thus, the onset of Irx expression is different between members within a cluster and between cognate genes (Houweling, 2001).

Whole mount in situ hybridization analysis of Irx1, -2, -3 and -5 expressions revealed, from E9 onwards, a striking similarity in expression between Irx1 and -2 in the ventral region of mesencephalon and the metencephalon, and between Irx3 and -5 in the dorsolateral region of mesencephalon, ventrolateral region of metencephalon and along the entire spinal cord. The Irx4 expression is observed in the metencephalon only. The expression of Irx1 and -2 and of Irx3 and -5, respectively, is almost identical in the mesencephalon and metencephalon. Irx1 and -2 show high levels of expression in the cerebellum and the sensory part of the pons. Expression is also found in the medial region of the motor part of the pons, close to the median sulcus. In the region of the emerging fifth cranial nerve of the pons, expression is observed as well. Irx3 and -5 are detected in the lateral region of the motor part of the pons, adjacent to the sulcus limitans, in the sensory part of the pons and weakly in the developing cerebellum. Irx4 shows a restricted pattern in the pons at the level of the sulcus limitans and in the cerebellum (Houweling, 2001).

Irx6 expression first becomes detectable at E11.5 in spots in the floor plate of the mesencephalon, in a region where the motor nuclei of CNIII and -IV will develop, and, like all other Irx genes, in two spots in the pons, where the motor nuclei of CNVII will develop. None of the genes are expressed in the dorsal part of the caudal metencephalon. At E13.5, Irx1/2 and Irx3/5 expression within the thalamus region remain identical. Irx1 and -2 are expressed at higher levels and the expression extends further caudal than Irx3 and -5. Analysis of the Irx expression patterns in the spinal cord reveal identical patterns of Irx1 and -2 and Irx3 and -5, respectively. Irx1 and -2 expression is found in the ventricular zone and mantle layer of the alar plate. Higher levels of expression are observed in the median plane, in the interneuron region. Weak expression is found in the basal plate, the floor plate and the mesodermal notochord. Expression in the spinal ganglia is not observed in every in situ hybridization experiment, indicating low levels of expression. Irx3 and -5 expression is found in the ventricular zone and mantle layer of the alar and basal plates but not in the floor and roof plates. In addition, Irx3 and -5 expression is observed in the motor neurons and notochord. Irx4 expression is not found in the spinal cord. Irx6 expression is observed only in the motor neurons at E11.5 and E13.5 and very weakly in the ventricular zone at E13.5 (Houweling, 2001).

At E9.5, Irx1 and -2 expression patterns overlap in the lateroventral regions of the otic vesicle, whereas Irx3 and -5 overlap in the medial region (adjacent to the hindbrain). At E11.5, strong Irx1 and -2 expression is observed in the region of the developing ganglion of CNV and -VIII. Positive spots for Irx1 and -2 and Irx3 and -5 were found in the ganglion of CNVII, medial to the cranial cardinal vein. In a region dorsal to the Irx1 and -2 expressing area, expression of Irx3, -5 and -6 is found in the ganglia of the CNVII and -VIII complex including the region connected to the caudal part of the otic vesicle. Irx4 is not detected in developing ganglia of cranial nerves CNV, -VII and -VIII or otic vesicle. All Irx genes are expressed specifically in a restricted area of the marginal zone of the neural retina; expression extends further cranial than caudal (Houweling, 2001).

All members of the IrxB cluster, Irx3, -5 and -6, are expressed in the mammary gland primordia, whereas Irx1, -2 and -4 transcripts are virtually absent. All Irx genes are expressed in the epidermis in a localized manner. However, the level of expression and the localization of expression differs between genes. Irx2 and -4 are highly expressed in most of the epidermis of the embryo. Irx1 shows less expression in the epidermis of the body and almost no expression in the epidermis of the head. Irx3 and -5 are only weakly expressed in the epidermis and Irx6 only weakly in the region of the limbs and thorax. In the underlying dermis Irx1 and -3 are found to be expressed most abundantly; Irx2 and -5 only weakly, and Irx4 and -6 are not detected in the dermis. This is one of the very few tissues showing a similarity in pattern between the cognate genes. Within the epidermis derived vibrissae primordia, all Irx genes except Irx6 are detectably expressed, the signal being localized in the cap of the follicle colocalizing with the germinal matrix. However, Irx2 and -4 are more strongly expressed than Irx1, -3 and -5 (Houweling, 2001).

At E13.5, relatively high levels of Irx1 and -2 expression is detected in the molar primordia and in the primordia of the upper incisor teeth. Lower expression levels of Irx3 and -5 are also detected, but the expression domain is expanded in the epithelial ectoderm distally from the molar primordia. No Irx4 and -6 expression is detected in the teeth primordia (Houweling, 2001).

The spatio-developmental patterns of Irx1-6 during limb development were examined between E9.5 and E14.5. Expression of Irx1 and -2 is first observed at E10.5 in the proximal region of the limb bud with a broader domain anteriorly. The expression becomes confined to the cartilage condensations of the digits (E11.5) and changes to a pattern of three stripes within the digit condensations at E13.5. Whole mount staining with a probe for Gdf-5 transcripts, a marker for joint forming regions, shows a very similar pattern at E14.5, indicating that Irx1 and -2 are specifically expressed in the joint forming regions. Irx4 is not expressed in the limbs. Irx3, -5 and -6 show similar patterns, albeit that Irx6 gives a a much weaker signal, indicating a low level of expression. First expression of Irx3 and -5 is observed at E9.5 in the mesenchyme of the proximal region of the limb bud over the antero-posterior axis, highest at the dorsal side. At E11.5, the proximal limb region expresses Irx3 and -5 in identical patterns with a stronger domain anteriorly. Irx6 can be visualized in whole mounts in a similar pattern compared to Irx3 and -5. From E13.5 onwards, the expression of Irx3 and -5 becomes confined to the interdigital mesenchyme. At this stage, Irx6 is clearly detected on sections in a pattern identical to Irx3 and -5 (Houweling, 2001).

At E11.5 and E13.5, Irx1 and -2 are expressed at high levels and in similar patterns in the mesoderm adjacent to the endoderm of the laryngo-tracheal groove, a region that coincides with the developing laryncheal cartilages and the arytenoids (paired cartilages of the larynx). The endoderm of the groove does not express these genes. Irx3 and -5 expression is restricted to the endoderm of the groove and foregut, adjacent to the Irx1 and -2 expressing the mesoderm. Irx4 and -6 are not expressed in the region of the groove (Houweling, 2001).

From E10.5 onwards, when the lung buds are expanding from the foregut, expression of Irx1, -2, -3 and -5 is detected at high levels in the epithelial layer of the lung buds and bronchia. Expression is absent from the surrounding lung mesenchyme. The expression patterns within the heart of Irx4, Irx1-5 and Irx6 have been described. Spatial and developmental expression patterns of Irx1 and -2 are identical, confined to a segment within the ventricles that includes the septum. Irx4 is expressed in the atrioventricular canal myocardium and the myocardium of the ventricles and, until E10.5, in the proximal outflow tract. Irx3 expression is confined to the trabeculations of the ventricles from E9 onwards. The patterns of Irx5 and -6 are identical within the heart, both confined to the endocardial layer that lines the atrial and ventricular chambers, but, interestingly, not in the endocardium lining the cushions. The expression of Irx6 is detected a day later (E10.5) than Irx5, possibly due to the much weaker expression of Irx6 (Houweling, 2001).

Gonads of E11.5 and E13.5 mice strongly express Irx3 and weakly express Irx5 whereas Irx1, -2, -4 and -6 are absent. The expression shows a sharp border between gonad and mesonephros. From E11.5 onwards, Irx1 and -2 expressing spots, not restricted to morphologically distinguishable structures, are observed in the pancreatic endoderm, whereas Irx3-6 transcripts are not detectable. In the metanephros the expression of the IrxA cluster genes, Irx1, -2, and, weakly, -4, coincides with the developing metanephrogenic blastema, which includes the developing glomeruli. Irx3 expression is observed in similar spots and additionally in longitudinal strands of cells at the margin between the cortex and medulla (Houweling, 2001).

From E10.5 onwards, the expression of Irx1 and -2 is detected in the somites. From E11.5, expression becomes confined to a segmental pattern along the spine that coincides with the region of the rib primordia. This was confirmed by the probing of serial sections for the muscle specific beta-myosin heavy chain and SERCA-2A genes, which show an exactly alternating pattern of the forming intercostal muscles (Houweling, 2001).

At E11.5, the precondensations of the otic cap express Irx1, -2, -3 and -5. At E13.5, the precondensations of the nasal cap cartilage express Irx3 and -5 in very similar patterns. At E13.5, the outer curvature of the stomach endoderm expresses Irx3 and -5. Irx1, -2, -4 and -6 are never detected in the endoderm of the gastrointestinal tract (Houweling, 2001).

This analysis reveals that the spatio-temporal expression patterns of genes within one cluster resemble each other from E9.5 onwards. The expression of Irx1 and -2 (on chromosome 13) and of Irx3 and -5 (on chromosome 8) is very similar in almost all the tissues examined. Furthermore, Irx3, -5 and -6 of the IrxB cluster are expressed in the mammary gland, while the genes of the other cluster are not. Likewise, Irx3, -5 and -6 are expressed in the interdigital mesenchyme of the limbs, whereas Irx1 and -2 are expressed in the precartilage condensations of the digits. Therefore, the clustered genes are coordinately expressed, indicating coordinated regulation at the level of the gene cluster (Houweling, 2001).

Distinct classes of neurons are generated at defined positions in the ventral neural tube in response to a gradient of Sonic Hedgehog (Shh) activity. A set of homeodomain transcription factors expressed by neural progenitors act as intermediaries in Shh-dependent neural patterning. These homeodomain factors fall into two classes: class I proteins are repressed by Shh and class II proteins require Shh signaling for their expression. The profile of class I and class II protein expression defines five progenitor domains, each of which generates a distinct class of postmitotic neurons. Cross-repressive interactions between class I and class II proteins appear to refine and maintain these progenitor domains. The combinatorial expression of three of these proteins -- Nkx6.1, Nkx2.2, and Irx3 -- specifies the identity of three classes of neurons generated in the ventral third of the neural tube (Briscoe, 2000).

The expression of certain class I (Pax7, Dbx1, Dbx2, and Pax6) and class II (Nkx2.2) proteins is controlled by Shh signaling in vitro. The expression of class I proteins is repressed by Shh signaling, and the more ventral the boundary of class I protein expression in vivo, the higher is the concentration of Shh required for repression of protein expression in vitro. Conversely, Shh signaling is required to induce expression of the class II protein Nkx2.2 in vitro. Repression of Irx3 requires ~3 nM Shh-N, a concentration greater than that required for repression of Pax7, Dbx1, and Dbx2 expression, but less than that required for complete repression of Pax6. Conversely, induction of Nkx6.1 requires ~0.25 nM Shh-N -- a concentration lower than that required for induction of Nkx2.2 (3-4 nM). Thus, the link between the domains of expression of class I and class II proteins in vivo and the Shh concentration that regulates their expression in vitro extends to Irx3 and Nkx6.1. These findings support the idea that the differential patterns of expression of all class I and class II proteins depend initially on graded Shh signaling (Briscoe, 2000).

The boundaries of progenitor domains are sharply delineated in vivo, raising questions about the steps that operate downstream of Shh signaling to establish the nongraded domains of expression of class I and class II proteins. It was asked whether the domain of expression of class I proteins might be constrained by the action of the class II protein that abuts the same domain boundary, or vice versa. To test this, individual homeodomain proteins were misexpressed in the chick neural tube in mosaic fashion, and the resulting pattern of class I and class II protein expression was examined. Examined was the interaction between the class I protein Pax6 and the class II protein Nkx2.2 -- proteins that exhibit complementary domains of expression at the pMN/p3 boundary. To assess the influence of Pax6 on Nkx2.2, Pax6 was misexpressed ventral to its normal limit and the resulting pattern of Nkx2.2 expression was examined. After electroporation of Pax6, small clusters of ectopic Pax6+ cells were detected within the p3 domain. These cells lack Nkx2.2 expression, whereas expression of Nkx2.2 is maintained by neighboring p3 domain cells that lack ectopic Pax6 expression, arguing for a cell-autonomous action of Pax6. The expression of other class I and class II proteins is not affected by the deregulated expression of Pax6. Thus, Pax6 acts selectively to repress Nkx2.2 expression in p3 domain cells. These results complement studies showing a requirement for Pax6 activity in defining the dorsal limit of the p3 domain in vivo. The loss of Nkx6.1 function results in a ventral expansion in the extent of the p1 domain, without any change in Shh signaling. It is noteworthy that the boundaries of each of the five progenitor domains are sharply defined, yet class II proteins have been identified only at the pMN/p3 and p1/p2 boundaries. Thus, additional class II proteins may exist, with patterns of expression that complement the three orphan class I proteins (Briscoe, 2000).

This study has relied on ectopic expression methods to address the roles of Nkx6.1, Nkx2.2, and Irx3 in specifying the fate of V2 neurons, MNs, and V3 neurons. The results show that Nkx2.2 activity is sufficient to induce V3 neurons, that Nkx6.1 activity in the absence of Irx3 induces MNs, whereas Nkx6.1 activity in the presence of Irx3 induces V2 neurons. The inferences derived from these gain-of-function studies are supported by the switches in neuronal fate that occur in mice in which individual class I and class II proteins have been inactivated by gene targeting. In mice lacking Pax6 activity, the dorsal expansion in the domain of Nkx2.2 expression is accompanied by an expansion in the domain of V3 neuron generation, and by the loss of MNs. Conversely, the loss of Nkx2.2 results in the loss of V3 neurons and in the ectopic generation of MNs within the p3 domain. In addition, the loss of Nkx6.1 activity depletes the ventral neural tube of many MNs and V2 neurons (Briscoe, 2000 and references therein).

How do class I and class II proteins control neuronal subtype identity? The final cell division of certain ventral progenitors is accompanied by the onset of expression of a distinct set of homeodomain proteins, notably MNR2 and Lim3. Ectopic expression of MNR2 is able to induce MN differentiation independent of dorsoventral position, and ectopic expression of Lim3 induces V2 neurons. The studies indicate that class I and class II proteins function upstream of MNR2 and Lim3. Thus, within the pMN and p2 domains, the actions of progenitor homeodomain proteins in specifying neuronal subtype identity are likely to be mediated through MNR2 and Lim3. Subtype determinant factors with equivalent functions may therefore be expressed by cells in the other ventral progenitor domains (Briscoe, 2000).

A set of seven homeodomain proteins defines five neural progenitor domains with a fundamental role in the organization of ventral neural pattern. The analysis of these homeodomain proteins suggests that ventral patterning proceeds in three stages: (1) the regulation of class I and class II proteins by graded Shh signals; (2) the refinement and maintenance of progenitor domain identity by cross-repressive interactions between homeodomain proteins, and (3) the translation of a homeodomain protein code into neuronal subtype identity. The central features of this model may apply to other vertebrate tissues in which cell pattern is regulated by local sources of extrinsic signals. Consistent with this idea, cross-regulatory interactions between transcription factors have been suggested to refine cell pattern in the embryonic mesoderm and in the pituitary gland. The principles of the model of ventral patterning outlined here resemble those involved in subdividing the Drosophila embryo. Graded Shh signaling subdivides the ventral neural tube into five domains, just as graded levels of the dorsal protein establish five distinct regions of the early Drosophila embryo, suggesting an upper limit to the number of distinct cell fates that can be generated in response to a single gradient signaling system. In addition, the graded anterioposterior distribution of maternally supplied factors in the Drosophila embryo is known to initiate the expression of a set of proteins encoded by the gap genes. Subsequent cross-regulatory interactions establish and maintain sharp boundaries in the expression of gap proteins, and their activities within individual domains control later aspects of cell pattern. Thus, in the neural tube and the Drosophila embryo, the cross-repression of genes whose initial expression is controlled by graded upstream signals provides an effective mechanism for establishing and maintaining progenitor domains and for imposing cell type identity (Briscoe, 2000 and references therein).

In the early Xenopus embryo, the Xiro homeodomain proteins of the Iroquois (Iro) family control the expression of proneural genes and the size of the neural plate. Xiro1 functions as a repressor that is strictly required for neural differentiation, even when the BMP4 pathway is impaired. Xiro1 and Bmp4 repress each other. Consistently, Xiro1 and Bmp4 have complementary patterns of expression during gastrulation. The expression of Xiro1 requires Wnt signaling. Thus, Xiro1 is probably a mediator of the known downregulation of Bmp4 by Wnt signaling (Gomez-Skarmeta, 2001).

A comparative lung expression analysis of the murine Irx1 and Irx2 genes was performed. At embryonic day 8.5 (E8.5), the Irx1 and Irx2 expression starts in the foregut region, where the laryngo-tracheal groove will form. The expression is prominent in the lung epithelium during glandular development. It declines at the end of the canalicular phase. The Irx1 and Irx2 expression domains were compared to Gli1, 2, 3 and Mash1. Their homologs in Drosophila are known as regulative partners of the iroquois complex. The Irx and Gli genes are coexpressed in the developing lungs. Their transcripts are not localized in the same cells but adjacent to each other in either mesenchymal or epithelial structures. Araucan, one of the prepattern genes of the iroquois complex, is regulated by cubitus interuptus and regulates in turn the gene activity of the achaete scute-complex (AS-C). The murine homologs of ci and AS-C are the Gli and Mash genes, respectively. Upstream as well as downstream factors in the iroquois regulation cascade in Drosophila are expressed in the developing mouse lung. All three Gli genes are expressed in the overlapping domains of the lung mesenchyme. At the same time, the Irx1 and Irx2 genes are active in the distal part of the epithelium. The branching morphogenesis of the lung depends upon interactions between the epithelium and the mesoderm. Mash1-positive cells can be detected on cross-sections. They are located in the proximity of the iroquois-positive epithelium and belong to pulmonary neuroendocrine cells and their precursors. In the same space of time, Irx and Gli genes are coexpressed in the developing lungs in adjacent tissues, whereas Mash1 transcripts can be found at the same time in the developing lungs but solely in single cells (Becker, 2001).

The gene for activin ßA is expressed in the early odontogenic mesenchyme of all murine teeth but mutant mice show a patterning defect where incisors and mandibular molars fail to develop but maxillary molars develop normally. In order to understand why maxillary molar tooth development can proceed in the absence of activin, the role of mediators of activin signalling in tooth development was explored. Analysis of tooth development in activin receptor II and Smad2 mutants shows that a similar tooth phenotype to activin ßA mutants can be observed. In addition, a novel downstream target of activin signalling, the Iroquois-related homeobox gene, Irx1, has been identified; its expression in activin ßA mutant embryos is lost in all tooth germs, including the maxillary molars. These results strongly suggest that other TGFß molecules are not stimulating the activin signalling pathway in the absence of activin. This was confirmed by a non-genetic approach using exogenous soluble receptors to inhibit all activin signalling in tooth development. These reproduced the genetic phenotypes. Activin, thus, has an essential role in early development of incisor and mandibular molar teeth but this pathway is not required for development of maxillary molars (Ferguson, 2001).

Although multiple axon guidance cues have been discovered in recent years, little is known about the mechanism by which the spatiotemporal expression patterns of the axon guidance cues are regulated in vertebrates. A homeobox gene, Irx4, is expressed in a pattern similar to that of Slit1 in the chicken retina. Overexpression of Irx4 leads to specific downregulation of Slit1 expression, whereas inhibition of Irx4 activity by a dominant negative mutant leads to induction of Slit1 expression, indicating that Irx4 is a crucial regulator of Slit1 expression in the retina. In addition, by examining axonal behavior in the retinas with overexpression of Irx4 and using several in vivo assays to test the effect of Slit1, it was found that Slit1 acts positively to guide the retinal axons inside the optic fiber layer (OFL). The regulation of Slit1 expression by Irx4 is important for providing intermediate targets for retinal axons during their growth within the retina (Jin, 2003).

The nephron, the basic structural and functional unit of the vertebrate kidney, is organized into discrete segments, which are composed of distinct renal epithelial cell types. Each cell type carries out highly specific physiological functions to regulate fluid balance, osmolarity, and metabolic waste excretion. To date, the genetic basis of regionalization of the nephron has remained largely unknown. This study shows that Irx3, a member of the Iroquois (Irx) gene family, acts as a master regulator of intermediate tubule fate. Comparative studies in Xenopus and mouse have identified Irx1, Irx2, and Irx3 as an evolutionary conserved subset of Irx genes, whose expression represents the earliest manifestation of intermediate compartment patterning in the developing vertebrate nephron discovered to date. Intermediate tubule progenitors will give rise to epithelia of Henle's loop in mammals. Loss-of-function studies indicate that irx1 and irx2 are dispensable, whereas irx3 is necessary for intermediate tubule formation in Xenopus. Furthermore, misexpression of irx3 is sufficient to direct ectopic development of intermediate tubules in the Xenopus mesoderm. Taken together, irx3 is the first gene known to be necessary and sufficient to specify nephron segment fate in vivo (Reggiani, 2007).

Limb skeletal pattern relies heavily on graded Sonic hedgehog (Shh) signaling. As a morphogen and growth cue, Shh regulates identities of posterior limb elements, including the ulna/fibula and digits 2 through 5. In contrast, proximal and anterior structures, including the humerus/femur, radius/tibia, and digit 1, are regarded as Shh independent, and mechanisms governing their specification are unclear. This study shows that patterning of the proximal and anterior limb skeleton involves two phases. Irx3 and Irx5 (Irx3/5) are essential in the initiating limb bud to specify progenitors of the femur, tibia, and digit 1. However, these skeletal elements can be restored in Irx3/5 null mice when Shh signaling is diminished, indicating that Shh negatively regulates their formation after initiation. These data provide genetic evidence supporting the concept of early specification and progressive determination of anterior limb pattern (Li, 2014).

Axon pathfinding relies on the ability of the growth cone to detect and interpret guidance cues and to modulate cytoskeletal changes in response to these signals. The murine POU domain transcription factor Brn-3.2 regulates pathfinding in retinal ganglion cell (RGC) axons at multiple points along their pathways and the establishment of topographic order in the superior colliculus. Using representational difference analysis, Brn-3.2 gene targets likely to act on axon guidance have been identified at the levels of transcription, cell-cell interaction, and signal transduction, including the actin-binding LIM domain protein abLIM. Evidence is presented that abLIM plays a crucial role in RGC axon pathfinding, sharing functional similarity with its C. elegans homolog, UNC-115. These findings provide insights into a Brn-3.2-directed hierarchical program linking signaling events to cytoskeletal changes required for axon pathfinding (Erkman, 2000).

To understand the molecular mechanisms by which Brn-3.2 exerts its effects on RGC axon guidance, candidate target genes have been identified using a modification of representational difference analysis. Thus far, screening has yielded three potentially novel genes, and seven with matching sequences in mouse EST and human genomic and cDNA databases, the structure and function of which have not been reported. A number of genes were obtained that share a very low degree of homology with known genes and cannot be classified at this time. In addition, five candidate target genes have been identified that represent previously characterized molecules, including the transcription factors Irx6, EBF/Olf-1, EBF/Olf-2, and a mouse homolog of rat Neuritin. Irx6 is a homeodomain transcription factor with homology to the Drosophila genes of the iroquois complex, Olf-1 and Olf-2, belonging to the early B cell factor (EBF) family of HLH transcription factors, and Neuritin, a GPI-anchored neuronal cell surface protein, are all well characterized. One RDA product that did not present homology to published sequences was ultimately identified by analysis of cDNA clones as part of the 3'UTR of the mouse homolog of the human actin binding zinc finger protein abLIM. The amino acid sequence of the LIM domain containing all four LIM motifs specific for the retinal isoform is highly conserved and shows 97% identity to the human sequence. In situ hybridization analysis of m-abLIM in E15.5 mice shows, in addition to its expression in the inner layer of the retina, expression in other neuronal structures including peripheral sensory ganglia, spinal cord, SC, and nonneural tissues such as the thymus (Erkman, 2000).

If these genes are regulated by Brn-3.2, their mRNA levels should decrease in Brn-3.2^-/- retina, and their spatial and temporal patterns of expression should support such an assumption. Indeed, comparison of mRNA levels in E15.5 wild-type and Brn-3.2^-/- retinas reveals a dramatic decrease in the levels of Irx6, Olf-1, m-abLIM, and Neuritin, and a modest effect on Olf-2 mRNA levels, indicating that even relatively small differences can be detected by the modified RDA protocol. Temporal expression patterns of Brn-3.2, Irx6, Olf-2, and m-abLIM were determined using adjacent sections of retina at different developmental stages. Expression of Brn-3.2 mRNA, first detectable in the retina at E11.5, precedes initial detectable expression of Irx6 and Olf-2 around E12.5, and m-abLIM around E13.5. Thus, these genes may represent components of a molecular cascade regulated by Brn-3.2 (Erkman, 2000).

The iroquois (iro) homeobox genes participate in many developmental processes both in vertebrates and invertebrates -- among them are neural plate formation and neural patterning. The Xenopus Iro (Xiro) function in primary neurogenesis has been studied in detail. Misexpression of Xiro genes promotes the activation of the proneural gene Xngnr1 but suppresses neuronal differentiation. This is probably due to upregulation of at least two neuronal-fate repressors: XHairy2A and XZic2. Accordingly, primary neurons arise at the border of the Xiro expression domains. In addition, XGadd45-gamma has been identified as a new gene repressed by Xiro. XGadd45-gamma encodes a cell-cycle inhibitor and is expressed in territories where cells will exit mitosis, such as those where primary neurons arise. Indeed, XGadd45-gamma misexpression causes cell cycle arrest. It is concluded that during Xenopus primary neuron formation, in Xiro expressing territories neuronal differentiation is impaired, while in adjacent cells, XGadd45-gamma may help cells stop dividing and differentiate as neurons (de la Calle-Mustienes, 2002).

XGadd45-gamma belongs to a three-member family that encodes the small (18 kDa), related proteins Gadd45-a, Gadd45-ß/MyD118 and Gadd45-gamma/CR6. Apparently, these proteins have similar functions in the arrest of cell growth and in apoptosis, although the details of their mechanism(s) of action are unknown. They all interact with the cyclin dependent kinase inhibitor p21^WAF1/CIP1, the proliferating cell nuclear antigen (PCNA, a component of a complex which includes cyclin-dependent kinases and p21^WAF1/CIP1), and Cdc2/Cyclin B1. Overexpression of each protein in cell culture promotes cell-cycle arrest and apoptosis. Some evidence supports that Gadd45-a inhibits Cdc2/Cyclin B1 kinase activity. In addition, these proteins bind to and activate the MTK1/MEKK4 kinase, which itself activates the p38/JNK kinase pathway that promotes apoptosis in response to environmental stress. However, the Gadd45 requirement for p38/JNK kinase activation remains controversial. In contrast, Gadd45-a, Gadd45-ß and Gadd45-gamma/CR6 are differentially activated in response to several signaling pathways and a variety of genetic and envionmental stresses, and show distinct patterns of expression in mice tissues. These data suggest that, despite their very similar composition, each protein may have some functional specificity (de la Calle-Mustienes, 2002).

During Xenopus early development, XGadd45-gamma is expressed in a dynamic pattern that prefigures many territories where cells will stop dividing. Examples include (during gastrulation) the dorsal mesoderm, and, in the neurula stages, the neuronal precursors, the prospective ciliated cells, the somites and the prospective cement gland. In all these territories, XGadd45-gamma mRNA transcription precedes mitotic quiescence. As expected for a cell-cycle inhibitor, overexpression of XGadd45-gamma at moderate levels reduces the number of dividing cells and this effect is not a consequence of an increase of cell death. Injection of high amounts of XGadd45-gamma mRNA promotes lethality. Gastrulation is impeded, and cells lose adherence and are extruded into the vitelline space. These XGadd45-gamma effects are similar to those observed with other cell-cycle inhibitors. Sites of XGadd45-gamma expression coincide, in some cases, with those of cyclin-dependent kinase inhibitor p27^XIC1. Moreover, overexpression of either XGadd45-gamma or p27^XIC1 arrests the cell-cycle in vivo. Thus, it is possible that Gadd45-gamma not only interacts with p21^WAF1/CIP1, but also with p27^XIC1. Interference with XGadd45-gamma function with a specific morpholino oligonucleotide did not affect the cell cycle (de la Calle-Mustienes, 2002).

This may be due to redundancy between different Gadd45 proteins. The spatial and temporal patterns of expression of Gadd45-gamma and the Notch ligand XDl1 largely coincide. Moreover, both XGadd45-gamma and XDl1 are positively regulated by proneural genes and negatively controlled by Notch signaling. According to the lateral inhibition model, activation of the Notch pathway within a cell, by signaling from neighboring cells, maintains the cell's mitotic potential and prevents its differentiation. In contrast, a cell that expresses high levels of Notch ligands and signals strongly, escapes lateral inhibition, exits the cell cycle and differentiates. XGadd45-gamma may provide a link between Notch signaling, cell-cycle arrest and differentiation. Thus, in the neural plate, cells with high levels of proneural genes have also high levels of XDl1 and XGadd45-gamma. The first allows them to escape lateral inhibition, and the second to exit the cell cycle. These cells can then differentiate. Mitotic arrest mediated by XGadd45-gamma probably occurs through interaction with cyclin and inhibitors of cyclin-dependent kinases. In neighboring cells, the Notch pathway is activated, proneural genes and XGadd45-gamma are downregulated, and cell-cycle arrest and differentiation cannot occur. It is of interest that induction of Gadd45 genes in cell culture stops the cell cycle in G1 phase. This phase is compatible with exiting the cell cycle, a requirement for terminal neuronal differentiation. Cells that differentiate outside the neural plate may resort to genes different from the proneural ones to accumulate Notch ligands and XGadd45-gamma (de la Calle-Mustienes, 2002).

This study compares the effects of overexpressing either Xiro1, -2 or -3 in neural development. To make comparisons more meaningful, equivalent constructs were prepared in the pCS2MT plasmid. The overexpression of each Xiro gene causes similar effects, although Xiro3 was approximately five to ten times more potent. Paradoxically, the overexpressions activated Xngnr1 and repressed neuronal differentiation. This may be explained at least in part by the finding that Xiro upregulates the neuronal repressors XHairy2A and XZic2. Indeed, it has been shown that XZic2 antagonizes development of Xngnr1-promoted ectopic neurons. XZic2 antagonizes Xngnr1-promoted XGadd45-gamma and XDl1 expression. Consistently with these findings, in wild type embryos the intermediate stripes of expression of XHairy2A and XZic2 are within the Xiro1 domains. Also in accordance with these results, in the prospective spinal chord, the Xiro1 domain is contained within the broader Xngnr1 domain and neurons arise at the border of the Xiro1 domain. Taken together, these observations suggest that Xiro proteins simultaneously participate in the activation of Xngnr1 and of genes that antagonize primary neuron formation (de la Calle-Mustienes, 2002).

Overexpressions of Xiro genes represses both XGadd45-gamma and XDl1 in territories where primary neurons arise. Consistently, in wild type embryos, XGadd45-gamma and XDl1 are expressed at the borders of Xiro domains. Moreover, XDl1 is activated in embryos expressing a Xiro1 chimera that converts the Xiro1 repressor into an activator (HD-GR-E1A). This activation occurs even in the absence of protein synthesis. Thus, XDl1 is probably directly repressed by Xiro. However, XGadd45-gamma is repressed by HD-GR-E1A, probably because Xngnr1 is also downregulated. Indeed, coinjection of HD-GR-E1A and Xngnr1 mRNAs rescues the expression of XGadd45-gamma. Thus, Xiro-mediated repression of XGadd45-gamma is probably indirect and may take place, at least in part, by Xiro-upregulated neuronal repressors. In this case, interference with Xiro function would suppress neuronal repressors, but would also downregulate Xngnr1, which is needed for XGadd45-gamma expression (de la Calle-Mustienes, 2002).

A model is proposed for the function of Xiro in neural patterning that integrates the above data. Xiro proteins, as well as other factors, participate in the activation of Xngnr1. Within the Xiro domains, Xngnr1 does not activate XDl1 or XGadd45-gamma, and cannot promote differentiation of primary neurons due to the upregulation by Xiro of neuronal repressors, such as XHairy2A and XZic2. In addition, Xiro probably represses XDl1 directly. Outside the Xiro domains, other factors, such as the Gli proteins, activate Xngnr1, which in turn promotes the expression of XDl1 and XGadd45-gamma in those cells that will become primary neurons. XDl1 switches on the lateral inhibition mechanism by which the Notch signaling pathway is activated in neighboring cells. This pathway downregulates proneural genes, XDl1 and XGadd45-gamma. As a consequence, these cells keep dividing and do not differentiate. In contrast, cells with high levels of Xngnr1, XDl1 and XGadd45-gamma escape lateral inhibition, exit the cell cycle (in part due to the presence of XGadd45-gamma) and differentiate as primary neurons. This differentiation is triggered by a genetic program activated by Xngnr1. Thus, Xiro proteins may help coordinate cell cycle and differentiation (de la Calle-Mustienes, 2002).