Shh-dependent formation of the ZLI is opposed by signals from the dorsal diencephalon

During the development of the vertebrate central nervous system, regional cell identities emerge through the actions of inductive signals that direct the fate of neural progenitor cells. Many of these inductive signals derive from discrete neural cell groups, and they influence cell pattern along both the dorsoventral and rostrocaudal axes of the neural tube. The fate of neural progenitors is determined largely by the position that they occupy with respect to orthogonally arrayed patterning signals.

Along the dorsoventral axis, the assignment of neural cell fate is achieved by a mechanism that is conserved along much of the length of the neural tube. Signals provided by two polar cell groups, the floor plate ventrally and the roof plate dorsally, impose distinct cell fates on intervening neural progenitor cells (Lee and Jessell,1999; Tanabe and Jessell,1996). Ventral cell fates are specified largely through a gradient of Sonic hedgehog (Shh) activity that emanates from the floor plate(Chiang et al., 1996; Ericson et al., 1995; Wijgerde et al., 2002),whereas the fate of many dorsal cell types is imposed by BMP family members secreted from the roof plate (Liem et al.,1997; Nguyen et al.,2000; Timmer et al.,2002). This basic system of polarized Shh and BMP signaling from midline cell groups operates within the spinal cord, hindbrain, midbrain and caudal forebrain (reviewed by Briscoe and Ericson, 2001; Lee and Jessell, 1999). Moreover, even in the rostral forebrain, where there is no overt floor plate or roof plate, Shh and BMP appear to impose regional pattern on neural cells (Barth and Wilson, 1995; Dale et al.,1997; Ericson et al.,1995; Furuta et al.,1997; Golden et al.,1999; Hebert et al.,2002; Shimamura and Rubenstein, 1997).

In contrast to the uniformity of dorsoventral patterning, the specification of rostrocaudal neural identity appears to obey a more fragmentary logic, with distinct organizing centers operating over different rostrocaudal domains of the neural tube. Three major signaling centers are known to control rostrocaudal neural pattern: the anterior neural ridge (ANR), the isthmic organizer (IsO) and the node. The ANR is positioned at the rostral extreme of the neural tube and directs rostrocaudal cell fates in the telencephalon, in part through the actions of Fgf8 (Houart et al., 1998; Shimamura and Rubenstein, 1997). The IsO is positioned at the junction between the midbrain and hindbrain (Broccoli et al., 1999; Li and Joyner,2001; Martinez-Barbera et al.,2001; Millet et al.,1999), and regulates the specification of cell types in the midbrain and rostral hindbrain, through the secretion of Wnt1 and Fgf8(reviewed by Liu and Joyner,2001). At more caudal levels of the neural tube that give rise to the spinal cord, rostrocaudal positional identity is influenced by node-derived signals, and Fgf8 is a major component of activity of the node(Liu et al., 2001). Thus, FGF signaling is a common feature of the activity of three distinct rostrocaudal organizing centers.

Within the diencephalon, however, the rostrocaudal patterning of cell types occurs independently of signals provided by the ANR and the IsO(Chi et al., 2003; Jaszai et al., 2003; Shanmugalingam et al., 2000),and is likely to depend on signals provided by the ZLI, a prominent structure that protrudes from the basal plate at the boundary between the prospective ventral thalamus and the dorsal thalamus(Kiecker and Lumsden, 2004). Shh is expressed within the ZLI (Echelard et al., 1993), and the acquisition of post-mitotic neural identities in adjacent thalamic tissues emerges in the wake of the ventral to dorsal progression of Shh expression within the ZLI(Larsen et al., 2001). Recent studies demonstrated that Shh signals are required for the specification of thalamic identitites, and that the likely source of these signals is the ZLI(Hashimoto-Torii et al., 2003; Kiecker and Lumsden, 2004). Thus, the ZLI, as with the IsO and the ANR, is aligned perpendicular to the main axis of the neural tube, but as with a dorsoventral organizing center,the floor plate expresses Shh. In some respects then, the ZLI may be an organizing center with properties characteristic of both rostrocaudal and dorsoventral patterning centers.

Although the ZLI has been implicated in diencephalic patterning, it is as yet unclear how the ZLI is formed. Some evidence about the early patterning mechanisms that govern the initial position of ZLI formation has emerged. At neural plate stages, cross-repressive interactions between the transcription factors Six3 and Irx3 establish a boundary that anticipates the position of formation of the ZLI (Kobayashi et al.,2002). After neural tube closure, Wnt8b expression marks the prospective ZLI and is flanked by domains of Lunatic fringe(Lfng) expression (Garcia-Lopez et al., 2004). Furthermore, ectopic Lfng expression represses ZLI formation, suggesting that the lateral limits of the ZLI are normally constrained by these adjacent domains of Lfng expression(Zeltser et al., 2001). Despite these insights, however, the factors responsible for inducing the ZLI,in the context of these positional constraints, remain unknown.

In this study, I have examined the molecular mechanisms that control ZLI formation in an avian forebrain explant system. Through fate-mapping experiments, I show that the ZLI differentiates in the alar plate of the diencephalon in response to inductive signals derived from the basal plate. Shh signaling from the basal plate is required to initiate Shhexpression within the ZLI, and, subsequently, long-range Shh signals provided by the basal plate and ZLI are needed for its dorsal progression. Finally, the dorsal limit of progression of the ZLI appears to be constrained in part by a limit in the range of Shh action, and in part by an opponent signal emanating from the dorsal diencephalon.

Fertilized White Leghorn chicken eggs (Charles River) were incubated at 38°C until the desired developmental stage(Hamburger and Hamilton,1951).

Two-color in situ hybridization was performed as described(Dietrich et al., 1997) with the following probes: Shh, Ptc2(Pearse et al., 2001), Lfng (Laufer et al.,1997), Wnt3a, Foxg1(Bell et al., 2001). Immunohistochemistry was performed as described(Yamada et al., 1993) with mouse anti-Shh (5E1) (Ericson et al.,1996) and sheep anti-GFP (Biogenesis). Alexa488- and Cy3-conjugated secondary antibodies were obtained from Molecular Probes and Jackson ImmunoResearch Laboratories, respectively. Images were obtained with a Nikon E800 microscope.

Stage 13-14 chick heads were hemissected along the longitudinal axis of the neural tube, the mesenchyme removed, and tissue containing the forebrain and midbrain regions cultured in the presence of surrounding ectodermal tissues on a Millipore filter in DMEM:F12 (Specialty Media) and 10% FCS (Hyclone). Before fixation with 4% paraformaldehyde (PFA), the explants were attached to the underlying filters with collagen (Cohesion).

DiI injection (Molecular Probes) and photooxidation were performed as described (Psychoyos and Stern,1996). Because the alar and basal plates could respond differently to factors in the culture medium, where possible, I assessed the relative length of the ZLI as a fraction of the length of the alar plate.

To block Hh signaling, the monoclonal antibody 5E1 (Developmental Studies Hybridoma Bank) was added to the culture medium at 25 ng/μl. Hh-Ag1.3 (10 nM to 1 μM; Curis) was added to the culture medium, which induced spinal cord progenitors in a concentration-dependent manner(Frank-Kamenetsky et al.,2002).

Intact forebrain explants were electroporated in a Sylgard dish using a concentric bipolar electrode (FHC, Bowdoinham, ME) to produce focal regions of misexpression. A 2.5 mg/ml DNA solution was injected above the target tissue and four 50-msec pulses of 15V were delivered by a T820 electrosquareporator(BTX, San Diego, CA). mShh-CD4 (Yang et al., 1997) and full-length mShh(Riddle et al., 1993) were cloned into the pCAGGS vector (Fukuchi et al., 1994). Each plasmid solution contained a mixture of 0.25 mg/ml pCAGGS-GFP plasmid (Momose et al.,1999). Explant manipulations were performed after electroporation. Explants were incubated at 37°C for 48 hours, embedded in collagen, and fixed as described above. Electroporated cells were visualized by staining with a sheep-anti-GFP antibody (Biogenesis).

To characterize the temporal dynamics of ZLI differentiation in the developing diencephalon, I monitored Shh and Ptc2expression. Shh was expressed in the ventral diencephalon from Hamburger and Hamilton stages 5-6(Hamburger and Hamilton,1951), and its dorsal limit of expression appeared to coincide with the alar/basal plate boundary of the diencephalon, defined by a morphologically visible sulcus at stage 13-14(Fig. 1A,J). Shhexpression was first detected in an extreme ventral domain of the prospective ZLI at stage 15, in a small group of cells that extended dorsally from the basal plate. From stage 15, this domain of Shh expression, predictive of ZLI differentiation, expanded dorsally at a rate of ∼20 μm/hour, to reach a length of 500-600 μm at stage 20(Fig. 1A-C). At this and later stages, Shh expression had extended to 75-80% of the length of the alar plate, but did not reach the dorsal midline of the diencephalon(Fig. 1C,J,K).

The expression of the Hh receptor Ptc2 is upregulated by Shh signaling, and its expression marks cells exposed to a high level of Hh activity (Pearse et al.,2001). From neural plate stages, Ptc2 was expressed by cells adjacent to the domain of Shh expression in the diencephalic basal plate, and thus delineates the ventral-most cells of the alar plate. Following the onset of ZLI differentiation at stage 15, Ptc2expression was cleared from the prospective ZLI, and two stripes of Ptc2 expression flanked the Shh domain in the ZLI and extended dorsally in parallel with the expansion of Shhexpression.

To examine the source and identity of signals that initiate the program of ZLI differentiation, I analyzed diencephalic development in an in vitro culture system (described in Materials and methods). After 48 hours, these explants achieved a size equivalent to that of stage 20 embryos grown in ovo. The onset of ZLI differentiation in vitro, as marked by Shh and Ptc2 expression, extended dorsally at ∼13 μm per hour(Fig. 1D-I,K), reaching a final position in the alar plate similar to that observed in stage 20 embryos in ovo. Thus, the progression of ZLI differentiation in vitro appears to proceed in a similar manner to that in vivo, albeit at a slightly slower rate.

I next investigated whether the progressive dorsal expansion of Shh expression in the ZLI reflects the migration of Shh-expressing cells from the basal plate, or the induction of ZLI differentiation in cells already positioned within the alar plate. To determine whether cells from the basal plate contribute to the ZLI, I injected carbocyanine dye (DiI) focally into the basal plate of stage 13-14 explants. The precise location of labeled cells with respect to the ZLI was determined by photo-conversion of DiI and analysis of the position of cells containing DiI-induced precipitate in relation to the domain of Ptc2 expression(Fig. 2A). After DiI injections into the basal plate, few if any labeled cells were detected in the ZLI after 48 hours (n=24; Fig. 2C-E). Thus, the dorsal migration of Shh-expressing basal-plate cells does not appear to underlie the progressive dorsal expansion of Shh expression that is predictive of ZLI differentiation.

To locate the cells that give rise to the ZLI, I compared the initial and final positions of DiI-labeled cells in each explant. In 88% (60/68) of explants, DiI-labeled cells occupied the same relative position along the dorsoventral axis (±15%) after 48 hours in culture(Fig. 2B), and cell movements that could account for the dorsal progression of ZLI differentiation were not observed. DiI injection into the ventral 40% of the diencephalon (relative initial position <0.4) resulted in final locations within the basal plate,whereas DiI injection in the middle third of the diencephalon (relative final position 0.4-0.6) resulted in cell locations within the ZLI (n=11,black circles in Fig. 2B,F-H). Consistent with fate-mapping studies performed at stage 9-10(Garcia-Lopez et al., 2004),these data suggest that at stage 13-14, cells destined to contribute to the ZLI initially occupy the alar plate. In turn, this finding implies that these cells acquire their ZLI character in response to an extrinsic inductive signal.

To address the origin of signals involved in ZLI differentiation, I assessed whether the basal plate is required for the induction of Shhexpression in the ZLI. Using the sulcus at the alar/basal plate boundary as a guide, I excised the ventral diencephalon and assessed ZLI formation in resulting dorsal explants grown for 24-48 hours. Two sets of dorsal (D)explants were analyzed: D2/3 explants, which contained most of the alar plate,and D1/2 explants, which lacked the ventral-most region alar plate(Fig. 3A). To confirm the accuracy of dissection, I assessed Shh and Ptc2 expression in dorsal explants fixed immediately after dissection (t=0 hours). In addition, to determine the ventral boundary of dorsal explants relative to the basal plate, I assessed Ptc2 expression in the complementary ventral region of excised diencephalic tissue (V explants) at t=0. All V1/2 and V1/3 explants expressed Shh and Ptc2 at the time of dissection, but expression of these genes was never detected in D1/2 and D2/3 explants (n=19, 20 respectively), providing evidence that dorsal explants do not contain basal-plate tissue(Fig. 3B,C,F,G,J,K).

I next assessed the extent of ZLI differentiation in D2/3 and D1/2 explants grown in vitro for 48 hours. Shh and Ptc2 were expressed in most D2/3 explants (n=41/57; Fig. 3D,E), although the length of the Shh expression domain varied. By contrast, none of the D1/2 explants expressed Shh or Ptc2 after the same period in vitro (n=0/24; Fig. 3H,I). The emergence of a ZLI-like domain of Shh expression in D2/3, but not in D1/2, explants suggests that cells in the ventral-most region of the alar plate initially received signals that are sufficient to initiate or sustain a program of ZLI differentiation.

I therefore considered whether the ZLI-inducing activity of the ventral alar plate is produced locally, or whether it is the result of signals secreted from the basal plate. To assess whether the basal plate is a source of Shh-inducing signals, I monitored Shh expression in D1/2 explants co-cultured with basal-plate tissue(Fig. 4A). D1/2 explants never expressed Shh when grown alone, but they did express Shhwhen the basal plate was grafted to the ventral side of the D1/2 explant(n=8/9; Fig. 4B-C). This finding supports the idea that the basal plate is the source of a signal(s) that initiates Shh expression in adjacent alar-plate tissue. In turn, the emergence of Shh expression in a ZLI-like domain in D2/3 explants grown alone suggests that the dorsal progression of ZLI differentiation rapidly acquires independence from ongoing signaling from the basal plate.

The basal plate of the diencephalon is a prominent site of Shhexpression, prior to the differentiation of the ZLI(Echelard et al., 1993),raising the question of whether the ZLI-inductive signal supplied by the basal plate involves Shh. To assess this, I analyzed ZLI differentiation in the presence of a monoclonal antibody, MAb-5E1, that blocks Hh signaling(Ericson et al., 1996). Preincubation of basal-plate explants with MAb-5E1 blocked their ability to induce Shh expression when subsequently co-cultured with D1/2 explants (n=9/9; Fig. 4D). This finding provides a first line of evidence that Shh signaling from the basal plate is required for the initiation of ZLI differentiation.

I also assessed the effect of blocking Hh signaling in intact forebrain explants that were not subdivided (intact explants). Exposure of stage 13-14 forebrain explants to MAb-5E1 for 48 hours did not alter Shhexpression in the basal plate, supporting the view that once Shhexpression in the basal plate has been established, its persistence does not require ongoing Hh signaling. By contrast, the domain of Shhexpression within the ZLI was dramatically reduced in all intact forebrain explants grown in the presence of MAb-5E1 (mean reduction of 70-80%; n=52; Fig. 5B,C). The persistence of Shh expression in the ventral domain of the ZLI of stage 13-14 explants exposed to MAb-5E1 could reflect the prior influence of Shh signaling. To test this, forebrain explants were exposed to MAb-5E1 at stages 10-12, and under these conditions Shh expression in the ZLI domain was virtually eliminated (n=9/9; Fig. 5D). Together, these findings support the idea that Shh signaling is required for the initiation of a program of ZLI differentiation.

Hh proteins can influence surrounding cells through their actions as long-range inductive signals (Briscoe et al., 2001; Struhl et al.,1997), through the local induction of secondary signals that themselves serve as long-range factors(Basler and Struhl, 1994; Tabata and Kornberg, 1994; Zecca et al., 1995), or through the cell-to-cell propagated induction of Shh expression in adjacent cells (Huangfu et al.,2003; Placzek et al.,1993; Struhl et al.,1997). To begin to determine which of these strategies is used to elicit ZLI differentiation, I first examined whether there is a continuous requirement for Shh signaling during the dorsal expansion of the ZLI. To assess this, I analyzed Shh expression in stage 13-14 explants exposed to MAb-5E1 at different times during the 48 hours explant culture period (Fig. 5A). I reasoned that a sensitivity to MAb-5E1 only at early times in culture would indicate a transient involvement of Shh signaling in ZLI differentiation, and suggest the induction of a secondary patterning signal. By contrast, I observed that the progressive expansion of Shh expression in the ZLI was arrested by MAb-5E1 addition at different times throughout the first 30 hours of the culture period (Fig. 5H). Moreover, the length at which ZLI progression was arrested increased as a function of the delay before the addition of MAb-5E1(Fig. 5E-G, blue curve in H). The rate of growth of the ZLI in forebrain explants grown in the presence of MAb-5E1 was nearly identical to that of untreated explants grown for 12-28 hours (Fig. 1K, red curve). Together, these observations are best explained by a continuous requirement for Shh signaling during ZLI differentiation, and argue against the induction of a secondary and molecularly distinct relay signal.

Shh expression in the basal plate extends along the length of the forebrain, whereas expression in the alar plate is restricted to the narrow stripe predictive of the ZLI. The competence of alar-plate cells to express Shh may therefore be restricted to the presumptive ZLI. To assess this, I examined the consequences of widespread exposure of diencephalic alar-plate cells to Hh activity. D1/2 explants were cultured in the presence of the small molecule Hh agonist 1.3 (Hh-Ag1.3) for 48 hours(Frank-Kamenetsky et al.,2002). Shh was expressed selectively in a stripe of cells that corresponded to the position of the ZLI(Fig. 6D). Thus, alar-plate cells rostral and caudal to the prospective ZLI are not competent to express ZLI markers upon exposure to Hh agonists. This finding implies the existence of a mechanism that confines Shh expression to the ZLI.

A second constraint on ZLI differentiation operates along the dorsoventral axis of the diencephalon, and is revealed by the failure of the ZLI to extend to the dorsal midline. The dorsal restriction in ZLI differentiation could be explained by a limit in the range of Shh action, and/or by the presence of a dorsally located inhibitor that blocks the progression of ZLI differentiation. To distinguish between these possibilities, I assessed ZLI differentiation in D1/2 explants exposed to a high concentration of Hh-Ag1.3. I reasoned that if the limit of ZLI differentiation is determined solely by the range of action of Shh from basal-plate cells, then this treatment should extend the Shh expression domain, and by inference ZLI differentiation, to the dorsal midline. However, I observed that exposure of intact forebrain or D1/2 explants to Hh-Ag1.3 did not expand the zone of ZLI differentiation, raising the possibility of the existence of a Shh-independent inhibitor of ZLI formation in the dorsal diencephalon. I next varied the concentration of Hh-Ag1.3, and found that lower concentrations resulted in shorter domains of Shh expression that always abutted the ventral boundary of the explant. Taken together, these observations suggest that cells in the dorsal diencephalon, by virtue of their expression of, or exposure to, an inhibitor,require higher levels of Hh to differentiate into ZLI than do their ventral counterparts (Fig. 6B-E). These findings raise the possibility that the limit to ZLI progression results from the presence of a dorsally restricted inhibitor of ZLI differentiation.

To assess the ability of diencephalic tissue to form a ZLI in the absence of potentially inhibitory signals from the dorsal diencephalon, I generated intermediate (I) explants by removing the dorsal fifth of the neural tube from dorsal (D) explants (Fig. 7A). To facilitate survival of the narrow intermediate explants, the ventral boundaries of the dorsal and intermediate explants were shifted to a position between D1/2 and D2/3 explants. Wnt3a is expressed in the dorsal midline of the diencephalon and midbrain and in the prospective thalamus from stage 12-13 (data not shown), and this marker was therefore used to assess the accuracy of the dissection and the maintenance of a diencephalic identity in the explant. The dorsal domain of Wnt3a expression was observed in both intact and dorsal explants, but was completely lost from intermediate explants (Fig. 7E), providing evidence that the dorsal extreme of the neural tube was removed completely. Moreover, Wnt3a expression was detected in intact, dorsal and intermediate explants (Fig. 7C-E) after two days in culture, suggesting that a diencephalic identity was maintained in the absence of ongoing signaling from ventral and/or dorsal tissues.

To address whether signals present at the dorsal diencephalic midline constrain Shh expression and ZLI differentiation, I assessed the effects of removing the dorsal fifth of a dorsal explant on cell differentiation in the remaining intermediate explant. Removal of the extreme dorsal diencephalic tissue in intermediate explants increased the average length of the ZLI by 45%, from 204 μm in dorsal explants (n=14) to 295 μm in intermediate explants (n=42) after exposure to low levels of Hh-Ag1.3 (Fig. 7B,F,G). Moreover, the length of the Shh domain reverted to dorsal explant-like levels when intermediate explants were grown in contact with excised dorsal fifth diencephalic tissue (D^*)(Fig. 7B,H). These data are consistent with the idea that a secreted signal from the dorsal diencephalon impairs the ability of alar-plate cells to acquire ZLI character in response to Hh signals.

I also examined the ventral to dorsal propagation of ZLI differentiation in dorsal versus intermediate explants elicited by a localized source of Hh signals, a scenario that is likely to approximate the process of ZLI differentiation in vivo. Stage 13-14 forebrain explants were focally electroporated with a membrane-tethered form of mShh (mShhCD4)(Yang et al., 1997). Dorsal explants electroporated with mShhCD4 did not express Shh at any position in the explant (n=43; Fig. 7F). Because mShhCD4 is active in other assays (data not shown), this finding suggests that electroporation of mShhCD4 achieves a low level of Shh signaling. By contrast,in mShhCD4-expressing intermediate explants, Shh was expressed throughout the ZLI domain, even when electroporation was restricted to the ventral aspect of the explant (n=4; Fig. 7G). Similar results were obtained with electroporation of a secreted form of mShh in intermediate versus intact forebrain explants (data not shown). These data provide additional support for the idea that the dorsal diencephalic region is the source of a factor that can block the progression of ZLI differentiation.

To examine more directly whether a dorsally derived signal opposes ZLI formation, I assessed whether grafts of dorsal diencephalic tissue can suppress the propagation of ZLI differentiation. The dorsal fifth of the diencephalon was cultured adjacent to the prospective thalamic territory in stage 13-14 explants, and the location of the grafted tissue relative to the ZLI after two days in culture was determined by in situ hybridization with probes against Wnt3a and Shh, respectively(Fig. 8A). Dorsal diencephalic tissue inhibited the propagation of Shh expression in the adjacent ZLI beyond the position of the graft (n=16; Fig. 8C,E). Control grafts of intermediate telencephalic tissue, marked by Foxg1 expression(Bell et al., 2001), did not affect the expression of Shh in the ZLI (n=9; Fig. 8B,D,F). Taken together,the complementary effects of the removal or addition of dorsal diencephalic tissue provide strong evidence that it provides a signal that opposes the propagation of ZLI differentiation.

The ZLI is a prominent landmark in the developing diencephalon and has been implicated in the patterning of adjacent thalamic structures(Hashimoto-Torii et al., 2003; Kiecker and Lumsden, 2004),yet little is known about its formation. In this study I provide evidence that the differentiation of the ZLI from progenitor cells in the alar plate is initiated by a Shh-dependent signal from the basal plate, and that the subsequent dorsal progression of ZLI differentiation is also directed by Shh signaling. The dorsal limit of ZLI differentiation appears to be defined by an inhibitory influence from the extreme dorsal region of the diencephalon that opposes the Shh-mediated differentiation of the ZLI. Diencephalic cells that lie rostral and caudal to the position of the ZLI are refractory to the inductive influence of Shh, thus confining the ZLI to a narrow stripe along the rostrocaudal axis of the diencephalon.

Previous in ovo cell-labeling studies have revealed that the ZLI forms within a narrow lineage-restricted compartment that is demarcated by borders of Lfng expression (Zeltser et al., 2001). However, in these in ovo studies the initial position of labeled cells was not determined precisely, and thus the cellular origin of the ZLI remained unclear. My findings help to resolve the issue of whether the ZLI emerges through the directed dorsal migration of Shh-expressing basal-plate cells or from de novo induction of ZLI character within alar-plate cells. The present fate-mapping studies in forebrain explants provide strong evidence that the program of ZLI differentiation is initiated in cells that occupy the ventral region of the alar plate, and subsequently propagates dorsally in the absence of extensive cell migration. These findings are consistent with a recent fate-mapping study that locates prospective ZLI territory in the alar plate of stage 9-10 chick embryos(Garcia-Lopez et al.,2004).

In principle, the inductive signals that initiate ZLI differentiation could be transferred in planar fashion from adjacent diencephalic tissues, a suggestion that has been favored(Echevarria et al., 2003; Echevarria et al., 2001; Kobayashi et al., 2002), or alternatively could be transmitted vertically from the basal plate. The observation that cells in the ventral-most region of the alar plate, present in D2/3 explants but not in D1/2 explants, are required to sustain a program of ZLI differentiation provides evidence that ventrally derived signals are involved. A plausible candidate for a ventral ZLI-initiating signal derived from the basal plate is Shh. The ability of basal-plate tissue to induce ZLI differentiation in adjacent dorsal explants was severely reduced by a blockade of Shh signaling within the basal plate, suggesting that Shh signaling from the basal plate is required to initiate ZLI differentiation.

In contrast to the absolute requirement for Shh signaling in ZLI formation in the chick, studies in the zebrafish system provide evidence that a subpopulation of cells in the ZLI is independent of Hh or basal plate-derived signals. Whereas treatment of chick forebrain explants with the Hh pathway inhibitors MAb 5E1 or cyclopamine (data not shown) abolishes Shhexpression in the ZLI, zebrafish embryos grown in the presence of cyclopamine exhibited a reduced domain of Shh expression in the ZLI(Mathieu et al., 2002). Divergent conclusions have emerged from analyses of ZLI formation in zebrafish smoothened (smu) mutants, in which signaling downstream of all Hh family ligands is disrupted(Holzschuh et al., 2003; Mathieu et al., 2002). Holzschuh et al. observed a requirement for Hh signaling at 24 hpf that was not observed at 28 hpf by Mathieu et al. This discrepancy could reflect a delay in ZLI formation in the smu mutants, and/or bona fide differences in the smu alleles analyzed. The presence of a reduced domain of Shh expression at the position of the ZLI in smumutants at 28 hpf could be explained by a gradual loss of the activity of maternal Smu protein, which coincides with the period of ZLI formation(Varga et al., 2001). This phenotype would be analogous to the reduced ZLI I observed when intact explants were grown in the presence of Shh-blocking antibodies(Fig. 5C,D). Alternatively, the presence of an Hh-independent subpopulation of cells in the ZLI in zebrafish could reflect a difference between the two species. A small domain of Shh expression at the position of the ZLI is also observed in 26% of zebrafish cyclops mutants, which lack a basal plate in the diencephalon (Sampath et al.,1998). Shh expression in these cells could result from inefficient Shh signaling from the subjacent notochord, or could represent a subpopulation of the ZLI that is independent of basal plate-derived signals. Although the nature of the ZLI phenotype in zebrafish mutants needs to be analyzed further, the existence of an Hh-independent subpopulation in the ZLI could reflect species differences analogous to those observed in another Shh-expressing tissue, the floor plate (reviewed by Strahle et al., 2004). In the chick and mouse, specification of both the medial floor plate (MFP) and lateral floor plate (LFP) are Hh-dependent(Chiang et al., 1996; Ericson et al., 1996). By contrast, LFP formation in zebrafish is dependent on Hh signaling, whereas formation of the MFP is not (Chen et al.,2001; Karlstrom et al.,1999; Schauerte et al.,1998; Varga et al.,2001).

I have found that ZLI differentiation in the alar plate progresses in the absence of sustained signaling from the basal plate, raising the issue of how ZLI differentiation is propagated dorsally. The dorsal propagation of ZLI differentiation can be arrested by a blockade of Hh signaling late in the period of explant culture, suggesting a requirement for ongoing Shh signaling during the period of ZLI differentiation. This finding does not support the possibility that Hh merely initiates ZLI differentiation, with its dorsal progression depending on a distinct secondary signal. Since there is not a continuous requirement for a basal-plate source of Shh for propagation of ZLI differentiation within alar-plate explants, the ZLI is the most likely source of Shh involved in this later phase of diffusible signaling. Taken together,my findings support the idea that basal plate-derived Shh signals initiate ZLI differentiation in the alar plate, and that Shh signals from the ZLI itself participate in the subsequent dorsal progression of ZLI differentiation.

Shh is expressed throughout the rostrocaudal axis of the diencephalic basal plate, raising the issue of how the rostrocaudal position of the ZLI is determined, with respect to the much broader ventral domain of expression of Shh, its inductive signal. At neural plate stages, several genes are expressed in domains that have boundaries at the position of the future ZLI,and some of these genes are likely to have a role in restricting ZLI differentiation. The domains of Six3 and Irx3 expression in the neural plate abut each other, and could influence the position of ZLI formation by establishing zones of non-competence in flanking neural tissue(Kobayashi et al., 2002).

At later developmental stages at which overt ZLI differentiation occurs, Six3 expression regresses rostrally so that it no longer abuts the Irx3 domain, consistent with the idea that other factors directly regulate the initiation and progression of ZLI differentiation. At stage 13, Lfng and Wnt3b are expressed in complementary domains that demarcate the borders of the ZLI(Garcia-Lopez et al., 2004; Garda et al., 2002; Zeltser et al., 2001). Moreover, ectopic Lfng disrupts cell sorting at the compartment boundaries and prevents Shh expression in the ZLI(Zeltser et al., 2001),suggesting that Lfng constrains ZLI differentiation in flanking thalamic tissues by repressing the potential for Shh expression. The maintenance of normal borders of Lfng expression in the absence of Shh signals is consistent with the idea that a distinct set of developmental cues, possibly provided by Six3, Irx3 and Wnts, establishes a prepattern in the diencephalon (Braun et al.,2003; Kobayashi et al.,2002; Kiecker and Lumsden,2004).

The ZLI never encroaches into the dorsal extreme of the diencephalon, and dorsal diencephalic tissue does not respond to Hh exposure with ZLI induction,suggesting that dorsal diencephalic tissue is normally unable to initiate ZLI differentiation. Following the removal of the dorsal extreme of the neural tube, the enhanced sensitivity of dorsal regions of the diencephalon to Shh signaling suggests that this tissue is exposed to a signal that suppresses ZLI differentiation. Moreover, the ventral-to-dorsal gradient of sensitivity of diencephalic tissue to different levels of Hh signaling during ZLI induction suggests that this inhibitor acts in a graded manner. The suppression of ZLI propagation adjacent to grafts of dorsal diencephalic tissue provides additional evidence for a secreted inhibitory factor. Candidate factors for such a secreted dorsal inhibitor include members of the BMP/GDF(Furuta et al., 1997; Golden et al., 1999; Lee and Jessell, 1999) and Wnt(Garda et al., 2002; Hollyday et al., 1995)families of signaling molecules. In caudal neural tissue, long-range BMP signals from the surface ectoderm and prospective roof plate have been proposed to antagonize floor-plate induction by Shh secreted from the notochord (Patten and Placzek,2002). Simultaneously, BMPs secreted from the roof plate influence late patterning in the brain and spinal cord by antagonizing Shh signaling(Liem et al., 2000; Liem et al., 1995; Ohkubo et al., 2002).

My findings suggest that the ZLI possesses features of both rostrocaudal and dorsoventral signaling centers. The node, IsO, ANR and ZLI secrete signaling molecules, notably FGFs, which control cell identities along the rostrocaudal axis. The ZLI, as with the IsO, is oriented perpendicular to the main axis of the neural tube, yet is distinct from other rostrocaudal organizers in that it expresses Shh, a signaling molecule associated with the two main dorsoventral organizers, the notochord and floor plate.

Although the ZLI influences the acquisition of thalamic cell identities along the rostrocaudal axis, the mechanism regulating ZLI differentiation resembles that involved in establishing sequential dorsoventral organizers. In particular, ZLI differentiation is induced in the diencephalic alar plate through a mechanism reminiscent of the induction of lateral floor-plate cells(LFP) in the ventral-most region of the neural tube. Here, Shh signaling from the MFP is required to induce LFP differentiation in adjacent cell populations in the neuroepithelium (Charrier et al.,2002; Ding et al.,1998; Matise et al.,1998; Schauerte et al.,1998). Moreover, floor-plate signaling appears to be opposed by long-range signals from the dorsal neural tube(Patten and Placzek, 2002). In contrast to the LFP, which occupies only a few cell layers in the ventral neural tube, long-range Shh signaling leads to the extensive propagation of ZLI differentiation along the dorsoventral axis of the diencephalon. Cells in the MFP and LFP are likely to play redundant roles in ventral patterning in the neural tube (Odenthal et al.,2000). By contrast, Shh signals emanating from the basal plate and ZLI appear to have distinct functions in patterning diencephalic cells along orthogonal axes. These dual and orthogonal sources of Shh signaling in the diencephalon may contribute to the emergence of the complex nuclear organization of the thalamus.

I am indebted to Tom Jessell for his advice and support throughout this project. This work was carried out in the laboratories of Claudio Stern and Ed Laufer, and I thank them for their advice, critical reagents and laboratory facilities. I also thank Dan Vasiliauskas, Ben Novitch and Hynek Wichterle for comments on the manuscript. This work was supported by an NICHHD RO3 grant.