The mammalian thalamus is located in the diencephalon and is composed of dozens of morphologically and functionally distinct nuclei. The majority of these nuclei project axons to the neocortex in unique patterns and play critical roles in sensory, motor, and cognitive functions. It has been assumed that the adult thalamus is derived from neural progenitor cells located within the alar plate of the caudal diencephalon. Nevertheless, how a distinct array of postmitotic thalamic nuclei emerge from this single developmental unit has remained largely unknown. Our recent studies found that these thalamic nuclei are in fact derived from molecularly heterogeneous populations of progenitor cells distributed within at least two distinct progenitor domains in the caudal diencephalon. In this study, we investigated how such molecular heterogeneity is established and maintained during early development of the thalamus and how early signaling mechanisms influence the formation of postmitotic thalamic nuclei. By using mouse genetics and in utero electroporation, we provide evidence that Sonic hedgehog (Shh), which is normally expressed in ventral and rostral borders of the embryonic thalamus, plays a crucial role in patterning progenitor domains throughout the thalamus. We also show that increasing or decreasing Shh activity causes dramatic reorganization of postmitotic thalamic nuclei through altering the positional identity of progenitor cells.
The thalamus is located in the vertebrate diencephalon and plays critical roles in controlling animal behavior. Each of the thalamic nuclei exhibits a unique pattern of gene expression and connectivity. In mammals, a vast majority of thalamic nuclei send axons to unique sets of distinct neocortical areas. For example, principal sensory nuclei (visual, somatosensory, and auditory) establish topographic and area-specific thalamocortical projections to primary sensory areas in the neocortex (Jones, 2007). Despite the functional importance of the thalamus, however, molecular mechanisms that control the specification of thalamic nuclei are not well understood. A particularly intriguing question is how a single developmental unit in the alar plate of caudal diencephalon [prosomere 2 in the study by Puelles and Rubenstein (2003)] can generate such a diverse array of postmitotic nuclei at later stages of development.
Numerous studies have shown that ventral-to-dorsal gradient of Shh activity specifies different neuronal subtypes throughout the embryonic CNS (Fuccillo et al., 2006). In the diencephalon, Shh is not only expressed in the basal plate, which is ventral to the thalamus, but is also expressed in the zona limitans intrathalamica (ZLI), a dorsally extending group of cells located immediately rostral to the early embryonic thalamus (Puelles and Rubenstein, 1993; Shimamura et al., 1995). In ovo electroporation and grafting studies in chick showed that ectopic expression of Shh in the caudal diencephalon and mesencephalon induces the expression of Gbx2 and reduces Pax6 (Kiecker and Lumsden, 2004; Vieira et al., 2005). Conversely, inhibition of Shh signaling by a dominant-negative form of Shh receptor Ptc1 reduced Nkx2.2, Ptc1, and Gbx2 expression (Kiecker and Lumsden, 2004). Together, these studies establish the importance of Shh signaling in the global regionalization of the diencephalon, particularly its role in specifying the identity of the thalamus as a whole. To date, however, whether Shh signaling also controls the identity of progenitor cells within the thalamus along the rostrocaudal axis and subsequently contributes to the specification of different thalamic nuclei has remained largely unknown.
We recently reported molecular heterogeneity of thalamic progenitor cells in the mouse and analyzed the postmitotic fate of each progenitor cell population using genetic lineage tracing methods (Vue et al., 2007). We proposed that the thalamic ventricular zone is marked by the expression of the basic-helix-loop-helix (bHLH) transcription factor Olig3, and is divided into two distinct progenitor domains, pTH-R and pTH-C (Fig. 1A,B). The pTH-R domain is rostroventrally located within the thalamic primordium, expresses transcription factors Nkx2.2 and Mash1, and contributes largely to nuclei that do not project to the cortex. These nuclei were previously not considered to be part of the postmitotic thalamus. The other domain, pTH-C, expresses Ngn1 and Ngn2 and gives rise to all the classical thalamic nuclei projecting to the cortex. Within pTH-C, Olig2, a bHLH protein, is expressed in a high-rostroventral to low-caudodorsal gradient, while the homeodomain protein Dbx1 is expressed in the opposite gradient. Because Nkx2.2, Olig2, and Dbx1 are known to be differentially regulated by Shh in the ventral spinal cord (Briscoe and Novitch, 2008), our previous study predicted that differential Shh signaling may impart not only the specification of pTH-R and pTH-C, but also the molecular heterogeneity of progenitor cells within the pTH-C domain. To test this hypothesis, we used conditional gene activation or deletion as well as in utero electroporation in mice to increase or decrease Shh activity in thalamic progenitor cells in temporally and spatially restricted manners. We found that altering the level of Shh activity shifts the positional gene expression patterns in both pTH-R and pTH-C. In addition, this shift results in reorganization of postmitotic thalamic nuclei.
Materials and Methods
Care and experimentation on mice were done in accordance with the Institutional Animal Care and Use Committee of the University of Minnesota. Noon of the day on which the vaginal plug was found was counted as embryonic day 0.5 (E0.5). Stages of early embryos were confirmed by morphology (Kaufman, 1992). To generate NesCre/+; R26SmoM2/+ compound heterozygous mice, we crossed heterozygous Nestin-Cre (NesCre/+) transgenic mice (Tronche et al., 1999) with homozygous ROSA26-stop-SmoM2-EYFP (R26SmoM2/SmoM2) mice (Jeong et al., 2004) and identified Cre-positive embryos by PCR. Cre-negative littermates were used as controls. NesCre/+; R26SmoM2/+ mice died by the end of the first postnatal day (P0). For Shh conditional knock-out, we crossed NesCre/+; Shhc/+ mice with Shhc/c mice. Shhc is a conditional allele for Shh (Lewis et al., 2001). PCR was done to identify NesCre allele as well as floxed and wild-type alleles of Shh. Smo conditional knock-out mice were generated by crossing NesCre/+; Smoc/+ mice with Smoc/c mice (Long et al., 2001). Postmitotic fates of Olig2-expressing progenitor cells were analyzed by crossing Olig2Cre/+ mice (Dessaud et al., 2007) with R26-stop-EGFP reporter mice. Olig2 mutant embryos were obtained by crossing between Olig2IRES-EGFP/+ mice (Mukouyama et al., 2006). For analysis of normal gene expression (Fig. 1) and in utero electroporation, timed pregnant CD1 mice (Charles River) were used. For all the other mice, colonies were maintained in C57B/6J background.
Production of Olig3Cre/+ mice.
We produced Olig3Cre/+ mice by homologous recombination in embryonic stem (ES) cells. In short, R1 ES cells were transfected with a targeting vector so that the entire coding region of the Olig3 gene is replaced by a cassette encoding Cre recombinase followed by frt-neo-frt (FNF) sequence (see Fig. 3K). G418-resistant clones were selected and analyzed by Southern blot analysis for homologous recombinants. Two positive clones were microinjected into blastocysts obtained from the mating of C57BL/6NCrl female mice with (C57BL/6J × DBA/2J) F1 male mice (performed at Transgenic Animal Model Core of University of Michigan), and male chimeras were further bred with C57BL/6J females to produce heterozygous Olig3Cre/+ mice. Olig3Cre/+ mice were bred with R26SmoM2/SmoM2 mice to obtain Olig3Cre/+; R26SmoM2/+ mice. To produce conditional Smo knock-out mice, we crossed Olig3Cre/+; Smoc/+ and Smoc/c mice. To monitor the spatial and temporal patterns of Cre-mediated recombination, NesCre/+ or Olig3Cre/+ mice were bred with R26-stop-EYFP reporter mice (Srinivas et al., 2001), and embryos were analyzed for the expression of EYFP and various progenitor cell markers.
In utero electroporation.
In utero electroporation was performed as described previously (Saito and Nakatsuji, 2001). Timed pregnant (E10.5 or E11.5) CD1 females were anesthetized by intraperitoneal injection of pentobarbital sodium (2.0 mg for 30 g dams). Embryos were visualized by illumination of the uterus with fiber optics, and a pulled micropipette (1B120F-6, World Precision Instruments) loaded with DNA was inserted into the third ventricle of each embryo. Approximately 0.5 μl of the DNA solution was injected with Picospritzer III (Parker Instruments). Four pulses of square-wave current were applied (35 V, 30 ms on and 100 ms off), using CUY21EDIT electroporator (NEPAGENE) and paddle- or needle-type electrodes. We used 4 μg/μl pCAG-Olig2 (from Masato Nakafuku, Cincinnati Children's Hospital Medical Center, Cincinnati, OH) plus 0.5 μg/μl pCAG-nuclear EGFP and 2 μg/μl pCAG-SmoM2-EYFP (from Andrew McMahon, Harvard University, Cambridge, MA). After the injection, the dam was sutured and allowed to recover until analysis.
In situ hybridization and immunohistochemistry.
In situ hybridization and immunohistochemistry were done based on Vue et al. (2007). In this study, goat anti-Sox1 antibody from R&D Systems was used at 1:100. cDNAs for BHLHB4, Gli1, Ptc1, Dlx2, Sox14, and Pdlim3 were obtained from Open Biosystems. Irx2 and Spry1/Spry2 cDNAs were obtained from Peter Gruss (Max Planck Society, Munich, Germany) and Gail Martin (University of California, San Francisco, San Francisco, CA), respectively. Sox1-positive cells in the intergeniculate leaflet (IGL) nucleus were counted and compared between Olig3Cre/+; Smoc/c and control embryos. Serial sections from each brain were collected onto 8 slides, and Sox1-positive cells within NPY-expressing IGL nucleus were counted with ImageJ software. Average cell counts per slide were compared between the two genotypes (4 thalami in each group), and Student's t test was performed to test the significance of difference. An SEM was used for presentation of data.
Analysis of cell proliferation in the thalamic ventricular zone of SmoM2-expressing or Shh knock-out brains.
Brains of E11.5 Olig3Cre/+; R26SmoM2/+ and control litter mates, or E12.5 NesCre/+; Shhc/c and control littermates were cryosectioned at 20 μm thickness and then immunostained with anti-phospho histone H3 (anti-PH3) antibody (1:100, Millipore #06-570) to label cells in M-phase. PH3-positive cells situated caudal to the ZLI and within the Olig3-expressing domain in the diencephalon were manually counted, and the numbers were compared between Olig3Cre/+; R26SmoM2/+ or NesCre/+; Shhc/c with their respective controls. Student's t test was performed to test the significance of difference, and an SEM was used for presentation of data.
Anterograde labeling of retinogeniculate axons and their targets in the thalamus.
Approximately 2 μl of cholera toxin B subunit [CTB (5 μg/μl), List Biological Laboratories or Invitrogen] was injected into the right retina of neonatal Olig3Cre/+; R26SmoM2/+ mice and their control littermates. After an overnight survival period, pups were perfused with 4% paraformaldehyde and processed for cryosectioning. Goat anti-CTB antibody (1:100, List Biological Laboratories) was used to detect the retinal ganglion cell axons as well as their targets in the contralateral diencephalon, the dorsal lateral geniculate (dLG) and ventral lateral geniculate (vLG) nuclei as well as IGL. We compared the areas of terminations by measuring the dLG between Cre+ pups and Cre− controls. The border between dLG and IGL was identified by DAPI staining and immunohistochemistry with anti-Sox2. Scanned images were analyzed by ImageJ software for area measurement. We measured the area of every section containing dLG (21–29 sections per brain, 20 μm thickness) and calculated the estimated volume of dLG by summing the values of area × thickness of each section. Student's t test was performed to test the significance of difference, and an SEM was used for presentation of data.
Shh signaling is graded within the thalamic progenitor zone
Within the diencephalon, Shh is expressed in the ZLI and the basal plate, which directly border the thalamus rostrally and ventrally, respectively (Fig. 1A). In this study we asked whether differential Shh signaling is responsible for setting up the expression patterns of the various transcription factors that we described previously (Fig. 1B) and later contributes to the specification of different thalamic nuclei. To determine whether the embryonic thalamus is indeed exposed to graded Shh activity, we analyzed the expression of Gli1 and Ptc1 in mouse embryos from E10.5 to E12.5. Gli1 and Ptc1 are direct target genes of the Shh transduction pathway, and their expression levels reflect the levels of Shh activity in the cell (Agren et al., 2004; Bai et al., 2004). We predicted that if thalamic progenitor cells are differentially patterned by Shh signaling, then both Gli1 and Ptc1 should be expressed in graded manners across pTH-R and pTH-C. Indeed, at both E10.5 and E12.5, Gli1 and Ptc1 were expressed in a high-rostroventral to low-caudodorsal gradient in wild-type brains (Fig. 1D–F,H–J; thalamus is defined by Shh and Olig3 in C and G); both genes were expressed at the highest levels in pTH-R, and their expression gradually tapered off caudally and dorsally away from the ZLI and the basal plate. With this finding, we hypothesized that the highest level of Shh signaling defines pTH-R, and progressively lower levels define different positional identity of progenitor cells within pTH-C depending on the distance from the ZLI and the basal plate. It was also evident that expression of Gli1 and Ptc1 is found only in progenitor cells of the thalamic ventricular zone (Fig. 1E,F,I,J) and not in postmitotic cells in the mantle zone, demonstrating that Shh signaling, whose direct output is gene activation mediated by Gli transcription factors, is restricted to neural progenitor cells.
Ectopic enhancement of Shh signaling strongly induces Gli1 and Ptc1 in the entire thalamic progenitor domains
Based on the above findings, we examined whether increased Shh activity in thalamic progenitor cells can cause a rostroventral shift in their positional identity. More specifically, we predicted that if caudodorsal progenitors are exposed to a higher level of Shh activity than normal, these progenitor cells would express rostroventral molecular markers rather than caudodorsal ones. To test this possibility, we crossed Nestin-Cre (NesCre/+) transgenic mice (Tronche et al., 1999) with ROSA26-stop-SmoM2-EYFP (R26SmoM2/SmoM2) mice (Jeong et al., 2004) to express the fusion protein of SmoM2-EYFP broadly in neural progenitor cells, including those in the thalamus. SmoM2 carries a missense mutation in the Smoothened (Smo) gene, which encodes a transmembrane Shh effector. SmoM2-expressing cells undergo cell-autonomous activation of Shh signaling independent of ligand binding (Xie et al., 1998). Thus, we expected that mis-expression of SmoM2 would cause elevated Shh signaling in all thalamic progenitor cells. By comparing NesCre/+; R26SmoM2/+ and Cre-negative, R26SmoM2/+ controls and wild-type embryos, we found that Cre-negative control and wild-type embryos show the same gene expression and morphology at all stages. In contrast, brains of NesCre/+; R26SmoM2/+ mice were larger in size than Cre-negative littermates, especially in the dorsal telencephalon.
Analysis of SmoM2-EYFP expression as detected by anti-EGFP antibody showed that in NesCre/+; R26SmoM2/+ embryos, Cre-mediated recombination starts to occur in the diencephalon by E9.5 (Fig. 2A–C,F–H). At this stage, Shh expression in the ZLI has just started (Fig. 2A,F), and Gli1 and Ptc1 were not clearly detectable in control embryos (Fig. 2D,E). However, in NesCre/+; R26SmoM2/+ embryos, Gli1 and Ptc1 were already expressed ectopically in thalamic progenitor cells near the ZLI (Fig. 2I,J). Subsequently, recombination progresses through the thalamus into the more caudodorsal region, and by E10.0, SmoM2-EYFP and Gli1 expression covered the entire thalamus labeled by Olig3 (Fig. 2K,L,M,N). By E10.5, most embryos we analyzed showed broad expression of SmoM2-EYFP and Gli1 in the entire diencephalon except the most dorsal portion (Fig. 2P,Q). Analysis of recombination using R26-stop-EYFP reporter mice showed a similar rostroventral to caudodorsal progression of EYFP expression (supplemental Fig. S1A–D, available at www.jneurosci.org as supplemental material). These results demonstrate that in NesCre/+; R26SmoM2/+ embryos, Shh signaling is elevated in a rostroventral to caudodorsal direction, and by E10.0, there is a uniform enhancement in signaling within the entire thalamic ventricular zone.
Increased Shh activity induces rostroventral markers in thalamic progenitor cells
We then asked whether the increased Shh signaling in NesCre/+; R26SmoM2/+ embryos alters the expression profile of transcription factors in the thalamus. In NesCre/+; R26SmoM2/+ embryos, expression of Shh in the ZLI was not affected (Fig. 3A,B,E,F). Nonetheless, expression of Nkx2.2 and Mash1, two markers for the rostral progenitor domain, pTH-R, was expanded (Fig. 3C,D,G,H). This expansion results in more than two-fold increase in the size of pTH-R and disrupts the normally distinct gene expression boundary between pTH-R and pTH-C. The once distinct boundary is now replaced by a “mixed zone (pTH-R/C)” in which some cells expressing only pTH-R markers (Nkx2.2 and Mash1) are intermingled with cells expressing only pTH-C markers (Ngn2 and Olig2) (Fig. 3G,H). Such a zone of mixed pTH-R and pTH-C identity was not seen in control embryos (Fig. 3C,D). Although expression of SmoM2-EYFP appeared ubiquitous and homogeneous throughout the thalamic ventricular zone of NesCre/+; R26SmoM2/+ embryos at E11.5 (supplemental Fig. S1E–H, available at www.jneurosci.org as supplemental material), it is possible that only a small number of cells in this zone expressed ectopic SmoM2 early enough to take on the pTH-R fate, whereas the remaining cells were irreversibly determined to become pTH-C progenitor cells.
SmoM2 acts cell autonomously to enhance Shh signaling (Xie et al., 1998). Thus, the observed changes in marker expression are likely to be caused by intrinsic changes in thalamic patterning. Nonetheless, since most CNS progenitor cells are forced to express SmoM2 in NesCre/+; R26SmoM2/+ embryos, it is possible that some secondary, indirect effects from outside the thalamus may have contributed to the expansion of pTH-R and/or the formation of the “mixed zone.” To investigate this possibility, we took two additional approaches to further restrict the SmoM2 expression spatially and temporally within the thalamus.
First, we performed in utero electroporation of SmoM2 plasmid (pCAG-SmoM2-EYFP) at E11.5 and analyzed the embryos a day later. We found that similar to NesCre/+; R26SmoM2/+ embryos, both Nkx2.2 (Fig. 3I,J) and Mash1 (data not shown) are expanded and are ectopically expressed more caudally on the electroporated side compared with the control side. As a second approach, we generated Olig3Cre mice (Fig. 3K), so that we can express Cre recombinase in pTH-R, pTH-C, and the ZLI, but not in surrounding regions of the forebrain such as the prethalamus, the habenula, the pretectum, the diencephalic basal plate or the telencephalon. Cre-mediated recombination as assessed with R26-stop-EYFP reporter mice (Fig. 3L–O) and ROSA26-stop-SmoM2-EYFP mice (supplemental Fig. S1I,J, available at www.jneurosci.org as supplemental material) faithfully followed the expression of the endogenous Olig3 gene. Thus, Olig3Cre mice are useful in thalamus-specific gene manipulation. We found that brains of Olig3Cre/+; R26SmoM2/+ are morphologically indistinguishable from control littermates, and some of these pups survive postnatally. Neuronal differentiation as determined by the expression of TuJ1, a neuronal marker, was also not affected in these embryos (supplemental Fig. S2A,B, available at www.jneurosci.org as supplemental material). Additionally, analysis of cell proliferation in Olig3Cre/+; R26SmoM2/+ embryos at E11.5 revealed that BrdU-incorporation and phospho-histone H3 (PH3), an M-phase marker, also appeared normal (supplemental Fig. S2A,B, available at www.jneurosci.org as supplemental material). Quantification of PH3-positive cells in Olig3-expressing thalamic progenitor cells showed that there was no significant difference in number between Olig3Cre/+; R26SmoM2/+ and control embryos (4440 ± 206.0, n = 2 for Olig3Cre/+; R26SmoM2/+, 4906 ± 449.5, n = 2 for Cre− controls; p = 0.45).
Analysis of progenitor markers in Olig3Cre/+; R26SmoM2/+ embryos showed similar changes in gene expression compared with NesCre/+; R26SmoM2/+ embryos; pTH-R was expanded, and a “mixed zone (pTH-R/C)” was formed between pTH-C and the expanded pTH-R (Fig. 3P–S). The results of these two additional approaches, in vivo electroporation and the use of Olig3Cre mice, further reinforce our conclusion that it is the intrinsic increase in Shh signaling that causes the caudodorsal shift of pTH-R markers in thalamic progenitor cells.
We next examined how the pTH-C domain was affected by the increase in Shh signaling. First, in NesCre/+; R26SmoM2/+ embryos, Olig3 was expanded caudodorsally into the pretectal/habenular region at E12.5 (Fig. 4A,G). At this stage, Gbx2 and RORα, markers of the postmitotic thalamus derived from the pTH-C domain (Vue et al., 2007), were also expanded accordingly (supplemental Fig. S3A–C,F–H, available at www.jneurosci.org as supplemental material). In contrast, Pax7, which is normally expressed in the pretectum and habenula, but not in thalamic progenitor cells, was not detected in NesCre/+; R26SmoM2/+ embryos (Fig. 4E,K). Moreover, expression of Mash1 in the caudal pretectum was replaced by Ngn2, which normally spans from pTH-C only into the rostral pretectum (Fig. 4F,L). Irx2 and BHLHB4, early postmitotic markers of the pretectum, were also reduced at E12.5 (supplemental Fig. S3D,E,I,J, available at www.jneurosci.org as supplemental material). These results show that a high Shh signal in the caudal diencephalon induces the thalamic progenitor domain, pTH-C, at the expense of the pretectum and habenula. Olig2, which is normally expressed in a rostral–ventral to caudal–dorsal gradient within pTH-C, was homogeneously expressed in this expanded pTH-C domain of NesCre/+; R26SmoM2/+ embryos (Fig. 4B,H). Another marker gene, Pdlim3, which is normally expressed in pTH-R as well as the rostroventral part of pTH-C (Fig. 4C) (Gray et al., 2004), also expanded (Fig. 4C,I). These results demonstrate that enhanced Shh signaling in the entire diencephalon transformed it into the rostral–ventral thalamus, which is composed of the pTH-R domain and the rostral–ventral part of the pTH-C domain (summarized in Fig. 9A,B1,B2,C1,C2).
Surprisingly, the expression of Dbx1, which is normally expressed in the caudodorsal part of pTH-C as well as in the pretectum and habenula, was not significantly reduced in NesCre/+; R26SmoM2/+ embryos (Fig. 4D,J), suggesting that Dbx1 is positively regulated by some unknown mechanisms operating in the caudodorsal diencephalon. Expression of Dlx2, a prethalamic marker, was not ectopically induced in the thalamic progenitor cells in NesCre/+; R26SmoM2/+ embryos (data not shown), consistent with a previous study showing that the thalamus and the prethalamus are distinctly prepatterned and express different sets of genes in response to Shh signaling (Kiecker and Lumsden, 2004).
To ensure that the above changes were also not the product of indirect effects, we again analyzed embryos electroporated with the pCAG-SmoM2-EYFP plasmid, as well as Olig3Cre/+; R26SmoM2/+ embryos. In electroporated embryos where an isolated population of progenitor cells in the pretectum were transfected, Olig2 was ectopically induced (Fig. 4S,T). In Olig3Cre/+; R26SmoM2/+ embryos, although the pretectum and the habenula are spared and the extent of Olig3 expression was similar to controls (Fig. 4M,P), Ptc1 and Olig2 were homogeneously elevated within the entire pTH-C domain (Fig. 4N,O,Q,R), demonstrating that thalamus-specific elevation of Shh signaling intrinsically induced the rostral pTH-C markers.
Expression of rostral postmitotic thalamic markers is expanded in SmoM2-expressing embryos
Does increased Shh signaling in thalamic progenitor cells lead to reorganization of postmitotic thalamus? To address this question, we examined the expression of postmitotic markers that are differentially expressed within nuclei of the late embryonic thalamus (Fig. 5).
First, we analyzed nuclei that are derived from the pTH-R domain. We previously found that progenitor cells in pTH-R generate part of the vLG as well as the IGL, two nuclei that do not contain neurons projecting to the neocortex (Vue et al., 2007). Consistent with this fate analysis and the expanded pTH-R markers in NesCre/+; R26SmoM2/+ as well as Olig3Cre/+; R26SmoM2/+ embryos, vLG/IGL was expanded caudodorsally along the lateral surface of the diencephalon (Fig. 5A,G for NesCre/+; R26SmoM2/+, and data not shown for Olig3Cre/+; R26SmoM2/+). These expanded nuclei are marked by expression of Nkx2.2, Sox14 (Fig. 5A–C,G–I), and Sox1 (data not shown). The expanded Nkx2.2 and Sox14 expression is consistent with previous studies using chick and mouse embryos (Hashimoto-Torii et al., 2003; Kiecker and Lumsden, 2004; Vieira et al., 2005) and establishes that IGL and the lateral part of vLG are specified by high Shh activity in pTH-R. Nkx2.2-positive cells were also found in a more caudolateral region that is away from the large cluster of expanded vLG/IGL cells (Fig. 5H, arrowheads). These cells did not express Sox2 but intermingled with Sox2-expressing cells (Fig. 5, compare G, H, J). This intermingling of two molecularly different cell types reflects our earlier observation that two completely separate progenitor cell populations, one with pTH-R identity and the other with pTH-C identity, are mixed in the SmoM2-expressing, thalamic ventricular zone (Fig. 3G,H).
Next, we tested whether markers of thalamic nuclei that are derived from the pTH-C domain are also shifted in SmoM2-expressing embryos. In control embryos, Sox2 is expressed in the rostroventral part of the pTH-C-derived thalamic mantle zone (Fig. 5D,M,O) (Vue et al., 2007). In E16.5 NesCre/+; R26SmoM2/+ brains, Sox2 expression was expanded caudally and dorsally, occupying essentially the entire diencephalic mantle zone (Fig. 5D,J). Similarly, another rostral nucleus marker, RORα, which is expressed more specifically in ventral posterior nucleus (VP) and dLG (Nakagawa and O'Leary D, 2003), was expanded caudally and dorsally in NesCre/+; R26SmoM2/+ brains compared with controls (Fig. 5E,K). To determine whether the expansion of these rostral nucleus markers is accompanied by the repression of caudal nucleus markers, we analyzed the expression of Gbx2. Although Gbx2 is initially expressed in the entire postmitotic thalamus (supplemental Fig. S3A,B,F,G, available at www.jneurosci.org as supplemental material), its expression later becomes restricted to medial and caudodorsal nuclei and is nearly complementary to that of Sox2 and RORα (Fig. 5F,N,Q). In NesCre/+; R26SmoM2/+ brains, Gbx2 expression was dramatically reduced, particularly in regions where both Sox2 and RORα are ectopically induced (Fig. 5L).
To better demonstrate the local effects of enhanced Shh signal on postmitotic thalamic development, we again analyzed SmoM2-electroporated embryos and Olig3Cre/+; R26SmoM2/+ embryos. When caudodorsal thalamus was electroporated at E11.5 and analyzed at E17.5, Sox2 was highly induced in the cohort of cells on the transfected side (Fig. 5M). Gbx2 was specifically repressed in these Sox2-expressing cells, but its expression on the control side was unaffected (Fig. 5M,N). In Olig3Cre/+; R26SmoM2/+ brains at E18.5, the overall morphology of the thalamic mantle zone was normal. However, Sox2 was ubiquitously expressed in the entire thalamus, not just in rostroventral nuclei as seen in the control (Fig. 5O,R). RORα was also significantly expanded caudodorsally in the lateral part of the thalamus (Fig. 5P,S). Furthermore, Gbx2 expression is almost completely absent in Olig3Cre/+; R26SmoM2/+ brains (Fig. 5Q,T). These data demonstrate that in Olig3Cre/+; R26SmoM2/+ embryos, the entire postmitotic thalamus is dominated by rostroventral identity.
Although the expansion of RORα implicates that rostroventrally located nuclei, VP, and dLG are increased in size in Olig3Cre/+; R26SmoM2/+ embryos, we used criteria other than molecular markers that would definitively prove that individual nuclei are altered. For this purpose, we used anatomical criteria to define thalamic nuclei. Specifically, we injected anterograde axon tracer CTB to the retina of the neonatal Olig3Cre/+; R26SmoM2/+ mice to label retinogeniculate axons in the dLG nucleus. Because Olig3 is not expressed in the retina or along the retinogeniculate pathway to the thalamus, Olig3Cre/+; R26SmoM2/+ mice provide a particularly useful model to evaluate the specific effects of enhanced Shh signal in thalamic progenitor cells on the size of postnatal thalamic nuclei. We found that the volume of the dLG nucleus as estimated by the CTB-labeled region in the thalamus, was increased by ∼3.6 times in Cre+ pups (0.086 ± 0.008 mm3 in Cre+ pups, n = 3; 0.024 ± 0.003 mm3 in Cre− pups, n = 2; p = 0.0091) (Fig. 5U–X).
In summary, we demonstrated that increasing Shh signaling in CNS progenitor cells results in expansion of rostral–ventral thalamic progenitor pools, pTH-R, and rostroventral pTH-C, at the expense of more caudal–dorsal thalamus as well as the pretectum and habenula (summarized in Fig. 9B1,B2,C1,C2). When the manipulation of the Shh signal was restricted to thalamic progenitor cells, pTH-R and rostroventral pTH-C domain still expanded and caudal–dorsal pTH-C was shrunken. These results strongly indicate that the high Shh signal biases thalamic progenitor cells toward more rostral–ventral fates in a cell-intrinsic manner. This conclusion is consistent with our observation that the number of PH3+ cells in the entire Olig3-expressing domain was similar between Olig3Cre/+; R26SmoM2/+ and control embryos (supplemental Fig. S2, available at www.jneurosci.org as supplemental material). Compatible with changes in progenitor identity, postmitotic thalamic nuclei are expanded or shrunken based on their putative locations of origin (summarized in Fig. 9B3,C3).
Because Gli1 and Ptc1 are not ectopically expressed in the mantle zone of SmoM2-expressing brains, the altered organization of the postmitotic thalamus is unlikely to be due to the direct activation of Shh signaling in postmitotic cells. Instead, we conclude that shifted identity of thalamic progenitor cells resulted in corresponding alteration of the identity of postmitotic thalamic nuclei.
Overexpression of the bHLH transcription factor Olig2 partially mimics the effect of SmoM2 expression in the thalamus
If Shh signal directly controls the identity of thalamic progenitor cells and indirectly influences the specification of postmitotic thalamic nuclei, it is likely that transcription factors that are regulated by Shh signal in progenitor cells translate their positional information onto their postmitotic progeny. Olig2 is normally expressed in the rostroventral portion of pTH-C, and is induced in progenitor cells that ectopically express SmoM2. These SmoM2-expressing embryos also showed induction of Sox2 and RORα caudally and dorsally throughout the thalamic mantle zone. Thus, it is possible that Olig2 functions to specify rostroventrally located thalamic nuclei that express Sox2 and RORα. To test this possibility, we first performed genetic lineage tracing of Olig2-expressing progenitor cells (supplemental Fig. S4A–C, available at www.jneurosci.org as supplemental material). We crossed Olig2Cre/+ mice (Dessaud et al., 2007) with ROSA26-stop-EGFP reporter mice, and analyzed E14.5 compound heterozygous embryos. At E11.5, GFP was expressed in the rostroventral part of pTH-C, as well as pTH-R (supplemental Fig. S4A, available at www.jneurosci.org as supplemental material). Expression of GFP in pTH-R suggests that there is transient expression of Olig2 in this progenitor domain before E10.5, which is similar to the observation in ventral spinal cord (Dessaud et al., 2007). At E14.5, EGFP was expressed in the rostroventral portion of the postmitotic thalamus, which largely overlapped with Sox2 expression (supplemental Fig. S4B,C, available at www.jneurosci.org as supplemental material). VP and the ventral part of medial geniculate (MGv) nuclei were two nuclei that had particularly strong EGFP expression. In contrast, EGFP was not found in more caudodorsal nuclei. Thus, Olig2-expressing thalamic progenitor cells give rise to rostroventral part of the postmitotic thalamus, consistent with the possibility that Olig2 plays a role in the specification of rostroventral nuclei such as VP and MGv.
We then tested whether Olig2 is necessary and/or sufficient for the specification of these nuclei. When we ectopically expressed Olig2 by in utero electroporation at E10.5 in the caudodorsal diencephalon, Sox2 was induced in the mantle zone only on the transfected side at E13.5 and E14.5 (supplemental Fig. S4E,F, available at www.jneurosci.org as supplemental material), which partially mimicked the effect of expressing SmoM2. Cell proliferation and neuronal differentiation did not seem to be affected in Olig2-transfected cells, and overexpressing EGFP alone did not induce Sox2 expression (data not shown). These results suggest that Olig2 mediates the Shh signal and confers at least part of the rostroventral identity to the postmitotic thalamus. In contrast, Olig2 null mice did not show altered expression of Sox2 or progenitor markers such as Mash1, Ngn2, and Dbx1 (data not shown). Thus, Olig2 is not necessary but is sufficient to induce at least some molecular features of the rostroventral thalamic nuclei.
Attenuation of Shh signal in thalamic progenitor cells reduces the expression of rostroventral markers and increases caudodorsal markers
Previous studies on chick embryos showed that ectopic expression of Ptc1Δloop2, a dominant-negative form of the Shh receptor, Ptc1, reduced the expression of Nkx2.2 and Gbx2, and increased the expression of Pax6, which is normally expressed highly in the caudal diencephalon (Kiecker and Lumsden, 2004). It was also reported that Gli2/3 double knock-out mice showed reduced expression of Gbx2, whereas Gli1 mutant mice showed reduced Sox14 expression (Hashimoto-Torii et al., 2003). These studies demonstrated the requirement of Shh signaling in establishing the identity of the thalamus as a whole. However, it is still unknown whether Shh is required for the patterned expression of progenitor markers in pTH-R and pTH-C, or for the normal organization of postmitotic thalamic nuclei. Shh null mice show pronounced reduction in size of the diencephalon by E9.0, which reflects the early role of Shh in cell proliferation and survival (Ishibashi and McMahon, 2002). Therefore, we analyzed conditional Shh knock-out mice by using the NesCre allele to avoid such early effects. Conditional Shh mutant mice were smaller than control littermates, but they were postnatally viable and have relatively normal gross morphology of the brain (data not shown). The smaller brain size is reflected in part by a significant decrease in the number of PH3-positive progenitor cells in the thalamic ventricular zone of conditional Shh knock-out (NesCre/+; Shhc/c) embryos (3437 ± 146.0, n = 2 for NesCre/+; Shhc/c, 6605 ± 472.0, n = 2 for Cre− controls; p = 0.024) (supplemental Fig. S2C,D, available at www.jneurosci.org as supplemental material).
In NesCre/+; Shhc/c embryos, Shh expression in the forebrain was lost by E10.5 (data not shown; Machold et al., 2003). Consistent with the loss of Shh in the ZLI and the basal plate, Ptc1 (data not shown) and Gli1 mRNA was undetectable at E10.5 and E12.5 in NesCre/+; Shhc/c embryos (Fig. 6B,G and not shown). Furthermore, markers for the pTH-R domain were not induced. At E12.5, we observed a complete loss of Nkx2.2 (Fig. 6C,H) and Mash1 (Fig. 6D,I) within the thalamic ventricular zone, although their expression in the prethalamic progenitor domain remained at very low levels (Fig. 6H,I). Thus, NesCre/+ mice are useful in deleting the Shh gene in the brain when Shh signal is still critically required for the induction and/or maintenance of the expression of pTH-R markers. In addition to Nkx2.2 and Mash1, expression of Olig2, a rostroventral pTH-C marker, was also lost in pTH-C of NesCre/+; Shhc/c embryos, although it was weakly detected in the prethalamus (Fig. 6C,H, arrowhead for the prethalamus). In contrast, markers that are expressed throughout the pTH-C domain, such as Olig3 (Fig. 6A,F) and Ngn2 (Fig. 6D,I) were still expressed in NesCre/+; Shhc/c embryos at E12.5 and earlier stages, although the size and intensity of the expression of Olig3 seemed to be reduced. Dbx1, which is normally restricted to the caudodorsal pTH-C, was expanded rostroventrally in NesCre/+; Shhc/c thalamus (Fig. 6E,J). A pretectal marker, Pax7, was still expressed (data not shown). Together, these findings demonstrate that Shh activity is required for progenitor cell identity in pTH-R and rostroventral pTH-C and for suppressing the caudodorsal pTH-C identity (summarized in Fig. 9B1,B2,D1,D2).
Next, to more directly test the intrinsic requirement of Shh signaling in the thalamus, we conditionally knocked out the gene encoding Smo, a transmembrane effector of Shh signaling. Two Cre drivers, NesCre/+ and Olig3Cre/+, were used. Both NesCre/+; Smoc/c and Olig3Cre/+; Smoc/c mice showed normal gross morphology of the brain and survived postnatally. We found that NesCre/+; Smoc/c embryos show lack of Ptc1 expression in the thalamus at E12.5 (Fig. 7B,G) despite the normal expression of Shh in the ZLI (Fig. 7A,F). In these embryos, we also found that Olig3 expression was reduced in the thalamic ventricular zone (Fig. 7A,F), and Mash1 and Olig2 were undetectable in pTH-R and pTH-C, respectively (Fig. 7C,D,H,I). Furthermore, Dbx1 expression was expanded rostroventrally (Fig. 7E,J). However, unlike in NesCre/+; Shhc/c embryos, Nkx2.2 was still detectable though not as robustly as in the control (Fig. 7C,H).
Olig3Cre/+; Smoc/c embryos at E12.5 showed normal expression of Shh in the ZLI (Fig. 7K,P) and reduction of Ptc1 expression in the thalamus but not in the prethalamus (Fig. 7L,Q), compatible with the restricted deletion of Smo and reduced Shh signal specific to the thalamus. The reduction of Ptc1 expression in the thalamus was less significant than in NesCre/+; Smoc/c embryos, where Ptc1 was undetectable in the thalamus and the prethalamus (Fig. 7G). Nevertheless, Olig3Cre/+; Smoc/c embryos showed reduction of Mash1-expressing cells in the thalamus (Fig. 7M,N,R,S). In addition, the decreased number of Mash1-expressing cells were intermingled with Ngn2-expressing cells, giving rise to the mixed zone (pTH-R/C) instead of pure pTH-R as we found in the control (Fig. 7N,S). Although Nkx2.2 and Olig2 were not consistently reduced in expression, Dbx1 expression was grossly expanded rostroventrally (Fig. 7O,T), as in the NesCre/+; Shhc/c and NesCre/+; Smoc/c embryos. Together, these results indicate that the Shh signal is intrinsically required for the induction and/or maintenance of molecular characteristics of the pTH-R domain and for the suppression of caudodorsal part of the pTH-C domain.
The dramatic changes in the progenitor cell identity with conditional deletion of the Shh or Smo gene prompted us to examine which populations of postmitotic thalamus are affected in the absence of Shh or Smo. In E17.5 NesCre/+; Shhc/c embryos, as expected from the loss of pTH-R markers at E12.5 (Fig. 6C,D,H,I), Nkx2.2 (Fig. 8A,C,F,H) and Sox14 (data not shown) in vLG/IGL were both dramatically reduced, confirming the requirement of Shh signaling for the vLG/IGL lineage. Similar changes were found in NesCre/+; Smoc/c embryos (Fig. 8K–N). We next analyzed the pTH-C derivative. Expression of Sox2 (Fig. 8B,G) and RORα (Fig. 8D,I) was profoundly reduced in NesCre/+; Shhc/c embryos. In contrast, Gbx2 expanded and occupied almost the entire thalamic mantle zone (Fig. 8E,J). Again, similar changes were observed in NesCre/+; Smoc/c embryos (Fig. 8K,M and data not shown). In Olig3Cre/+; Smoc/c mice, IGL, as marked by Sox1 and neuropeptide Y (NPY) expression, was dramatically reduced in size and cell number. Sox1-positive cells in this nucleus was reduced by ∼70% (Olig3Cre/+; Smoc/+: 337.5 ± 13.80 cells/slide, Olig3Cre/+; Smoc/c: 103 ± 24.93 cells/slide, n = 4; p = 0.0002). This reduction likely reflects the reduction of Mash1-positive cells in the pTH-R at E12.5 (Fig. 7M,N,R,S).
In summary, decreasing Shh signaling resulted in reduction of rostral–ventral progenitor pools, pTH-R, and the rostroventral part of the pTH-C domain, and expansion of caudodorsal progenitor pools (summarized in Fig. 9B1,B2,D1,D2). This change in thalamic progenitor cell identity corresponds to alterations in postmitotic populations (Fig. 9B3,D3). Although conditional deletion of Shh or Smo with two different Cre lines caused different degrees of reduction in Shh signal, we observed consistent changes in the progenitor cell populations as well as their postmitotic derivatives. These observations confirm the cell-intrinsic requirement of Shh signal in the specification of rostral–ventral progenitor cells in the thalamus.
Relationship between Shh and Fgf signaling pathways in thalamic patterning
A recent study by Kataoka and Shimogori (2008) reported that Fgf8 is expressed in the dorsal midline of the caudal diencephalon and that this expression domain continues to a region immediately rostral to the dorsal tip of the ZLI, suggesting that Fgf signaling may have a role in thalamic patterning. Focal in vivo electroporation of Fgf8 near the ZLI increased the size of pTH-R and its derivative, “Rim domain,” although overexpression of Fgf8 did not change the expression pattern of Shh or Ptc1 (Kataoka and Shimogori, 2008). These results suggest that Fgfs may play a patterning role independent of Shh signaling in the thalamus by regulating gene expression in pTH-R progenitor cells. Alternatively, Fgf8 may be downstream of the Shh pathway and mediates part of the roles that Shh plays. In fact, Shh induces the expression of Fgf4 in early limb development (Laufer et al., 1994), and complete Shh null embryos show reduced expression of Fgf15 in mouse diencephalon at E8.5 (Ishibashi and McMahon, 2002). To assess the role of Shh in Fgf8 expression and Fgf signal in the thalamus, we analyzed the expression of Fgf8 and its downstream targets, Spry1 and Spry2 (Liu et al., 2003) in NesCre/+; R26SmoM2/+ and NesCre/+; Shhc/c embryos (supplemental Fig. S4, available at www.jneurosci.org as supplemental material). We found that in both control and NesCre/+; R26SmoM2/+ embryos, Fgf8 is expressed in the dorsal midline as well as in the region immediately rostral to the dorsal part of the ZLI (supplemental Fig. S4B,F, available at www.jneurosci.org as supplemental material). Expression of Spry1 and Spry2 was also similar in the control and SmoM2-expressing embryos (supplemental Fig. S4C,D,G,H, available at www.jneurosci.org as supplemental material). In addition, NesCre/+; Shhc/c embryos at E12.5 showed no significant reduction of expression of Fgf8, Spry1, or Spry2 (supplemental Fig. S4J–L, available at www.jneurosci.org as supplemental material) near the ZLI. These results indicate that the changes to the pTH-R domain observed in NesCre/+; R26SmoM2/+ and NesCre/+; Shhc/c embryos are not due to alterations in Fgf8 expression or Fgf signal.
In this study, we used two lines of mice expressing Cre recombinase to conditionally increase or decrease Shh signaling in the thalamus, and performed in vivo electroporation to elevate the signal. When Shh signaling was globally increased using the NesCre allele, the entire caudal alar diencephalon, but not the surrounding brain regions including the midbrain, took on the molecular properties of the thalamus, which is consistent with previous studies using chick embryos (Vieira et al., 2005). Our current study further demonstrates that the differential Shh signal plays a crucial role not only in the specification of the most rostroventral progenitor domain of the thalamus (pTH-R) and its postmitotic derivatives, but also in the patterning within the caudodorsal thalamic progenitor domain (pTH-C) and the formation of the cortex-projecting thalamic nuclei (summarized in Fig. 9).
A high Shh activity positively regulates the identity of the pTH-R domain and its postmitotic derivatives in the thalamus
Previous studies revealed that a high Shh activity positively regulates the expression of Nkx2.2 and Sox14 in chick and mouse thalamus (Hashimoto-Torii et al., 2003; Kiecker and Lumsden, 2004; Vieira et al., 2005). We find that in the mouse, both Nkx2.2 and Sox14 are expressed in the lineage of the rostral progenitor domain, pTH-R. In our present study, ectopic expression of SmoM2 resulted in increased Shh activity in thalamic progenitor cells as shown by the expression of the target genes Gli1 and Ptc1, and expansion of pTH-R markers. Postmitotic populations derived from pTH-R progenitor cells also increased. Conversely, conditionally deleting Shh using the NesCre allele completely abolished the expression of Nkx2.2 and Mash1 as well as the postmitotic markers of the pTH-R derivative. Deleting the Smo gene with either NesCre or Olig3Cre allele also reduced expression of pTH-R markers in progenitor cells, and resulted in corresponding decrease in its postmitotic derivative, further confirming the cell-autonomous requirement of Shh signaling in this cell lineage. Interestingly, even when SmoM2 is ectopically expressed in the entire thalamic progenitor domains either with NesCre or Olig3Cre allele, only part of the thalamus became pTH-R; the caudodorsal part still expressed pTH-C markers (Fig. 3G,H,R,S). Because the Cre-mediated recombination seems to progress in a rostroventral to caudodorsal direction in these Cre mice, it is possible that the rostroventral part starts to express SmoM2 earlier than caudodorsal part. If SmoM2 can induce ectopic expression of pTH-R markers only before a certain stage of development, the differential onset of SmoM2 expression across the thalamic ventricular zone might explain the partial transformation of pTH-C into pTH-R. In addition, caudodorsally located thalamic progenitor cells may be less sensitive to enhanced Shh signaling than rostroventral cells. This may explain why in vivo electroporation of SmoM2 at E11.5 caused the ectopic expression of pTH-R markers only in the most rostroventral part of pTH-C (Fig. 3I,J).
However, it is also possible that establishment of pTH-R identity requires not only high Shh signaling, but also activity of other signaling molecules that are expressed in or near the ZLI or the basal plate. An intriguing possibility is Fgf8, which was recently shown to play an independent role in specifying pTH-R and its derivatives (Kataoka and Shimogori, 2008). In this current study, we found that increasing Shh activity does not enhance the expression of Fgf8 or its downstream targets, Spry1 and Spry2, and that sustained expression of Shh is not required for the expression of Fgf8, Spry1, or Spry2 in the diencephalon. Thus, if activation of both Shh and Fgf8 pathways is required for the expression of pTH-R markers, the lack of increased Fgf8 signaling in SmoM2-expressing thalamus may account for the limited expansion of pTH-R domain. However, since Fgf8 is expressed only near the dorsal tip of the ZLI, not in the entire ZLI, it is unclear whether endogenous Fgf8 signaling is required for all the progenitor cells in the pTH-R domain. Further studies are needed to identify the collaboration of signaling pathways responsible for the pTH-R identity.
How does Shh activity control the patterning of the pTH-C domain and the cortex-projecting thalamic nuclei?
Compared with the contribution of Shh signaling to the pTH-R lineage, understanding the role of Shh activity in the caudodorsal progenitor domain, pTH-C, is more challenging. The pTH-C domain contains highly heterogeneous progenitor populations characterized by graded expression of transcription factors such as Olig2 or Dbx1. In addition, the many dozens of distinct, cortex-projecting thalamic nuclei that are generated from this domain are not always identifiable by marker gene expression alone, and the use of cytoarchitectural features to discriminate individual nuclei is also limited. Because of these difficulties, how these nuclei are patterned and specified during embryogenesis is not well understood. We approached this question by first identifying and characterizing the molecular differences within pTH-C, which we found contributes to all the cortex-projecting thalamic nuclei. This analysis indicated that heterogeneity within this progenitor domain may contribute to the specification of multiple thalamic nuclei (Vue et al., 2007). Our current study now shows for the first time that the level of Shh signaling controls the identity of pTH-C progenitor cells, and that this in turn regulates the sizes and identity of postmitotic thalamic nuclei.
We showed that high Shh signaling positively regulates the expression of Olig2 and Pdlim3 in the rostral part of pTH-C. Conversely, conditional deletion of Shh or Smo with the NesCre/+ allele diminishes the expression of Olig2 in pTH-C, while it expands the expression of Dbx1, a marker for caudodorsal pTH-C. In addition, increasing or decreasing Shh signaling altered regional patterns of gene expression in the postmitotic thalamus in a way that was consistent with the changes in progenitor cells. Among the postmitotic markers we analyzed, RORα and Sox2 are expressed more strongly in the nuclei that are derived from the rostroventral part of the pTH-C domain. In SmoM2-expressing embryos, Sox2 and RORα are expanded to occupy the entire rostrocaudal extent of the postmitotic thalamus, consistent with the expanded expression of Olig2 and Pdlim3. Shh and Smo knock-out brains show the opposite overall changes. These results clearly demonstrate that differential Shh activity controls not only the identity of pTH-C progenitor cells but also the spatial organization of the postmitotic thalamic nuclei. When SmoM2 was conditionally expressed using the Olig3Cre allele, the dLG nucleus, which is derived from the rostroventral part of pTH-C, was indeed expanded based on anterograde axon tracing from the retina. This confirms an intrinsic role of Shh signaling in the formation of a specific thalamic nucleus that projects to the cortex.
A recent study by Szabó et al. (2009) used Foxb1-Cre mice to broadly delete Shh in the developing CNS and found grossly abnormal thalamus and aberrant expression of postmitotic markers at E18.5. They also found that markers broadly expressed in thalamic progenitor cells such as Olig3 and Ngn2 were still expressed in the mutant mice. Based on these observations, they argued that Shh is mainly required in postmitotic cells in the thalamus to promote migration and aggregation into nuclei as well as axonal extension to the cortex rather than neuronal fate acquisition. In contrast, our current study strongly indicates that Shh acts primarily on thalamic progenitor cells. First, markers that are normally expressed in graded manners within the pTH-C domain such as Olig2, Pdlim3, and Dbx1 were significantly affected when Shh activity was altered. The observed alteration of the postmitotic thalamus reflected the changes in the expression of these progenitor cell markers. Second, direct target genes of Shh signaling such as Gli1 and Ptc1 were expressed only in thalamic progenitors, even though SmoM2 was mis-expressed in both progenitor and postmitotic cells after Cre-mediated recombination. Gli1 and Ptc1 expression was diminished in progenitor cells when the Shh signal was perturbed. Based on these observations, we conclude that Shh acts primarily on thalamic progenitor cells, and the graded Shh signal is essential to confer thalamic progenitors with position-dependent molecular heterogeneity, which forms a basis for generating a diverse set of postmitotic thalamic nuclei. Although the study by Szabó et al. (2009) and our current study used different Cre lines to delete the Shh gene, both observed the complete absence of the pTH-R marker Nkx2.2. Given the early and broad deletion of Shh in their conditional mutant mice, it is also likely that expression of graded pTH-C markers were altered as well. Thus, we think the different conclusions of the two studies have resulted from different interpretations of potentially similar data. Although we cannot completely rule out the possibility that Shh plays some role in postmitotic differentiation of thalamic neurons such as migration or axonal projection, demonstration of such roles would require different sets of experiments where Shh signal is altered in postmitotic cell-specific, conditional gene targeting in mice, or by in vitro assays that exclude the involvement of progenitor cells.
If Shh signaling regulates the expression of transcription factors in thalamic progenitor cells, then do these transcription factors play a functional role in specifying the formation of the various nuclei along the rostral–caudal and dorsal–ventral axes? We found by overexpression experiments that Olig2 can induce rostroventral thalamic nucleus marker Sox2 in the dorsal–caudal thalamus. However, Olig2 mutant mice show normal expression of Sox2, indicating that other factors are also involved. Further studies are needed to identify such genes and to test the individual roles of each of the other transcription factors.
In conclusion, we believe that this study provides a unique perspective in neural development in several respects. First, we showed how Shh patterns a single developmental unit in the embryonic diencephalon (pTH-C) to generate multiple neuronal groups that are grossly arranged in similar rostrocaudal and dorsoventral locations to those of their precursor cells. This feature makes the developing thalamus a good model system to study how early patterning events contribute to the specification of multiple neuronal fates that share some basic properties such as the projection to the neocortex and use of glutamate as neurotransmitter. Second, connectivity and functions of thalamocortical projections are relatively well understood, especially for those from the principal sensory thalamic nuclei. In this study, by using Olig3Cre mice, we were able to generate postnatally viable mice in which Shh signaling was specifically enhanced or reduced in the thalamus without primarily affecting the development of the cortex or brain regions along the thalamocortical pathway. These mice will allow us to examine at the systems level the neuroanatomical and functional consequences of altering the early patterning of the thalamus.
This work was supported by Minnesota Medical Foundation, Grant-in-Aid Program of University of Minnesota Graduate School, National Institute of Neurological Disorders and Stroke Grant R01 NS049357, and Whitehall Foundation. We thank Aya Taniguchi, Anu Jayabalu, Christina Kazemzadeh, Lauren Bolopue, and Melody Lee for help in gene targeting, maintaining mouse colonies, and genotyping embryos; Yasuhiko Kawakami and members of the Nakagawa and Koyano laboratories for discussion; James Briscoe, Martyn Goulding, Peter Gruss, Gail Martin, Andrew McMahon, and Masato Nakafuku for plasmids; Andras Nagy for R1 ES cells; Jun-ichi Miyazaki for CAG promoter; Tomomi Shimogori and Tetsuichiro Saito for advice on in utero electroporation; Kyuson Yun for NestinCre/+ mice; and Developmental Studies Hybridoma Bank for antibodies. R26SmoM2/SmoM2, Shhc/c, and Smoc/c mice were developed by Andrew McMahon's laboratory and were obtained from The Jackson Laboratory. University of Minnesota Mouse Genetics Facility provided neomycin-resistant fibroblasts for ES cell feeders, and Thom Sanders, Elizabeth Hughes, and Tina Jones and University of Michigan Transgenic Animal Model Core contributed to the production of Olig3Cre chimeras. We thank Paul Letourneau and Steve McLoon for critical reading of early versions of the manuscript and Masato Nakafuku for encouragement and helpful discussion.
- Correspondence should be addressed to Yasushi Nakagawa at the above address.