Isl1 is required for promoting TRN cell identity and repressing non-TRN prethalamic cell fates in the TRN

To explore whether Isl1 is required for prethalamic differentiation, we conditionally inactivated Isl1 expression by crossing an Isl1F/F mouse line, in which the homeodomain (exon 4) is flanked by loxP sites [38], with a Dlx5/6-Cre mouse line that drives Cre-mediated recombination in the ventral forebrain [39]. During embryogenesis, Isl1 is principally expressed in the TRN, a prethalamic derivative (Fig. S1A-C). In Dlx5/6-Cre; Isl1F/F embryos, Isl1 staining was almost completely lost in the prethalamus as well as in other forebrain regions (Fig. S1D-F). We first analyzed the effects of Isl1 loss on prethalamic differentiation by staining for Hes5 and Sox2, which mark proliferative progenitors, as well as for the postmitotic neuronal markers Elavl3/4 (HuC/D). At E12.5, Hes5/Sox2 and HuC/D, which exhibit mutually exclusive expression, were similarly expressed in control (Dlx5/6-Cre; Isl1F/+) and Isl1-deficient embryos (Fig. S2A-D). We next examined early patterning in the conditional mutants by analyzing the distribution of Ascl1, Gsx1/2, Helt, Olig2, and Nkx2.2, which are specifically expressed in prethalamic progenitors. None of these markers were significantly altered in Dlx5/6-Cre; Isl1F/F embryos compared to control littermates (Fig. S2E-J).

Prethalamic progenitors differentiate and populate into heterogeneous nuclear groups, including the TRN, zona incerta (ZI), and ventral lateral geniculate nucleus (vLG). At E14.5 and E16.5, these three major nuclei can be identified by postmitotic regional markers [40, 41]. The TRN consists of a sheet of cells surrounded by the rostral ZI (rZI), vLG and zona limitans intrathalamica (zli) and can be identified by staining for Dlx1 and Meis2 (Fig. 1A2,B2,C2,D2). Dlx1 was also expressed in the rZI, vLG, and zli (Fig. 1A2, B2), while Meis2 was mainly detected in the TRN (Fig. 1C2, D2). Arx, Pax6, and Lhx1/5 were expressed in the rZI and vLG at distinct levels (Fig. 1E2, F2, G2, H2, I2, J2, K2, L2). Arx and Lhx1/5 were also detected in the zli, and Pax6 expression was observed in the rostral half of the zli [40,41,42] (Fig. 1E2, F2, G2, H2, I2, J2, K2, L2). In contrast, the TRN does not express Arx, Pax6, or Lhx1/5 at detectable levels (Fig. 1E2, F2, G2, H2, I2, J2, K2, L2; Fig. S3). In the absence of Isl1, compared with that in controls, Dlx1 was similarly expressed in the rZI, vLG, and zli, but Dlx1 expression was undetectable in the TRN at E14.5 and E16.5 (Fig. 1A4, B4). In addition, Meis2 expression was almost completely absent in the TRN of Dlx5/6-Cre; Isl1F/F embryos (Fig. 1C4, D4). Concomitantly, the loss of Isl1 function led to the ectopic induction of Arx, Pax6 and Lhx1/5 in the TRN (Fig. 1E4, F4, G4, H4, I4, J4, K4, L4; Fig. S3), suggesting the derepression of TRN-depleted genes in the TRN in the absence of Isl1. We further examined the mRNA level of Gad1, which encodes the GABA synthetic enzyme glutamate decarboxylase. Loss of Isl1 did not result in significant changes in Gad1 expression in the prethalamic regions (Fig. 1M4, N4), indicating that the prethalamic complex maintains a GABAergic neurotransmitter phenotype in the absence of Isl1, but the neuronal subtype composition in the complex was altered. Therefore, these observations demonstrate that Isl1 is required for the acquisition of TRN identity and for suppressing non-TRN prethalamic neuronal fates in the TRN.

Fig. 1

TRN neuronal identity is dependent on Isl1 function. (Top) Schematic showing the wild-type anatomy of the prethalamus. (A-N) In situ hybridization of prethalamic markers on control (Dlx5/6-Cre; Isl1F/+) and mutant (Dlx5/6-Cre; Isl1F/F) coronal sections. The red boxed areas are magnified as indicated. The patterns of gene expression in the prethalamus at E14.5 were similar to those at E16.5. (A-B) Dlx1 was normally expressed in all prethalamic nuclei, including the rZI, TRN, and vLG, as well as the central diencephalic organizer zli. Dlx1 expression in the TRN of control embryos was high (A2, B2), while Isl1-deficient embryos lost Dlx1 expression specifically in the TRN but not in the vLG, rZI or zli (A4, B4). (C-D) Within the prethalamus of control embryos, Meis2 expression was limited to the TRN (C2, D2). Loss of Isl1 resulted in the abrogation of Meis2 expression (C4, D4). (E-L) Arx, Pax6, and Lhx1 were normally expressed in the rZI, zli, and rostroventral parts of the vLG but were not detectable in the TRN. Lhx5 is also weakly expressed in the rZI, zli, and vLG but not in the TRN. In the absence of Isl1 function, the expression of these genes was largely unaffected in the rZI, zli, and vLG but was ectopically induced in the TRN. (M, N) A GABAergic neuronal marker Gad1 was similarly expressed in control and Isl1 mutants. (O) Quantification of in situ hybridization. The pixel intensity values of TRN or vLG were grouped and averaged from three different sections per embryo. (n = 5 embryos; Student’s t-test; **p < 0.01, ***p < 0.001, ****p < 0.0001) (P) A schematic diagram indicating that Dlx1+Meis2+ TRN cells lose their identity and instead acquire molecular features of the remainder of the prethalamus. Cx, cortex; Hyp, hypothalamus; Pt, prethalamus; rZI, rostral zona incerta; Th, thalamus; TRN, thalamic reticular nucleus; vLG, ventral lateral geniculate nucleus; zli, zona limitans intrathalamica. Scale bar, 200 μm

Isl1 is required for normal TCA navigation in the TRN

As the TRN is a diencephalic contingent that TCAs first encounter and cross, we considered potentially important factors of the TRN in normal TCA projection. Therefore, we determined the effect of Isl1 deletion on the progression and navigation of thalamic axons as they cross the TRN. To specifically label TCAs in the developing forebrain, we crossed the Dlx5/6-Cre; Isl1F allele with the Gbx2GFP allele, which expresses GFP from the Gbx2 locus [43, 44]. We examined the distribution of GFP+ TCA in coronal and sagittal sections through the forebrain in control (Gbx2GFP;Dlx5/6-Cre; Isl1F/+) and mutant (Gbx2GFP;Dlx5/6-Cre; Isl1F/F) embryos. At E14.5 ~ E16.5, in control embryos, GFP+ thalamic axons exiting the thalamus were found to extend rostrally through the prethalamus in smooth and parallel arrays of fascicles (Fig. 2A4, A5, A6, A13 and Fig. 2B1, B3, B5). In contrast, in Isl1 mutant embryos, many thalamic axons fasciculated abnormally to form much denser, thicker and more disordered bundles in the prethalamus (Fig. 2A10, A11, A12, A13, red arrows, white asterisks; B2, B4, B6, white arrows). TCAs within the thalamus are obscured by the intense GFP fluorescence expressed by cell bodies. Therefore, the TCA projection was also examined by an antibody against medium chain neurofilament (NF-M) expressed on TCAs [45]. Strikingly, abnormally thick bundles were also detected even within the thalamus, and some of the GFP+ neurofilament+ (NF+) fibers were disoriented and failed to extend rostrally through the prethalamus but instead were misrouted caudally into the pretectum (Fig. 2A9, white arrows; B2, B4, B6, white arrowheads; C4’, C4” white arrowheads). Normally, thalamic axons exiting the prethalamus make a sharp turn to avoid the hypothalamus and enter the telencephalon (Fig. 2A1-A6). In contrast, in Isl1-deficient embryos, some thalamic axons failed to cross the boundary between the diencephalon and telencephalon and instead invaded the hypothalamic region ventrally (Fig. 2A12, yellow arrowheads). Moreover, many thalamic axons that entered the ventral telencephalon were misrouted ventrally at E16.5 (Fig. 2A7-A11, red arrowheads). We also observed GFP+ axons in the subplate layer of the cortex, but the majority of them stalled at the midpoint of the cortical areas (Fig. 2A7-A11, white arrowheads).

Fig. 2

Loss of Isl1 leads to TCA pathfinding defects in the prethalamus and thalamus. (A) Immunohistochemistry for GFP on coronal sections of control (Gbx2GFP;Dlx5/6-Cre; Isl1F/+) (A1-A6) and mutant (Gbx2GFP;Dlx5/6-Cre; Isl1F/F) (A7-A12) embryos at E16.5. The red arrows in A10 and A11 indicate abnormal TCA projections in the prethalamus of Isl1 mutant embryos. White asterisks in A10, A11, and A12 indicate abnormally large bundles of TCAs crossing the prethalamus and entering the ventral telencephalon in Isl1 mutants compared to controls. GFP+ TCAs misrouted caudally to the habenula (white arrows in A9). Red asterisks in A10-A12 mark abnormal fasciculation of TCAs within the thalamus. GFP+ TCAs misrouted ventrally in the ventral telencephalon (A7-A11, red arrowheads) and hypothalamus (A12, yellow arrowhead). In the cortex, Isl1 mutant TCAs were shorter than those of control embryos (A7-A11, white arrowheads). (A13) Examples of the measurements taken for the quantification. Yellow dashed lines in A4, A5, A10, A11 indicate the position of the lines used to quantify the number of GFP+ fluorescent signals crossing the prethalamus. (n = 5 embryos; Student’s t-test; *p < 0.001). (B) Double immunofluorescence for GFP and NF-M on sagittal sections of control and Isl1 mutant embryos at E14.5. Control TCAs extended in parallel and oriented ventrolaterally, forming a thin fasciculus in the thalamus and prethalamus, including the TRN. In contrast, Isl1 mutant TCAs formed an abnormally thick fasciculus and extended randomly in the thalamus (white arrowheads) and prethalamus (white arrows). (C) Double immunofluorescence for GFP and NF-M on sagittal sections of control and Isl1 mutant embryos at E18.5. In the thalamus, GFP+ TCAs formed abnormally thick axon bundles orienting randomly and misrouting caudally into the pretectum (white arrowheads in C4’,C4”). Cx, cerebral cortex; Eth, epithalamus; Hyp, hypothalamus; Prt, pretectum; Pt, prethalamus; Th, thalamus. Scale bars, 200 μm

As Isl1 expression is also lost in the ventral telencephalon, we next sought to determine the developmental origin of TCA defects in Isl1-deficient mutants. We therefore generated telencephalic-specific inactivation by crossing the Isl1F/F allele with the Foxg1-IRES-Cre mouse line, in which an IRES-Cre was inserted in the 3’ UTR of the Foxg1 locus to prevent Foxg1 haploinsufficiency and ectopic Cre expression found in the widely used Foxg1-Cre mice [46]. In Foxg1-IRES-Cre; Isl1F/F embryos, Isl1 expression was completely lost in the telencephalon but remained in other brain regions, including the prethalamus (Fig. S4E-H). We then crossed the Foxg1-IRES-Cre; Isl1F/F allele with the Gbx2GFP allele. In Gbx2GFP; Foxg1-IRES-Cre; Isl1F/F embryos, GFP+ NF+ thalamic axons emerging from the thalamus converged to form normal TCA bundles as they crossed the prethalamus in a pattern similar to those in controls (Fig. S5B, D, F, H). In the ventral telencephalon, some GFP+ NF+ axons were disorganized and ran ectopically toward the ventral floor (Fig. S5E-H, arrowheads). Thalamic axons that entered the cortex exhibited delayed outgrowth compared to those of controls (Fig. S5B, C, D, F, G, H, arrows). Therefore, deletion of telencephalic Isl1 can cause TCA defects to a lesser extent than loss of Isl1 in both the diencephalon and telencephalon. Taken together, these results suggest that thalamo-cortical connectivity is dependent on the functions of Isl1 from the telencephalon and diencephalon, but diencephalic Isl1 may play a more significant role in TCA projection.

Isl1 represses the expression of Efna5, which inhibits thalamic neurite outgrowth

To further understand how defective TRN differentiation resulting from Isl1 loss leads to abnormal fasciculation and misrouting of TCAs in the prethalamus, we examined the distribution of axon guidance molecules in the developing prethalamus. Quantitative RT-PCR analysis in E13.5 prethalamus revealed that Efna5, a member of the Eph receptor-interacting ligand family was significantly upregulated in Isl1-deficient prethalamus (Fig. 3A). We then verified the differential expression of Efna5 by RNA in situ hybridization. In control embryos, Efna5 was largely undetectable in the TRN (Fig. 3B1, B7), while we observed strong upregulation of Efna5 in Dlx5/6-Cre; Isl1F/F embryos (Fig. 3B4, B10). Epha4 and Epha7, which encode the receptors for Efna5, appeared to be unchanged in the absence of Isl1 (Fig. 3B2, B3, B5, B6, B8, B9, B11, B12). To examine whether ectopic upregulation of Efna5 might affect TCA navigation into the prethalamus, thalamic explants were prepared from control (Gbx2GFP;Dlx5/6-Cre; Isl1F/+) or mutant embryos (Gbx2GFP;Dlx5/6-Cre; Isl1F/F) and cocultured with aggregates of Efna5-expressing cells. When cocultured with mock-transfected cells, thalamic explants derived from control or mutant embryos grew long neurites (Fig. 3C). In contrast, both control and mutant explants exhibited significantly shorter neurites when cocultured with Efna5-expressing cells (Fig. 3C). These results suggest that Isl1 is required to repress the expression of Efna5, which interferes with the normal growth of TCA passing through the prethalamus.

Fig. 3

Loss of Isl1 derepresses the expression of Efna5, a repellent for TCAs, in the TRN. (A) Differential expression of Efna5 in the prethalamus of E13.5 mutant embryos (Dlx5/6-Cre; Isl1F/F) compared with control embryos (Dlx5/6-Cre; Isl1F/+) based on qRT-PCR. Prethalamic Efna5 expression was significantly elevated in Isl1 mutant embryos (two-way ANOVA with Sidak’s multiple comparison test; ****p ≤ 0.001). (B) RNA in situ hybridization for Efna5, Epha4, and Epha7 on coronal sections of control and Isl1 mutants at E13.5 and E14.5. Loss of Isl1 resulted in significant upregulation of Efna5 in the TRN (red dashed areas), despite similar staining in other brain regions. Scale bar, 200 μm. (B13) Quantification of in situ hybridization. The pixel intensity values of TRN were grouped and averaged from three different sections per embryo. (n = 3 embryos; Student’s t-test; **p < 0.01) (C) Thalamic explants from control (Gbx2GFP;Dlx5/6-Cre; Isl1F/+) and mutant (Gbx2GFP; Dlx5/6-Cre; Isl1F/F) embryos were cocultured with Efna5-expressing cells (outlined by a white dashed line). Explants were photographed, and outgrowth from the distal and proximal sides of the explants was quantified by GFP fluorescence-occupied area. The values were averaged from two explants per embryo, and the averaged values from three independent experiments were plotted. A significant increase in the distal-proximal (D:P) fluorescence ratio was observed in the presence of Efna5-expressing cells (two-way ANOVA with Sidak’s multiple comparison test; *p < 0.05). Pt, prethalamus; Th, thalamus; TRN, thalamic reticular nucleus

Prethalamo-thalamic axons (PTAs) precede TCAs, and loss of PTAs by Isl1 inactivation causes the pathfinding defects of TCAs in the thalamus

Our findings that abnormal fasciculation and misrouting of TCA were also evident within the thalamus itself led us to investigate whether Isl1 deletion affected patterning or regionalization of the thalamus, although the expression pattern of Isl1 does not suggest a role in thalamic development. We first analyzed the effects of Isl1 loss on early thalamic differentiation. Hes5, Sox2, Elavl3/4, and Olig2 were similarly expressed in control and Isl1-deficient thalamus (Fig. S2A-D, I). We next examined the expression profiles of a panel of transcription factors whose functions influence TCA development. Loss of Neurog2 or Lhx2 disrupts normal TCA growth and pathfinding [47, 48]. Previous genetic studies demonstrated crucial functions of Gbx2 and Tcf7l2 in regulating correct TCA guidance [44, 49]. During development, we did not observe significant alterations in the distribution of these factors in Isl1-deficient thalamus (Fig. S6). We also examined the distribution of additional postmitotic neuronal markers or axon guidance molecules, and none of these markers expression were altered in the thalamus of Dlx5/6-Cre; Isl1F/F embryos (Fig. S6). Overall, our gene expression analysis indicated that Isl1 deletion does not significantly perturb the molecular regionalization of the thalamus. Therefore, TCA abnormalities are unlikely to result from obvious defects in thalamic development.

Neurons within the prethalamus extend axons to the thalamus. These prethalamo-thalamic axons (PTAs) have been suggested to act as pioneer axons for TCA navigation [50,51,52]. However, the developmental and spatiotemporal relationships between these two axonal projections remain largely unknown. We thus investigated whether PTAs are involved in correct TCA formation within the thalamus. To mark PTAs, we crossed the Dlx5/6-Cre allele with the R26-tdTomato mouse line. At E13.5 and E16.5, we observed tdTomato+ axons that emerged from the prethalamus and extended into the thalamus of Dlx5/6-Cre; R26-tdTomato embryos (Fig. 4A). To address the temporal relationships between PTAs and TCAs, we simultaneously labeled these two axonal pathways by crossing the Gbx2GFP reporter line onto the Dlx5/6-Cre; R26-tdTomato mouse line. At E11.5, a significant portion of tdTomato+ PTAs emerged from the marginal zone of the developing prethalamus and extended caudally through the thalamus (Fig. 4B1, arrowheads), while we did not observe TCAs entering the prethalamus, indicating that PTAs precede TCAs. To assess the spatial relationships between these axonal projections within the thalamus, we visualized TCAs with NF-M immunofluorescence because after E11.5, GFP+ TCA fibers are masked by the bright GFP staining expressed by cell bodies. We found that tdTomato+ prethalamic fibers and NF-M+ thalamic fibers exhibited close associations in the thalamus (Fig. 4C3, C9). To address whether Isl1 is required for the formation of PTAs, we crossed mice carrying the floxed Isl1 allele and R26-tdTomato with the Dlx5/6-Cre line. Isl1 deletion in the prethalamus caused a profound reduction in the number of tdTomato+ axons projecting from the prethalamus to the thalamus in E13.5 and E14.5 embryos (Fig. 4C4, C10, C13). In the thalamus of control embryos, we observed tdTomato+ axons and NF-M+ TCAs in ordered and parallel arrays (Fig. 4C2, C3, C8, C9), while NF-M+ TCAs exhibited erroneous and disorganized trajectories in the thalamus of Isl1-deficient embryos (Fig. 4C5, C6, C11, C12), suggesting a potential pioneering role of PTAs in guiding thalamic axons.

Fig. 4

Isl1 is required for the formation of prethalamo-thalamic pioneer axons. (A) Immunohistochemistry for tdTomato on coronal sections of Dlx5/6-Cre; R26-tdTomato embryos at E13.5 and E16.5. Prethalamo-thalamic axons (PTAs) labeled by tdTomato extended into the thalamus, forming ordered and parallel projections. (B) Costaining for tdTomato and GFP on coronal sections of Dlx5/6-Cre; R26-TdTomato; Gbx2GFP embryos at E11.5. PTAs labeled with tdTomato extended into the thalamus, while GFP+ thalamic axons had yet to enter the prethalamus. (C) Costaining for tdTomato and NF-M on coronal sections of control (Dlx5/6-Cre; Isl1F/+:R26-tdTomato) and mutant (Dlx5/6-Cre; Isl1F/F:R26-tdTomato) embryos at E13.5 and E14.5. Loss of Isl1 caused a visible reduction in the number of tdTomato+ axons projecting from the prethalamus to the thalamus in E13.5 and E14.5 mutant embryos. Control NF-M+ thalamic axons formed ordered and parallel projections, but mutant NF-M+ thalamic axons aggregated into thick bundles that ran laterally. (C13) Examples of the measurements taken for the quantification. Yellow dashed lines in C1, C4, C7, C10 indicate the position of the lines used to quantify the number of axon bundles crossing the thalamus. (two-way ANOVA with Sidak’s multiple comparison test; *p < 0.05, **p < 0.01, ***p < 0.001). (D) Implanting the chemical tracer Neurovue into the prethalamus of control (Dlx5/6-Cre; Isl1F/+) at E13.5 visualized PTA projections across the thalamus. Coronal sections from five independent experiments revealed a defective pattern in the outgrowth of PTAs in mutant embryos (Dlx5/6-Cre; Isl1F/F) similar to that shown by tdTomato immunohistochemistry in Dlx5/6-Cre; Isl1F/F; R26-TdTomato embryos. Pt, prethalamus; Th, thalamus. Scale bars: 200 μm

Isl1 is also expressed in the ventral telencephalon, and neurons in the ventral telencephalon project to the thalamus [53]. To confirm whether the tdTomato+ axons that were lost in Dlx5/6-Cre; Isl1F/F;R26-tdTomato embryos originated from the prethalamus, we injected the neuronal tracer Neurovue into the prethalamus of E13.5 control and Dlx5/6-Cre; Isl1F/F embryos. Neurovue placed in the prethalamus of control embryos successfully labeled axons projecting to the thalamus, while we did not observe Neurovue+ axons passing through the thalamus in the absence of Isl1 (Fig. 4D). Taken together, these results suggest that Isl1 regulates correct TCA projections within the thalamus by promoting proper formation of PTAs.

Prolonged Isl1 function is required to maintain TRN neuron fate

Given that Isl1 expression persists in the TRN during late embryogenesis (Fig. S1C), our observations raise the intriguing possibility that Isl1 has a continued role in prethalamic development. To address whether the dependency of TRN development on Isl1 is temporally regulated, we took advantage of a newly generated mouse line that drives Cre-mediated recombination in the forebrain. This driver relies on a Sox2-bound regulatory element (SBR), which was previously shown to direct the transcription of a reporter gene to the telencephalon and diencephalon at late developmental stages [54]. To determine the efficiency of Cre-mediated recombination, SBR-Cre transgenic mice were generated and crossed with a R26-tdTomato reporter strain. At E16.5, strong tdTomato staining was detected in the developing forebrain (Fig. S7A, B, C). This SBR-Cre allele was then crossed with Isl1F/F mice to generate conditional mutant progeny. Isl1 staining was almost completely lost in the prethalamus of SBR-Cre; Isl1F/F embryos at E16.5 (Fig. S7J-O). Like in the prethalamus of Dlx5/6-Cre; Isl1F/F mutants, Dlx1 and Meis2 expression was significantly downregulated in the TRN of SBR-Cre; Isl1F/F embryos (Fig. 5G, H, M). Moreover, we observed ectopic activation of Arx, Pax6, and Lhx1/5 in the TRN of SBR-Cre; Isl1F/F mutants in a pattern similar to that of Dlx5/6-Cre; Isl1F/F mutants (Fig. 5I-L, M), indicating that Isl1 continues to control TRN identity during embryogenesis.

Fig. 5

Isl1 continues to regulate TRN identity during late embryogenesis (A-L) In situ hybridization for prethalamic markers on control (SBR-Cre; Isl1F/+) and mutant (SBR-Cre; Isl1F/F) coronal sections at E16.5. Dlx1 expression in the TRN but not in other prethalamic nuclei was lost in Isl1 mutant embryos (G). Meis2 expression, which was restricted to the TRN of control embryos, was abrogated in mutant embryos (H). Loss of Isl1 resulted in ectopic induction of Arx, Pax6, and Lhx1/5 in the TRN (I-L). (M) Quantification of in situ hybridization. The pixel intensity values of TRN or vLG were grouped and averaged from three different sections per embryo (n = 3 embryos; Student’s t-test; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Scale bars, 200 μm

Ascl1 is required for promoting TRN neuron fate and PTA outgrowth and regulates Isl1 expression in the prethalamus

The TCA pathfinding defects resulting from the loss of Ascl1 [37] closely resemble the defects in TCAs caused by Isl1 deficiency. Ascl1 mutant embryos exhibit abnormal growth and pathfinding of TCAs in the developing forebrain [37]. To explore whether these errors are due to the disruption of prethalamic factors, we analyzed the Ascl1KI allele, in which the entire coding region of Ascl1 was replaced with GFP [55]. Similar to the phenotype described for Isl1- deficient mutants, the loss of Ascl1 caused severe downregulation of Dlx1 in the TRN of the prethalamus (Fig. 6A6, A16). In contrast to the Isl1- deficient mutants, Dlx1 expression was also lost in the vLG and zli in the absence of Ascl1 (Fig. 6A6, A16). Meis2 expression was greatly reduced in the TRN of Ascl1 mutants in a pattern similar to that of Isl1- deficient mutants (Fig. 6A7, A17). The narrow band corresponding to Arx, Pax6, or Lhx1 expression along the zli was mostly missing, and the expression of these genes was also lost in the vLG in the absence of Ascl1 (Fig. 6A8, A9, A10, A18, A19, A20). In contrast, Arx, Pax6, and Lhx1 expression was ectopically induced in the TRN of Ascl1KI/KI mutants (Fig. 6A8, A9, A10, A18, A19, A20). These data suggest that, similar to Isl1, Ascl1 is required to promote TRN identity and repress non-TRN prethalamic cell fates within the TRN and that unlike Isl1, Ascl1 is also important for the differentiation of other prethalamic nuclei.

We next analyzed circuit formation between the prethalamus and thalamus. Consistent with previous observations [37], NF-M+ thalamic axons formed abnormally thick fascicles and extended randomly in the prethalamus (Fig. 6B3, B4). Axonal arrangement was also severely disrupted within the thalamus of Ascl1KI/KI mutants. NF-M+ thalamic axons formed thin fascicles and were arranged in parallel in the control thalamus, whereas those in Ascl1 mutants formed thick bundles and crossed each other with abnormal overlapping course of axons (Fig. 6B3, B4, arrows). Although defective TCAs within the thalamus could be a secondary consequence of a total failure in morphological maturation of the prethalamus, we assessed the potential importance of PTAs in the establishment of thalamocortical connectivity. We therefore crossed mice carrying the Ascl1 null allele and R26-tdTomato with the Dlx5/6-Cre line. Loss of Ascl1 resulted in a marked reduction in the number of tdTomato+ axons projecting from the prethalamus to the thalamus in E13.5 and E14.5 embryos (Fig. 6C). When the Neurovue tracer was injected into prethalamic regions, Neurovue labeling accumulated in the prethalamus of both control and mutant embryos, but Neurovue+ axons were not apparent in the thalamus of Ascl1 mutants (Fig. 6D), confirming that Ascl1 deletion disrupts the formation of axons that project from the prethalamus to the thalamus. These results demonstrate that Ascl1 and Isl1 have a significant degree of overlapping regulatory function in prethalamic differentiation and circuit formation.

Fig. 6

Ascl1 controls TRN cell fate and PTA outgrowth and is essential for Isl1 expression in the prethalamus. (A) In situ hybridization for prethalamic markers on control (Ascl1KI/+) and mutant (Ascl1KI/KI) coronal sections at E14.5 and E16.5. In Ascl1 mutants, Dlx1 expression was lost in the TRN as well as in the rZI, vLG, and zli (A6, A16). Loss of Ascl1 resulted in severe downregulation of Meis2 expression in the TRN (A7, A17). Arx, Pax6, or Lhx1 expression in the vLG and zli of Ascl1KI/KI mutants was downregulated but ectopically induced in the TRN (A8-A10, A18-A20). (A21) Quantification of in situ hybridization. The pixel intensity values of TRN or vLG were grouped and averaged from three different sections per embryo (n = 5 embryos; Student’s t-test; **p < 0.01, ***p < 0.001, ****p < 0.0001). (B) Immunohistochemistry for NF-M on coronal sections of control (Ascl1KI/+) and mutant (Ascl1KI/KI) embryos at E14.5 and E16.5. In control embryos, NF-M+ thalamic axons extended in ordered and parallel arrays that were less fasciculated in the thalamus and prethalamus (B1,B2) than in mutant embryos (B3,B4). Moreover, in Ascl1 mutants, thick NF-M+ bundles crossed each other with abnormal overlapping course of axons (arrows in B3,B4). (C) Immunohistochemistry for tdTomato on coronal sections of control (Dlx5/6-Cre; R26-TdTomato; Ascl1KI/+) and mutant (Dlx5/6-Cre; R26-TdTomato; Ascl1KI/KI) embryos at E13.5 and E14.5. The control prethalamus formed ordered and parallel tdTomato+ projections to the thalamus (C1,C3), while very few PTAs were detected in the mutant embryos (C2,C4). (C5) Examples of the measurements taken for the quantification. Yellow dashed lines in C1-C4 indicate the position of the lines used to quantify the number of axon bundles crossing the thalamus. (n = 3 embryos; two-way ANOVA with Sidak’s multiple comparison test; **p < 0.01, ***p < 0.001, ****p < 0.0001). (D) Implanting the chemical tracer Neurovue into the prethalamus of control (Ascl1KI/+) and mutant embryos (Ascl1KI/KI) at E13.5. Compared to those in controls (D1), no Neurovue+ PTAs were labeled in the thalamus in the absence of Ascl1 (D2). (E) In situ hybridization for Isl1 and Shh on control (Ascl1KI/+) and mutant (Ascl1KI/KI) coronal sections at E12.5. Shh is a specific marker for the zli. Loss of Ascl1 abrogated Isl1 expression in the TRN. (E5) Quantification of in situ hybridization. The pixel intensity values of TRN were grouped and averaged from four different sections per embryo (n = 4 embryos; Student’s t-test; ****p < 0.0001).Pt, prethalamus: Hyp, hypothalamus; Pt, prethalamus; rZI, rostral zona incerta; Th, thalamus; TRN, thalamic reticular nucleus; vLG, ventral lateral geniculate; zli, zona limitans intrathalamica. Scale bars, 200 μm

We then investigated the gene regulatory hierarchy between Isl1 and Ascl1. In the developing prethalamus, Ascl1 expression is restricted to ventricular zone progenitors (Fig. S2E), whereas Isl1 is expressed both in the progenitor and postmitotic regions (Fig. 6E1, E2). When in situ hybridization for Isl1 on coronal sections was performed with Shh, a zli-specific marker, to locate the affected area more precisely, we found that Isl1 expression was lost in the prethalamus of Ascl1KI/KI mutant embryos (Fig. 6E3, E4). In contrast, Ascl1 expression was largely unaffected in the prethalamus of Dlx5/6-Cre; Isl1F/F mutants (Fig. S2E). These observations revealed a crucial role for Ascl1 in the regulation of Isl1 expression in the prethalamus.

Isl1 is necessary for Ascl1-mediated fate specification and neurite outgrowth in in vitro cultured prethalamic neurons

The above-described observations led us to ask whether Isl1 is an essential component of Ascl1-mediated genetic programs that support neuronal fate specification and connectivity of the TRN. To address this question, we established primary cell cultures from the prethalamus of E13.5 Ascl1KI/+ and Ascl1KI/KI embryonic brains. To verify whether cultured cells were indeed derived from TRN tissues, cells that had been cultured for 2 days in vitro (DIV) were immunostained with Isl1, Dlx1, and Pax6 antibodies. The majority of primary cultured cells from control TRN tissues were Isl1+, Dlx1+ and Pax6−, while Ascl1-deficient cells exhibited Isl1−, Dlx1− and Pax6+ immunoreactivity (Fig. 7A, B), indicating TRN cell identity. We next examined neurite outgrowth of neurons in primary cultures from Ascl1KI/+ and Ascl1KI/KI embryos using Tuj1, a neuron-specific class III beta-tubulin that is widely used to visualize neuronal extensions. We measured the number of neurites, the length of the longest neurite, and the total length of neurites to assess neurite outgrowth in primary cultures. Consistent with our embryo experiments, Ascl1KI/KI primary cultured neurons developed fewer neurites than did Ascl1KI/+ neurons at DIV2 and DIV6 (Fig. 7C, D). The longest neurites and the total length of neurites in Ascl1KI/KI neurons were shorter than those in Ascl1KI/+ neurons (Fig. 7C, D). To assess the ability of Isl1 to rescue the TRN phenotypes caused by Ascl1 deficiency, an Isl1 expression construct was transfected into Ascl1KI/KI TRN cells that had been cultured for 1 DIV together with a tdTomato reporter expression vector to identify the transfected cells. After 6 additional days of culture, the tdTomato+ cells were evaluated for neuron identity and neuritogenesis. Compared with that in mock-transfected Ascl1KI/KI cells, the number of Dlx1+ cells in primary cultures of Ascl1KI/KI cells transfected with the Isl1 expression construct increased, while the number of Pax6+ cells in Isl1-transfected cells decreased (Fig.