Satb2 Regulates the Differentiation of Both Callosal and Subcerebral Projection Neurons in the Developing Cerebral Cortex

Abstract

The chromatin-remodeling protein Satb2 plays a role in the generation of distinct subtypes of neocortical pyramidal neurons. Previous studies have shown that Satb2 is required for normal development of callosal projection neurons (CPNs), which fail to extend axons callosally in the absence of Satb2 and instead project subcortically. Here we conditionally delete Satb2 from the developing neocortex and find that neurons in the upper layers adopt some electrophysiological properties characteristic of deep layer neurons, but projections from the superficial layers do not contribute to the aberrant subcortical projections seen in Satb2 mutants. Instead, axons from deep layer CPNs descend subcortically in the absence of Satb2. These data demonstrate distinct developmental roles of Satb2 in regulating the fates of upper and deep layer neurons. Unexpectedly, Satb2 mutant brains also display changes in gene expression by subcerebral projection neurons (SCPNs), accompanied by a failure of corticospinal tract (CST) formation. Altering the timing of Satb2 ablation reveals that SCPNs require an early expression of Satb2 for differentiation and extension of the CST, suggesting that early transient expression of Satb2 in these cells plays an essential role in development. Collectively these data show that Satb2 is required by both CPNs and SCPNs for proper differentiation and axon pathfinding.

Keywords: Ctip2; Fezf2; callosal projection neuron; corticospinal tract; development; pyramidal neuron; specification.

Figure 1.

Emx1-Cre efficiently ablates Satb2 in neocortical pyramidal neurons. Coronal sections of P4 neocortex were assessed for Cre recombination efficiency. Animals carried the Ai9 RFP reporter allele to mark cells that have undergone recombination. Satb2 (green, A–F) is readily observed in Emx1-Cre⁺;Satb2^+/lacZ;Ai9⁺ (control) neocortex in Layers 2–6 (A), but completely lost in Emx1-Cre⁺;Satb2^lox/lacZ;Ai9⁺ (mutants; B). (C, C′) High power confocal pictures of the superficial layers of control animals show that the vast majority of Satb2⁺ neurons coexpress RFP, which, due to its high mRNA stability is detected almost ubiquitously except for nuclei. Note that RFP⁻ cells (blue in C; arrowheads in C and C′) do not express Satb2, suggesting that these non-recombined cells are either interneurons or endothelial cells. (D, D′) High power confocal pictures of superficial layers of a mutant shows complete loss of Satb2. (E, F) High power confocal pictures of Layer 5 in control (E) and mutant (F) neocortex shows loss of Satb2 in the deep layers of mutants. Scale bars: 200 μm in B for A, B; 50 μm in F for C–F.

Figure 2.

Molecular changes in superficial layers of Emx1-Cre;Satb2 mutant mice. In situ hybridization on coronal P1 sections. (A–B′) Crystallin µ (CRYM), a SCPN marker, is expressed by Layer 5 neurons of Emx1-Cre⁺;Satb2^+/LacZ (controls; A, A′), but is strikingly upregulated across the entire cortical wall of Emx1-Cre⁺;Satb2^lox/lacZ (mutants; B, B′). (C–D) Immunohistochemistry for Fog-2 (red in C, D) and Ctip2 (green in C, D) on coronal sections of P4 brains shows strong expression of Fog-2 in Layer 6 and the superficial layers of cingulate cortex in controls (arrowhead in C), but a loss of Fog-2 expression in mutant cingulate cortex (D). Scale bars: 1 mm in B for A, B; 100 µm in B′ for A′, B′, 500 µm in D for C, D.

Figure 3.

Satb2-deficient superficial layer neurons adopt some electrophysiological characteristics of deep layer neurons. (A–C) Layer 2/3 neurons of somatosensory cortex were whole-cell patch-clamped in current-clamp mode and recorded in acute cortical slices prepared from P25 mutant and control animals. Satb2 mutant upper layer neurons acquired a hyperpolarization-activated current in response to a 100 pA current step. (A) Overlay of representative responses of patch-clamped neurons from Layer 2/3 control (Emx1-Cre⁺;Satb2^+/LacZ; black) and mutant animal (Emx1-Cre⁺;Satb2^lox/LacZ; red). Peak of responses (arrowheads) and end-of-step (open arrowheads) show differences in sag responses for mutants (red) and controls (black). (B) Quantification of voltage differences between the peak and end-of-step responses to the hyperpolarizing current step reveals a significant difference for the mutants (red in B) compared with controls (black in B). (C) Quantitation of sag responses and after-depolarization in control versus mutant animals shows significant increases in Satb2 mutant neurons compared with controls. (D–E) In situ hybridization for HCN1 on coronal E18.5 sections reveals a striking upregulation of HCN1 expression in the Satb2-deficient cortex (E) compared with control (D). Scale bar: 1 mm in E (D, E).

Figure 4.

Laminar fate restriction of pyramidal neurons within cortical layers. (A–F) Retrograde tracing from subcortical targets reveals laminar restriction of pyramidal neurons in sagittal sections of P17 Emx1-Cre;Satb2 animals. (A, B) Retrograde labeling from the thalamus reveals labeled neurons (red in A, B) in Layer 6 of Emx1-Cre⁺;Satb2^+/lacZ (control; A) and Emx1-Cre⁺;Satb2^lox/lacZ (mutant; B). (C, D) Retrograde labeling from the cerebral peduncle at the pons/midbrain junction reveals neurons in Layer 5 (red in C, D) in control (C) and mutant (D) Emx1-Cre;Satb2 animals. Sections were counterstained with Fog-2 (green in C, D), which shows high expression in Layer 6 and weak expression in some Layer 5 neurons. (E, F) Backlabeling of neurons in Layer 5 of control (E) and mutant (F) brains from retrograde injections into the anterior pretectal nucleus. Animals were hemizygous for golli-GFP (green in E, F), which prominently labels neurons of Layer 6 and some Layer 5 neurons. (G–J) Tamoxifen-inducible Nestin-Cre^ERT2;Satb2 mutants confirm that aberrant subcortical projections in Satb2 mutants arise from the deep layers, and not from the superficial layers. (G, I) Sagittal sections of a P4 Nestin-Cre^ERT2;Satb2^lox/lacZ mutant treated with a single dose of Tamoxifen at E11.5 reveal β-galactosidase expression (red in G–J) across the entire cortical wall (G) and β-galactosidase⁺ axons in the internal capsule (arrowhead in G), thalamus (open arrowhead in G) and along the CST in the brainstem (I). (H, J) Sagittal sections of a P4 Nestin-Cre^ERT2;Satb2^lox/lacZ mutant treated with a single dose of Tamoxifen at E15.5 shows β-galactosidase expression confined to superficial layers (H) and the absence of β-galactosidase⁺ subcortical axons (J), suggesting that Satb2-deficient superficial CPNs do not contribute to subcortical projections. (K, L) Anterograde axonal labeling using Phaseolus vulgaris leucoagglutinin (PHA-L) from somatosensory cortex of P15 Emx1-Cre⁺;Satb2^lox/+ (control; K) and Emx1-Cre⁺;Satb2^lox/lacZ (mutant; L) shows labeled axons along the corpus callosum of controls (filled arrowhead in K), but a dramatic reduction in mutants (L). Instead, mutant axons project ventrally (asterisk in L) and an increase in ipsilateral projections (open arrowheads in K and L) is observed in Emx1-Cre;Satb2 mutants. Arrows in K and L indicate injection sites. Scale bars: 100 μm in A for A–F; 1 mm in H (for G, H); 500 μm in J (for I, J) and K (for K, L).

Figure 5.

Loss of Satb2 leads to a failure of the CST. (A–H) Emx1-Cre;Satb2 animals heterozygous for Fezf2^AP allow visualization of the CST using AP. (A, B) Sagittal sections of P4 brains reveal AP⁺ CST axons within the cerebral peduncle (open arrowheads in A, B). (A) In Emx1-Cre⁺;Satb2^lox/+;Fezf2^AP controls, the CST can be observed in the ventral brainstem (arrowhead in A) and crossing into the spinal cord at the pyramidal decussation (asterisk in A). In Emx1-Cre⁺;Satb2^lox/LacZ;Fezf2^AP mutants, labeling is present in the cerebral peduncle (open arrowhead in B), but no AP is detected in the brainstem (arrowhead in B) or pyramidal decussation. (C, D) Cross sections through the brainstem at the levels indicated by the dashed lines in A and B reveal the AP⁺ CST at the ventral surface of controls (arrowheads in C and C′), but a loss of the CST in Emx1-Cre;Satb2 mutants (D, D′). (E, F) Whole-mount ventral views of AP histochemistry on P4 brains confirm the failure of CST formation in Emx1-Cre;Satb2 mutants. In controls, AP staining marks CST axons in the cerebral peduncle (open arrowhead in E), along the pons (arrowhead in E), the pyramids (arrow in E), and pyramidal decussation (asterisk in E). In Satb2-deficient animals, some AP reactivity is found at the cerebral peduncle (open arrowhead in F), but the tract fails to extend into the brainstem (F). (G, H) Cross sections through cervical spinal cord of P15 animals reveal a loss of CST axons in Emx1-Cre;Satb2 mutants. In controls, AP labels CST axons in the ventral dorsal funiculus (arrow in G), but no AP signal is detected in the mutant spinal cord (arrow in H). (I–L) Protein kinase C γ (PKCγ) immunohistochemistry confirms the loss of CST in Satb2 mutants. Immunohistological analysis of P15 sagittal brain sections reveals the PKCγ⁺ CST in controls at the level of the cerebral peduncle, pons (open arrowhead in I), pyramids (arrowhead in I), and pyramidal decussation (asterisk in I). In Emx1-Cre;Satb2 mutants, strong PKCγ labeling is found in the cerebral peduncle (arrow in J), but only weak label at the pons (open arrowhead in J) and very weak signal along the pyramids (arrowhead in J). (K, L) Cross sections through cervical spinal cord of P15 animals reveals PKCγ labeling of the CST in the ventral dorsal funiculus of controls (arrowhead in K), but the absence of the tract in mutants (arrowhead in L). PKCγ is also expressed by interneurons of the spinal cord (open arrowheads in K, L). Sections were colabeled with L1. Scale bars: 1 mm in B (A, B), D (C, D), D′ (C′, D′), F (E, F), 250 μm in H (G, H), 1 mm in J (I, J), 250 μm in L (K, L).

Figure 6.

Colocalization of Satb2 and Ctip2 during embryonic development suggests early role of Satb2 in corticospinal motor neuron differentiation. (A–L) Immunohistological analysis of Satb2 (red in A–J) and Ctip2 (green in A–J) reveals overlap during embryonic development of the cerebral cortex. (A, B) At E13.5, the time when Layer 5 neurons are being born, very little Satb2 immunoreactivity is observed in the cerebral cortex (A). At high power magnification (B; boxed area in A), few Ctip2⁺ cells coexpress Satb2 (arrows in B). (C, D) At E14.5, Satb2 expression emerges more robustly, with highest expression levels found in the lateral areas of developing cortex (arrow in C). (D) In motor cortex (boxed area in C), some Ctip2⁺ cells coexpress Satb2 (arrows in D). (E, F) At E15.5, Satb2 expression has expanded from lateral neocortex into motor cortex, and Satb2 marks the majority of neurons in developing Layer 4. (G, H) At E16.5, robust Satb2 expression is apparent in the developing upper layers of the cerebral cortex. High power magnification (H; boxed area in G) reveals considerable overlap between Satb2 and Ctip2 in maturing Layer 5 neurons. (I, J) at P4, robust Satb2 expression is observed across the entire cortical wall (I) and magnification of Layer 5 (J; boxed area in I) shows a fraction of Ctip2⁺ neurons coexpressing Satb2. Scale bars: 250 μm in A (A, C, E, G), 500 μm in I, 50 μm in J (B, D, F, H, J).

Figure 7.

Molecular changes in Layer 5 of Emx1-Cre;Satb2 mutants suggest incomplete differentiation of SCPNs. In situ hybridization on coronal P1 sections shows a loss of the SCPN differentiation markers Foxo1 (A–B) and Foxp2 (C–D) in Layer 5 motor cortex (arrowheads in A–D) of Emx1-Cre⁺;Satb2^lox/LacZ mutants (B, D) compared with Emx1-Cre⁺;Satb2^lox/+ controls (A, C). High power magnification reveals expression of Foxo1 in Layer 5 of controls (A′) and loss of Foxo1 in Layer 5 motor cortex of Emx1-Cre;Satb2 mutants (B′). Foxp2 expression is detected in both Layer 5 (arrowhead in C′) and Layer 6 of control animals, but is lost specifically in Layer 5 of Emx1-Cre;Satb2 mutants (D′). Scale bars: 1 mm in D for A–D, 100 μm in D′ for A′–D′.

Figure 8.

Rbp4-Cre targets recombination to Layer 5 pyramidal neurons. Rbp4-Cre activity was assessed by Ai9-RFP expression in Rbp4-Cre⁺;Ai9⁺ animals. (A–F) Immunohistological analysis of RFP (in red in A–I) and Ctip2 (green in A–F) shows the onset of RFP expression in coronal sections of E16.5 brains (A, B). (A) Very few RFP⁺ cells are seen in the developing cortex at E16.5. (B) High power magnification of motor cortex shows that the majority of RFP⁺ neurons coexpress Ctip2 (arrowheads in B). (C) At E17.5, only a fraction of Ctip2⁺ neurons coexpress RFP. (D–F) Coronal sections of P4 animals show robust RFP expression in Layer 5 neurons (arrowhead in D) and RFP⁺ axons in the internal capsule (arrow in D). (E) The majority of Ctip2⁺ neurons coexpress RFP; some RFP⁺ cells are also found above the white matter (arrowhead in E). (F) High power magnification of Layer 5 reveals that the majority of the large Ctip2⁺ Layer 5 neurons coexpress RFP, and the majority of the RFP⁻;Ctip2⁺ neurons are neurons with smaller nuclei. (G) A coronal section of a P4 brain shows RFP in the anterior commissure (arrowhead in G). (H, I) Sagittal sections of P4 brains reveal RFP⁺ axons in the cerebral peduncle (arrowheads in H), in the CST in the brainstem (arrowhead in I), in the pyramidal decussation (arrow in I) and in descending fibers in the spinal cord (open arrowhead in I). Scale bars: 250 μm in A and E; 100 μm in C for B, C; 50 μm in F; 1 mm in D, G and H; 500 μm in I.

Figure 9.

Early Satb2 expression is required for proper CST development. The Ai9-RFP reporter allele was used to visualize Cre-recombined neurons and corticospinal tracts in different mutant Satb2 strains. (A–B′) Coronal sections of P4 brains were analyzed for recombination efficiency in Rbp4-Cre⁺;Satb2^lox/+;Ai9⁺ controls (A, A′) and Rbp4-Cre⁺;Satb2^lox/LacZ;Ai9⁺ mutants (B, B′), respectively. (A, B) Low power magnification views show Satb2 expression (green in A–B′) in all cortical layers of both control and mutant, and Rbp4-Cre-induced RFP expression (from the Ai9 reporter) in Layer 5 neurons of both genotypes. (A′, B′) High power magnification of boxed areas in A and B. In controls, a subset of RFP⁺ Layer 5 neurons coexpressed Satb2 (arrowheads in A′), similar to the coexpression of Satb2 and Ctip2 in Figure 6. (B′) In Rbp4-Cre;Satb2 mutants, the fraction of Satb2⁺;RFP⁺ neurons is dramatically reduced. The small number of Satb2⁺;RFP⁺ neurons (arrowhead in B′) may reflect a slow Satb2 protein turnover in Cre-recombined postmitotic neurons. Satb2⁺;RFP⁻ neurons (green in B′) are CPNs that are not targeted by Rbp4-Cre. (C, D) Early recombination using Emx1-Cre reveals RFP⁺ CST axons in sagittal sections along the brainstem (arrowhead in C) and in the pyramidal decussation (open arrowhead in C) of P4 Emx1-Cre⁺;Satb2^lox/+;Ai9⁺ controls (C). In Emx1-Cre⁺;Satb2^lox/LacZ;Ai9⁺mutants, labeled axons occupy the cerebral peduncle (asterisk in D), but few labeled axons are visible at the pons (open arrow in D) or brainstem (arrowhead in D). (E, F) Late, Layer 5 specific recombination using Rbp4-Cre robustly labels the CST along the brainstem (arrowheads in E, F), pyramidal decussation (open arrowheads in E, F) and in descending fibers of the spinal cord (arrows in E, F) of both control (E) and mutant (F) animals at P4. (G–J) RFP and PKCγ immunohistochemistry on cross sections through the cervical spinal cord of P15 animals: (G, H) Early recombination with Emx1-Cre leads to robust RFP expression in the control CST within the ventral dorsal funiculus, which colabels with PKCγ (G). In mutants both PKCγ and RFP are missing, suggesting the CST fails to reach the spinal cord (H). (I, J) In contrast, late recombination induced by Rbp4-Cre does not disrupt CST development; RFP and PKCγ are detected in the ventral dorsal funiculus in both controls (I) and mutants (J). Scale bars: 250 μm in B for A, B; 50 μm in B′ for A′, B′; 1 mm in F for C–F; 100 μm in J for G–J.