The Fezl coding region was targeted with a cassette containing a 3-kb homology sequence upstream of the start codon, an EGFP-IRES-PLAP cassette (EGFP, enhanced GFP; IRES, internal ribosomal entry site; PLAP, human placental alkaline phosphatase), a floxed PGK-neo cassette, and a 4.3-kb homology sequence downstream of Fezl followed by a RSV-TK cassette (RSV-TK, Rous sarcoma virus promoter driving expression of thymidine kinase gene) (Fig. 6A, which is published as supporting information on the PNAS web site). The linearized construct was electroporated into E14a embryonic stem cells, which were subjected to positive (neo) and negative (TK) selection then screened by Southern analysis to select homologous recombinant clones. Heterozygous Fezl^+/- mice were mated to β-actin-cre CD1 mice to excise the pgk-neo selection cassette, and these Fezl^+/- mice were mated to generate Fezl^-/- mutants.

In situ hybridization using ³⁵S-labeled probes was performed as described (17). PLAP activity was detected with AP staining buffer (0.1 mg/ml 5-bromo-4-chloro-3-indolyl phosphate/1 mg/ml nitroblue tetrazolium in 100 mM Tris·Hcl, pH 9.5/100 mM NaCl). For anterograde axon tracing, three injections of biotinylated dextran amine (BDA) (10% in saline; Molecular Probes) were placed in the motor, somatosensory, or visual cortex of adult mice (see Supporting Text, which is published as supporting information on the PNAS web site). For retrograde labeling, rhodamine fluorescent latex microspheres (Lumafluor, New York) were injected bilaterally into the pyramidal decussation or superior colliculus of adult mice. Survival times were 2 weeks (BDA) or 1 week (microspheres). To visualize BDA, 45-μm coronal or sagittal sections were processed with ABC VectaStain Kit (Vector Laboratories) and diaminobenzidine. Each labeling experiment used three to six mutants and three to six littermate controls.

The Fezl ORF encodes a protein of 455 amino acids that contains six C₂H₂-type zinc-finger motifs and an eh1 domain, which is conserved in several transcriptional repressors (12, 14). To generate Fezl-deficient mice, we replaced the entire Fezl ORF with a promoterless EGFP-IRES-PLAP (EGFP, enhanced GFP; IRES, internal ribosomal entry site) cassette (18). Southern analysis and PCR genotyping confirmed germ-line transmission of the targeted allele (Fig. 6 A and B).

Consistent with a previous study of Fezl mutants (19), heterozygous Fezl^+/- mice produced homozygous Fezl^-/- progeny at a Mendelian ratio, and locomotor activity tests revealed that Fezl^-/- mice were twice as active as littermate controls (control, 161.4 ± 64.3 counts per day; Fezl^-/-, 335.5 ± 126.7 counts per day, P = 0.0039). Mutant brains appeared grossly normal, with clear evidence of cortical lamination (Fig. 6 C-F). Although defects in subplate neurons and thalamocortical projections have been reported in Fezl mutants (19), subplate neurons were present in the mutants generated here, as revealed by cresyl violet staining (Fig. 6 E and F) and expression of Tbr1 (Fig. 7 C and D, which is published as supporting information on the PNAS web site), and cytochrome oxidase staining revealed normal barrel fields in somatosensory cortex (Fig. 8, which is published as supporting information on the PNAS web site).

PLAP Recapitulates Fezl Expression in Mutant and Heterozygous Mice. To visualize Fezl-expressing neurons and their axons, we replaced the Fezl ORF with enhanced GFP (EGFP) and PLAP. For unknown reasons, EGFP could not be detected, but PLAP mRNA and protein activity were easily visualized. In situ hybridization revealed that the pattern of PLAP expression was essentially identical to that of Fezl mRNA (Fig. 9 A and B, which is published as supporting information on the PNAS web site) in both Fezl^+/- and Fezl^-/- brains, with expression in layers 5 and 6 (Fig. 9 C-F) and in the hippocampus and hypothalamus (data not shown) (15). These data suggest that deep-layer neurons were present in Fezl mutant brains, and that these cells continue to express the Fezl mutant allele. BrdUrd-labeling studies confirmed that layer 5 and 6 neurons were generated at the appropriate time, migrated normally, and survived into postnatal life (data not shown). However, layer 5 appeared thinner and more densely populated with smaller cells in Fezl mutants (Fig. 6 C-F).

Disruption of Molecular Markers for Deep Cortical Layers in Fezl Mutant Mice. To explore the fates of layer 5 and 6 neurons in Fezl^-/- brains, we performed in situ hybridization at embryonic day (E)14 and postnatal day (P)0 and P15 for genes that distinguish these neurons from those in other layers. The expression of some deep-layer genes, such as Otx1 (20) and Tbr1 (11), was unaffected in the Fezl^-/- cortex (Fig. 7 A-D), suggesting that Fezl-deficient mice produce layer 5 and 6 neurons [including corticospinal motor neurons (CSMNs), which express Otx1 (20)]; however, the expression of other deep-layer markers was severely disrupted (Table 1). Two mRNAs expressed in layer 5, ER81 and Foxo1, were greatly reduced or completely absent in the Fezl^-/- cortex (Fig. 1 A-D). Similarly, the expression of the layer 6 markers Grg4 and Foxp2 was severely reduced or abrogated in mutants (Fig. 1 E-H). Although expression of these genes was disrupted in the deep layers of cortex, expression in other brain regions was unaffected (e.g., Foxp2 in striatum; Fig. 1F).

Open in a separate window

Fig. 1.

In situ hybridization for genes normally expressed by deep layer neurons at P15. (A and B) ER81 is normally expressed in layer 5 and hippocampus, but expression in layer 5 is severely reduced in Fezl mutants. (C and D) Expression of Foxo1 in layer 5 is reduced in Fezl^-/- mice. (E and F) Foxp2, which is normally expressed in layer 6 and striatum, is lost from layer 6 in Fezl^-/- brains. (G and H) The expression of Grg4 in layers 5 and 6 is severely reduced in mutants. (Scale bar, 0.5 mm.)

The altered expression of deep-layer markers suggested that Fezl mutant mice might have defects in the specification or differentiation of neurons in layers 5 and 6. To ascertain whether these cells acquired alternative laminar fates, we performed in situ hybridization for seven genes normally expressed in other cortical layers. All were expressed normally in Fezl mutants (Fig. 7 E-H; Table 2, which is published as supporting information on the PNAS web site), suggesting that layer 5 and 6 neurons did not adopt upper-layer identities.

Defective Axonal Projections Revealed by PLAP Staining. The PLAP marker targeted to the Fezl locus was used to visualize the axons of Fezl-expressing mutant neurons at P0, P8, and P15. In Fezl^+/- and Fezl^-/- animals, PLAP marked the cell bodies and dendrites of neurons in the cortical plate and hippocampus (Fig. 2). In addition, major axon tracts associated with layer 5 and 6 subcortical projections were clearly labeled and showed dramatic changes in Fezl^-/- brains (Fig. 2 K-R). Layer 5 and 6 neurons project subcortical axons by growing through the internal capsule, then diverging into the dorsal thalamus (layer 6) or toward the spinal cord (layer 5). The fasciculation of PLAP-labeled axons in the internal capsule was disorganized in Fezl^-/- brains (Fig. 2 K and L), and it appeared that a greater number of labeled axons entered the thalamus in mutants (Fig. 2 B, E, M, and N). However, because PLAP reveals fasciculated axons and not more diffuse projections, it was difficult to discern whether thalamic targeting was altered.

Open in a separate window

Fig. 2.

PLAP staining reveals defective axonal projections in Fezl^-/- brains at P0. (A-F) Sagittal sections showing PLAP staining in Fezl^+/- brains (A-C) and Fezl^-/- brains (D and E). (A and D) PLAP stains axons in the internal capsule and cerebral peduncle of control and mutant brains. PLAP activity is strong in the anterior commissure in Fezl^-/- brains (D) but much weaker in heterozygotes (A). (B and E) A subset of axons in both controls and mutants leaves the internal capsule and enters the thalamus (th). (C and F) PLAP prominently marks the pyramidal tract in heterozygotes (arrowhead), but this labeling is absent in Fezl mutants. (G-P) Coronal sections showing PLAP staining in Fezl^+/- (G, I, K, M, O, and Q) and Fezl^-/- brains (H, J, L, N, P, and R). (G-J) The corpus callosum is apparent in heterozygotes (G and I) but absent in Fezl mutants (H and J). Some axons in Fezl^-/- mice take an aberrant pathway into the septum (arrowhead in H). Mutants show increased PLAP in the external capsule and posterior limb of the anterior commissure (J, arrowhead) compared with controls (I). (K and L) PLAP marks axons in the internal capsule (arrowhead), but these axons appear disorganized in mutants. (M and N) More PLAP positive axons in the cerebral peduncle turn dorsally into the thalamus (arrowhead) in Fezl^-/- brains than in controls. (O and P) PLAP activity in the cerebral peduncle (arrowhead) is dramatically reduced in Fezl^-/- brains. (Q and R) As axons approach the pyramidal decussation, PLAP marks the pyramidal tract (arrowhead) in controls, but little or no staining is visible in mutants. ic, internal capsule; cp, cerebral peduncle; pt, pyramidal tract; th, thalamus; cc, corpus callosum; ac, anterior commissure; ec, external capsule.

The axons of layer 5 neurons project to a variety of subcortical targets, including the superior and inferior colliculi, pons, deep cerebellar nuclei, and spinal cord (1). Extensive defects were seen in these projections in Fezl mutants. In control brains at P0, PLAP-positive axons descended through the cerebral peduncle and pyramidal tract to the pyramidal decussation, where they formed the corticospinal tract (CST) (Fig. 2 A-C) (1). In Fezl^-/- mutants, PLAP-labeled axons were present in the cerebral peduncle, but staining decreased dramatically at caudal levels (Fig. 2 O and P), with little or no PLAP activity apparent in the pyramidal tract (Fig. 2R). Given that Fezl^-/- mice have two copies of IRES-PLAP at the Fezl locus rather than one, the absence of PLAP labeling in the CST of Fezl^-/- mice suggested that layer 5 axons failed to develop normally.

The absence of CST axons raised the possibility that layer 5 neurons innervated abnormal targets. Indeed, PLAP revealed axon pathways in mutant brains that were not prominently labeled in heterozygotes. The external capsule and anterior commissure showed light PLAP labeling in Fezl^+/- mice (Fig. 2I). However, even after taking into account the dosage of the mutant allele, much stronger PLAP activity was detected in these regions in Fezl^-/- brains (Fig. 2 J), suggesting that axons from Fezl-expressing neurons that normally extend through the internal capsule were diverted onto alternative inappropriate pathways.

The corpus callosum, which showed only faint PLAP staining in heterozygotes, was completely absent in Fezl^-/- mice (Fig. 2 G-J). PLAP-positive axons failed to cross the midline and instead accumulated into Probst bundles (Fig. 2 H and J) or turned ventrally into the septum (Fig. 2H, arrowhead). It is not clear why the callosum is missing in Fezl-deficient brains, but this suggests that early Fezl expression may be required for midline development.

Subcortical Projection Defects Revealed by Retrograde Labeling. PLAP effectively reveals axon bundles but lacks the sensitivity to detect individual axons. To confirm the changes in axonal projections, we retrogradely labeled cortical neurons from the pyramidal decussation or superior colliculus with fluorescent latex microspheres. As expected, injections into the pyramidal decussation of control mice marked layer 5 neurons throughout the motor and somatosensory areas, but not the visual cortex (Fig. 3 A and B). However, in Fezl^-/- mutants, only a very small number of layer 5 neurons, found exclusively in the somatosensory cortex, extended axons to the pyramidal decussation (Fig. 3 C and D). Injections into the superior colliculus of controls labeled layer 5 neurons in visual and somatosensory cortex (Fig. 10 A and B, which is published as supporting information on the PNAS web site). In Fezl^-/- mice, very few layer 5 neurons were labeled, and these cells were confined to caudal visual cortex (Fig. 10 C and D). These experiments confirmed that most layer 5 cortical neurons in Fezl^-/- mice failed to extend axons to subcortical targets.

Open in a separate window

Fig. 3.

Defects in layer 5 projections to the pyramidal decussion in Fezl^-/- mutants. Sagittal sections showing retrograde labeling from the pyramidal decussation. The extent of retrograde labeling is indicated by arrowheads. Fluorescence images have been inverted in A and C. (A) Retrograde labeling normally marks layer 5 neurons throughout the motor and somatosensory areas. (B) High-power view of the region delineated by the box in A. (C) In Fezl^-/- mutants, only layer 5 neurons within somatosensory cortex (arrowheads) retain projections to the spinal cord. (D) High-power view of the boxed region in C, showing sparse labeling in the mutant.

Anterograde Tracing Reveals Defects in Axon Targeting. PLAP labeling suggested that Fezl-expressing neurons form aberrant axonal projections in Fezl mutants. To further explore this issue, we injected the anterograde tracer BDA into one of three cortical areas (motor, somatosensory, or visual), then visualized axon projections in brain sections (Fig. 4; Fig. 11, which is published as supporting information on the PNAS web site). BDA labeling in Fezl mutants was distinct from that in control brains. In addition to callosal defects (Figs. ​(Figs.4B4B and 10B), neurons from somatosensory and visual cortex extended fewer axons into the superior colliculus (Figs. 4 C and D and 11 E and F). Dramatic changes were seen for motor and somatosensory cortex projections to the hindbrain and spinal cord; no labeled axons were visible in the pyramidal decussation or CST of Fezl^-/- mutants (Fig. 4 G-J).

Open in a separate window

Fig. 4.

Abnormal axon projections from the somatosensory and motor cortices of Fezl^-/- mice. Sagittal sections showing BDA-labeled axons traced from somatosensory (A-F) or motor cortex (G-J) of adult mice. (A and B) BDA injections into somatosensory cortex normally label axons that innervate neighboring cortical areas. Axons in Fezl^-/- mice fail to cross the corpus callosum and instead form Probst bundles (asterisk), and some labeled axon bundles enter rostral motor areas (arrowhead). (C and D) Control axons from somatosensory cortex project to the superior colliculus, which contains many fewer labeled axons in Fezl mutants. (E) Somatosensory axons innervate the reticular nucleus, the lateral part of the ventral posterior nucleus and the posterior nucleus of the thalamus. (F) Thalamic projections in Fezl^-/- mice show a grossly normal pattern, but more axons appear to innervate the medial ventral posterior nucleus, pretectum, and zona inserta. (G and H) BDA-labeled axons from motor cortex normally branch in the central pons, which contains many fewer axons in mutants. (I) Axons from motor cortex extend through the pyramidal tract, cross in the pyramidal decussation, then descend in the CST. (J) No axonal labeling was visible in the pyramidal decussation or CST of Fezl^-/- mutants. mot, motor cortex; wm, white matter; som, somatosensory cortex; hip, hippocampus; sc, superior colliculus; ic, inferior colliculus; rt, reticular thalamic nucleus; vp, ventral posterior nucleus; pt, pretectum; and zi, zona inserta. (Scale bar, 0.5 mm.)

The BDA-labeling patterns support the view that the axons of layer 5 neurons adopted alternative trajectories. As seen with PLAP, Fezl mutants showed an increased number of axons extending through the external capsule and anterior commissure, with a small number then extending dorsally into the contralateral cortex (Fig. 11D). A second projection was observed in the ipsilateral cortex, where BDA injections into somatosensory cortex of mutants labeled axon bundles extending into motor cortex (Fig. 4 A and B). Collectively, these experiments both confirmed the targeting defects revealed by PLAP and suggested that axons were redirected along inappropriate trajectories to novel targets.

We also examined thalamic projections after BDA injections. Regardless of the area injected, labeled axons appeared to choose appropriate thalamic targets in Fezl^-/- mice. For example, somatosensory axons were present in the reticular nucleus, the lateral part of the ventral posterior nucleus, and the pretectum in both control and mutant animals (Fig. 4 E and F), although stronger labeling was seen in the zona inserta of Fezl^-/- mice.

Fezl Regulates the Expression of Several Genes Expressed in CSMNs. To investigate the molecular mechanisms underlying the defects observed in the subcortical projections of layer 5 neurons, we used in situ hybridization to examine the expression of 125 genes involved in forebrain patterning, axon guidance, and the development of CSMNs. Mutations in genes such as netrin1 (21), DCC and Unc5h3 (22), and slit1/2 (23) cause defects in corpus callosum and pyramidal tract development; thus, we examined the expression of a variety of axon guidance molecules, including netrins and their receptors, Robo and slit family members, and semaphorins and plexins. Of 36 genes tested, no obvious differences in expression were detected at E14, E15, E17, or P0 (Fig. 12 and Table 2, which are published as supporting information on the PNAS web site), suggesting that most major axon guidance molecules were expressed normally in Fezl-deficient brains. A second explanation for the altered axon trajectories in Fezl mutants is that Fezl is required for patterning the environment through which axons migrate. However, examination of markers that distinguish the major dorsal and ventral domains of the forebrain revealed no differences between controls and mutants (data not shown; see Table 2 for list of probes).

Another explanation for the defective projections in Fezl^-/- mice is a failure in the specification or differentiation of subcortical projection neurons. Recent gene profiling experiments identified several genes, including Fezl, that are expressed in specific subtypes of layer 5 neurons (24). We explored the expression patterns of 35 of these genes and found that many, including Ctip2, Riken 2010001O09, Crim1, NFH, Lix1, and Synaptotagmin 9, showed a loss or reduction of expression in Fezl^-/- animals (Fig. 5, Table 1). However, not all CSMN genes were lost in mutants. Expression of the CSMN general identity gene Contactin 6 (Fig. 5 O and P) and the later-expressed gene Netrin G1 (data not shown) was maintained in Fezl-deficient layer 5 neurons, indicating that these neurons still occupy their correct positions and retain at least some aspects of CSMN differentiation. In addition, the expression of the CSMN-specific gene Igfbp4 was expanded into layer 6 in mutants (Fig. 5 G and H); thus, although the projections of layer 6 neurons appeared normal, their molecular identities were substantially altered.

Open in a separate window

Fig. 5.

Reduced expression of CSMN genes in Fezl mutant mice. Sagittal sections from P15 (A-L) or P0 (M-P) brains. (A and B) Ctip2 is normally expressed in layers 5 and 6 of cerebral cortex, hippocampus (hip) and striatum (str), but cortical expression is lost in Fezl^-/- mice. (C and D) Crim1 is expressed in layers 2, 3, and 5 in control brains, but layer 5 expression is lost in mutants. (E and F) Neurofilament heavy chain (NFH) in layer 5 is severely reduced in mutants. (G and H) Expression of Igfbp4 expands into layer 6 in Fezl mutants. (I and J) The expression of Lix1 in layer 5 is severely reduced in mutants. (K and L) Layer 5 cells express Synaptotagmin 9 in controls but not in Fezl mutants. (M and N) Expression of Riken 2010001O09 in layer 5 at P0 is lost in the Fezl mutants. (O and P) The CSMN general identity gene Contactin 6 is expressed normally in mutants. (Scale bar, 0.5 mm.)

The most striking loss of expression in Fezl^-/- mice involved the CSMN early development gene Ctip2 (24), which encodes a zincfinger TF. In control brains, Ctip2 was expressed in layer 5, weakly in layer 6, and in hippocampus and striatum (Fig. 5A). In Fezl^-/- mice, Ctip2 expression in striatum and hippocampus was maintained, but expression of Ctip2 in the cortex was completely abrogated (Fig. 5B). Previous work revealed defects in CST development in Ctip2^-/- mice, which show a phenotype remarkably similar to that of Fezl mutants, with CST axons failing to extend past the pons in both cases (24). That both Fezl and Ctip2 are expressed in layer 5 CSMNs, the similarity in their mutant phenotypes, and the absence of Ctip2 expression (as well as that of other deep layer-specific genes) in Fezl mutants suggest that Fezl acts upstream of Ctip2 to regulate multiple aspects of subcortical projection neuron differentiation (24).

Although much progress has been made in elucidating the mechanisms of neuronal fate determination in the vertebrate spinal cord (8, 9), little is known about the mechanisms that create specificity in higher brain regions such as the cerebral cortex. Here, we report that Fezl, a zinc-finger transcriptional repressor expressed in layers 5 and 6, regulates the differentiation and connectivity of layer 5 subcortical projection neurons. We found that in Fezl-deficient mice, Fezl-expressing deep-layer neurons failed to extend axons to the tectum, pons, and spinal cord and instead formed inappropriate projections. These defects could not be explained by alterations in the expression of axon guidance ligands or receptors or by changes in forebrain patterning. The generation and migration of deep-layer neurons appeared normal, because PLAP mRNA was robustly expressed in the deep layers of Fezl mutants, and many deep-layer markers (including many CSMN-specific genes) were expressed normally in mutants. These findings suggest that Fezl regulates specific features of deep-layer differentiation, including the expression of Ctip2, a TF that is essential for CST development (24).

The transcriptional control of cell fate decisions and differentiation plays a critical role in the elaboration of distinct neuronal phenotypes throughout the nervous system (8, 9). Among the genes affected by the Fezl mutation, several [such as the ETS gene ER81 (25)] regulate the differentiation and connectivity of other neuronal subtypes. Another Fezl target, Grg4, is a member of the Groucho family of transcriptional corepressors, which are recruited by homeodomain TFs to regulate progenitors fates along the dorsal-ventral axis of the spinal cord (9). Interference with repression suggests that ventral neuron identities in the spinal cord are achieved through the repression of repressors (26). Interestingly, the recruitment of Groucho corepressors is mediated by the engrailed homology (eh1) domains of homeodomain proteins (27). Fezl also contains an eh1 motif (12), raising the possibility that Groucho family members may interact with Fezl to regulate differentiation in the developing cortex. However, our data suggest that Fezl does not actively suppress the acquisition of alternative laminar identities during cortical development, because we saw no expansion of markers for upper layer neurons into the deep layers, nor did we detect an expansion of markers for layers 5 or 6 (apart from Igfbp4) into inappropriate layers.

Fezl mutant mice exhibit profound defects in the elaboration of subcortical axonal projections. The growth and refinement of subcortically projecting axons from layer 5 neurons in rodents have been studied extensively (1). In neonates, layer 5 neurons throughout the cortex extend simple unbranched axons into the spinal cord. The parental axons then branch and invade additional targets, including the pons, colliculi, and deep cerebellar nuclei. Later, axon branches are selectively eliminated in a manner appropriate for the areal position of each neuron (1). In Fezl-deficient mice, CST projections were not detected by PLAP or BDA labeling, although a very small number of cells could be retrogradely labeled from the pyramidal decussation. PLAP- and BDA-labeled axons were also observed in inappropriate regions, suggesting that Fezl-expressing neurons have diverted their axons to novel destinations (including the anterior commissure, rostral cortex, striatum, and dorsal thalamus) in Fezl mutants.

Although Fezl mutants displayed severe defects in the guidance and targeting of CST axons, molecules that regulate the corticospinal axon growth and guidance (1, 28-32), as well as related family members and receptors, were expressed normally in Fezl-deficient brains. Proper patterning of the basal forebrain and hypothalamus is also essential for pyramidal tract development (32), but examination of basal forebrain patterning in Fezl mutant mice revealed that ventral forebrain and hypothalamic markers were expressed normally. Thus, although we have not exhausted the list of all known axon guidance molecules, it is likely that the layer 5 subcortical projection defects are caused by a misregulation of guidance molecules that remain to be identified and/or the failure of axons to respond properly to known guidance cues.

The anatomical and morphological development of CSMNs has been studied extensively (33), but the molecular mechanisms that control their specification, differentiation, survival, and connectivity are largely unknown. The recent advent of gene expression profiling, coupled with the purification of layer 5 CSMN and corticotectal neurons, has enabled the identification of specific patterns of gene expression in CSMNs (24). This previous study showed that Fezl is expressed throughout CSMN differentiation, and that Ctip2, which is also expressed in CSMNs, is required for normal CST development. In Ctip2-deficient mice, corticofugal axons failed to extend past the pons to reach the spinal cord (24), a phenotype reminiscent of that in Fezl-deficient mice (although Ctip2 mutant mice die perinatally, precluding an analysis of mature patterns of connectivity). The similar phenotypes of Fezl-deficient and Ctip2 mutant mice suggest that these two genes may act in a common pathway to control CSMN development. The earliest expression of Ctip2 is observed in postmitotic neurons in lateral cortex at E14, with expression spreading medially over time (24). Fezl, in contrast, is first detected in the forebrain ventricular zone at E8.5, and expression remains strong during the production and differentiation of deep-layer neurons (19). This timing suggests that Fezl acts upstream of Ctip2 to control the differentiation of deep-layer neurons. Indeed, we found that in the absence of Fezl function, Ctip2 expression in the cerebral cortex was completely abrogated, even though expression in hippocampus and striatum was normal. Collectively, our data suggest that Fezl is a key upstream regulator of corticospinal projection neuron differentiation.

Note. After the completion of these experiments, a report suggesting that Fezl is required for specification of CSMNs and the formation of corticospinal projections was published by Molyneaux et al. (34).

Author contributions: B.C., L.R.S., and S.K.M. designed research; B.C. and L.R.S. performed research; B.C., L.R.S., and S.K.M. analyzed data; and B.C., L.R.S., and S.K.M. wrote the paper.

Conflict of interest statement: No conflicts declared.

Abbreviations: BDA, biotinylated dextran amine; CSMN, corticospinal motor neuron; CST, corticospinal tract; PLAP, human placental alkaline phosphatase; TF, transcription factor; Pn, postnatal day n; En, embryonic day n.