a, The locations of CNNEs (E1-4) analyzed in this study are indicated. b, Deletion of each CNEE from the Fezf2-Gfp BAC and transgenesis. A positive selection neomycin-cassette (Neo) flanked by homology arms was inserted by homologous recombination, resulting in the deletion of each CNNE. After the removal of Neo by flippase, the modified BACs (ΔE1-4) were used for transgenesis. c-g, Whole-mount brains of P0 control Fezf2-Gfp (c) and founder mutant (d-g) transgenic mice. n ≥ 3 founders per mutant line. In Fezf2-Gfp mice (c), GFP was expressed in the neocortex (Ncx), olfactory bulb (OB), hypothalamus (Hyp) (arrowhead), and CS axons in the pons (arrow). The deletion of E4 (g), but not E1-3 (d-f), led to a specific loss of GFP expression in the Ncx and CS axons (arrow), but not the OB or Hyp.

a, Generation of E4 knockout mice (E4^−/−). b, Duplex PCR genotyping of wildtype and mutant alleles using primers P1, P2, and P3. c, Analysis of neocortical Fezf2 expression by quantitative (q) RT-PCR. Normalized to Gapdh, Fezf2 mRNA levels were significantly reduced in the E4^−/− compared to heterozygous littermate controls (E4^+/−). (P = 2.1×10⁻⁶, one-tailed Student’s t-test, n = 3 per genotype). Error bars represent s.e.m. d, A schematic depiction of CS system. e-g, Sagittal (top row) and coronal (bottom row) sections of the pons from wildtype (e), cortex-specific Fezf2 knockout (f), and E4^−/− (g) mice immunostained for axon marker L1CAM (red) and PRKCG (green). The near complete loss of CS axons (arrowheads and dashed outlines) in the E4^−/− is a phenocopy of cortex-specific Fezf2 deletion. Scale bars represent 200 µm.

a, Chromatin immunoprecipitation (ChIP) from Neuro-2a cells expressing V5-tagged SOX4 and SOX11. Captured DNA was analyzed by PCR using primers specific for E1-4. SOX4 and SOX11 bound and recruited Pol II to E4 but not E1-3. b, Analysis of SOXC transactivation using empty (pGL4) or E1-4 containing luciferase vectors. The activity of E4, but not E1-3, was significantly increased by Sox4 (≥ 4 fold) or Sox11 (≥ 13 fold), but not Sox12 (≤ 1.5 fold). c-d, Analysis of functional elements within E4 using luciferase vectors containing deletion fragments of E4 (E4F1-4) (c). E4F2, but not the other fragments, was significantly activated by co-transfection of Sox4 (≥ 5 fold) or Sox11 (≥ 9 fold), but not Sox12 (≤1.5 fold) (d). e, Analysis of putative SOX binding sites (SB1-3) using mouse (Mo) E4F2 luciferase vectors mutagenized by substitution (blue lowercase nucleotides) with zebrafish (Ze) sequence. Targeted mutation of SB2 (MoE4F2-m2) significantly diminished the trans-activating ability of SOX4 or SOX11. f, Electrophoretic mobility shift assay (EMSA) with biotin-labeled SB2 DNA. SOX4-V5 and SOX11(1-276)-V5 shifted wildtype (arrowhead) but not mutated SB2 DNA. SOXC-SB2 complexes were super-shifted (open arrowhead) by an anti-V5 antibody. g, A schematic model of E4 regulation by SOX5 and SOXC. h-i, Analysis of competition between SOX5 and SOXCs using the E4-containing luciferase vector. Decreasing concentrations of co-transfected Sox5 led to a dose-dependent increase in E4 activation in response to Sox4 or Sox11 (h), whereas decreasing concentrations of co-transfected Sox4 or Sox11 led to a dose-dependent increase in Sox5 repression of E4 (i). One-tailed Student’s t-test; **P<0.01, ***P<0.001. n = 4 per condition. Error bars represent s.e.m.

a, The requirement of SOX4 and SOX11 for E4 transactivation was determined in neurons cultured from E14.5 heterozygous littermate control and Sox4;Sox11 cdKO mice and transfected with the E4 luciferase construct. The activity of E4 in double knockout neurons was not significantly above background (1.1 fold. b, Analysis of neocortical Fezf2 expression by qRT-PCR in cortex-specific Sox4 and/or Sox11 mutants. Normalized to Gapdh, Fezf2 mRNA levels were dramatically reduced in cdKO but not Sox4 or Sox11 mutants. c-d, Sagittal (top row) and coronal (bottom row) sections of the control and cdKO pons immunostained for L1CAM (red) and PRKCG (green). The dramatic loss of L1CAM and PRKCG-positive CS tract axons in the cdKO (arrowhead and dashed outline) is similar to that of the cortex-specific Fezf2 mutant. Errors bars represent s.e.m. One-tailed Student’s t-test; NS=Not significant, **P<0.01, ***P<0.001. n = 4 per genotype. Scale bars represent 200 µm.

a, Hierarchical clustering of E4F2 sequences from seven vertebrates. Percentage nucleotide identity relative to mouse E4F2 is indicated in parentheses. b, Analysis of species differences in Sox11 activation of E4F2. Sox11 trans-activated the E4F2 sequence of five vertebrates (≥ 7.3 fold) and Xenopus (2.8 fold). In contrast, the activity of zebrafish E4F2, which is divergent in SB1 and SB2, was not activated by Sox11. c, Murinization (red uppercase nucleotides) of zebrafish SB2 (ZeE4F2-m2) partially rescued the loss of transactivation of zE4F2 by Sox11. d-e, Cell-autonomous rescue of mouse Fezf2 loss-of-function by zebrafish Fezf2 (ZeFezf2). In utero electroporation (IUE) of Fezf2^fl/fl neocortical wall at E12.5 with Cre and CRE-responsive Gfp plasmids. Fezf2-deficient L5 neurons do not form CS tract at P0 (d). Co-electroporation of a ZeFezf2 plasmid cell-autonomously rescued the formation of CS tract by Fezf2-deficient neurons (e). Errors bars represent s.e.m. One-tailed Student’s t-test; *P<0.05, **P<0.01, ***P<0.001. n = 4 per condition. Scale bar represents 200 µm.