Disruption of SoxB1-dependent Sonic hedgehog expression in the hypothalamus causes septo-optic dysplasia

Abstract

Septo-optic dysplasia (SOD) is a congenital brain anomaly that results in pituitary, optic nerve, and midline forebrain defects. The etiology of SOD is poorly understood, with the majority of cases being sporadic. In rare instances, SOD is caused by mutations in Sox2, Sox3, or Hesx1, but how this manifests in disease is not entirely certain. We demonstrate here that mouse embryos lacking Sonic hedgehog (Shh) in the prospective hypothalamus exhibit key features of SOD, including pituitary hypoplasia and absence of the optic disc. The hypothalamic source of Shh is required to maintain gene expression boundaries along the anteroposterior and mediolateral neural axes that are important for proper pituitary and eye development, respectively. We further reveal that Sox2 and Sox3 are dose-dependent regulators of Shh transcription that directly bind and activate a long-range Shh forebrain enhancer. These data indicate that reduced levels of Shh expression in the hypothalamus cause SOD.

Figure 1. The hypothalamic source of Shh is required for proper pituitary morphology at E12.5

(a,b,d,e) Whole mount in situ hybridization for Shh and Gli1 on bisected brains from control (SBE2cre; Shh^+/loxp) and Shh^Δhyp (SBE2cre; Shh^−/loxp) embryos. In the hypothalamus of control embryos, Shh is expressed in progenitors of the ventromedial nucleus (VM). Shh is missing from the VM domain of Shh^Δhyp embryos, but other forebrain regions, including the zona limitans intrathalamica (ZLI) and preoptic area (POA) maintain Shh expression. Gli1 marks Shh-responsive cells in ventral and anterior hypothalamic nuclei, including the mammillary (MM), premammillary (PM), arcuate (Arc), suprachiasmatic (SCN), anterior hypothalamus posterior region (AHP), and anterior hypothalamus anterior region (AHA). With the exception of the AHA and MM, all other hypothalamic regions show reduced Gli1 expression in Shh^Δhyp embryos. (c,f) Nkx2.1 immunostaining on midsagittal sections through control and Shh^Δhyp embryos counterstained with DAPI (blue). Nkx2.1 staining is present in the ventral hypothalamus and medial ganglionic eminence (MGE) of Shh^Δhyp embryos. However, the posterior pituitary (PP) is dysmorphic and ectopically positioned compared to control embryos. (g,j) In situ hybridization for Tbx2 on midsagittal sections confirms the ectopic PP in Shh^Δhyp versus control embryos. (h,k) Double labeling of the POA and PP with Foxg1 and Tbx2, respectively, shows that the intervening region corresponding to the retrochiasmatic hypothalamus (RCH) is reduced in Shh^Δhyp embryos. (i,l) Nkx2.1 and Lhx3 immunostaining reveals that the anterior pituitary (AP) is duplicated and shifted anteriorly in Shh^Δhyp mutants. D, dorsal; A, anterior.

Figure 2. Shh is required for proper anteroposterior patterning in the hypothalamus

(a) Diagram of a midsagittal section through a control E10.5 brain delineating the approximate expression domains of anteroposterior patterning markers in the hypothalamus. Bmp4 and Fgf10 are co-expressed in the posterior ventral midline (blue), while Shh, Six6, and Tcf4 mark the anterior ventral midline (red). (b,c) At E9.5, Bmp4 expression is rostrally expanded (brackets) in Shh^Δhyp embryos. Bmp4 is also ectopically expressed in the oral ectoderm of mutants (arrow). (d,e) Fgf10 expression is expanded rostrally in Shh^Δhyp versus control embryos. Notice that the vertical line extending from the optic recess to the rostral extent of the Fgf10 expression domain (horizontal line) is truncated in Shh^Δhyp compared to control embryos. (f,g) Shh transcription is repressed in the hypothalamus of Shh^Δhyp embryos. (h, i) Six6 expression is downregulated in the hypothalamus of Shh^Δhyp embryos. (j, k) Tcf4 expression is unaffected in the hypothalamus of Shh^Δhyp embryos. D, dorsal; A, anterior.

Figure 3. The optic disc fails to form in Shh^Δhyp embryos

(a–j) Coronal sections through the eyes of control and Shh^Δhyp embryos at E14.5. (a,b) RGC axons labeled with neurofilament (NF) bundle and exit the eye at the optic disc in control, but not Shh^Δhyp embryos. (c–f) Pax2 and Netrin1 mark cells of the optic disc (arrowheads) and optic stalk (arrows). The optic disc is absent in Shh^Δhyp mutants. (g–j) Shh and Gli1 expression is unaltered in the eyes of Shh^Δhyp versus control embryos. (k–v) Coronal sections through the hypothalamus of E10.5 embryos at the level of the eye. (k–n) The expression of Shh in the ventral midline of the hypothalamus and Gli1, in the optic stalk (OS) is lost in Shh^Δhyp mutants. (o–t) The expression of Vax1, Pax2, and Netrin 1 is reduced in the ventral retina (below dotted line) of Shh^Δhyp mutants, but is preserved in the optic stalk. Pax6 staining in the dorsal retina (above dotted line) expanded ventrally in Shh^Δhyp embryos. (u,v) The weak Bmp7 expression in the ventral midline of the hypothalamus, optic stalk, and ventral retina in control embryos is absent in Shh^Δhyp mutants. (w) Diagram of the eye and optic stalk in control and mutant, tilted to expose the ventral optic fissure (OF). The fissure and invagination of cells (curved arrows) does not extend to the stalk in mutants. The level of section in panels x-cc is shown. Labels: dR, dorsal retina; L, lens; vR, ventral retina; dOS, dorsal optic stalk; vOS, ventral optic stalk. (x–cc) Sections cut parallel to the eye surface at E10.5. Pax2⁺ cells invaginate at the lips of the optic fissure (arrows) in control and Shh^Δhyp at surface and mid-cup levels. However, Pax2⁺ optic disc precursors at the interface between the eye cup and stalk (arrowhead) fail to invaginate in Shh^Δhyp. D, dorsal; L, lateral; A, anterior.

Figure 4. SoxB1 proteins are dose dependent regulators of Shh expression in the anterior hypothalamus

(a–d) Coexpression of Shh, Sox2 and Sox3 on midsagittal sections through the hypothalamus of wild type embryos at E10.5. (e–p) In situ hybridization for Shh (e–h), Fgf10 (i–l) and Six6 (m–p) on midsagittal sections through the hypothalamus of control and SoxB1 mutant embryos at E10.5. The position of the infundibulum is marked with a dashed line, and the arrows point to the posterior boundaries of Shh and Six6 expression. A subtle reduction in the intensity and posterior demarcation of Shh and Six6 expression was detected in Sox3^Y/− and Sox3^+/−; Sox2^+/− mutants, concomitant with an anterior expansion in Fg10 expression. Sox3^Y/−; Sox2^+/− mutants showed a profound loss of Shh and Six6 expression in the anterior hypothalamus and corresponding gain in Fgf10.

Figure 5. Sox2 binds directly to SBE2 and is necessary for its activation

(a) Schematic map showing the distribution of genes/exons (black rectangles) and regulatory elements (colored rectangles) within the 1Mb genomic region upstream of Shh on mouse chromosome 5. The SoxB1 binding site (red) in SBE2 is highly conserved across species. (b) EMSAs performed with Cos-1 cell extracts transfected with pcDNA3-Flag (lane 1) or pcDNA3-Flag-Sox2 (lanes 2–11) expression vectors, incubated with a radiolabeled SBE2 wild type (WT) probe. The arrow points to the specific protein/DNA complex. Increasing concentrations of unlabeled wild type probes overlapping the SoxB1 site in SBE2 (lanes 3–5), or MHP (lanes 6–8) effectively competed for Sox2 binding, whereas the mutated SBE2 (ΔSoxB1) probe did not (lanes 9–11). (c) Sox2 regulates SBE2 activity in vitro. Cos-1 cells were co-transfected with either WT (blue) or ΔSoxB1 (magenta) versions of SBE2-luciferase constructs and increasing amounts of pcDNA3-Flag-Sox2 plasmids. Sox2 activated SBE2-luc (WT) in a dose-dependent manner, but not SBE2-luc (ΔSoxB1). Each point on the curve represents an average of three experiments performed in triplicate (*p<0.02, Student’s t-test). (d) Human mutations in Sox2 affect SBE2 binding and/or activation. Top: Schema of Sox2 protein showing amino acid substitutions arising from point mutations in either the HMG DNA-binding or transactivation domains. Middle: Cos-1 cell lysates transfected with pcDNA3-Flag (lane 1), pcDNA3-Flag-Sox2 (lane 2), or pcDNA3-Flag-muSox2 (lanes 3–5) were analyzed for binding to a radiolabeled SBE2(WT) probe. No complex formation was observed for the L97P mutant (lane 4), whereas, the A191T and Q177X mutants (lanes 3 and 5) showed normal or increased DNA binding, respectively. αFlag immunoblot demonstrates that wild type and mutant Sox2 proteins were expressed at equivalent levels, although the size of Q177X is smaller than WT. Bottom: Mutant Sox2 proteins showed significantly impaired transactivation of the SBE2 (WT) luciferase reporter construct. Each bar represents the average of three experiments performed in triplicate. Asterisk represents a statistically significant difference from wild type (p<0.05). (e) ChIP from embryos using anti-Sox2, Sox3 or IgG antibodies. QPCR results from three independent experiments revealed SBE2 enrichment in Sox2 and Sox3 bound chromatin isolated from forebrain and posterior trunk regions of E10.5 mouse embryos (*p<0.01, Student’s t-test). A control sequence, 6.5 kb downstream of SBE2, was not enriched in Sox2 or Sox3 bound chromatin. Error bars in all graphs represent standard error of the mean.

Figure 6. SBE2 activity in the hypothalamus is dependent on a highly conserved SoxB1-binding site

X-gal staining of representative transgenic embryos carrying either wild type (WT; a,c,e,g) or SoxB1 deleted (ΔSoxB1; b,d,f,h) SBE2-lacZ reporter constructs at E10.5. The dashed lines in (a,b) show the plane of section in panels (c–h). The number of embryos showing representative reporter activity over the total number of transgenic embryos is indicated for each construct (a,b).