Inhibition of Sox2-dependent activation of Shh in the ventral diencephalon by Tbx3 is required for formation of the neurohypophysis

Abstract

Tbx2 and Tbx3 are two highly related members of the T-box transcription factor gene family that regulate patterning and differentiation of a number of tissue rudiments in the mouse. Both genes are partially co-expressed in the ventral diencephalon and the infundibulum; however, a functional requirement in murine pituitary development has not been reported. Here, we show by genetic lineage tracing that Tbx2(+) cells constitute the precursor population of the neurohypophysis. However, Tbx2 is dispensable for neurohypophysis development as revealed by normal formation of this organ in Tbx2-deficient mice. By contrast, loss of Tbx3 from the ventral diencephalon results in a failure to establish the Tbx2(+) domain in this region, and a lack of evagination of the infundibulum and formation of the neurohypophysis. Rathke's pouch is severely hypoplastic, exhibits defects in dorsoventral patterning, and degenerates after E12.5. In Tbx3-deficient embryos, the ventral diencephalon is hyperproliferative and displays an abnormal cellular architecture, probably resulting from a failure to repress transcription of Shh. We further show that Tbx3 and Tbx2 repress Shh by sequestering the SRY box-containing transcription factor Sox2 away from a Shh forebrain enhancer (SBE2), thus preventing its activation. These data suggest that Tbx3 is required in the ventral diencephalon to establish a Shh(-) domain to allow formation of the infundibulum.

Keywords: Diencephalon; Infundibulum; Neurohypophysis; Pituitary gland.

Fig. 1.

Genetic lineage tracing of Tbx2⁺ cells in the developing hypothalamus-pituitary axis. (A-G) Immunofluorescent detection of GFP (green) of the R26^mTmG reporter allele (A-G) and co-staining of Emcn (red) (G) on midsagittal (A,B,D) and coronal sections (C,E,F,G) of Tbx2^cre/+;Rosa^mTmG/+ embryos at the indicated stages. Section planes of C,E,F are indicated by white lines in B,D. The region depicted in G is indicated in F. Blue, DAPI. inf, infundibulum; pd, pars distalis; pn, pars nervosa; pt, pars tuberalis; rp, Rathke's pouch; vd, ventral diencephalon.

Fig. 2.

Loss of Tbx3 affects morphogenesis of the pituitary gland. Hematoxylin and Eosin staining of midsagittal sections of control (wt) and Tbx3KO embryos at the indicated stages. ah, adenohypophysis; inf, infundibulum; rp, Rathke's pouch; vd, ventral diencephalon.

Fig. 3.

The ventral diencephalon is hyperproliferative in Tbx3-deficient embryos. (A,B) Analysis of cell death by the TUNEL assay (A) and of cell proliferation by the BrdU incorporation assay (B) on midsagittal sections of control and Tbx3KO embryos. Cranial is to the left. Stages are as indicated in the figure. (B) False color overlay of DAPI (blue) and BrdU signals (red). The domains analyzed for proliferation correspond to the Tbx3 expression domain and are labeled within the figures. (C) Quantification of BrdU-positive cells. E9.5 (n=5), wild type (wt) versus mutant: vd: 36.7±3.3% versus 44.7±2.7%, P=0.003; rp: 47.2±4.1% versus 40.2±4.6%, P=0.036. E10.5 (n=3), wild type versus mutant: vd: 13.9±1.6% versus 46.7±3.3%, P=0.0001; rp: 42.1±2.1% versus 34.1±3.7%, P=0.03. Error bars indicate s.d. *P<0.05; **P<0.01; two-tailed Student's t-test. (D) In situ hybridization analysis on midsagittal sections of E10.5 control and Tbx3KO embryos. rp, Rathke's pouch; vd, ventral diencephalon.

Fig. 4.

Molecular analysis of the developing adenohypophysis in Tbx3KO embryos. In situ hybridization analysis on midsagittal sections of control (wt) and Tbx3KO embryos at the indicated stages. Cranial is to the left. ah, adenohypophysis; rp, Rathke's pouch.

Fig. 5.

Molecular analysis of the developing ventral diencephalon in Tbx3KO embryos. In situ hybridization analysis on midsagittal sections of E10.5 (A) and E9.5 (B) control and Tbx3KO embryos. Cranial is to the left. rp, Rathke's pouch; vd, ventral diencephalon.

Fig. 6.

TBX2 and TBX3 repress SBE2 activation by SOX2. (A) Schematic of Flag-tagged SOX2 and HA-tagged TBX2 showing the position of functional domains: SOX2, high mobility group/DNA-binding domain (HMG), activation domain (AD); TBX2, T-box DNA-binding domain (DB), repression domain (RD), as well as site mutations. (B) Luciferase reporter assays performed in COS-1 cells co-transfected with SBE2-luciferase and expression constructs for SOX2, TBX2/3 or both. Asterisk indicates statistically significant differences from transfections with SOX2 alone (P<0.05, Student's t-test). (C) Co-IP experiments with Flag-tagged SOX2 and HA-tagged TBX2 normalized for equivalent expression levels in COS-1 cells (lanes 6, 7). The empty vector (pcDNA3.1) serves as a negative control (lanes 1-5). Input lanes (2, 4, 6) serve as positive controls. IPs were performed with an anti-HA antibody (upper panel), followed by western blot (WB) with an anti-Flag antibody, or vice versa (lower panel). (D) Co-IP experiments with Flag-tagged SOX2 containing different point mutations in the HMG domain (R100E/M104E; R115E; or R100E/M104E/R115E) and HA-tagged TBX2, co-transfected in COS-1 cells. Co-IP was performed with anti-HA antibody followed by western blot with anti-Flag antibody. (E) Co-IP experiments with HA-tagged TBX2 containing mutations in the DNA-binding domain (R122E/R123E) or a deletion of the repression domain (ΔRD) and Flag-tagged SOX2 expressed in COS-1 cells. Co-IP experiments were performed with an anti-Flag antibody followed by western blot with an anti-HA antibody. (F) Luciferase reporter assays performed in COS-1 cells co-transfected with SBE2-luciferase and expression constructs for SOX2, and different variants of TBX2 that disrupt complex formation with SOX2. Error bars in all graphs represent s.e.m.

Fig. 7.

SOX2 binding to SBE2 is blocked by TBX2. (A) ChIP-QPCR results from COS-1 cells co-transfected with an SBE2-luciferase construct and either Flag-tagged SOX2, isoforms of HA-tagged TBX2, or both. QPCR results from three independent experiments reveal a significant enrichment of SBE2 DNA in SOX2 (black)- and TBX2^wt (gray)-bound chromatin when transfected independently, but not together, compared with IgG controls (white) (Student's t-test, *P<0.05). TBX2^R122E/R123E was not recruited to SBE2 and had no influence on SOX2 binding. TBX2^ΔRD was enriched on SBE2 but did not affect SOX2 recruitment. (B) Top: Schematic showing position of candidate T-box binding sites in SBE2 in relation to the Sox2 site (SoxB1). Asterisk marks nucleotide mismatches with T-box consensus sequence. Bottom: EMSAs performed with COS-1 cell extracts transfected with HA-tagged TBX2 or TBX2^R122E/R123E (negative control) and radiolabeled probes overlapping a consensus T-box binding site (Cons T-box bs) or candidate SBE2 T-box site1. TBX2 complexes migrate similarly when bound to T-box bs or SBE2 T-box site1 (lower arrowheads). Supershifted complexes form in the presence of anti-HA antibody (upper arrowheads). (C) ChIP-QPCR results from COS-1 cells co-transfected with an SBE2-luciferase construct containing mutations in T-box site1 and Flag-tagged SOX2 (black), HA-tagged TBX2 (gray) or both. (D) Luciferase reporter assays performed in COS-1 cells co-transfected with SBE2-luciferase constructs (wt, or SBE2 T-box site1 mutation), and expression constructs for SOX2, TBX2 or both. Asterisk indicates statistically significant differences from transfections with SOX2 alone (P<0.05, Student's t-test). (E) Schematic showing the transcriptional regulation of Shh in the infundibular region of the ventral diencephalon. For details see text.