Edited by Richard M. Harland, University of California, Berkeley, CA, and approved December 7, 2015 (received for review February 17, 2015)
The relevance of transcription factor 7-like 1 (TCF7L1) during hypothalamo–pituitary (HP) axis development remains unknown. Using mouse genetics, we show that TCF7L1 acts as a transcriptional repressor to regulate the expression of the hypothalamic signals involved in pituitary formation. In addition, we screened a cohort of human patients with forebrain and/or pituitary defects and report two independent missense variants, p.R92P and p.R400Q, in human TCF7L1. Functional studies in vitro and rescue experiments in zebrafish mutants deficient for tcf7l1a and tcf7l1b show that the p.R92P and p.R400Q variants exhibit reduced repressing activity compared with wild-type TCF7L1. In summary, we identify TCF7L1 as a determinant for the establishment of HP axis development and as a potential candidate gene to be mutated in congenital hypopituitarism.
Aberrant embryonic development of the hypothalamus and/or pituitary gland in humans results in congenital hypopituitarism (CH). Transcription factor 7-like 1 (TCF7L1), an important regulator of the WNT/β-catenin signaling pathway, is expressed in the developing forebrain and pituitary gland, but its role during hypothalamo–pituitary (HP) axis formation or involvement in human CH remains elusive. Using a conditional genetic approach in the mouse, we first demonstrate that TCF7L1 is required in the prospective hypothalamus to maintain normal expression of the hypothalamic signals involved in the induction and subsequent expansion of Rathke’s pouch progenitors. Next, we reveal that the function of TCF7L1 during HP axis development depends exclusively on the repressing activity of TCF7L1 and does not require its interaction with β-catenin. Finally, we report the identification of two independent missense variants in human TCF7L1, p.R92P and p.R400Q, in a cohort of patients with forebrain and/or pituitary defects. We demonstrate that these variants exhibit reduced repressing activity in vitro and in vivo relative to wild-type TCF7L1. Together, our data provide support for a conserved molecular function of TCF7L1 as a transcriptional repressor during HP axis development in mammals and identify variants in this transcription factor that are likely to contribute to the etiology of CH.
Congenital hypopituitarism (CH) is a complex condition defined by the deficiency of one or more pituitary hormones and can be present in isolation or as part of a syndrome (1⇓–3). Septooptic dysplasia (SOD) is a rare form of CH (1 in 10,000) that manifests in conjunction with defects in the telencephalon (e.g., corpus callosum and septum pellucidum) and/or eyes (e.g., optic nerve hypoplasia) and is associated with high morbidity and occasional mortality (4).
Abnormal embryonic development of the hypothalamo–pituitary (HP) axis is a major cause of CH. The pituitary gland, comprising the anterior and posterior lobes, is an organ of dual embryonic origin, and derives from oral and neural ectoderm, respectively. In the mouse at around 9.5 d postcoitum (dpc), secreted signals from the hypothalamic primordium such as fibroblast growth factor (FGF) 8, FGF10, bone morphogenetic protein 4 (BMP4), and sonic hedgehog (SHH) induce Rathke’s pouch (RP), the primordium of the anterior pituitary (5, 6). Upon specification, RP progenitors proliferate rapidly, exit the cell cycle, and activate the expression of cell fate commitment genes such as POU domain, class 1, transcription factor 1 [Pou1f1 (Pit1)], Tpit, and Sf1. This process culminates in the differentiation of all hormone-producing cells by the end of gestation (7, 8): somatotrophs [growth hormone (GH)], lactotrophs [prolactin (PRL)], thyrotrophs [thyroid-stimulating hormone (TSH)], gonadotrophs [follicle-stimulating hormone (FSH) and luteinizing hormone (LH)], corticotrophs [adrenocorticotropic hormone (ACTH)], and melanotrophs [melanocyte-stimulating hormone (MSH)] (2, 9). A small proportion of cells remain in the postnatal anterior pituitary as tissue-specific stem cells, which have the capacity to self-propagate and differentiate in vivo and in vitro (10⇓–12).
Mutations in several developmental genes regulating HP development in the mouse have been identified in patients suffering from CH and SOD (4). One of these is the homeobox gene Hesx1/HESX1, which acts as a transcriptional repressor of the WNT/β-catenin signaling pathway (13, 14).
The stability of β-catenin, which is a strong transcriptional activator devoid of a DNA binding domain, is critical for normal WNT/β-catenin function (15⇓–17). In the absence of WNT ligands, β-catenin is phosphorylated by a destruction complex and degraded rendering the pathway inactive (18, 19). The binding of WNT ligands leads to β-catenin dephosphorylation and stabilization (20, 21). β-Catenin can then translocate into the nucleus and interact with mainly, but not exclusively, members of the T-cell factor/lymphoid enhancer factor (TCF/LEF) family to activate the expression of target genes (22, 23). In mammals, TCF/LEF factors include TCF7 (formerly TCF1), transcription factor 7-like 1 (TCF7L1) (formerly TCF3), TCF7L2 (formerly TCF4), and LEF1, which all bind the same consensus DNA motif 5′-(A/T)(A/T)CAAAG-3′ and have a β-catenin–interacting domain at the N terminus (24, 25). It is believed that, in the absence of stable β-catenin, TCF/LEF factors can repress target genes of the pathway through interactions with corepressors (26, 27), although cell context-specific functions have been demonstrated where specific factors are mostly required as either activators (28, 29) or repressors (30⇓⇓–33).
In the developing HP axis, this pathway coordinates proliferation of RP progenitors and differentiation of hormone-producing cells (34⇓⇓⇓⇓⇓⇓–41). TCF7L1 has been shown to play critical roles in maintenance of stem cell pluripotency (42, 43), tissue homeostasis of the skin epithelia (44), cell lineage determination during gastrulation (45, 46), and brain development in vertebrates (47, 48). However, the function of this important transcriptional factor during HP axis development has not been investigated to date. Moreover, despite the solid evidence demonstrating the critical function of the WNT/β-catenin pathway during HP axis development in the mouse, to date, mutations in components of this pathway have not been identified in patients with CH. In this study, we reveal that the β-catenin–independent repressor activity of TCF7L1 is required for normal HP axis formation in mice and humans.
The Hesx1Cre/+ (49), Tcf7l1flox/flox (44), and Tcf7l1ΔN/ΔN (33) have been previously described. We have previously shown that the Hesx1-Cre mouse line drives efficient recombination of loxP-flanked DNA in RP progenitors from 9.5 dpc (49, 50). Hesx1Cre/+;R26loxP-YFP/+ 18.5 dpc embryos display widespread yellow fluorescent protein (YFP) expression in the vast majority of cells of the anterior pituitary, suggesting that the Hesx1-expressing RP progenitors give rise to all hormone-producing cells (Hesx1-cell lineage). In addition to RP, Hesx1Cre/+;R26loxP-YFP/+ and Hesx1Cre/+;R26loxP-lacZ/+ embryos show reporter expression in the ventral telencephalon and preoptic hypothalamic area from 9.5 dpc (49⇓–51). Tcf7l1 is expressed in the forebrain, including the hypothalamic area, as well as in the anterior pituitary until 14.5 dpc (37). To reveal the function of Tcf7l1 during normal HP axis development, we used the Tcf7l1flox/flox conditional mouse line (44, 47). To generate pituitary-specific Tcf7l1 genetic deletions, Tcf7l1flox/flox mice were initially crossed to β-actin:Cre animals (52) to generate Tcf7l1+/− heterozygous mice. These mice were subsequently crossed to Hesx1Cre/+ to generate Hesx1Cre/+;Tcf7l1+/− double heterozygotes, which upon crossing with Tcf7l1flox/flox mice, were used to generate the Hesx1Cre/+;Tcf7l1flox/− mice and embryos analyzed in this study. Genotyping of mice and embryos was carried out by PCR on ear punch biopsies or pieces of tissue from embryos digested in DNAreleasy (Anachem) as per manufacturer’s instructions. The data presented in this work are representative of examples of at least three individual embryos per genotype. All of the experiments performed in mice were carried out according to UK Home Office guidance and approved by a local ethics committee.
Patients with CH disorders were recruited into the study from both national and international pediatric and adult endocrinology centers between 1998 and 2005. A total of 215 probands (129 male, 86 female) were screened for mutations within TCF7L1. These included 31 patients with CH without any midline or eye defects, 6 with holoprosencephaly (HPE), 148 patients with SOD (characterized by optic nerve hypoplasia in association with hypopituitarism and/or midline defects), and 30 patients with anophthalmia (n = 12) or microphthalmia (n = 18). Ethics committee approval was obtained from the Institute of Child Health/Great Ormond Street Hospital for Children Joint Research Ethics Committee (Institute of Child Health, London, United Kingdom). Informed written consent was obtained from the parents and, where applicable, the patients before collection of samples and genomic analysis.
The entire coding region of TCF7L1 (NM_031283) was PCR-amplified and subjected to direct-sequencing in our cohort. Detailed PCR and sequencing conditions are in SI Materials and Methods. For any mutations identified, control databases were consulted, including 1000 Genomes (www.1000genomes.org), Single Nucleotide Polymorphism Database (dbSNP), Exome Variant Server (EVS), and the Exome Aggregation Consortium (ExAC) (comprising >61,000 genomes) Browser (exac.broadinstitute.org). Patients 1 and 2 were screened and found not to carry mutations in other relevant genes involved in CH and SOD, specifically patient 1 was screened for mutations in HESX1, CHD7, KAL1, PROKR2, FGFR1, and FGF8; and patient 2 was screened for mutations in HESX1, SOX3, KAL1, PROKR2, FGFR1, and FGF8.
Histological processing of embryos and in situ hybridization on paraffin sections was performed as previously described (53, 54). The antisense riboprobes used in this study [Gh, glycoprotein hormone α (Cga), Pomc1, Prop1, Lh, Tsh, Fgf10, Fgf8, Bmp4, LIM homeobox protein 3 (Lhx3), Bmp4, T-box 2 (Tbx2), T-box 3 (Tbx3), and Pit1] have been described (51, 53⇓–55). Human TCF7L1 riboprobe was generated using a full-length cDNA cloned into pCMV-Sport6 (SalI 5′, NotI 3′) (clone no. IRATp970D0681D, IMAGE no. 6141641; SourceBioscience).
Embryos were fixed in 4% (wt/vol) paraformaldehyde and processed for immunodetection as previously described (11, 51). Detection of hormones was carried out using antibodies for (α-ACTH) (10C-CR1096M1), α-TSH (NHPP AFP-1274789), α-PRL (NHPP AFP-425-10-91), α-LH (NHPP AFP-C697071P), and α-FSH (AFP-7798-1289) (Developmental Studies Hybridoma Bank) at a 1:1,000 dilution. α-SHH (AF464; R&D Systems) and α-TCF7L1 (33) were used at a 1:100 dilution and signal amplified using the Tyramide Signal Amplification kit (NEL741001KT; Perkin-Elmer). α-Cleaved Caspase 3 was used at a 1:300 dilution (9661; Cell Signaling Technologies) and α-phosphorylated-histone H3 at a 1:1,000 dilution (06-570; Upstate). The mitotic index of pituitary periluminal progenitors represents the percentage of proliferating cells (phosphorylated histone H3-positive cells) from total number of DAPI-positive nuclei. Anti-phosphorylated histone H3 (1:1,000) (rabbit polyclonal; Upstate) was used for this study.
HEK293T cells were grown in DMEM supplemented with 10% FBS. Two reporters were used to assess the repressing activity of human wild-type TCF7L1 and the p.R92P and p.R400Q variants: (i) the TOPflash reporter plasmid carrying six wild-type binding sites for the TCF/LEF1 factors upstream of the firefly luciferase gene; (ii) the LEF1-promoter-luc reporter, which contains a DNA fragment (−6713 to −1 bp) of the human LEF1 promoter inserted upstream of a luciferase reporter gene (33). Approximately 250,000 cells were seeded per well of 24-well tissue culture plates and the following day subjected to DNA transfection. A total of 100 ng of TOPflash or 200 ng of LEF1-promoter-luc reporter were cotransfected with constructs expressing wild-type and mutant hTCF7L1 proteins (10 and 50 ng) and 80 ng of Renilla luciferase plasmid as internal control using Lipofectamine 2000 (Life Technologies) according to the manufacturer’s protocol. The total amount of DNA was kept constant in all of the transfections by complementing with pBluescript plasmid up to 350 ng. Activation of the TOPflash reporter was achieved by culturing the cells in the presence of 10 μM SB216763 (Sigma), a potent GSK3-β inhibitor, resulting in the stabilization of β-catenin. The LEF1-promoter-luc was activated with 30 μM SB216763. The repressing activity of wild-type hTCF7L1 and the variants was assessed 48 h posttransfection using the Dual-Luciferase Reporter Assay System (Promega) following the manufacturer’s instructions. Total levels of repression caused by the hTCF7L1 p.R92P and p.R400Q proteins were represented as a percentage of the maximum repression caused by wild-type hTCF7L1. Experiments were repeated four times in triplicates and statistical analysis was done using one-way ANOVA with Tukey’s post hoc analysis.
Zebrafish were bred and raised under standard conditions. hdl/tcf7l1am881−/−;tcf7l1bzf157Tg+/− (from now on tcf7l1a−/−; tcf3l1b+/−) zebrafish embryos were generated from crossing tcf7l1a+/−;tcf7l1b+/− males with tcf7l1a+/− females (56), raised at 28 °C and staged according to Kimmel et al. (57) Embryos were microinjected with 100 pg (5 nanoliters) of mRNA encoding either wild-type, p.R92P, or p.R400Q hTCF7L1 at one to two cell stage. mRNA encoding wild-type or mutant hTCF7L1 proteins was coinjected with mRNA encoding GFP, and embryos with low or no GFP fluorescence were excluded from the analysis. tcf7l1a alleles were genotyped by KASP assays (LCG Group). The tcf7l1b viral insertion was identified by genomic DNA PCR using the following primers: GSP2 forward, 5′-GGATACCGTGTGATCGTTTCTC-3′; and LTR2 reverse, 5′-TCTGTTCCTGACCTTGATCTGA-3′.
Mendelian ratios were evaluated using the χ2 test. For the luciferase assays, a one-way ANOVA was used, and for the zebrafish experiments, an unpaired t test was conducted. Values less than 0.05 were considered statistically significant. Quantitative data are presented as means ± SEM. Clonogenic potential of control and Hesx1Cre/+;Tcf7l1+/− mutant stem cells was evaluated using a paired t test. Total cell counts of control and Hesx1Cre/+;Tcf7l1+/− mutant pituitaries were analyzed by unpaired t test.
Genotyping analysis of 9.5–18.5 dpc embryos from crosses between Hesx1Cre/+;Tcf7l1+/− and Tcf7l1flox/flox showed a normal Mendelian distribution of genotypes, implying that the lack of Tcf7l1 in the Hesx1 cell lineage is not embryonic lethal (Table 1). Hesx1Cre/+;Tcf7l1flox/− mutant embryos displayed variable degrees of anterior forebrain defects, including eye defects (i.e., microphthalmia or anophthalmia) and telencephalic abnormalities (i.e., small or absent telencephalic vesicles), as previously described (47) (Fig. S1C). Additionally, we observed neural tube closure defects resulting in exencephaly in around 10% of these mutants (Fig. S1B). Genotyping analyses of mice at 3 wk of age revealed no significant differences of the expected Mendelian ratios for the Hesx1Cre/+;Tcf7l1flox/− mutants compared with control littermates, but perinatal death was observed in some Hesx1Cre/+;Tcf7l1flox/− mice (Table 1). Most of the surviving mutants did not exhibit any gross morphological defects, but around 25% of these showed dwarfism with reduced size and weight compared with their littermates, suggesting a potential functional compromise of the HP axis (Fig. S1 A, D, and E). We observed unilateral microphthalmia with no compromise of the telencephalic vesicles in around 20% of the Hesx1Cre/+;Tcf7l1flox/+ embryos as previously reported (47).
Genotypes of offspring from Hesx1Cre/+;TCF7L1+/− x Tcf7l1fl/fl genetic crosses
Conditional deletion of Tcf3/Tcf7L1 results in forebrain defects and partially penetrant dwarfism. (A) Hesx1Cre/+;Tcf7l1fl/− mutant adult mouse exhibiting reduced body size compared with a control wild-type littermate. (B and C) Wild-type and Hesx1Cre/+;Tcf7l1fl/− mutant embryos showing exencephaly. (D and E) Growth graphs of Hesx1Cre/+;Tcf3fl/− mutants and control embryos. Note the severe growth failure in the mutants (triangle) relative to the control littermates (circles).
Histological and in situ hybridization analyses of Hesx1Cre/+;Tcf7l1flox/− mutant and control embryos at 17.5 dpc revealed two clearly discernable pituitary phenotypes. (i) Group 1 embryos, ∼25% of the Hesx1Cre/+;Tcf7l1flox/− mutants, displayed severe morphological defects and ectopic pituitary tissue was often observed in the roof of the oropharyngeal cavity (Fig. 1 A′′–H′′). In these mutants, morphologically distinguishable posterior or intermediate lobes could not be observed. This phenotype was associated with severe forebrain defects. Hormone-producing cells were detected in these mutants, even in the ectopically located pituitary tissue. (ii) Group 2 embryos, accounting for ∼75% of Hesx1Cre/+;Tcf7l1flox/− mutants, showed a less severe phenotype with mild pituitary hyperplasia and presence of pituitary cleft bifurcations (Fig. 1 D′–H′). Posterior and intermediate lobes appeared morphologically normal in mutants in this group. Total cell counting of dissociated pituitaries at 18.5 dpc revealed average cell counts of 75,300 ± 5,227.33 for the Hesx1Cre/+;Tcf7l1flox/− mutants (n = 5) and 66,200 ± 7,585.62 for controls (n = 8). Although these differences did not reach statistical significance (P = 0.0584), the data suggest a trend toward mild hyperplasia in group 2 mutants. Differentiation of hormone-producing cells occurred normally with no apparent differences between Hesx1Cre/+;Tcf7l1flox/− group 2 mutants and control embryos. The Sox2-positive stem cell compartment was analyzed in vitro by culturing dissociated cells in stem cell-promoting medium, which revealed no differences in clonogenic potential between group 2 mutants and controls (Fig. S2).
Abnormal pituitary morphogenesis in Hesx1Cre/+;Tcf7l1fl/− mutants. In situ hybridization (A–C) and immunohistochemistry (D–H) of transverse histological sections of the pituitary gland of control embryos (A–H) and Hesx1Cre/+;Tcf7l1fl/− mutants (A′–H′ and A′′–H′′) at 17.5 dpc. (A′–H′) Mildly affected embryos show pituitary hyperplasia and cleft bifurcations (arrows in D′ and E′) but mostly normal expression of differentiation markers. (A′′–H′′) Severely affected pituitaries exhibit dysmorphic pituitary tissue that is ectopically located in the oropharyngeal cavity (arrowheads), but hormone-producing cells are present. AP, anterior pituitary; BS, basisphenoid bone; Cga, glycoprotein hormone α; Gh, growth hormone; OC, oral cavity; Pomc1, pro-opiomelanocortin-α; PP, posterior pituitary. Pictures are representative of five embryos per genotype. (Scale bar: H′′, 100 μm.)
Clonogenic potential of pituitary stem cells at 18.5 dpc and 1 d postnatally (P1). Pituitaries were dissociated to single cell suspensions and cultured in stem cell-promoting media as described. Note that there are not significant differences in the proportion of clonogenic cells between genotypes.
Similar morphological abnormalities, including hyperplasia, cleft bifurcations, and ectopically located pituitary tissue, were observed in Hesx1Cre/+;Tcf7l1flox/− mutants at 13.5 and 15.5 dpc. Nevertheless, expression of Lhx3, Pomc1, Prop1, and Pit1 at 13.5 dpc as well as Lhx3, Pomc1, Cga, and Pit1 at 15.5 dpc was detected in the developing pituitary of all mutants analyzed (Fig. 2 A–P). Together, these analyses suggest that loss of Tcf7l1 in Hesx1-expressing cells leads to aberrant pituitary morphogenesis but normal cell differentiation in the majority of the embryos, with low penetrance of a phenotype characterized by ectopic pituitary tissue, possibly of anterior lobe identity, in the oropharyngeal cavity concomitant with defective terminal differentiation.
Increased proliferation of RP progenitors but normal patterning of the developing anterior pituitary in Hesx1Cre/+;Tcf7l1fl/− mutants. (A–P) In situ hybridization on sagittal sections through the pituitary gland of control and Hesx1Cre/+;Tcf7l1fl/− mutant embryos (stage and probes are indicated; anterior to the left). (A–H) Mildly affected pituitaries show bifurcations and moderate expansion of the expression domains of Lhx3 and Prop1 at 13.5 and 15.5 dpc (arrows). (I–P) Severely affected pituitaries exhibit enlarged expression domains attributable to the abnormal morphogenesis of the developing anterior pituitary, which extends into the oropharyngeal cavity (arrowheads). (Q–R) Anti–phospho-histone H3 immunofluorescent staining on sagittal histological sections of a control embryo and a Hesx1Cre/+;Tcf7l1fl/− mutant. (S) Quantitative analyses showing a statistically significant increase in the mitotic index in the Hesx1Cre/+;Tcf7l1fl/− developing pituitaries relative to control littermates at 13.5 dpc but not at 15.5 dpc. AP, anterior pituitary; BS, basisphenoid bone; Inf, infundibulum; OC, oropharyngeal cavity. Pictures are representative of seven embryos per genotype. ***P < 0.05 (one-way ANOVA). (Scale bar: H and P, 100 μm.)
During early pituitary development, the synergistic expression of Bmp4, Fgf8, and Fgf10 within the developing hypothalamus is essential for the induction of Lhx3 in the region of the oral ectoderm fated to become RP (2). The expression domains of Fgf8, Fgf10, and Bmp4 in the hypothalamic anlage (Fig. 3 A–L) and of Lhx3 in the underlying RP (Fig. 3 M–P) were rostrally expanded in the Hesx1Cre/+;Tcf7l1flox/− mutants relative to controls at 9.5 dpc. Enlargement of the Lhx3-expression domain suggests the recruitment of additional oral ectoderm into RP epithelium, which may contribute to the hyperplasia observed at subsequent developmental stages. The expression of SHH, another critical signal required for normal RP development (58, 59), in the preoptic area of the hypothalamus was shifted anteriorly and slightly reduced in the Hesx1Cre/+;Tcf7l1flox/− mutants in comparison with controls (Fig. 4 I–L). In agreement with this observation, the expression domains of Tbx3 and to a lesser extent Tbx2, which normally act as repressors of Shh expression in the hypothalamus (55), were rostrally shifted, thus invading the caudal region of the preoptic area, where SHH expression appears reduced (Fig. 4 A–H).
Dysregulation of Fgf8, Fgf10, and Bmp4 expression in the hypothalamus of Hesx1Cre/+;Tcf7l1fl/− mutants. In situ hybridization on sagittal histological sections revealing the expression of Fgf10 (A–D), Fgf8 (E–H), and Bmp4 (I–L) in the prospective hypothalamus and Lhx3 (M–P) in RP (anterior to the left). (A–H) Fgf10 and Fgf8 expression domains are anteriorized along the neural epithelium of the ventral diencephalon (i.e., prospective hypothalamus) overlaying the oral ectoderm of RP in the Hesx1Cre/+;Tcf7l1fl/− mutants (compare arrowheads in C and D and G and H, respectively). C and D and G and H represent enlarged images of the dotted squared areas in A and B and E and F, respectively. (I–L) The expression domain of Bmp4 is rostrally expanded in Hesx1Cre/+;Tcf7l1fl/− prospective hypothalamus compared with control embryos (compare arrowheads K and L). K and L represent enlarged images of the squared dotted areas in I and J. (M–P) The expression domain of Lhx3 is rostrally extended in Hesx1Cre/+;Tcf3fl/− compared with the control embryo (arrows in O and P), indicating an expansion of the RP epithelium. O and P are enlarged images of the dotted square areas in M and N respectively. VD, ventral diencephalon. Pictures are representative of five embryos per genotype. (Scale bar: P, 100 μm.)
SHH and its regulators Tbx2 and Tbx3 are misexpressed in the developing hypothalamus of Hesx1Cre/+;Tcf7l1fl/− mutants. (A–H) In situ hybridization on sagittal sections of 10.5-dpc embryos (anterior to the left) reveals the anterior shift of the Tbx3 expression domain, and to a lesser extent Tbx2, within the developing hypothalamus of the Hesx1Cre/+;Tcf7l1fl/− mutants compared with the control embryos. (I–L) Immunostaining showing a reduction of SHH expression in the caudal region of the preoptic area in a Hesx1Cre/+;Tcf7l1fl/− mutant relative to a control. Note that the reduction of SHH expression in the caudal region of the preoptic area appears to overlap with the anterior extension of Tbx3 expression. VD, ventral diencephalon. Pictures are representative of three embryos per genotype. (Scale bar: P, 100 μm.)
Proliferation analysis revealed a mild but statistically significant increase in the mitotic index of RP periluminal progenitors in Hesx1Cre/+;Tcf7l1flox/− mutant pituitaries at 13.5 dpc (P < 0.05), but this increase was transient and not observed at 15.5 dpc (P > 0.05) (Fig. 2 Q–S). This phenotype is entirely consistent with the deletion of Tcf7l1 in both the hypothalamus and RP in the Hesx1Cre/+;Tcf7l1flox/− mutant pituitaries (Fig. S3 C and D). Caspase 3 immunostaining revealed the presence of a few apoptotic cells in the ventral regions of RP in both the mutant and control embryos (Fig. S3 A and B). Together, these studies suggest that Tcf7l1 is required for proper patterning of the prospective hypothalamus and for establishment of the normal expression domains of critical hypothalamic signals involved in the induction and proliferation of RP progenitors.
Immunostaining against Caspase 3 and TCF7L1. Histological sagittal sections of Hesx1Cre/+;Tcf7l1fl/− mutant and control embryos at 10.5 dpc (anterior to the left). (A and B) Caspase+ve cells are detected mostly in the ventral region of RP (arrowhead) in both genotypes. (C and D) TCF7L1+ve cells (green) are detected throughout the hypothalamus and RP regions in the wild-type embryo but are absent in the Hesx1Cre/+;Tcf7l1fl/− mutant. (Scale bar: 100 μm.)
Next, we sought to assess whether the requirement of TCF7L1 within the developing hypothalamus and pituitary gland was dependent or independent of β-catenin. To achieve this, we used a recently generated mouse line (Tcf7l1ΔN/ΔN) expressing a TCF7L1 mutant protein lacking the β-catenin–interacting domain at the N terminus. This truncated form of TCF7L1 (TCF7L1ΔN) shows similar DNA–binding and -repressing activities to wild-type TCF7L1, but the mutant’s interaction with β-catenin is abolished (33).
In situ hybridization analysis revealed the absence of morphological defects in the pituitary gland and the normal expression of terminal differentiation markers (Gh, Pomc, Cga, Tsh, and Lh) in Tcf7l1ΔN/ΔN mutants relative to control embryos at 18.5 dpc (Fig. 5 I–R). Likewise, the expression domain of Fgf10 within the hypothalamus and Lhx3 in the developing RP were comparable between genotypes (Fig. 5 A–H). Note that Tcf7l1ΔN/ΔN mutants show a general developmental delay and are smaller than control littermates, which accounts for the slight size differences in the expression domains (Fig. 5 C and D) (33). The absence of any apparent defects in Tcf7l1ΔN/ΔN mutants demonstrates that TCF7L1 acts primarily as a transcriptional repressor within the developing hypothalamus and pituitary gland and that TCF7L1’s function is independent of β-catenin. Additionally, these data imply that the defects observed in Hesx1Cre/+;Tcf7l1flox/− mutants are a consequence of the derepression of TCF7L1 transcriptional targets.
Expression of a mutant form of TCF7L1 lacking the β-catenin-interacting domain (Tcf7l1ΔN/ΔN) is sufficient to sustain normal hypothalamic-pituitary axis development. (A–H) In situ hybridization on sagittal sections showing normal expression domains of Fgf10 (A–D) and Lhx3 (E-H) in Tcf7l1ΔN/ΔN mutants and control wild-type littermates at 9.5 dpc (anterior to the left). C, D, G, and H are enlarged images of the dotted squared areas. The apparent difference in size in the Tcf7l1ΔN/ΔN mutants relative to the controls is caused by an overall developmental delay in the mutants. (I–R) In situ hybridization on transverse histological sections through the pituitary gland of Tcf7l1ΔN/ΔN mutants and control embryos at 18.5 dpc reveal no gross differences in the levels of expression of several differentiation markers. Note that pituitary morphogenesis is also normal in Tcf7l1ΔN/ΔN mutants relative to controls. AL, anterior lobe; BS, basisphenoid bone; PL, posterior lobe; VD, ventral diencephalon. Pictures are representative of five embryos per genotype. (Scale bars: H and R, 100 μm.)
The forebrain and pituitary defects observed in the Hesx1Cre/+;Tcf7l1flox/− mutants are very similar to those previously described in Hesx1-deficient embryos, suggesting that these transcriptional repressors may control the same genetic program (49, 54). Because mutations in HESX1 have been associated with congenital hypopituitarism in humans, including SOD (60, 61), we hypothesized that mutations in TCF7L1 may also be implicated in these conditions. Indeed, in situ hybridization showed that TCF7L1 is expressed during human embryonic development within brain tissue, including the hypothalamus, and the developing pituitary gland (Fig. 6).
TCF7L1 is expressed in the developing hypothalamic-pituitary axis, central nervous system and eyes during human embryogenesis. In situ hybridization against hTCF7L1 on coronal (A–C and G–I) or sagittal (D–F) histological sections at Carnegie stage (CS) 13, 18, and 20. (A–C) Note the expression of hTCF7L1 in the ventral diencephalon neuroepithelium (i.e., prospective hypothalamus) (arrowheads in B), RP progenitors (arrows in B), and developing optic cups (arrows in C) at CS13. (D–F) At CS18, expression of hTCF7L1 is observed throughout the neuroepithelium of the hind-, mid-, and forebrain, including the hypothalamus (arrowheads in E) and telencephalon (arrows in F). Expression is also detected in the developing anterior pituitary gland (arrows in E). (G–I) At CS20, expression of hTCF7L1 is detected in the anterior and posterior lobes of the pituitary gland (H), neural retina (arrows in I), eyelid (arrow in G), olfactory epithelium of the nasal cavity, and trigeminal ganglia. B, C, E, F, H, and I are enlarged images of the dotted squared areas. AP, anterior pituitary; BS, basisphenoid bone; hyp, hypothalamus; NC, nasal cavity; NR, neural retina; OC, optic cup; ON, optic nerve; PP, posterior pituitary; VD, ventral diencephalon. (Scale bars: A, D, and G, 500 μm; B, C, E, F, H and I, 100 μm.)
To investigate this possibility, we screened a total of 215 patients with hypopituitarism including SOD for mutations in the TCF7L1 locus by Sanger sequencing. Two heterozygous sequence variants resulting in amino acid changes in highly conserved residues of TCF7L1 were identified in two unrelated patients (clinical details are presented in Table 2 and SI Text).
Phenotypes in patients with TCF7L1 variations
Patient 1 of Caucasian (Finnish; nonconsanguineous) descent, showing forebrain defects (i.e., partial agenesis of the corpus callosum, thin right optic tract, and thin anterior commissure), and normal pituitary function, carried a missense G-to-C sequence variant at position 275 (c.275G > C), which causes the substitution of arginine 92 by proline (p.R92P) (Fig. 7 A and B). This variant was not identified in online databases (listed in Materials and Methods) or 392 additional Caucasian controls (including 200 healthy Finnish). The variant was inherited from the father and was also present in the paternal uncle, who were both asymptomatic.
Identification of hTCF7L1 heterozygous missense variants in patients with SOD. (A) Electropherogram showing two heterozygous missense variants in hTCF7L1: p.R92P and p.R400Q. (B) Both DNA variants result in the substitution of highly conserved amino acids. R92 is not conserved in Xenopus and zebrafish. (C) Transient luciferase assays on HEK293 cells cotransfected with either TOPflash (left graph) or hLEF1-promoter-luc (right graph) reporters and constructs expressing wild-type, p.R92P, or p.R400Q hTCF7L1 proteins. Note the significant reduction in repressing activity of the p.R92P and p.R400Q hTCF7L1 variants relative to wild-type TCF7L1 using both reporters. **P < 0.05 (one-way ANOVA). (D) Western blot analysis of transfected HEK293 cells used in C with a specific anti-TCF7L1 antibody detects the expression of the wild-type and p.R92P or p.R400Q hTCF7L1 proteins at similar levels. GAPDH was also detected in the same membrane as loading control.
Patient 2, of Pakistani (nonconsanguineous) descent, was diagnosed with classic SOD and showed small optic nerves and chiasm, an absent posterior pituitary, a very small anterior pituitary and absent septum pellucidum. The proband harbored a missense G-to-A sequence variant (c.1199G > A), resulting in an arginine to glutamine change at position 400 (p.R400Q) (Fig. 7 A and B). Although this variant was not identified in our healthy controls or the dbSNP or the Exome Sequencing Project (ESP) databases, the variant is present in 14 individuals in the ExAC database with an allelic frequency of 2.958 × e−5 in heterozygosity. The variant was inherited from the unaffected mother, and two unaffected siblings also carried the variant.
We sought to assess the capacity of wild-type and hTCF7L1 p.R92P and p.R400Q proteins to repress WNT/β-catenin-mediated transcriptional activation in vitro. We used two different reporters in transfected HEK293 cells cultured in the presence of a GSK3β inhibitor. Using the TOPflash reporter, the repressing activity of the TCF7L1 (p.R92P) variant was reduced by 34–40% (10 ng and 50 ng, respectively; P < 0.05) relative to wild-type TCF7L1 (Fig. 7C, left graph). Likewise, the repressing activity of the TCF7L1 (p.R400Q) variant on the TOPflash reporter was diminished by 41–55% (10 ng and 50 ng, respectively; P < 0.05) compared with wild-type TCF7L1 (Fig. 7C, left graph). A reduction of the transcriptional repressing activity of these variants was also observed when using the hLEF1-promoter-luc reporter [30% reduction for TCF7L1 (p.R92P) and 27% reduction for the TCF7L1 (p.R400Q) (P < 0.05); Fig. 7C, right graph]. In these transfection experiments, wild-type and mutant hTCF7L1 proteins were expressed at similar levels, as revealed by Western blot analysis using a specific anti-TCF7L1 antibody (Fig. 7D). A potential molecular mechanism underlying the decreased repressing activity of the TCF7L1 (p.R92P) variant could be the increased phosphorylation of this protein by the Homeodomain-Interacting Protein Kinase 2 (HIPK2) compared with wild-type hTCF7L1 (Fig. S4 and SI Text). No dominant-negative effects were observed in cotransfection experiments combining wild-type and hTCF7L1 p.R92P and p.R400Q variants (Fig. S5).
The hTCF7L1 p.R92P mutant protein is hyperphosphorylated by HIPK2. Western blot analysis using antibody to detect phosphorylated TCF7L1 reveals an increase in HIPK2-mediated phosphorylation of hTCF7L1 (p.R92P) mutant protein compared with wild-type TCF7L1.
Absence of dominant-negative effects of the hTCF7L1 p.R92P and p.R400Q variants. Transient luciferase assays on HEK293 cells cotransfected with TopFlash reporter and constructs expressing wild-type hTCF7L1 and either p.R92P or p.R400Q hTCF7L1 proteins. Note the significant reduction in repressing activity of the p.R92P and p.R400Q hTCF7L1 variants relative to wild-type TCF7L1 and the absence of effect when combined with wild-type hTCF7L1. **P < 0.05 (one-way ANOVA).
Finally, to assess the repressing activities of the two variants in vivo, we took advantage of a zebrafish model with a lower gene dosage of the hTCF7L1 orthologs tcf7l1a and tcf7l1b. tcf7l1a−/− and tcf3l1b+/− single mutants show no phenotype, but tcf7l1a−/−;tcf3l1b+/− double mutants exhibit an eyeless phenotype attributable to the ectopic activation of the Wnt/β-catenin signaling pathway and derepression of targets of this pathway in the forebrain (including the eye field) (30, 31). We injected double tcf7l1a−/−;tcf7l1b+/− double mutant zebrafish embryos, which normally develop no eyes, with mRNA encoding either wild-type hTCF7L1 or the two variants identified, with the aim of determining the variants’ ability to rescue the eyeless phenotype. Microinjection of wild-type hTCF7L1 mRNA fully rescued the eyeless phenotype of tcf7l1a−/−;tcf3l1b+/− double mutants (Fig. 8; n = 18 mutants, three independent experiments). However, only 64.5% (n = 31 mutants, three independent experiments) and 49.65% (n = 24 mutants, two independent experiments) of the eyeless tcf7l1a−/−;tcf3l1b+/− double mutants were rescued by injection of mRNA encoding the hTCF7L1 p.R92P and p.R400Q variants, respectively (Fig. 8). These results support the in vitro findings previously presented and add further evidence to the notion that the hTCF7L1 variants p.R92P and p.R400Q exhibit reduced repressing activity in vivo.
Incomplete rescue of the eyeless phenotype of double tcf7l1a/b zebrafish mutants with the human hTCF7L1 p.R92P and p.R400Q variants. Lateral view of 32-h postfertilization zebrafish embryos (anterior to left). Wild-type embryos injected with 100 pg of hTCF7L1 (C), p.R92P (E), and p.R400Q (H) mRNA show no phenotype compared with wild-type untreated embryos (A). tcf7l1a−/−;tcf3l1b+/− double mutants fail to develop eyes (B). Injection of tcf7l1a−/−;tcf3l1b+/− double mutants with 100 pg of wild-type hTCFL1 mRNA rescues the eyeless phenotype (D). Injection of tcf7l1a−/−;tcf3l1b+/− double mutants with 100 pg of hTCF7L1 p.R92P (F and G) or p.R400Q (I and J) results in incomplete rescue of the eyeless phenotype. (K) Percentage of embryos with normal eyes is shown in blue, and percentage of embryos with very small or no eyes is shown in red. p.R92P: 64.5% SD 12.41, P < 0.009; p.R400Q: 49.65% SD 5.93, P < 0.0006 (unpaired t test).
In this study, we have provided genetic and molecular evidence demonstrating that the repressing activity of the transcription factor TCF7L1 is required for normal establishment of the HP axis in mice and humans.
The pituitary defects of the Hesx1Cre/+;Tcf7l1flox/− mouse mutants are very similar to those observed in Hesx1−/− embryos; increased proliferation leading to abnormal pituitary morphogenesis but overall normal terminal differentiation (14, 54). We show the absence of mTCF7L1 protein in both the hypothalamus and the developing RP by 10.5 dpc in Hesx1Cre/+;Tcf7l1fl/− mutants, thus validating the genetic approach. Within the hypothalamus, we have revealed a critical role for mTCF7L1 in the regulation of anteroposterior patterning, because there is an overall anterior shift in the expression domains of caudal hypothalamic signals and markers (e.g., Fgf8, Fgf10, Bmp4, Tbx2, and Tbx3), which is concomitant with a reduction in SHH expression in the anterior hypothalamic area (Figs. 3 and 4). This abnormal hypothalamic signaling results in the recruitment of additional ectoderm into RP, as detected by the rostral expansion in the Lhx3 expression domain and increased proliferation in the developing pituitary of Hesx1Cre/+;Tcf7l1fl/− mutants at 13.5 dpc. In addition, there may be a contribution to the phenotype attributable to the lack of TCF7l1 within RP, which we cannot rule out.
There is a marked variability of expressivity of the pituitary phenotype in Hesx1Cre/+;Tcf7l1fl/− mouse mutants (Figs. 1 and 2), which is reminiscent of that observed in Hesx1-deficient embryos. In most of the Hesx1Cre/+;Tcf7l1fl/− mutants (group 2), there is an overall increase in pituitary tissue, and anterior, posterior, and intermediate lobes can be identified. However, in the most severely affected mutants (group 1), the posterior and intermediate lobes cannot be morphologically detected from 13.5 to 17.5 dpc. Incidentally, some Hesx1Cre/+;Tcf7l1fl/− mutants at 9.5 dpc show a lack of contact between the developing infundibulum and RP, which may contribute to the development of the ectopic pituitary in the roof of the oropharyngeal cavity. It is likely that this phenotypic variability observed from 13.5 dpc is brought about by the differential dysregulation of the hypothalamic signals normally involved in induction and proliferation of RP progenitors, an idea that is difficult to test experimentally.
Previously, we have shown that the cooperation between HESX1 and TCF7L1 is required to repress the activation of the WNT/β-catenin pathway in the anterior neural plate of zebrafish and mouse embryos (47). Our data are in support of the idea that TCF7L1 acts as a repressor during normal pituitary development in mice. Functional studies in mice have demonstrated a temporally specific function of the WNT/β-catenin signaling pathway during pituitary organogenesis. This pathway must be repressed at early stages of pituitary development and activated from 14.5 dpc for normal differentiation of the Pou1f1-cell lineage (37). In agreement with this notion, the permanent activation of the WNT/β-catenin pathway in RP progenitors at 9.5 dpc, by expression of a degradation-resistant form of β-catenin, results in severe pituitary hyperplasia during embryogenesis (50). Mice deficient for Tcf7l2, a TCF/LEF factor shown to act as either an activator or a repressor depending on the cell context, show anterior pituitary hyperplasia but no defects in the differentiation of hormone-producing cells (41). The similarities in the pituitary phenotype between Tcf7l2−/−, Hesx1−/−, and Hesx1Cre/+;Tcf7l1flox/− mutants suggest that these factors are likely to act as transcriptional repressors of WNT/β-catenin pathway targets during HP axis development. Indeed, the expression of a mutant form of TCF7L1 unable to interact with β-catenin, but maintaining repressing activity does not lead to any hypothalamic or pituitary defects, demonstrating that TCF7L1 is required as a repressor (Fig. 5).
Our data also provide evidence that the repressing activity of TCF7L1 is required for normal HP axis development in humans. Firstly, we show the expression of hTCF7L1 mRNA in the developing human embryo (Fig. 6). Furthermore, we have identified two TCF7L1 variants, p.R92P and p.R400Q, in unrelated families with SOD, which substitute highly conserved residues, and have shown that these variants are functionally compromised. These mutant proteins show reduced repressing activity relative to wild-type TCF7L1 using both the TOPflash reporter plasmid and a luciferase reporter containing a fragment of the human LEF1 promoter (Fig. 7) (33). In addition, we show that the TCF7L1 p.R92P and p.R400Q variants cannot rescue the eye defects of zebrafish mutants with a reduced gene dosage of tcf7l1 (Fig. 8). Given the presence of the variants in the phenotypically unaffected father and paternal uncle, our data suggest that the variants are not uniquely responsible for disease development but may be variably penetrant. The concept of variable penetrance has been observed in a variety of human genetic disorders. These disorders include Kallmann syndrome, where recent data have suggested an oligogenic basis to the disorder (62). Additionally, mutations in SHH and GLI2, a mediator of SHH, are associated with variably penetrant holoprosencephaly. Sequence variants in LHX4 and GLI2 are also associated with variably penetrant CH (63, 64). Of note, variable phenotypic expressivity is also observed in the Hesx1Cre/+;Tcf7l1fl/− mouse mutants and in the tcf7l1a−/−;tcf3l1b+/− zebrafish mutants injected with the hTCF7L1 variants identified in this study.
In summary, we identify Tcf7l1 as a determinant for the establishment of the HP axis and as a potential candidate gene to be mutated in congenital hypopituitarism.
Patient 1 had severe psychomotor retardation; focal generalized epilepsy; hypotonia; skeletal defects; a high-arched, narrow, and vaulted palate; crowded teeth with large central incisors; normal genitalia; normal thyroid function, prolactin, and morning cortisol; normal bone age; and a growth velocity of 4.4 cm/y between 7 and 9 y of age.
Patient 2 had normal genitalia; normal thyroid function and cortisol secretion initially; developmental delay; responded well to GH but became fatigued with reduced cortisol secretion (mean 24 h cortisol, 85 nmol/L at 9.2 y), needing hydrocortisone replacement; poor sleep pattern requiring melatonin; and commencement of spontaneous puberty at 9.5 y.
HEK293T cells were grown in DMEM supplemented with 10% FBS. The cells were transfected using a mixture of polyethylenimine (PEI) (23966-2; Polysciences) at a 1:3 wt/wt ratio of DNA to PEI in serum free DMEM. The cells were either transfected with wild-type or mutant TCF7L1 plasmids or together with a myc-tagged HIPK2 plasmid. The cells were allowed to grow for 48 h before they were collected in a lysis buffer [1% Triton-X-100, 50 mM NaCl, 1 mM EDTA, 50 mM Tris (pH 7.6)] containing protease and phosphatase inhibitors. Cell lysates (15 μg of protein per sample) were separated on SDS/PAGE gels and subjected to Western blotting. The membranes were probed with anti-TCF7L1 and anti–phospho-TCF7L1 antibodies and with anti-Myc for detecting myc-HIPK2 (65, 66). The blots were visualized by enhanced luminescence using the FUJI LAS-3000 imaging system.
Sequencing primers are:
TCF7L1_e1-2 forward: TCAAACTTTGTTGCGGCGGCT
TCF7L1_e1-2 reverse: GGTGCGTCCACCCAAGCAAG
TCF7L1_e3 forward: TATGTGCCCGCTGGGACA
TCF7L1_e3 reverse: CTTTACTTTCATCCAGGCAC
TCF7L1_e4 forward: CGTTTGGAAATTGACCCAGT
TCF7L1_e4 reverse: ATTGTTAGCCCCAAGCACTC
TCF7L1_e5 forward: AGGACTGAATGGGATGTTGC
TCF7L1_e5 reverse: CCATGATTTGAACCCAGGTC
TCF7L1_e6-7 forward: GGATGCTCCCATTTGATGTT
TCF7L1_e6-7 reverse: CCCTGTGCCTGCTCATTAAC
TCF7L1_e8 forward: GCCATTAAGCATCTCCCAAA
TCF7L1_e8 reverse: CTGTGATGCCTGCTGAGTGA
TCF7L1_e9-10 forward: ACTTGAATTAGCATCCAGGC
TCF7L1_e9-10 reverse: GGGTGAGCCTCTGTCTCA
TCF7L1_e11 forward: TAAAGGGCAACGTCCTGTCT
TCF7L1_e11 reverse: CTGGGAAAATGCCTTCACAT
TCF7L1_e12 forward: GCCAAAATCATCCCTGTGTC
TCF7L1_e12 reverse: CCACTTTACAGAACAGTCCA
The HIPK2 plays a critical role in the regulation of TCF7L1 phosphorylation and repressing activity in vivo and in vitro. In the absence of WNT ligands, TCF7L1 phosphorylation is normally low and repression activity is increased. In contrast, WNT stimulation or HIPK2 overexpression leads to TCF7L1 phosphorylation and derepression of target gene expression (65, 66). We sought to assess whether the loss of repressing activity of the TCF7L1 (p.R92P) mutation could be explained by the abnormal phosphorylation of this protein by HIPK2. Unfortunately, we were not able to carry out these experiments on the TCF7L1 (p.R400Q).
HEK293 cells were transfected with constructs expressing either wild-type or TCF7L1 (p.R92P) proteins with or without HIPK2-expressing plasmids, and the phosphorylation levels of TCF7L1 were assayed using a specific antibody. In the absence of exogenous HIPK2, the mutant protein showed a modest increase in phosphorylation compared with wild-type TCF7L1 (compare lanes 1 and 3 in Fig. S4). However, this difference was exacerbated upon cotransfection of TCF7L1 proteins with HIPK2 (compare lanes 2 and 4 in Fig. S4). This marked increase in phosphorylation of the TCF7L1 (p.R92P) mutant protein would be predicted to result in reduced binding affinity for the protein’s target genes and derepression of target gene expression. Together, these in vitro data suggest that the TCF7L1 (p.R92P) mutant protein is potentially pathogenic and may contribute to the phenotype observed in patient 1. In addition, these results strongly support the hypothesis that the repressing activity of TCF7L1 is required for normal forebrain and pituitary development in humans.
We thank Paul Le Tissier for comments on the manuscript. We thank the Developmental Studies Hybridoma Bank (University of Iowa) and the National Hormone and Peptide Program (Harbor–University of California, Los Angeles Medical Center) for providing some of the antibodies used in this study. We also thank Dr. A. Kispert for the Tbx2 and Tbx3 riboprobes and Dr. K. Itoh and Dr. S. Janssens for exploratory studies on HIPK2 phosphorylation. The human embryonic and fetal material was provided by the Joint Medical Research Council (MRC)/Wellcome Trust Human Developmental Biology Resource (www.hdbr.org) (Grant 099175). This work was supported by Wellcome Trust Grants 086545 and 084361, MRC Grant 164126, and by the National Institute for Health Research Biomedical Research Centre at Great Ormond Street Hospital for Children National Health Service Foundation Trust and University College London. M.T.D. is funded by Great Ormond Street Children’s Hospital Charity. V.S., A.G., and C.G.-M. are currently been supported by an Early Career Fellowship from the Medical College of Saint Bartholomew’s Hospital Trust. S.W.W. was supported by Wellcome Trust and MRC funding.
↵1Present address: Centre for Endocrinology, William Harvey Research Institute, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, London EC1M 6BQ, United Kingdom.
↵2Present address: Kinghorn Centre for Clinical Genomics, Garvan Institute of Medical Research, Darlinghurst, NSW 2010, Australia.
↵3Present address: St. Vincent’s Clinical School, University of New South Wales, Sydney, NSW 2052, Australia.
↵4Present address: Division of Craniofacial Development and Stem Cell Biology, King’s College London, London SE1 9RT, United Kingdom.
↵5M.T.D. and J.P.M.-B. contributed equally to this work.
Author contributions: C.G.-M., R.M.Y., S.Y.S., and J.P.M.-B. designed research; C.G.-M., M.J.M., V.S., R.M.Y., G.C., L.C.G., S.A.J., S.P., A.G., B.B., M.K., R.E.B., E.R., R.V., T.L., F.G., S.N.M.G., N.C., C.L.A., and J.P.M.-B. performed research; C.-I.W., J.T., M.S., S.W.W., B.J.M., and M.T.D. contributed new reagents/analytic tools; C.G.-M., M.J.M., V.S., R.M.Y., G.C., L.C.G., S.P., A.G., M.K., R.E.B., E.R., R.V., T.L., F.G., S.N.M.G., N.C., C.L.A., and J.P.M.-B. analyzed data; and C.G.-M., R.M.Y., and J.P.M.-B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1503346113/-/DCSupplemental.
