Hedgehog signaling update^†‡

Studies of the C-Terminal Cholesterol Moiety

Many studies have been carried out with contradictory results, which suggest that the C-terminal cholesterol moiety either aids or hinders hedgehog transport [Wang et al., 2007; Varjosalo and Taipale, 2008]. Early studies of the Drosophila imaginal discs suggested that the addition of cholesterol restricts the range of hedgehog movement, but its absence extends the range of movement [Porter et al., 1996; Burke et al., 1999]. Mouse limb studies have suggested that the cholesterol moiety is critical for long range signaling [Lewis et al., 2001], although a more recent study suggests that cholesterol modification restricts the spread of the Shh gradient in the limb bud [Li et al., 2006]. Studies in which both the C-terminal cholesterol moiety and the N-terminal palmitate are both absent, hedgehog can escape the producing cell without even requiring Dispatched [Mann and Beachy, 2004]. Resolving such contradictory findings will be critical in understanding the precise function of cholesterol in the trafficking and activity of hedgehog protein [Wang et al., 2007].

In Drosophila, hedgehog co-purifies with lipoprotein particles. These lipoproteins are necessary for hedgehog signaling [Panakova et al., 2005], suggesting that lipophilic modifications, such as cholesterol serve, at least in part, to localize Shh to certain areas of the plasma membrane critical for proper hedgehog trafficking and activity [Wang et al., 2007].

Enhancer Element Control of Shh Tissue Expression

The action of multiple enhancer elements acting independently control the transcription of Shh in different tissues. “Two independent enhancers—Shh floor plate enhancer 1 (SFPE1) and SFPE2, located at −8 kb and in intron 2, respectively—act to direct reporter expression to the floor plate of the hindbrain and spinal cord. A third element in intron 2, Shh brain enhancer (SBE1), directs reporter expression to the ventral midbrain and caudal diencephalon. The more distal elements, SBE2, SBE3, and SBE4, which are located >400 kb upstream of Shh transcription start site (TSS) drive reporter expression in the ventral forebrain. The combined activity of these enhancers appears to cover all regions of Shh transcription along the anteroposterior axis of the mouse neural tube” [Varjosalo and Taipale, 2008].

“The enhancer controlled Shh expression in the zone of polarizing activity (ZPA) of limb buds, mammal and fish conserved sequence 1 (MFCS1), is located even further upstream of the start site, at −1 Mb in intron 5 of the Lmbr1 gene. This element is the only enhancer in Shh that has been analyzed also by loss-of-function studies, which demonstrate that MFCS1 is necessary for Shh expression in the mouse ZPA. In humans, germline mutations within the conserved MFCS1 element can cause limb malformations characterized by preaxial polydactyly. Although many enhancers that drive Shh expression have been identified, very little is known about the specific transcription factors that control their activity” [Varjosalo and Taipale, 2008].

Different Effects Produced by Hedgehog Signals

Hedgehog signals can produce different effects. In Drosophila, Hh can act as an on/off switch that regulates only adjacent cells as, for example, in the dorsal embryonic ectoderm [Ingham and Hidalgo, 1993; Thérond et al., 1999]. Hh can also act as a short-range morphogen that spreads more than 10–20 cell diameters to control several different fates as a function of its concentration. For example, a model for understanding the short-range morphogen activity in the Drosophila wing disc is based on the bifunctional transcriptional regulator, Cubitus interruptus (Ci) [Strigini and Cohen, 1997]. In the absence of Hh, Ci becomes truncated, which represses target genes. Low levels of Hh block the truncation of Ci, which causes derepression of some target genes, but activated Ci is not made. At higher levels of Hh, Ci is fully converted to the activator form, and with the aid of Fused (Fu) moves to the cell nuclei [Méthot and Basler, 2000; Hooper and Scott, 2005].

Furthermore, Hedgehog can act as a long-range morphogen that spreads over many cell diameters as a function of different concentrations that specify a series of cell fates. For example, in the mouse neural tube, it has been suggested that different Shh levels are directly translated into different ratios of Gli activator/Gli repressor, which control cell fate [Jacob and Briscoe, 2003; Stamataki et al., 2005; Huangfu and Anderson, 2006].

Cell Surface and Extracellular Matrix Proteins

Several cell surface and extracellular matrix proteins have been identified and these have added increased complexity to hedgehog signaling. Cdo and Boc in vertebrates and their counterparts ihog and boi in Drosophila help regulate hedgehog signaling in a positive manner by binding to the hedgehog protein [Tenzen et al., 2006; Yao et al., 2006; Zhang et al., 2006a]. This enhances signaling in a synergistic interaction with Ptch1 [Hooper and Scott, 2005].

Shifted (Shf) is also found in both Drosophila and in the vertebrates. Shf is the counterpart of WIF1. However, unlike WIF1, which acts as an antagonist in Wnt signaling (see Similarities of Hedgehog and Wnt Signaling Section), Shf appears to function in a positive manner by regulating the extracellular distribution of hedgehog ligand [Glise et al., 2005; Gorfinkiel et al., 2005].

Mammalian Scube2 and its zebrafish ortholog you are extracellular regulators of hedgehog signaling [Kawakami et al., 2005; Woods and Talbot, 2005; Hollway et al., 2006]. Although it is not yet clear how Scube2 and you function, they act upstream of Smoothened and may possibly regulate the distribution of hedgehog ligand, or alternatively, perhaps permit hedgehog signaling indirectly by antagonizing the BMP pathway [Wang et al., 2007].

Tectonic is required to activate hedgehog signaling in the mouse and apparently acts downstream of Smoothened [Reiter and Skames, 2006].

Hedgehog Interacting Protein 1 (Hip1) and Growth Arrest Specific 1 (Gas1)

Hip1, found in vertebrates but not in Drosophila, encodes a membrane-bound glycoprotein that binds Shh with attenuates hedgehog signaling, thus reducing its effective range [Chuang and McMahon, 1999]. Gas1 is a 45 kDa glycophosphatydlinosotol (GPI)-linked protein originally for arresting the cell cycle when overexpressed [Del Sal et al., 1992]. It has been suggested that Gas1 attenuates hedgehog signaling, thus reducing its effective range [Lee et al., 2001]. Kang et al. 2007 found in “cooking with Gas1” that Gas1 is associated with Cdo in patterning the neural tube and the midface. In Figure 6, Ptch2 interacts with both Hip1 and Gas1 and together with ligand binding regulates the range and level of hedgehog signaling [Wang et al., 2007].

Intracellular Transduction

Costal 2 (Cos2), a microtubule-associated kinesin-like motor protein, so important in Drosophila transduction [Kamal and Goldstein, 2002; Mandelkow and Mandelkow, 2002] plays no role in mammalian transduction. In fact, the putative mammalian homologs of Cos2, Kif7, and Kif27, do not play a significant role in mammalian transduction [Wang et al., 2007]. Evidence is accumulating to suggest that the cilium may provide a function analogous to that of Cos2 by linking Smo to the modulation of Gli transcription factors [Corbit et al., 2005; Haycraft et al., 2005; Huangfu and Anderson, 2005]. Suppressor of fused (Sufu) has acquired an essential function in mammals. Fused (Fu), a serine–threonine kinase, so essential in regulating hedgehog signaling in Drosophila, plays no obvious role in the mouse [Chen et al., 2005; Merchant et al., 2005], but may be required in other vertebrates such as in zebrafish. The relative roles of Gli orthologs also differ between the zebrafish and the mouse [Karlstrom et al., 2003]. Although differences between Drosophila and the mouse are not uncommon in signal transduction, the striking differences in regulation in Drosophila, zebrafish, and the mouse are not typical of other pathways and, in fact, may be related to the presence or absence of acquisition or loss of cilium-based mechanisms of pathway regulation [Wang et al., 2007].

Rab23

The term Rab stands for Ras-like in rat brain. Rab proteins are small GTPases of the Ras superfamily that regulate intracellular membrane trafficking. A null mutation open brain (opb^−/−) results in an open neural tube involving the head and spinal cord, abnormal somites, polydactyly, and poorly developed eyes [Günther et al., 1994]. The opb mutation was found to arise from the Rab23 gene [Eggenschwiler et al., 2001]. Rab23 is a negative regulator of hedgehog signaling [Evans et al., 2003]. Rab23 acts downstream of Smo but upstream of Gli2, Gli3, and the IFT proteins [Huangfu and Anderson, 2006]. The human RAB23 gene, which maps to 6p11, encodes a 237 amino acid protein. Jenkins et al. 2007 showed that RAB23 mutations caused Carpenter syndrome and because of the open neural tube defect in Rab23 null mice Jenkins et al. 2007 suggested a species difference in the requirement for Rab23 during early development. Other known human RAB mutations include autosomal dominant Charcot–Marie–Tooth disease type 2B (RAB7 mutations) and autosomal recessive Griscelli syndrome type 2 (RAB27A mutations) [Ménasché et al., 2000; Verhoeven et al., 2003].

Gli Gene Family

GLI1 was identified originally as an amplified gene in a human glioblastoma [Kinzler et al., 1987, 1988; Kinzler and Vogelstein, 1990]. GLI1, GL2, and GLI3 (Table III) encode transcription factors that share five highly conserved tandem C₂H₂ zinc fingers and a conservative histidine–cysteine linker sequence between zinc fingers [Rupert et al., 1988]. The three GLI genes vary in size: GLI3 > GLI2 > GLI1. GLI1 has two isoforms; GLI2 has three alternatively spliced exons; and GLI3 has only one isoform. GLI1, encoding a 1,106 amino acid protein, maps to 12q13; GLI2, encoding 810, 829, 1241, or 1,258 amino acids, maps to 2q14; GLI3, encoding 1,595 amino acids, maps to 7p13 [Rupert et al., 1988, 1990; Matsumoto et al., 1996; Reifenberger et al., 1996; Biesecker, 2004]. The functions and phenotypic effects of GLI1, GLI2, and GLI3 are summarized in Table III. GLI3 has both activator and a repressor forms; the activator form utilizes cAMP response element-binding protein (CBP) as a co-activator [Partanen, 2001].

Unfortunately, Human Krüppel-Related Gene 4 (HKR4), which maps to 8q24.3, has been misclassified as a member of the GLI gene family and was called GLI4 by Kas et al. 1996. Also, in Drosophila, the term Gli refers not to the Gli gene family, but to the Drosophila gene Gliotactin [Rupert et al., 1988].

Recent studies show that the molecular mechanism of Gli3 processing requires the same set of kinases required for Ci in Drosophila: PKA, CKI, and GSK3β [Jia et al., 2004; Zhang et al., 2006b; Tempe et al., 2006; Wang and Li, 2006]. In vertebrates, PKA begins the sequential phosphorylations of Gli3 and leads to the binding of βTrCP, the vertebrate homolog of Slimb, and the subsequent processing of Gli3 [Tempe et al., 2006; Wang and Li, 2006]. In contrast, both Gli1 and Gli2 are degraded instead of being processed. The degradation of Gli2 is inhibited by hedgehog signaling [Pan et al., 2006]. The same kinases that act on Gli3 also phosphorylate Gli2, leading to Gli2 association with βTrCP and ubiquitin proteasome-mediated degradation. Gli1 undergoes PKA- and βTrCP-dependent degradation [HuntZicker et al., 2006]. Thus, the processing of Gli3 and the degradation of Gli1 and Gli2 share similar mechanisms, but result in different outcomes [Wang et al., 2007].

Several other regulator factors are involved in processing Ci and Gli. HIB, a hedgehog-induced protein that contains MATH and BTB domains, is a negative regulator of hedgehog signaling in Drosophila that promotes ubiquitination and degradation of Ci. SPOP, the mammalian homolog of HIB, is involved in Gli degradation [Zhang et al., 2006b]. In contrast, PI3-kinase-dependent activation of Akt regulates Shh signaling by controlling phosphorylation of Gil2 [Riobo et al., 2006]. In the chicken, the Talpid3 gene is essential for Gli3 processing and Gli activator functions [Davey et al., 2006].

Cyclic AMP Response Element-Binding Protein (CBP)

The cAMP-regulated enhancer (CRE) binds many transcription factors including CRE-binding protein (CRB), which is activated as a result of phosphorylation by protein kinase A (PKA) at a single amino acid residue, serine133. CBP is a 2,441 amino acid 265 kDa nuclear protein, which is a co-activator for a number of transcription factors including GLI3. CBP maps to 16p13.3 [Chrivia et al., 1993; Petrij et al., 1995].

Target Genes

Hedgehog signaling affects almost all portions of the vertebrate body plan during development. This is made possible by the various cellular responses to Hedgehog, including (1) the type of responding cell, (2) the dose of Hedgehog received, and (3) the timing during which the cell is exposed [Varjosalo and Taipale, 2008].

Target genes (1) can be positive or negative [Varjosalo and Taipale, 2008], (2) may be identified by genome-wide expression profiling [McGlinn et al., 2002; Xu et al., 2006], and (3) are being discovered on an ongoing basis [Varjosalo and Taipale, 2008].

Target genes include, among others, BMP4 [Astorga and Carlsson, 2007], FGF4 [Laufer et al., 1994], VEGF [Pola et al., 2001], Myf5 [Gustafsson et al., 2002], Ptch1, Ptch2, Nkx2.2, Nkx2.1, Rab34 [Vokes et al., 2007], Pax6, Pax7 [Wang et al., 2007], Pax9, Jagged1 [McGlinn et al., 2002], genes involved in cell growth and division (e.g., N-Myc) [Oliver et al., 2003], and many transcription factors (other members of the Myod/Myf, Pax, Nkx, Dbx, and Irx families) [Pierani et al., 1999; Gustafsson et al., 2002; Jacob and Briscoe, 2003; Vokes et al., 2007; Varjosalo and Taipale, 2008].

Tables of Mutations

The following sections contain both text and tables of mutations for disorders of the hedgehog signaling network. In recent years, recommendations for molecular mutations have changed a number of times so that yesterday's nomenclature is replaced with today's nomenclature and it is predictable that there will be further recommendations for tomorrow's nomenclature. This being the case, I have elected to maintain the original nomenclature from each table to make it more user-friendly for readers who wish to review the original sources.

Nevoid Basal Cell Carcinoma Syndrome and PTCH1 Mutations

The nevoid basal cell carcinoma syndrome (Fig. 10) is an autosomal dominant disorder characterized by skin manifestations (multiple basal cell carcinomas in about 90%, epidermal cysts, and palmar and plantar pits), skeletal defects (bifid ribs, cervical spina bifida occulta, kyphoscoliosis, pectus excavatum, Marfanoid habitus in some cases, short fourth metacarpals, and postaxial polydactyly in about 1–4%), CNS abnormalities (medulloblastoma in about 3–5%, calcification of the falx cerebri, bridging of the sella turcica, and meningioma in about 1% or less), craniofacial features (macrocephaly, mild hypertelorism, well developed supraorbital ridges resulting in an appearance of sunken eyes, relative mandibular prognathism, and cleft lip/palate in about 5%), oral manifestations (odontogenic keratocysts of the jaws in about 90%), and other findings (lymphomesenteric cysts, ovarian fibromas, often bilateral in about 15%, and cardiac fibroma in about 3%) [Gorlin, 1987; Cohen, 1999].

The syndrome is caused by mutations in PTCH1, the human homolog of the Drosophila segment polarity gene Ptch1. PTCH1 is a tumor suppressor gene located at 9q22.3. It functions as a cell cycle regulator, stopping cell division in the absence of SHH ligand and permitting cell division when ligand binding occurs [Cohen, 1999].

Generally, for a tumor suppressor gene to be inactivated, two hits are required. The first hit involves a mutation in one allele which can be dominantly inherited if present in a germ cell. The second hit involves the other allele, resulting in loss of heterozygosity by deletion, mitotic nondisjunction, or mitotic recombination. PTCH1 is probably not a classic tumor suppressor gene because the first hit causes major features such as a Marfanoid habitus, macrocephaly, relative mandibular prognathism, spina bifica occulta, postaxial polydactyly, and cleft lip/palate. Loss of heterozygosity has been demonstrated in basal cell carcinomas, odontogenic keratocysts, and medulloblastomas [Cohen, 1999].

PTCH1 germline mutations for the nevoid basal cell carcinoma syndrome are summarized in Table IV and mutations for sporadic basal cell carcinomas are summarized in Table V. Germline mutations (see Fig. 11) are mainly of the truncating type and they occur most commonly in the large extracellular loops, the large intracellular loop, and the N-terminal region. Germline missense mutations occur predominantly in the transmembrane domains, particularly transmembrane 4, which is located in the SSD. Missense mutations in sporadic basal cell carcinomas are clustered in the first large extracellular loop, the large intracellular loop, and in the C-terminal region [Lindström et al., 2006]. No correlations have been observed between the position of germline mutations and the phenotype of the nevoid basal cell carcinoma syndrome [Wicking et al., 1997; Lindström et al., 2006]. Some instances of PTCH1 mutations for the nevoid basal cell carcinoma syndrome represent examples of somatic mosaicism [Wicking et al., 1997].

A deletion (9q22.1–q22.32) has been noted to include PTCH1 and ROR2 (found in autosomal recessive Robinow syndrome); the authors speculated that pulmonary valve stenosis found in their patient with nevoid basal cell carcinoma syndrome might result from haploinsufficiency of ROR2 [Nowakowska et al., 2007]. Happle 2000 suggested that the few known instances of unilaterally located multiple basal cell carcinomas in the absence of other abnormalities suggests the possibility of a nonsyndromic type of hereditary multiple basal cell carcinomas. One PTCH2 mutation has been reported in a family with nevoid basal cell carcinoma syndrome [Fan et al., 2008] and rare instances of PTCH2 mutations have been noted in instances of isolated basal cell carcinoma and medulloblastoma [Smyth et al., 1999].

DISP1 Mutations

The discussion of DISP1 mutations follows the discussion of PTCH1 and PTCH2 because all three are 12-pass transmembrane proteins with a SSD (see Fig. 3). DISP1, originally known as DISPA based on its murine ortholog DispA, maps to 1q42 [Ma et al., 2002]. Roessler et al. 2009b reported two families with truncating mutations in DISP1 (Table VI). Shaffer et al. 2007 used array-based comparative genomic hybridization identified a microdeletion of 1q41q42. Overlap with DISP1 resulted in features suggestive of microforms of holoprosencephaly, such as developmental delay, seizures craniofacial dysmorphism, microcephaly, cleft palate, and short stature, but no frank holoprosencephaly. The phenotypic features in the two families reported by Roessler et al. 2009b with DISP1 mutations were similar to those found by Shaffer et al. 2007; they suggested that a model of holoprosencephaly based on multiple genetic and/or environmental insults—the multiple hit hypothesis [Ming and Muenke, 2002] may provide the best framework for a more comprehensive understanding of the complex interplay of various predisposing factors in holoprosencephaly.

Holoprosencephaly

Holoprosencephalic phenotypes are illustrated in Figure 12. Table VII summarizes the variability in their facial features and their CNS anomalies. All genes whose mutations may result in holoprosencephaly are summarized in Table VIII; they are characterized under the followings headings: gene symbols, their meanings, their map locations, and their functions. Table IX summarizes the mutated genes, the variability in their holoprosencephalic phenotypes, and their frequencies.

Some Causes of Holoprosencephaly Involving Cholesterol

Table X summarizes some causes of holoprosencephaly involving cholesterol and Figure 13 shows a partial, simplified cholesterol biosynthesis pathway of some of these disorders. Two teratogens—AY9944 and BM15.776 block Δ⁷-sterol reductase, resulting in the accumulation of 7-dehydrocholesterol and reduced serum cholesterol with the production of holoprosencephaly in rats [Barbu et al., 1984; Kolf-Clauw et al., 1996, 1997; Gofflot et al., 1999]. Triparanol is a teratogen that blocks Δ²⁴-sterol reductase, resulting in the accumulation of desmosterol and reduced serum cholesterol with the production of holoprosencephaly in rats [Roux, 1964]. Lrp2 is the gene responsible for the receptor megalin. An Lrp2^−/− null mutation for megalin results in holoprosencephaly in the mouse [McCarthy et al., 2002]. The range plant Veratrum californicum when eaten by pregnant ewes produces holoprosencephaly in newborn lambs; two of the isolated compounds—cyclopamine and jervine—have some similarities to cholesterol (Fig. 14) [Keeler and Binns, 1966; Keeler, 1975]. Both cyclopamine and jervine have also been produced experimentally in other animals [Cohen and Shiota, 2002]. In humans, about 5–6% of Smith–Lemli–Opitz syndrome cases are associated with holoprosencephaly [Kelley et al., 1996].

Impaired Cholesterol Biosynthesis in Human Holoprosencephaly

Of 228 cases of human holoprosencephaly, 9.6% had impaired cholesterol biosynthesis. Subgroups included: (1) one with Smith–Lemli–Opitz syndrome; (2) five with abnormalities of known holoprosencephaly genes (which suggested that defective regulation in cholesterol biosynthesis might add further impairment to hedgehog signaling); and (3) five with an accumulation of cholesterol precursors with variably reduced cholesterol synthesis (which suggested that newly synthesized precursors might have defective transport from the endoplasmic reticulum to the plasma membrane) [Haas et al., 2007; Haas and Muenke, 2010] (see also sections titled “Lipid Adducts Attached to Hedgehog Ligand,” “Studies of the C-Terminal Cholesterol Moiety,” “Sterol Sensing Domain,” and “DHCR7 Mutations and Smith–Lemli–Opitz syndrome”).

Nodal Signaling Pathway

The NODAL signaling pathway plays a central role in the induction of mesoderm and endoderm, and in the specification of left/right asymmetry (Fig. 15, Table VIII). NODAL is a ligand of the TGFβ superfamily. It signals through a heterodimer of activin receptors (ActR1B and ActR2A/B) with Tdgf1 as a co-receptor [Shen, 2007] (Fig. 15). When NODAL binds to the receptor complex, ActR2A/B phosphorylates ActR1B. In turn, ActR1B then phosphorylates Smad2 and Smad3, which interact with Smad4 and activate downstream genes by recruiting FoxHl (Fig. 15).

NODAL mutations may cause holoprosencephaly, but more commonly result in cardiac anomalies and left/right axis malformations [Roessler et al., 2009c; Solomon et al., 2010] (Fig. 15). FOXH1 mutations can result in cardiac anomalies or in holoprosencephaly [Roessler et al., 2008]. One TDGF1 mutation has resulted in a midline brain malformation and another mutation with frank holoprosencephaly [de la Cruz et al., 2002].

Bone Morphogenetic Protein Signaling Pathway

Bone morphogenetic proteins (BMPs) are members of the TGFβ superfamily. They signal through a heterodimer of BMP receptors (BMPR1 and BMPR2). The binding of BMP ligands to their receptor complexes leads to phosphorylation of BMPR1 by BMPR2. In turn, BMPR1 phosphorylates Smad 1/5/8, and this interacts with Smad4, which then becomes a downstream effector [Geng and Oliver] (Fig. 16).

Chordin (Chd) and Noggin (Nog) are secreted antagonists that directly interact with BMP ligands, preventing them from binding to their receptors (Fig. 16). Twisted gastrulation (Tsg), an antagonist as well, promotes the formation of a stable BMP/Chd/Tsg complex that prevents the binding of BMP ligands to their receptors. However, the role of Tsg is complicated, and it can also utilize Chd as a substrate for metalloprotease tolloid, which cleaves Chd at specific sites. When Tsg promotes Chd degradation, BMP ligands are then able to signal through their receptors [Geng and Oliver, 2009] (Fig. 16).

Multiple Hit Hypothesis for Holoprosencephaly

Single known genetic mutations for holoprosencephaly in humans account for about 28% of all cases (see Table IX; 10 different genes plus 5–6% of Smith–Lemli–Opitz cases caused by DHCR7 mutations [Dubourg et al., 2007; Kelley and Hennekam, 2000]. A few examples have double heterozygote mutations (see section titled “SHH Mutations”; SHH + ZIC2; 2 examples of SHH + deletion TGIF; GLI2 + PTCH1; and 2 different SIX3 mutations). Double heterozygotes in mice have been discussed by Ming and Muenke 2002.

It is most probable that human cases of holoprosencephaly are multifactorial, being caused by a combination of numerous genetic and environmental factors, and this subject has been discussed by Roessler et al. 2008. These authors found reduced NODAL signaling strength with mutations in several pathway members. Their molecular analysis presages future studies of holoprosencephaly by whole genome analysis.

Multifactorial inheritance includes genetic alterations during embryogenesis, involving for example, the prechordal plate, the position of the anterior portion of the notochord, and early neural tube development [Cohen et al., 1971; Geng and Oliver, 2009]. Although it has been difficult to study how such factors interact in humans [Cohen, 2006; Dubourg et al, 2008; Roessler et al., 2008], studies of a host of genetic and environmental factors in zebrafish, mice, and other animal models have been reviewed by Cohen and Shiota [2002].

In Table XI, mouse models of an alobar holoprosencephaly-like phenotype can result from reduced Nodal signaling, either by (1) introducing a hypomorphic allele (flox) into the Nodal or Cripto null background or (2) by functional inactivation of Gdf1 and genetic deletion of one Nodal allele [Geng and Oliver, 2009].

Although no mutations in the BMP signaling pathway cause holoprosencephaly, Table XI shows mutations in BMP antagonists that can result in an alobar holoprosencephaly-like phenotype in mice either (1) by digenic inheritance or (2) in Tsg^−/− embryos, which can only be observed on a C57BL/6 murine background [Geng and Oliver, 2009].

SHH Mutations

Shh gene expression has been demonstrated in the floor plate of the neural tube. Targeted gene disruption in the mouse produces cyclopia with absence of ventral neural tube cells. The Shh mutation is recessive, resulting in a severe phenotype [Chiang et al., 1996]. In humans, SHH maps to 7q36. In contrast to the mouse, SHH mutations are heterozygous, and the holoprosencephalic phenotype is more variable [Roessler et al., 1996; Nanni et al., 1999].

SHH mutations of the missense, nonsense, deletion, and frameshift types have been identified in exons 1, 2, and 3 (Table XII) [Nanni et al., 1999; Muenke and Beachy, 2001]. Nonsense, deletion, and frameshift mutations may truncate SHH-N, altering biological activity, or interfering with the autoprocessing reaction by affecting SHH-C (Fig. 2). Thus, haploinsufficiency results in holoprosencephaly. Great variability in expression characterizes autosomal dominant SHH families, some being minimally affected with holoprosencephalic microforms and others being severely affected. No genotype/phenotype correlations have been found [Nanni et al., 1999; Muenke and Beachy, 2001; Marini et al., 2003].

Numerous missense mutations occur throughout the entire coding region SHH (Table XII) and their functional significance is more difficult to interpret. If SHH mutations involve haploinsufficiency, holoprosencephaly results, but whether missense mutations alone are sufficient to cause holoprosencephaly has been questioned. Roessler et al. 1996 identified two missense mutations in exon 2, involving an invariant amino acid, Trp117 (Table XII) that immediately follows the crucial α helix-1 motif; they suggested that mutations in Trp117 might destabilize SHH-N. It is also possible that other genes in the sonic hedgehog signaling network or in other developmental pathways might work synergistically with SHH missense mutations. At least three patients have been reported with SHH mutations together with additional gene mutations—one with a ZIC2 mutation and two with deletions involving TGIF [Nanni et al., 1999; Gripp et al., 2000]. There are other double mutations which do not involve SHH: GLI2 and PTCH1 [Rahimov et al., 2006] and double SIX3 mutations [Ribeiro et al., 2006a]. In mice, double heterozygotes result in holoprosencephaly, whereas single heterozygotes do not (Table XIII) [Nomura and Li, 1998; Ming and Muenke, 2002]. Nanni et al. 2001 reported an unusual autosomal dominant family with a single maxillary central incisor and a novel SHH mutation (331A → T, resulting in Ile111Phe), not found in the holoprosencephaly population. El-Jaick et al. 2005 reported a case of alobar holoprosencephaly in a proposita whose mother had only a single maxillary central incisor; an SHH mutation, also involving the 111th amino acid residue, was identified (332T → A, resulting in Ile111Asp).

Schimmenti et al. 2003 reported a three generation pedigree in which the proband and his mother had iris and uveoretinal colobomas without optic nerve involvement. A novel SHH 24 bp deletion in the 3′ end of the gene was identified.

BMP4 and hedgehog signaling genes (SHH, PTCH1) are expressed co-spatially and co-temporally in the eye, brain, and hand of developing human embryos; Bakrania et al. 2008 identified four cases of anophthalmia–microphthalmia with mutations in both BMP4 and SHH/PTCH1, suggesting synergism between the two pathways in ocular development.

ZIC2 Mutations

ZIC2, which maps to 13q32, is a homolog of the Drosophila odd-paired-like (opa) gene [Aruga et al., 1996] and the zebrafish odd-paired-like (opl) gene. Muenke et al. 2002 suggested that Zic2 is expressed in the dorsal neural tube, in contrast to Shh, which is expressed in the ventral neural tube. It appear that during midgastrulation, Zic2 functions precede those of Shh [Warr et al., 2008].

In humans, heterozygous ZIC2 mutations cause holoprosencephaly. Most patients have alobar or semilobar holoprosencephaly, but lobar holoprosencephaly and syntelencephaly have been noted on occasion [Brown et al., 2001; Muenke et al., 2002]. Before ZIC2 was identified and mapped to 13q32, Cohen 1989a,b observed that in patients with del(13q), facial dysmorphism was mild. Brown et al. 2001 studied 28 patients with ZIC2 mutations and found no instances of cyclopia, ethmocephaly, cebocephaly, or median cleft lip, but rather a normal or mildly dysmorphic face. Solomon et al. 2009b found ZIC2 mutations in 8.5% of probands with holoprosencephaly in large series: 153 individuals from 116 unrelated kindreds, including 137 patients with molecularly determined ZIC2 mutations and 16 patients with whole-gene deletions. A novel and distinct facial phenotype was described and included bitemporal narrowing, upslanting palpebral fissures, short nose with anteverted nares, broad philtrum, and large ears.

ZIC2 mutations have been identified by many authors [Brown et al., 1998, 2001, 2005; Orioli et al., 2001; Dubourg et al., 2004; Roessler et al., 2009a]. In humans, heterozygous ZIC2 mutations cause holoprosencephaly either (1) by small insertions or deletions, leading to frameshift and premature stop codons, thus truncating the protein, or (2) by alanine tract expansion [Brown et al., 1998, 2001; Dubourg et al., 2002]. Brown et al. 2002 observed that many mutations occurred in the carboxy-terminus outside the DNA-binding domain, including some patients with identical expansions of a carboxy-terminal alanine tract. In these instances, DNA-binding is normal, suggesting that the transactivating function of the ZIC2 protein is interrupted, and the protein–protein interactions are critical for normal activity. Muenke et al. 2002 found that 2 of 19 families with ZIC2 mutations had 3 pregnancies resulting in anencephaly. Targeted Zic2 mutations in mice can produce neural tube defects as well as holoprosencephaly [Nagai et al., 2000]. Brown et al. 2002 found no ZIC2 mutations in human neural tube defects, but suggested a possible association with polyhistidine tract polymorphism in the ZIC2 gene. The quintessential study of ZIC2 mutations is that of Roessler et al. 2009a who analyzed 21 previously reported cases and 62 novel mutations (Table XIV). They found loss-of-function to be the predominant mechanism in ZIC2 mutations.

SIX3 Mutations

The so/SIX family of transcription factors consists of a distantly related group of homeobox-containing genes characterized by the presence of a SIX domain (contiguous homology domain), which participates in transcriptional activation [Kawakami et al., 1996]. Six3 is expressed in the developing forebrain and the eyes and in all vertebrates including humans [Oliver et al., 1995; Kobayashi et al., 1998; Loosli et al., 1999; Lacbawan et al., 2009]. Human SIX3 maps to 2p21 [Schell et al., 1996; Wallis et al., 1999]. Geng et al. 2008 and Jeong et al. 2008 showed that Six3 is direct regulator of Shh expression by demonstrating a cross-regulatory loop between Shh and Six3 in the ventral forebrain.

In a cohort of 800 patients with holoprosencephaly, Lacbawan et al. 2009 detected SIX3 mutations in 4.7% of affected probands. SIX3 mutations result in significantly more severe holoprosencephaly than in other cases of nonchromosomal, nonsyndromic holoprosencephaly [Ribeiro et al., 2006a; Lacbawan et al., 2009].

Domené et al. 2008 developed sensitive assays of SIX3 function in zebrafish and confirmed that 89% of the deleterious mutations were loss-of-function alleles. Lacbawan et al. 2009 found that the more severe forms of holoprosencephaly demonstrated greater loss-of-function. The female/male ratio was 1.5:1 and maternal inheritance was twice as common as paternal inheritance. Fourteen percent of SIX3 mutations occur de novo. In familial cases, clinical variability is wide and penetrance is estimated to be at least 62%. The most commonly associated clinical findings in decreasing order of frequency are hypotelorism, microcephaly, cleft lip-palate, seizures, premaxillary agenesis, diabetes insipidus, single maxillary central incisor, and coloboma.

SIX3 mutations are summarized in Table XV together with the associated references. Ribeiro et al. 2006a reported one patient with a double SIX3 mutations (c.206G → A, resulting in c.Gly69Asp, and c.406dupGC, frameshift). Ming et al. 2002 and Domené et al. 2008 reported a double SIX3/PTCH1 mutation (SIX3: c.278C → A, resulting in p.Ala93Asp, and PTCH1: c.1165G → A, resulting in p.Ala393Thr).

TGIF Mutations

TG-interacting factor (TGIF), which maps to 18p11.3, modulates TGFβ components that cause holoprosencephaly in animals. TGIF is an atypical homeodomain protein that is localized in the nucleus [Wallis and Muenke, 1999; Muenke and Cohen, 2000; Muenke and Beachy, 2001]. TGIF acts as a co-repressor of SMAD2/SMAD4 and in recruiting histone deacetylases to the complex [Wotton et al., 1999]. TGIF can also bind to the RXR binding site of the cellular retinol-binding protein II promoter (CRBPII RXRE) [Bertolino et al., 1995], but TGIF and RXR compete for binding to the promoter element [Bertolino et al., 1996]. Thus, TGIF can repress retinoic acid-regulated gene transcription. Perhaps TGIF mutations may produce loss of function as a repressor, resulting in overactivity of retinoic acid-regulated genes, simulating the effect of excessive retinoic acid exposure in humans [Lammer et al., 1985] and in mice [Webster et al., 1986; Sulik et al., 1988, 1995]. Helms et al. 1994, 1997 demonstrated an interrelationship between the SHH and retinoic acid pathways in the chick.

To date, at least six heterozygous missense mutations in TGIF have been identified in human holoprosencephaly [Gripp et al., 2000; Dubourg et al., 2004; Herbergs et al., 2002]. The mutation reported by Herbergs et al. 2002 was found in one parent who had hypotelorism and microcephaly.

PTCH1 Mutations

PTCH1 is part of the hedgehog signaling network, Helms et al. 1997 demonstrated that retinoic acid inhibited SHH and PTCH1 expression in the craniofacial primordia of chick embryos; the polarizing activity in these tissues was abolished. These findings are of particular interest because human mutations in both SHH [Roessler et al., 1996, 1997; Nanni et al., 1999] and PTCH1 [Ming et al., 1998] have been shown to cause holoprosencephaly.

PTCH1, which maps to 9q22.3, is a 12-pass transmembrane protein for which SHH is the ligand. PTCH1 mutations have been identified in patients with nevoid basal cell carcinoma syndrome, isolated basal cell carcinoma, medulloblastoma, meningioma, neuroectodermal tumor, breast cancer, squamous cell carcinoma of the esophagus, and trichoepithelioma [Cohen, 1999]. Most mutations have resulted in protein truncation. However, Ming et al. 1998 reported four PTCH1 missense mutations—two in extracellular loops that may be required for SHH binding, and two intracellular loops (Table XVI).

Ribeiro et al. 2006b reported five probands with PTCH1 mutations (Table XVI) and highly variable phenotypes with holoprosencephaly in four cases and a holoprosencephaly-like phenotype in one case. Three mutations had not been reported previously (Ala443Gly, Val751Gly, andVal908Gly). Two patients had the same Val908Gly mutation but were phenotypically different: alobar holoprosencephaly, absent nasal septum, midline cleft lip-palate, and large ears in one case, and lobar holoprosencephaly, hypertelorism, clefting of the nose, severe bilateral microphthalmia, and single maxillary central incisor in the other. One patient had a Thr105Met mutation, holoprosencephaly-like facial features, and a normal MRI [Ribeiro et al., 2006b]. Ming et al. 2002 reported an identical mutation, but with alobar holoprosencephaly.

Mansilla et al. 2006 reported three instances of isolated cleft lip-palate with PTCH1 missense variants in probands. An unaffected relative or relatives had the same PTCH1 variant (Table XVII).

GLI2 Mutations

GLI2, which maps to 2q14 [Matsumoto et al., 1996] is part of the hedgehog signaling network. Transgenic mice overexpressing Gli2 in cutaneous keratinocytes developed multiple basal cell carcinomas [Grachtschouk et al., 2000]. Roessler et al. 2002 studied 300 familial and sporadic cases of holoprosencephaly. They identified seven sequence variations in GLI2. Functional studies showed four mutations with loss of function. Two mutations truncated the zinc finger domain. Another mutation was in the invariant splice donor consensus sequence. A frameshift mutation was found segregating in a large pedigree associated with panhypopituitarism.

Later, Roessler et al. 2005 reported a previously unidentified GLI2 sequence encoding an additional 328 amino acids at the amino terminus; their studies showed a strong inhibitory effect on transcriptional activity of wild-type GLI2 and established that this domain is essential for dominant-negative activity in GLI2 mutants. Roessler et al. 2005 noted that haploinsufficiency was associated in three affected families, whereas dominant-negative properties in the GLI2 mutant protein reduced transcriptional activity even further in two other affected families.

Rahimov et al. 2006 reported four patients with GLI2 mutations and their strikingly diverse phenotypes (Table XVIII), including one case with a double GLI2/PTCH1 mutation and a holoprosencephaly-like phenotype and another case with associated branchial arch anomalies.

FOXH1 Mutations

FOXH1 maps to 8q24.3 [Zhou et al., 1998]. To assess the multiple hit hypothesis in the context of NODAL signaling, Roessler et al. 2008 examined the genomes of individuals with congenital heart defects, laterality, and holoprosencephaly for mutations in NODAL, GDF1, CFC1, TDGF1, FOXH1, and SMAD4. Gene products throughout the pathway were associated with conotruncal heart defects in about 5–10% of subjects, but far less commonly with isolated laterality (∼1%) or isolated holoprosencephaly (∼1%).

TDGF1 Mutations

TDGF1 (teratocarcinoma-derived growth factor 1) (CRIPTO), an epidermal growth factor (EGF)-cycteine-rich (CFC) family member, maps to 3p21-p23 [Ciccodicola et al., 1989; Saccone et al., 1995]. In mice, Tdgf1 is essential for germ and layer formation, positioning of the anteroposterior axis, and neural patterning [Shen and Schier, 2000]. EGF–CFC proteins are co-factors in Nodal signaling. They are membrane-associated, glycosylated, and have a characteristic EGF-like motif and CFC domain [Shen and Schier, 2000]. de la Cruz et al. 2002 screened 83 familial and 327 sporadic cases of holoprosencephaly. They identified two loss-of-function mutations in the CFC domain of TDGF1. One patient had semilobar holoprosencephaly and median orofacial clefting; the other had a dysplastic forebrain, hypoplastic corpus callosum, and absent septum pellucidum.

Fatty Liver in Holoprosencephaly

Fatty infiltration of the liver and other related liver alterations are common in patients with holoprosencephaly of at least five different etiologies, including SHH, SIX3, TGIF, FOXH1, and ZIC2 [Solomon et al., 2009a].

DHCR7 Mutations and Smith–Lemli–Opitz Syndrome

DHCR7 mutations cause Smith–Lemli–Opitz syndrome, an autosomal recessive disorder characterized by microcephaly, mental retardation, ptosis of the eyelids, epicanthic folds, strabismus, anteverted nares, micrognathia, cleft palate in some instances, characteristic Y-shaped 2–3 syndactyly, postaxial polydactyly, cardiovascular anomalies, and renal anomalies, together with ambiguous genitalia and cryptorchidism in males (Fig. 17). The syndrome is variably expressed from severe, such as holoprosencephaly in 5–6%, resulting in lethality, to minimal facial dysmorphism and mild mental impairment [Kelley et al., 1996; Gorlin et al., 2001; Nowaczyk et al., 2001; Steiner and Martin, 2007].

DHCR7 maps to 11q12-q13. Mutations in DHCR7 (Δ⁷-sterol reductase) cause Smith–Lemli–Opitz syndrome. This results in an increased concentration of the precursor, 7-dehydrocholesterol with a reduced concentration in serum cholesterol (Fig. 13).

Mutations in the Smith–Lemli–Opitz syndrome are of the missense, nonsense, and splice site types. IVS8-1G > C has been identified in 29% of the patients [Kelley and Hennekam, 2000; Krakowiak et al., 2000; Witsch-Baumgartner et al., 2000; Yu et al., 2000a,b]. Nine cases of holoprosencephaly have been reported to date with molecular analysis in only two of them; one was homozygous for IV8-1G > C and one was a double mutation (IV8-1G > T and deletion of the entire 3^rd and 4^th exons) in a case of cyclopia [Weaver et al., 2010]. Although double mutations, such as for DHCR7 and, for example, SHH might be a possibility, this has not been reported to date. All known double mutations for holoprosencephaly have been reviewed in the section titled “Multiple Hit Hypothesis for Holoprosencephaly.”

Sterol depletion, but with still enough for the cholesterol moiety attachment to hedgehog may affect the activity of Smo [Cooper et al., 2003] (see section on Smoothened and Fig. 5). Willnow et al. 1996 showed that absence of megalin in knockout mice (Lrp⁻/Lrp⁻) produced holoprosencephaly (Table X). Dehart et al. 1997, using a rodent model for Smith–Lemli–Opitz syndrome, showed that cholesterol deficiency could interfere with embryonic plasma membrane function and cell-to-cell interaction.

The cholesterol supply during embryogenesis is an important factor affecting the Smith–Lemli–Opitz syndrome phenotype. Alipoprotein E (apo E), a constituent of lipoproteins in the plasma and body fluids, is involved in maternal–embryonal cholesterol transport. A simplified model of the role of apo E in this transport is shown in Figure 18. A genetic polymorphism in the apo E gene is characterized by three common alleles: 2, 3, and 4, which differ by base substitutions in two codon, resulting in alterations in two amino acid residues: 112 (Cys > Arg) and 158 (Arg > Cys). Witsch-Baumgartner et al. 2004 showed a significant correlation between clinical severity in Smith–Lemli–Opitz syndrome patients and apo E genotypes: a maternal apo 2 genotype was associated with a severe phenotype, whereas maternal apo E genotypes without the 2 allele were associated with a milder phenotype.

The human blood–brain barrier closes to cholesterol transport early in embryonic gestation [Plotz et al., 1968]. If maternal–fetal cholesterol transport could be induced before closure of the blood–brain barrier, neurologic and behavioral abnormalities might be ameliorated. Recently, Jenkins et al. 2008 presented some data suggesting that high density lipoprotein (HDL) particles from Smith–Lemli–Opitz syndrome fetuses are better acceptors of cholesterol than HDL particles from normal fetuses, which raises the possibility that if more cholesterol could be transferred to the fetal circulation, the fetus would be capable of accepting it and circulating it in HDL particles. Although the role of maternal–fetal transport across the placenta has been controversial [Wollett, 2005], evidence supporting this pathway has been accumulating [Lindegaard et al., 2008].

The transfer of cholesterol between lipoproteins and cells depends on cholesterol transporters and receptors. ATP-binding cassette transporter Al (Abca1) is a cellular transporter protein that mediates the efflux of cholesterol from cells to lipid poor apo A1, contributing to the formation of HDL particles [Oram and Vaughn, 2000]. The ATP-binding cassette transporter G1 (Abcg1), belonging to the same family of transporters, appears to be involved in the efflux from cells also, but the acceptors of cholesterol are HDL₂ and HDL₃ particles [Wang et al., 2004]. Scavenger receptor class B type 1 (Sr-b1) is a multi-ligand receptor that mediates the cellular uptake of cholesterol from HDL particles, as well as the cholesterol efflux from cells [Acton et al., 1997; Xu et al., 1997; de la Llera-Moya et al., 1999; Thuahnai et al., 2004].

Abca1, Abcg1, and Sr-b1 mRNAs are all expressed in the placenta of the mouse [Christiansen-Weber et al., 2000; Hatzopoulos et al., 1998; Langmann et al., 2003]. The Smith–Lemli–Opitz syndrome mouse model is incapable of de novo synthesis of cholesterol in utero. Lindegaard et al. 2008 showed that Abca1 and possibly Sr-b1 contribute to transporting maternal cholesterol to the developing mouse fetus and suggest that modulating maternal–fetal transport has the potential for in utero therapy of human Smith–Lemli–Opitz syndrome.

Greig Cephalopolysyndactyly Syndrome and GLI3 Mutations

Greig cephalopolysyndactyly syndrome is an autosomal dominant disorder characterized by broad and prominent forehead; cutaneous syndactyly; postaxial polydactyly of the hands more commonly than preaxial polydactyly; and ocular hypertelorism with secondary telecanthus (Fig. 19) [Biesecker, 2004, 2006; Balk and Biesecker, 2008]. More severely affected patients may also have developmental delay, mental retardation, and on occasion, seizures and hydrocephalus, making up a variant disorder known as Greig cephalopolysyndactyly contiguous gene syndrome [Johnston et al., 2003, 2007]. At the mild end of the spectrum, some patients have the limb anomalies with either subtle or no craniofacial defects, showing that the boundary between syndromic and nonsyndromic preaxial polydactyly and cutaneous syndactyly is not clear [Biesecker, 2006].

GLI3 is a zinc-finger transcription factor gene which maps to 7p14.1. Greig syndrome and Pallister–Hall syndrome (vide infra) are allelic, both being caused by GLI3 mutations [Vortkamp et al., 1991; Kang et al., 1997; Wild et al., 1997]. Greig syndrome mutations are summarized in Table XIX [Johnston et al., 2005].

Acrocallosal syndrome, an autosomal recessive disorder characterized by polysyndactyly, macrocephaly, prominent forehead, hypertelorism, agenesis or dysgenesis of the corpus callosum, mental retardation, and/or developmental delay with or without seizures and CNS anomalies [Schinzel, 1982]. Distinction between Greig cephalopolysyndactyly syndrome and acrocallosal syndrome in sporadic instances can be challenging, but demonstrating haploinsufficiency for GLI3 through molecular testing unambiguously identifies Greig syndrome and excludes acrocallosal syndrome [Johnston et al., 2003].

Pallister–Hall Syndrome and GLI3 Mutations

Characteristic features that distinguish Pallister–Hall syndrome from Greig cephalopolysyndactyly syndrome include hypothalamic hamartoma, bifid epiglottis, and insertional polydactyly (Fig. 20) [Biesecker, 2006]. Other findings include osseous syndactly and may include seizures, precocious puberty, panhypopituitarism or less commonly growth hormone deficiency, laryngeal cleft, atypical lung lobation, and imperforate anus. The disorder has autosomal dominant inheritance with great variability in phenotypic expression from mild to severe. Patients at the mild end of the phenotypic spectrum are often familial. Sporadic cases may be mild, moderate, or severe. A Pallister–Hall syndrome patient with any of the following medical problems—laryngeal cleft, adrenal insufficiency, or anal atresia—constitute a medical emergency. Immediate securing the airway is the first priority if respiratory distress is present, which can be caused by a laryngeal cleft, whereas the diagnostic feature, bifid epiglottis is symptom free. The most important emergency with panhypopituitarism is adrenal insufficiency and should be treated if present. Imperforate anus with its subsequent bowel obstruction needs immediate attention [Biesecker, 2004].

GLI3 mutations that cause Pallister–Hall syndrome and Greig cephalopolysyndactyly syndrome are distinct in two ways. First, the mutations causing Greig syndrome include translocations, large deletions, frameshift, missense, nonsense, and splice site mutations (Table XIX). In contrast, mutations in Pallister–Hall syndrome are almost completely limited to frameshift and nonsense mutations (Table XX). Second, the positions of the frameshift mutations in the two disorders are distinctive. Mutations in the first third of the gene cause Greig syndrome, whereas mutations in the second-third of the gene cause primarily Pallister–Hall syndrome [Johnston et al., 2005; Biesecker, 2006]. Thus, it is clear that these two allelic disorders have different modes of pathogenesis: functional haploinsufficency causes Greig cephalopolysyndactyly syndrome and a truncated functional repressor causes Pallister–Hall syndrome [Johnston et al., 2005; Biesecker, 2006].

Craig et al. 2008 identified somatic chromosomal abnormalities with loss of heterozygosity within GLI3 in 15% of isolated (non-Pallister–Hall) hypothalamic hamartomas (n = 55).

Carpenter Syndrome and RAB23 Mutations

Carpenter syndrome is an autosomal recessive disorder characterized by craniosynostosis, most commonly involving the sagittal and metopic sutures; short fingers with clinodactyly; commonly but not always preaxial polydactyly of the feet; sometimes postaxial polydactyly; and other abnormalities, such as congenital heart defects, short stature, obesity, and mental deficiency (Fig. 21) [Gorlin et al., 2001]. Carpenter syndrome is caused by RAB23 mutations [Jenkins et al., 2007] (Table XXI).

Rubinstein–Taybi Syndrome and CBP Mutations

Rubinstein–Taybi syndrome is characterized by growth deficiency, mental retardation, microcephaly, downslanting palpebral fissures, curved nose in profile with the septum extending below the alae nasi, mild micrognathia, broad thumbs, and broad great toes (Fig. 22). Many other abnormalities have been reported including, among others, strabismus (58%), heart defects (32%), and various tumors (nasopharyngeal rhabdomyosarcoma, intraspinal neurilemoma, neuroblastoma, meningioma, and pheochromocytoma [Gorlin et al., 2001].

CBP,2 which maps to 16p13.3, is a transcriptional co-activator involved in different signal transduction pathways that regulates the expression of many genes and plays an important role in the regulation of cell growth, cellular differentiation, and tumor suppression. CBP mutations cause Rubinstein–Taybi syndrome [Petrij et al., 1995; Coupry et al., 2002]. Microdeletions in CBP have been found in about 15% of patients [Gorlin et al., 2001]. Missense mutations, nonsense mutations, and splice site mutations have also been identified [Coupry et al., 2002; Bartsch et al., 2005; Schorry et al., 2008] (Table XXII).

Schorry et al. 2008 studied genotype–phenotype correlations with mutation types: truncating large deletions, single amino acid substitutions, or no CBP mutations. Growth deficiency in height and weight were observed more frequently in patients with no mutation. A trend towards lower IQ was found in patients with large deletions. Rubinstein–Taybi syndrome is etiologically heterogeneous. Some cases without CBP mutations have been found to have EP300 mutations and preliminary evidence has shown that some cases may lack broad thumbs. Other patients with Rubinstein–Taybi syndrome may be caused by an as yet unidentified gene [Schorry et al., 2008].