Cis-regulatory underpinnings of human GLI3 expression in embryonic craniofacial structures and internal organs


Author to whom all correspondence should be addressed.



The zinc finger transcription factor Gli3 is an important mediator of Sonic hedgehog (Shh) signaling. During early embryonic development Gli3 participates in patterning and growth of the central nervous system, face, skeleton, limb, tooth and gut. Precise regulation of the temporal and spatial expression of Gli3 is crucial for the proper specification of these structures in mammals and other vertebrates. Previously we reported a set of human intronic cis-regulators controlling almost the entire known repertoire of endogenous Gli3 expression in mouse neural tube and limbs. However, the genetic underpinning of GLI3 expression in other embryonic domains such as craniofacial structures and internal organs remain elusive. Here we demonstrate in a transgenic mice assay the potential of a subset of human/fish conserved non-coding sequences (CNEs) residing within GLI3 intronic intervals to induce reporter gene expression at known regions of endogenous Gli3 transcription in embryonic domains other than central nervous system (CNS) and limbs. Highly specific reporter expression was observed in craniofacial structures, eye, gut, and genitourinary system. Moreover, the comparison of expression patterns directed by these intronic cis-acting regulatory elements in mouse and zebrafish embryos suggests that in accordance with sequence conservation, the target site specificity of a subset of these elements remains preserved among these two lineages. Taken together with our recent investigations, it is proposed here that during vertebrate evolution the Gli3 expression control acquired multiple, independently acting, intronic enhancers for spatiotemporal patterning of CNS, limbs, craniofacial structures and internal organs.


Early embryonic patterning is regulated by interplay of complex signaling pathways. Functional connections among signaling molecules of a particular signal transduction pathway occurring at many levels, including the interaction of extracellular signaling molecules with their receptors, the import of signals from receptors to the nucleus, or the interpretation of a signal by the activation or repression of target genes through transcription factors. In humans, the impairment of these interactions at any level can lead to diseases, including birth defects and cancers. The Hedgehog (HH) signaling pathway, first elucidated in Drosophila, plays a pivotal role in organogenesis and differentiation in many different vertebrate structures including the nervous system, limbs, skeleton, lung, gut, and genitourinary system (Anderson et al. 2012). In vertebrates, secreted glycoproteins sonic hedgehog (SHH), Indian hedgehog (IHH), and desert hedgehog (DHH), encoded by paralogous genes, act as morphogens triggering a string of reactions in target cells resulting in activation or repression of downstream target genes, preferentially by a family of GLI transcription factors (GLI1, GLI2, and GLI3) (Abbasi et al. 2009).

The zinc-finger transcription factor GLI3, like its paralogues GLI1 and GLI2, acts as a primary transducer of SHH signaling in a context dependent combinatorial fashion (Ruiz i Altaba et al. 2003). GLI3 and GLI2 can serve both as transcriptional activators or repressors, whereas GLI1, whose expression is transcriptionally regulated by GLI2 and GLI3, appears to play a secondary role in potentiating the SHH response (Ding et al. 1998; Motoyama et al. 1998; Bai & Joyner 2001; Bai et al. 2002).

Mutations in the human GLI3 gene cause a variety of dominant developmental defect syndromes subsumed under the term “GLI3 morphopathies” (Radhakrishna et al. 1999), including Greig cephalopolysyndactyly syndrome (GCPS) (Vortkamp et al. 1991; Wild et al. 1997; Kalff-Suske et al. 1999), Pallister-Hall syndrome (PHS) (Kang et al. 1997b), postaxial polydactyly type A (PAPA) (Radhakrishna et al. 1997), and preaxial polydactyly type IV (PPD-IV) (Radhakrishna et al. 1999). Mutations affecting murine Gli3, such as extra toes (Xt), anterior digit deformity (add), and polydactyly Nagoya (Pdn), serve as models for GLI3 morphopathies (Pohl et al. 1990; Schimmang et al. 1992, 1994; Hui & Joyner 1993). All GLI3 morphopathies show malformations of the autopod, i.e. polydactylies or syndactylies. In addition, craniofacial abnormalities are associated with GCPS, and in the most severe form, PHS, other developmental malformations occur, such as hypothalamic hamartoma, visceral anomalies, anus atresy, epiglottis and larynx defects (Biesecker & Ondrey 1998). Mouse embryos with homozygous Gli3 deficiency show pleiotropic and lethal congenital malformations with distinct preaxial limb polydactylies (Schimmang et al. 1992; Hui & Joyner 1993).

A multitude of studies in mice and other model organisms have suggested that a GLI-code, the interplay of the GLI proteins expressed in a quantitatively and temporally fine tuned pattern in adjacent domains, provides a basic morphogenetic tool that is used over and over again in embryonal development. It is applied for patterning along the anteroposterior axis (Brewster et al. 2000), induction of sclerotome (Buttitta et al. 2003), morphogenesis of the neural tube (Persson et al. 2002; Wijgerde et al. 2002; Ruiz i Altaba et al. 2003), the prosencephalon (Rallu et al. 2002) and cerebellum (Corrales et al. 2004), for anterior-posterior limb patterning (Robert & Lallemand 2006), chondrocyte differentiation (Koziel et al. 2005), skeletal muscle formation (McDermott et al. 2005), and prostate gland development (Pu et al. 2004). This multitude of tasks demands a tight temporal and spatial control of Gli gene expression.

Previously, we identified a set of ancient gene regulatory elements controlling expression of the human GLI3 gene. Eleven intronic Fugu-human conserved noncoding elements (CNEs) from the introns of GLI3 acted in transiently transfected cultured cells in a cell-type dependent fashion as activators or repressors of reporter gene expression (Abbasi et al. 2007, 2010; Paparidis et al. 2007). By expressing reporter genes under the control of the human GLI3-CNEs in zebrafish embryos, we demonstrated that the activating or repressor potential of the CNEs observed in human cell culture was retained in vivo in a teleost fish. To a large extent reporter expression induced by these elements was in agreement with the endogenous zebrafish gli3 expression (Abbasi et al. 2007). In addition we used transgenic assays to show that GLI3 intronic CNEs, which are able to activate transcription in cell cultures and zebrafish, can induce reporter gene activation at sites of endogenous Gli3 expression also in chicken and mice (Abbasi 2010). For transgenic mice assays we reported intra-GLI3 CNEs induced transgene expression to many known regions of endogenous Gli3 transcription in limb buds, as well as along the anterior–posterior and dorsal–ventral axis of the developing mouse neural tube (Paparidis et al. 2007; Abbasi 2010). It is important to note that, in addition to its critical role in limb and neural tube development, mouse Gli3 is known to be involved in a multitude of other patterning mechanisms, and similarly, its expression pattern is highly complex and well defined, involving a number of cell and tissue types (other than limb and neural tube). However the cis-acting sequence elements regulating GLI3 expression in embryonic tissues other than neural tube and limbs have not yet been reported (Abbasi et al. 2007, 2010; Paparidis et al. 2007; Alvarez-Medina et al. 2008; Abbasi 2011; Coy et al. 2011). In this study we used a transgenic mouse assay to test previously defined GLI3 intronic CNEs for their ability to target reporter gene expression to sites of endogenous Gli3 expression in tissue domains other than neural tube and limbs. Our data demonstrates that reporter gene expression patterns induced by a subset of these elements mimicking precisely the reported endogenous Gli3 expression in craniofacial structures, eye, and a subset of internal organs. Comparison of the expression pattern in mouse with the expression data in zebrafish obtained with the same set of CNEs (Abbasi et al. 2007) demonstrates that in addition to sequence conservation, the target site specificity of at least some of these elements remain preserved over the course of 450 million years since teleosts diverged from the lineage leading to modern mammals.

Material and methods

Reporter constructs

Candidate enhancer sequences (CNE1, 6, 9 and 10) were polymerase chain reaction (PCR) amplified using the high fidelity herculase enhanced DNA polymerase (Stratagene) with primers containing restriction site tags. Amplified DNA was purified using the QIAquick PCR purification kit (Qiagen). Purified PCR products were then subjected to restriction site digestion and subsequently cloned into the multiple cloning site (Kpn1 site) of the p1230 vector (a generous gift of R. Krumlauf) in front of the human β-globin promoter driving lacZ reporter gene. Recombinant reporter expression constructs were transfected into Top10 competent bacterial cells (Invitrogen) and subsequently isolated and purified using the Qiagen plasmid purification kit (Qiagen). To control for the presence of any point mutations generated during PCR amplification, the clones were sequenced on an ABI 377 automated sequencer (Applied Biosystems).

Generation of transgenic mice

Inserts were separated from vector sequences as described previously (Paparidis et al. 2007) and diluted for injection into CB6F2 or FVB/N zygotes in 10 mmol\L Tris, pH 7.5, 0.1 mmol\L ethylenediaminetetraacetic acid (EDTA) , pH 8.0 buffer in a final concentration of 1–3 μg\mL. Injected oozytes were transferred by PolyGene AGRumlang, Switzerland, or IMT Transgenic Mouse Unit, Philipps University Marburg, Germany, into the oviduct of foster mice. The amount of DNA applied cannot be determined with certainty, but it is estimated that 1–2 picoliter are microinjected into each male pronucleus of fertilized eggs. G0 embryos were allowed to develop to term, and by using genomic DNA (extracted from tail or ear tissue using standard protocols) at least three offspring carrying recombinant constructs were identified by PCR (primers: “XgalF”, 5′-CAACAGTTGCGCAGCCTGAATG-3′;“XgalR”,5′GTGGGAACAAACGGCGGATTG-3′) and used for breeding with the respective wild type animals. Transgenic males were subsequently used as studs with wild type females to maintain transgenic lines and to generate embryos of different age for lacZ expression analysis.

Embryo staining and histological analysis

The time of gestation was calculated taking noon of the day of detection of a vaginal plug as embryonic day 0.5 (E 0.5). Embryos were harvested at 9.5, 10.5, 11.5, 12.5 and 13.5 dpc. Expression of the transgene was detected by staining embryos overnight in X-gal buffer. Embryos were dissected free of extraembryonic membranes (which were retained for genotype analysis) then fixed in 0.5% glutaraldehyde at 4°C for 30 min to 2 h, depending on their developmental stage, washed with phosphate-buffered saline (PBS) (containing 2 mmol\L MgCl2), and stained overnight in X-gal reaction buffer ([containing 35 mmol\L K3Fe[CN]6, 35 mmol\L K4Fe[CN]6 and 2 mmol\L MgCl2]) containing 0.1% Xgal at 37°C. The staining reaction was stopped by washing the embryos in PBS. The embryos were postfixed overnight in 0.5% glutaraldehyde at 4°C.

To analyze internal organs some of the embryos were dehydrated, embedded in paraffin wax and were sectioned at 10–40 μm. Sections were deparaffinized with xylene and mounted for histological analysis. At least two transgenic mice embryos for each CNE were analyzed.

By using the genomic DNA (extracted from tail or ear tissue using standard protocols), the primers listed in parentheses were used for the PCR genotyping of mice carrying recombinant constructs (“XgalF”, 5′-CACAGTTGCGCAGCCTGAATG-3′; “XgalR”, 5′-GTGGGAACAAACGGCGGATTG-3′).

Sequence data and comparative analysis

Approximately 1 Mb of the human genome, encompassing GLI3 (ENSG_106571) as well as GLI3 orthologous sequences of Chimpanzee (ENSPTRG00000019117), Rhesus (ENSMMUG_13614), Mouse (ENSMUSG_21318), Rat (ENSRNOG_14395), Dog (ENSCAFG_3535), Cow (ENSBTAG_10671) Chick (ENSGALG_12329), Opossum (ENSMODG_2714), Frog (ENSXETG_1856), and Fugu (SINFRUG_153715) were retrieved from the ENSEMBL genome browser ( Multispecies sequence comparison was performed by using the Shuffle-LAGAN alignment tool kit (Brudno et al. 2003). Human sequence was used as the baseline and annotated by using the exon/intron information available at ENSEMBL genome browser. Shuffle-LAGAN alignment was visualized with the VISTA visualization tool (Mayor et al. 2000). The conservation was measured using a 60 bp window and a cutoff score of 50% identity, with human sequence as a base line.


Identification of Tetrapod-Teleost conserved GLI3 intronic enhancers by comparative sequence analysis

Multi-species alignment of human GLI3 genomic sequence with orthologous intervals form other vertebrate species localized 12 intronic conserved non-coding elements, showing at least 50% identity over 60 bp window down to Fugu (Fig. 1). These enhancers are distributed across almost the entire GLI3 interval, with two elements in each of introns 2, 3, 4, and 10 and one in each of introns 1, 6, and 13 (Fig. 1). The GLI3 specific gene regulatory functions of 11 of these human elements were previous tested using human cell lines and zebrafish embryos. Additionally, the spatiotemporal aspects of CNEs 1, 2, 6, 9, 10, and 11, which had acted as enhancers in cell cultures and zebrafish embryos, were determined in transgenic mice assay. Previously, we focused on the potential of CNEs 1, 2, 6, 9, and 11 to replicate endogenous Gli3 expression pattern during development of the limbs and the central nervous system (CNS) (Abbasi et al. 2010). Here we extend our previous work to discuss in detail the potential of CNEs 1, 6, 9 and 10 to drive lacZ reporter gene expression in craniofacial structures, eye, and subset of internal organs of transgenic mice embryos. For each of the enhancer elements the embryonic mice from stable transgenic lines have been analyzed at different time points of development by whole mount staining and using histological sections to determine the time course and the location of reporter X-Gal signals reflecting expression of the reporter gene.

Figure 1.

SLAGAN alignment of the genomic region encompassing GLI3. In each panel, human genomic GLI3 DNA sequence from ENSEMBL is aligned with chimpanzee, rhesus, dog, cow, mouse, rat, opossum, chicken, frog, and Fugu orthologous regions. Alignment parameters are explained in the methods section. Conserved non-coding elements that we tested previously for their cis-regulatory potential are enclosed in black rectangles. Elements with asterisk (*) symbols are those with the potential to regulate reporter expression in tissues other than brain and limbs (described in this study). Conserved coding and non-coding sequences are depicted in blue and pink, respectively. Blue arrows above the plot depict the direction of GLI3 transcription. Ex and CNE stand for exon and conserved non-coding element, respectively.

The spatiotemporal aspects of CNE1 and CNE6 activity reflects GLI3 functions in the development of craniofacial structures

The primary activity domains of 935 bp CNE1 and 862 bp CNE6 cis-regions are central nervous system and limbs, respectively, and has been described in detail previously (Abbasi et al. 2010). In addition to brain and limbs, the reporter activity induced by these two intronic elements was replicating the Gli3 functions in the development of craniofacial structures. For CNE1 cis-region, at E9.5 strong lacZ expression (from E9.5-onward) was observed in the first (maxillary and mandibular components are shown by white and black arrowheads, respectively) and second branchial arch regions (Fig. 2A). At E11.5 the expression is maintained in the maxillary components (Fig. 2A–C), but no reporter activity was detected in the mandibular components of the first arch (Fig. 2B). In day E11.5 and E12.5 embryos the reporter expression was also observed in the facial mesenchyme, rostrally extending to nasal processes (Fig. 2B,C). In histological analysis of CNE1 carrying whole-mount embryos it appeared that within the facial regions lacZ expression is present in the medial and lateral nasal processes, precartilage primordium of nasal capsule, Meckel's cartilage, lateral palatine process and in the dental lamina (Fig. 2D–F).

Figure 2.

CNE1 and CNE6 governed lacZ expression, mimic the craniofacial aspects of endogenous Gli3 expression in a non-redundant fashion. Expression of a β-galactosidase reporter gene in transgenic mouse embryos. Whole mount views of E9.5 (A), 10.5 (G), E11.5 (B, H and I), and E12.5 (C, J and K) embryos. At E9.5 (A), within the facial region CNE1 mediated reporter expression was detected in the maxillary/mandibular component of the first branchial arch (white and black arrowheads in panel A, labeled as I) and in the second arch (black arrowhead in panel A, labeled as II). At E11.5 (B), the expression is maintained in a maxillary component of yjr first arch, and in the second branchial arch, strong reporter expression was also observed in the lateral nasal process. Sectioning of the facial region (D and E) (at the level shown with dotted lines in panel C) indicate β-galactosidase staining in the precartilage primordium of the nasal capsule of E12.5 embryo. (F) Red and black dotted circles highlight reporter expression signals in the precartilage primordium of Meckel's cartilage and lateral palatine process, respectively, whereas in the same panel the open arrowhead shows β-galactosidase staining within dental lamina (tooth primordium). CNE6 induced reporter expression starts in the rostroventral domain of forebrain at day 11.5 (white arrowhead in panel H). At E11.5 and E12.5 (panels I and J) the CNE6 induced reporter signal is more prominent in the head mesenchyme in the region of nasal process. Transverse section of the E13.5 embryo at the level shown with a black dotted line in K, revealed transgene expression specifically within precartilage primordium of the nasal capsule (L). ba2, second branchial arch; dl, dental lamina; flb, forelimb bud; hlb, hindlimb bud; llp, lateral palatine process; lnp, lateral nasal process; mc, Meckel's cartilage; mnp, medial nasal process; mx1, maxillary component of first branchial arch; nc, nasal cavity; ns, nasal septum; st, stomodaeum; t, tongue.

In addition to limbs (Abbasi et al. 2010), the CNE6 directed reporter expression was evident in rostro-ventral telencephalon by E10.5 (compare panel 2G with 2H), and later, by E11.5, in the head mesenchyme in the region of nasal process (Fig. 2I–L).

The CNE10 mediated lacZ expression is largely confined to foregut derivatives, eye and mammary placodes

The whole mount view of embryos from a stable transgenic line carrying the 1133 bp CNE10 revealed lacZ expression in the eye, visceral organ, mammary placodes, and in the external genitalia region (Figs 3A–C and 4A). Additionally, the detailed histological examination of E11.5 embryos demonstrated the CNE10 mediated lacZ expression in visceral organs, pharynx, esophagus, tracheal duct, wall of stomach, dorsal pancreatic primordium, lumen of duodenum, proximal loop of midgut, mesonephric vesicles and mesonephric (Wolffian) duct (Fig. 3D–I). β-galactosidase was also seen in Rathke's pouch, hepatic primordium and gall bladder (Fig. 3F,G).

Figure 3.

CNE10 induced lacZ expression in mammary placodes and internal organs of transgenic mice embryos. Whole mount views of E11.5 (A) and E12.5 (B & C) embryos carrying a reporter construct encompassing the 1133 bp CNE10. In E11.5 embryos CNE10 mediates lacZ expression in visceral organs (arrow in A). LacZ expression was observed in mammary placodes (B, numbered as 2, 3, and 4) and external genitalia (arrow in C). Note, mammary placodes 1 and 5 are covered by forelimb and hindlimb and are not visible. Sagittal sections of the transgenic E11.5 embryos (D–I). (D) Arrowheads indicate β-galactosidase staining in the mesonephric vesicles. (E) Open and black arrowheads display reporter expression in the mesonephric duct and along the wall of the stomach (weak staining), respectively. (F) Staining in Rathke's pouch and pharynx. (G) Reporter expression in the hepatic primordium (liver), gall bladder, lumen of duodenum and proximal loop of midgut. (H) Strong expression is seen in the pancreas. (I) Black and open arrowheads show reporter expression in the esophagus and tracheal duct. dpp, dorsal pancreatic primordium; es, esophagus; gb, gall bladder; hp, hepatic primordium; ld, lumen of duodenum; md, mesonephric duct; ph, pharynx; plm, proximal loop of midgut; rp, Rathke's pouch; s, stomach; td, tracheal duct.

Figure 4.

CNE10 cis-region was found to drive evolutionary conserved expression within vertebrate eye. A is a whole mount view of E12.5 mouse embryo, where CNE10 induced lacZ expression is prominent in eye (arrowhead). Sectioning through the (transgenic mouse) eye revealed lacZ expression specifically within the sensory layer of the retina (B). In (C) CNE10 induced GFP expression (fluorescent view) is shown in the retinal cell of a live day 2 zebrafish eye. Whole mount anti-GFP immunostaining (D) revealed GFP expression within the lens epithelial cell layer of day 3 zebrafish embryo. l, lens; r, retina; slr, sensory layer of retina.

Sagittal sections through the eyes of transgenic mouse embryos carrying CNE10 revealed lacZ expression specifically within the sensory layer of the retina (Fig. 4B). Eye-specific functions of the CNE10 cis-region appeared to be highly conserved in teleost fish (Fig. 4C,D).

CNE9 induced reporter activity was detected in the somites of developing mice embryos

Brain and spinal cord specific activity of CNE9 in transgenic mice assay was largely non-redundant to CNE1 induced reporter expression in these tissues (Abbasi et al. 2010). In addition to neural tube the transgenic embryos carrying approximately 1.2 kb CNE9 fragment showed X-gal staining in the inter-limb somites at E11.5 (Fig. 5A,B). The histological examination of inter-limb region revealed lacZ activity specifically in the ventral-lateral lip of dermomyotome (Fig. 5C).

Figure 5.

CNE9 cis-region induced lacZ expression was detected in the interlimb somites. (A and B) Whole mount views of E11.5 embryos. In addition to hindbrain (hb) and spinal cord (sc) stable transgenic lines carrying the 669 bp CNE9, induced expression in the interlimb (indicated by bracket symbol) somites (som). (C) Sectioning at the level shown with dotted lines in panel-B revealed reporter activity within ventral-lateral lip of the interlimb somite (arrowheads).


GLI3 expression pattern and functions during embryogenesis are highly complex and well defined

The mouse Gli3 is involved in a multitude of patterning mechanisms during early embryonic development, and similarly its expression pattern is highly complex and well defined. Gli3 is strongly expressed dorsally in the entire brain region and also in the cell populations within ventral aspects of the brain, and is known to play a key role not only in the dorsal-ventral patterning but also anterior–posterior patterning of telencephalon, diencephalon, midbrain and hindbrain (Matise et al. 1998; Aoto et al. 2002; Kuschel et al. 2003; Tyurina et al. 2005). In the spinal cord the Gli3 functions are required for normal MN (motor neurons) differentiation, for the correct spatial patterning of V0-V2 interneurons (intermediate spinal cord) and for the development of floor plate cells and V3 interneurons (ventral spinal cord) (Bai et al. 2004). Interestingly, although Gli3 is extensively expressed in the dorsal spinal cord, no obvious phenotype is seen there in Gli3 mutants (Bai et al. 2004). Mouse Gli3 is extensively expressed in the developing facial mesenchyme and has critical roles in the medial nasal process, lateral nasal process, maxillary process, palatal and tooth development (Mo et al. 1997; Hardcastle et al. 1998; Aoto et al. 2002). Furthermore Gli3 functions are also required for the normal development of ear (Hui & Joyner 1993). Gli3/ embryos exhibit a variety of eye abnormalities ranging from microphthalmia to the absence of any remnant of eye tissue (Johnson 1967; Franz & Besecke 1991; Tyurina et al. 2005; Furimsky & Wallace 2006). The protein is expressed in the stalk region of the optic field, in the neural retina, retinal pigment epithelium, lens and surface ectoderm (Aoto et al. 2002; Zaki et al. 2006). Within the somite, Gli3 expression is initially widespread, and becomes restricted rapidly to the dorsal medial lip of the dermamyotome (DML, the precursor cells for epaxial musculature) and the ventral lateral lip of the dermamyotome (VLL, the progenitor cells for hypaxial muscles), (McDermott et al. 2005). Gli2/Gli3 double mutants display a variety of foregut defects from esophageal atresia, tracheo-esophageal fistula, severe lung phenotype, smaller than normal hepatic and pancreatic buds, to complete absence of lung, trachea and esophagus (Motoyama et al. 1998). Consistent with the genetic studies in mice, in some Pallister-Hall syndrome patients, foregut malformations including lung lobulation defects, tracheal stenosis and tracheo-esophageal fistula have been observed (Verloes et al. 1995; Kang et al. 1997a). Furthermore, Gli3 has also been implicated in the normal development of stomach (Kim et al. 2005), another foregut derivate. These data suggest key roles of Gli3 in the normal development of several foregut derivates. Human GLI3 has also been shown to play a role in the normal development of pituitary gland (Kang et al. 1997a). Gli3 mutations are found to cause renal dysplasia/aplasia in humans and mice (Kang et al. 1997a; Bose et al. 2002), and similarly, its mRNA has been detected in the embryonic urogenital structures (Hu et al. 2006). Mouse Gli3 has also been shown to perform a critical role in development of external genitalia and is broadly expressed there (Haraguchi et al. 2001). A genetic study in mice has revealed the importance of Gli3 functions in the normal development of mammary line and mammary placodes (Veltmaat et al. 2006), and by immunohistochemistry the Gli3 protein was localized in all mammary buds (Hatsell & Cowin 2006). The GLI3 associated dominant genetic disorder Greig cephalopolysyndactyly syndrome (GCPS) and mouse mutant extra-toes (Xt) have shown a crucial role of this gene in limb development (Vortkamp et al. 1991; Hui & Joyner 1993). Subsequent studies in mice revealed multiple functions of Gli3 in limb patterning and morphogenesis.

Reflecting the complex roles of Gli3 in early mouse embryogenesis, previous results indicated that multiple evolutionary conserved GLI3 associated human cis-regulators control almost the entire known repertoire of endogenous Gli3 expression in neural tube and limbs (Abbasi et al. 2007, 2010; Paparidis et al. 2007; Alvarez-Medina et al. 2008; Coy et al. 2011). However the genetic mechanisms controlling GLI3 expression in tissues other than brain/limbs such as craniofacial structures and internal organs were unknown and have been investigated in this study.

Cis-regulatory underpinnings for GLI3 expression in craniofacial structures

Consistent with Gli3 roles in the normal development of craniofacial structures and its widespread expression there, the CNE1 was found to drive expression in numbers of facial domains including nasal processes, the derivates of branchial arch material, including maxillary and mandibular components of jaw, and Meckel's cartilage. Furthermore, in the oral cavity CNE1 recapitulated Gli3 functions in palatal and tooth development.

In addition to the limbs, CNE6 directed transgene expression at early stages of development (up to E10.5) within a rostroventral domain of the telencephalon. By E11.5 the strong lacZ activity was seen in the facial region within the precartilage primordium of the nasal septum. This spatial shift in transgene expression from early rostroventral telencephalon to the facial region is consistent with the observation that mesenchymal cells in the later region are derived from the migration of neural crest cells from the forebrain (Noden 1983; Couly et al. 1998). Even though both CNE1 and CNE6 directed reporter expression in the nasal system, intriguingly, close histological analysis revealed a separation of their target sites, that is, CNE1 induced reporter activity specifically within the nasal capsule, whereas CNE6 was active within the nasal septum (compare reporter expression in Fig. 2E,L).

Cis-regulatory control of GLI3 expression in several internal organs, mammary placodes, eye and external genitalia

Consistent with Gli3 functions in foregut development (Motoyama et al. 1998), the CNE10 cis-region was found to drive transgene expression in several foregut derivatives, which included pharynx, esophagus, tracheal duct, liver, gall bladder, pancreatic bud and along the wall of stomach and in the lumens of duodenum (Fig. 4). The CNE10 activity also coincides with Gli3 functions and its expression within embryonic urogenital structures (Schimmang et al. 1992). Gli3 has been shown to be expressed within several structures of eye (Aoto et al. 2002; Zaki et al. 2006), in agreement with this, at E11.5 CNE10 drove lacZ expression in the sensory layer of retina, by E12.5 the expression within the eye was more widespread. This element also indicated activity within the presumptive pituitary formatting region (Rathke's pouch). Consistent with localization of mouse Gli3 mRNA in all of mammary buds and in the external genitalia (Hatsell & Cowin 2006), the CNE10 induced reporter expression strongly in these tissues.

Cis-region reside within intron-13 of GLI3 regulates a subset of its known expression in developing somites

Within the somites, Gli3 expression is widespread, and known to play a vital role in epaxial and hypaxial myotome formation (McDermott et al. 2005). Consistent with its known roles there, a cis-regulatory region within intron-13 of GLI3 upregulates reporter expression specifically within the ventral-lateral lip of dermomyotome (Fig. 5C). Remaining aspects of endogenous Gli3 expression in somites were not recapitulated by this enhancer, suggesting the importance of normal genomic context in defining the full repertoire of enhancer function or alternatively other unknown cis-region(s) might work in conjunction with CNE9 to set down the complete aspects of Gli3 function in developing somites.

Intra-GLI3 enhancers depicts the preservation and divergence of target site specificity during the course of vertebrate evolution

Previously we have analyzed the 10 independent CNEs in zebrafish embryos (Abbasi et al. 2007). Seven of these intra-GLI3 CNEs, including CNE1, 2, 9, 10 not only recapitulated the known expression repertoire of zebrafish gli3 but also showed considerable functional redundancy with respect to the site of expression (Abbasi et al. 2007). In contrast, for a similar subset of CNEs (CNE1, 2, 9, 6, 11 10) the functional data from mice suggest that multiple independent enhancers control the expression of Gli3 in distinct developmental domains, largely in a non-redundant fashion. Additionally, the comparative mouse-fish data showed that some of these ancient enhancers diverged with respect to target embryonic domains in which they dictate expression in each of these animals, whereas other retained the specificity of their action at least in part. For instance, in zebrafish the CNE1 and CNE9 directed reporter expression primarily in neuronal subpopulations in brain and spinal cord, and similarly these elements were highly active in the neural tube of mice (Abbasi et al. 2007, 2010). The forebrain specific regulatory functions of ultra-conserved element, CNE2 appeared to be exceptionally conserved among fish and tetrapod lineages (Paparidis et al. 2007).

The most prominent sites of CNE10 activity in zebrafish were various subdivisions of eye. In addition, it also induced reporter expression in the lower jaw primordial (Abbasi et al. 2007). In mice the functions of the same element seem to be conserved with respect to eye (Fig. 4); however, CNE10 was unable to show any activity in the mandibular arch region of the mouse embryos. Furthermore CNE10 governed reporter expression in several foregut derivatives and in the urogenital structures of developing mouse embryos, whereas no reporter expression was observed in the comparable structures of zebrafish embryos (Abbasi et al. 2007).

These findings demonstrate that that even though Gli3 in fish and mammals shared evolutionary conserved multiple non-coding sequence elements functioning as cis-acting regulators, the functions of this ancient “gene regulatory catalogue” might have diverged at two levels: In mammals these cis-regulators attained a higher level of functional modularity by abolishing the potential for redundant expression control: second in order to cope with differential developmental and anatomical needs of fish and mammals the target site specificity of some of these elements has diverged significantly among these two lineages. This sort of functional differentiation might have been achieved either through changes in the overall span of enhancers or through turnover of transcriptional factor binding site inputs.


Taken together with our previous reports, we defined a complex cis-regulatory landscape for GLI3, an important downstream effector of Hh signaling cascade. Cis-regulatory repertoire of GLI3 seems to be docked largely inside the intronic intervals of this gene and this arrangement remained preserved since the divergence of tetrapod and teleost lineages 450 million years ago. We showed that this ancient catalogue of cis-acting sites controlling GLI3 expression at almost all known sites of its endogenous activity, that is, neural tube, limbs, craniofacial structures, eyes, internal organs, mammary buds and in the external genitalia. Considerable functional redundancy was observed among the activity of these elements in zebrafish, whereas in mice, higher level of functional modularity was seen; multiple elements controlling expression independently in multiple tissues largely in a non-redundant fashion. This higher modularity signals the reduced pleiotropic effects of mutational events and thus implicates cis-acting sites crucial for developmental and morphological evolution. In addition these cis-regulatory elements will help in understanding the genetic basis of those potential GLI3 associated human birth defects, which cannot be attributed to a mutation in coding sequence of this gene. In such cases these enhancers can be searched for those mutations that can potentially affect the space and time availability of GLI3 transcript during embryogenesis.


This work was funded by Deutsche Forschungsgemeinschaft (Gr373-21; GRK 767), Stiftung P.E. Kempkes, Marburg, Germany and Higher Education Commission, Islamabad, Pakistan. For the maintenance of transgenic mice we acknowledge members of the Central Animal Facility of the Medical Faculty of the Philipps-University Marburg.

Author contributions

Wet lab experiments were designed and performed by K.-H.G., A.A.A., and R.M. with advice in histological analysis by A.S., and S.K. Computational analyses were performed by A.A.A. The manuscript was written by K.-H.G. and A.A.A. All authors read and approved the final publication.

Conflict of interest

The authors have declared that no competing interests exist.