Functional Characterization of a Heterozygous GLI2 Missense Mutation in Patients With Multiple Pituitary Hormone Deficiency

Abstract

Context:

The GLI2 transcription factor is a major effector protein of the sonic hedgehog pathway and suggested to play a key role in pituitary development. Genomic GLI2 aberrations that mainly result in truncated proteins have been reported to cause holoprosencephaly or holoprosencephaly-like features, sometimes associated with hypopituitarism.

Objective:

Our objective was to determine the frequency of GLI2 mutations in patients with multiple pituitary hormone deficiency (MPHD).

Design:

Patients were selected from participants in the Genetics and Neuroendocrinology of Short Stature International Study (GeNeSIS) program. Patients with mutations within established candidate genes were excluded.

Patients:

A total of 165 patients with MPHD defined as GH deficiency and at least 1 additional pituitary hormone deficiency were studied regardless of the presence of extrapituitary clinical manifestations.

Main Outcome Measures:

Prevalence of GLI2 variations in MPHD patients was assessed and detailed phenotypic characterization is given. Transcriptional activity of identified GLI2 variants was evaluated by functional reporter assays.

Results:

In 5 subjects, 4 heterozygous missense variants were identified, of which 2 are unpublished so far. One variant, p.R516P, results in vitro in a complete loss of protein function. In addition to GH deficiency, the carrier of the mutation demonstrates deficiency of thyrotrope and gonadotrope function, a maldescended posterior pituitary lobe, and polydactyly, but no midline defects.

Conclusions:

For the first time, we show that heterozygous amino acid substitutions within GLI2 may lead to MPHD with mild extrapituitary findings. The phenotype of GLI2 mutations is variable, and penetrance is incomplete. GLI2 mutations are associated with anterior pituitary hypoplasia, and frequently, ectopy of the posterior lobe occurs.

Studies in zebrafish, chicken, and rodents suggest that sonic hedgehog (Shh), a morphogen that is expressed in the ventral diencephalon and the oral ectoderm except the Rathke's pouch primordium, has a role in early steps of pituitary ontogenesis. Deprivation of Shh expression or action results in pituitary hypoplasia and malfunction (1, 2). The Shh receptor patched (Ptch1) as well as the Shh effector transcription factors Gli1, Gli2, and Gli3 are expressed in Rathke's pouch, indicating competence of the developing gland to receive and respond to Shh signals (3). Gli2 knockout in mice results in severe skeletal abnormalities including absence of vertebral body and intervertebral discs, shortened limbs and sternum, cleft palate, and tooth defects (4). Pituitaries of Gli2 knockout mice are absent in some but not all offspring (5). Moreover, Gli2-mediated Shh signals have been shown to be involved in dorsoventral and anteroposterior patterning and brain development (6–8).

Human GLI2 (MIM 165230) is a 1586-amino acid protein (UniProtKB accession number P10070), which is encoded by 13 exons (ENSEMBL accession number ENSG00000074047; GenBank accession number NM_005270.4) on chromosome 2q14. The functional domains of the human GLI2 protein are not yet well understood. In the murine Gli2 ortholog, a central DNA-binding domain consisting of 5 zinc fingers is flanked N-terminally by a repressor domain and C-terminally by a transactivation domain (9). Accordingly, a human variant lacking the N-terminal repressor domain (ΔN) in vitro shows a 30-fold higher transcriptional activity compared with the full-length protein (10).

In humans, defects of SHH secretion and signaling during early stages of development are considered to be one of the main causes of holoprosencephaly (HPE) (11, 12). The clinical phenotype is characterized by a remarkable variability even among patients with identical mutations (11, 13) and may or may not comprise pituitary defects (14, 15). Similarly, several genomic GLI2 aberrations have been reported that are phenotypically characterized by a variable degree of pituitary and extrapituitary malformations. In addition to its involvement in carcinogenesis (16), mutations within the GLI2 gene have been initially reported to cause HPE or HPE-like features and polydactyly (17). Since then, GLI2 sequence variants have been identified in patients with craniofacial anomalies ranging from severe midline defects to absence of clinical findings in mutational carriers (18–20). In some patients, these findings have been reported to occur associated with hypopituitarism. Multiple pituitary hormone deficiency (MPHD) may also represent the sole clinical manifestation, although polydactyly is a frequently observed concomitant abnormality. However, the dominant inheritance pattern is characterized by incomplete penetrance and highly variable expressivity. The lack of classical Mendelian inheritance may be in part attributable to co-occurring mutations in interacting genes, although final evidence is lacking (21). Interestingly, all patients with GLI2 mutations reported so far that have pituitary perturbations carry mutations that cause introduction of a premature termination codon and thereby have a truncated GLI2 protein (10, 17, 19, 22). Recently, GLI2 missense mutations have been reported for patients with HPE and HPE-like features as well as isolated cleft lip and palate with polydactyly and branchial arch anomalies but without involvement of the pituitary gland (18, 21). The pathogenic impact of these missense variants awaits in vitro analyses. So far, molecular characterization of GLI2 missense and loss-of-function mutations by Roessler et al (10, 17) revealed lack of GLI2 activity for those mutations resulting in truncated GLI2 proteins but wild-type activity for the missense mutations.

Here we describe a patient with MPHD and polydactyly, who carries a heterozygous missense mutation within the GLI2 gene. In vitro analysis supports our assumption of a causative relationship of the mutation with the patient's phenotype.

Subjects and Methods

Subjects

We studied 168 patients with MPHD (165 pedigrees; mean age 9.13 years at the time of study entry [range 0.11–28.39 years]) from 15 countries selected from the Genetics and Neuroendocrinology of Short Stature International Study (GeNeSIS) program sponsored by Eli Lilly and Company (Indianapolis, Indiana). All patients had GH and TSH deficiency. ACTH and/or cortisol deficiency was reported in 62 and gonadotropin deficiency in 38 patients; however, because 87 children (63%) were prepubertal at evaluation (≤10 years), FSH/LH deficiency might evolve later on during development. Abnormal magnetic resonance imaging (MRI) was documented for 96 cases (anterior pituitary hypoplasia, 65 cases; ectopic or absent posterior pituitary, 52 cases; other anomalies affecting pituitary stalk, infundibulum, and sella turcica, 15 cases). In addition, DNA of 50 healthy subjects of normal height was investigated as a control. Written informed consent was obtained from all subjects and/or their legal guardians. The study was approved by the local institutions' ethics review boards according to the Declaration of Helsinki as required.

Screening for GLI2 mutations

Fragments of the GLI2 gene were PCR amplified from peripheral blood genomic DNA, prescreened by denaturing HPLC (WAVE System; Transgenomic, Glasgow, United Kingdom), and samples that showed an abnormal elution pattern were further analyzed by direct sequencing (ABI PRISM 310 DNA analyzer; Applied Biosystems, Foster City, California).

Database mining

The frequency of identified variants in publicly accessible variant databases was ascertained by performing gene- or sequence-centered queries at NCBI's dbSNP, the National Heart, Lung, and Blood Institute Exome Sequencing Project, and the 1000 Genomes Project (23). Allele frequency data of the study cohort presented herein were compared with the corresponding allele frequencies as stated for the published control populations by two-sided Fisher's exact test. Evaluation to predict the possible impact of the amino acid substitutions was obtained using the Polymorphism Phenotyping (PolyPhen-2) tool. URLs and accession dates are given in the Supplemental Material (published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org).

Functional analyses

Mouse embryonic fibroblast (NIH-3T3) or murine corticotrophinoma (AtT-20) cells, in which the Shh pathway is active (24, 25), as well as COS7 cells were transfected with pCS2 expression vectors containing either wild-type or mutant GLI2 cDNA to provide expression of GLI2 proteins for luciferase reporter and EMSAs, respectively. Plasmids containing full-length GLI2 (pCS2-GLI2fl) and GLI2 missing the N-terminal repressor domain (pCS2-GLI2ΔN) were kindly provided by Dr E. Roessler (National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland). Variants were introduced into wild-type cDNA by site-directed mutagenesis alone or in combination according to their native occurrence.

As reporter plasmid in cotransfection luciferase assays we used a pGL4.10 vector containing an 8-fold repeat of the 3′-Gli–binding site from Hnf3β floor plate enhancer in front of the chicken δ-crystallin minimal promoter that drives expression of the firefly luciferase (pGL4.10-8 × 3′-Gli-BS). The construct was generated from the original 8 × 3′-Gli BS–LucII plasmid kindly provided by Dr H. Sasaki, Osaka University, Osaka, Japan (26, 27). Reporter assays were performed using the dual-luciferase reporter assay system according to the recommendations of the manufacturer (Promega, Madison, Wisconsin).

Nonradioactive EMSAs were basically performed according to the LightShift chemiluminescent EMSA kit protocol (Thermo Fisher Scientific, Bonn, Germany) with some minor modifications. A 5′-biotin–labeled oligonucleotide comprising the consensus GLI-binding site and the surrounding sequence as present in the human PTCH1 promoter (5′-GTTGCCTACCTGGGTGGTCTCTCTACTTT; consensus GLI-binding site highlighted) was used as probe. An oligonucleotide with the same sequence but without biotin label was used in competition experiments. GLI2 protein expression was verified by Western blotting.

Detailed protocols are provided in the Supplemental Material.

Results

We identified 5 subjects in 5 families bearing nonsynonymous GLI2 variations in a cohort of 165 patients with MPHD (3.0%). Before the analysis of the GLI2 gene, mutations in the POU1F1, PROP1, HESX1, LHX3, and LHX4 genes were excluded in all patients. One patient had a GLI2 dinucleotide substitution affecting two amino acids (patient C, p.[M1444I; L1445F]); two patients, who were not related to each other, carried an identical combination of 2 missense variations on one allele (patients A and B, p.[M1352V; D1520N]). Two more patients (patient D, p.R516P; patient E, p.R1543H) had single amino acid substitutions. All variations were heterozygous, and none of them was found in 50 German control individuals. However, all but one of the identified variations (p.R516P) have been found in large population exome sequencing projects (Table 1). The p.R1543H variant was identified only in the Exome Sequencing Project (ESP) cohort. Both the dinucleotide substitution and the monoallelic double variation have been reported previously (Table 1). Only variants p.R516P and p.R1543H show a high phylogenic conservation, a fact that is also reflected in the Polyphen prediction as damaging (Table 1 and Supplemental Figure 1).

Table 1.

Summary of GLI2 Variations Identified in This Study, Allele Frequency Data and Literature Review

Amino Acid Positiona	Nucleotide Positionb	PolyPhen	dbSNP ID	Allele Frequencies			Other Studies
				This Study (Controls)	ESPc	1000 Genomesc	Allele Frequency (Controls)	Context	Ref.
R516P	1547G>C	Probably damaging		0/100
M1352V	4054A>G	Benign		0/100	101/6863	11/2188	0/192	CFA	18d
					P = .335	P = .684	2/162	SHFM	32
D1520N	4558G>A	Probably damaging	rs114814747	0/100	106/6914	11/2188	0/192	CFA, left anophthalmia	18d
					P = .242	P = .684	2/162	SHFM	32
M1444I	4332G>A	Benign		0/100	9/7011	13/2188	3/110	MPHD	19e
					P = .369	P = 1.000	0/>200	CFA	20f
							1/330	SHFM	32
							2/348	SHFM	31
L1445F	4333C>T	Probably damaging		0/100	9/7011	13/2188	3/110	MPHD	19e
					P = .369	P = 1.000	1/330	SHFM	32
							2/348	SHFM	31
R1543H	4628G>A	Probably damaging		0/100	5/7015
					P = .241

Abbreviations: CFA, craniofacial anomalies; SHFM, split hand/split foot malformation.

^aAffected amino acid position (UniProtKB P10070); underlines indicate variants occurring in cis on the same allele.

^bPosition of the affected nucleotide within the GLI2 open reading frame (GenBank NM_005270).

^cP values (Fisher's exact test) are shown. Statistics as well as restrictions and possible caveats for PolyPhen, ESP, and 1000 Genomes data are described in Subjects and Methods. For ESP, only European-American allele data are shown.

^dBertolacini et al (18) identified the p.M1352V and p.D1520N variants in distinct patients.

^eVariants p.M1444I and p.L1445F described by Franca et al (19) are co-occurring with the putative pathogenic p.L788fsX794 GLI2 mutation.

^fRahimov et al (20) report the p.M1444I variant without the frequently co-occurring p.L1445F variation.

Patients

Patients presented with MPHD affecting all pituitary hormones with only minor differences. Postnatal growth before GH replacement therapy was retarded (range −3.1 to −4.6 SD score [SDS]). Extrapituitary manifestations were, if occurring at all, generally mild and comprised postaxial polydactyly (patient D), crowding of midfacial features, and an antimongoloid slant in the first years of life as well as severe right-sided visual impairment (patient E). Severely afflicted vision in this patient was due to hypoplasia of the right optic nerve with an absent right-sided chiasm as revealed by MRI. In addition, MRI in all patients disclosed a hypoplastic anterior lobe of the pituitary, and most cases showed abnormally developed or located neurohypophysis and pituitary stalk to a variable extent. Because patient D is carrier of the causative pathogenic p.R516P GLI2 mutation, a detailed clinical description is given below; case reports of all other patients bearing GLI2 variants are provided in the Supplemental Material.

Patient D

Patient D carries a single point mutation (c.1547G>C) in exon 10 that changes arginine at position 516 to proline (p.R516P). The patient, a girl from Australia and second of three children of a Caucasian marriage, was presented because of short stature at the age of 5.6 years with a height of 97.2 cm (−3.13 SDS) and a weight of 13.4 kg (−3.16 SDS) (Figure 1). Bone age assessed at various occasions was retarded by approximately 1.5 years. Pituitary testing revealed a peak GH of 0.5 ng/ml in glucagon testing. Accordingly, IGF-I was decreased to 11.5 (normal 53–160) ng/ml. Central hypothyroidism became evident at age 10.9 years with TSH of 0.98 (normal 0.64–4.67) mIU/L and free T₄ of 6.9 (normal 10–23) pmol/L. T₄ replacement was commenced. By age 13 years, she had no breast development with prepubertal estradiol and gonadotropin measurements (LH <1 U/L; FSH 4.1 U/L; estradiol 64 pmol/L). Puberty was induced with increasing estrogen supplements from age 14 years with cyclical hormones introduced at age 17 years, which induced menarche and continue to be administered. Adrenal axis function appeared to be normal at several stimulation tests performed between 7 and 18 years. Prolactin levels measured at 2 occasions were in the low-normal range (95 and 120 mU/L at 10.5 and 13.0 years, respectively).

Figure 1.

Patient D. A, Growth chart of the patient, showing the final height of father (filled square) and mother (filled circle). The period of GH replacement therapy is marked at the bottom of the chart. Percentiles are shown as indicated. B, Family tree of patient D. The proband (P) is the second of 3 children of a nonconsanguineous Caucasian marriage and as his father carrier of the p.R516P mutation (arrows). Asterisks indicate individuals for whom molecular genetic analysis has been performed. Filled symbol, pan-hypopituitarism; hatched symbols, polydactyly. C, MRI performed at 7 years. Sagittal scans show mild anterior pituitary hypoplasia (arrow) and a maldescended posterior lobe (arrowhead).

Recombinant human GH replacement therapy was initiated at the age of 7 years (0.175 mg/kg/wk). During the first year of GH treatment, height velocity increased 2-fold from 5.1 to 10.4 cm/y (Δheight [SDS] 1.1). Continuation of GH replacement over nearly 9 years resulted in a normal height of 168.8 cm (0.89 SDS) at the patient's age of 17.6 years (Figure 1). At that time, she weighed 69.9 kg (1.15 SDS; body mass index of 24.5 [0.86 SDS]). After cessation of GH therapy stimulated GH, IGF-I and IGF-binding protein-3 serum concentrations returned to subnormal values, thus confirming the GH-deficient status.

No midline defects were reported. However, the patient had unilateral polydactyly (a small extra digit on 1 hand), which was corrected by simple ligation in the first week of life. MRI performed at 7 years showed mild anterior pituitary hypoplasia and a maldescended posterior lobe (Figure 1). Formal IQ testing (Wechsler Intelligence Scale for Children [WISC-R], WISC-IV) performed at the age of 10.5 years demonstrated a full-scale IQ of 84.

Genetic evaluation of the GLI2 gene in the parents revealed inheritance of the c.1547G>C mutation from the father (Figure 1). Both parents are of normal height (mother, 169.0 cm; father, 176.5 cm) with no endocrine, dysmorphic, or neural abnormalities. Also, the patient's older sister presented without any obvious findings, whereas her younger brother had postaxial polydactyly with 1 extra digit on each hand. However, he grew normally, suggesting an unaffected pituitary gland. Unfortunately, both siblings were not available for genetic analysis.

Functional analysis

To evaluate whether the identified variations interfere with the basic transactivation and DNA-binding properties of the GLI2 transcription factor, we performed luciferase reporter assays. Transference of the promoter construct from the original backbone of the pδ51 LucII plasmid into the optimized pGL4.10 reporter plasmid (see Subjects and Methods) led to an approximately 10-fold increase of reporter activity induced by GLI2ΔN compared with wild-type GLI2 in NIH-3T3 cells (Supplemental Figure 2). Disruption of the GLI-binding sites completely abolished transactivation by both full-length and suppression-domain–devoid GLI2 wild-type effector constructs (data not shown).

Introduction of the p.[M1352V; D1520N] (patients A and B) or p.[M1444I; L1445F] (patient C) double variants into GLI2ΔN reduced transcriptional activity to 75.2% or 59.5%, respectively, which was significantly different from wild-type activity (Figure 2A). The same variations within full-length GLI2 resulted in a decrease of luciferase activity to 64.7% (p.[M1352V; D1520N]) and 74.8% (p.[M1444I; L1445F]). In contrast, the p.R516P mutation completely abolished the transcriptional activity of both full-length and the repressor-domain–deficient GLI2 protein isoforms. Surprisingly, the p.R1543H identified in the most severely affected patient (patient E) did not compromise transcriptional activity under the conditions tested within this study.

Figure 2.

Proteins encoded by mutated GLI2 genes display distinct degrees of their transactivation properties. Expression vectors for mutant GLI2 and wild-type (WT) proteins (full-length [fl] and N-terminally truncated [ΔN]) were transiently cotransfected into NIH-3T3 cells with a luciferase reporter gene under the control of an 8-fold repeat of the Hnf3β Gli-binding site. Promoter activity was assayed by measuring luciferase activity 24 hours after transfection. Negative controls (vector only [VO]) received equivalent amounts of empty expression vector plasmid. Activities measured as relative light units (RLUs) are the mean of at least triplicate assays ± SEM. RLUs were normalized and compared (where not otherwise indicated) with the activity of the corresponding WT construct. Significance levels are indicated as follows: *, P ≤ .05; **, P ≤ .01; ***, P ≤ .001; n.s., not significant. MD.VN refers to p.[M1352V; D1520N]; ML.IF to p.[M1444I; L1445F]. A, Activity of the double and single variants as found in our patients. B, For the double mutants, each variant was cloned and tested separately and compared with the corresponding double and WT construct.

Similar results were obtained when cotransfection experiments using the same constructs were performed in the murine corticotrophinoma cell line AtT-20 (data not shown).

Because 3 of 5 patients in our study carried (known) monoallelic double variants, we sought to elucidate the impact of each individual variation. The rationale behind this approach was to clarify whether the synergistic effect of two variant amino acid residues might have a more damaging consequence for the affected protein than a single substitution alone or, alternatively, if either amino acid alteration might relieve a more detrimental effect of the other one. Luciferase reporter assays of p.M1352V and p.D1520N as well as p.M1444I and the p.L1445F single variants compared with the corresponding double variation in NIH-3T3 cells did not provide strong evidence for either of the suggested mechanisms (Figure 2B).

Specific GLI2 mutations have been previously shown to exhibit dominant properties toward the wild-type protein. Cotransfection of p.R516P along with wild-type GLI2 did not result in a dominant inhibition of wild-type GLI2 transcriptional activity (Figure 3A). Because dominant-negative effects have been shown to require an intact DNA-binding ability of GLI2, we tested the DNA-protein interaction using EMSAs. As expected, wild-type GLI2 strongly associated with the PTCH1 promoter-derived probe, and competition with the corresponding unlabeled oligonucleotide abrogated the signal of the slowly migrating complex (Figure 3B). In contrast, p.R516P GLI2-expressing lysates did not demonstrate an electrophoretic shift of the DNA probe signal above background, suggesting loss of the DNA-binding capacity of the mutant protein.

Figure 3.

p.R516P does not show dominant-negative properties toward GLI2 wild-type protein and fails to interact with the cognate GLI2 DNA-binding site. A, Dual-luciferase assays were basically performed as described for Figure 2. NIH-3T3 cells were cotransfected with equal amounts of wild-type and p.R516P expression plasmids, or the same amount of each individual plasmid supplemented by the empty vector (vector only [VO]) to avoid differences in transfection efficiency. Relative light units (RLUs) were normalized and compared with the activity of the corresponding WT construct. Significance levels are indicated: ***, P ≤ .001; n.s., not significant. B, EMSAs using whole-cell lysates of COS7 cells transfected with the indicated GLI2 cDNA plasmids or a vector control (VO). Probe was a biotin-labeled PTCH1 promoter-derived GLI-binding site oligonucleotide as described in Subjects and Methods. Competition experiments (+) were performed using a 100-fold excess of the unlabeled PTCH1 oligonucleotide. Expression of wild-type and mutant GLI2 proteins was confirmed by Western blotting using an anti-Myc-tag antibody (WB α-Myc).

Discussion

GLI transcription factors are the major effector proteins of the SHH pathway. Human SHH mutations have been shown to cause HPE, sometimes affecting the pituitary in a dominant manner but with incomplete penetrance. Accordingly, several variations of the human GLI2 gene, which is regarded as the major GLI family effector in humans, have been identified in the past. Carriers of these GLI2 variations present with clinical manifestations characterized by some phenotypic overlap with SHH mutational carriers. In the present investigation, we screened patients with MPHD for GLI2 variants regardless of the presence of extrapituitary clinical manifestations.

In 5 subjects, 4 heterozygous variant alleles were identified, of which 2 represent novel, so far unpublished, missense variations (Table 1). The 4 variant carriers that were available for further clinical examinations showed pan-hypopituitarism and a hypoplastic anterior pituitary. Two of the 4 patients had an ectopic neurohypophysis, and 1 patient had no identifiable posterior pituitary tissue. Abnormal location of the neurohypophysis is not a common feature of MPHD but appears to be a frequent finding in GLI2 mutational carriers. An ectopic or undescended posterior pituitary lobe in the context of hypopituitarism has otherwise been reported only for carriers of HESX1, SOX3, OTX2, and LHX4 mutations. The first GLI2 mutations were identified in patients presenting with hypopituitarism and variable manifestations within the HPE spectrum, but detailed description of pituitary imaging, particularly with respect to the posterior lobe, was not available (10, 17). In a well-investigated cohort of patients with isolated GH deficiency or MPHD, Franca and colleagues (19) recently reported on 3 index probands from Brazil without HPE or HPE-like features, who bear GLI2 loss-of-function mutations (due to early peptide chain termination), and all demonstrated an ectopic or invisible posterior pituitary lobe. Maldevelopment of the posterior pituitary is in congruence with findings in mice, where in all Gli2 mutants, the posterior lobe was absent (28).

In addition, the 2 carriers of the novel variations p.R516P and p.R1543H presented with extrapituitary findings, ie, polydactyly or craniofacial anomalies, respectively. Polydactyly has been found with high frequency in patients with pituitary anomalies and GLI2 mutations and appears to be an extrapituitary hallmark of this condition (10, 17, 19). However, the appearance of an extra digit in GLI2 patients is not necessarily associated with pituitary insufficiency but has also been described in patients with midline defects without reported hypopituitarism (18, 29).

So far, GLI2 mutations with proven pathogenic impact have been exclusively identified in a heterozygous state and comprised nonsense and frameshift mutations, all leading to truncated proteins lacking a functional C-terminal transactivation domain. Loss of function has been demonstrated for some of them by distinct in vitro approaches, but experimental reasoning is so far limited to the mutations identified by Roessler et al. (10, 17). Evaluation of the influence of the GLI2 variations identified within the present study on the transcriptional activity of the GLI protein revealed a quantitative effect for the double variants (25%–40% decrease) but a complete loss of function for the p.R516P mutation. In the latter, loss of transcriptional activity was independent from the cell system applied and did not change with presence or absence of the N-terminal suppressor domain. Arginine residue 516 resides within the third zinc finger of the DNA-binding domain of GLI2, which consists of 5 consecutive Cys₂His₂ fingers, whose zinc finger motifs 4 and 5 make the strongest base contacts (30). In the crystal structure of a GLI-DNA complex, the side chain of Arg516 makes direct contact through a hydrogen bond to a phosphodiester oxygen of the DNA backbone (30). Replacement of proline for arginine can be assumed to result in the loss of this hydrogen bond because proline cannot act as hydrogen bond donor. This assumption was confirmed by EMSAs that demonstrated that the p.R516P GLI2 protein is not able to form complexes with a consensus GLI-binding site anymore.

Severity of the clinical manifestations in our study appeared to worsen when GLI2 variants occurred within protein domains with critical functional importance; p.R516P is assigned to the DNA-binding zinc finger domain, whereas p.R1543H resides within the essential transactivation domain. Failure to detect any loss of transcriptional activity as observed for the p.R1543H variant might be due to the fact that our test system relies on an artificial minimal promoter construct. The action of additional factors (such as coactivators), operating in the physical context of native GLI2 target genes, might be affected by the conservative histidine for arginine substitution in the native environment but not in our system. Alternatively, indispensability of the C-terminal peptide has been demonstrated for the mouse ortholog (9), but functional importance might differ in the human GLI2 protein. Additional studies employing more specific experimental models are necessary.

Although the phenotype of the p.R516P carrier was modestly affected compared with GLI2 patients at the severe end of the HPE spectrum, we investigated whether the proline for arginine substitution demonstrates a dominant-negative effect on the wild-type GLI2 protein. Cells cotransfected with wild-type and mutant GLI2 constructs revealed an induction of the reporter gene comparable to cells expressing wild-type GLI2 alone. Such a lack of functional dominance was also reported for a zinc-finger-3/4–lacking protein independent from the presence or absence of the N-terminal suppressor domain, and thus our results support the observation by Roessler et al (10) that dominant-negative action requires intact zinc finger domains and DNA-binding activity. The pathogenic impact of the p.R516P mutation, therefore, is probably the consequence of a dose effect.

In summary, evidence for a causative pathogenic effect, inferred from variant distribution within control cohorts and functional evaluation, can be provided for the p.R516P GLI2 mutation. Clinical manifestations of GLI2 mutational carriers are highly variable and may include HPE and HPE-like findings, midline facial anomalies, and maldevelopment of the anterior and posterior pituitary lobes leading to variable hormone deficiencies of the adenohypophysis and neurohypophysis, and may also comprise defects in the morphogenesis of the extremities resulting in preaxial or postaxial polydactyly and possibly being associated with the split hand/split foot malformation (31, 32). However, in conditions of pan-hypopituitarism, GLI2 mutations appear to be frequently accompanied by an ectopic neurohypophysis and polydactyly. Striking clinical heterogeneity hallmarked by a lack of consistently occurring clinical findings as well as incomplete penetrance of specific GLI2 mutations point to a modifying impact exerted by interacting genes (where common GLI2 variants with modest functional impact might act themselves as modifiers) and pathways within an individual genetic background (33, 34). Currently, the concept of autosomal random monoallelic expression is emerging, which indicates that a fraction of genes throughout the genome are expressed from only one randomly selected allele. Although the alleles that become transcribed differ within a single organism, there is a remarkable clonal stability (eg, within a given tissue), which might result in the preferential expression of a heterozygous disease-causing protein variant (35).