Novel Lethal Form of Congenital Hypopituitarism Associated With the First Recessive LHX4 Mutation

Abstract

Background:

LHX4 encodes a member of the LIM-homeodomain family of transcription factors that is required for normal development of the pituitary gland. To date, only incompletely penetrant heterozygous mutations in LHX4 have been described in patients with variable combined pituitary hormone deficiencies.

Objective/Hypothesis:

To report a unique family with a novel recessive variant in LHX4 associated with a lethal form of congenital hypopituitarism that was identified through screening a total of 97 patients.

Method:

We screened 97 unrelated patients with combined pituitary hormone deficiency, including 65% with an ectopic posterior pituitary, for variants in the LHX4 gene using Sanger sequencing. Control databases (1000 Genomes, dbSNP, Exome Variant Server, ExAC Browser) were consulted upon identification of variants.

Results:

We identified the first novel homozygous missense variant (c.377C>T, p.T126M) in two deceased male patients of Pakistani origin with severe panhypopituitarism associated with anterior pituitary aplasia and posterior pituitary ectopia. Both were born small for gestational age with a small phallus, undescended testes, and mid-facial hypoplasia. The parents' first-born child was a female with mid-facial hypoplasia (DNA was unavailable). Despite rapid commencement of hydrocortisone and T₄ in the brothers, all three children died within the first week of life. The LHX4(p.T126M) variant is located within the LIM2 domain, in a highly conserved location. The absence of homozygosity for the variant in over 65 000 controls suggests that it is likely to be responsible for the phenotype.

Conclusion:

We report, for the first time to our knowledge, a novel homozygous mutation in LHX4 associated with a lethal phenotype, implying that recessive mutations in LHX4 may be incompatible with life.

LHX3 and LHX4 are members of the LIM-homeodomain (LIM-HD) transcription factor protein family. The primary structure of LIM-HD proteins has been conserved through evolution. They characteristically possess two zinc-coordinated amino-terminal LIM domains and a DNA-binding homeodomain. The LIM domains are multifunctional, mediating interactions that modulate complex formation, target gene transactivation, DNA-binding affinity, and protein stability, among other roles (1). LIM-HD proteins interact with partners, such as the LIM domain-binding protein 1 (Ldb1; also known as nuclear LIM interactor or CLIM2), Ldb2/CLIM1, R-LIM, melanocyte-specific gene-related gene 1, selective LIM-domain binding protein, and the pituitary transcription factor PIT1. Of the mammalian LIM-HD proteins, ISL1, ISL2, LHX2, LHX3, and LHX4 have been implicated in pituitary development (2).

LHX4/Lhx4 is expressed in the developing hindbrain, cerebral cortex, pituitary gland, and spinal cord (3). Lhx4 and the related Lhx3 gene are expressed at mouse embryonic day 9.5 in Rathke's pouch, the primordium of the pituitary gland. By embryonic day 12.5, Lhx4 is expressed in the tissue that will become the anterior pituitary (AP), whereas Lhx3 expression continues throughout the pouch. Thereafter, Lhx4 transcription is reduced with lower levels than Lhx3 in the mature gland (4). Lhx3 and Lhx4 are differentially expressed in subpopulations of adult pituitary cells (5).

The LHX4 protein acts as a transcriptional regulator during pituitary gland and nervous system development. In the pituitary, LHX3/4 proteins have been implicated in the regulation of genes including prolactin (Prl), TSHβ, FSHβ, CGA (encoding αGSU, the common α-glycoprotein subunit of hormones such as TSH and FSH), and the PIT1 transcription factor (5, 6).

In mice, Lhx3 null mutant pituitary precursor cells cease to proliferate before differentiation, whereas in Lhx4 null mutants, these cells differentiate in reduced numbers. A lack of proliferation in Lhx4 mutants causes failure to respond to inductive signals and subsequent misregulation of other transcription factor genes, eg, Lhx3, inevitably leading to increased cell death (5). Therefore patients with LHX4 mutations may have a partial loss of LHX3 function. LHX3 and LHX4 work in conjunction to form a definitive Rathke's pouch and regulate proliferation and differentiation of pituitary lineages. Mice homozygous for Lhx4 mutations die shortly after birth with immature lungs that fail to inflate, whereas heterozygous mice appear normal (7). Lhx4 null mice exhibit incomplete pituitary gland development. In humans, heterozygous autosomal dominant LHX4 mutations are associated with variable and variably penetrant combined pituitary hormone deficiency (CPHD) (5) and are considered to be due to haploinsufficiency rather than dominant-negative effects (8).

Many transcription factors are known to play a role in the etiology of congenital hypopituitarism (eg, HESX1, PROP1, POU1F1, LHX3, LHX4, OTX2, SOX2, and SOX3), but only 5–15% of cases have an associated genetic variant identified (9). This study analyzed DNA extracted from a cohort of patients with hypopituitarism for mutations in the LHX4 gene; we report a novel recessive LHX4 mutation associated with a lethal form of congenital hypopituitarism.

Patients and Methods

Patient cohorts

DNA was extracted from blood samples taken from 97 patients (male:female ratio, 1.1:1) with an ectopic posterior pituitary (EPP); 62 have CPHD, 22 have isolated GH deficiency, and 13 are in the septo-optic dysplasia spectrum. Septo-optic dysplasia was defined by the presence of two of the three classical triad features: optic nerve hypoplasia; midline forebrain defects, eg, agenesis of the corpus callosum and absent septum pellucidum; and pituitary hypoplasia with variable hypopituitarism (10). Patients were recruited from national/international centers between 1998 and 2013. Ethical committee approval was obtained from the University College London Institute of Child Health/Great Ormond Street Hospital for Children Joint Research Ethics Committee, and written consent was obtained from patients and/or parents.

Direct sequencing analysis

Patient DNA samples were screened for LHX4 (ENST00000263726) mutations. Detailed PCR/sequencing conditions are available upon request. For any novel/potentially pathogenic variants identified in LHX4, control databases were consulted, including 1000 Genomes (www.1000genomes.org), dbSNP, Exome Variant Server (EVS), and the ExAC Browser (http://exac.broadinstitute.org/).

Western blot analysis

Western blot analysis was performed as previously described (11) using protein extracts from cells transfected with plasmids expressing wild-type (wt) and mutant LHX4. Rabbit anti-FLAG polyclonal primary antibody (Sigma) and goat antirabbit IgG HRP-linked secondary antibody (Cell Signaling Technology) were used. The negative control was pcDNA3.1 empty vector (data not shown).

Functional studies

Cell culture

HEK293T cells, derived from human embryonic kidney in which LHX4 protein has not been detected, were maintained in DMEM supplemented with 10% calf serum in the presence of penicillin/streptomycin (Invitrogen) and passaged when 80% confluent.

Preparation of constructs for qualitative analysis

Identified mutations were introduced into the full-length human LHX4 cDNA in the pcDNA3.1/FLAG(-)C expression vector (Invitrogen) by site-directed mutagenesis using the QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene; Agilent Technologies) and verified by direct sequencing analysis.

Cell transfection and luciferase assays

HEK293T cells were seeded onto six-well plates at 5 × 10⁵ cells/well and cotransfected with 500 ng of αGSU/Cga reporter plasmid, containing the murine αGSU target gene promoter (−507 to +46 bp) upstream of a luciferase reporter gene, and 250 ng of LHX4 expression vector per well, using FuGene6 transfection reagent (Roche). These concentrations were consistent with previously established LHX4 transfection studies (5). All wells had equal DNA concentrations, and control cultures received pcDNA3.1 empty vector. Total cell protein was determined by the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), and luciferase activity was normalized to the amount of total protein. Further assays involved transient transfection in 24-well plates using equivalent plasmid ratios, this time with the prolactin reporter plasmid containing the rat gene enhancer (−1831 to −1539 bp)/promoter (−422 to +33 bp) upstream of a luciferase reporter gene (Prl-Luc). In addition, half of the described dose of the expression vector was transfected to analyze dose dependency; pcDNA3.1 empty vector was added to this lower dosage to maintain the same concentration of DNA per well. Subsequently, PIT1 expression vector replaced this pcDNA3.1 empty vector so as to evaluate LHX4 and PIT1 synergy. In these assays, cells were transfected with 50 ng/well of Renilla-luciferase reporter, and results were normalized to Renilla-luciferase activity. All wells received mCherry vector (10 ng) to visualize transfection efficiency under the microscope 24 hours after transfection, upon which cells were lysed and luciferase activities measured using the Dual Luciferase Reporter Assay System (Promega) on a luminometer (FLUOstar Optima; BMG Labtech). Results are shown as means ± SD of three independent experiments in triplicate.

Results

Mutational analysis

Direct sequencing analysis revealed a novel homozygous missense variant in LHX4 (c.377C>T, p.T126M) (Figure 1D) in two male patients (pedigree 1). This variant was completely absent from the 1000 Genomes, dbSNP, and EVS databases and absent in homozygous form in the ExAC Browser. It is only present twice in the latter in heterozygous form: in a South Asian and a European individual, respectively, out of a total of 121 140 alleles. This variant results in a predicted substitution of a threonine residue instead of a methionine within a highly conserved region of the LIM2 domain of the protein (Figure 2A). Online prediction models were consulted for this variant: Polyphen2 (score 1.000, possibly damaging), and SIFT (score 0, damaging). All patients screened in this study were negative for variants in all other candidate genes analyzed: LHX3, HESX1, PROP1, PIT1, and SOX3. This approach, however, cannot exclude possible deletions located in other regions of the genome.

Figure 1.

A, The nonconsanguineous pedigree showing three generations. Black bold squares (IIIb, IIIc) indicate presence of the homozygous variant. A line through the circle and squares indicates that the patient is deceased. Half colored circles and squares represent carriers of the variant in heterozygous form, and question marks indicate that the DNA is unavailable. B and C, MRI of patient IIIb. B, Sagittal view showing complete absence of the AP and EPP. C, Magnified section of the MRI (outlined by white rectangle in panel B). D, A novel homozygous missense variant (c.377C>T) was identified in exon 3 of LHX4 in patients IIIb and IIIc, shown by 'N′ and indicated by the arrow with the WT sequence below. E and F, Chest x-rays of patients IIIa and IIIb, respectively, showing classical respiratory distress syndrome.

Figure 2.

A, The threonine residue (represented by the green T) at location p.T126 of LHX4 is conserved across evolution and also in other LIM-HD proteins (LHX3, LHX6, LHX8) and is substituted by methionine in patients IIIb and IIIc. B, A protein prediction model suggests the threonine at p.T126 interacts with an arginine residue at p.R103. This interaction is predicted to be disrupted when this threonine is substituted for methionine (p.M126) as seen in patients IIIb and IIIc. C, A schematic diagram of the LHX4 protein showing the predicted outcomes of other published mutations of LHX4 and the novel homozygous p.T126M variant described here (red font).

Patient phenotypes

Auxology, endocrine investigations, and magnetic resonance imaging (MRI) brain scan data for all affected patients are summarized in Table 1.

Table 1.

Patient Clinical Data

Patient	IIIa	IIIb	IIIc
Sex	Female	Male	Male
GA, wk	35	35	38
Birth weight SDS (weight, kg)	−2.3 (2.4)	−2.4 (2.44)	−2 (2.6)
Birth length SDS (length, cm)	−2.2 (46)	−0.02 (51)	−1.8 (47.5)
FT4, ng/dL	—	0.20 (NR, 0.93–1.7)	0.35 (NR, 0.93–1.7)
Basal TSH, mU/L	—	0.01	<0.01
Basal cortisol, μg/dL	—	2.9	<1.1
ACTH	—	Undetectable	Undetectable
GH at time of hypoglycemia, μg/L	—	—	<0.05
Basal prolactin, μg/L	—	—	<0.5
Other features	Antimongoloid slant of eyes	Poor muscle tone	Poor tone and reflexes
	Persistent hyponatremia	Small phallus (1.7 cm) and absent scrotal rugae	Small phallus (1.5 cm) and absent scrotal rugae
	Poor muscle tone	Undescended testes	Left undescended testis
	Poor respiratory effort, grunting (started on ventilation)	Poor respiratory effort, grunting (started on ventilation)	Low blood glucose
	CXR, atelectasis and RDS-like picture	CXR, atelectasis and RDS-like picture	Persistent hyponatremia
	Low-set crumpled ears	Low-set crumpled ears	Poor respiratory effort, grunting (started on ventilation)
	Small upturned nose with depressed nasal bridge	Small upturned nose with depressed nasal bridge	CXR, atelectasis and RDS-like picture
	Mid-facial hypoplasia	Mid-facial hypoplasia	Low-set crumpled ears
	—	—	Small upturned nose with depressed nasal bridge
	—	—	Mid-facial hypoplasia

Abbreviations: FT4, free T₄; GA, gestational age; NR, normal range; RDS, respiratory distress syndrome. Dashes indicate no value for that patient for the particular hormone.

Pedigree

The nonconsanguineous Pakistani family (Figure 1A) consisted of three siblings, one female and two males, who all presented at birth. A daughter (IIIa) was born first with a birth weight of 2.4 kg (−2.3 SDS) and a birth length of 46 cm (−2.2 SDS) (DNA not available). She had poor muscle tone and respiratory effort at birth and developed grunting and persistent hyponatremia. Examination revealed mid-facial hypoplasia, a small nose with depressed nasal bridge, antimongoloid slant of eyes, and low-set crumpled ears. She had no clefting of lip or palate. She was started on nasal continuous positive airway pressure followed by ventilation, but she died on the fifth day of life without endocrine status evaluation. The second child, a son (IIIb), was born with a birth weight of 2.44 kg (−2.4 SDS) and a birth length of 51 cm (−0.02 SDS). He was born by emergency cesarean section due to fetal bradycardia and presented with poor muscle tone suggestive of a neuromuscular disorder and respiratory distress marked by grunting, and he was started on ventilation. On the third day of life he showed a lactate dehydrogenase of 2044 IU/L (normal range 220–540), a creatine phosphokinase of 929 IU/L (normal range 6–129), and lung opacification on chest X-ray. Examination revealed low-set crumpled ears, small upturned nose with depressed nasal bridge, mid-facial hypoplasia, hypoplastic nipples and nails, micropenis (1.7 cm), and absent scrotal rugae with bilaterally undescended testes. Pituitary function revealed ACTH and TSH deficiency. The third child, a male (IIIc), was born with a birth weight of 2.6 kg (−2 SDS) and a birth length of 47.5 cm (−1.8 SDS). He had low blood glucose and respiratory distress with grunting, and he was started on ventilation. Examination revealed poor tone and diminished reflexes, low-set crumpled ears, small upturned nose with depressed nasal bridge, mid-facial hypoplasia, micropenis (1.5 cm), absent scrotal rugae, and a palpable right testis with a left undescended testis. Persistent hyponatremia and intestinal perforation complicated the clinical picture, and endocrine testing revealed complete ACTH, TSH, and prolactin and probable GH deficiencies (refer to Table 1 for endocrine results obtained during an episode of hypoglycemia). Both brothers showed AP aplasia and an EPP (Figure 1, B and C); MRI was not performed on their sister (IIIa). All three siblings had atelectasis of the lungs on x-ray with a respiratory distress syndrome-like picture, which worsened in the brothers before death (Figure 1, E and F); no skeletal dysplasia or cardiac defect was present. Despite the rapid commencement of hydrocortisone and T₄ treatment in the brothers, all three children died within the first week of life with presumed fulminant sepsis, with klebsiella sepsis confirmed as the cause of death in the first-born male. The parents are healthy young adults with no endocrine abnormalities.

Gene activation assays

Both wtLHX4 and LHX4(p.T126M) similarly activated the αGSU-Luc reporter (Figure 3A). When activating Prl-Luc, the constructs [wtLHX4, LHX4(p.T126M)] resulted in a dose-dependent increase in luciferase activity (Figure 3B). When each of these constructs was cotransfected with wtPIT1, there was a significant (P < .05) increase in activity compared to transfection with LHX4 on its own. However, there was no significant difference in synergistic activity with wtPIT1 between the wtLHX4 and the mutant construct (Figure 3B). The significant increase in activity between wtLHX4 transfection alone and wtLHX4 cotransfection with wtPIT1 was replicated (P < .001) in a second set of transfection assays using activation of the Prl-Luc reporter (Figure 3C). The known partial loss of function mutation LHX4(p.R84C), located between the LIM domains, was used as a comparator. There was no significant difference in activation of the Prl-Luc reporter when transfecting the LHX4(p.R84C), both alone and in the presence of the wtLHX4, compared to wtLHX4 alone. Cotransfection of LHX4(p.R84C) with wtPIT1 led to an increase in transcriptional activity compared to wtLHX4 alone; however, activation was significantly lower (P < .05) than that achieved by coexpression of wtLHX4 and wtPIT1 (Figure 3C).

Figure 3.

A–C, Transient transfection assays in HEK293T cells using the LHX4(p.T126M) variant. Results are shown as means ± SD of three independent experiments, with each assay performed in triplicate. Error bars depict the SEM across experiments. A, Luciferase activity on the αGSU promoter after transfection with wt (wtLHX4) and LHX4(p.T126M) expression constructs, respectively, cotransfected with the αGSU-Luciferase reporter construct. Luciferase activity was normalized to the amount of total protein. EV, empty vector as a negative control. B, Luciferase activity on the prolactin promoter (Prl-Luc) after transfection with wtLHX4, wtPIT1, and LHX4(p.T126M). The lower dose (31.25 ng/well) is indicated by lowercase letters, and the higher dose (62.5 ng/well), which is the concentration consistent with previous studies, is indicated by uppercase letters. Luciferase activity was normalized to Renilla-luciferase activity. There was a significant increase in luciferase activity between wtLHX4 transfection alone and wtLHX4 cotransfection with wtPIT1 on activating the Prl-Luc in the presence of wtPIT1 (P < .05), indicated by the asterisk. C, Luciferase activity on Prl-Luc after transfection with two doses, as described in panel B, of each expression construct: wtLHX4, wtPIT1, and LHX4(p.R84C) alone, in addition to wtLHX4 and LHX4(p.R84C) (lower dose) cotransfected with wtPIT1 (lower dose). The significant increase in luciferase activity between wtLHX4 transfection alone and wtLHX4 cotransfection with wtPIT1 was replicated in this assay on activating the prolactin promoter but to a higher degree (P < .001), indicated by triple asterisks. There was a significant difference in luciferase activity between wtLHX4 cotransfection with wtPIT1 and LHX4(p.R84C) cotransfection with wtPIT1 (P < .05), indicated by single asterisk.

Discussion

The homozygous LHX4 mutation identified in the three siblings from pedigree I causes a change in a highly conserved region of the LIM2 domain of the protein and has outcomes that parallel the Lhx4 homozygous null mutant mouse model (7). Mutations affecting LIM domains have been known to affect transactivation. Previous functional studies on the partial loss of function mutation, LHX4 (p.R84C), showed reduced activation of the αGSU and TSHβ reporters and inactivity on the PIT1 promoter reporter gene (5). The LHX4(p.V101A) mutant was unable to activate the PIT1 and FSHβ subunit gene promoters (12). The LHX4(p.V75I) located in the LIM1 domain was associated with a partial impairment of the capacity to transactivate PIT1 and αGSU, without any dominant-negative effects (13).

In our study, having shown that the LHX4(p.T126M) variant does not result in a change of function in one context, namely activation of the αGSU-Luc promoter, we analyzed the synergistic activity of LHX4 with PIT1. There was no significant difference in synergistic transactivation of the Prl-Luc reporter between wtLHX4 and LHX4(p.T126M) when cotransfected with wtPIT1. The known partial loss of function mutation p.R84C was used as a known control in this assay and showed the expected significantly lower synergistic activity with wtPIT1 compared to wtLHX4. Because the wtLHX4 and the LHX4(p.T126M) showed similar positive activities, it is suggested that both are well expressed. In addition, Western blot analysis showed that LHX4(p.T126M) produced a protein product of comparable size to wtLHX4 (data not shown).

Because transactivation of both the αGSU-Luc and Prl-Luc promoters was not affected by LHX4(p.T126M), we suggest that DNA-binding and gene activation is perhaps not the mechanism affected by this variant. It is understood that there are other promoters to explore in this context such as TSHβ, PIT1, and FSHβ; however, we cannot be certain that these are genuine physiological targets of LHX4 because they may be expressed at different developmental stages. The complete target gene set of LHX4 is unclear, and many mechanisms taking place in vivo (including partner interactions) might underlie the defect associated with this mutation.

Mutations located in the LHX4 homeodomain, such as p.L190R and p.A210P, cause loss of DNA-binding (5). The p.T126M substitution is not located in the homeodomain and therefore would not be predicted to affect DNA binding; our observations of preserved gene activation are consistent with this hypothesis. A heterozygous frameshift mutation, p.T99NfsX53, was previously identified in LIM2 of LHX4 in two brothers with GH and TSH deficiencies, pituitary hypoplasia, and a poorly developed sella turcica. The youngest also had corpus callosum hypoplasia and an ectopic neurohypophysis. This mutation led to homeodomain truncation and loss of transcriptional activity on the PIT1, Prl, and GH promoters due to abolished DNA binding (14).

LIM domains are involved in protein-protein interaction, as observed in both LHX3 and LHX4. Sloop et al (15) previously analyzed the binding abilities of purified LHX3 protein (mutant and wt) with partners Ldb1 and PIT1. Through in vitro binding assays, they tested whether the LHX3(p.Y111/116C) substitution, located in the LIM2 domain, disrupts structure and binding to putative partner proteins. Interaction of LHX3(p.Y111/116C) with PIT1 was reduced compared to wtLHX3, which may explain the reduction in the ability of LHX3(p.Y111/116C) to activate the Prl promoter in the presence of PIT1 (15).

The presence of another genetic abnormality in the genotype of our patients, contributing to their fatal phenotype, cannot be ruled out. However, impaired complex formation, such as decreased binding of LHX4 to partners such as Ldb1, may be a possible mechanism whereby LHX4(p.T126M), especially in its homozygous state, leads to this phenotype.

A protein prediction model (www.RasMol.Org) showed that the threonine at position 126 in LHX4 likely interacts with arginine at position 103 (Figure 2B). Substitution of this threonine by methionine may disrupt this interaction by affecting stability of the protein. Additionally, the threonine is on the surface within a turn of the protein (Figure 2B), suggesting involvement in protein-protein interaction.

The p.T126M variant was present in both brothers (IIIb and IIIc) in pedigree 1. DNA of the sister (IIIa) was not available; however, her clinical presentation suggests that she probably carried the homozygous variant. Previously, patients carrying heterozygous LHX4 mutations have manifested variable CPHD incorporating GH ± TSH, prolactin, ACTH, LH, and FSH deficiencies. Phenotypes have included dysmorphic features, a small AP, an EPP, a poorly developed sella turcica, Chiari malformation, respiratory distress syndrome, and corpus callosum hypoplasia (5, 14, 16, 17). Both brothers in pedigree 1 presented with panhypopituitarism (Table 1) and showed AP atrophy and an EPP, similar to previously reported carriers of LHX4 mutations.

A maternally inherited heterozygous 1.5-megabase microdeletion in 1q25.2q25.3, including LHX4 (q25.2), was reported in a CPHD patient with minor physical anomalies, suggestive of a midline defect, and heart failure (8). A de novo interstitial deletion of chromosome 1q24.3q31.1, incorporating LHX4, was defined by array-comparative genomic hybridization in a patient with pituitary hormone deficiencies, severe cognitive impairment, bilateral cleft lip/palate and other associated abnormalities. The deletion of LHX4 was considered to be largely causative of the diminished growth and CPHD in this patient (18).

To date, described LHX4 mutations have been heterozygous, often variably penetrant with outcomes likely caused by haploinsufficiency (19) (Figure 2C). This is the first report describing a homozygous variant in a patient. Given the lethality of recessive mutations in rodents (7), the homozygous p.T126M variant is likely to be responsible for this pedigree's lethal phenotype. All three siblings presented with respiratory distress echoing the mouse model, in which null mutants died within the first week of life from immature lungs that failed to inflate. This study suggests that LHX4 is of fundamental importance in maintaining life across species and that recessive mutations in LHX4 are lethal.