Identification of novel candidate pathogenic genes in pituitary stalk interruption syndrome by whole‐exome sequencing

Abstract Pituitary stalk interruption syndrome (PSIS) is a type of congenital malformation of the anterior pituitary, which leads to isolated growth hormone deficiency or multiple hypothalamic‐pituitary deficiencies. Many genetic factors have been explored, but they only account for a minority of the genetic aetiology. To identify novel PSIS pathogenic genes, we conducted whole‐exome sequencing with 59 sporadic PSIS patients, followed by filtering gene panels involved in pituitary development, holoprosencephaly and midline abnormality. A total of 81 heterozygous variants, distributed among 59 genes, were identified in 50 patients, with 31 patients carrying polygenic variants. Fourteen of the 59 pathogenic genes clustered to the Hedgehog pathway. Of them, PTCH1 and PTCH2, inhibitors of Hedgehog signalling, showed the most frequent heterozygous mutations (22%, seven missense and one frameshift mutations were identified in 13 patients). Moreover, five novel heterozygous null variants in genes including PTCH2 (p.S391fs, combined with p.L104P), Hedgehog acyltransferase (p.R280X, de novo), MAPK3 (p.H50fs), EGR4 (p.G22fs, combined with LHX4 p.S263N) and SPG11 (p.Q1624X), which lead to truncated proteins, were identified. In conclusion, genetic mutations in the Hedgehog signalling pathway might underlie the complex polygenic background of PSIS, and the findings of our study could extend the understanding of PSIS pathogenic genes.


| INTRODUC TI ON
Recently, a whole-exome sequencing (WES) study was performed in 24 Chinese patients with isolated PSIS by Guo et al 5 , who identified several heterozygous mutations in genes associated with Notch, Shh and Wnt signalling pathways. Another study performed involving 20 isolated PSIS patients from the Netherlands suggested a non-Mendelian polygenic aetiology of PSIS. 6 Despite the fact that dozens of genes have been associated with PSIS, fewer than 5% of cases can be explained by known pathogenic genes, and genetic aetiology in sporadic patients is still largely undetermined. In the current study, we performed a WES study on 59 isolated patients with PSIS to identify novel germline mutations that might contribute to sporadic PSIS.
The findings of our current study could extend the understanding of PSIS pathogenic genetic aetiology.

| Exome sequencing and bioinformatics analysis
The WES was performed using SureSelect v5 reagents (Agilent Technologies) to capture exons and the HiSeq X Ten platform (Illumina) for subsequent sequencing. Alignment was carried out with respect to the human genome assembly hg19, followed by recalibration and variant calling. Mutation sites of the genes were annotated with ANNOVAR. The gene mutations were filtered in three panels, which were constructed from the OMIM database, includ-  Table S1. Then, candidate pathogenic variants were considered based on nucleotide and amino acid conservation and pathogenicity prediction by bioinformatics tools including PolyPhen-2, SIFT, MutationTaster and CADD. We excluded the variants with population allele frequencies greater than 0.3% in the 1000 Genomes Project. Finally, variants that were recurrent in more than one patient or that were null mutations were of concern and discussed. The STRING database was used to infer the protein-protein interactions of the identified pathogenic genes. The Sanger sequencing of both forward and reverse strands was used to further confirm the candidate pathogenic variants; the primer sequences are provided in Table S2.
Among them, GLI2, PTCH2 and PRKAR2A had the same variant in more than two patients. Of 50 patients with potentially pathogenic variants, 31 had more than one candidate variant, suggesting a polygenic genetic aetiology of PSIS.
In panel 2 associated with holoprosencephaly malformation, candidate pathogenic variants were present in PTCH1/2, LRP2, TCTN1, CAD, HHAT, STIL and VIPR2 (Table 2). Of them, PTCH1/2 and TCTN1 had the same variants in more than one patients and HHAT had a nonsense mutation. Mutations in PTCH1 and PTCH2 were the most frequent, with an overall incidence of 22% (13/59). Four missense variants in PTCH1 and three missense and one frameshift variants in PTCH2 were identified in 13 patients (Table S3). The frameshift variant of PTCH2 is a known pathogenic variant of basal cell naevus syndrome. In panel 3 associated with midline abnormality, recurrent candidate pathogenic variants were present in ROBO2, GPSM2, ATR and PRKAR2A.
Frameshift mutations were found in MAPK3, and nonsense mutations were found in EGR4 and SPG11 (Table 2).

| Novel pathogenic genes associated with PSIS
Well-documented pathogenic variants of PSIS (HESX1, LHX4, PROP1, PROKR2, OTX2, CDON, SOX3, GPR161, POU1F1, GLI1, GLI2, OTUD4, ROBO2 and Shh) based on the ClinVar tool were not found, and only several rare, candidate pathogenic variants were found in GLI1, GLI2, LHX4, CDON, ROBO2 and OTUD4. These variants were suggested to be damaging based on in silico prediction and low allele frequencies, but the interpretation of these variants was classified as unknown significance by ClinVar. Among 59 genes, the variants that led to truncation of the protein or de novo mutations, forming homozygous or compound heterozygous variants, were considered pathogenic and discussed in detail as follows (Table 3). These were

| Case 2. De novo variant of HHAT (P23)
P23 was a 21-year-old man with short stature and CPHD (GH, TSH, ACTH and gonadotropin deficiency; Figure 3A). He had the perinatal complication in which feet appear first. The patient harboured a stop-gain mutation in HHAT (c.C838T, p.R280X), a de novo mutation that was not detected in his parents. p.R280X was interpreted as a pathogenic gene with evidence of PVS1 (null variant), PS2 (de novo) and PM2 (absent from controls). HHAT is a hedgehog acyltransferase, and diseases associated with HHAT include chondrodysplasia-pseudohermaphroditism syndrome and ancylostomiasis. However, P23 did not have clinical phenotypes of these diseases. Besides the HHAT nonsense mutation, P23 also had a maternal missense mutation in NIN (c.C5894G, p.S1965C).

| Case 3. Frameshift variant of MAPK3 (P54)
P54 was an 18-year-old female who experienced hypoxic coma for 2 days due to amniotic fluid aspiration after caesarean delivery
He had the perinatal complication of abnormal foetal position (breech delivery

| Case 5. Nonsense variant of SPG11 (P41)
P41 was a 19-year-old man, with short stature, CPHD (GH, TSH, ACTH and gonadotropin deficiency) and typical MRI characteristics of PSIS ( Figure 3D). He had perinatal complications with breech delivery and a history of hypoxia at birth. A stop-gain mutation in SPG11 (c.C4870T, p.Q1624X) was found, which was inherited from his mother. SPG11 p.Q1624X was interpreted as a likely pathogenic gene with evidence of PVS1 (null variant), PM2 (absent from controls) and PP3 (damaging based on in silico prediction). His mother, who had the same mutation, was asymptomatic.

| Case 6. Compound heterozygous variants of GLI2 and PTCH2 (P58)
P58 was a 28-year-old man. He had perinatal injury (feet appear first and history of hypoxia at birth). We found a missense mutation in GLI2 (c.G1570A, p.A524T) inherited from his mother ( Figure 3E). and were predicted to be possibly damaging by MutationTaster. P15 had two sisters with a normal phenotype, and the possibility of combined mutations in CDON and GLI2 was excluded by genetically test.

| D ISCUSS I ON
In the present study, based on WES of 59 isolated patients, we identified five novel candidate pathogenic genes for PSIS, including PTCH2, HHAT, MAPK3, EGR4 and SPG11 (Table 3)

| HHAT
HHAT is a hedgehog acyltransferase, required for the post-translational palmitoylation of Hedgehog proteins. Abdel-Salam et al 21 reported a biallelic novel missense HHAT variant that might cause syndromic microcephaly and cerebellar-vermis hypoplasia. HHAT mutations can also be indicative of severe acrania-holoprosencephaly-agnathia craniofacial defects. Loss-of-function HHAT in mouse models leads to holoprosencephaly, which mimics the severe condition observed in humans. 22 Previous studies have provided clinical evidence for the essential roles of HHAT in human testicular organogenesis and embryonic development. PSIS was suggested as a mild form of holoprosencephaly, 23 and P23 had a de novo stopgain mutation in HHAT ( Figure 3A). It is highly possible that HHAT p.R280X is a novel pathogenic gene of PSIS.

| MAPK3
Recurrent MAPK3 missense mutations have been found in neurodevelopmental diseases, such as ASD, ID and NDDs. 24 MAPK3 is a key regulator of the syndrome involved in axon targeting and the regulation of cortical cytoarchitecture. 25 Besides pituitary hormone deficiency, P54 ( Figure 3B) actually had certain aspects of mental retardation, presenting with problems in understanding and lacking the ability of comprehensive memory and language expression. Her mother also had the same mutation, although she was asymptomatic for PSIS. The patient had definite hypoxia due to amniotic fluid aspiration. This would act as an environmental exposure, which promotes dominance of the MAPK3 frameshift mutation. We suspected the MAPK3 p.H50fs mutation to be a novel PSIS pathogenic gene with a wide range of midline abnormalities; however, this needs to be confirmed by more studies.

| EGR4
Early growth response protein (EGR4) is a transcriptional regulator that is required for mitogenesis and differentiation. EGR4 has been reported to participate in fertility development during the regulation of LH secretion or posterior hindbrain development. 26 Consistent with the EGR4 function in fertility, P48 ( Figure 3C) showed poor responses to HCG (human chorionic gonadotropin) substitution therapy. Substitution therapy was initiated with levothyroxine and hydrocortisone, and delayed puberty was treated with HCG. After more than 1 year of treatment with HCG, the patient still had lower LH and FSH levels. Although the testicles became larger, the patient still had azoospermia, as suggested by a sperm test.

| SPG11
SPG11 is a transmembrane protein that is phosphorylated upon DNA damage. Mutations in SPG11 comprise a major cause of spastic paraplegia with a thin corpus callosum. 27 It is expressed ubiquitously in the nervous system but most prominently in the cerebellum, cerebral cortex, hippocampus and pineal gland. Loss-of-function SPG11 was identified in hereditary spastic paraplegia patients. 28 P41 ( Figure 3D) had a SPG11 nonsense mutation in c.C4870T, which was absent in the control population. For P41, the possibility of spastic paraplegia was excluded, and we suspected that the SPG11 p.Q1624X mutation is a novel PSIS pathogenic gene involved in nervous system development.

ACK N OWLED G EM ENTS
The present study was financially supported by the National

CO N FLI C T O F I NTE R E S T
There is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this article.