Characterization of dominant-negative growth hormone receptor variants reveals a potential therapeutic target for short stature

Abstract

Objective

Growth hormone insensitivity (GHI) encompasses growth restriction, normal/elevated growth hormone (GH), and low insulin-like growth factor I (IGF1). “Nonclassical” GHI is poorly characterized and is rarely caused by heterozygous dominant-negative (DN) variants located in the intracellular or transmembrane domains of the GH receptor (GHR). We sought to determine the molecular mechanisms underpinning the growth restriction in 2 GHI cases.

Methods and Design

A custom-made genetic investigative pipeline was exploited to identify the genetic cause of growth restriction in patients with GHI. Nanoluc binary technology (NanoBiT), in vitro splicing assays, western blotting, and flow cytometry, characterized the novel GHR variants.

Results

Novel heterozygous GHR variants were identified in 2 unrelated patients with GHI. In vitro splicing assays indicated both variants activated the same alternative splice acceptor site resulting in aberrant splicing and exclusion of 26 base pairs of GHR exon 9. The GHR variants produced truncated receptors and impaired GH-induced GHR signaling. NanoBiT complementation and flow cytometry showed increased cell surface expression of variant GHR homo/heterodimers compared to wild-type (WT) homodimers and increased recombinant human GH binding to variant GHR homo/heterodimers and GH binding protein (GHBP) cleaved from the variant GHRs. The findings demonstrated increased variant GHR dimers and GHBP with resultant GH sequestration.

Conclusion

We identified and characterized 2 novel, naturally occurring truncated GHR gene variants. Intriguingly, these DN GHR variants act via the same cryptic splice acceptor site, highlighting impairing GH binding to excess GHBP as a potential therapeutic approach.

Introduction

Postnatal growth is a multifactorial process that includes regulation by the growth hormone (GH)- insulin-like growth factor I (IGF1) signaling pathway. Rare genetic defects in genes encoding key proteins of the GH-IGF1 axis have been identified in subjects with severe short stature. These include GHR^{1, 2}STAT5B,³IGF1,⁴IGF2,⁵ IGFALS,⁶ and PAPPA2⁷. The commonest genetic defects are homozygous or compound heterozygous mutations in the extracellular domain (ECD) of the GH receptor (GHR), causing “classical” or severe GH insensitivity (GHI), also known as Laron syndrome (OMIM: 262500).¹

Over 90 GHR gene mutations have been identified, associated with a clinical phenotype of GHI.⁸ The GHR is a class I cytokine receptor comprising key structural domains: a signal peptide, an ECD, a transmembrane domain (TD), and an intracellular domain (ICD).⁹ The primary function of the ECD is to enable GH binding and facilitate receptor dimerization.¹⁰ The ECD is cleaved by the metalloproteinase, tumor necrosis factor-alpha converting enzyme, to produce GH binding protein (GHBP). GHBP binds to circulating GH in a 1:1 ratio, prolonging the half-life of GH.¹¹ The TD facilitates anchoring of the GHR to the cell membrane and the ICD comprises a proline-rich motif which acts as a binding site for Janus Kinase (JAK) 2, which initiates intracellular signaling⁹ and a ubiquitin dependent-endocytosis motif which facilitates GHR internalization.¹²

Nonclassical or mild-moderate GHI is poorly characterized and frequently underdiagnosed.¹³ The evolution of next-generation genetic sequencing technologies has facilitated the genetic characterization of GHI and expanded this rare disorder spectrum. Eight heterozygous dominant-negative (DN) GHR variants (OMIM #604271) located in the TD or ICD of the GHR have been described.^14‐19 The Centre for Endocrinology at the William Harvey Research Institute is an international referral center for patients with undiagnosed growth restriction and suspected GHI.^{13, 20} Our work has expanded the GHI continuum through the identification of novel genetic variants in patients referred with GHI.^21‐23 We identified and functionally characterized 2 novel DN GHR variants located in the ICD of the GHR in 2 unrelated patients with nonclassical GHI, gaining novel insights into GHR physiology.

Methods

Study approval

Informed consent for genetic research was obtained from patients and/or their parents. Ethical approval was granted from the East of England Cambridge East Research Ethics Committee (REC reference: 17/EE/0178). This research was compliant with the Declaration of Helsinki.

Subjects

Proband 1 (P1) presented aged 16.5 years with short stature (height SDS −3.2) and small for gestational age [birth weight (BW) SDS −2.4] (Table 1). He had early postnatal hypoglycemia which was conservatively managed. Clinical assessment showed only subtle dysmorphisms with relative macrocephaly. A skeletal survey identified borderline mesomelic shortening in the upper limb (radius to humerus ratio was 0.7; mean 0.75 +/− 0.2) without Madelung deformity. There was a slight lumbar hyperlordosis. His mother was also short with a similar phenotype (height SDS −2.4). There was no history of consanguinity and no further familial history of growth restriction. Serum GHBP was within the normal reference interval in P1.

Proband 2 (P2) presented at 14.6 years with short stature (height SDS −2.7). He had a normal BW (BW SDS 0.2), no dysmorphic features and no significant past medical history. His parents were not consanguineous, and there was no family history of short stature. His biochemistry was in keeping with GHI with elevated GH peak (57.5 µg/L) and IGF1 deficiency (SDS −3.0; with no increase following an IGF generation test). He had a markedly elevated GHBP of 14567pM (reference 536-3634pM).

GHBP Assays

Serum GHBP levels were first measured by a modification of the ligand immunofunctional assay (LIFA)²⁴ using an in-house monoclonal anti-GHBP antibody.²⁵ The assay is designed to only measure intact GHBP molecules which can bind recombinant GH. Within-assay coefficient of variation (CV) was 9.4% at 115 pmol/L and 6.1% at 1550 pmol/L. At the same concentrations, between-assay CVs were 8.5% and 10.9%, respectively. The lower limit of quantification was 69 pM and the linear range was 69–4000 pM.

GHBP concentrations were also measured by an in-house, time-resolved fluorescence immunoassay (IFMA) based on monoclonal antibodies.²⁶ The assay is designed to measure total GHBP (exon 3-retaining and exon 3-deleted forms). It is standardized against recombinant nonglycosylated GHBP with concentration assigned by amino acid analysis (Protein Laboratories Rehovot Ltd). Within-assay CVs were 3.4% at 312 pM and 3.4% at 2034 pM. At the same concentrations, between-assay CVs were 16.0% and 11.7%, respectively. The lower limit of quantification was 80 pmol/L, and the linear range covered concentrations between 80 and 4880 pmol/L. Reference intervals for both GHBP assays were established from a local cohort of healthy adults.

Genetic and bioinformatic analysis

Isolation of Genomic DNA from peripheral blood leukocytes was performed using a Nucleon BACC2 Genomic DNA Extraction Kit (GE Healthcare) using the manufacturer's instructions. Next generation sequencing was undertaken using a short stature gene panel as previously outlined.²¹ Ingenuity Variant Analysis filtered genetic variants from the variant Call Files based on several parameters (Figure S1) followed by bioinformatic analysis of pathogenicity.²¹

UK biobank data analysis

We performed a look-up of GHR variant rs199960137 (c.876-15T > G) in whole-exome sequencing (WES) data for height within the UK Biobank. Height was derived using data from field 50. Data processing, quality control and variant annotation was performed as previously described.²⁷ In total, data was available for 420 162 individuals of European genetic ancestry. To test for potential sex-specific effects of the variants on height, we conducted a linear model in sex-combined and sex-stratified data between carrier status and height adjusting for the standard covariates [age, exome-sequencing batch and the first 10 principal components (and sex for the sex-combined analysis)]. This analysis was conducted using the “lm” function in the “stats” package in RStudio (v2022.07.1). Sexual dimorphism was ascertained by comparing the association effect sizes between the male- and female-only analyses, as outlined below (where f denotes the female association summary statistics and m denotes the male ones).²⁸

In vitro splicing assay

An in vitro splicing assay was performed as previously described²² using the Exontrap cloning vector pET01 (MoBiTec GmbH) and DNA fragments from P1, P2, and WT (control). Primers are listed in Table S1.

Creation of GHR mutant and nanoluc small and large bit fusion vectors by Gibson assembly

A pcDNA3.1 expression vector containing the entire coding sequence of the WT GHR was utilized to create the GHR variants (MUT1 and MUT2). Primers were designed using Benchling assembly wizard (Benchling Biology Software 2020, https://benchling.com) to mimic the deleted 26 bps of the MUT GHRs as well as the additional T > G nucleotide change seen in the spliced product of P2 (Table S1) creating MUT1 and MUT2 plasmids. Nanoluc small BiT (SmBiT) and large BiT (LgBiT) were cloned on to the N terminus of the WT, MUT1 and MUT2 GHR constructs connected with a flexible Glycine (Gly)-Serine (Ser) linker. Constructs were generated following the Gibson assembly methodology according to the manufacturer's instructions (Gibson Assembly Master Mix, New England Biolabs) as previously described.²² Two fusion proteins with Nanoluc SmBiT and LgBiT were generated for each receptor giving a total of 6 GHR WT and MUT constructs (GHRWT-LgBiT, GHRMUT1-LgBiT, GHRMUT2-LgBiT, GHRWT-SmBiT, GHRMUT1-SmBiT, and GHRMUT2-SmBiT).

Expression of constructs in mammalian HEK293 cell line and GH stimulation

Human embryonic kidney 293 (HEK293) cells were chosen for the experiments due to their ease of growth, maintenance, and transfection. HEK293 cells were grown and seeded into 6-well plates.²² At 80% confluency, the cells were transfected in duplicate with pCDNA3.1, GHR WT, MUT1, and MUT2 constructs and co-transfected in a 1:1 ratio with GHR WT: MUT constructs using lipofectamine 2000 reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. After 24 hours, cells were serum starved and incubated for a further 24 hours. Cell lysates and conditioned media (CM) were harvested at baseline and following treatment with recombinant human GH (rhGH; 500 ng, 0.5 μg/mL) (Life Technology) for 20 minutes. Western blotting was performed as previously outlined.²²

Nanoluc Binary Technology complementation and binding assays

Protein–protein interactions were assessed by Nanoluc Binary Technology (NanoBiT) using the GHR WT, MUT1, and MUT2 LgBiT and SmBiT plasmids. See Supplemental Data for the detailed protocol.

Flow cytometry

HEK293 cells were transiently transfected with GHR WT, MUT, and WT:MUT Nanoluc Sm/Lg BiT constructs. Following serum starvation, the cells were rinsed three times with phosphate buffered-saline (PBS), harvested in 500 µL assay buffer (1 × Dulbecco's phopsphate-buffered saline (DPBS), 1 mM ethylenediaminetetraacetic acid (EDTA), 25 mM Hepes, 1% BSA), and centrifuged at 500 rpm for 5 minutes. Cells were fixed in 4% paraformaldehyde for 10 minutes at RT. The pellet was suspended in 500 µL primary Antibody [Biotechne NanoLuc (Nluc) Luciferase Antibody MAB10026] diluted in buffer 1:500 and incubated under agitation for 30 minutes at 4 °C. Cells were once again washed three times with the assay buffer. The pellet was resuspended in buffer with 1:1000 secondary antibody (Abcam goat anti-mouse IgG H&L Alexa Fluor488 ab150113) and incubated under agitation for 30 minutes at 4°C. The cells were then centrifuged at 500 rpm for 5 minutes, resuspended in buffer and strained through a 40 μM cell strainer. Data was obtained from the BD LSR Fortessa Cell Analyzer and the flow cytometry results were analyzed using FlowJo v10.8 Software (BD Life Sciences).

Statistics

Experiments were repeated using at least 3 biological replicates. Competitive binding dissociation curves were fitted using nonlinear regression one site Fit Log IC50 and One phase decay analysis. Flow cytometry data was analyzed using the FlowJo v10.8 Software (BD Life Sciences). All statistical tests, curve fitting and graph generation were performed using GraphPad Prism 9, Version 9.3.1, San Diego, CA, United States. A P-value of <.05 was considered significant for all analyses.

Results

Discovery of novel heterozygous GHR mutations in intron 8 and exon 9 in the probands

Proband 1

A heterozygous GHR variant, c.876-15T > G, rs199960137 (MUT1) was identified in intron 8 (Figure 1A), confirmed by Sanger sequencing and segregation showed it to be inherited from the mother who was also short (Figure 1B). MUT1 had a Genome Aggregation Database (GnomAD) allele frequency of 0.03% and a Combined Annotation Dependent Depletion (CADD) score <10 (Table 1). It was predicted to negatively impact splicing by the Human Splicing Finder software (GENOMNIS SAS company: http://umd.be/HSF3/) by disrupting the polypyrimidine tract before the exon 9 canonical splice site. MaxEntScan score²⁹ predicted that the disruption of this region reduced the efficacy of splicing at the canonical splice site by 20% with recognition of a cryptic acceptor splice site within exon 9 by the spliceosome. This causes the splicing of 26 base pairs (bp) from exon 9, a frameshift and premature stop codon at position 298 with the formation of a 297 amino acid truncated GHR protein (Figure S2A). Interestingly, this variant was identified in the UK Biobank (UKBB) whole-exome sequencing data on height (beta = −1.44 cm, P = 7.1×10⁻⁶, N carriers = 242). Carriers had a shorter average height than the UKBB average both sex-combined and sex-stratified (167.3 cm, range 147–189 cm vs 168.8 cm range 132-205 cm, P = 2.46 × 10⁻²) (Table 2). There was no evidence for a sex-dimorphic effect when running a linear model on sex-stratified data (P_het = 0.74).

Proband 2

The heterozygous GHR c.902T > G, p.V301G (MUT2) variant in exon 9 arose de novo (Figure 1C). MUT2 was confirmed by Sanger sequencing. Both parents and sibling had wild-type (WT) GHR sequencing. MUT2 was novel and not listed in the Exome Aggregation Consortium (ExAC) or GnomAD databases. It was predicted damaging by sorting intolerant from tolerant (SIFT), probably damaging by Polyphen-2 and had a CADD score of 27.7 (Table 1). MUT2 was predicted to activate the same cryptic acceptor splice site in exon 9, resulting in a mutant exon 9, 26 bp smaller than the WT exon 9 GHR (Figure S2B).

An in vitro splicing assay showed alternative splicing for the GHR variants

Both GHR variants (MUT1; c.876-15T > G and MUT2; c.902T > G, p.V301G) were predicted to activate the same alternative splice acceptor site resulting in abnormal splicing and deletion of 26 bp of GHR exon 9 (Figure 2A and B). Polymerase chain reaction (PCR) gel electrophoresis of the cDNA splicing products showed a smaller band in both probands. Sanger sequencing confirmed a mutant exon 9 GHR, 26 bp smaller than WT GHR (Figure 2C). This resulted in a frameshift and the formation of truncated proteins comprising 297 amino acids. These proteins differ from each other by one amino acid, with P1 having an additional amino acid substitution to serine at position 292 (Figure S2). Interestingly, WT GHR exon 9 was also present in P1 indicating the in vitro production of both WT and truncated mutant GHR exon 9 cDNA. The truncated GHR proteins were predicted to be nonfunctional as the key signaling and internalization domains located in the ICD are absent.

Both wild-type and truncated mutant GHRs can be stably expressed

Western Blot analysis demonstrated the expression of WT and truncated MUT GHR proteins (Figures 3A and B). When WT and MUT constructs were transfected into HEK293 cells in a 1:1 ratio to replicate the heterozygous state of the probands, both the WT GHR (80 kDa) and truncated MUT GHRs (55 kDa) were identified.

The GHR variants impair GHR signaling

A major downstream growth signaling pathway of the GHR is the JAK-STAT pathway.³⁰ Reduced GH-induced STAT5b phosphorylation was detected in cell lysates from HEK293 cells transiently expressing GHR WT and MUT constructs in a 1:1 ratio (Figure 3B). Furthermore, pSTAT5b was not detected in cell lysates from cells transfected with MUT constructs alone, demonstrating that our MUT GHRs are nonfunctional and that they exert a dominant-negative effect on WT GHR signaling.

NanoBiT complementation assays showed increased levels of mutant GHR dimers

NanoBiT provides an innovative experimental methodology to assess protein interactions in live cells.^31‐35 The GHR exists at the cell membrane as a homodimer and has been shown previously to form heterodimers with mutant GHRs.³⁶ We applied NanoBiT to understand how these variants influence GHR dimerization and expression (Figure 4A) and confirmed that the addition of NanoBiT polypeptides did not alter GHR function (Figure S3). The assay was optimized, and the detectable luminescence signal was demonstrated to be specific to GHR interaction with an observed reduction in luminescence values following the addition of increasing quantities of untagged GHR (receptor without NanoBiT) (Figure S4A).

NanoBiT complementation assays in live cells showed significantly increased levels of GHR MUT homo/heterodimers in comparison to WT GHR homodimers, quantified by the increased fold change in luminescence readings of MUT1:MUT1 (P < .01) and WT:MUT1 (P < .05) GHR homo/heterodimers compared to WT:WT homodimers (Figure 4B). Similar results were demonstrated for MUT2; with an increased fold change of luminescence readings for MUT2:MUT2 (P < .0001) and WT:MUT2 (P < .05) GHR homo/heterodimers compared to WT:WT homodimers (Figure 4C). These results demonstrate increased cell surface time and levels of GHR MUT homo/heterodimers.

Flow cytometry showed increased cell surface expression of mutant GHRs

NanoBiT complementation assays detect receptor dimerization in all compartments of the cell thus, flow cytometry utilizing unpermeabilized HEK293 cells was performed to quantify GHR expression at the cell surface. Briefly, cells were gated to remove debris, dead cells and doublets and cells staining positive for NanoLuc (Nluc) Luciferase antibody which detects the Nanoluc LgBiT fragment were selected to determine surface expression of the WT and MUT GHRs (Figure 5). Flow cytometry demonstrated significantly increased cell surface expression of WT:MUT1 (P < .01) and WT:MUT2 (P < .05) GHR heterodimers and MUT1:MUT1 (P < .05) and MUT2:MUT2 (P < .05) GHR homodimers compared to WT:WT GHR homodimers (Figure 6) and corroborated that the NanoBiT results were reflective of GHR surface expression.

Nanobit complementation assays showed increased GH binding to mutant GHR dimers and GHBP cleaved from mutant GHRs

We investigated what effects these increased MUT surface expression had on GH binding by quantifying recombinant human GH-SmBiT (rhGH-SmBiT) binding to live cells transiently expressing WT and MUT GHR-LgBiT (Figure 7A). The assay was optimized, and the detectable luminescence signal was specific to rhGH-SmBiT-GHR-LgBiT interaction with an observed reduction in luminescence values following the addition of increasing quantities of untagged rhGH (ligand without NanoBiT) (Figure S4B). A significant increase in luminescence signal for rhGH-SmBiT binding to GHR WT:MUT1 (P ≤ .05), WT:MUT2 (P ≤ .01), MUT1 (P ≤ .0001) and MUT2 (P ≤ .001) compared to cells expressing GHR WT (Figure 7B) supported ligand sequestration to the mutant GHR homo- and heterodimers. As there appeared to be more mutant receptors at the surface, we investigated whether this equated to elevated cleaved GHBP due to the increased residence time at the plasma membrane. The CM which contained cleaved GHBP with LgBiT (Figure 7A) was analyzed from cells transiently expressing WT, MUT and WT:MUT GHR-LgBiT. There was a significant increase in luminescence signal for rhGH-SmBiT binding to GHR in live cells in the absence of GHBP, WT:MUT1 (P ≤ .01), WT:MUT2 (P ≤ .001), MUT1 (P ≤ .001) and MUT2 (P ≤ .001) (Figure 7C) as well as to the GHBP cleaved from WT:MUT1 (P ≤ .05), WT:MUT2 (P ≤ .01), MUT1 (P ≤ .0001) and MUT2 (P ≤ .0001) (Figure 7D) compared to cells transiently expressing WT GHR. This demonstrated that the presence of MUT GHR leads to increased rhGH ligand binding to GHBP.

Competitive binding of rhGH-SmBiT and untagged rhGH to WT and MUT GHR heterodimers and homodimers

Some MUT GHRs have reduced binding affinity to the GH ligand³⁶ however there is no published data for WT:MUT GHR heterodimers. We utilized NanoBiT to assess the binding affinity of rhGH to the WT:MUT GHR heterodimers (Figure 8A). High luminescence readings were seen for cells transiently expressing GHR-LgBiT MUT homodimers as well as WT:MUT heterodimers in comparison to WT homodimers on the addition of rhGH-SmBiT (Figure 8B) in keeping with the increased mutant GHR cell surface expression. The addition of increasing concentrations of untagged rhGH reduced the luminescence signal through competitive binding to the GHRs and displacement of rhGH-SmBiT (Figures 8C and D). Displacement of WT and MUT GHR homo- and heterodimers followed a typical sigmoidal curve with an absolute calculated IC₅₀ value of 3.42 × 10⁻⁸ M for GHR WT, 9.78 × 10⁻⁸ M for WT:MUT1, 3.30 × 10⁻⁸ M for GHR WT:MUT2, 4.70 × 10⁻⁸ M for GHR MUT1 and 6.22 × 10⁻⁸ M for GHR MUT2 (Table 3). There was a shift in the nonlinear regression curve to the right for the MUT homodimers which correlates to reduced binding affinity of these GHRs in comparison to WT GHR homodimers. The sigmoidal curve generated for WT:MUT heterodimers was comparable to the WT homodimers demonstrating no significant difference in binding affinity.

Discussion

The application of whole genomic sequencing has increased the diagnostic rates for many genetic conditions. Our center utilizes a targeted genetic investigative pipeline which includes a short stature whole genome panel to enhance the genetic diagnostic rate in patients with GHI.²¹ The novel GHR variants identified in intron 8 and exon 9 activate the same cryptic acceptor splice site within exon 9 of the GHR. The splicing process, catalyzed by the spliceosome, facilitates the removal of noncoding intronic DNA and the formation of mature messenger RNA through the identification of splice elements.³⁷ Variants impacting splicing are estimated to account for a third of disease-causing mutations³⁸ and GHR splicing defects are an established cause of GHI.^{39, 40} Alternative splicing of GHR exon 9 has been observed in “normal” subjects by RT-PCR of normal human liver cells.³⁶ An estimated 40%-60% of human genes have been shown to have alternative splice forms which contribute to the complexity of the human genome.⁴¹ Interestingly, both our GHR mutations encode a truncated mutant GHR which differ by just one amino acid. This is the first published report of naturally-occurring GHR mutations leading to activation of this known alternative exon 9 splice site and presenting clinically with growth restriction. It highlights the significance of this cryptic splice site in the pathogenesis of growth restriction and physiological height variation.

The phenotype of our patients was heterogeneous despite the activation of the same alternative splice acceptor site. The intronic P1 GHR variant was inherited from his mother who was also short. This variant was predicted to disrupt the polypyrimidine tract; a region rich in pyrimidine nucleotides before the 3′ end of the intron that promotes assembly of the spliceosome. The in vitro splicing assay detected WT GHR which suggests that the disruption to the polypyrimidine tract reduced the efficacy of canonical splicing but did not eradicate it. This intronic variant is predicted to have varied effects on the splice acceptor site with some of the splice products being WT GHR. This is consistent with P1's mother having a milder phenotype. Correspondingly, the UK Biobank data showed that individuals carrying MUT1, were on average shorter and there was also a notable height variation of carriers within this cohort. Our experimental data revealed that both variants act similarly. This differs from the in vivo/patient data which revealed varying degrees of GH resistance. Interestingly, P1 had IGF1 and GHBP levels in the normal range which would lead to low clinical suspicion of GHI and demonstrates the diagnostic difficulties in this spectrum of disorders. Experimentally, we utilized a 1:1 WT:MUT GHR ratio to replicate the heterozygous state. In vivo, the ratio of MUT to WT GHR generated would be variable and this is a limitation of this study. Splicing is a complex process and transcript heterogeneity has been shown to contribute to phenotypic variability in patients with the same genetic mutation.⁴² Our group has previously demonstrated this concept with the homozygous GHR 6Ψ pseudoexon variant where the mean 6Ψ:WT GHR transcript ratios ranged from 29-71:1 for patients with the same mutation and higher 6Ψ:WT transcript ratios were associated with more severe short stature.⁴² Utilizing a range of WT:MUT GHR ratios and/or patient dermal fibroblasts, would improve the understanding of the biochemical and clinical heterogeneity observed in individuals with these DN GHR variants. Undoubtedly, other genetic and environmental factors may also contribute to the variable phenotype observed.⁴³ Furthermore, height is a polygenic trait so, despite the detailed genetic analysis utilized by our GHI genetic investigative pipeline including whole exome sequencing, there remains the possibility that unidentified genetic variant(s) may also contribute to the diverse presentation.

The GHR variants were confirmed experimentally to be DN mutations which impaired GHR signaling. Mutant GHRs have previously been shown to dimerize with WT GHR and both our mutant GHRs demonstrate increased cell surface expression of MUT:MUT homodimers and WT:MUT heterodimers. There was increased rhGH binding to GHR MUT:MUT homodimers and WT:MUT heterodimers due to the increased availability of these mutant receptor dimers at the cell surface. These results suggest that sequestration of GH to mutant GHRs reduces the amount available to signal through functional WT GHR. Elevated or normal GHBP is a characteristic feature of DN GHR mutations and high levels of GHBP have been shown to correspond to reduced GH-induced signaling which is partially reversed by increasing rhGH treatment levels in vitro, albeit to supraphysiological levels.¹⁶ We showed that our mutant GHRs produce elevated GHBP that binds rhGH and sequesters it, reducing binding to the functional WT GHR. Our results also indicated reduced binding affinity of the mutant GHR homodimer. In contrast, there was no difference in binding affinity for the WT:MUT GHR heterodimer which gave similar results to the WT GHR homodimer. This causes an additive reduction in the availability of GH to bind and signal through the WT GHR.

The nonclassical, milder clinical presentation associated with DN GHR mutations causes diagnostic confusion and delayed access to appropriate treatments such as recombinant human IGF1 (rhIGF1). Both patients had delayed diagnoses attributed to their milder presentations. P1 had normal IGF1 and GHBP levels. This is recognized in nonclassical GHI and compounds the diagnostic challenges. They were both post-pubertal and had completed their linear growth at the time of diagnosis. As such, growth-promoting therapy such as rhIGF1 was not appropriate but may have improved their height prognosis if instituted earlier. P1 was distressed by his height and underwent leg lengthening surgery following genetic diagnosis. Nonclassical GHI is characterized by a range of severity of growth failure with varied clinical and biochemical features including low/normal IGF1 and normal/elevated GHBP levels.⁴⁴ The precise physiological function of GHBP has not been fully elucidated but GHBP levels are influenced by age, body composition (percentage of body fat), and pregnancy.^{45, 46} GHBP has been targeted in the generation of both long-acting GHR antagonists in the treatment of acromegaly,⁴⁷ a disease associated with GH excess and long-acting GH formulations for managing patients with GH deficiency.⁴⁸ The results of these studies provide promising proof of concept data for possible future clinical applications. Targeting GHBP to reduce the sequestration of rhGH is a model that requires further investigation. Although this approach would not alter the presence of nonfunctional mutant GHR homodimers and WT:MUT heterodimers on the cell surface, selective targeting of GHBP would potentially increase the bioavailability of GH to bind to functional WT GHR and alleviate the GH resistance (Figure S5). The exact mechanisms underlying GHBP production/regulation necessitates additional evaluation and we highlight the need to investigate GHBP as a potential therapeutic target for growth restriction.

In summary, our experimental data illustrate the importance of early genetic testing of patients with nonclassical GHI features. We have identified and molecularly characterized 2 novel DN GHR variants in 2 unrelated patients with nonclassical GHI showing that they activate the same cryptic acceptor splice site within exon 9 of the GHR. We demonstrated increased cell surface expression of both MUT receptor homo- and heterodimers compared to the WT GHR homodimers with sequestration of GH to the mutant GHRs and the cleaved GHBP produced by the mutant receptors. There are conceivable limitations of rhIGF1 monotherapy as the important independent growth-promoting effects of GH, are not addressed by this approach.⁴⁹ This highlights the fundamental need to expand the therapeutic options available for patients with DN GHR variants. Future studies targeting GHBP to prevent GH sequestration is a logical novel approach.

Acknowledgments

The UK Biobank (UKBB) research was conducted using the UKBB Resource under application 9905. We are grateful to Dr Ruth Rose for her work in generating the rhGH-SmBiT ligand which was performed at the protein purification facility at the Queen Mary University of London. We are also grateful to Professor Richard Ross for the generous gift of the pcDNA3.1 GHR expression vector and to patients, families, and referring clinicians for their valuable participation in this research.

Supplementary material

Supplementary material is available at European Journal of Endoicrinology online.

Funding

This work was supported by grants and fellowships from NIHR Advanced Fellowship (grant number NIHR300098) awarded to H.L.S., Barts Charity Clinical Research Training Fellowship (grant number MGU0591) awarded to A.A., Sandoz Limited UK research grant (grant number 1010180) awarded to H.L.S., the 2018 European Society for Paediatric Endocrinology (ESPE) Research Fellowship awarded to EC and funding from the Medical Research Council (unit programs: MC_UU_12015/2, MC_UU_00006/2) awarded to J.R.B.P.

Conflict of interest

J.R.B.P. is an employee and shareholder of Adrestia Therapeutics.

Data availability

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

Authors’ contributions

P.J.M., H.L.S., E.C., and A.A. designed the study methodology. A.A. and E.C. performed the experimental work and conducted data acquisition and analysis. A.M. and T.L. also contributed to data collection. E.C., H.L.S., and A.A. collected clinical data and provided patient material. K.S. performed the GHBP assays. J.R.B.P. and L.R.K. performed the UKBB data analysis and interpretation. A.A., P.J.M., and H.L.S. coordinated the project and wrote the report. All authors contributed to data interpretation and critical appraisal of the manuscript.

References

Author notes