A de novo PRPF8 Pathogenic Variant in Transient Severe Hypophosphatemia with Delayed Puberty and Growth Failure

Hypophosphatemia is an uncommon finding in children. It results from increased renal loss of phosphate, impaired or decreased intestinal absorption of phosphate, decreased dietary intake, or altered cellular redistribution [1]. Increased renal phosphate wasting can be due to increased parathyroid hormone concentration, vitamin D deficiency, or abnormal renal phosphate handling due to increased FGF23 or defective tubular phosphate reabsorption. Decreased gastrointestinal absorption is observed upon ingestion of phosphate-binding agents or chronic diarrhea [1]. Severe malnutrition, including anorexia nervosa, can also result in hypophosphatemia [1]. Reference values for plasma phosphate concentrations vary according to age, children and adolescents having higher values than adults [2, 3].

Hypophosphatemia has multiple effects on various organ systems. In growing children, low phosphate concentration affects hypertrophic chondrocytes, causing delayed mineralization of the growth plates [1]. Typical symptoms include impaired bone growth leading to short stature, deformities, and signs of rickets. The other symptoms include bone pain, muscle weakness, dental problems, and fatigue. Profound hypophosphatemia affects the whole body, including the respiratory system, heart and brain function, blood circulation, and muscles [4].

Several genetic causes for hypophosphatemia have been recognized. Phosphate-regulating endopeptidase X-linked (PHEX, OMIM*300550), ectonucleotide pyrophosphatase (ENPP1, OMIM*173335), fibroblast growth factor 23 (FGF23, OMIM*605380), and dentin matrix acidic phosphoprotein 1 (DMP1, OMIM*600980) are some of the known genes related to hypophosphatemia [4‒6]. X-linked hypophosphatemia is the most common of these rare diseases; it is caused by pathogenic variants in the PHEX gene [4]. However, although most genetic forms of hypophosphatemia can be linked to the presently known genes related to abnormal phosphate metabolism, there are still unsolved cases in which patients with apparently genetic hypophosphatemia have no disease-causing variants in the known genes.

We describe an adolescent who presented with consistently low plasma phosphate concentrations, delayed puberty, and growth retardation. Extensive genetic studies revealed a de novo heterozygous pathogenic variant in PRPF8 (called also PRPC8), encoding a pre-mRNA processing factor 8. Parallel to genetic studies, the patient was diagnosed with growth hormone (GH) deficiency due to a possible partially empty sella, which may be a novel manifestation of defective PRPF8 function. Hypophosphatemia was alleviated with GH treatment.

We recruited a Finnish patient with unexplained hypophosphatemia and his unaffected family members to a genetic study, approved by the Ethics Committee of HUS Helsinki University Hospital. All subjects gave a written informed consent before participation. Written informed consent was obtained from the patients for publication of the details of their medical case and any accompanying images. Clinical and biochemical data were obtained from the patient’s medical records.

DNA was extracted from whole blood using the Geneaid DNA Isolation Kit (Geneaid Biotech Ltd., New Taipei City, Taiwan) according to the standard protocol. The gene panel including the known genes for hypophosphatemia and some other metabolic bone disorders (ALPL, CLCN5, CYP27B1, CYP2R1, DMP1, ENPP1, FAH, FGF23, KL, PHEX, SLC34A1, SLC34A3, VDR) was analyzed by Blueprint Genetics, Espoo, Finland.

The family trio underwent whole exome sequencing (WES) at Blueprint Genetics, Espoo, Finland. WES involved preparing and fragmenting samples for sequencing using adapters and a bead-based selection method with the Illumina platform; the GRCh37 (hg19) reference genome was used in the analysis. Further details are presented in the online supplementary material (for all online suppl. material, see https://doi.org/10.1159/000540249). The analysis for copy number variations in known disease-associated genes (>3,750 genes) was performed using two different genetic databases (the Clinical Genomics Database, and the Developmental Disorders Genotype-Phenotype Database).

In addition, we analyzed CNVs from exome sequenced NGS data alignment files (.bam). Four different programs were utilized: Copy Number Inference From Exome Reads (CoNIFER) v0.2.2, eXome-Hidden Markov Model (XHMM) v1.1, ExomeDepth v1.1.15, and COpy number Detection by EXome sequencing (CODEX) v1.18.0. Default settings were used for each program during the analysis. The results were combined with a minimum 1-bp overlap and filtered using a logistic regression model (R 3.6.3) to predict true-positive results [7]. To assess their frequency in common CNV databases, the results were annotated with an in-house upgraded version of cnvScan [8] with 1000 Genomes, DGV, DECIPHER, ExAC, and gnomAD (v 2.1) SVs and an in-house CNV database (n = 268 previously analyzed exome sequenced samples). The annotated results were filtered for de novo CNVs with a maximum minor allele frequency of 1% in any database with 50% reciprocal overlap.

Data annotation for WES analysis regarding single nucleotide variations (SNVs) was performed using the ANNOVAR software [8], and SNVs and short indel variations were analyzed with VarAFT 2.16 (http://varaft.eu). The SNVs, indels, and CNVs from the trio family were filtered by this workflow (online suppl. Fig. 1), and de novo dominant variants according to the following criteria: (1) the variant segregated appropriately, considering the different models of inheritance; (2) the variant was in the coding (exons) or splicing region or a short indel; (3) minor allele frequency in different databases (gnomAD, Sisu, VarSome, dbSNP) was <0.1%. Various predictions were used to classify the variants, like SIFT/PROVEAN [9], UMD-Predictor [10], and Combined Annotation Dependent Depletion (CADD) Score [11], and the variants were classified (pathogenic, likely pathogenic, uncertain significance, likely benign, and benign) according to the ACMG criteria [12].

Various resources were used to determine the function of the gene, including Google Scholar (https://scholar.google.com/), PubMed (https://pubmed.ncbi.nlm.nih.gov/), Musculoskeletal KP (https://msk.hugeamp.org/), GeneCard (https://www.genecards.org/), and GWAS Catalog (https://www.ebi.ac.uk/gwas/). The Human Protein Atlas was used to evaluate the protein expression in various human tissues (https://www.proteinatlas.org/ENSG00000174231-PRPF8/tissue). Cobalt identified conserved regions in the protein (https://www.ncbi.nlm.nih.gov/gene/10594/ortholog/?scope=7742), and STRING, protein-protein interactions (https://string-db.org/). ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) was used to evaluate the clinical consequences. Sanger sequencing was used to confirm the three variants detected from whole exome data in PRPF8, DLX4, and ELFN2. Primer sequences are available upon request.

The index patient is presently a 17-year-old boy born at 35+2 weeks of gestation by C-section with a birth weight of 2,800 g. During the neonatal period, parameters of calcium (P-Ca-ion 1.40 mmol/L; reference range 1.2–1.5 mmol/L on 6th day of life) and phosphate metabolism (P-Pi 2.04 mmol/L; reference range 1.5–2.6 mmol/L on 6th day of life) were normal. Early growth and development were normal until 8 years, and after this there was a slight decrease (−0.5 SD) from 8 to 10 years. He had no bone pain or lower limb deformities. Iron stores (P-ferritin, 44 μg/L; reference range 20–195 μg/L) and alkaline phosphatase (180 U/L; reference range: 125–405 U/L) were normal. He did not use glucocorticoid treatment during childhood.

At 14 years, he was investigated for slow growth (Fig. 1) and delayed puberty (Table 1). He had a mid-parental target height of 180 cm (mother 163 cm, father 179 cm). He had normal height-adjusted weight (0%) during childhood and slightly below normal median (−15%) after pubertal growth spurt. No specific learning difficulties were observed. A head MRI showed a small adenohypophysis of 3 mm in height, but otherwise normal structures, consistent with a partially empty sella (Fig. 2). The patient had no symptoms of idiopathic intracranial hypertension, also known as a pseudotumor cerebri. Laboratory investigations showed low concentrations of pubertal hormones (P-LH 0.1 IU/L, P-FSH 0.6 IU/L, and S-testosterone 0.5 nmol/L) consistent with his prepubertal Tanner stage (Table 2). Importantly, he had significantly subnormal plasma phosphate concentration (fP-Pi 0.57 mmol/L; reference range 1.1–1.8 mmol/L) (Fig. 3) on several control visits, while other parameters of calcium homeostasis (plasma and urinary calcium, parathyroid hormone, 25-OH-vitamin D) were normal. C-terminal FGF23 concentration (cFGF23) was repeatedly within the reference range (85 kRU/L, and 48 kRU/L; reference range 26–110 kRU/L) but regarded high in respect to significant hypophosphatemia (Table 2). U-Pi/U-Crea ratio was 2.4 mmol/mmol, which was also considered high in the presence of significant hypophosphatemia. Kidney function was normal (P-Crea 49 μmol/L; reference range: 50–95 μmol/L). Before starting phosphate supplementation, the ratio of tubular maximum reabsorption of phosphate (TmP/GFR) (0.43 mmol/L, reference range: 1.15–2.44 mmol/L) was low indicating renal phosphate wasting. Further, 1,25(OH)₂-vitamin D level was inappropriately normal (144 pmol/L; reference range 48–190 pmol/L). Due to low phosphate concentration, oral phosphate supplementation (500 mg twice daily) was started, and phosphate concentrations increased moderately from 0.57 mmol/L to 0.7–0.9 mmol/L. Phosphate concentrations were measured in the morning without phosphate supplementation. Based on GH stimulation tests, the patient was diagnosed with GH deficiency (arginine stimulation test values <2 μg/L twice, the first test not primed, and the second test primed with testosterone 65 mg daily during 7 days, S-IGF-1 17 nmol/L) (Table 2).

Hypophosphatemia prompted extensive investigations. The patient had not sustained any fractures and had no dental problems. The diet was normal and included irregular consumption of dairy products and vitamin D supplements. Laboratory tests to exclude celiac disease, inflammatory bowel disease, and renal failure were all normal. Tumor-induced osteomalacia was ruled out based on total body MRI, which did not show any lesions. cFGF23 levels were not markedly increased (48–85 kRU/L; reference range 26–110 kRU/L), which speaks against tumor-induced osteomalacia. In radiographs of the long bones, knee (Fig. 2), and wrist, patchiness in the bone structure was suspected and the knee metaphyses showed mild irregularity but no overt signs of rickets.

GH treatment was started with a standard dose (30–35 μg/kg daily), leading to a good growth response (Fig. 1). During GH treatment, the patient’s phosphate concentration started to improve. The phosphate supplement could be discontinued, and P-Pi remained within normal range (P-Pi 1.12–1.46 mmol/L) without oral phosphate medication. After starting phosphate supplementation and GH treatment, TmP/GFR increased to normal level (1.17 mmol/L), and after stopping phosphate supplementation, it remained normal (0.93 mmol/L). The patient’s puberty progressed slowly from Tanner G2P2 to G3P3 after the start of GH treatment (S-testosterone 1.3 nmol/L); no testosterone treatment was used. The bone maturation remains significantly delayed with a bone age of 13.6 years at a chronological age of 16.6 years.

The persistent hypophosphatemia prompted genetic investigations. None of the 13 genes related to hypophosphatemia and metabolic bone diseases were found to contain potential pathogenic variants (ALPL, CLCN5, CYP27B1, CYP2R1, DMP1, ENPP1, FAH, FGF23, KL, PHEX, SLC34A1, SLC34A3, and VDR). Furthermore, no potential pathogenic copy number variants were found in the known hypophosphatemia genes or in the whole exome CNV analysis. We also excluded in the exome data exonic, flanking intronic, and UTR region variants in PHEX and other known genes for hypophosphatemia.

Exome data analysis identified five variants that segregated with the phenotype: two nonsynonymous compound heterozygous variants in both the DLX4 and the ELFN2 genes, and a de novo heterozygous stop-gain variant in PRPF8 (online suppl. Table 1). The de novo heterozygous PRPF8 variant, present only in the patient, was regarded as the strongest candidate. The stop-gain variant NM_006445.4:c.5548C>T p.(Arg1850*) is located in exon 35 of the (100%) conserved RNase H homology domain (Fig. 4, 5; online suppl. Table 1). The PRPF8 variant p.(Arg1850*) was not present in ExAC or SISu, but frequencies of 0.00008 and 6.841e-7 were reported in dbSNP (GO Exome Sequencing Project, global population) and gnomAD v4.0.0 databases. It was predicted to be damaging by UMD prediction and VarSome germline classification, and of uncertain significance by ClinVar prediction. The PRPF8 variant p.(Arg1850*) CADD score value was very high 48, which predicted that the variant could cause potential damage to the protein and significantly modify its structure in the RNase H homology domain which is essential for DNA stability. The variant was predicted according to ClinVar to create a premature translational stop signal p.(Arg1850*), causing an absent or disrupted protein product. Decipher predicted a high probability of being loss-of-function intolerant (pLI) score (pLI 1); high pLI scores (pLI ≥0.9) indicate extreme loss-of-function intolerance and, thus, a deleterious effect on gene function.

PRPF8 plays an essential role in splicing, and PRPF8 encodes U2- and U12-dependent spliceosomes, which are crucial components of pre-mRNA splicing. According to the Human Protein Atlas (https://www.proteinatlas.org/), PRPF8 RNA is expressed in many tissues, such as the pituitary gland, parathyroid glands, bone marrow, and kidneys (Fig. 5). PRPF8 may also play a role in skeletal development or hypophosphatemia. PRPF38A has been previously linked to bone mass development [16], and the proteins PRPF8 and PRPF38A interact with each other [17]. According to Harmonizome, a comprehensive knowledge portal about genes and proteins [18], hypophosphatemia and hyperphosphaturia-linked protein SLC9A3R1 interacts with PRPF8. However, using the STRING analysis tool, no protein-protein interactions were found between PRPF8 and known proteins associated with hypophosphatemia.

We describe a patient with prolonged severe hypophosphatemia, delayed puberty, growth failure, and a partially empty sella associated with GH deficiency. In search for genetic causes for hypophosphatemia, we identified a rare de novo pathogenic PRPF8 variant p.(Arg1850*) which may have contributed to both defective pituitary development and hypophosphatemia. The patient’s phosphate concentration normalized with GH treatment, suggesting that it was partially due to GH deficiency. Our findings suggest that PRPF8 variants may contribute to neurodevelopment and possibly phosphate homeostasis.

PRPF8 is a scaffolding component of a highly conserved spliceosome complex that mediates the ordered assembly of spliceosomal proteins and snRNAs [19, 20]. In previous studies, pathogenic variants of PRPF8 have been associated with ophthalmological problems [15], such as retinitis pigmentosa, and cancers like hepatocellular carcinoma [21], breast cancer [22], and myeloid malignancies [23]. The mechanism underlying the disruption of gene function is known [24, 25], and our predictions point to similar disruptions in PRPF8 function. Recently, heterozygous variants in PRPF8 were related to neurodevelopmental disorders. Patients with heterozygous defects in PRPF8 were considered to have a new neurodevelopmental syndrome with various symptoms, including hypotonia, autism, feeding problems, development delay, and growth and bone development disorders like scoliosis [15]. However, there was no information regarding the phosphate levels of mutation-positive individuals.

Our patient had slow growth and delayed puberty and was found to have GH deficiency and an empty sella. O’Grady et al. [15] reported normal brain MRI findings in some patients with PRPF8 variants, while some had ventriculomegaly or other findings, but empty sella was reported in no one. Our patient had delayed puberty and GH deficiency, suggesting a dysfunctional pituitary gland, in addition to growth failure. We hypothesize that empty sella together with delayed puberty, growth failure, and GH deficiency may be linked to the PRPF8 p.(Arg1850*) stop-gain variant and could thus reflect part of the spectrum of neurological abnormalities and other symptoms linked to PRPF8 variants by O’Grady et al. [15].

This possible link to neurodevelopmental issues is also supported by previous associations between autism and PRPF8 [26], and central nervous system defects in a zebrafish mutant [27]. Further evidence is needed to confirm the role of PRPF8 variants in pituitary developmental defects, including GH deficiency.

Our patient had prolonged hypophosphatemia combined with low TmP/GFR, inappropriately low 1,25(OH)₂-vitamin D, and inappropriately high FGF23 concentration, suggesting a genetic hypophosphatemia. Based on Hartley et al. [28] (2022), the cFGF23 level was borderline and would not reliably differentiate FGF23 dependent or FGF23 independent hypophosphatemia. Surprisingly, prolonged hypophosphatemia and TmP/GFR normalized with GH treatment. Previous studies have indicated that GH monotherapy improves low phosphate levels and improves longitudinal growth in children with XLH [29, 30]. In our patient, both GH therapy and pubertal spurt contributed to improved growth. In previous studies, GH treatment increased both iFGF23 and cFGF23, and 1,25(OH)₂-vitamin D levels [29, 31]. Based on human and animal studies, the effect of increased phosphate levels is mediated by IGF-1 [29, 30, 32]. IGF-1 stimulates phosphate transport in the renal proximal tubular cells by affecting Na-Pi cotransporters through IGF-1R and increases the maximal tubular phosphate reabsorption rate [33], as observed by the improved TmP/GFR ratio during GH treatment in our patient. It is unclear whether PRPF8 participates in phosphate homeostasis, especially in renal phosphate handling. PRPF8 is expressed in the pituitary gland and the kidneys, which suggests that it may also play a role in mineral handling on the renal level.

To our knowledge, hypophosphatemia has not been previously linked to pathogenic variants in PRPF8. According to Harmonizome results [18], PRPF8 pathogenic variants could affect phosphate metabolism. The Harmonizome database indicates an interaction between PRPF8 and sodium/hydrogen exchanger regulatory factor-1 (SLC9A3R1, or NHERF1). Sneddon et al. [34] reported that SLC9A3R1 and sodium-dependent phosphate cotransporter-2A (NPT2A, or SLC34A1), which regulates phosphate homeostasis primarily in the renal tubule, bind together via the PDZ ligand. However, more information is needed regarding the interactions between PRPF8, SLC9A3R1, and SLC34A1, to confirm the effect on phosphate metabolism. Unfortunately, the interactions in databases are not sufficient evidence for PRPF8 effect on low phosphate.

On the other hand, phosphate homeostasis in our patient has normalized with GH treatment, suggesting that the role of PRPF8 in phosphate handling is not of major significance. More clinical and translational studies are needed to determine the effects of this stop-gain PRPF8 variant on different tissues and organs, such as the pituitary gland, bone, and kidney. Repeated measurements of iFGF23, cFGF23, and 1,25(OH)₂-vitamin D levels after starting GH treatment were unfortunately not conducted in this study.

In summary, we describe a rare PRPF8 variant in an adolescent presenting with prolonged severe hypophosphatemia and poor growth. The patient was found to have a partially empty sella and GH deficiency. Hypophosphatemia normalized with GH therapy, probably via the effects of IGF-I on the renal tubular reabsorption of phosphate. Early identification of the cause of hypophosphatemia is crucial for proper treatment and genetic counseling and contributes to improved treatment outcomes. GH deficiency should be included in the differential diagnostics in patients presenting with hypophosphatemia and poor growth.

We thank the subjects who participated in this study. We also thank research nurse Päivi Turunen, MSc, Iuliia Savenko, and biomedical laboratory scientist Mira Aronen.

This study protocol was reviewed and approved by the Ethics Committee of HUS Helsinki University Hospital (Approval No. HUS/404/2018). All subjects gave a written informed consent before participation. Written informed consent was obtained from the patients for publication of the details of their medical case and any accompanying images.

The authors declare no conflict of interest relevant to this study.

This work was supported by the Sigrid Juselius Foundation, Novo Nordisk Foundation, Emil Aaltonen Foundation, Finnish Cultural Foundation, Finnish Pediatric Research Foundation, Academy of Finland, Orion Foundation, and Folkhälsan Research Foundation.

L.K. performed the formal analysis, was responsible for funding acquisition and methodology, and wrote the original draft of the manuscript. P.S. was responsible for conceptualizing, investigating, reviewing, and editing the manuscript. S.R. performed the formal analysis, was responsible for the methodology, and reviewed and edited the manuscript. M.K.M. was responsible for conceptualizing, investigating, reviewing, and editing the manuscript. M.P. performed the formal analysis, was responsible for methodology and supervision, and reviewed and edited the manuscript. O.M. was responsible for the conceptualization, funding acquisition, investigation, methodology, supervision, and reviewing and editing of the manuscript. All authors read and accepted the final version of the manuscript.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author Outi Mäkitie: outi.makitie@helsinki.fi.