Genetic analysis of 39 erythrocytosis and hereditary hemochromatosis‐associated genes in the Slovenian family with idiopathic erythrocytosis

Abstract

Background

Erythrocytosis is a condition with an excessive number of erythrocytes, accompanied by an elevated haemoglobin and/or haematocrit value. Congenital erythrocytosis has a diverse genetic background with several genes involved in erythropoiesis. In clinical practice, nine genes are usually examined, but in approximately 70% of patients, no causative mutation can be identified. In this study, we screened 39 genes, aiming to identify potential disease‐driving variants in the family with erythrocytosis of unknown cause.

Patients and Methods

Two affected family members with elevated haemoglobin and/or haematocrit and negative for acquired causes and one healthy relative from the same family were selected for molecular‐genetic analysis of 24 erythrocytosis and 15 hereditary haemochromatosis‐associated genes with targeted NGS. The identified variants were further analysed for pathogenicity using various bioinformatic tools and review of the literature.

Results

Of the 12 identified variants, two heterozygous variants, the missense variant c.471G>C (NM_022051.2) (p.(Gln157His)) in the EGLN1 gene and the intron variant c.2572‐13A>G (NM_004972.3) in the JAK2 gene, were classified as low‐frequency variants in European population. None of the two variants were present in a healthy family member. Variant c.2572‐13A>G has potential impact on splicing by one prediction tool.

Conclusion

For the first time, we included 39 genes in the erythrocytosis clinical panel and identified two potential disease‐driving variants in the Slovene family studied. Based on the reported functional in vitro studies combined with our bioinformatics analysis, we suggest further functional analysis of variant in the JAK2 gene and evaluation of a cumulative effect of both variants.

The genetic analysis was conducted on family with idiopathic erythrocytosis (IE), that is, erythrocytosis of unknown cause. Affected family members were included in the study based on the inclusion criteria for congenital erythrocytosis (eg elevated haemoglobin/haematocrit, absence of variants in some other myeloproliferative neoplasm genes and absence of secondary erythrocytosis). For the first, we included 39 erythrocytosis and haemochromatosis‐associated genes in genetic analysis with targeted NGS. Based on the inheritance pattern and selection of variants with minor allele frequency below 0.05, one missense variant in the EGLN1 gene and one intron variant in the JAK2 gene were identified in the affected family members.

1. INTRODUCTION

Erythrocytosis is a heterogeneous group of disorders that occur when there is an increase in the red blood cell (RBC) mass by more than 125% of the predicted value for specific body height and weight, which defines increased haemoglobin levels and/or haematocrit. ¹ , ² Erythrocytosis is classified according to its cause, as primary or secondary, and as congenital or acquired. Primary erythrocytosis occurs when there is an intrinsic defect in the erythroid progenitor cells that is associated with expanded proliferation of RBCs, with erythropoietin (EPO) levels remaining below normal. On the contrary, secondary erythrocytosis occurs when the defect is extrinsic to the erythroid compartment and is instead associated with normal or inappropriately high EPO levels. ³ Most cases of erythrocytosis are acquired, which defines erythrocytosis secondary to various cardiac, pulmonary or renal diseases, or to external hypoxia, or where erythrocytosis is the consequence of somatic gene variants. Polycythaemia vera is the most common primary acquired erythrocytosis, which is characterized by the somatic variant p.Val617Phe or variants in exon 12 of the Janus kinase 2 (JAK2) gene. ⁴ The JAK2 protein is a tyrosine kinase that has important roles in the signal transduction for proliferation and differentiation of myeloid cells. JAK2 binds to the intracellular domains of the cytokine receptors that lack catalytic activity, including the EPO receptor (EPOR). When ligands such as the cytokine EPO bind to their receptors, they induce conformational changes and phosphorylation events that activate JAK2, with the consequent signal transduction via the downstream JAK2/STAT5, MAPK/ERK and PI3K/AKT pathways. In haematopoiesis, JAK2 is a critical mediator of an effective erythropoiesis. ⁵

Congenital erythrocytosis, which is also known as familial erythrocytosis, is a rare clinical condition that is present from birth due to a germline defect in one of the various genes involved in the regulation of oxygen homeostasis. ⁴ , ⁶ For males, the haemoglobin levels are usually >185 g/L (with haematocrit >0.52), and a little lower for females, at >165 g/L (for haematocrit >0.48). ¹ , ² Eight types of congenital erythrocytosis have been classified according to Online Mendelian Inheritance in Man (ECYT1‐8). ⁷ Primary congenital erythrocytosis is defined as type 1 (ECYT1), and this is caused by variants in the EPOR gene. Secondary congenital erythrocytosis is classified into remaining seven types (ECYT2‐8), and these are caused by variants in genes involved in the oxygen‐sensing pathway (VHL, EGLN1, EPAS1, EPO) or variants that affect the haemoglobin oxygen affinity (HBB, HBA1, HBA2, BPGM) (Table 1). ⁶ , ⁷ , ⁸ , ⁹ Oxygen homeostasis is regulated through a complex pathway that is mediated by the essential transcription factors known as hypoxia‐inducible factors (HIFs). HIFs can up‐regulate a number of target genes involved in maintenance of sufficient tissue oxygenation, including the hormone EPO, which regulates proliferation and differentiation of erythroid progenitors. Transcription complex HIF is a dimer, regulated through oxygen‐dependent regulation of alpha subunit. In humans, three isoforms of alpha subunit are known, with HIF2α [officially termed endothelial PAS domain protein 1, EPAS1] as major isoform involved in erythropoiesis. At normal oxygen concentrations (ie, normoxia), the HIF‐α subunit is hydroxylated by HIF‐prolyl hydroxylase 2 (PHD2) [officially termed EGLN1] at prolyl residues in its oxygen‐dependent degradation domain. Post‐translational hydroxylation regulates the oxygen‐dependent stability of EPAS1, as it allows binding of the von Hippel‐Lindau tumor suppressor (VHL) , which results in ubiquitination, followed by degradation by proteasomes. Under hypoxic conditions (eg 1% O₂), oxygen availability is limited, and thus, this hydroxylation is diminished, which prevent interactions with the VHL protein. This results in stabilization of HIF2α and formation of the stable HIF dimer complex that is responsible for the subsequent transcription activation of the targeted genes. ¹⁰

TABLE 1.

Classification of congenital erythrocytosis.

Type	OMIM #	Inheritance a	Gene	Gene location	Protein (synonym)
ECYT1	1333100	AD	EPOR	19p13.2	Erythropoietin receptor
ECYT2	263400	AR	VHL	3p25.3	von Hippel‐Lindau tumor suppressor
ECYT3	609820	AD	EGLN1	1q42.2	Egl−9 family hypoxia‐inducible factor 1 (HIF‐prolyl hydroxylase 2, PHD2)
ECYT4	611783	AD	EPAS1	2p21	Endothelial PAS domain protein 1 (hypoxia‐inducible factor 2 subunit alpha, HIF2α)
ECYT5	617907	AD	EPO	7q22.1	Erythropoietin
ECYT6	617980	AD	HBB	11p15.4	Haemoglobin subunit beta
ECYT7	617981	AD	HBA1, HBA2	16p13.3	Haemoglobin subunit alpha
ECYT8	222800	AR	BPGM	7q33	Bisphosphoglycerate mutase

Abbreviations: AD, autosomal‐dominant; AR, autosomal‐recessive

^aInheritance of the specific CE type was attained from OMIM database. ⁷

The majority of the variants associated with ECYT1‐8 have been collected in the Global Variome shared Leiden Open Variation Database (LOVD), which is supported by the LOVD 3.x software. ¹¹ Along with the nine genes responsible for ECYT1‐8, other key genes are involved in the erythropoiesis pathway and are hence plausible target genes for the development of erythrocytosis. ¹² Because of heterogenic genetic background, the cause for erythrocytosis remains unknown in 70% of patients with indication of congenital erythrocytosis that have been screened for known pathogenic variants in genes causative for ECYT1‐8. Those patients are therefore diagnosed with erythrocytosis of unknown cause, so‐called idiopathic erythrocytosis. No proper diagnosis and prognosis can be made by clinicians, and further genetic studies are needed for patients with idiopathic erythrocytosis. ¹² , ¹³

Enhanced erythropoiesis can be also induced by iron overload, as iron is necessary for the synthesis of haemoglobin. ¹⁴ An excess iron in the blood could be a sign of hereditary haemochromatosis, a well‐known disorder caused by defects in genes that are involved in iron metabolism, transport and balance. The most common hereditary haemochromatosis is type 1, associated with homozygous or compound heterozygous variants in homeostatic iron regulator (HFE) gene. ¹⁵ , ¹⁶ Some authors have observed high frequency of heterozygous HFE variants among patients with idiopathic erythrocytosis, indicating the involvement of haemochromatosis genes in the development of erythrocytosis. ¹⁵

The aim of the present study was to investigate the hereditary gene variants of multiple genes involved in erythropoiesis, in a family with indication of congenital erythrocytosis. An expanded selection of 24 erythrocytosis and 15 hereditary haemochromatosis‐associated genes was included in the targeted molecular‐genetic analysis with next‐generation sequencing (NGS). Here, we report on these NGS screening results where we describe variants with plausible causative effects on increased erythropoiesis.

2. MATERIALS AND METHODS

2.1. Patients

The index patient (Figure 1, individual I:2) was selected from the individuals followed at the University Medical Centre Ljubljana (UMC) over an 8‐year period according to the diagnostic algorithm for erythrocytosis. ¹⁷ The inclusion criteria were as follows: (a) haemoglobin and/or haematocrit above reference values at least twice over 2 months; (b) absence of variants JAK2 p.Val617Phe and JAK2 exon 12; (c) absence of any defined cause of secondary acquired erythrocytosis; and (d) absence of variants in genes for the thrombopoietin receptor (MPL), calreticulin (CALR) and receptor tyrosine kinase (KIT). ¹⁷ The index patient was a 68‐year‐old male who was referred to the Clinical Department of Haematology at UMC Ljubljana in 2013, due to his high RBC count of 6.36 x 10¹² cells/L, high haemoglobin of 221 g/L and high haematocrit of 0.650. His EPO level was in the normal range (11.4 IU/L). The patient also had elevated blood pressure and heart rate. The patient was prescribed for therapeutic phlebotomy, and at the last follow‐up in 2019, he had a slightly increased haemoglobin of 171 g/L. Upon medical history revision, he reported that his son also had clinical signs of erythrocytosis.

FIGURE 1.

Identification of a heterozygous nucleotide variants in the EGLN1 and JAK2 genes (A) Family pedigree and segregation of the variants identified within the family. (B) Next‐generation sequencing (NGS) and (C) Sanger sequencing, showing heterozygous EGLN1 variant in the two affected patients , but not in the unaffected family member. (D) Results of NGS showing intron variant in the JAK2 gene in both of the affected family members, but not in the unaffected family member. Squares, circles, represent males, females, respectively. Filled symbols, patients with clinical diagnosis of erythrocytosis; open symbols, unaffected family members; slashed symbol, subject is deceased. Carriers of heterozygous EGLN1 and JAK2 variants are indicated as ‐/M. Arrow indicates mutation point

The son (Figure 1, individual II:1) of the index patient was 41 years old, and at his first appointment in 2015, he presented with a RBC count of 6.29 × 10¹² cells/L, high haemoglobin of 192 g/L and high haematocrit of 0.552, while his EPO level was normal (8.3 IU/L). The son was a regular blood donor, and so he was not referred for a phlebotomy. These two patients were tested for polycythaemia vera and hereditary haemochromatosis and were negative for variants p.Val617Phe and exon 12 in the JAK2 gene, and also for variants p.Cys282Tyr, p.His63Asp and p.Ser65Asp in the gene for homostatic iron regulator (HFE). Neither of these two patients showed any pulmonary, cardiac or renal abnormalities.

The wife and mother of these elder and younger patients, respectively (Figure 1, individual I:1), is deceased, but had had normal RBC counts, haemoglobin levels and haematocrit, and no signs of erythrocytosis.

All three affected and healthy family members were, together with a reference DNA control NA12878, included in the NGS genetic testing for congenital erythrocytosis. The study was approved by the Slovenian Ethical Committee, No. KME 115/07/15.

2.2. Genetic analysis

Patient's peripheral blood was collected for genetic analysis, together with written informed consent signed by all patients. Granulocytes were isolated from collected peripheral blood, and the genomic DNA was extracted from 1 to 2 × 10⁷ cells using QIAamp DNA mini kits (Qiagen). The patients underwent targeted NGS, with a custom gene panel that covered the target regions of 39 selected genes: 21 genes previously associated with congenital erythrocytosis and polycythaemia vera, ¹² three additional erythrocytosis‐associated genes (PKLR, TET2 and GATA); and 15 genes previously associated with hereditary haemochromatosis. ¹⁸ All of the selected genes were targeted for exon regions, while EPO, VHL and some of the other genes had the first intron, promoter and enhancer regions also included ¹⁹ , ²⁰ (Table 2).

TABLE 2.

List of sequenced genes and regions

Association	Exon	Intron 1	Promotor and enhancer
Erythrocytosis‐associated genes a	BHLHE41, BPGM, EGLN1, EGLN2, EGLN3, EPAS1, EPO, EPOR, GFI1B, HBA1, HBA2, HBB, HIF1A, HIF1AN, HIF3A, JAK2, KDM6A, OS9, PKLR, SH2B3, VHL, ZNF197, GATA1 and TET2	VHL, EPO, EPOR, HBB, HBA1, HBA2	EPO
Hereditary haemochromatosis‐associated genes b	HFE, HJV, HAMP, TFR2, SLC40A1, FTH1, TF, B2 M, CP, FTL, CDAN1, SEC23B, SLC25A38, STEAP3 and ALAS2	‐	‐

BHLHE41 gene for basic helix‐loop‐helix e41, BPGM gene for bisphosphoglycerate mutase, EGLN1 gene for egl 1, EGLN2 gene for egl, EGLN3 gene for egl 3, EPAS1 gene for endothelial PAS domain protein 1, EPO gene for erythropoietin, EPOR gene for erythropoietin receptor, GFI1B gene for growth factor‐independent 1B transcriptional repressor, HBA1 gene for haemoglobin subunit alpha 1, HBA2 gene for haemoglobin subunit alpha 2, HBB gene for haemoglobin subunit beta, HIF1A gene for hypoxia‐inducible factor 1 subunit alpha, HIF1AN gene for hypoxia‐inducible factor 1 subunit alpha inhibitor, HIF3A gene for hypoxia‐inducible factor 3 subunit alpha, JAK2 gene for Janus kinase 2, KDM6A gene for lysine demethylase 6A, OS9 gene for OS9 PKLR gene for pyruvate kinase L/R, SH2B3 gene for SH2B adaptor protein 3, VHL gene for von Hippel‐Lindau tumor suppressor, ZNF197 gene for zinc finger protein 197, GATA1 gene for GATA binding protein 1, TET2 gene for tet methylcytosine dioxygenase 2, HFE gene for homeostatic iron regulator, HJV gene for hemojuvelin BMP co‐receptor, HAMP gene for hepcidin antimicrobial peptide, TFR2 gene for transferrin receptor 2, SLC40A1 gene for solute carrier family 40 member 1, FTH1 gene for ferritin heavy chain 1, TF gene for transferrin, B2M gene for beta‐2‐microglobulin, CP gene for ceruloplasmin, FTL gene for ferritin light chain, CDAN1 gene for codanin 1, SEC23B SEC23 homolog B, SLC25A38 gene for solute carrier family 25 member 38, STEAP3 STEAP3 metalloreductase, ALAS2 gene for 5’‐aminolevulinate synthase 2.

^aAdopted from Camps et al. (2016). ¹²

^bAdopted form Lanktree et al. (2017). ¹⁸

Libraries were prepared using Nextera DNA library preparation kits (Illumina), with enrichment performed by probe hybridization approach using custom gene panel (Integrated DNA Technologies) followed by sequencing (MiniSeq, Illumina). The disease‐risk variants identified were validated by Sanger sequencing (GATC Biotech). Sanger sequencing and prior PCR amplification were performed with custom‐designed primers (Integrated DNA Technologies; available upon request).

2.3. Bioinformatics analysis

The sequencing analysis was performed with built‐in bioinformatics tools (Illumina) and variant annotation with an online tool (Variant Interpreter Illumina). The sequences were aligned to reference genome hg19 (GRCh37). To remove variants with low sequencing quality, the following filters were used within the variant caller: genotype quality (ie GQX value) <30 or not present; quality by depth <2; root mean square mapping quality <20; strand bias > −10; and read depth <1. The variants were first selected based on the relationship between genotypes and phenotypes; this selection was for variants identified as heterozygous in the affected family members and not in the healthy family members, or identified as homozygous in the affected family members and as heterozygous in the healthy family members. For the final selection, the variants with minor allele frequencies (MAFs) <0.05 in the European population were filtered, as MAF <0.05 is distinct for low‐frequency variants. ²¹ We selected data from a European Non‐Finnish population managed by the GnomAD genome and GnomAD exome databases, and data from a European population managed by the 1000 Genomes database. Before filtering, the presence of high‐frequency variants in the JAK2 and HFE genes that cause polycythaemia vera and hereditary haemochromatosis was also assessed. The focus of our analysis was on small variants that involved one or a few nucleotides, such as single nucleotide variants (SNVs) and small insertions/deletions (INDELs).

With the aim to determine the degree of variant position conservation during evolution, conservation analysis was performed with the ConSurf server (https://consurf.tau.ac.il/). ²² The ConSurf server predicts the evolutionary conservation of amino acids or nucleotides based on multiple alignment and phylogenetic relations of homologous sequences that result in position‐specific conservation scores. Continuous conservation scores are divided into scale of nine colour grades for visualization, from the most variable position (grade 1), through intermediately conserved position (grade 5), to the most conserved position (grade 9). To each colour grade, a confidence interval is assigned to each conservation score. If the interval spans four or more colour grades, less than six homologous sequences are aligned and the conservation score is unreliable. ²² The parameters for the conservation analysis of protein sequences were as follows: CSI‐BLAST search algorithm; 3 iterations; E‐value cut‐off 0.0001; protein sequence database UNIREF90; number of analysed homologues 150; minimal 35% identity between homologues; maximal 95% identity between homologues; multiple sequence alignment algorithm ClustalW; and method for calculation of the evolution rate Bayesian paradigm. For conservation analysis of nucleotide sequences, we manually searched for homologous sequences using Basic Local Alignment Search Tool (BLAST) program BLASTN (https://blast.ncbi.nlm.nih.gov/Blast.cgi) ²³ and perform multiple sequence alignment with ClustalW (https://www.ebi.ac.uk/Tools/msa/clustalo/). ²⁴ For calculation of evolution rate in ConSurf server, we selected Bayesian paradigm.

The pathogenicity of variants in coding regions was assessed using in silico prediction tools CADD score (https://cadd.gs.washington.edu/snv), ²⁵ PolyPhen‐2 (http://genetics.bwh.harvard.edu/pph2/), ²⁶ SIFT (https://sift.bii.a‐star.edu.sg/), ²⁷ MutPred2 (http://mutpred.mutdb.org/index.html), ²⁸ SNPs&GO (https://snps.biofold.org/snps‐and‐go/snps‐and‐go.html), ²⁹ PANTHER (http://www.pantherdb.org/tools/csnpScoreForm.jsp), ³⁰ PhD‐SNP (https://snps.biofold.org/phd‐snp/phd‐snp.html), ³¹ PROVEAN (http://provean.jcvi.org/index.php) ³² and Mutation Taster 2 (http://www.mutationtaster.org/) ³³ as described by Schiemann and Stowell, 2016, ³⁴ Bris et al. 2018, ³⁵ Wang et al. 2020. ³⁶ The pathogenicity prediction of intron variants and impact on splicing features were analysed using tools CADD score, ²⁵ the RegSNP‐intron (https://regsnps‐intron.ccbb.iupui.edu/), Human Splicing Finder (http://www.umd.be/HSF/), ³⁷ IntSplice (https://www.med.nagoya‐u.ac.jp/neurogenetics/IntSplice/index.html) ³⁸ as reviewed in Ohno et al. 2018 ³⁹ and Lin et. al. 2019. ⁴⁰

With the CADD (Combined Annotation Dependent Depletion) tool, the prediction of the deleteriousness of variants results in ‘raw’ score and ‘PHRED‐scaled’ score. A PHRED‐scaled score expresses the rank of variant pathogenicity; for example, a score of 10 indicates that the variant is predicted to be in the 10% of the most deleterious substitutions and a score of 20 between the 1% most deleterious. The suggested threshold is between 10 and 20, and we set the cut‐off score at >15. ²⁵ The PolyPhen‐2 (Polymorphism Phenotyping vs.‐2.0) predicts the effect of an amino acid change with classifiers ‘benign’, ‘possibly damaging’ and ‘probably damaging’ and scores from 0.0 to 1.0: variants with values closer to 0.0 are classified as benign and variants with values closer to 1.0 are more confidently predicted as probably damaging. Variants with score over 0.50 were predicted to be pathogenic. In bioinformatics analysis with PolyPhen‐2, we present values obtained with HumVar prediction model, which is preferred model for diagnostics of Mendelian diseases. ²⁶ The prediction tool SIFT (Sorting Intolerant From Tolerant) categorizes impact of amino acid change on protein function based on SIFT score ranging from 0 to 1: <0.05 is classified as damaging and >0.05 as tolerated. ²⁷ The output of MutPred2 (Mutation prediction vs.‐2.0) consists of a general score that ranges between 0.0 and 1.0, with a higher score indicating greater probability to be pathogenic. The cut‐off general score >0.5 was considered to be pathogenic. ²⁸ The prediction tool SNPs&GO uses the reliability index (RI) to evaluate how reliable is the prediction, with 0 being the most unreliable and 10 being the most reliable. Variants are predicted as neutral polymorphism (neutral) or disease related (disease) when the probability score is >0.5. ²⁹ PANTHER (Protein Analysis Through Evolutionary Relationships) estimates the probability of a variant to impact protein function based on evolutionary preservation of a position in protein. The likelihood of deleterious effect increases with the longer preservation time. The thresholds values are >450 million years (my) for ‘probably damaging’ prediction, between 200 my and 450 my for ‘possibly damaging’ and <200 my for ‘probably benign’ prediction. ³⁰ The prediction tool PhD‐SNP (Predictor of Human Deleterious SNP) classifies variants into ‘neutral polymorphism’ or ‘disease related’ with values from 0 to 1 and the decision threshold for disease causing is >0.5. It also uses reliability index (RI) ranges from 0 to 10, from the most unreliable to the most reliable prediction. Prediction was used with 20‐fold cross‐validation. ³¹ PROVEAN (Protein Variant Effect Analyzer) uses PROVEAN score for binary classification of variants into either deleterious or neutral. The threshold is set to > −2.5 for neutral predictions and < −2.5 for deleterious predictions. ³² The prediction tool Mutation Taster 2 employs a Bayes classifier to predict a variant as one of the four possible types: ‘disease causing’, that is probably deleterious, ‘disease‐causing automatic’, that is known to be deleterious, ‘polymorphism’, that is probably harmless and ‘polymorphism automatic’, that is known to be harmless. It also uses a probability values of the predictions ranging between 0 and 1, with values closer to 1 as more reliable prediction. ³³ The RegSNP‐intron is a tool that predicts the pathogenicity of intron variants with probability scores 0.00 −0.36 for benign variants; 0.36–0.45 for possibly damaging variants and 0.45–1.00 for damaging variants. ⁴⁰ The HSF (Human Splicing Finder) allows the identification of all splicing features through multiple algorithms and also the prediction of the impact of intron variants on these features. ³⁷ IntSplice is a tool that predicts a splicing consequence of intron variant close to the 3’ end of an intron, and the result is either an abnormal or normal splicing. ³⁸

The variants were named in agreement with the standard international nomenclature guidelines of the Human Genome Variation Society. ⁴¹

3. RESULTS

3.1. Genetic analysis

Next‐generation sequencing revealed 74, 76 and 79 small variants in individuals II:1, I:2 and I:1, respectively. Twelve variants identified in genes associated with erythrocytosis and hereditary haemochromatosis were consistent with autosomal‐dominant or autosomal‐recessive disease inheritance patterns in the family. Three variants in the EPAS1 and VHL genes were also identified in a reference DNA control NA12878, while the remaining nine variants in the CDAN1, EGLN1 and JAK2 genes were present only in family members (Table 3). After filtering, three variants in the VHL, JAK2 and EGLN1 genes were recognized as low‐frequency variants (MAF < 0.05) with an autosomal‐dominant inheritance pattern. Variant c.341‐325_341‐324delTT (NM_000551.3) in the VHL gene will not be discussed further here, as it is located in a part of the sequence that is rich in thymine (ie the T nucleotide), and therefore, the deletion of two T nucleotides might be due to sequencer error. Next‐generation sequencing confirmed the absence of polycythaemia vera‐causing variant p.Val617Phe and exon 12 variants in the JAK2 gene and hereditary haemochromatosis‐causing variants p.Cys282Tyr, p.His63Asp and p.Ser65Asp variants in the HFE gene in both of these patients and in healthy family member.

TABLE 3.

List of all small variants in the analysed family trio consistent with autosomal‐dominant or autosomal‐recessive inheritance patterns

Genomic location (hg19)	Gene	Position	HGVS coding DNA/ HGVS protein	SNV ID	MAF a	Average DP	Inheritance in family b
chr. 15:43017426	CDAN1	Exon 27	NM_138477.2:c.3474A>C / p.(Leu1158=)	rs16957091	0.23	288	AD
chr. 15:43018486	CDAN1	Intron 24	NM_138477.2:c.3204+22C>T	rs2305085	0.19	101	AD
chr. 15:43020983	CDAN1	Exon 20	NM_138477.2:c.2671C>T / p.(Arg891Cys)	rs8023524	0.19	421	AD
chr. 15:43021563	CDAN1	Intron 17/ splice region	NM_138477.2:c.2408‐3C>T	rs12905385	0.19	348	AD
chr. 15:43021986	CDAN1	Intron 16/ splice region	NM_138477.2:c.2352+8C>T	rs12594483	0.12	231	AD
chr. 1:231557164	EGLN1	Exon 1	NM_022051.2:c.471G>C / p.(Gln157His)	rs61750991	0.03	93	AD
chr. 2:46603671c	EPAS1	Intron 8/ splice region	NM_001430.4:c.1035‐7C>G	rs7557402	0.47	184	AR
chr. 2:46605954 c	EPAS1	Intron 11	NM_001430.4:c.1554+48G>C	rs7598371	0.46	91	AR
chr. 9:5050706	JAK2	Exon 6	NM_004972.3:c.489C>T / p.(His163=)	rs2230722	0.31	297	AD
chr. 9:5089661	JAK2	Intron 19	NM_004972.3:c.2572‐13A>G	rs780797578	0.0001	140	AD
chr. 9:5090934	JAK2	Intron 22	NM_004972.3:c.3059+23A>T	rs2274649	0.29	104	AD
chr. 3:10187858 c	VHL	Intron 1	NM_000551.3:c.341‐325_341‐324delTT	‐	0.0004	131	AD

^aMAF in European population was selected from a European Non‐Finnish population managed by GnomAD genome database.

^bInheritance in family was determined based on the relationship between genotypes and phenotypes: variants identified as heterozygous in the affected family members and not in the healthy ones are defined with AD; variants identified as homozygous in the affected family members and as heterozygous in the healthy family member are defined with AR.

^cVariants also identified in a reference DNA control NA12878.

Variants with MAF <0.05 are indicated in bold. HGVS, Human Genome Variation Society; SNV, single nucleotide variant; MAF, minor allele frequency; DP, read depth; CDAN1, gene for codanin 1; EGLN1 gene for egl‐9 family hypoxia inducible factor 1, EPAS1 gene for endothelial PAS domain protein 1, JAK2, gene for Janus kinase 2; VHL, gene for von Hippel‐Lindau tumour suppressor; AD, autosomal dominant; AR, autosomal recessive.

Both affected family members (father I:2, son II:1) were heterozygous for two SNVs in the JAK2 and EGLN1 genes, while unaffected family member (individual I:1) did not show either SNV (Table 4; Figure 1A). A missense variant in exon 1 of the EGLN1 gene was a heterozygous G > C substitution at nt. c.471, which caused an amino acid change at residue 157 from glutamine to histidine (Figure 1B,C). This is a known variant, designated with SNV ID number rs61750991. According to the ClinVar online database, this variant is classified in association with congenital erythrocytosis type 3 (ECYT3) as benign. ⁴² A non‐coding variant in intron 19 of the JAK2 gene (rs780797578) is located 13 base pairs from the exon 20 (c.2572‐13A>G), in proximity to the splicing region (Figure 1D).

TABLE 4.

Two single nucleotide variants consistent with inheritance pattern and with minor allele frequency below 0.05 in the Slovene family studied

Genomic location (hg19)	Gene	HGVS coding DNA / HGVS protein	SNV ID	Location	MAF (database)	Functional predictions	Patient	Genotype	VRF
chr. 1:231557164	EGLN1	c.471G>C /p.(Gln157His)	rs61750991	Exon 1	0.03 (GnomAD) 0.03 (GnomAD Exome) 0.03 (1000 Genomes)	CADD: low pathogenicity PolyPhen−2: benign SIFT: tolerated MutPred2: low pathogenicity SNPs&GO: neutral PANTHER: probably benign PhD‐SNP: neutral PROVEAN: neutral Mutation Taster 2: polymorphism ClinVar: benign	I:2 II:1 I:1	Heterozygous Heterozygous WT	0.5 0.5 ‐
chr. 9:5089661	JAK2	c.2572‐13A>G	rs780797578	Intron 19	0.0001 (GnomAD) 0.00003 (GnomAD Exome)	CADD: low pathogenicity RegSNP‐intron: benign HSF: potential alteration on splicing IntSplice: normal splicing	I:2 II:1 I:1	Heterozygous Heterozygous WT	0.5 0.5 ‐

Abbreviations: HGVS, Human Genome Variation Society; SNV, single nucleotide variant; MAF, minor allele frequency; VRF, variant read frequency; EGLN1, gene for egl‐9 family hypoxia‐inducible factor 1; JAK2, gene for Janus kinase 2; WT, wild type.

3.2. Bioinformatics analysis

A conservation analysis of amino acids in EGLN1 protein using ConSurf server showed that Gln157 is a low conserved amino acid, indicated as a variable position (grade 3), which rapidly evolved. The confidence interval was (4, 2) (Supplementary Figure S1). An amino acid change in p.Gln157His was predicted to have mild effects on protein function according to all nine prediction algorithms (Table 4; Supplementary Table S1).

The ConSurf conservation analysis of 132 nucleotides at the 3′end of JAK2 intron 19 showed that nucleotide position c.2572‐13A is intermediately conserved (grade 6). The confidence interval was (8, 5) (Supplementary Figure S2). The pathogenicity predictions of JAK2 variant c.2572‐13A>G using CADD score and RegSNP‐intron tools showed that variant has low deleterious effect on protein function. Similarly, the prediction with IntSplice indicated that variant has no potential impact on splicing. However, the analysis with splice prediction tool HSF revealed that variant c.2572‐13A>G could have an effect on splicing, since potential new donor site was identified because of nucleotide change.

4. DISCUSSION

The aim of this study was to identify causative or disease‐associated variants in patients with erythrocytosis of unknown cause, in a family with obvious blood parameters and other signs for congenital erythrocytosis. The total number of total SNVs and small insertion/deletions found in the target regions of 39 genes in two affected and one healthy family member varied from 74 to 79. The called genotypes were consistent with autosomal‐dominant or autosomal‐recessive inheritance patterns of the disease at only 12 variant sites (approximately 15%) (Table 3). As expected, only a few were missense variants located in coding regions, while the majority of variants were in introns (Table 3). Through evolution, introns have been recognized as a much less conserved parts of sequences and their potential variants have mainly had indirect impacts on structural changes to proteins. Our sequencing data show that a large part of the variants identified were common variants, with MAF >0.05 in European populations, while only three variants identified within the family had frequencies <0.05. As congenital erythrocytosis is a rare clinical disorder, we focused only on the low‐frequency variants. Overall, one SNV in the coding region of the EGLN1 gene and one SNV in the intron region of the JAK2 gene were examined in detail.

The EGLN1 missense variant c.471G>C/ p.(Gln157His) was already identified in previous studies. ⁴³ , ⁴⁴ Albiero et al. ⁴³ identified this variant in two family members (father and son) with increased haemoglobin, where the father was also positive for JAK2 p.Val617Phe. ⁴³ Also Ladroue et al. ⁴⁴ reported this variant in a patient with erythrocytosis. ⁴⁴ Potential germline mutations in EGLN1 lead to stabilization of EPAS1 under normoxic conditions, which results in abnormal EPO levels and secondary erythrocytosis. ⁴⁵ To date, a total of 39 different EGLN1 variants have been identified in patients with erythrocytosis, which comprise heterozygous missense, frameshifts and nonsense variants. ¹¹ The patients carrying these variants had predominantly normal to elevated EPO levels, similar to the phenotypes of the affected father and son from the present study, who had EPO levels in the normal range. The variant c.471G>C segregated in the family with an autosomal‐dominant inheritance pattern (Figure 1A–C), which is consistent with an autosomal‐dominant inheritance mode that was recorded for the variants in the EGLN1 gene, that are causative for the ECYT3 (Table 1). The majority of variants reported in previous studies lie between or near the C‐terminal catalytic domain, which is responsible for the HIFα/2‐oxoglutarate or ferrous iron binding. Nonsense and frameshift variants can have substantial effects on the EGLN1 protein in terms of impaired function, as these can produce a protein truncated in its C‐terminal region; in contrast, missense substitutions have less clear effects. ⁴⁵ Variant p.(Gln157His) is positioned in the first exon of the EGLN1 gene, approximately 100 amino acids from the MYND zing‐finger domain at the N‐terminus (amino acids 21–58) and almost equally distant from the β substrate recognition loop at the C‐terminus (amino acids 241–251). ⁴⁶ The conservation analysis showed low evolutionary conservation of residue p.Gln157 (Supplementary Figure S1), although not all causative variants are positioned at fully conserved residues (eg p.Asn203Lys, p.Lys204Glu, p.Gly285Arg, p.Lys291Ile). ⁴⁵ According to in silico pathogenicity predictions, substitution of glutamine with histidine at residue 157 will have a mild effect on the protein structure and function (Table 4, Supplementary Table S1), which corresponds to a benign interpretation in the ClinVar database. ⁴² Ladroue et al. ⁴⁴ also performed in vitro hydroxylation assays and showed that this variant has no functional effect on the protein. ⁴⁴

Variant c.2572‐13A>G is located in intron 19 of the JAK2 gene (Table 4; Figure 1D). Somatic JAK2 variants are disease‐causing events in patients with polycythaemia vera, as types of myeloproliferative neoplasms. On the other hand, JAK2 germline variants rarely correlate with haematological disorders. They have been described, however, in patients with polycythaemia vera and other myeloproliferative neoplasms, such as essential thrombocythemia and primary myelofibrosis. ⁴⁷ , ⁴⁸ , ⁴⁹ Kapralova et al. ⁵⁰ reported on two heterozygous missense germline variants p.Glu846Asp and p.Arg1063His that they identified in a patient with erythrocytosis and abnormal megakaryopoiesis of the bone marrow. ⁵⁰ They showed that both of these variants lead to erythrocytosis with megakaryocyte abnormalities through hyper‐activation of the JAK2/STAT5 signalling pathway via EPOR. ⁵⁰ Similarly, they suggested that combined impact of two variants, p.Glu846Asp in theJAK2 gene and p.Gln157His in the EGLN1 gene, resulted in erythrocytosis phenotype in one patient. ⁵¹ Interestingly, we found the same EGLN1 variant in our studied family, which could imply on the cumulative effect of the two identified variants in JAK2 and EGLN1 gene.

The JAK2 protein has four important domains: the FERM domain (amino acids 37–380) and the SH2 domain (amino acids 401–482) have roles in receptor and ligand binding; the JAK homology 2 (JH2) pseudo‐kinase domain (amino acids 545–809) acts as a negative regulator of the adjacent JAK2 homology 1 (JH1) kinase domain (amino acids 849–1124). ⁵ , ⁴⁶ The variants associated with haematopoietic disorders lie predominantly in the pseudo‐kinase and kinase domains; for instance, the most prevalent somatic variant p.Val617Phe lies in the JH2 domain and disrupts the JH2‐JH1 auto‐inhibitory interaction, which leads to increased JAK2 kinase activity and activation of its downstream effectors, independent of cytokine binding. ⁵ Variant c.2572‐13A>G identified in the present study is located in intron 19, which is positioned between exons 19 and 20, and these code for the JH1 domain. Previous in silico models have shown that variants in the JH1 domain can have effects on prolonged activation of STAT5 in a low‐cytokine environment. ⁵⁰ The bioinformatic analysis with three prediction tools showed that variant has no impact on splicing or protein function, while the algorithm of Human Splicing Finder pointed to a potential alteration of splicing event (Table 4, Supplementary Table S1). The effect of variant c.2572‐13A>G should be further tested in vitro. In accordance with possible impact of variant on protein function were also results from the ConSurf server, as the conservation analysis showed that nucleotide position c.2572–13 was intermediately conserved through evolution (Supplementary Figure S2). Intron variant c.2572‐13A>G is a known variant, as it has already been assigned to the SNV ID number rs780797578, although this variant has not been cited in any publication, and also no clinical significance is reported in the ClinVar database. ⁴²

To study the development of erythrocytosis in the present family, the cumulative effects of both identified variants need to be assessed. In addition, the cumulative effects of common variants (ie MAF >0.05) should not be neglected, as small contributions of common variants can explain large proportions of those at disease risk in the population described by an infinitesimal model. ²¹ It has been reported that inherited common genetic variants can establish the background for a rare disease, or even rare disorders that are typically considered to be monogenic. ⁵² , ⁵³

5. CONCLUSIONS

In the 39 sequenced erythrocytosis and haemochromatosis‐associated genes, we identified two variants in one Slovene family with suspicion of congenital erythrocytosis: c.471G>C/ p.(Gln157His) in the EGLN1 gene and c.2572‐13A>G in the JAK2 gene. Hitherto, neither of these variants can be interpreted as solely causative for the development of erythrocytosis and the cumulative effects of both variants and also between other variants should be assessed. The impact of variant p.(Gln157His) on EGLN1 function was well described by Ladroue et al. ⁴⁴ in which the functional analysis showed no impairment of hydroxylation of EPAS1. ⁴⁴ On the contrary, further functional assays for variant c.2572‐13A>G in the JAK2 gene are necessary to better understand the impact of this intron variant on the regulation of protein function.

In the current study, we targeted exon regions and some regulatory regions of 39 genes involved in erythropoiesis, which are not all part of commercially available gene panels for congenital erythrocytosis. As far as we know, this is the first time that this extended selection of both erythrocytosis and hereditary haemochromatosis‐associated genes was included in the molecular‐genetic analysis of erythrocytosis. We believe, this selection of genes represents a good initial step toward better diagnosis of patients with idiopathic erythrocytosis. For further investigation of the aetiology of congenital erythrocytosis cases, we propose full screening of the non‐coding regions of erythrocytosis‐associated genes for all statistically significant variants that are potentially associated with increased expansion of RBCs, followed by whole‐exome sequencing or whole‐genome sequencing, if necessary.

CONFLICT OF INTEREST

RK and AV are employees of Kemomed Ltd., Kemomed Research and Development. The other authors declare that they have no conflict of interest.

Supporting information

Supplementary Material

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.