Autosomal Dominant Erythrocytosis Caused by Non‐Renal Erythropoietin (EPO) Due to EPO c.‐136 G>A Germline Mutation
Lucie Lanikova, Dusan Hrckulak, Veronika Zimolova, Felipe R. Lorenzo, Jihyun Song, Katerina Vecerkova, Olga Babosova, Linda Berkova, Vladimir Korinek, Steve Elliott, Josef T. Prchal
TL;DR
A new EPO gene mutation causes a type of inherited anemia by making red blood cells in non-kidney tissues.
Contribution
The study identifies a novel germline EPO promoter mutation causing non-renal erythropoietin production in autosomal dominant erythrocytosis.
Findings
The EPO c.-136 G>A mutation increases EPO expression in Hep3B cells under normoxic and hypoxic conditions.
Patient samples show a more basic EPO isoform pattern, indicating reduced renal and increased non-renal EPO expression.
The mutation creates a new hypoxia response element, leading to persistent non-renal EPO production.
Abstract
We previously reported a five‐generation kindred with autosomal dominant erythrocytosis associated with a novel germline promoter variant in the erythropoietin (EPO) gene (EPO c.‐136 G>A). This mutation creates a new hypoxia response element (HRE) consensus sequence on the reverse strand suggesting a gain of function mutation. CRISPR/Cas9‐edited Hep3B cells harboring the c.‐136 G>A variant had increased EPO mRNA and protein expression under both normoxic and hypoxic conditions compared to wild‐type cells; functional assays confirmed the activity of the c.‐136 G>A variant‐induced EPO. Isoelectric focusing analyses of patient urine and plasma showed a more basic EPO isoform pattern, consistent with the reduced sulfated N‐glycan contribution, suggesting decreased renal and increased non‐renal expression. Luciferase reporter assays confirmed increased transcriptional activation of the…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4- —National Institute for Cancer Research
- —ELIXIR‐CZ: Czech National Infrastructure for Biological Data
- —Grantová Agentura, Univerzita Karlova10.13039/100007543
- —Agentura Pro Zdravotnický Výzkum České Republiky10.13039/501100009553
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsErythropoietin and Anemia Treatment · Biosimilars and Bioanalytical Methods · Erythrocyte Function and Pathophysiology
Introduction
1
Hereditary erythrocytosis is classified in the online mendelian inheritance in man (OMIM) database into distinct categories based on molecular etiology including high hemoglobin oxygen affinity from mutations in globin genes or low 2,3‐diphosphoglycerate (DPG) from BPGM (encoding bisphosphoglycerate mutase) gene mutations, mutations affecting oxygen‐sensing HIFs' pathways genes—von Hippel–Lindau (VHL), EGLN1 (encoding PHD2), and EPAS1 (encoding HIF‐2α), or the EPOR (EPO receptor) [1, 2]. However, in a significant proportion of patients with familial or idiopathic erythrocytosis, the underlying etiology remains unidentified [3].
Here, we report a genetic alteration in a family we studied for over a decade, in whom we have identified a 5′ untranslated region (UTR) EPO i.e., EPO ^c.‐136 G>A^ mutation [4]. This EPO ^c.‐136 G>A^ variant has also been subsequently identified as a candidate pathogenic non‐coding mutation in two additional unrelated families with erythrocytosis; however, its underlying causative molecular mechanism remains unknown [5]. We now present experimental data supporting the concept of EPO‐driven erythrocytosis with a non‐renal glycosylation pattern as the cause of hereditary autosomal dominant erythrocytosis.
In adult humans, EPO, a glycoprotein hormone, is produced primarily by peritubular interstitial cells in the kidney, while approximately 10%–30% is synthesized by hepatic cells in the liver [6, 7]. Studies using transgenic mice expressing GFP [8] or LacZ [9] reporters under control of the Epo promoter have shown high expression of Epo only in the kidney and liver. However, low levels of EPO production have also been reported in other human tissues such as brain [10], pancreas, or erythroid progenitors [11] (see tissue profiling of EPO from The Genotype Tissue Expression GTEx and The Human Protein Atlas (HPA) projects available in Figure S1) [12, 13]. EPO expression in both the kidney and liver increases in response to hypoxia, which is essential for regulating erythropoiesis [8]. During fetal development, hepatic EPO production predominates, but this expression is markedly downregulated after birth, accompanied by a switch to renal expression [10, 14, 15]. However, tissue expression can be altered by mutations in the EPO promoter. In mice, a G‐to‐T mutation in the GATA box located 30 bp upstream from the Epo transcription start site leads to constitutive Epo expression in other tissues including epithelial cells of the lung, thymus, liver bile duct, and distal collecting cells of the kidney [8].
Identifying the tissue source of EPO expression has long been challenging, but carbohydrate analysis has provided a key tool. Carbohydrate attached to EPO is a critical determinant of its biological activity, serum half‐life, protein stability and biodistribution [16]. Carbohydrate chains are capped by negatively charged sialic acids, variable in number, that result in a banding pattern of charged isoforms detectable by isoelectric focusing (IEF) [17]. Negatively charged sulfate on N‐linked carbohydrates contributes to additional heterogeneity. EPO produced in human kidney is heavily sulfated [18] with up to 12 attached sulfates, whereas liver‐derived EPO [17] or recombinant human EPO (rHuEPO) produced in Chinese hamster ovary cells (CHO) lack sulfate [18, 19]. These differences allow IEF or lectin binding to distinguish renal from non‐renal EPO. Originally developed for detecting rHuEPO doping [17, 20, 21], these methods have since proven useful for tracing endogenous EPO sources. For instance, a recent study [22] demonstrated that EPO produced in individuals carrying mutations in the EPO promoter exhibited a basic glycosylation profile, in contrast to the typical acidic profile, consistent with a shift from renal to non‐renal expression. The molecular mechanism underlying this tissue‐specific switch remains unknown.
Methods
2
Patient Samples and Measurements of Blood Parameters
2.1
A total of 22 family members from a large pedigree with dominantly inherited erythrocytosis were clinically evaluated. Measurements of hematocrit (HCT) and EPO were carried out at the ARUP Laboratories, Salt Lake City during the time period of 2011–2012. Mutations in globins, BPGM, EPOR genes, as well as germline JAK2, EPAS1, EGLN1, and VHL mutations associated with augmented oxygen‐sensing pathway were excluded by Sanger sequencing.
Whole Exome Sequencing (WES) Analysis
2.2
WES was performed using SureSelect (SSELXT Human All Exons+UTR (V5) System (Agilent)) on five affected individuals, and the presence or absence of EPO ^c.‐136 G>A^ variant was subsequently confirmed by Sanger sequencing in 15 family members.
Sarcosyl Polyacrylamide Gel Electrophoresis (SAR‐PAGE)/Western Blotting
2.3
Plasma samples were obtained from six members of the studied family (three affected and three unaffected). Samples were analyzed following a validated protocol published previously [23]. Venous plasma samples were processed via magnetic bead‐based immunopurification and incubated overnight with rabbit anti‐human EPO antibody (LS‐C11323, LS Bio) followed by capture with anti‐rabbit IgG magnetic beads (Dynabeads M‐280). After sequential washing, EPO was eluted at 95°C for 5 min. Eluates were resolved by SAR‐PAGE on 10% bis‐tris gels, transferred to PVDF membranes, and probed with biotinylated anti‐EPO primary antibody (# MAB2871, clone AE7A5, R&D Systems), followed by HRP‐conjugated streptavidin detection. Chemiluminescent imaging was performed using an Amersham Imager 600.
Isoelectric Focusing (IEF)
2.4
The same six plasma eluates prepared for SAR‐PAGE/Western blotting were used for IEF analysis. In addition, 15 mL urine samples from two affected members underwent buffering, size filtration, and immunopurification using anti‐EPO immunoaffinity isolation plates (StemCell Technologies). After overnight incubation, retentates were eluted at 95°C for 5 min. IEF was performed according to the validated protocol outlined in the world anti‐doping agency technical document TD2021EPO.
Cell Lines
2.5
The EPO ^c.‐136 G>A^ mutation was introduced into the human hepatoma cell line Hep3B (provided by RC Skoda, Baylor College of Medicine) using a CRISPR/Cas9‐based strategy. Hep3B cells were maintained in RPMI medium containing 5% fetal bovine serum (Thermo Fisher Scientific) and antibiotics. 200 000 cells were resuspended in buffer R and nucleofected with the Neon Transfection System (Thermo Fisher Scientific) using the following parameters: 1.275 V, 40 ms, 1 pulse. The nucleofection mixture contained 1.6 μL of Cas9‐GFP (10 μg/μL, IDT Alt‐R S.p. Cas9‐GFP V3), 5 μg of sgRNA, and 2 μg of donor template (Table S1). Since the cells did not survive single‐cell sorting following nucleofection, they were instead sorted for green‐fluorescent protein (GFP) as 20 cells per well into 96‐well plates. Allele‐specific PCR (Table S1) screening identified three heterozygous pools, which were cultured at low density on 15 cm^2^ plates to allow single colonies to grow. Individual colonies were then manually picked to isolate single‐cell clones, designated 54, 172, and 181. The veracity of these clones was confirmed by Sanger sequencing. In a subsequent round of targeting, a homozygous clone (designated C2‐41) was generated. To induce hypoxia, the cells were cultured for 48–72 h in a hypoxia chamber (Stem Cell Technologies) flushed with a calibrated mixture of gases containing 1% O_2_, 5% CO_2_ and 94% N_2_.
RNA‐Seq Analysis
2.6
RNA from wild‐type Hep3B cell line and EPO ^c.‐136 G>A^ Hep3B clone 172 was extracted with the RNAeasy mini kit (Qiagen) and cDNA was synthesized using the KAPA RNA HyperPrep kit (Roche) according to the manufacturer's instructions. Sequencing was performed by NextSeq 550 system (Illumina) using NextSeq 500/550 High Output Kit v2. Nf‐core/rnaseq pipeline v1.4.2 [24, 25] was used to process raw reads, including sequencing adaptors and low‐quality reads removal with Trim Galore v0.6.4/cutadapt [26] v2.5 and mapping to GRCh38 reference genome (Ensembl annotation version 101) [27] with HISAT2 [28] v.1.0. Exon splicing was visualized using Integrative Genomics Viewer (IGV) [29] v2.8.13. The data were validated by Oxford Nanopore sequencing of cDNA‐based PCR amplicons. PCR products were generated using proof‐reading KOD One PCR Master Mix (Toyobo, Japan) and primer sets spanning the entire cDNA (primer sequences are listed in Table S1). Sequencing libraries were prepared with the Oxford nanopore ligation sequencing kit V14 (SQK‐LSK114‐XL, ONT) without barcoding, using 50 ng of input DNA. For sequencing, 7.5 ng of library DNA was loaded onto a MinION Mk1B platform with a new R10.4.1 flow cell (FLO‐MIN114, ONT). Reads were basecalled using Dorado (v0.9.6, ONT 2025, RRID:SCR_025883) with the sup model ([email protected]) and only reads with quality scores ≥ 20 were retained. Filtered reads were mapped to the human reference genome GRCh38.p14, and primary alignments with mapping quality ≥ 20 were used for downstream analysis.
Real‐Time PCR Assay
2.7
RNA was isolated using TRI reagent (Merck) and residual DNA was removed by DNA‐free DNase treatment and removal reagents (Ambion, Thermo Fisher Scientific). 1000 ng of DNA‐free RNA was reverse‐transcribed using the first strand cDNA transcriptor synthesis kit (Roche). Gene expression experiments were performed on the LightCycler 480 system (Roche) with TaqMan probe Hs01071097 m1 and reference gene 4333761F RPLP0. The data were validated using SYBR Green Master Mix (Roche) and two primer sets: E2–E4 and E4–E5 (primer sequences are listed in Table S1).
ELISA
2.8
The supernatant from Hep3B cells cultured in normoxia or hypoxia was collected for enzyme linked immunosorbent assay (ELISA) measurement and quantified using LEGEND MAX human erythropoietin (EPO) ELISA Kit according to manufacturer protocol (BioLegend). In order to normalize the results, the total amount of the protein in each sample was measured by micro BCA protein assay kit (Thermo Fisher Scientific) according to manufacturer protocol.
Functional Analysis of EPO Activity
2.9
Viability of Ba/F3‐EPOR cells in the presence of different dilutions of Hep3B cells supernatant generated in hypoxia was quantified by CellTitre‐Blue reagent (Promega) and Perkin–Elmer Envision reader. The supernatant from 1 × 10^7^ Hep3B cells per sample was harvested after 72 h in hypoxia (1% O_2_). The Ba/F3‐EPOR cell (5 × 10^5^ cells per sample) were grown in DMEM medium containing 10% fetal bovine serum (Thermo Fisher Scientific) and antibiotics, supplemented with 1 U/mL EPO (rhEPO, R&D Systems). Before each experiment, cells were washed five times with PBS to remove residual EPO. Human EPO (rhEPO, R&D Systems) was used as an internal control.
Dual Luciferase Transcriptional Assays
2.10
The effect of G>A substitution in EPO ^c.‐136 G>A^ variant on EPO transcriptional activity was measured using firefly luciferase assays in Hep3B, HepG2 and HEK293 cells. Reporter plasmid (EPO/Luc reporter) was constructed with EPO 5′ UTR fragment (hFEPO_out, hREPO_out, Table S1) ligated and cloned into the pGl4.26 reporter backbone (Promega). The schematic diagram of the reporter construct containing the human EPO promoter is shown as Figure S2. The cells were transfected with Lipofectamine 2000 (Thermo Fisher Scientific), and expression of EPO was induced for 72 h in hypoxia (1% O_2_). The luciferase assay was performed according to the supplier's protocol using Dual‐Glo Luciferase Assay System (Promega) when Renilla was used as internal control (the ratios of reporter:Renilla:HIF‐2/HIF‐1 plasmids are indicated for each experiment in figure legends).
Chromatin Immunoprecipitation (ChIP)
2.11
Wild‐type Hep3B cell line and EPO ^c.‐136 G>A^ Hep3B clone 172 (25 × 10^6^ cells) were processed according to the ChIP manual (iDeal ChIP‐qPCR Kit, Diagenode). Chromatin was fragmented using focused ultrasonication (ME220, Covaris). The following antibodies were used: HIF‐1α (NB100‐479 and NB100‐105, Novus Biologicals), HIF‐2α (NB100‐122 and NB100‐132, Nobus Biologicals) and KLF9 (701 888, Invitrogen). DNA from immunoprecipitates was used as a template for qPCR on EPO promoter and control regions using primers listed in Table S1 and LightCycler 480 SYBR Green I Master Mix (04887352001, Roche). Resulting Ct values were used to calculate the recovery, expressed as a % of input DNA sample. Rabbit IgG antibody and human myoglobin exon 2 primers were used as controls (Diagenode).
Results
3
We studied nine affected and 13 non‐affected relatives from a five‐generation kindred with autosomal dominant familial erythrocytosis (Figure 1A). Affected family members had erythrocytosis with elevated HCT: males had a mean HCT of 55.7% (normal range: 42%–51%), and females had a mean HCT of 50.9% (normal range: 36–46). Additionally, these patients demonstrated moderately increased EPO levels for their elevated HCT but remaining within the normal range (mean 6.25 mIU/mL in non‐affected versus 8.83 mIU/mL in affected subjects; normal range 5–25 mIU/mL), p = 0.1741 (Figure 1B). No splenomegaly, leukocytosis, thrombocytosis, arterial hypoxia, or abnormalities in oxygen affinity (p50) were observed, all having normal hemoglobin/O_2_ affinity (p50). Other known inherited causes of erythrocytosis were excluded. In 2013, we reported whole‐exome sequencing of five affected individuals, which identified a novel heterozygous 5′ UTR EPO variant located 136 nucleotides upstream from the ATG start codon of the EPO gene (NM_000799.4:c.‐136 G>A; ENST00000252723.3:c.‐136 G>A) (Figure 1C) [4]. The mutation segregated with the erythrocytosis phenotype in 15 examined family members: the seven affected subjects were heterozygous for this variant (Figure 1A, red arrows), whereas eight non‐affected relatives were negative (Figure 1A, black arrows), suggesting its causative role in their erythrocytosis. We previously hypothesized that the interaction with HIF‐2, the principal transcriptional regulator of EPO, may be enhanced, explaining the molecular basis of erythrocytosis in the affected members of this family; but in those original experiments, we could neither exclude it nor prove it.
Pedigree, clinical characteristics, and genetic analysis of a family with erythrocytosis. (A) Pedigree. Individuals depicted in black represent patients with erythrocytosis, those in white are non‐affected members, and those in red denote family members who did not participate in the study. The red arrow indicates family members from whom DNA was obtained for whole‐exome sequencing, while the dashed red and black arrows indicate individuals in whom the EPO c.‐136 G>A promoter mutation was confirmed or ruled out, respectively, by Sanger sequencing. (B) Panel shows hematocrit (HCT) and erythropoietin (EPO) levels of non‐affected and affected family members; data are shown as all members and according to gender: ■male, ●female. Plots include HCT, EPO, and EPO versus HCT. (C) Sanger sequencing confirmed results from WES—presence of EPO c.‐136 G>A promoter mutation in five affected family members with erythrocytosis. [Color figure can be viewed at wileyonlinelibrary.com]
To investigate the functional impact of the novel EPO variant, we introduced EPO ^c.‐136 G>A^ mutation into EPO‐producing Hep3B cells using CRISPR/Cas9‐mediated homologous recombination with single‐stranded donor oligonucleotides. Firstly, we performed RNA‐seq analysis to determine whether the c.‐136 G>A variant alters EPO mRNA processing similarly to previously reported EPO c.32delG variant causing also autosomal dominant familial erythrocytosis [30]. In that case, a single‐nucleotide deletion in exon 2 of the EPO gene caused a frameshift and alternative EPO mRNA transcripts, leading to the increased production of functional EPO protein with a shortened signal peptide and a novel N‐terminus. In contrast, our analysis revealed no evidence of alternative EPO mRNA transcripts in the CRISPR/Cas9‐edited cells (Figures 2A and S3). RNA‐seq analysis of splicing junctions suggests that c.‐136 G>A variant abolishes production of physiological, yet minor, EV‐3 isoform of EPO that lacks exon 3 and has no clear effect on erythropoiesis [31]. However, validation by deep nanopore sequencing also detected EV‐3 isoform in EPO ^c.‐136 G>A^ Hep3B cell line (Figure S3) excluding its contribution to the observed phenotype.
*Analysis of EPO c.‐136 G>A promoter mutation causing autosomal dominant familial erythrocytosis. (A) RNA‐seq analysis. Exon splicing was visualized using integrative genomics viewer (IGV) v2.8.13. IGV shows alignment to chromosome 7 (100 721 000–100 723 000 bp) when coverage track diagram (gray histogram) shows read depth at different positions (left). Sashimi plot shows read coverage (histogram peaks) and splicing junctions (curved arcs) (right). Validation experiment by Oxford Nanopore sequencing is provided as Figure S3. (B) Quantification of EPO transcripts by quantitative reverse‐transcriptase polymerase chain reaction produced in normoxia (21% O2) and hypoxia (1% O2). Data are normalized to RPLP0 house‐keeping gene. WT denotes the parental Hep3B cell line, ctr corresponds to an unedited Hep3B cell line, clones 54, 172, and 181 are heterozygous EPO c.‐136 G>A Hep3B edited lines and clone C2‐41 is homozygous for EPO c.‐136 G>A mutation. Shown are means ± SD of at least three independent experiments. Two‐tailed t test: *p < 0.05, ***p < 0.005. C. EPO protein was measured by ELISA (BioLegend) and EPO concentrations in normoxia (21% O2) and hypoxia (1% O2) are shown as means ± SD of three independent experiments. Results are normalized to input of proteins. (D) Functional activity of EPO produced by Hep3B cells, measured in culture supernatants by a proliferation assay with the Ba/F3‐EPOR cell line. Serial dilutions of 10 IU/mL of recombinant human EPO (rhEPO, R&D Systems) were used for reference. Viability of Ba/F3‐EPOR cells was quantified by CellTitre‐Blue reagent and Perkin–Elmer Envision analyzer. Data are shown as luminescence (RLU) at different dilutions. Values represent the means ± SD of three independent experiments. (E) EPO transcriptional activity induced by 1% hypoxia was measured using firefly luciferase reporter (EPO/Luc) in Hep3B and HepG2 cells. Luminescence was measured 72 h after transfection (Lipofectamine 2000, Thermo Fisher Scientific) and Renilla counts were used as internal control (ratio 1:4 was used for Renilla vs. EPO/Luc reporter). Shown are means ± SD of at least three independent experiments done in duplicates. Two‐tailed t test: *p < 0.05, ***p < 0.005. (F) EPO transcriptional activity induced by 1% hypoxia was measured using firefly luciferase reporter (EPO/Luc) in HEK293 cells in the presence of HIF‐2α/HIF‐1α. Luminescence was measured 72 h after transfection (Lipofectamine 2000, Thermo Fisher Scientific) and Renilla counts were used as internal control (ratio 1:4:4 was used for Renilla vs. EPO/Luc reporter vs. HIF‐2α/HIF‐1α). Shown are means ± SD of at least three independent experiments done in duplicates. Two‐tailed t test: p < 0.05. [Color figure can be viewed at wileyonlinelibrary.com]
We next measured EPO transcript levels in wild‐type and EPO ^c.‐136 G>A^ Hep3B cell lines in normoxia and hypoxia. Hypoxia significantly increased the relative expression of the EPO ^c.‐136 G>A^ allele in all three heterozygous EPO ^c.‐136 G>A^ Hep3B cell lines (Figure 2B). EPO mRNA levels in a homozygous clone were significantly increased and detected even in normoxia, unlike in wild‐type Hep3B cells (Figures 2B and S4). To assess the functional consequences of this mutation, we quantified EPO secretion from heterozygous EPO ^c.‐136 G>A^ Hep3B clones and wild‐type controls under normoxic and hypoxic conditions. Hypoxia induced a 2.7‐fold increase in EPO production in mutant cells compared to wild‐type Hep3B cells (Figure 2C). Moreover, the supernatant from EPO ^c.‐136 G>A^ mutant clones induced a stronger proliferative response in EPO‐dependent Ba/F3‐EPOR cells than supernatant from wild‐type cells, indicating preserved biological activity and elevated EPO levels (Figure 2D). Since the mutation is located in a CpG island, we performed bisulfite sequencing, which confirmed that the EPO promoter was largely unmethylated in both wild‐type and mutant clones (Figure S5).
To further investigate the transcriptional activity of the EPO ^c.‐136 G>A^ variant, we cloned wild‐type EPO and mutant EPO sequences upstream of a luciferase reporter gene (Figure S2) and transfected these constructs into two EPO‐producing cell lines, Hep3B and HepG2. The mutant promoter significantly increased reporter activity (Figure 2E). To assess its interaction with HIF‐1 and HIF‐2, we co‐transfected the EPO‐luc reporter construct with HIF‐1α and HIF‐2α expression plasmids. The activity of the mutant EPO ^c.‐136 G>A^ promoter was further enhanced in cells with elevated HIF‐2α and to a lesser extent by HIF‐1α levels (Figure 2F). Validation experiments using dose‐dependent protein detection (Figure S6A) and a hypoxia response element luciferase assay (Figure S6B) revealed differences in expression efficiency, explaining increased induction by HIF‐2α compared to HIF‐1α. We hypothesized that the mutation enhances non‐canonical HIF binding, thereby increasing EPO production. Bioinformatic analysis of the promoter sequence supported this hypothesis, revealing that the c.‐136 G>A substitution creates new putative binding sites for several transcription factors, including HIF (Figure 3A). Furthermore, the high evolutionary conservation of this promoter region across species suggests its functional importance (Figure S7).
Predicted transcription factor binding at the human EPO promoter harboring the c.‐136 G>A variant and validation by ChIP analysis. (A) Exons are depicted as black rectangles (I–V), and regulatory elements, including upstream and downstream hypoxia response elements (5′‐HRE, 3′‐HRE), liver and kidney enhancer (LIE, KIE), and negative regulatory elements (NRE, NRLE) are shown as labeled boxes. The promoter hypoxia‐responsive element (pHRE) near the transcription start site is indicated in blue. The exact base pair distances between elements are not drawn to scale. The schematic diagram was adapted and modified based on data previously published [32]. (B) Predicted transcription factor binding sites at the EPO promoter region in the human sequence (JASPAR 2024) [33]. The position of the c.‐136G>A variant is marked in red. Binding sites for KLF family members (KLF9, KLF13, KLF17), SOX18, and the ARNT:HIF‐lα complex are indicated by arrows, the direction of each arrow corresponds to the DNA strand on which the respective transcription factor is predicted to bind. Detailed analysis is available as Figure S7. ChIP assays were performed with wild‐type (WT) or EPO c.‐136 G>A (172) cell lines. Left: binding of HIF‐1α, HIF‐2α, or control IgG to the EPO promoter was analyzed under hypoxia (1% O2). Right: binding of KLF9 at wild‐type and mutant promoter regions under normoxia (21% O2) and hypoxia (1% O₂). hEPO_ChIPctrl: control primers at 3′UTR HRE; hEPO_ChIP1: primers spanning the c.‐136G>A region; hPGK_ChIP1: control genomic locus; hMB: myoglobin promoter negative control. Data are expressed as percentage of input recovery. [Color figure can be viewed at wileyonlinelibrary.com]
However, ChIP experiments failed to confirm direct HIF‐2 or HIF‐1 binding to the mutated EPO promoter (Figure 3B), suggesting a possible alternative regulatory mechanism(s). Based on in silico predictions, the EPO ^c.‐136 G>A^ variant also generates potential binding sites for members of Kruppel‐Like Factor (KLF) family (KLF9, KLF13, and KLF17), as well as SOX18, which may likewise contribute to the observed phenotype (Figure 3B). Due to the unavailability of ChIP‐grade antibodies, we were able to confirm only the enrichment of KLF9 on the mutated promoter under hypoxic conditions; however, further studies are required to establish its impact on EPO expression.
Our results demonstrate that, in the liver‐derived Hep3B cells, the EPO ^c.‐136 G>A^ mutation enhances EPO transcription and secretion and that these effects, unlike those of wild‐type EPO, are further augmented by hypoxia, independently of detectable direct HIF binding. Another study, published this year, identified three different novel non‐coding mutations in the EPO promoter and intron 1 that modulated uncharacterized regulatory elements with high HIF‐2 responsiveness. Using patient‐derived induced pluripotent stem cells differentiated into hepatocyte‐like cells showed, similarly to our Hep3B model, that these mutations significantly increased EPO transcription [22].
Isoelectric focusing analysis revealed that EPO from patients with altered promotor and intron 1 sequences exhibited glycosylation patterns similar to hepatic EPO detected in newborns compared to the more acidic profile of EPO produced in renal cells in adults [22]. To investigate whether the EPO ^c.‐136 G>A^ mutation also had EPO that undergoes non‐renal glycosylation, we evaluated circulating EPO in plasma samples from six members of the studied family. The IEF profile of EPO ^c.‐136 G>A^ shows a shift in the distribution of EPO isoforms toward the basic region, compared to endogenous EPO from non‐affected family members (Figure 4A,B). This pattern aligns with previously characterized differences in sulfation of N‐linked carbohydrates between renal and non‐renal sources of EPO, where reduced contribution from sulfated glycans is known to result in a more basic isoform distribution [18, 19, 35, 36, 37]. A similar shift was seen in urine samples. SAR‐PAGE on all these samples showed no difference in size relative to rHuEPO (Figure 4C). An EPO‐IEF test was introduced approximately 25 years ago to detect illicit administration of rHuEPO by athletes seeking to enhance performance [17, 21]. Thus the basic IEF profile observed in affected subjects could potentially be interpreted as “suspicious” by an antidoping laboratory employing the EPO‐IEF test. However, the antidoping community has largely switched to a SAR‐PAGE assay, in which the increased molecular size of rHuEPO in blood or urine samples, compared with endogenous EPO, is indicative of doping. Therefore, IEF‐positive results in individuals carrying the EPO ^c.‐136 G>A^ variant would not be suspected of doping under current testing protocols. One of the non‐affected relatives, subject IV‐4 (53 years at the time of sampling, now deceased), exhibited a basic IEF profile (Figure 4B, red asterisk) despite carrying wild‐type EPO sequence and having normal blood counts (HCT repeatedly measured at 45.6%). The medical records revealed significant cardiovascular morbidities but we have been unable to get details about whether he had a recent transfusion or possible renal abnormalities that could suppress renal EPO production and explain the observed basic EPO‐IEF [38, 39]. As the subject is now deceased, the appropriate analyses are not possible.
Identification of a basic non‐renal profile in patients with EPO c.‐136 G>A variant. (A) Schematic representation of N‐glycosylation on the EPO protein. EPO contains three N‐linked glycosylation sites at Asn24, Asn38, and Asn83. Each attached carbohydrate can have 1 to 4 branches which affects the amount of sialic acid on the termini and the amount of sulfate attached to N‐acetylglucosamine (GlcNAc). N‐linked carbohydrate can have considerable microheterogeneity [34]. Shown is the tetra antennary form. The X/x summarize the relative abundance of sulfated N‐glycans and their distribution in purified urinary EPO (designated as renal) versus rHuEPO expressed in CHO cells (designated as non‐renal) based on published biochemical characterization [18, 19, 35, 36, 37]. Sulfation is indicated by black circles and occurs on the C6 position of GlcNAc attached to mannose. Pink circles = sialic acid, blue squares = galactose, red squares = GlcNAc, green squares = mannose, violet squares = fucose. (B) IEF of EPO from plasma and urine samples from selected individuals in a pedigree. Each lane corresponds to a different family member, labeled with their respective pedigree codes. A red asterisk () denotes a relative with cardiovascular disease who carries an unaltered (wild‐type) EPO promoter sequence and displays a basic EPO glycosylation profile. Plasma and urine samples are shown separately as indicated. A mixture of rHuEPO and darbepoetin alfa was used as a standard. For enhanced visualization of the signal, a membrane after longer exposure was used; however, the standard is shown with a shorter exposure. Original, uncropped membranes are shown in Figure S8. (C) SAR‐PAGE analysis of EPO from the same individuals (left). The right panel includes sample from healthy subject administered rHuEPO, and negative control sample for comparison. A red asterisk () denotes a relative with cardiovascular disease who carries an unaltered (wild‐type) EPO promoter sequence and displays a basic EPO glycosylation profile. A mixture of rHuEPO and darbepoetin alfa was used as a standard. [Color figure can be viewed at wileyonlinelibrary.com]
Discussion
4
Our findings suggest EPO ^c.‐136 G>A^ is a gain‐of‐function promoter mutation that contributes to hereditary erythrocytosis by introducing a novel HRE consensus motif. In CRISPR/Cas9‐edited Hep3B cells, the mutant promoter increased transcription of EPO mRNA and its translated protein even under normoxia, with further augmentation during hypoxia. These data indicate both constitutive activation of baseline expression and preserved oxygen responsiveness of this EPO mutation (Figures 1 and 2). The luciferase reporter assays demonstrate that the EPO ^c.‐136 G>A^ substitution directly enhances transcriptional activity in a cis‐regulatory manner (Figure 2). Although the variant creates a putative HIF‐binding site, direct ChIP did not detect binding of HIF‐1 or HIF‐2 at the mutant promoter. Due to technical limitations of ChIP in the GC‐rich region of the human EPO promoter, we cannot definitely exclude that the c.‐136 G>A variant creates a HIF‐responsive element. Our results may reflect the complex, context‐dependent nature of HIFs, or other transcription factors, interactions with the mutant EPO promoter (Figure 3).
Previous studies have shown that hypoxia‐induced EPO transcription relies on multiple regulatory elements acting in tissue‐specific contexts. In hepatic cells, HIF‐2 primarily binds a distal 3′ enhancer, while promoter‐proximal HREs exhibit only weak engagement [32]. In contrast, neuronal and embryonic cells utilize tandem HREs within the promoter itself [32, 40]. Additional studies will be necessary to elucidate the precise molecular mechanisms by which the EPO ^c.‐136 G>A^ variant modulates EPO transcription within its native chromatin context. (Figure 3). Further, several other transcription factors may be involved, for example, KLF9 (Figure 3B, Figure S7); however, its role in erythropoiesis is currently unknown.
It has been suggested that the glycosylation profile of proteins remains inherently linked to the tissue‐specific expression of glycosylation enzymes [41]. EPO produced in renal cells carries up to 12 negatively charged sulfate groups, while EPO synthesized in non‐renal tissues contains markedly lower levels of sulfation (Figure 4A) [6, 10, 11, 18]. This has been observed in patients with renal failure, where the shift in EPO production to the liver results in a predominance of more basic, non‐sulfated isoforms [39]. A shift toward more basic isoforms occurs also with vigorous exercise [42], in erythrocytosis due to liver cirrhosis [22] and with administration of erythropoiesis‐stimulating agents including rHuEPO, darbepoetin alfa [17, 43] and HIF stabilizers [44, 45]. A study of cyclists doping with rHuEPO during 1998 Tour de France showed a marked decrease of acidic EPO isoforms [21]. Kidney (sulfated) EPO is the dominant source in circulation under normal conditions but kidney EPO may increase further with hemorrhage, aplastic anemia and also in patients with Chuvash erythrocytosis due to mutations in the VHL gene and other genes in HIFs pathway [2, 18, 46, 47]. The biological effect of sulfation is unknown. Despite structural differences, sulfation appears to have little to no impact on EPO's biological activity. EPO purified from human urine, which contains sulfated isoforms, exhibits equivalent activity in vitro and in vivo compared to recombinant EPO produced in CHO cells, which is largely unsulfated [48].
A decline of acidic EPO isoforms is evident in our patients having an EPO ^c.‐136 G>A^ mutation in the EPO gene promoter (Figure 4B). There is also a shift to basic isoforms with other promoter and intron 1 mutations that result in erythrocytosis [22]. Martin et al. [22] interpreted the altered IEF pattern of patients and the observed erythrocytosis as an effect of reduced sialylation and suggested faster clearance by hepatic asialoglycoprotein receptors [49]. Another possibility was increased EPO specific activity of basic EPO isoforms. However, the IEF shift observed is more consistent with diminished N‐glycan sulfation than with altered sialylation. As noted above, sulfated and unsulfated EPO have similar activity, and the asialoglycoprotein receptor clearance hypothesis has been disproven [50].
We hypothesize that erythrocytosis and the loss of acidic EPO isoforms in EPO ^c.‐136 G>A^ carriers results from abundant EPO expression from non‐renal tissues leading to increased hemoglobin and tissue oxygenation, which in turn suppresses hypoxia‐inducible signaling and EPO transcription in the kidney. This homeostatic feedback explains the observed “silencing” of renal (sulfated) EPO production without requiring a direct cis‐acting inhibitory effect of the mutant allele in the kidney. While our data suggest that EPO ^c.‐136 G>A^ carriers may have persistent hepatic or other non‐renal tissue EPO transcription, the direct identification of the EPO producing tissue is currently lacking. The application of spatial transcriptomics [51, 52] could resolve this question by cell‐specific analysis of EPO transcription in an EPO ^c.‐136 G>A^ animal model or specific human tissues [53]. Martin et al. [22] also did not provide direct evidence that the EPO in the affected patients originated in hepatocytes, even though they suggested a liver shift. It is conceivable that promoter mutations might activate EPO transcription in multiple organ systems (e.g., liver hepatocytes, lung fibroblasts, or even marrow stromal cells) [8], all of which may produce “hepatic‐like” EPO isoforms.
It is possible that promoter mutations may cause a direct shift to liver expression due to an alteration in regulatory mechanisms that turned on liver expression or turned off renal expression as suggested by Martin et al. [22] However, there is no convincing evidence for this, and it is difficult to reconcile how multiple promoter mutations in different locations would all have the same phenotype. It is also difficult to reconcile how this could occur in heterozygous subjects who have one normal EPO allele.
The identification of the EPO ^c.‐136 G>A^ promoter mutation as a cause of autosomal dominant erythrocytosis highlights the critical role of non‐coding regulatory variants in human disease. This expands the known genetic landscape beyond classical coding mutations in EPOR or hypoxia‐sensing pathway genes and underscores the diagnostic value of sequencing EPO regulatory regions in patients with JAK2‐negative erythrocytosis with normal or moderately elevated EPO levels. Our findings illustrate how disruption of tissue‐specific gene regulation can result in aberrant extra‐renal EPO production, impacting erythropoiesis. More broadly, this case exemplifies how regulatory mutations can disrupt gene expression programs and should be considered in other endocrine or developmental disorders.
Author Contributions
L.L. conceived this project, provided financial support, designed the experiments, performed the research, analyzed data and wrote the manuscript; D.H. and V.Z. designed and performed ChIP assay, performed part of the bioinformatic analysis and edited the manuscript; F.R.L. designed and performed pedigree studies, identified this EPO mutation, and was responsible for patients' samples collection; L.B., O.B. designed and performed Hep3B gene targeting, J.S. helped to design CRISPR/Cas9 construct, and edited the manuscript; K.V. analyzed sequencing data; V.K. provided financial support and edited the manuscript, S.E. designed some experiments, analyzed the data and wrote the manuscript; J.T.P. identified and studied affected family, was responsible for coordination of the clinical part of the study and identification of this EPO mutation, designed some experiments, provided financial support, and wrote the manuscript.
Funding
This work was supported by the National Institute for Cancer Research, Programme EXCELES (ID Project No. LX22NPO5102) ‐ Funded by the European Union‐Next generation EU, the ELIXIR‐CZ: Czech National Infrastructure for Biological Data (LM2023055), Grantová Agentura, Univerzita Karlova (GA UK No. 238121), and Agentura Pro Zdravotnický Výzkum České Republiky (NW26‐08‐00340).
Ethics Statement
This study was conducted in accordance with the Declaration of Helsinki. The study protocol was reviewed and approved by the University of Utah institutional ethics committee. Written informed consent was obtained from all participants prior to inclusion in the study. For participants under the age of 18 years, written informed consent was obtained from a parent or legal guardian.
Conflicts of Interest
S.E. is a former employee of Amgen Inc., but currently receives no financial compensation from Amgen. S.E. is an inventor of erythropoiesis‐stimulating agents–related patents, but is not the assignee, so he receives no personal financial benefit.
Supporting information
Data S1: Supporting Information.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1I. Noumani , C. N. Harrison , and M. F. Mc Mullin , “Erythrocytosis: Diagnosis and Investigation,” International Journal of Laboratory Hematology 46, no. S 1 (2024): 55–62.38695361 10.1111/ijlh.14298 · doi ↗ · pubmed ↗
- 2J. T. Prchal , “Primary and Secondary Erythrocytoses/Polycythemias,” in Williams Hematology, 10th ed., ed. K. Kaushansky , J. T. Prchal , L. J. Burns , M. A. Lichtman , M. Levi , and D. C. Linch (Mc Graw‐Hill Education, 2021), 941–961.
- 3C. Bento , “Genetic Basis of Congenital Erythrocytosis,” International Journal of Laboratory Hematology 40, no. S 1 (2018): 62–67.29741264 10.1111/ijlh.12828 · doi ↗ · pubmed ↗
- 4F. Lorenzo , R. Margraf , S. Swierczek , K. Hickman , K. Voelkerding , and J. Prchal , “A Novel EPO Gene Mutation in a Family With Autosomal Dominant Polycythemia,” Blood 122, no. 21 (2013): 950.
- 5J. C. Taylor , H. C. Martin , S. Lise , et al., “Factors Influencing Success of Clinical Genome Sequencing Across a Broad Spectrum of Disorders,” Nature Genetics 47, no. 7 (2015): 717–726.25985138 10.1038/ng.3304 PMC 4601524 · doi ↗ · pubmed ↗
- 6S. Elliott and A. M. Sinclair , “The Effect of Erythropoietin on Normal and Neoplastic Cells,” Biologics 6 (2012): 163–189.22848149 10.2147/BTT.S 32281 PMC 3402043 · doi ↗ · pubmed ↗
- 7B. K. Kragesteen , A. Giladi , E. David , et al., “The Transcriptional and Regulatory Identity of Erythropoietin Producing Cells,” Nature Medicine 29, no. 5 (2023): 1191–1200.10.1038/s 41591-023-02314-737106166 · doi ↗ · pubmed ↗
- 8N. Obara , N. Suzuki , K. Kim , T. Nagasawa , S. Imagawa , and M. Yamamoto , “Repression via the GATA Box is Essential for Tissue‐Specific Erythropoietin Gene Expression,” Blood 111, no. 10 (2008): 5223–5232.18202227 10.1182/blood-2007-10-115857 · doi ↗ · pubmed ↗