Gain-of-function mutations in RPA1 cause a syndrome with short telomeres and somatic genetic rescue

Abstract

Human telomere biology disorders (TBD)/short telomere syndromes (STS) are heterogeneous disorders caused by inherited loss-of-function mutations in telomere-associated genes. Here, we identify 3 germline heterozygous missense variants in the RPA1 gene in 4 unrelated probands presenting with short telomeres and varying clinical features of TBD/STS, including bone marrow failure, myelodysplastic syndrome, T- and B-cell lymphopenia, pulmonary fibrosis, or skin manifestations. All variants cluster to DNA-binding domain A of RPA1 protein. RPA1 is a single-strand DNA-binding protein required for DNA replication and repair and involved in telomere maintenance. We showed that RPA1E240K and RPA1V227A proteins exhibit increased binding to single-strand and telomeric DNA, implying a gain in DNA-binding function, whereas RPA1T270A has binding properties similar to wild-type protein. To study the mutational effect in a cellular system, CRISPR/Cas9 was used to knock-in the RPA1E240K mutation into healthy inducible pluripotent stem cells. This resulted in severe telomere shortening and impaired hematopoietic differentiation. Furthermore, in patients with RPA1E240K, we discovered somatic genetic rescue in hematopoietic cells due to an acquired truncating cis RPA1 mutation or a uniparental isodisomy 17p with loss of mutant allele, coinciding with stabilized blood counts. Using single-cell sequencing, the 2 somatic genetic rescue events were proven to be independently acquired in hematopoietic stem cells. In summary, we describe the first human disease caused by germline RPA1 variants in individuals with TBD/STS.

Graphical abstract

Figure 1.

Clinical characteristics of patients with identified RPA1 variants. (A) Germline RPA1 variants identified in 4 pedigrees using exome sequencing. Black filled, open dotted, and open symbols denote affected individuals with heterozygous (het) RPA1 mutations, unaffected carriers, and unaffected family members without mutations, respectively. (B) Top panel: schematic of the RPA1 gene with 17 coding exons illustrated as black lines and untranslated regions shown in red (NM_002945.5). The 3 unique patient variants are located on exons 8, 9, and 10. Bottom panel: Human RPA1 protein illustration with 4 oligonucleotide/oligosaccharide–binding fold domains F, A, B, and C with amino acid boundaries shown below. Oligonucleotide/oligosaccharide–binding folds A, B, and C are ssDNA-binding domains. Red arrows indicate location of 3 RPA1 missense alterations. (C) BM findings in patient cohort (top panel): P1 BM aspirate smears (Wright-Giemsa staining) at time of clinical presentation at age 10 years showing marked reduction in cell content with hypoplasia of all lineages (left image) and megaloblastic maturation of erythroid precursors (middle image). P2 BM infiltrated with myeloblasts (right image) consistent with MDS with excess blasts. Pulmonary findings by chest computed tomography (CT) imaging in patient cohort (middle panel): P2 chest CT scan (left image) during hospitalization showing necrotizing pneumonitis with diffuse ground-glass opacities and large air-filled cavities with differential diagnosis, including pulmonary GVHD, opportunistic infection(s), pulmonary fibrosis, or a combination of the aforementioned conditions. P3 presented at age 58 years with cough and exertional dyspnea with chest CT scan (middle image) findings showing intralobular reticulations with traction bronchiectasis and mild honeycombing, with left asymmetric and basal, subpleural predominance. At 61 years, P3 chest CT scan (right image) showed significant progression of asymmetric lung fibrosis with left predominance and massive basal honeycombing. Mucocutaneous abnormalities in P1 (bottom panel): oral leukoplakia at 19 years of age (top left image) with mild improvement at 25 years of age (top right image), nail dystrophy (bottom left image), and reticular skin pigmentation on the ventral neck (bottom right image). (D) Telomere length analysis by flow cytometry–based FISH was conducted in lymphocytes of P1 (red circles) and P4 (blue circles). P1 telomere length is less than the first percentile at 24 and 27 years of age (red circles). P4 telomere length is at the fifth percentile at 1.75 years of age and less than the first percentile at age 3.33 years (blue circles). All measurements were performed in triplicate, and mean telomere length was calculated in kilobases in relation to the internal control (bovine thymocytes) with known telomere length. A total of 356 healthy controls used for calculation of the first (solid line, bottom), fifth (dashed line, bottom), 95th (dashed line, top), and 99th (solid line, top) percentile curves. (E) TRF analysis in peripheral blood DNA from P2 and family and P3 compared with healthy age–matched control (Ctrl), digested with HinfI and RsaI enzymes followed by separation on 0.7% agarose gel.

Figure 2.

All RPA mutant heterotrimeric proteins exhibit increased affinity for ssDNA, and RPA^V227A and RPA^E240K possess increased capacity to unfold human telomeric G-quadruplex (h-telG4) DNA. (A) Schematic depiction of the experimental schemes for the Förster resonance energy transfer (FRET)-based assays. Binding of RPA:RPA1^WT (RPA^WT), RPA:RPA1^V227A (RPA^V227A), RPA:RPA1^V270A (RPA^V270A), and RPA:RPA1^E240K (RPA^E240K) proteins was monitored by using 1 nM dT30 ssDNA molecules (top), 1 nM dT15, or TTAGGGTAAGGGTAA telomeric DNA sequence (middle) labeled with Cy3 and Cy5 fluorescent dyes at the 5′ and 3′ ends, respectively. High FRET corresponds to free ssDNA, and low FRET reflects RPA binding. Unfolding of the h-telG4 was monitored by using the (TTAGGG)₅ sequence (bottom). FRET between the Cy3 and Cy5 dyes calculated for the h-telG4 in the absence of proteins and in buffer containing K⁺ corresponds to 100% folded quadruplex, whereas the FRET value of h-telG4 in the presence of saturating concentrations of RPA^E240K in buffer containing Li⁺ corresponds to 100% unfolded h-telG4. (B) Stoichiometric binding (1 RPA: 1 dT30 molecule) was observed for RPA^E240K, and nearly stoichiometric binding was observed for RPA^WT, RPA^V227A, and RPA^V270A. The arrows mark the respective protein concentrations at inflection points of the 2-line linear regression fit. dT15 (C) and TTAGGGTAAGGGTAA (D) telomeric DNA sequence binding to RPA^WT, RPA^V227A, RPA^V270A, and RPA^E240K. The data were fitted to a quadratic binding equation. The calculated K_ds with respective fitting errors are listed in supplemental Table 3. (E-F) Melting of the h-telG4 DNA, stabilized by the presence of 100 mM KCl. (E) Extent of the h-telG4 melting reactions was calculated from the plateaus of each respective time course. (F) h-telG4 melting rates for RPA^WT, RPA^V227A, RPA^V270A, and RPA^E240K were calculated from the slopes of FRET change during the first 20 seconds of each time course (supplemental Figure 4). The data were fitted to a quadratic binding equation. The calculated apparent K_ds with respective fitting errors are listed in supplemental Table 3. In all panels, the data are shown as average for 3 independent experiments. Error bars represent standard deviation. Where not shown, error bars are smaller than the data points.

Figure 3.

Human RPA1^E240K iPSC exhibit telomere shortening and reduced hematopoietic potential. (A) Left panel: Healthy control iPSC (RPA1^WT) following CRISPR/Cas9–guided homozygous c.718G>A (p.E240K, as E240K) modification within endogenous RPA1 locus (RPA1^E240K) confirmed with Sanger analysis. Right panel: Illustration of iPSC monolayer-based differentiation to HPs and subsequently to erythroid and myeloid cell lineages. Day 10 HP cells were fluorescence-activated cell-sorted and cultured in erythroid or myeloid differentiation media for 14 days. In parallel, iPSC-derived HP were further cultured until day 21 to assess for expression of pan-hematopoietic markers. (B) Immunoblot analysis of RPA1 expression in RPA1^WT and RPA1^E240K iPSC whole-cell extracts with histone H3 as loading control. Telomere length in RPA1^WT passage 17 and RPA1^E240K passage 12 iPSCs (C) and iPSC-derived HP cells (D) using quantitative FISH. Graphs represent mean ± standard error of the mean (SEM) of 1 of 3 independent experiments (Student t test, ****P < .0001). (E) Decreased percentage of CD43⁺CD45⁺ RPA1^E240K hematopoietic cells compared with RPA1^WT at days 16 and 21. Data represent mean ± SEM of 2 independent experiments (Student t test, *P = .03, **P = .0074). (F) Graphical representation of CD71⁺CD235⁺ erythroid cells from iPSC-derived erythroid cultures at day 14. Data represent mean ± SEM of 4 independent experiments (Student t test, ***P = .0002). (G) Plot representation of CD45⁺CD11b⁺ myeloid cells from iPSC cultures. Data represent mean ± SEM of 2 independent experiments (Student t test, **P = .0018). EPO, erythropoietin; IL-3, interleukin-3; GMCSF, granulocyte-macrophage colony-stimulating factor; GCSF, granulocyte colony-stimulating factor; SCF, stem cell factor; TPO, thrombopoietin.

Figure 4.

Natural evolution of disease and somatic genetic rescue in P1. (A) White blood cells (WBC, triangles), hemoglobin (Hgb, circles), and platelets (diamonds) plotted over 23 years for P1 (pedigree 1). Red arrow indicates time of clinical presentation. (B) Illustration of germline RPA1 variant in exon 9 and somatic mutation in exon 16 with respective DNA Sanger electropherograms from BM. Bottom table depicts variant allelic frequencies from exome sequencing performed in BM and skin fibroblast DNA and RNA sequencing from BM when patient was aged 20 years. (C) Copy number neutral UPD encompassing RPA1 locus at 17p13.3 (red arrow) identified by using an SNP array. Serial SNP array analysis in BM granulocytes shows UPD expansion over time (denoted by purple brackets). (D) Schematic of RPA1 locus (gray bar) with germline (c.718) and somatic (c.1735) mutational spots 17 kb apart. Three haplotype orientations between c.718 and c.1735 identified in marrow DNA of P1 at age 19 years from 2 independent experiments using digital droplet polymerase chain reaction: left haplotype, wt/wt (c.718G wild type/c.1735G wild type) denoted by black boxes; middle haplotype, mut/wt (c.718A mutant/c.1735G wild type) denoted by green and black boxes; and right haplotype, mut/mut (c.718A mutant/c.1735T mutant) denoted by green and red boxes. (E) Ultra-deep amplicon sequencing of bone marrow DNA and RNA targeting position of RPA1 somatic mutation (c.1735) confirms near total loss of mutant RNA. (F) Longitudinal deep sequencing in BM samples from diagnosis to age 25 years showing decrease in allele frequency of the germline c.G718G>A variant (red line) and increase of the somatic c.1735G>T mutation (blue line). (G) Single cells from P1 BM at ages 13 and 17 years were sequenced for germline (RPA1:chr17:1782314:G>A) and somatic (RPA1:chr17:1798378:G>T) mutational positions using single-cell DNA (scDNA) sequencing Tapestri Platform. Violin plot shows 3 clonal populations, including homozygous wild type (blue, RPA1^WT/WT; rescue clone 1 = UPD17p), heterozygous RPA1^E240K/WT (gold, native state hematopoiesis), and heterozygous c.718G>A with concurrent c.1735G>T stop-gain (red, RPA1^{E240K/WT + K579*}= rescue clone 2). (H) Tapestri single-cell multi-omic analysis combining DNA mutation data and surface protein expression performed in P1 BM at age 17 years. Panels depict 3 clones (color coding identical to that in panel G) constructed from 2110 high-quality cells with normalized protein expression of markers for hematopoietic stem and progenitor cells (CD34), stem cells (CD90), progenitors (CD38), and terminally differentiated cells, including myeloid (CD11b), B-lymphoid (CD19), and T-lymphoid (CD3) cells.