Heterozygous RTEL1 variants in bone marrow failure and myeloid neoplasms

Judith C. W. Marsh, Fernanda Gutierrez-Rodrigues, James Cooper, Jie Jiang, Shreyans Gandhi, Sachiko Kajigaya, Xingmin Feng, Maria del Pilar F. Ibanez, Flávia S. Donaires, João P. Lopes da Silva, Zejuan Li, Soma Das, Maria Ibanez, Alexander E. Smith, Nicholas Lea, Steven Best, Robin Ireland, Austin G. Kulasekararaj, Donal P. McLornan, Anthony Pagliuca, Isabelle Callebaut, Neal S. Young, Rodrigo T. Calado, Danielle M. Townsley and Ghulam J Mufti

Data supplements

Article Figures & Data


  • Figure 1.

    Heterozygous RTEL1 variants. (A) Linear representation of RTEL1 isoform 3 (1300 amino acids; NM_001283009.1). Colored boxes indicate the conserved RTEL1 domains, and lines indicate the positions of individual variants in the gene. Twenty-one novel (black) and 2 previously described (red) heterozygous variants were identified in 8 patients from the NIH cohort (n = 59) and in 19 patients from the KCH cohort (n = 457). Variants are represented by triangles, circles, and squares, which also represent the type of the variant (missense variants, stop-gained variants, or splicing variants, respectively). Blue triangles represent 2 compound heterozygous patients. Splicing variants are indicated by dashed lines. *Eleven variants had a CADD phred score of >15 (this score was selected to predicted RTEL1 deleteriousness; supplemental Table 2). (B-G) Ribbon representation of the 3D structure models of the RTEL1 domains. Amino acid changes observed in patients are colored in red and displayed in an atomic representation in the figure within their close neighborhood. Panel B focuses on variants located on the ARCH and HD2 domains and panel C on the first harmonin-N-like domain. Panels D-G focus on variants located on the C4C4-RING domain, helicase domain 1, the FeS cluster, and helicase domain 2, respectively. The RTEL1 3D structure information obtained by comparative modeling predicted 14 of 23 RTEL1 variants as substitutions that may destabilize the domain 3D structure in which they are located or affect the protein partner interactions of RTEL1 (supplemental data). (H) The frequency of potential deleterious RTEL1 missense and LoF in the ExAC database (n = 60 706), KCH (n = 457), and NIH (n = 59) cohorts. Missense variants are enriched in the NIH cohort but not in the KCH cohort when compared with the ExAC database. However, the KCH cohort had a very significant enrichment in LoF variants when compared with ExAC. (I) Frequency of missense RTEL1 variants stratified by CADD phred score in the ExAC database, and the KCH and NIH cohorts. The CADD predicts deleteriousness of a given variant. All variants with CADD over 15 from KCH and NIH had CADD over 20 as well. The CADD of 15 was selected as a cutoff to predict a variant as deleterious on the basis of the frequency identified in the ExAC database. HD1, helicase domain 1; HD2, helicase domain 2; HNL-1, first harmonin-N-like domain.

  • Figure 2.

    Schematic diagram representing RTEL1 variants identified in this study. (A) We screened 516 patients from 2 independent cohorts (the KCH and NIH), identifying 23 RTEL1 variants in 27 unrelated patients. The number of patients with pathogenic/likely pathogenic, uncertain significance, or likely benign variants are presented by colored bars according to their clinical diagnosis. (B) Twenty-five patients carried heterozygous RTEL1 variants, whereas 2 patients were biallelic. The number of patients with LP, US, and LB variants are summarized in the figure. Heterozygous LP variants were enriched in the group of patients suspected to have inherited BMF (see supplemental Figure 1 for cohort’s data). *Patients with hypoMDS and not suspected to have constitutional BMF. †Four RTEL1 variants identified in patients NIH-1 and NIH-2 were associated with an autosomal recessive form of AA because they were compound heterozygous. ‡Despite being classified as LB, the P871L variant was seen together with the D719N variant in compound heterozygosity.

  • Figure 3.

    Pedigrees of 6 families identified with RTEL1 variants. Pedigrees from proband KCH-11, KCH-9, KCH-8, NIH-1, NIH-2, and NIH-5, respectively. The RTEL1 variant presented in the family is described above each pedigree. Arrows indicate the probands of each family. Open circles and squares show females and males who are noncarriers, respectively. Half-filled or solid circles and squares represent individuals who are heterozygous or homozygous for the RTEL1 variant, respectively. A line through the circle or square indicates individuals who are deceased. When tested, TL and clinical features are described in the figure under each individual. Relatives who lack clinical and mutational data are indicated by a question mark. For families NIH-1 and NIH-2, probands were identified as compound heterozygous (half-red, half–light red, square). In NIH-1, the proband’s father and mother carried the D719N (half-red square) and the P871L (half–light red circle) variants, respectively, both heterozygous. In NIH-2, the father was not screened, but the mother carried the G951S variant in heterozygosity. BMF, bone marrow failure; CY/HD, compound heterozygous for C282Y and H63D hereditary hemochromatosis genes; KCH, King’s College Hospital cohort; NIH, National Institutes of Health cohort; MAA, moderate aplastic anemia; TLCO, transfer factor of the lung for carbon monoxide.

  • Figure 4.

    Impact of RTEL1 variants on telomere maintenance. (A) Normalized single-strand 3′ overhang measurement by nondenaturing Southern blot of patients from the NIH cohort and 2 siblings of NIH-5 (S1- and S2-NIH5). Relative 3′ overhang signals were determined by normalizing the sum of chemiluminescent signal (ΣODi) from each column in the nondenaturing membrane (overhang signals) by the telomeric signal in the denatured membrane (representing total genomic DNA). The background signal (sample treated with Exo1) was subtracted from total ΣODi for all the samples. Error bars represent the standard deviations of independent experiments. Patients’ 3′ overhang signals were then normalized by an average of 3′ overhang signals from healthy individuals used as a control in every experiment and plotted. The 3′ overhang signals of patients below a 95% confidence interval of 3′ overhang measurements from controls (gray interval in the graphic) were considered eroded. (B) Normalized single-strand 3′ overhang measurement of 14 patients from the NIH cohort without RTEL1 variants in which DNA was available. Three patients had telomeric overhangs below a 95% confidence interval of healthy controls that were considered short. (C) Western blot analysis. (Left) Whole extracts of bulk 293T cells stably expressing recombinant RTEL1-FLAG wild-type or one of the following RTEL1 variants: M652T, D719N, G951S, and F1262L. Control used was 293T infected with an empty vector. (Right) Whole extracts (input) of isolated RTEL1-FLAG WT or 293T clones with the P82L, M652T, D719N, G951S, and F1262L variants were immunoprecipitated with anti-FLAG to evaluate RTEL1 and TRF2 interaction. Protein expression was analyzed with antibodies as indicated. (D) TRF2 expression in both bulk-infected 293T cells and isolated clones normalized by the control (empty vector). Error bars represent the standard deviations between TRF2 expression in bulk 293T and isolated clones. (E) Clustergram of the telomere biology gene expression levels in 4 patients with RTEL1 variants from the NIH cohort and 2 controls using the RT2 Profiler PCR array system (Qiagen). Control samples were used as a reference for normalization. For comparison, the RTEL1 and TRF2 fold-change expression in relation to controls were plotted and shown in the graphic below the heatmap. The RTEL1 gene expression was validated by real-time PCR using a Taqman probe that detected the boundary of exon 7-8 in all isoforms and another probe that detected the RING domain in isoform 3. (F) Phi29-dependent t-circle amplification assay in patients’ peripheral blood. The t-circles were detected in a lower amount in patients with heterozygous variants than in biallelic patients. The asymptomatic mother of NIH-2 (P1-NIH2) and a healthy control (CT1) presented the highest number of t-circles. DNA extracted from VA13 cells was used as a control of the assay. IPP, immunoprecipitation; RT-PCR, reverse transcription polymerase chain reaction.

  • Figure 5.

    Schematic representation of RTEL1 roles in telomere dysfunction. In normal conditions, RTEL1 promotes G4 quadruplex and t-loop unwinding for DNA replication. The RING-domain mutations inhibit RTEL1-TRF2 interaction and then t-loop resolution.5 Thus, abnormal SXL4/1 activation persistently cuts t-loops, leading to telomere shortening as well as the accumulation of t-circles in cells. Short telomeres, commonly seen in patients with mutated RTEL1, trigger cell senescence and apoptosis via p53, leading to hematopoietic stem cell depletion in BM. However, some RTEL1 mutations were related to excessive 3′ overhang attrition in the absence of telomere shortening. In the context of cells maintaining their telomere lengths, the sustained DNA damage response activation caused by uncapped telomeres may promote leukemogenesis by molecular pathways distinct from typical accelerated telomere attrition associated with very short telomeres. Also, different RTEL1 variants modulated TRF2 expression in vitro, which may be an alternative mechanism related to RTEL1 dysfunction rather than impaired t-loop disassembly alone. DDR, DNA damage response; HSC, hematopoietic stem cell.


  • Table 1.

    Description of 27 patients with RTEL1 variants

    Patient IDAge/sexTL*RTEL1 variantsOther germline variantsRTEL1 ACGMHematologic diagnosisSomatic anomalies and other clinical featuresFamily historyOutcome
    Heterozygous pathogenic or likely pathogenic variants
     KCH-136/M<1stR974X, c.2920 C>T§TERT, R756C c.2266 C>TPathogenicFamilial hypoMDSAbnormal lung function: TLCOc 72%Brother has hypoMDSResponded to cyclosporine
     KCH-325/M<1stR974X, c.2920 C>T§NonePathogenicHypoMDSSuspected to have inherited diseaseFH of cancerNo treatment
     KCH-829/F<1stc.301+1G>ANoneLPICUS: neutropeniaBicuspid aortic valve, high arched palatePedigree (Figure 3)Treated with G-CSF
     KCH-1031/F<1stR986X, c.2956 C>T§NonePathogenicICUS: neutropeniaDystrophic nails, neutropenia for 16 yNoneTreated with G-CSF
     KCH-1154/F<10thR986X, c.2956 C>T§Not testedPathogenicMacrocytosis, hypocellular BMDystrophic nailPedigree (Figure 3)No treatment
     KCH-1530/F<1stc.301+1G>ANoneLPAA/PNHNoneNoneTreated with IST, LTFU
     NIH-431/FNormalM652T, c.1955 T>CSLX4, T750M c.2249 C>TLPICUS: neutropeniaNormal BM cellularityNoneNo treatment
    Biallelic pathogenic or likely pathogenic variants
     NIH-132/M<1stD719N, c.2155 G>A P871L, c.2612 C>TNoneLPMAALiver cirrhosis, pulmonary fibrosisPedigree (Figure 3)Responded to androgen (danazol)
     NIH-26/F<1stG951, c.2851 G>A F1262L, c.3786 C>GNoneLPSAAProlonged thrombocytopeniaPedigree (Figure 3)Awaiting HSCT
    Variants of uncertain significance
     KCH-271/M<10thV178A, c.533 T>CNoneUSMDS (RAEB1)Laryngeal squamous cell cancer, age of 60 y; small cell lung cancer, age of 66 yFH of cancerTreated with 5-azacitidine, progressed to AML, died
     KCH-437/M<1stA549T, c.1645 G>ANoneUSHypoMDSReticulate skin pigmentation, skin warts, liver fibrosis, portal hypertension, esophageal varicesFH of cancerNo treatment
     KCH-533/F<10thQ397E, c.1189 C>GNoneUSICUSNoneNoneEvolved to hypoMDS
     KCH-919/F<1stD220N, c.658 G>ATERT, A1062T c.3184 G>AUSMAAShort stature, reticulate skin pigmentation, dystrophic nails, Lichen planus, epiphoraPedigree (Figure 3)No treatment
     KCH-1233/M<10thA990V, c.2969 C>TNoneUSBMFS, ICUSLearning difficulties, epilepsy, urogenital anomalies, hyperostosis cranium, occult spina bifidaNoneG-CSF, died lymphoma
     KCH-1373/M<1stP992L, c.2975 C>TNoneUSCMML1NoneFH of cancerNo treatment
     KCH-1453/F<1stQ397E, c.1189 C>GNoneUSThrombocytopenia/MAAShort statureFH of cancer, brother has short TLPartial response to IST, relapsed
     KCH-1629/F<1stP824S, c.2470 C>TNoneUSMAAPresented in pregnancyNoneNo treatment
     KCH-1738/M<1stV402A, c.1205 T>CNoneUSHypoMDSNoneNoneNo treatment
     KCH-1963/M<1stD942N, c.2824 G>ANoneUSAMLNoneNoneChemotherapy, MUD HSCT, died GVHD
     NIH-317/MNormalP884L, c.2651 C>TNoneUSSAANoneFH of cancerTreated with IST/EPAG and evolved to −7/MDS
     NIH-563/FNormalP82L, c.245 C>TNoneUSMAAMitral valve prolapsePedigree (Figure 3)Progressed to SAA on EPAG and evolved to MDS/AML; died following HSCT of relapsed AML
     NIH-732/F<1stA902G, c.2705 C>GTERC, r.287 C>G§USMAAFrequent miscarriages, early graying of hairMaternal early graying of hair, MDS in maternal uncle, BMF in paternal grandmotherNo treatment
    Likely benign variants
     KCH-652/M<1stM320T, c.959 T>CTERT, A279T c.835 G>ALBHypoMDSCeliac disease, hepatosplenomegaly, abnormal lung functionNoneMUD HSCT; complete response
     KCH-738/F<10thP867L, c.2600 C>TNoneLBMacrocytosisShort stature, normal BM cellularityNoneNo treatment
     KCH-1873/MNormalM320T, c.959 T>CNoneLBMDS (RAEB2)NoneNoneBest supportive care; died
     NIH-623/FNormalP867L, c.2672 C>TTERT, R537C c.1609 C>TLBMAAEczemaFather and brother with early graying of hair, mother with macrocytic anemiaNo treatment
     NIH-847/MNormalP996H, c.2987 C>ANoneLBhypoMDSNoneSister with SLE, paternal grandfather with polycythemia veraResponded to EPAG
    • The RTEL1 variants identified in our study were annotated using the isoform 3 (1300 aa; NM_001283009.1).

    • CMML, chronic myelomonocytic leukemia; EPAG, eltrombopag; F, female; FH, family history; GVHD, graft-versus-host disease; HSCT, hematopoietic stem cell transplantation; ID, identification; IST, immunosuppressive therapy with antithymocyte globulin and cyclosporine; LTFU, loss to follow-up; M, male; MAA, moderate aplastic anemia; MUD, matched unrelated donor; PNH, paroxysmal nocturnal hemoglobinuria; RAEB, MDS with refractory anemia with excess blasts; SAA, severe AA; SLE, systemic lupus erythematosus; TLCO, transfer factor of the lung for carbon monoxide.

    • * <1st, TL below the first percentile of age matched controls (very short telomeres); <10th, TL below tenth percentile (short telomeres).

    • For the KCH cohort, the targeting next-generation sequencing panel had 12 candidate genes. For the NIH cohort, the targeting next-generation sequencing panel had 49 candidate genes.

    • ACMG consensus criteria.

    • § Variants previously reported as pathogenic.