For initial explorative analysis, the relative telomere length (RTL) was measured from DNA of granulocytes using quantitative multiplex real-time polymerase chain reaction (PCR) according to Cawthon (29). In-house reference values from peripheral blood (PB) and bone marrow (BM) granulocytes of 90 healthy BM donors (age 2–18 years) were used for percentile calculation. RTL [telomere to single copy gene (T/S) ratio] was calculated as median from at least two independent triplicate runs. The intra-assay coefficient of variation (CV) ranged from 0 to 6% (mean 3.75%), and inter-assay CV was 2–11% (mean 7%). Telomere length was validated using metaphase telomere/centromere-fluorescence in situ hybridization (T/C-FISH) in a second laboratory, as previously described (30).

Primary fibroblasts were grown in DMEM medium containing 20% FCS and 1% P/S and tested mycoplasma free. For hematopoietic cell cultures, mononuclear cells (MNCs) were isolated from BM of patients and healthy control using Ficoll-based density gradient centrifugation, and magnetically isolated CD34⁺ cells (MACS Miltenyi) were cultured for 7 days with IL6, SCF, and FLT3-L. T-cell proliferation, flow cytometric analysis of lymphocyte subsets, T-cell receptor Vβ-repertoire, and T-cell cytokine production were assessed as previously reported (31, 32). For analysis of ex vivo survival, MNCs were cultured for 6 days. At days 0, 1, 2, 3, and 6, 2 × 10⁵ cells were stained with Annexin V (AV) and PI (BD Biosciences) and analyzed by flow cytometry. Degranulation of T- and NK-cells was assessed as previously reported (33).

Primary fibroblasts from healthy control, P1, and a patient with known DC and TERC mutation were fixed for 5 min in 2% vol/vol paraformaldehyde in PBS, washed in PBS, and stained in β-galactosidase fixative solution (X-gal) in 5 mmol/l potassium ferricyanide, 5 mmol/l potassium ferrocyanide, and 2 mmol/l MgCl₂ in PBS for 16 h at 37°C. Controls and patient cells were analyzed at the same passage number, 200 cells were counted per well, and staining performed in triplicates.

Primary fibroblasts were seeded at a density of 4,000 cells/cm². Parallel cultures were grown in DMEM with GlutaMAX (Gibco) and supplemented with 15% FBS (PAN). For flow cytometry, 48 h cultures were left untreated or exposed to 10 ng/ml MMC (Medac) or initially irradiated with 1.5 Gy from a linear accelerator. Cells were detached using 1× (0.05%) trypsin (diluted from trypsin 0.5%-EDTA 0.2% solution 10×, PAA), pelleted, and stained in medium containing 15 µg/ml Hoechst dye 33342 (Molecular Probes) for 30 min in the dark. Gates were set on vital cells via propidium iodide (PI, 1 µg/ml) exclusion. Split samples were stained with 1 µg/ml 4′,6-diamidino-2-phenylindole (DAPI; Molecular Probes) in buffer containing 154 mM NaCl, 0.1 M Tris pH 7.4, 1 mM CaCl₂, 0.5 mM MgCl₂, 0.2% BSA, and 0.1% NP40 in dH₂O. Univariate flow histograms were recorded on a triple-laser equipped LSRII flow cytometer (Becton Dickinson) using UV excitation of Hoechst 33342 or DAPI, and 488-nm excitation of PI. Resulting cell cycle distributions reflecting cellular DNA content were quantified using the MPLUS AV software package (Phoenix Flow Systems).

Fibroblasts were exposed to MMC at final concentrations of 0, 10, 50, or 100 ng/ml for the final 24 h of culture. For the last 3 h, 16 µl of Colcemid Solution (10 µg/ml; PAA) per milliliter of growth medium were added. Metaphase preparation followed standard procedures. Detachment of cells was evaluated by trypsin as above. Pellets were subjected to hypotonic treatment using 10 ml of pre-warmed 0.075 M KCl for 10 min at 37°C. Nuclei and metaphases were fixed using freshly prepared, ice-cold 100% methanol, and glacial acetic acid 3:1. A minimum of 50 complete metaphases from Giemsa-stained slides of each MMC concentration were scored quantitatively for breakage rates and analyzed qualitatively for types of chromosome aberrations.

V(D)J recombination assays were performed as described previously (34). In short, human primary RTEL1-proficient or RTEL1-deficient dermal fibroblasts were transfected using the Cell Line V Nucleofector Kit (Lonza, Cologne, Germany). Transfections were performed with 1.2 µg pcWT-RAG1, 1.8 µg pcWT-RAG2, 8.0 µg pMACS11-19VDJ, or pMACS11-19Flip, and 1.0 µg pcDNA6/myc-His Version A (Invitrogen, Life Technologies, Darmstadt, Germany). After 48 h, cells were harvested and analyzed by immunofluorescence flow cytometry. Transfected fibroblasts were detected with biotin–anti mouse H-2Kk antibody (BD Biosciences, San Jose, CA, USA) against the truncated murine major histocompatibility complex class I protein H-2Kk additionally encoded by the pMACS11-19VDJ and pMACS11-19Flip plasmids. The biotin-labeled antibodies were detected by streptavidin–peridinin chlorophyll protein staining (BD Biosciences). Assays were performed three times independently for the two tested cell lines. The arithmetic means of the three values and the corresponding SDs were calculated.

The analysis of telomeric T-circles was performed as previously described (27) using patient fibroblasts (P1, P3), T-lymphoblasts (P1 and family members, P5), and cell lines [VA13 ALT, Rtel1^−/− mouse embryonic fibroblasts (MEFs), Rtel1^F/F MEFs, and BJhTERT] and as positive control fibroblasts with homozygous mutation p.Arg1264His known to exhibit T-circle loss.

The essential clinical and laboratory data of our patients are summarized in Tables 1 and 2, key clinical featured of the index patient are shown in Figure 1, and the pedigrees of the investigated families are shown in Figure 2A.

Patient P1 (Family 1) was brought to our attention for recurring bacterial and viral pneumonias manifesting from 9 months of age. Pancytopenia, B/NK-cell lymphopenia, and hypogammaglobulinemia were subsequently noted. In short span, the manifestation of leukoplakia and ataxia associated with cerebellar hypoplasia led to the clinical diagnosis of HHS (Figures 1A–C). Between 2 and 3 years of age, P1 developed refractory non-infectious diarrhea and long-segment esophageal narrowing. An increased apoptotic rate within the colon and esophageal mucosa was detected by endoscopy (Figures 1E,F). Due to rapidly progressing BMF (Figure 1G) and immunodeficiency, P1 was transplanted from her HLA-identical grandmother (further details are described below). Similar clinical symptoms were observed in P2, a second-degree cousin of P1 (Figure 2A). Within first 3 months after birth, P2 had failure to thrive and microcephaly, with recurrent infections manifesting from age of 5 months. He died at the age of 19 months due to systemic CMV infection.

Two further patients (P3 and P4) belong to Family 2 (Figure 2A). P3 initially developed leukoplakia at age of 12 months; however, his leading medical concerns were developmental delay and BMF diagnosed at age of 2 years and 9 months. He was successfully transplanted from a 9/10 matched unrelated donor (MUD) at 5.2 years of age. Identical homozygous RTEL1 mutation was identified in his twin sister (P4) by family screening. P4 initially presented with mild chronic diarrhea, high MCV and NK-cell lymphopenia, and short telomeres at the age of approximately 4 years, but later also developed mild leukoplakia (Tables 1 and 2).

P5 from Family 3 (Figure 2A) suffered from recurrent infections beginning from 6 years of age and progressive BMF after EBV infection at 9 years of age. In contrast to P1–P4, he developed severe mucosal fragility, nail dystrophy, and liver cirrhosis with secondary hypersplenism (Table 1). He became transfusion dependent for RBC and platelets at 15 years of age. Oxymetholone therapy initiated at age of 20 years later resulted in transfusion independency for platelets. However, infectious complications intensified and he died of sepsis after pneumonia at 21 years of age.

P6 from Family 4 (Figure 2A) was born to non-consanguineous parents with an uneventful family history. Postnatal development was normal and he was asymptomatic at diagnosis when he was 17 years old. Complete blood count performed prior to surgery for a shoulder dislocation revealed thrombocytopenia and elevated MCV (Tables 1 and 2). These findings prompted further investigations and a BM biopsy was compatible with the histological diagnosis of hypocellular refractory cytopenia. Telomere analysis and full DC-genetic workup eventually identified RTEL1 deficiency. Family history revealed no malignancies in all patients or their first-degree relatives.

At the time when our index patients P1 and P2 were studied, disease-causing mutations in RTEL1 were not yet reported. We therefore used Sanger sequencing to exclude mutations in nine known DC-related genes, followed by homozygosity mapping and ES. Three variants (COL9A3^Arg103Gln, PRIC285^Glu615Lys, and RTEL1^Trp456Cys) segregated with the disease (Table 3). RTEL1 was the most plausible candidate gene, given its essential role in telomere maintenance and coinciding reports on RTEL1 mutations (Figure 2B). The homozygous mutation p.Trp456Cys affects a protein region with a very high degree of homology in all 17 species and is located in a β-sheet (Figures 3A,C). Targeted re-sequencing identified four additional RTEL1 mutations in P3-P6 (Figures 2A and 3B). Parents of twins P3 and P4 are first-degree cousins and transmit the heterozygous missense mutation p.Ile425Thr; both P3 and P4 are homozygous, while their healthy siblings are heterozygous carriers (Figure 2A) and have normal RTL (not shown). The RTEL1 amino acid Ile425 is located in the α-helix and found to be conserved in 16 out of 17 tested species (Figures 3B,D). Finally, a compound heterozygous state with one truncating splice site mutation c.2652 + 5G>A in intron 28 and a frameshift mutation c.3730delTG; p.Cys1244ProfsX17 in exon 34b of RTEL1 was identified in P5 (Figure 2A). P6 was the only patient in this study who carried solely a heterozygous nonsense mutation p.Val796AlafsX4 resulting in premature stop codon prior to both C-terminal harmonin domains. This mutation could not be investigated on functional level in primary patient cells due to lack of available material.

The longest four of the seven Ensembl-annotated, protein coding isoforms are shown in Figure 4A. Mutation nomenclature in this manuscript refers to the 1300aa isoform (NM_001283009), which along with the 1219aa isoform (NM_016434) is predominantly expressed in human cells, and also served as a reference for the majority of reported mutations (Figure 2B). Notably, the 1243aa isoform (NM_032957) that compared to other isoforms holds a longer exon5 (+24aa) was detected at very low levels using RT-PCR in peripheral blood MNCs of patients and controls, while it was absent in fibroblasts, pointing toward tissue-dependent specificity of alternative transcripts (Figure 4A). Furthermore, we noticed higher expression of long transcript (present in both 1300aa/1400aa isoforms) in healthy MNC as compared to fibroblasts (Figure 4B). Finally, the longest 1400aa transcript was neither detectable in blood nor in fibroblasts using RT-PCR targeting exon 36 (not shown).

Both homozygous missense mutations identified in P1–P4 did not affect mRNA expression in RTEL1-deficient patient cells (Figure 4B). As demonstrated by RT-PCR and cDNA sequencing, P5’s RTEL1 c.2652 + 5G>A mutation abolishes the original donor splice site, resulting in transcription prolongation of exon 28, adding 159 bp out of intron 28 (Figure 4C, left panel). A novel premature stop codon (aa +13) results in loss of the proliferating cell nuclear antigen-interacting protein PIP motif, RING domain, as well as recently identified harmonin-like domains (46). The mutated transcript showed similar mRNA expression signal to the wild-type allele. The second frameshift mutation in P5 was predicted to be protein truncating however did not affect mutant mRNA level as shown by cDNA sequencing (Figure 4C, right panel).

Patients with RTEL1 deficiency exhibited very short telomeres (below the first percentile of age-matched controls) in BM, PB, and skin fibroblasts measured using qPCR and metaphase T/C-FISH (Figures 5A,B). By contrast, parents of P1–P4 had normal telomere length (Figure 5B). To investigate if heterozygous carrier status affects telomere homeostasis under replicative stress, we investigated T-cell blasts of P1’s parents. Upon long-term expansion, heterozygous RTEL1^Trp456Cys lymphocytes exhibited significant telomere shortening (Figure 5C), while homozygous RTEL1^Trp456Cys cells died prematurely (not shown). RTEL1-deficient fibroblasts of P1 and P3 showed poor growth and premature cellular senescence (Figure 5D). T-cell phenotyping in P3 and P4 revealed that the majority of CD4 T-cells had a naïve phenotype, while almost 50% of CD8 T-cells were terminally differentiated effectors with senescent phenotype (Figure 5E), as determined by CD57 co-expression (47, 48). The proportion of CD57⁺ T-cells increased over time in P3, despite absence of severe viral infections. Moreover, increased spontaneous apoptosis was observed in MNCs of P3 and P4 (Figure 5F). However, short-term proliferation, degranulation and effector cytokine production of T-cells were unaffected (Table 2).

Experimental evidence from studying RTEL1-deficient embryonic stem cells, C. elegans and human cell lines indicate that RTEL1 is required for maintaining genomic integrity and plays a key role in regulating homologous recombination (HR) (25–27). It is essential for the repair of mitotic and meiotic double strand breaks and its loss results in uncontrolled HR leading to chromosomal breakage. To test the functional effect of RTEL1 mutations on the genomic integrity in our patients, we investigated spontaneous and crosslinker-induced chromosomal damage (Figure 6). Chromosome preparations from fibroblasts of P1 showed elevated levels of chromatid/iso-chromatid rather than chromosome breaks and reunion figures (Figures 6A,D). End-to-end chromosome fusions were not observed. The spontaneous and MMC-induced G2 phase fractions of fibroblasts from P1 and P3 were significantly increased compared to normal controls (P < 0.001) (Figure 6E). By contrast, low dose of ionizing radiation (1.5 Gy) did not result in any abnormal findings (Figure 6F). Without treatment, we observed metaphases with three and five breaks, rarely ever seen in normal controls (Figure 6G). Unexpectedly, repeated chromosomal breakage analysis performed on hematopoietic cells from our patients yielded normal results (Table 1).

We had the opportunity to investigate the B-cell distribution in BM and blood of P1 at age 22 months during an acute adenovirus infection and after clearance 3 months later (Figures 7A–C). During acute infection, we observed a differentiation arrest at the level of pre-BI-progenitors, resulting in the absence of pre-BII-progenitors and immature B-cells; and correspondingly very low numbers of peripheral B-cells (Figure 7B). After clearance of infection all previously diminished B-cell populations normalized (Figure 7C). As the cellular immunity relies on continuous replenishment from hematopoietic stem cells (HSC), we additionally investigated the proliferative potential of primary RTEL1-deficient BM cells. While total BM and CD34⁺ cells of P3 and P5 failed to expand in vitro, after prolonged culture the CD34⁺ population nearly disappeared (Figures 7D,E).

The progressive B-cell lymphopenia is a frequent observation in the context of impaired somatic recombination, thus we also examined V(D)J recombination in RTEL1-deficient fibroblasts of P1. As depicted in Figure 8A, V(D)J recombination efficiency was unaffected and comparable to healthy controls. Based on recently published experimental data, RTEL1 has been proposed to dismantle T-loops during replication thus preventing catastrophic cleavage of telomeres as a whole extra-chromosomal T-circle (27). Surprisingly, we did not detect elevated T-circle formation in any of our patients or carriers (in fibroblasts, BM and PB) using rolling circle amplification assay when compared to RTEL1 null mouse cells and RTEL1-deficient human cells carrying the homozygous RTEL1^Arg1264His-mutation (Figure 8B), thus suggesting impaired T-loop disassembly may not be the only underlying cause of the disease.

Allogeneic HSCT was performed on P1 and P3 at 3.1 and 5.2 years, respectively. Both patients received non-manipulated BM after reduced intensity conditioning with fludarabine, thiotepa, and antithymocyte globulin. While P3 received BM from a 9/10 MUD, P1 was transplanted from the 48 years old HLA-identical grandmother who was healthy and had RTL within normal range. At HSCT, RTEL1 deficiency was not yet identified in the family and we were unaware of the donor’s heterozygous state for the RTEL1^Trp456Cys-mutation. Full donor chimerism was achieved from day +22 onward and remained stable. At last follow-up (day +1,530), P1 showed persisting hematological and immunological reconstitution. There were no signs of graft-versus-host-disease or chemotherapy-related toxicity. Chronic diarrhea improved with clearance of viral infections. Nonetheless, her esophageal stenosis persists over time and requires frequent dilatation to allow normal oral nutrition. Similar to other recently described DC patients P1 also developed multiple pulmonary arteriovenous malformations with need for continuous oxygen support (20). The course of HSCT in P3 was uneventful. Full donor chimerism was achieved from day +30 onward. At last follow-up at 2 years 6 months post HSCT the patient was 7 years 8 months old, showed good developmental progress and normal hematological and immunological function. His oral leukoplakia showed moderate progress but he did not show signs of esophageal stricture or other mucosal problems so far.