Revertant Somatic Mosaicism by Mitotic Recombination in Dyskeratosis Congenita

Abstract

Revertant mosaicism is an infrequently observed phenomenon caused by spontaneous correction of a pathogenic allele. We have observed such reversions caused by mitotic recombination of mutant TERC (telomerase RNA component) alleles in six patients from four families affected by dyskeratosis congenita (DC). DC is a multisystem disorder characterized by mucocutaneous abnormalities, dystrophic nails, bone-marrow failure, lung fibrosis, liver cirrhosis, and cancer. We identified a 4 nt deletion in TERC in a family with an autosomal-dominant form of DC. In two affected brothers without bone-marrow failure, sequence analysis revealed pronounced overrepresentation of the wild-type allele in blood cells, whereas no such skewing was observed in the other tissues tested. These observations suggest that this mosaic pattern might have resulted from somatic reversion of the mutated allele to the normal allele in blood-forming cells. SNP-microarray analysis on blood DNA from the two brothers indeed showed independent events of acquired segmental isodisomy of chromosome 3q, including TERC, indicating that the reversions must have resulted from mitotic recombination events. Subsequently, after developing a highly sensitive method of detecting mosaic homozygosity, we have found four additional cases with a mosaic-reversion pattern in blood cells; these four cases are part of a cohort of 17 individuals with germline TERC mutations. This shows that revertant mosaicism is a recurrent event in DC. This finding has important implications for improving diagnostic testing and understanding the variable phenotype of DC.

Introduction

Dyskeratosis congenita (DC [MIM 127550]) is a multisystem bone-marrow-failure syndrome that was initially described as a combination of nail dystrophy, abnormal skin pigmentation, and oral leukoplakia. Affected persons exhibit a susceptibility to aplastic anemia, lung fibrosis, liver cirrhosis, and cancer.¹ The clinical presentation of patients is highly variable both between and within families.

Compared to age-matched controls, persons with DC have abnormally short telomeres.² Telomeres are complex DNA-protein structures at the end of chromosomes, and they protect the chromosomes from damage and, thereby, maintain chromosome stability.³ Telomeres shorten with each cell division and ultimately activate a DNA-damage response that leads to apoptosis or cell-cycle arrest.⁴ In humans, telomerase-based telomere elongation is the major mechanism that counteracts this process of telomere shortening.⁵ After birth, telomerase activity is restricted to germ cells, some stem cells and their immediate progeny, activated T cells, and monocytes.⁶ Approximately half of the DC patients have mutations in one of six genes that encode components of either the telomerase complex (DKC1 [MIM 300126], TERC [MIM 602322], TERT [MIM 187270], NOP10 [MIM 606471], and NHP2 [MIM 606470]) or the telomere shelterin complex (TINF2 [MIM 604319]). Mutations in these genes result in failure to maintain telomere length, and they primarily affect tissues that turn over most rapidly, including bone marrow, skin, and nails.

Individuals with DC display considerable variability in clinical severity, which might be partly explained by genotype-phenotype correlations. In general, individuals with DKC1 and TINF2 mutations exhibit an earlier and more severe presentation than do individuals with a mutation in one of the other genes.^1,7 This finding was recently confirmed by studies in iPS cells; such studies showed that DKC1 mutations result in a more severe telomere-maintenance defect than do mutations in TERT.⁸ Variable clinical phenotypes are also observed among individuals with mutations in similar DC-associated genes or even between carriers within affected families. The disease might even present as isolated aplastic anemia or idiopathic pulmonary fibrosis.^9–12 In autosomal-dominant DC, disease anticipation is one of the factors underlying variable expression, i.e., when phenotypes are present earlier and more severely in successive generations.^13,14 A sustained decrease in telomere length through generations is probably responsible for this process.^13,14

Phenotypic expression of an inherited mutation can also be influenced by a mosaic tissue distribution caused by its reversion to a normal allele. This phenomenon is infrequently observed and might be recognized if a person presents with a milder-than-expected clinical course or with a mixture of phenotypically normal and abnormal cells.¹⁵ Mechanisms that might explain reversion include mitotic gene conversion, back mutation, intragenic mitotic recombination, and the occurrence of compensatory mutations. Here, we report on revertant somatic mosaicism resulting from mitotic recombination in six DC-affected persons carrying germline TERC mutations.

Subjects and Methods

Subjects and DNA Samples

All persons or their legal representatives provided informed consent for the DNA studies and the collection of clinical data. The studies were performed according to the guidelines of the local ethical committees. Clinical information of the seven individuals from the Dutch family was obtained through clinical investigation in the departments of Human Genetics and Pulmonology (by authors M.C.J.J. and Y.H., respectively) of the Radboud University Nijmegen Medical Centre and by chart review. DNA was isolated from peripheral blood cells, frozen liver and lung tissue, and cultured fibroblasts via standard procedures.

To screen for additional examples of reversion in TERC, we selected subjects from the DC registry, which is located at the Royal London Hospital and administered by Dr. Dokal. In this study, we included 17 individuals who had a germline TERC mutation, who did not develop bone-marrow failure, and who thus did not undergo an allogeneic stem-cell transplantation (Tables 1 and 2 and Table S1, available online).

Table 1.

Clinical Features of the Dutch Family Members Affected by Dyskeratosis Congenita

	II:1	II:3	II:6	II:7	III:6	III:7	III:12
Reticular pigmentation	-	+	?	+	-	-	+
Dystrophic nails	+	+	?	+	-	+	+
Leukoplakia	-	-	?	-	-	-	-
Bone-marrow failure	-	-	-	-	thr	+ (34 years)	+ (15 years)
Pulmonary fibrosis	+ (57 years)	+ (58 years)	+ (53 years)	+ (57 years)	-	+ (33 years)	+ (17 years)
Liver cirrhosis	-	-	-	-	+ (21 years)	-	-
Osteoporosis	+ (57 years)	+ (65 years)	?	?			+
Malignancy	-	+ (66 years)a	-	-	-	-	-
Age at death (years)	59	66	55	59	32	alive	26

The presence (+) or absence (-) of each feature is reported for each individual. Age at first presentation is indicated where known. The following abbreviations are used: thr, thrombocytopenia; ?, data not available.

^aSquamous cell cancer of the head and neck region.

Table 2.

TERC Mutation Carriers and Somatic Reversion

Family	Subjects with DCa	Gender	TERC Mutation	Reference	Age at Molecular Diagnosis (Years)	Percentage of Overrepresented WT Peaksb	Reversion Visible in Array Plot	Significance of the Allelic Imbalancec
1	III:12, liver DNA	M	n.54_57del	-	26	86%	no	1
1	II:1	M	n.54_57del	-	58	100%	yes	5.26 × 10-97
1	II:1, fibroblast DNA					40%	NA	NA
1	II:7	M	n.54_57del	-	57	100%	yes	1.58 × 10-84
1	II:7, lung DNA					60%	NA	NA
2	1	M	n.96_97del	Vulliamy et al. (2004)14	11	44%	no	1
3	2	M	n.408C>G	Vulliamy et al. (2001)16	18	-	no	1
4	3	M	n.110_113del	Vulliamy et al. (2002)11	51	42%	no	1
5	4	M	n.54_57del	Vulliamy et al. (2006)19	53	88%	no	7.91 × 10-25
5	5	F	n.54_57del	Vulliamy et al. (2006)19	18	36%	no	0.67
6	6	F	n.79del	Vulliamy et al. (2006)19	77	54%	no	1
5	7	F	n.54_57del	Vulliamy et al. (2006)19	25	88%	yes	1.76 × 10-100
5	8	F	n.54_57del	Vulliamy et al. (2006)19	29	68%	no	1
3	9	F	n.408C>G	Vulliamy et al. (2001)16	0	-	no	1
7	10	M	n.48A>G	Vulliamy et al. (2006)19	54	-	no	1
8	11	F	n.2G>C	Marrone et al. (2007)32	54	-	no	1
8	12	F	n.2G>C	Marrone et al. (2007)32	28	-	no	1
9	13	M	n.176A>C	Vulliamy et al. (2011)33	5	-	no	1
10	14	M	n.110_113del	Vulliamy et al. (2011)33	54	76%	no	2.75 × 10-11
11	15	F	n.95_96del	Vulliamy et al. (2011)33	69	88%	yes	3.95 × 10-49
12	16	M	n.107G>T	Vulliamy et al. (2011)33	55	-	no	0.38
13	17	F	n.36C>T	Vulliamy et al. (2011)33	45	-	no	1

For the bold individuals in column 2, we found significant evidence of revertant somatic mosaicism. The following abbreviations are used: WT, wild-type; M, male; F, female; and NA, data not available.

^aDNA was isolated from peripheral blood unless stated otherwise.

^bPercentage of WT peaks that are higher than the corresponding mutant peak in the sequencing chromatogram. For each sample with a small deletion in TERC, 50 nonsynonymous base-pair positions (double peaks in the chromatogram) just downstream of the mutation were scored. Twenty samples with a frameshift mutation in BRCA1 were scored as a control cohort, resulting in a median of 48% (scores ranging from 34% to 58%) WT peaks that are higher than the corresponding mutant peaks.

^cAs calculated by the mosaic homozygosity reporter; the formulas are described in the Subjects and Methods section and summarized in the legend of Table S2.

Mutation Screening

We screened TERC and TERT for mutations by denaturing high-performance liquid chromatography and by using direct sequence analysis with the primers and PCR conditions previously described.^16,17 All mutations in TERC are numbered according to the reference sequence with RefSeq accession number NR_001566.1.

Genome-Wide SNP Genotyping

DNA samples were hybridized on a SNP 6.0 array according to the manufacturer's (Affymetrix, Santa Clara, CA, USA) protocols. Copy-number and allele-specific genotyping analyses were performed with the Affymetrix Genotyping Console v2.1 software and Nexus Copy Number 5.0 software, respectively (BioDiscovery, El Segundo, CA, USA).

Sorting of Blood-Cell Lineages

Peripheral-blood mononuclear cells (PBMCs) were isolated by Ficoll-Paque 1077 (GE Healthcare, Buckinghamshire, UK) density gradient centrifugation (20′, RT, 700 g, without brakes). The granulocytes were harvested from the flow through and kept on ice until DNA isolation. We further purified the fraction monocytes from the PBMCs by adhering them to a plastic surface for one hour in a CO₂ incubator and directly lysing them in QIAamp lysisbuffer (QIAGEN, Hilden, Germany). T and B cells were purified by means of direct cell labeling with CD3- and CD19- magnetic beads, respectively, according to the manufacturer's (Miltenyi Biotec GmbH) protocol, and they were also kept on ice until DNA isolation.

Mosaic Homozygosity Reporter

We developed a method of identifying mosaic homozygosity by detecting allelic imbalance in telomeric regions. This method assesses significant shifts in the allele-contrast signal of heterozygous SNPs obtained from Affymetrix SNP 6.0 arrays. The heterozygous range (0–0.5) of the absolute allele contrast (from now on referred to as the heterozygous allele contrast [HAC]) of the 3q telomeric region is compared with other telomeric regions of similar length. The length was taken as the number of SNPs on the SNP 6.0 array from TERC to the end of 3q.

We tested differences in HAC values between the 3q telomeric region and 34 telomeric regions elsewhere on the genome by performing a Wilcoxon rank test. We took the mean of the p values and, to correct for multiple testing, multiplied this number by the amount of telomeric comparisons (n = 34). All telomeric regions were included, except for those in which the window of SNPs crossed the centromere or those that had no probe coverage, i.e., 13p, 14p, 15p, 17p, 18p, 19p, 19q, 21p, and 22p. The formulas used for this method are depicted in the legend of Table S2.

Results

Family History

A Dutch family with autosomal-dominant DC (Figure 1 and Table 1) was identified through a 26-year-old proband (III:12). This proband presented with multiple features, i.e., dystrophic nails, reticular pigmentation of the neck and chest, a gray lock of hair, aplastic anemia, lung fibrosis, liver disease, and avascular necrosis of the hips, compatible with DC. The family history was positive for lung fibrosis. In addition, a cousin died because of liver cirrhosis, and another cousin was recently diagnosed with aplastic anemia.

Figure 1.

Pedigree of the Dutch Family Affected by DC

The proband (III:12) is marked by an arrow.

Underrepresentation of a Mutant TERC Allele in Two Affected Family Members

Sequence analysis of DNA derived from liver tissue of the proband revealed no pathogenic mutation in TERT and a heterozygous 4 bp deletion (n.54_57del) in exon 1 of TERC (Figure 2). TERC encodes the telomerase RNA component that serves as a template in the process of telomere elongation.¹⁸ This deletion disturbs the template domain of TERC and is, therefore, considered to be pathogenic. A partly overlapping deletion has previously been reported in an individual who presented with hypoplastic myelodysplasia and nail dystrophy at 13 years of age.¹⁹

Figure 2.

Sequence Analysis of TERC

Sequence analysis of TERC revealed a heterozygous 4 bp deletion, n.54_57del (RefSeq accession number NR_001566.1), in the proband (III:12). The wild-type allele was observed more than the mutated allele in DNA isolated from peripheral blood cells from the father (II:7) and an uncle (II:1) of the proband. Analysis of DNA from the father's lung tissue and from the uncle's cultured fibroblasts revealed that the mutation was heterozygous. These data are in concordance with somatic events in the hematopoietic compartments of II:7 and II:1; these events resulted in loss of the germline mutation in a considerable fraction of the peripheral blood cells.

Subsequent segregation analysis confirmed the presence of the mutation in all affected family members. In the father (II:7) and an uncle (II:1) of the proband, we observed discrepancies in bidirectional sequencing-peak heights in independent PCR amplicons from blood-cell-derived DNAs, suggesting an underrepresentation of the mutant TERC allele (Figure 2). This phenomenon appeared to be limited to hematopoietic cells, given that the mutation was present in an equimolar heterozygous state (Table 2) in DNA derived from lung tissue from II:7 and fibroblasts from II:1 (Figure 2). The tissue-restricted nature of this mosaicism, combined with recurrence in at least two family members, indicates that these events might have resulted from a somatic reversion of the mutated allele to a normal state.

SNP-Array Analysis Reveals Mitotic Recombination

To further explore this possibility, SNP-array analysis was performed on DNA derived from peripheral blood cells of the two affected brothers (II:1 and II:7). Indeed, this analysis revealed the presence of acquired uniparental disomies (UPDs) of a large chromosome 3q segment, which included TERC in both brothers (Figure 3). Moreover, in person II:1, the UPD region was flanked by a small genomic stretch with an intermediate level of homozygosity, suggesting the presence of two subpopulations of cells with differently sized segments of acquired UPD involving the same parental chromatid. Therefore, we conclude that the underrepresentation of the mutant TERC allele in blood DNA of person II:1 was caused by at least two independent events of mitotic recombination and that this recombination mechanism explains the reversion in both individuals.

Figure 3.

Mitotic Recombination on Chromosome 3q Results in Isodisomy of TERC

B-allele frequencies (y axis) of chromosome 3 of proband III:12 (his DNA was derived from the liver), his father, II:7, and his uncle, II:1 (their DNA was derived from peripheral blood). This analysis revealed UPD of the long arm of chromosome 3 in subject II:7. This is likely the result of a mitotic recombination event in the centromeric region (indicated by the triangle) of chromosome 3q. In subject II:1, mosaic UPD was observed, and two adjacent regions of chromosome 3q had different degrees of mosaicism. This suggests the presence of two subpopulations of cells (marked as 1 and 2) resulting from independent mitotic-recombination events with different breakpoints.

Mosaic Reversion Patterns in Distinct Blood Cell Lineages

In order to determine the origin of the two subpopulations of cells with differently sized segments of acquired UPD, we resequenced the TERC deletion in DNA isolated from different blood cell lineages of person II:1 (Figure S1A). Whereas the B cell lineage again showed a subfractional reversion, we observed normal heterozygous levels of the mutant allele in the T cells and an almost complete loss of the mutated allele in the myeloid lineage (granulocytes and monocytes). These sequencing results were reflected in the amount of UPD observed by SNP-array analysis (Figure S1B). Furthermore, both subpopulations exhibiting distinct mitotic-reversion events were observed in B cells, granulocytes, and monocytes. These data indicate that both events of mitotic recombination must have occurred in an early hematopoietic progenitor cell, i.e., before the segregation of myeloid- and lymphoid-differentiation pathways. Our finding that mitotic reversion is not visible in the B-allele-frequency plot of the T cells can be explained by the remarkable longevity of T-lymphocyte subsets.²⁰

Revertant Somatic Mosaicism in Additional Individuals Carrying a Germline TERC Mutation

For mitotic-reversion analysis, we selected from 12 families 17 individuals with a germline TERC mutation and with no history of bone-marrow failure (Table 2 and Table S1). TERC was resequenced in DNA derived from peripheral blood cells of those individuals, and SNP-array analyses were performed. Visual inspection of the B-allele-frequency plot revealed the presence of mosaic homozygosity in two cases (subjects 7 and 15; Figure 4). Subsequently, we used a quantitative method (mosaic homozygosity reporter; see Subjects and Methods and Table S2) to detect imbalances in the intensity ratio of heterozygous SNPs, and we identified two additional persons (subjects 4 and 14) with significant imbalances in the allelic ratios of heterozygous calls on the 3q arm; these imbalances are indicative of a partial loss of the mutant allele (Table 2). Two of these four novel subjects, a father (subject 4) and his daughter (subject 7), are related, and they carry the same mutation (n.54_57del) in TERC as the one observed in the Dutch family. Subjects 14 and 15 were also affected by small deletions n.110_113del and n.95_96del, respectively. The sequencing chromatograms showed that directly downstream of the deletions in all six individuals with significant reversion (detected by the mosaic homozygosity reporter) there was a disparity between the peak heights of the wild-type and the shifted mutant sequences (Table 2). These data illustrate that mosaic reversion of the mutant allele to the wild-type form is a recurrent event in individuals carrying a TERC mutation.

Figure 4.

Additional DC-Affected Individuals Show Revertant Somatic Mosaicism

SNP-array analysis revealed stretches of UPD on chromosome 3q in peripheral blood DNA of subjects 7 and 15, as observed in the B-allele-frequency plot. The starting points of these homologous stretches are indicated by a triangle. No indication of UPD was observed in the SNP-array data of chromosome 3q in subjects 4 and 14. However, quantitative analysis of the SNP data revealed an allelic imbalance in these samples (Table S2), indicating the presence of a significant population of cells homozygous for a region on 3qter. The sequence data were scored for an imbalance in peak heights between the wild-type and mutant peaks, as summarized in Table 2. All four samples showed an overrepresentation of higher wild-type peaks.

Discussion

We report reversion of a TERC mutation in six individuals from four families affected by autosomal-dominant DC. A mosaic pattern of somatic reversion becomes apparent when normal cells have a selective growth advantage over surrounding mutant cells, implying that the affected genes are expressed in regenerating organ systems like skin and blood.¹⁵ This holds true in particular for diseases with an underlying mechanism resulting in genomic instability or high mutation rates; two such diseases are Bloom syndrome [MIM 210900] and Fanconi anemia [MIM 227650], which are both caused by gene defects in DNA-repair pathways.^21,22 Ichthyosis with confetti [MIM 609165], caused by mutations in KRT10 [MIM 148080], is a recently described example of a skin disorder displaying multiple events of reversion.²³ In this condition, normal skin spots appear early in life and increase in number and size over time. Each normal spot results from a separate event of loss of heterozygosity on chromosome 17q, which harbors KRT10, via mitotic recombination.

Shortening of the telomeres conceivably initiates the events of mitotic recombination in DC-affected individuals. Several studies have demonstrated chromosomal instability in DC.^24–26 Telomere shortening triggers this instability by inducing inappropriate attachment of the chromosome ends through nonhomologous end joining or homologous recombination pathways.²⁷ The mitotic recombinations on chromosome 3q might also result from stochastic events, meaning that the reversion in DC is mainly driven by the selective advantage of cells with two functional copies of TERC. In this study, we have specifically screened for additional cases of reversion caused by mitotic recombination. Consequently, we might have missed reversion events caused by other mechanisms, including back mutation and the occurrence of compensatory mutations.

Altogether, we have recorded seven different events of reversion caused by somatic recombination in the six individuals presented here. On the basis of this observation and given the fact that DC meets all the criteria for a disease that is prone to revertant mosaicism, we expect that reversion will occur more often in persons affected by DC. This creates opportunities for the identification of novel genes associated with DC. Searching for overlapping homozygous regions in DC-affected individuals who did not develop bone-marrow failure might lead to the identification of a causative mutation, as was previously shown for mutations in KRT10 in Ichthyosis with confetti.²³

In this study, we used several tests to identify reversion caused by mitotic recombination. Quantitative interpretation of the SNP 6.0-array data by the mosaic homozygosity reporter that we have developed turned out to be the most sensitive test. In individuals 4 and 14 (Figure 4), we did not observe a stretch of mosaic isodisomy in the B-allele-frequency plot, but the mosaic homozygosity reporter revealed the presence of a significant population of cells homozygous for a region on 3qter. Retrospectively, we also detected that the wild-type allele was more highly represented in the sequencing chromatograms of all six samples with a reversion event than in those with normal heterozygosity on 3q. This latter analysis is only possible in the case of small deletions resulting in a shifted mutant allele because it is much more difficult to estimate the relative frequencies of two different alleles through a single point mutation. Furthermore, an overrepresentation of higher wild-type peaks was present in DNA isolated from liver tissue of proband III:12; however, this could not be confirmed by the mosaic homozygosity reporter. Therefore, we conclude that after careful interpretation of the sequencing chromatogram, one could suggest that a reversion event is present, but performing this analysis exclusively is not enough to draw firm conclusions.

All individuals with revertant somatic mosaicism in this study carried a small deletion in TERC. One explanation could be that these mutations are more deleterious than missense mutations and that after correction, they result in a greater selection advantage, which leads to revertant mosaicism. In the literature, however, we do not find evidence of such genotype-phenotype correlations. Time could be another factor influencing the chance that revertant mosaicism would occur in an individual. The mean age at molecular diagnosis of the six persons with reversion in our cohort is 53 years, whereas the persons without reversion have a mean age of 34 years.

The potential presence of somatic reversion in DC has important implications for clinical diagnostics. It is common practice that analysis of the DC-associated genes is performed on DNA isolated from peripheral blood cells of a proband suspected to have DC. In case no pathogenic mutation is found, an obvious conclusion can be that the phenotype in the family is caused by an aberration in a gene that has yet to be identified. On the basis of our findings, we recommend sequence analysis on DNA extracted from other cells, such as skin fibroblasts, particularly in individuals without bone-marrow failure.

Observing the reversion of a germline TERC mutation is also important for the development of future DC therapies. Currently, the only definitive treatment for bone-marrow failure in DC is hematopoietic stem cell transplantation (HSCT). The highly toxic conditioning regimens, which are thought to be a prerequisite for donor cell engraftment, are associated with significant adverse side effects, particularly on lung and liver tissue.^28–30 Dietz and colleagues reported that short-term survival after HSCT in DC-affected individuals is possible with a nonmyeloablative conditioning regimen.³¹ Our data show that, especially in the younger mononuclear cell lineages, reverted progenitor cells have a strong selective advantage over uncorrected cells. Therefore, isolation of autologous reverted stem cells could probably circumvent an allogeneic stem cell transplantation in a subset of individuals with DC.

In conclusion, we have observed somatic revertant mosaicism of a germline mutation in TERC in six individuals from four families affected by DC, indicating that reversion is a recurrent event in DC. This finding is important for improving diagnostic testing and understanding the variable phenotype of DC. In addition, we have developed a mosaic homozygosity reporter for the interpretation of SNP-array data in the search for genomic regions with low mosaic UPD.

Web Resources

The URLs for data presented herein are as follows:

• GenBank, http://www.ncbi.nlm.nih.gov/genbank/

• Online Mendelian Inheritance in Man (OMIM), http://www.omim.org/