Rothmund-Thomson syndrome (RTS, MIM#268400) is a rare autosomal recessive genodermatosis, characterized by wide clinical expressivity, primarily accounted by locus heterogeneity, but to be further deepened considering allele heterogeneity and genetic modifiers [1
]. Approximately 400 RTS patients are reported in the literature and generally all present as a hallmark sign the cutaneous erythematosus rash appearing within the first two years of life at sun-exposed areas, mainly on the face, then evolving into post-inflammatory chronic poikiloderma, a permanent lesion characterized by skin atrophy with hypo- and hyper-pigmented areas and telangiectasia [2
]. Other common features manifested in early childhood are growth delay, hyperkeratosis and sparse and thin hair, eyelashes and eyebrows, while premature aging is observed in adult age [2
]. However, only two-thirds of clinically diagnosed patients have biallelic alterations in RECQL4
gene (MIM#603780) which encodes a protein belonging to RecQ helicase family, represented in human by five members (RECQL1, BLM, WRN, RECQL4 and RECQL5) all with a fundamental role in safeguarding genome stability [3
]. The predominant subset of RECQL4
-mutated patients represents RTS type-II or Thomson-like entity, characterized by poikiloderma, skeletal defects and cancer predisposition (osteosarcoma (OS) and less frequently squamous cell carcinoma of the skin, hematological tumors and other malignancies) [4
]. The RTS type-I or Rothmund-type includes by definition patients negative to RECQL4
mutations who present like the RTS-II ones poikiloderma, ectodermal dysplasia and growth retardation but differ in displaying the hallmark sign of bilateral juvenile cataracts: the RTS-I genetic defect is so far unknown [1
Intriguingly, the most fearful sign of cancer predisposition, together with genome instability and premature aging, is shared by RTS-II, Bloom (MIM#210900) and Werner (MIM#277700) syndromes, all caused by defective functioning of RecQ helicases [6
]. In addition to RTS, RECQL4
biallelic alterations are also responsible for Baller-Gerold (BGS, MIM#218600) and RAPADILINO (MIM#266280) syndromes. All three RECQL4
-associated diseases are clinically characterized by growth delay, bone alterations and cancer predisposition, and the underlying biallelic pathogenic variants exert their effect by a loss-of-function mechanism [7
]. RTS is the most prevalent and hence best characterized entity, while RAPADILINO is a Finnish inheritance disorder so far reported in about 20 cases [8
] and BGS is even rarer with less than a dozen of cases confirmed by molecular diagnosis [9
gene, mapping to 8q24, encodes for a 1208 amino acids DNA helicase, a multi-domain protein with a multi-functional role in essential processes of DNA metabolism, including DNA replication [10
], DNA repair of double-strand breaks [11
], nucleotide excision repair [12
] and base excision repair [13
], telomere maintenance [14
], p53 transport to mitochondrion [15
] and mitochondrial DNA biogenesis [16
Like the other members of RecQ family, RECQL4 protein is characterized by the central highly conserved RecQ helicase domain (spanning amino acid residues 489–850 and encoded by exons 8–15), while the N- and the C-terminal are unique among the family members. The RECQL4 N-terminus, a Sld2-like domain (1–388 aa residues), essential for the initiation of DNA replication shows at least two nuclear targeting signals (encompassing amino acids 37–66 and 363–492) [3
]; a stretch of lysine residues (aa 376–386) subject to p300 acetylation [18
] and a mitochondrial localization sequence (first 84 amino acids) [15
]. At the C-terminal a R4ZBD domain (aa 836–1045) [19
], structurally different but functionally comparable to RQC domain [20
], and potential nuclear export signals are found [21
The involvement of different RECQL4 domains in carrying on the multiple functions needed for the safeguard of genome stability predicts a notable heterogeneity of mutant alleles depending on their intragenic location and their combination, considering the known prevalence of compound heterozygous versus homozygous genotypes reported for RTS patients [1
]. While pathogenic variants disrupting the helicase domain are classified as deleterious for RECQL4 as well as for WRN and BLM, variants in the N-terminus region of RECQL4, which for its role in DNA replication initiation and replication fork progression is indispensable for cell viability [22
], have likely a sub-lethal effect, consistent with the few mutations, never in the homozygous condition, identified in this region [7
]. Current poor knowledge of the specific roles of RECQL4 domains and subdomains precludes to rank different mutations according to the compromised domains and functions, but the increasing number of novel characterized pathogenic variants makes it worthwhile to detail the spectrum of the observed RTS phenotypes, especially as regards cancer outcome.
We herein report on five unrelated families with RTS-II affected members who highlight the huge variability of the clinical presentation depending on the strength of the underlying pathogenic variants and their homozygous or heterozygous combination. We have characterized the causative RECQL4 genetic lesions and have explored their pathogenic effect by in silico predictions and transcripts analyses to address mutation-phenotype correlation, especially in relation to cancer development.
The RTS-II patients herein described well summarize the remarkable clinical breadth of Rothmund-Thomson syndrome including mild as well as severe phenotypes at high risk for tumor development, primarily dependent on the pathogenic variants at the causative RECQL4 gene.
Patients #19 and #29 presenting with a phenotype mainly restricted to cutaneous alterations, are homozygous for the unreported pathogenic variants c.3054A>G, in exon 17, and c.3236G>T, in exon 18, hence sparing the helicase domain, encoded by RECQL4
exons 8-15. Both alterations lead to mis-splicing which is predicted by several softwares in the case of patient #19 and is demonstrated for patient #29 (Figure 2
a). In both cases mis-splicing impacts the RECQL4 C-terminal region (RQC) which comprises a zinc-binding domain and a winged helix domain involved in protein–protein interactions and regulation of the helicase activity, respectively [20
]. Identification of pathogenic variants affecting the RQC region, especially in homozygous state, is helpful to dissect the function of RECQL4 domains less explored than the key helicase domain, but involved in its regulation. We can infer by our young adult patients #19 and #29 presenting with an overall mild phenotype, that their splicing variants, though much less deleterious compared to truncating mutations, are able to reproduce all the main cutaneous findings of RTS, likely due to defective function of this RQC terminal region. Moreover, we also speculate that a putative residual helicase activity of the proteins encoded by the observed aberrant transcripts might contribute to the mild phenotype of patients #19 and #29.
Instead, a severe phenotype characterizes the affected individuals of families B and C. Osteosarcoma developed in two siblings of family B, found to be compound heterozygous for c.1048_1049del and c.1878+32_1878+55del deletions in exon 5 and intron 11, and in the index case of family C, carrying the recurrent complex c.[1568G>C;1573delT] and the novel c.3021_3022del alterations in exons 9 and 17.
Out of these pathogenic variants, two (c.[1568G>C;1573delT]; c.1878+32_1878+55del) directly impact and one (c.1048_1049del) leads to the lack of the helicase domain confirming the known assumption that patients with at least one RECQL4
truncating mutation disrupting the helicase domain have an increased risk to develop osteosarcoma at young age (mean age 11 y) [2
]. Moreover, these three alterations have been already reported in literature in RTS patients with tumor outcome, as surveyed in Table 2
Both pathogenic variants of affected members from family B modify the helicase coding sequence and should contribute to cancer development. The IVS11 c.1878+32_1878+55del deletion has been reported only once in a patient who developed OS in childhood, while out of the eight cases in the literature with c.1048_1049del deletion in exon 5 only one adult patient developed Hodgkin’s lymphoma at 35 years: the others may not have developed tumors due their young age (<13 y) at the analysis timepoint. It is worth noting that the c.1048_1049del allele is transcribed in an aberrant transcript with a premature stop codon escaping, at least partially, nonsense mediated decay (NMD), as we detected it by RT-PCR analysis, while the transcript of the c.1878+32_1878+55del allele leading to the internal in-frame deletion of exon 11 should not trigger NMD.
Moreover, the transcript carrying r.1048_1049del deletion might potentially be translated into a prematurely truncated protein that is predicted to lack the helicase domain and the lysine-rich region of nuclear targeting signal 2, which should cause cytoplasmic RECQL4 accumulation upon acetylation of the lysine residues (K376, K380, K382, K385, K386) by the p300 acetyltransferase [18
In family C, where only one of two affected siblings developed OS at young age, the helicase domain is perturbed by the paternal exon 9 c.[1568G>C;1573delT] alteration, which is definitely the most recurrent RECQL4
pathogenic variant [1
]. It has been associated to RTS and so far reported in 21 patients, seven of whom developed OS. Furthermore, c.[1568G>C;1573delT] has been found in heterozygous condition in one RAPADILINO patient [8
] and 4 BGS siblings [8
], comprising two fetuses and one child deceased at 3 years of age. The maternal c.3021_3022del in exon 17 is unreported and falls in gene region coding for RECQL4 C-terminus.
provides a general overview of the RECQL4
pathogenic variants detected to our knowledge in 35 patients who developed cancer presenting with all the three RECQL4
-associated diseases (28 RTS, 6 RAPADILINO and 1 BGS patients) [8
]. Out of 35 RECQL4
-mutated patients, 34 carried at least one pathogenic variant affecting the helicase domain, confirming the association between deleterious mutations in the RECQL4
gene and cancer predisposition first assessed by Wang in 2003 [35
Switching back to the RTS cancer families herein described, consistent with the location and predicted effect of the pathogenic variants the cancer predisposition appears more penetrant in family B where two siblings developed OS versus one in family C, as a second RTS affected member is cancer free at current age of 31 years. Nevertheless in both families OS, occurring at two different sites, was fatal at young age to the affected sibling. Apart cancer outcome, the phenotype of affected members is complete and severe in both families, also including rare RTS findings, although intrafamilial variability (for instance in gastrointestinal signs in family B siblings) is apparent and can be only ascribed to modifiers loci not shared by the siblings.
Cancer has also occurred in patient #38 who however developed rhabdomyosarcoma, an RTS-unusual tumor, and displays atypical skin RTS phenotype. Though she manifests the additional clinical features required to establish the RTS diagnosis as sparse/absent hair/eyelashes/eyebrows, hyperkeratosis, growth delay, diarrhea in infancy, bone, teeth and nail defects, she does not show the classic poikiloderma. Even if representing the hallmark cutaneous defect of RTS, poikiloderma is quite rare in the RECQL4
-related disorders RAPADILINO [36
] and Baller-Gerold [9
] making atypical skin changes in suspected RTS patients worth to be considered.
Patient #38 is also unsolved by molecular diagnosis as she is heterozygous carrier of a very rare c.2412_2420del pathogenic variant and is here included for sake of completeness, in keeping with the literature where at least 11 patients carrying only one heterozygous mutation have been described [26
]. Interestingly, the c.2412_2420del alteration of patient #38 falls in the region encoded by exon 14, which undergoes a physiological alternative splicing. An increased amount of the alternative in-frame transcript, encoding for a likely functional protein lacking 66 amino acids of the helicase domain, has been detected in two RTS siblings characterized by mild phenotype carrying the c.2272C>T mutation which is abolished by exon 14 alternative splicing [38
]. Unfortunately, patient #38 RNA was not available to investigate an expression change of the physiological alternative transcript r.2266_2463del.
In conclusion, deep phenotyping of a small cohort of patients with the ultra-rare Rothmund-Thomson type-II syndrome emphasizes the remarkably different clinical presentation with demand in some instances of patient management by an integrated team of professionals, including the oncologist. Characterization at DNA and RNA level of patients’ pathogenic variants is a valuable tool to predict their effect on overall phenotype and disease evolution, allowing for providing patients with oncological surveillance and personalized care of medical complications. Last, expanding the repertoire of RECQL4
gene variants affecting protein domains which a scarcely known role is instrumental to address functional studies [39
] aimed at precise mapping of intertwined functions of the multifunctional RECQL4 protein.
4. Materials and Methods
4.1. Biological Material from Affected and Healthy Individuals from RTS Families
Affected individuals of five unrelated families (A, B, C, D, E) with a clinical diagnosis of suspected or probable RTS were referred to our laboratory by clinical geneticists and dermatologists. All patients and their family members were enrolled in this study after obtainment of appropriate informed consent to genetic analysis and photos collection.
4.2. DNA Isolation and RECQL4 Mutational Analysis
Genomic DNA from peripheral blood of the index cases and their family members was isolated with Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) according to standard protocols.
The entire RECQL4
gene (NG_016430.1) (21 exons and 20 introns with the exception of IVS12 minisatellite) was amplified using GoTaq®
Flexi DNA polymerase (Promega) and primers listed in Table S2
PCR products were sequenced according to the manufacture’s protocol using Big Dye Terminator v.3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and primers listed in Table S2
on the ABI PRISM 3130 sequencer (Applied Biosystems). Electropherograms were analysed with ChromasPro software 1.7.4 (Technelysium Pty Ltd., South Brisbane, Australia) using the wild type sequence of the RECQL4
gene (GenBank: NG_016430.1) as reference.
4.3. Cell Cultures
EBV-transformed lymphoblastoid cell lines (LCLs) were established from peripheral blood lymphocytes of patient #42, her parents and five healthy controls.
LCLs were cultured in complete RPMI-1640 medium (EuroClone, Milano, Italy) supplemented with 10% fetal bovine serum (Lonza, Walkersville, MD, USA) and 1% penicillin, streptomycin and ampicillin in a 37 °C humidified incubator with 5% CO2.
4.4. RNA Isolation, RT-PCR and cDNA Analysis
TRI Reagent (Sigma, St Louis, MO, USA) was used to isolate total RNA of patients #29, #39, his affected brother and his parents, #42, healthy controls from white blood cells, and from LCLs of family C, according to manufacturer’s protocols.
After DNase I (RNase-free, New England Bio-Labs, Inc., Ipswich, MA, USA) treatment, 500 ng of total RNA were used to synthesize cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) with random hexamers. All samples were reverse transcribed in two independent experiments.
Different fragments of RECQL4
transcript were amplified using GoTaq® Flexi DNA polymerase (Promega) and primers listed in Table S3
Amplicons were run on 2% agarose gel, the bands were eluted with MinElute Gel Extraction Kit (Qiagen, Milano, Italy) and then sequenced as described above.
Nucleotide sequences were compared to the major RECQL4 transcript reference sequence [GenBank: NM_004260.3].
4.5. Bioinformatic Analyses
ExAC Browser of Broad Institute (http://exac.broadinstitute.org/gene/) and 1000 Genomes database (http://www.1000genomes.org/home
] were checked to assess the presence/absence of detected alterations in variations repositories.
4.6. High-Resolution CGH-Array
As regards patients #19 and #38, high-resolution array-based comparative genomic hybridization (CGH-array) analysis was performed, on 700 ng genomic blood DNA, using the high resolution SurePrint G3 Human CGH Microarray Kit 2 × 400 K in accordance with the manufacturer’s instructions. (Agilent Technologies, Palo Alto, CA, USA). Images were captured using the Agilent Feature Extraction 220.127.116.11 software and chromosomal profile was acquired using the ADM-2 algorithm provided by Agilent CytoGenomics v.18.104.22.168 software (Agilent Technologies).
Coordinates of Copy Number Variants (CNVs) referred to the Human Genome assembly GRCh37/hg19 (https://genome.ucsc.edu/
). CNV classification was performed according to the Database of Genomic Variants (http://projects.tcag.ca/variation/
, release March 2016) to exclude common polymorphic CNVs with a frequency >1% in healthy control subjects. The CNV classification by clinical relevance was performed according to the guidelines suggested by Miller [51
] and successively by the American College of Medical Genetics [52
SNP-array analysis was performed on samples of family #19 using the Human OmniExpress Exome-8 Bead Chip (Illumina Inc., San Diego, CA, USA) containing 960,919 loci derived from phases I, II and III of the International HapMap project. The array includes over 274,000 functional exonic markers, delivering unparalleled coverage of putative functional exonic variant selected from 12,000 individual exome and whole-genome sequences.
A total of 200 ng of gDNA (50 ng/µL) for each sample was processed according to Illumina’s Infinium HD Assay Super protocol. Normalization of raw image intensity data, genotype clustering and individual sample genotype calls were performed using Illumina’s Genome Studio software v2011.1 (cnvpartition 3.2.0). The CNVs were mapped to the human reference genome hg19 and annotated with UCSC RefGene. Allele detection and genotype calling were performed with Genome Studio software, B allele frequencies (BAFs) and log R ratios were exported as text files for PennCNV analysis.