Terminal osseous dysplasia with pigmentary defects and cardiomyopathy caused by a novel FLNA variant.

Terminal osseous dysplasia with pigmentary defects (TODPD), also known as digitocutaneous dysplasia, is one of the X-linked filaminopathies caused by a variety of FLNA-variants. TODPD is characterized by skeletal defects, skin fibromata and dysmorphic facial features. So far, only a single recurrent variant (c.5217G>A;p.Val1724_Thr1739del) in FLNA has found to be responsible for TODPD. We identified a novel c.5217+5G>C variant in FLNA in a female proband with skeletal defects, skin fibromata, interstitial lung disease, epilepsy, and restrictive cardiomyopathy. This variant causes mis-splicing of exon 31 predicting the production of a FLNA-protein with an in-frame-deletion of 16 residues identical to the miss-splicing-effect of the recurrent TODPD c.5217G>A variant. This mis-spliced transcript was explicitly detected in heart tissue, but was absent from blood, skin, and lung. X-inactivation analyses showed extreme skewing with almost complete inactivation of the mutated allele (>90%) in these tissues, except for heart. The mother of the proband, who also has fibromata and skeletal abnormalities, is also carrier of the FLNA-variant and was diagnosed with noncompaction cardiomyopathy after cardiac screening. No other relevant variants in cardiomyopathy-related genes were found. Here we describe a novel variant in FLNA (c.5217+5G>C) as the second pathogenic variant responsible for TODPD. Cardiomyopathy has not been described as a phenotypic feature of TODPD before.

INTRODUCTION

Terminal osseous dysplasia with pigmentary defects (TODPD, OMIM #300244) is one of the 10 X‐linked filaminopathies caused by pathogenic variants in FLNA (Breuning et al., 2000; Brunetti‐Pierri et al., 2010; Horii et al., 1998; Sun et al., 2010; Wade et al., 2020). FLNA is widely expressed and its protein structure includes a rod domain of 24 structurally homologus “repeats” many with specific functions as indicated by the phenotypic consequences of missense variants (Wade et al., 2020). Depending on the location of such variants within FLNA, a wide spectrum of disorders involving multiple organs with a clear phenotype–genotype correlation have been observed (Robertson, 2005). Loss‐of‐function variants are mainly associated with periventricular nodular heterotopia (PH) with secondary epilepsy, interstitial lung disease, connective tissue abnormalities and valvular cardiac anomalies (Wade et al., 2020). Gain‐of‐function variants, those that confer an alteration of some functions mediated by the protein, are associated with otopalatodigital spectrum disorders (OPD‐SD) characterized by a skeletal dysplasia, dysmorphic facial features and a variety of organ malformations. Within these OPD‐SD there is often clinical overlap, however, overlap with patients carrying FLNA loss‐of‐function variants is exceptional (Wade et al., 2020).

TODPD is part of the OPD‐spectrum disorders and is a male‐lethal disease with X‐linked dominant inheritance, characterized by skeletal dysplasia with brachydactyly, camptodactyly, limb deformities, scoliosis, pectus excavatum, and shortened and bowed long bones. Dysmorphic features include recurrent digital fibromata during infancy, multiple frenulae and pigmentary changes of the skin, hypertelorism, colobomata, and short stature (Bhabha et al., 2016; Sun et al., 2010). Only one single recurrent variant (c.5217G>A) in the FLNA‐gene is known to be responsible for TODPD (Sun et al., 2010). Here, we describe the clinical and molecular evaluation of two patients, a daughter (proband) and mother, with TODPD exhibiting an extended phenotype and a novel c.5217+5G>C variant in FLNA.

METHODS

Genetic workup

Genomic DNA was obtained from peripheral blood lymphocytes from both the proband and mother according to standard protocols. Genomic DNA isolation from skin, lung and heart tissue from the proband obtained at autopsy was performed with a standard protocol using a Maxwell RSC isolation machine and the Maxwell Tissue isolation kit (Cat no. AS1030). Targeted Sanger sequencing of the FLNA‐gene was performed according to standard protocols. Additionally, trio‐based whole‐exome sequencing (WES; 150 bp) with parents—focused on genes associated with cardiomyopathy and the HPO‐terms “cardiomyopathy” and “restrictive cardiomyopathy”—was performed on a NextSeq sequencer platform (Illumina) with a coverage of ×20 at greater than 95% of the sequences.

RNA analysis

Total RNA was obtained from peripheral blood using PAXgene tubes and the PAXgene blood RNA kit (Qiagen). In addition, total RNA was isolated from skin, lung, and heart tissue from the proband, obtained postmortem. Tissue was homogenized in Trizol reagent (Invitrogen) using a MagNa Lyser Green Beads (Roche). After homogenization Trizol was removed from the tube and used in the standard Trizol RNA extraction protocol. cDNA was produced by reverse transcription‐PCR with use of Superscript III kit (Invitrogen). PCRs were performed using FLNA specific primer sequences located on the transition of exon 29 and 30 (ACAGTGTCAATCGGAGGTCA) and in exon 32 (TGGGCGTAGGTGTACTGTG; Merck Sigma‐Aldrich). Sequencing of the cDNA was performed with the Brilliant Dye Terminator Cycle Sequencing system (Thermofisher Scientific) and Big Dye Terminator kit (Applied Biosystems) and analyzed using CodonCode Aligner version 8.0.2.

X‐chromosome inactivation analysis

The X‐chromosome inactivation pattern was tested with the HUMARA (human androgen receptor gene) assay as described by Allen et al. (1992) using the methylation‐sensitive restriction enzymes (HhaI and HpaII). The highly polymorphic CAG trinucleotide repeat in exon 1 of the AR gene (Xq12) was amplified with and without prior methylation‐sensitive restriction enzyme digestion. PCR products were run on an ABI3100 sequencer (Thermofisher Scientific) and results were analyzed using the software Genemapper v5 (Thermofisher Scientific). Degree of skewing was calculated by quantification as previously described (Lau et al., 1997). X‐inactivation was considered skewed when the ratio between the alleles was more than 90:10.

RESULTS

Case report

The proband was a 5‐year‐old girl, the first child of nonconsanguineous parents. The pregnancy was uncomplicated. She was born by sectio caesarea because of a narrow shaped pelvis of mother. Her birthweight was 2290 g at 39 weeks (<−2.5SD). Cognitive and motor milestones were normal: she sat at 8 months, walked at 11 months and was raised bilingual without problems. At the age of 6 months she presented for the first time with multiple digital fibromata, oral frenulae, and skeletal anomalies of the hands including bilateral hypoplasia of metacarpals 1 and 2, bilateral brachydactyly of the phalanges of digit II, and camptodactyly of the distal phalange of digit III (Figure 1(a–c)). She had additional dysmorphic features, including short stature (<−2.5SD and <−1.5SD of target height) and low weight (<−2.5SD); translucent skin, sparse hair, mild hypertelorism and a prominent forehead. Based on these clinical features TODPD was suspected, but could not be confirmed as the single recurrent variant for TODPD c.5217G>A in FLNA was not found with Sanger sequencing. However, a FLNA variant of unknown significance c.5217+5G>C was detected. At age 2 years, lumbar scoliosis was diagnosed (Figure 1(d)) a sign commonly reported in patients with TODPD. At age 3 years 10 months, she unexpectedly presented at the intensive care unit with severe heart failure due to restrictive cardiomyopathy with severe diastolic and mild systolic dysfunction and dilated atria and ventricles. Mild mitral valve insufficiency was present, without structural abnormalities of the valves. The aorta had a normal diameter. WES‐trio analyses with proband and parents—including filtering for genotypes that could mediate an autosomal recessive and dominant, de novo, and X‐linked disorder—yielded no other candidates for a cardiomyopathy phenotype. At age 4 years she developed respiratory distress with low CRP levels. CT of the chest revealed bilateral signs of interstitial lung disease, including groundglass opacities throughout the lungs, fibrotic changes, with bullae, bronchiectasis basal, and middle lobes, with destruction of lung parenchyma (Figure 1(e)). At age 5 years, she presented twice with tonic seizures which spontaneously stopped after few minutes and responded to levetiracetam treatment. Cerebral imaging to exclude PH was planned, but shortly after she deceased from heart failure. Autopsy revealed severe air retention of the basal lung lobes confirming emphysema or lung bullae, no signs of inflammation were observed. The heart weight was increased weight (95 g ref: 72–98 g) with bilateral dilated atria and ventricles and extensive fibroelastosis. Autopsy of the brain was not performed on request of the parents.

FIGURE 1

Phenotype of the proband. (a) Digital fibroma at age 2; (b) Camptodactyly of the distal phalange digit III (box); (c) Brachydactyly of the phalange of digit II (dashed box) and hypoplasia of metacarpal 1 and 2 age 2 (lined box); (d) X‐ray of the spine in standing pose at age 3 showing mild flattening of lumbar vertebrae with sinistroconvex lumbal scoliosis with a Cobb's curve of about 25°; (e) computed tomography after IV contrast of the chest (axial [left] and coronal reconstruction [right]), demonstrating diffuse ground glass opacities in all lobes (*). The middle lobe shows destruction of normal architecture, with formation of bullae and bronchiectasis (white arrow). In the lingula there is volume loss with bronchiectasis and formation of smaller bullae (dashed arrow). There is dilatation of the atria and milder dilatation of the left ventricle (^); (f) MRI‐heart (4 chamber cine T2 TRUFI) of mother at age 33 presenting noncompaction cardiomyopathy in end‐diastolic phase. Arrow indicating non compact myocardium of the left ventricle. (g) Pedigree of three generations of the family, indicating a TODPD‐phenotype (black left side), carriership of the c.5217+5G>C FLNA variant (black upper right quadrant) and cardiomyopathy (black lower right quadrant). Men are depicted with squares, women with circles

The two healthy brothers and the father of the proband did not show any signs of TODPD and echocardiography showed no signs of cardiomyopathy. Segregation analysis revealed absence of the c.5217+5G>C variant in FLNA in all three. The mother of the proband was carrier of the c.5217+5G>C FLNA variant and was diagnosed with noncompaction cardiomyopathy (noncompaction to compaction layer thickness ratio of 2.8:1 at end diastole; Figure 1(f)). She had multiple digital fibromata in infancy which were surgically removed and has multiple oral frenulae, unilateral camptodactyly of digit III, and short stature (1.43 m [<−4.4SD and <−3.3SD of target height]). The maternal grandmother did not carry the variant, suggesting that variant had occurred de novo in the mother of the proband (Figure 1(g)). It is unknown whether the maternal grandmother and mother's sisters and half‐sisters (same mother) have cardiomyopathy, as cardiac screening has not been performed. As far as could be ascertained, there are no other family members with clinical features of TODPD or cardiomyopathy.

Identification of the c.5217+5G>C p.Val1724_Thr1739del variant

Sanger sequencing of FLNA revealed the NM_001110556.2:c.5217+5G>C variant (Figure 2(a)). This noncoding FLNA‐variant has not been associated with TODPD or other human disease before and is absent in 179,736 reference alleles from gnomAD (Lek et al., 2016) and in ClinVar (Landrum et al., 2018). In silico prediction software (SpliceSiteFinder‐like, MaxEnsScan, NNSPLICE, and GeneSplicer) all predict the variant to functionally affect transcript splicing. RNA analyses by RT‐PCR of FLNA transcripts revealed normal splicing in blood. Only after the proband deceased, a RNA mis‐splicing effect (r.5170_5217del) could be observed in heart tissue: the variant results in an in‐frame deletion of 16 residues (p.Val1724_Thr1739del) in exon 31 (Figure 2(b)). This transcript was less abundant than the transcript derived from the paternal allele. This aberrantly spliced transcript could not be detected in blood, skin, and lung tissue using RT‐PCR.

FIGURE 2

Identification and characterization of the variant in FLNA. (a) Sanger sequencing of FLNA in proband (left) and mother (right) revealed the NM_001110556.1(FLNA):C.5217+5G>C variant. (b) Sequence traces of the RT‐PCR reactions from (top to bottom): A control (peripheral blood); skin tissue from the proband—also representative for lung tissue and peripheral blood; and heart tissue from the proband. Only in heart tissue from the proband aberrant splicing is present. After nucleotide position r.5169, the frame‐shifted and wildtype sequences are superimposed. Above the sequence trace the deduced sequence is shown. The sequence in black shows the normal transcript (part of exon 31) and the red sequence shows the aberrant transcript (part of exon 32 with the position of the reverse primer binding site underlined) resulting in a transcript that is 48 nucleotides shorter, NM_001110556.1(FLNA): R.5170_5217del

X‐chromosome inactivation analyses

X‐chromosome inactivation studies performed in peripheral blood DNA revealed an extremely skewed pattern of inactivation in the proband (>90%) and complete skewing in mother (100%). Identification and comparison of the X‐chromosomes by the repeat polymorphism in the AR‐gene revealed the mutated allele to be inactivated in the proband, explaining the absence of the mis‐spliced transcript in blood. Likewise, there was also an extremely skewed pattern of inactivation of the mutated X‐chromosome in skin and lung tissue of the proband. In heart tissue, however, there was random inactivation with approximately 60% expression of the mutated allele.

DISCUSSION

We describe the clinical and molecular evaluation of two patients, a daughter (proband) and her mother, with terminal osseous dysplasia with pigmentary defects and a novel c.5217+5G>C variant in FLNA. This intronic variant is located five nucleotides downstream from the position where the single recurrent c.5217G>A FLNA variant occurs that is responsible for all previously reported cases of TODPD. The new variant described here causes identical mis‐splicing in exon 31 leading to the deletion of 16 residues in the resulting FLNA protein (p.Val1724_Thr1739del). These data further support the contention that this specific deletion is the underlying mechanism leading to the TODPD phenotype. The mis‐splicing effect was undetectable in blood of the proband and her mother due to extreme skewing resulting in the inactivation of the maternal (mutated) X‐allele, a known phenomenon in TODPD (Sun et al., 2010). This mis‐splicing effect could however be demonstrated in affected heart tissue at postmortem. This case reveals a second causative FLNA‐variant associated with TODPD. Furthermore, it illustrates that splicing analysis in affected tissues can, if available, provide more insight into the pathogenicity of variants in FLNA.

We observed a clear variability in phenotypic severity in proband and her mother, the proband being more severely affected than her mother. This is possibly due to a difference in X‐inactivation between mother and daughter, where mother has a more extreme X‐inactivation pattern than the proband. The nearly complete inactivation of the mutated allele in most tissues could reflect of selection against cells expressing the mutant allele early in development. We were not able to test this in mother because affected tissue (heart) is not available.

In addition to the well‐known phenotypic features of TODPD, the proband and mother exhibited restrictive‐compaction and noncompaction cardiomyopathy respectively. Although cardiac valvular dysplasia is a known consequence of filaminopathy (OMIM #31440), cardiomyopathy has not been described as a phenotypic feature of TODPD before. Although an alternative pathogenic cause for the cardiomyopathy within this family cannot be fully excluded despite extensive DNA analysis, it is plausible that it could be part of an extended phenotypic spectrum of TODPD. Segregation analysis of the FLNA variant and cardiac evaluation was in accordance with this, although cardiac examination of the noncarrier maternal grandmother could not be performed. Furthermore, the mis‐spliced transcript was observed explicitly in heart tissue, where X‐inactivation was random and extreme X‐inactivation of the mutated allele was absent. Considering the function of FLNA as an important player in cross‐linking actin—essential for cell shape and motility—(Revenu et al., 2004), a disrupted FLNA protein might impact heart‐muscle structure and function. Therefore, cardiac screening might be considered in patients with TODPD caused by these two FLNA variants.

It is possible that the c.5217+5G>C variant not only leads to the FLNA gain‐of‐function transcript compatible with TODPD, but also to FLNA loss‐of‐function transcripts, since the proband exhibited phenotypic features that are also seen in patients with FLNA loss‐of‐function variants. Childhood interstitial lung disease is a recently recognized clinical consequence of loss‐of‐function FLNA variants (Sasaki et al., 2019). Although we could not confirm the mis‐splicing effect in the lung due to extreme skewing of X‐inactivation, it might be possible that specific interstitial cell types are principally affected. Epilepsy secondary to periventricular nodular heterotopia is also a common feature in loss‐of‐function FLNA disorders (Wade et al., 2020). PH was not confirmed in our proband as she died before a MRI‐scan could be taken and brain autopsy was not performed. If this hypothesis is true, the co‐occurrence of these loss‐of‐function phenotypic features and the TODPD phenotype might be explained by the capability of the c.5217+5G>C variant to produce more transcripts other than the r.5170_5217del transcript that individually confers loss‐of‐function and gain‐of‐function. These aberrant transcripts could have been be degraded by nonsense mediated mRNA decay (NMD) or have resulted in large PCR fragments or other nonproductive spliceforms that in aggregate result in a reduction in FLNA protein and were therefore not detected in our RT‐PCR reaction. The r.5170_5217del transcript was less abundant than the transcript derived from the paternal allele. Unfortunately we were not able to test this hypothesis as we do not possess cultures of tissue to inhibit NMD prior to RNA analysis. In general, we propose that, given the exceptional combination of phenotypic features within the OPD‐spectrum and the loss‐of‐function category, physicians should be aware of co‐occurring pathology in patients with pathogenic FLNA variants.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

L. Rumping: Contributed to the concept and design of the study, coordinated the study, wrote the manuscript, contributed to the clinical care of the proband and family. M.W. Wessels: Contributed to the concept and design of the study, contributed to the clinical care of the proband and family. A.V. Postma: Contributed to the acquisition and interpretation of the data. J. van Schuppen: Contributed to the acquisition and interpretation of the data. M.A. van Slegtenhorst: Contributed to the acquisition and interpretation of the data. J.J. Saris: Contributed to the acquisition and interpretation of the data. J.P. van Tintelen: Contributed to the acquisition and interpretation of the data, contributed to the clinical care of the proband and family. S.P. Robertson: Contributed to the concept and design of the study. S.M. Maas: Contributed to the concept and design of the study, contributed to the clinical care of the proband and family. R.H. Lekanne Deprez: Contributed to the concept and design of the study. All co‐authors critically revised the manuscript and provided final approval of the manuscript to be published.