We collected all probands referred for cardiovascular genetic evaluation at the same reference laboratory because HCM or RCM showing an FLNC-mRod2 variant, and who were available for study. We excluded all individuals with another causative variant to explain the phenotype. After the first evaluation, we selected those families with saw-tooth appearance at cardiovascular imaging for an in-depth clinical study.

We performed a comprehensive clinical assessment including a detailed evaluation of the patient's clinical history, pedigree, creatine kinase, 12-lead electrocardiogram, Holter monitoring, and expert echocardiography assessment. Left ventricular ejection fraction dysfunction was assessed by the Simpson biplane method.11 Diastolic function was evaluated according to current recommendations.12 Left ventricle (LV) myocardium appearance was analyzed with a special focus on hypertrabeculation, according to current diagnostic criteria for NCCM with 2 well-defined layers (thin compacted and thick noncompacted layers).13 Saw-tooth myocardium was identified as the presence of multiple deep projections and invaginations in a dense and compacted myocardium.

The study was approved by the local ethics committee. Informed written consent for inclusion in the study was obtained from all participants or, in the case of minors or deceased individuals, from first-degree family members.

Blood samples from probands were obtained for genetic sequencing using a phenotype-guided next-generation sequencing panel at the same institution. In probands with an overlap or nondefined phenotype, we used a next-generation sequencing panel including genes related or candidates for cardiac disease. Coding exons and intronic boundaries of 247 genes related to inherited cardiovascular diseases (table 1 of the supplementary data) were captured using a custom probe library.

For the pathogenicity assessment of genetic variants, we considered information such as frequency in public databases (eg, the Human Gene Mutation Database the Single Nucleotide Polymorphism Database, NHLBI GO Exome Sequencing Project and ClinVar or in the Genome Aggregation Database,14–16 its previous description in the literature and preservation among species. After identification of rare FLNC variants in the index patients, we performed a genetic cascade screening among all available relatives using Sanger DNA sequencing. For the de novo variants, paternity was confirmed when available. Finally, the logarithm of the odds score17 and American College of Medical Genetics and Genomics (ACMG)18 classification were evaluated. For detailed methods regarding the genetic analysis see supplementary data.

A complete histological, histochemical and immunohistochemical characterization of cardiac tissue belonging to 3 explanted hearts (patient from family F and family C, patients 2 and 3) was performed. Tissue samples were fixed for at least 48 hours in 10% neutral buffered formaldehyde, washed, dehydrated in ascending concentration of ethanol, cleared in xylol, and finally embedded in paraffin following a conventional protocol.19

Histological section of 5 μm thickness were obtained, dewaxed, hydrated and underwent the following steps: a) general histological examination was conducted with hematoxylin-eosin; b) the cardiomyocytes myofibrils were histochemical stained with Heidenhain hematoxylin; c) the organization and distribution of the main extracellular matrix (ECM) components were evaluated by Picrocirius collagen staining, Gomori's metal reduction technique for reticular fibers, Orcein histochemical staining for elastic fibers, and Alcian Blue staining for proteoglycans. In addition, collagens type I and IV (COL-I and COL-IV respectively) were identified by immunohistochemistry; d) the distribution pattern of FLNC and connexin-43 (Cx-43) were determined by immunohistochemistry. Technical details of the antibodies used are summarized in table 2 of the supplementary data.

In addition to the histological characterization, a semiquantitative analysis of the cell/collagen ECM area ratio was carried out in 3 different sections from 3 different tissue slides stained with the Picrocirius technique by using Image J software (National Institute of Health, United States) and following a previously described procedure.20 Furthermore, 20 points were selected in 3 sections from each condition stained with the Heidenhain hematoxylin method and the intensity of the histochemical reaction for myofibrils was determined. Finally, the cell width was calculated in 20 cells per image selected in 3 images from each condition.

To detect specific histological changes attributable to FLNC-mRod2 variants, we compared the histological findings with a nonpathological heart tissue sample (control), and with an FLNC-truncating variant (FLNC-tv) carrier sample from our historical cohort. The FLNC-tv case showed the recognized overlapping phenotype of dilated cardiomyopathy and arrhythmogenic cardiomyopathy.9,10

The wild-type plasmid pCMV6-FLNC (212462; OriGene Technologies, United States) was used as a template to insert the variants p.P2301L, p.E3224K and p.R2340W using the Quik Change Lightning Kit (Agilent Technologies, United States) in combination with appropriate oligonucleotides (table 3 of the supplementary data). The FLNC encoding complementary deoxyribonucleic acid (cDNA) of all plasmids was verified by Sanger sequencing (Macrogen, Netherlands). HT1080 (DSMZ, German Collection of Microorganisms and Cell Culture, ACC315, Germany) and H9C2 cells (ATCC, CRL-1446, United States) were cultured in μSlide chambers (ibidi GmbH, Germany) using Dulbecco's Modified Eagle Medium supplemented with 10% fetal calf serum and penicillin/streptomycin. Lipofectamin 3000 (Thermo Fisher Scientific, United States) was used for cell transfection according to the manufacturer's instructions. Twenty-four hours after transfection, cells were washed with phosphate buffered saline and were fixed for 10 minutes at room temperature using 4% Histofix (Carl-Roth, Germany). Immunocytochemistry analysis was performed as recently described using the TCS SP8 confocal microscope (Leica Microsystems, Germany).4

Statistical analyses were conducted using SPSS Statistics software, version 21.0. (IBM Corp, United States). Continuous variables are reported as mean value±standard deviation for each measurement. For quantification of histological variables, we performed the Mann-Whitney U test (nonparametric). All categorical variables are expressed as frequencies and percentages.

In total, we recruited 21 unrelated families from 7 reference centers who were genetically tested for HCM/RCM and found to carry rare FLNC-mRod2 variants (figure 1 of the supplementary data). A total of 11 families (52%) with 20 individuals (37 [23.7-52.7] years) showed 15 cases with a cardiac phenotype consisting of HCM or RCM with a marked and atypical left ventricular hypertrabeculation (LVHT) distorting LV with the presence of numerous cross bridging muscular projections. These protrusions of muscular bridges in the LV delineated deep crypts in a single, dense, and compacted myocardium (figure 1), which we have identified as “saw-tooth” appearance, different from the 2 characteristic layers of the NCCM (thin compacted and thick noncompacted layers). Furthermore, this singular LVHT was confirmed in 3 explanted hearts (figure 2).

Figure 1.

Cardiac magnetic resonance (top and center panels) and echocardiogram (lower panel) images showing the saw-tooth left ventricular hypertrabeculation appearance with muscular projections and deep crypts distorting the normal left ventricular architecture. A, patient III.1 from family C; B, patient II.2 from family A; C, patient from family E.

(0.24MB).

Figure 2.

Echocardiogram images (A) and macroscopic images of explanted heart (B) from the index patient of family C, aged 24 and 36 years, respectively. Echocardiogram images showed left ventricular muscular protrusions. Pathological images showed the atypical and marked hypertrabeculation with profound recesses.

(0.24MB).

Among the total cohort of 20 individuals, median age at diagnosis was 37 [23.7-52.7] years and 12 (60%) were male. HCM was more frequent (13; 65%) than RCM (4; 20%), and 15 individuals (75%) showed a saw-tooth appearance. A total of 16 individuals (80%) showed marked diastolic dysfunction (grade II or III) and 10 (50%) had atrial fibrillation, sometimes in adolescence or early adulthood.

Patients HCM showed an increased interventricular septum (14.3±7 mm), a dilated left atrium (45.6±10.4 mm) and an initially preserved left ventricular ejection fraction (58.5±10.3%). A total of 2 individuals from family A and 2 from family C had bicuspid aortic valve. In addition, 9 patients underwent cardiac magnetic resonance with late gadolinium enhancement assessment, but none of them showed late gadolinium enhancement scar. All of them exhibited an abnormal electrocardiogram: 5 showed T wave inversion in precordial leads and 8 in inferior and precordial leads, 5 had right bundle branch block and 2 showed an unspecific intraventricular conduction delay. None of the patients had low voltages or atrioventricular block. Holter monitoring did not identify increased premature ventricular beats but 1 did identify nonsustained ventricular tachycardia. None of them had symptoms, signs, or a history of skeletal myopathy (mean creatine kinase level 137±58 U/L). Cascade genetic screening showed mild cosegregation of 6 variants with the phenotype (families A, C, D, H, J, K; figure 3), while the remaining 5 cases exhibited a de novo presentation (table 1, figure 2 of the supplementary data, table 4 of the supplementary data). Noncarriers showed manifestations of cardiomyopathy.

Figure 3.

Pedigrees of families with mild cosegregation. Squares indicate males, circles indicate females, slashes indicate deceased individuals, black shading indicates cardiomyopathy phenotype, black stripe indicates silent carrier, N indicates absence of phenotype, (?) indicates not available for evaluation. The arrows + P indicate the proband. Heterozygous carriers (E1 −/+) and noncarriers (E1 −/−). ICD, implantable cardioverter-defibrillator; HF, heart failure; HTx, heart transplant; VT, sustained ventricular tachycardia.

(0.34MB).

The remaining 10 families identified with FLNC-mRod2 variants without the saw-tooth appearance consisted of 4 obstructive HCM (1 with LVHT but no saw-tooth trait) patients, 1 HCM on postmortem analysis, 1 apical HCM, 1 RCM and 3 individuals without cardiomyopathy.

After a median follow-up of 6.49 [3.8-21.33] years after phenotype diagnosis, 11 (55%) patients showed symptomatic heart failure (HF) New York Heart Association (NYHA)≥II, with 3 of them requiring 3 a heart transplant due to advanced HF in their thirties. Three patients (15%) died from HF. Furthermore, 4 (26.6%) patients had syncope during follow-up and 2 had sustained ventricular tachycardia (SVT) and received an implantable cardioverter-defibrillator (ICD).

Figure 4 represents the location of the 9 rare variants identified in the FLNC ROD2 domain. All the missense variants observed in this cohort were located in the ROD2 domain, specifically in 18 to 21 Ig-like domains. All identified variants were absent in control population databases and bioinformatics predictors showed that they were likely deleterious. ACMG classification identified 4 variants as likely pathogenic and 5 as variants of unknown significance (tables 5 and 6 of the supplementary data). A second relevant-causative genetic variant was not identified in any of the probands, and paternity was confirmed in 4 of the de novo cases. No CNVs were detected in the selected cohort. In all these cases, the quality of the sample was adequate to carry out this analysis.

Figure 4.

Representative illustration of the FLNC protein with the localization of included FLNC variants.

(0.28MB).

Hematoxylin-eosin staining revealed moderately larger cardiac cells in FLNC-tv heart samples compared with FLNC-mRod2 and controls (figure 5A). It also revealed a lower density of myofibrils and a slight decrease in intercellular space in FLNC-mRod2-affected tissues, resulting in a more compact tissue organization pattern in these specific samples (figure 5B). These results were confirmed by quantitative histological analyses (figure 3 of the supplementary data). The immunohistochemical analysis of FLNC confirmed the intracellular localization of this protein in all samples, but in FLNC-tv-affected samples, this protein was irregularly distributed and totally absent at the intercalated discs (ID) (figure 5C). In contrast, in FLNC-mRod2-affected tissues, it was present within the contractile apparatus, such as in the control group, but it was particularly positive at the ID level.

Figure 5.

Histological pattern, myofibril histochemistry, FLNC immunohistochemistry and ECM remodeling analyses in FLNC genetically affected heart tissues. A, hematoxylin-eosin staining of control and genetically compromised heart tissue samples (FLNC-tv and FLNC-mRod2) showing the increase in the intercellular space in FLNC-tv and compaction in FLNC-mRod2. B, Heideinhain's hematoxylin for myofibrils demonstrating the minor density in FLNC-mRod2. C, Intracellular distribution of FLNC as determined by immunohistochemistry. Note the absence of FLNC in the intercalated disc of the FLNC-tv sample. D, Histochemical identification of fibrillar collagens by Picrosirius staining (in red) with an increase in these fibers in genetically affected samples. Immunohistochemical identification of collagen type I (E) and type IV (F). Scale bar: 20μm. CTR, control; FLNC-mRod2, missense variant in FLNC-Rod2 domain; FLNC-tv, filamin C truncating variant.

(1.32MB).

On the other hand, the—remodeling analysis revealed fibrotic changes in genetically affected samples vs the control (figure 5D), showing a clear increase in fibrillar collagen fibers. In FLNC-mRod2-affected samples, we observed this increase in collagen fibers around the cardiomyocytes, exhibiting a well-defined and compacted pattern, different to that observed for FLNC-tv (abundant and irregularly organized) (figure 5E-F). Finally, the fibrotic process was semiquantitatively confirmed, showing that the increased collagen content reached 49.2±7.9% in FLNC-tv followed by 29±2.5% in FLNC-mRod2 genetically affected samples, both being considerably higher than the 26.5±2.2% collagen content observed in the control (figure 3 of the supplementary data). Moreover, the remodeling of the collagen network—fibrotic response—was accompanied by an alteration of the reticular fibers and acid proteoglycans content, intensity, and pattern (figure 4 of the supplementary data). All these histological findings in FLNC-mRod2-affected samples were consistent in the 3 samples.

HT1080 and H9C2 cells were transfected with expression plasmids for wild-type and mutant FLNC (p.P2301L, p.E2334K and p.R2340W). Confocal analyses revealed a comparable cellular localization of wild-type and mutant FLNC in transfected cells (figure 6 and figure 5 of the supplementary data) indicating that pathogenic protein aggregation was unlikely for the described missense variants.

Figure 6.

H9C2 cells were transfected with expression plasmids for wild-type and mutant filamin C (p.P2301L, p.E2334K and p.R2340W). c-Myc tagged filamin C is shown in red. F-actin (green) was contained using phalloidin conjugated with Alexa-488. Representative cell images are shown. Scale bars = 10 μm.

(0.98MB).

The main limitation of this study is the limited number of patients evaluated and its retrospective design. A total of 5 of the 9 rare variants are considered variants of unknown significance, thus limiting the conclusions that can be drawn. For histology, given the nature of the samples, we included only a limited number of independent heart samples affected by this particular genetic condition, limiting the possibility of quantitative and statistical analyses. Cardiac magnetic resonance tissue characterization is limited by the retrospective nature of the study. Furthermore, HT1080 cells may not reproduce human cardiomyocyte physiology.