MIB2 variants altering NOTCH signalling result in left ventricle hypertrabeculation/non-compaction and are associated with Ménétrier-like gastropathy

Abstract

We performed whole exome sequencing in individuals from a family with autosomal dominant gastropathy resembling Ménétrier disease, a premalignant gastric disorder with epithelial hyperplasia and enhanced EGFR signalling. Ménétrier disease is believed to be an acquired disorder, but its aetiology is unknown. In affected members, we found a missense p.V742G variant in MIB2, a gene regulating NOTCH signalling that has not been previously linked to human diseases. The variant segregated with the disease in the pedigree, affected a highly conserved amino acid residue, and was predicted to be deleterious although it was found with a low frequency in control individuals. The purified protein carrying the p.V742G variant showed reduced ubiquitination activity in vitro and white blood cells from affected individuals exhibited significant reductions of HES1 and NOTCH3 expression reflecting alteration of NOTCH signalling. Because mutations of MIB1, the homolog of MIB2, have been found in patients with left ventricle non-compaction (LVNC), we investigated members of our family with Ménétrier-like disease for this cardiac abnormality. Asymptomatic left ventricular hypertrabeculation, the mildest end of the LVNC spectrum, was detected in two members carrying the MIB2 variant. Finally, we identified an additional MIB2 variant (p.V984L) affecting protein stability in an unrelated isolated case with LVNC. Expression of both MIB2 variants affected NOTCH signalling, proliferation and apoptosis in primary rat cardiomyocytes.

In conclusion, we report the first example of left ventricular hypertrabeculation/LVNC with germline MIB2 variants resulting in altered NOTCH signalling that might be associated with a gastropathy clinically overlapping with Ménétrier disease.

Introduction

Whole exome sequencing (WES) is a powerful tool in identifying disease-associated variants in a wide variety of genetic diseases (1). In this study, we performed WES on a family of Italian descent with multiple members presenting with an autosomal dominant gastropathy with Ménétrier-like features (2). Ménétrier disease is a premalignant disorder of the stomach presenting with gross cerebriform hypertrophy of gastric mucosa confined to the body and fundus with sparing of the antrum (3,4). In affected patients epidermal growth factor receptor (EGFR) signalling is enhanced due to local overproduction of transforming growth factor-α (TGF-α) and this results in expansion of surface mucous cells in the body and fundus of the stomach (3,5–8). Therefore, patients with Ménétrier disease can be treated with Cetuximab, a monoclonal antibody blocking EGFR signalling (9). Nevertheless, the underlying molecular defect(s) that results in upregulation of TGF-α in Ménétrier disease is still unknown (3).

Results

Familial occurrence of Ménétrier disease has been rarely reported (4,10,11). Unraveling the molecular defects underlying rare familial forms might provide hints to understand the molecular basis of Ménétrier disease. WES was performed on four affected members of the family (II.2, II.4, IV.2 and IV.5, Family A; Fig. 1A). Like classic Ménétrier syndrome, the gastric lesions in the family with Ménétrier-like syndrome (2) showed EGFR overexpression (Fig. 1B). Given the autosomal dominant inheritance, we focused on single nucleotide variations and small insertions and deletions (indels) in the heterozygous state. We selected only rare heterozygous variants affecting all four affected members that fell within coding regions and were not predicted to generate synonymous amino acid changes. Eight single nucleotide substitutions passed this filtering step (Supplementary Material, Table S1). Among these, six were described in Exome Aggregation Consortium (ExAC) database (available at: http://exac.broadinstitute.org; accessed May 02, 2016) with frequency >0.01 and were discarded. The two remaining variants in nucleoporin-98 (NUP98) and mind-bomb 2 (MIB2) were confirmed by Sanger sequencing and analysed for segregation with the phenotype among affected members. We observed segregation in all ten affected members with available DNA only for the c.2225T > G (p.V742G) heterozygous variant in the MIB2 gene (Fig. 1A). Subject III.3 was an obligate carrier having transmitted the disease to his child but was asymptomatic, although he never underwent gastric endoscopy and thus, stomach alterations cannot be excluded. Gastric endoscopy failed to show stomach abnormalities in asymptomatic subject III.4 who did not carry the MIB2 variant. The p.V742G variant had an allele frequency of 0.05% in European non-Finnish population and a cumulative frequency of 0.15% in control populations of the ExAC database (Supplementary Material, Table S2). This variant was detected in homozygous state in two Asian individuals in ExAC. Nevertheless, the variant changes a highly conserved valine into glycine at the amino acid position 742 that is predicted to be deleterious by multiple missense prediction programs (Supplementary Material, Table S3). The Val742 is positioned towards the C-terminus in a highly conserved ankyrin repeat (a.a. 696–780; Pfam ID: PF12796; E-value = 2.1e-10) suggesting that the p.V742G variant might result in structural and/or functional effects disturbing the hydrophobic packing between α helices, which destabilize the helix-turn-helix fold of the ankyrin repeat (Fig. 2A) (12,13). Clustering of disease-related variants in regions of the three-dimensional (3D) structure is used as supporting evidence for the pathogenicity (14) and known cancer-related variants, such as p. A752V (liver) and p. D775N (lung), are also positioned in this ankyrin repeat and nearby the Val742 according to the Structure-PPi method (Fig. 2A) (15).

Figure 1.

(A) Pedigree of the family A with inherited gastropathy and/or left ventricular hypertrabeculation. (B) EGFR staining in gastric specimens from two affected individuals from the pedigree A and a control (Ctrl). (C) Pedigree B with an isolated case of LVNC (legend in panel A refers to both families A and B). (D) Cardiac ultrasound and MRI from the one subject of the pedigree A harboring the p.V742G variants (IV.2) and from the isolated case with the p.V984L variant (II.1) of pedigree B.

Figure 2.

(A and B) Mapping of MIB2 variants (in red) onto the 3D models of ankyrin-repeat and zinc finger RING-type motif. The positions of known cancer-related variants p.A752V (liver), p.D775N (lung) and p.I979M (breast) (in blue) and ligand-binding cysteine residues (in black) are also shown. (C) Western blotting with antibody against the FLAG tag after incubation of purified wild-type and mutated MIB2 proteins with E1, E2 and FLAG-ubiquitin proteins. The quantitation of the intensities of protein bands corresponding to ubiquitinated FLAG normalized for MIB2 protein band are shown in (D). Data are expressed as mean ± SD (n = 3 per group; t-test: *P < 0.05 vs WT). (E) Representative western blotting of HEK293 cells transfected with MIB2^WT, MIB2^V742G, MIB2^V984L or control vector (3×FLAG) and incubated with cycloheximide (CHX) up to 4 h or treated with proteasome inhibitor MG132 for 4 h showed decreased stability of MIB2^V984L mutant. (F) Relative quantification of the amounts of MIB2^WT, MIB2^V742G and MIB2^V984L proteins. In the graph, the amount of MIB2 was set at 100% in samples maintained 4 h in the presence of proteasome inhibitor MG132. Results are expressed as mean ± SD (n = 3 per group; two-way ANOVA, *P < 0.05).

Because MIB1 mutations were recently found to be responsible for left ventricle non-compaction (LVNC) (16), we evaluated 4 family members for this cardiac defect by heart ultrasound, even though none of them had cardiac complaints. The cardiac ultrasound followed by cardiac Magnetic Resonance Imaging (MRI) revealed in two (III.6 and IV.2) of four (III.2, III.6, IV.2, IV.5) family members carrying the MIB2 variant enhanced apical trabeculation of the left ventricle, which is consistent with an incomplete phenotype of LVNC (Fig. 1A and D). None of them had a non-compacted to compacted (NC/C) ratio >2 at cardiac MRI (III.6 = 1.4:1; IV.2 = 1.2:1), they had normal LV size (left ventricular end diastolic diameter 54mm and 42mm for III.6 and IV.2, respectively) and septal and posterior wall thicknesses (diastolic septal thickness 10mm and 8mm for III.6 and IV.2, respectively; diastolic posterior wall thickness 9mm and 8mm, respectively). They also showed preserved LV ejection fraction (EF) (III.6 = 60%; IV.2 = 70%) and diastolic function (preserved early diastolic velocity/late diastolic velocity ratio in both individuals), with no signs of concomitant hypertrophic and/or dilated cardiomyopathy. They were both asymptomatic for palpitations and syncope episodes.

We next evaluated a small group of subjects (n = 7) with cardiomyopathy and left ventricular hypertrabeculation or isolated LVNC who previously underwent extensive clinical testing for a panel of 38 cardiomyopathy genes (see Methods) by next generation sequencing. We identified a c.2950G > C (p.V984L) variant in MIB2 in a 47-year-old African American woman with typical LVNC who presented with symptoms of paresthesia and chest pain (Family B; Fig. 1C). Echocardiogram revealed an EF of 48% and mild left atrium (LA) and right atrium (RA) dilation along with hypertrabeculated LV consistent with the diagnosis of LVNC (Fig. 1D). The cardiac MRI identified increased left ventricular trabeculation with deep intratrabecular recesses of the apex extending to the mid LV, with a NC/C ratio at the LV apex of 5:1, thus confirming the diagnosis of LVNC (Fig. 1D). In addition, the systolic function was found to be only mildly reduced with an EF of 46%, while the LA and RA appeared normal in size. There were no apparent valve abnormalities and there was no delayed gadolinium enhancement. The electrocardiogram (ECG) showed sinus rhythm with premature supraventricular complexes, nonspecific T wave abnormality with a QTc interval of 450 ms (Supplementary Material, Fig. S1). The affected individual did not have gastrointestinal complaints. She belongs to a family with several siblings and she is the only member who currently has a diagnosis of LVNC. Both maternal and paternal ancestry is African American but consanguinity was denied. The family history was negative for diagnoses of hypertrophic cardiomyopathy and/or non-ischemic dilated cardiomyopathy, sudden death or implantable cardioverter defibrillator placement although both parents suffered a myocardial infarction consistent with coronary artery disease and some of her siblings had unspecified cardiac symptoms. Echocardiogram in the mother did not identify any major LV abnormalities. Both parents and other family members were unavailable for genetic testing, thus segregation analysis was not possible.

The p.V984L variant is not present in Exome Variant Server (EVS) (enriched for African Americans) while it is reported in 1 out 102,118 alleles (9.8×10⁻⁶) in ExAC (Supplementary Material, Table S2). The variant changes a valine into leucine at the conserved amino acid position 984 but is not predicted to be deleterious by missense prediction programs (Supplementary Material, Table S3). Nevertheless, the variant is predicted to affect the 3D stability of the protein (Supplementary Material, Table S3). The Val984 is located in the linker region between two zinc-finger RING type motifs (a.a. 945–1064; Pfam ID: PF13920; E-value = 1.6e-06) and in the vicinity of the ligand-binding cysteine residues and also of the breast cancer variant p.I979M (Fig. 2B). Moreover, Val984 is conserved in the homolog MIB1 protein and the same region of MIB1 was previously found to be affected by a mutation responsible for LVNC (16).

MIB2 was previously shown to undergo auto-ubiquitination (17). To investigate whether the MIB2 variants result in impaired protein function, we performed an in vitro functional ubiquitination assay (Fig. 2C and D). Purified recombinant wild-type MIB2^WT, and mutant MIB2^V742G and MIB2^V984L proteins were incubated with E1, E2 and FLAG-ubiquitin proteins to measure auto-ubiquitination activity by western blotting. The protein band detected by the antibody against the FLAG tag was reduced for MIB2^V742G compared to the wild-type protein. Therefore, MIB2^V742G is less auto-ubiquitinated compared to the wild-type protein, suggesting a reduced ubiquitin ligase activity. In contrast, MIB2^V984L was not found to have reduced auto-ubiquitination activity (Fig. 2C and D). Nevertheless, the levels of this mutant protein were significantly reduced by western blotting performed with two different antibodies recognizing different epitopes in transfected HeLa cells (Supplementary Material, Fig. S2A) even though no significant differences were observed in MIB2 mRNA levels and GFP protein levels used as an internal control of transfection efficiency (Supplementary Material, Fig. S2A and B). To investigate the degradation rate of MIB2 mutants, HEK293 cells were first transfected with FLAG-tagged MIB2^WT, MIB2^V742G or MIB2^V984L and then treated with cycloheximide (CHX) to block protein synthesis; MG132 was used to inhibit proteasome degradation. Western blot analysis for FLAG-MIB2 showed significantly reduced degradation rate for MIB2^V742G compared to the wild-type protein while the MIB2^V984L could only be detected after MG132 treatment, suggesting protein instability and rapid proteasome degradation (Fig. 2E and F).

Previous studies have shown that both MIB1 and MIB2 regulate NOTCH activity (18,19). Therefore, we evaluated NOTCH signalling genes in RNA extracted from white blood cells of two affected members from Family A (III.6 and IV.5) that were compared to age- and gender-matched controls. Significant reductions of HES1 and NOTCH3 expression were detected in both individuals while NRARP expression was significantly down-regulated in subject III.6 compared to corresponding controls and showed a trend in reduction in subject IV.5 (Fig. 3). Taken together, these data show decreased NOTCH signalling in white blood cells carrying the p.V742G variant. Blood sample for RNA analysis was not available for the isolated case of LVNC carrying the p.V984L variant (Family B, II.1; Fig. 1C).

Figure 3.

Expression of NOTCH signalling genes in blood leukocytes of two affected members from the pedigree (III.6 and IV.5) compared to age- and gender-matched controls (n = 4). Mean ± SEM is shown (one-sample t-test: *P < 0.05 vs controls). The SEM for each subject is calculated on technical triplicates.

To investigate the consequence of MIB2 variants on NOTCH signalling, neonatal rat cardiomyocytes were infected with adeno-associated viral vectors (rAAV6) expressing MIB2^WT, MIB2^V742G or MIB2^V984L and were also transfected with a plasmid bearing the luciferase gene under the control of an artificial NOTCH1 responsive element containing four repeats of the C Promoter-binding Factor 1 (CBF1) binding sites (20). AAV6 vectors have shown robust and persistent transgene expression in cardiomyocytes (21). As control of NOTCH-mediated transactivation, cardiomyocytes were grown in the presence of the γ-secretase inhibitor DAPT [(N-[N-(3,5-difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester)], which specifically inhibits NOTCH receptor signalling (22). Transcriptional activation of NOTCH regulatory elements by either MIB2^V742G or MIB2^V984L was reduced compared to MIB2^WT (Fig. 4A). Cardiomyocytes transduced with the vector encoding MIB2^V742G also showed decreased vitality compared to MIB2^WT control measured by luminescence through the ATPLite assay; only a trend in reduction of vitality was observed in cells transduced with the MIB2^V984L (Fig. 4B). To determine whether MIB2 mutants affect cardiomyocyte proliferation, incorporation of thymidine analogue 5-ethynyl-2’-deoxyuridine (EdU) was evaluated in cardiomyocytes transduced with rAAV6 encoding wild-type or MIB2 variants. Transduced cardiomyocytes showed similar mRNA levels of MIB2 following infection with rAAV6 expressing MIB2^WT, MIB2^V742G or MIB2^V984L (Supplementary Material, Fig. S3). MIB2^V742G variant decreased the number of proliferating rat cardiomyocytes compared to MIB2^WT (Fig. 4C and D). A trend in reduction of proliferating cells was detected in cells transduced with MIB2^V984L (Fig. 4C and D).

Figure 4.

(A) NOTCH signalling activity measured by 4XCBF1-Luc reporter assay. Primary rat cardiomyocytes were transfected with 4XCBF1 firefly luciferase reporter and renilla-luciferase reporter for normalization. DAPT treatment reduced 4XCBF1-luciferase reporter activation and was used as a specific transactivation control. Mean ± SD is shown (n = 8 per group; two-way ANOVA and Tukey’s post-hoc: *P < 0.05; ***P < 0.005 versus MIB2^WT). ATP-Lite (B) and proliferation (C,D) assays on freshly isolated rat neonatal cardiomyocytes transduced with rAAV6 vectors encoding MIB2^WT, MIB2^V742G, or MIB2^V984L. Green, α-actinin; red, EdU; blue, DNA (DAPI). (D) Quantification of the total number of α-actinin+/EdU+ cells (proliferating cardiomyocytes) after 4 days of culture. Mean ± SD is shown (n = 8 per group; one-way ANOVA and Tukey’s post-hoc: * P < 0.05; **P < 0.01 versus MIB2^WT).

Discussion

We report herein the first example of an inherited disorder with left ventricular hypertrabeculation/LVNC due to MIB2 mutations altering NOTCH signalling. Individuals of a large pedigree analysed herein harbouring the p.V742G variant in MIB2 gene also showed clinical features of a gastropathy that closely resembles Ménétrier disease (2). Within the affected family members with the Ménétrier-like disease, at least one individual (III.3) carrying the p.V742G variant was free of gastric symptoms suggesting incomplete penetrance of the stomach disease. Variable expressivity of the gastric phenotype was also observed (2). Nevertheless, the low frequency of the p.V742G variant in the control population and the rare occurrence of Ménétrier-like syndrome suggest that the MIB2 variant is not the only cause for the gastropathy. In addition to MIB2 mutations, other genes or environmental factors could be needed for the clinical expression of the gastropathy. Ménétrier disease has been previously linked to infection with cytomegalovirus (CMV), particularly in children (23), and Helicobacter pylori (3). Interestingly, both CMV and Helicobacter pylori have been found to downregulate NOTCH expression (24,25). However, it should be noted that no evidence of infections by CMV and Helicobacter pylori was detected in the index cases of the family (2). Nevertheless, additional yet unidentified genetic, epigenetic (26) or environmental factors might affect NOTCH signalling resulting in expression of the disease.

NOTCH signalling is critical for cell-fate determination and is involved in maintaining the balance between cell proliferation and differentiation (27). The gastric epithelium is continuously regenerated by gastric stem cells that give rise to various cell types including parietal cells, chief cells, surface mucous cells mucous neck cells and enteroendocrine cells. Progenitor cells are found at the isthmus of the gastric glands. From this region, precursor cells give rise to surface mucous cells that move to the luminal surface of the glands, parietal cells that move to the base of the glands, and mucous neck cells which also migrate down and further differentiate into chief cells. NOTCH signalling is required to maintain the gastric stem cell compartment (28). Therefore, it is not surprising that alteration of NOTCH signalling in the stomach might lead to gastric hypertrophy. Cross-talk between EGFR and NOTCH has been previously reported (29) and could be responsible for the increased EGFR expression observed in the patients. Although some differences exist between Ménétrier-like observed in the family and classic Ménétrier disease (2), we speculate that NOTCH pathway might also be involved in the pathogenesis of Ménétrier syndrome.

Several genes have been linked to LVNC and they encode proteins involved in cellular energy, muscle development, ion channel formation, or components of muscle filaments (30–32). Altogether, these defects show that LVNC is caused by at least two final common pathways: a primary pathway (such as the sarcomere) and a developmental pathway (such as the NOTCH pathway) (33,34). Evidence from mouse developmental studies show that hypertrabeculation results from altered regulation of cell proliferation, differentiation and maturation during the formation of the ventricular wall, particularly if the NOTCH signalling pathway is affected (35). Inactivation of NOTCH pathway genes such as Notch1, Rbpj, or Fkbp1a in mice results in disruption of trabeculation of ventricular chamber and spongy myocardial wall (36–38). Further support to the role of the NOTCH pathway in LVNC was confirmed in humans with autosomal dominant LVNC and germline mutations in MIB1 gene (16). Similarly to MIB2 variants reported herein, two MIB1 mutations previously found in patients with LVNC also affect the ankyrin repeat and the zinc-finger RING type motif (16), thus suggesting the important roles of these domains in protein function or stability. Interestingly, MIB2 gene mapping to the 1p36 region is deleted in patients with 1p36 deletion syndrome (39) who are frequently found to have LVNC, which tends to improve over time (40,41).

The diagnosis of LVNC relies on non-invasive imaging studies, usually echocardiography and cardiac MRI. However, the diagnostic criteria for both methods are debated and are currently based on measurements of the ratio of the thickness of the non-compacted layer to that of the compacted layer (42). Such criteria were present in two subjects with left ventricular hypertrabeculation/LVNC carrying the p.V742G variant in MIB2 gene. Healthy individuals and competitive athletes may fulfill current imaging criteria for diagnosis of LVNC presenting with normal left ventricular size and preserved systolic and diastolic functions. This has been considered a benign variant (37,43). Therefore, it has been proposed that the extent of myocardial compaction may be a continuous spectrum within the population ranging from asymptomatic individuals to end stage heart failure. Based on the lack of cardiac symptoms or altered heart function in the individuals with p.V742G variant and the very low frequency of MIB2 variant in the population, we argue that at least a subgroup of individuals with benign LVNC might harbour MIB2 mutations. It remains to be ascertained whether and which additional hits, either genetic or environmental, are needed for the development of overt cardiac disease. The identification and evaluation of a larger number of individuals harbouring MIB2 mutations will allow understanding the full spectrum of cardiac and gastric phenotypes associated with variants in this gene.

In conclusion, we found that germline MIB2 variants altering NOTCH signalling result in a spectrum of left ventricular hypertrabeculation/LVNC and might be associated with hypertrophic gastropathy clinically overlapping with Ménétrier disease.

Subjects and Methods

Clinical findings of family A are described in greater details elsewhere (2). Subject II.1 of family B was part of a highly selected case series of subjects with cardiomyopathy and left ventricular hypertrabeculation or isolated LVNC. This individual underwent clinical genetic testing for a next generation sequencing panel of 38 genes associated with various forms of cardiomyopathies including ACTC1, ACTN2, ANKRD1, CSRP3, DES, EMD, LAMP2, LMNA, MTND1, MTND5, MTND6, MTTD, MTTH, MTTI, MTTK, MTTL1, MTTL2, MTTM, MTTQ, MTTS1, MTTS2, MYBPC3, MYH7, NEXN, PLN, RBM20, SCN5A, SGCD, TAZ, TCAP, TNNC1, TNNI3, TNNT2, TPMI, TTN, TTR, VCL, ZASP.

Immunohistochemistry

Gastric biopsies were obtained by endoscopy procedures and were formalin fixed, alcohol dehydrated, paraffin processed and embedded. Representative biopsies of gastric body were 4 µm cut and mounted on poly-lysine coated slides, deparaffinized, rehydrated and treated at 95°C for 45 min in 10 mM citrate buffer (pH 6.0) to unmask epitopes. Sections were subsequently rinsed with 3% hydrogen peroxide at room temperature to quench endogenous peroxidase activity. The antigen was visualized using the avidin-biotin complex method with an automated instrument from Ventana Medical Systems. Anti-EGFR (3C6) primary antibody (Ventana) was incubated at 37°C for 40 min and after the reaction with secondary antibody and peroxidase, tissues were stained for 5 min with 3,3'-Diaminobenzidine (DAB) chromogen and counterstained with hematoxylin, air dried and cover-slipped. Image capture was performed using Leica DM1000 microscope.

Whole exome sequencing

Whole exome sequencing was performed on DNA samples extracted from peripheral blood of four subjects in two different sessions. The first samples from subjects IV.2 and IV.5 of family A were performed by hybridization of shotgun fragment libraries to the Agilent SureSelect Human All Exon v1 (Agilent Technologies) in-solution capture assays and libraries were sequenced using the SOLiD system v3.5 (Life Technologies) according to manufacturer’s instructions. The sequences were analysed using an automated custom pipeline. Sequencing reads were first colour-corrected using SOLiD Accuracy Enhancer Tool (SAET), then mapped to the reference genome (UCSC, hg19 build) using the software BioScope v1.3 (Life Technologies) and duplicate reads were removed using Picard (http://picard.sourceforge.net). Single nucleotide variations (SNV) and in-del mutation calling analyses were carried out using the appropriate BioScope calling module: diBayes algorithm with medium stringency settings and the SOLiD Small Indel Fragment Tool, respectively.

Samples from subjects II.2 and, II.4 of family A were processed by hybridization of shotgun fragment libraries to the Illumina TruSeq Exome Enrichment Kit (Illumina) and libraries were sequenced using the HiSeq1000 system (Illumina) according to the manufacturer’s instructions. The sequences were analysed using an automated custom pipeline also available as a web resource (44). Paired sequencing reads were aligned to the reference genome (UCSC, hg19 build) using the Burrows-Wheeler Alignment (BWA) tool (45) and post-alignment process and duplicate removal was performed using SAMtools (46) and Picard. Further processing (local realignment around in-del and base recalibration) and SNV and in-del calling were performed with Genome Analysis Toolkit (47).

The called SNV and indel variants produced with both platforms were annotated using ANNOVAR (48) with the relative position in genes using the RefSeq (49) gene model, amino acid change, presence in dbSNP (v137) (50) and in known disease-associated databases (51–53), frequency in ExAC (v0.3), NHLBI EVS, 1,000 genomes project (54), multiple cross-species conservation (55,56), and prediction scores of damaging on protein activity (57–61). The annotated results were imported into an in-house variation database used to make comparisons between samples and filter results. The alignments at candidate positions were visually inspected using the Integrative Genomics Viewer (62). MIB2 and NUP98 mutations were confirmed by Sanger sequencing (Primm).

Computational analysis

Prediction of deleterious the effect of the identified variants was performed by a combined approach using: 1) prediction of variant pathogenicity; 2) prediction of protein stability changes; and 3) functional annotations that overlap or are in close proximity to the affected residues. Human MIB2 (UniProt accession: Q96AX9) is a multi-domain protein with 1070 amino acids whose 3D structure has not been yet determined. MIB2 gene encodes for 17 protein-coding transcripts, according to Ensembl database (http://www.ensembl.org/), therefore in the present manuscript the nucleotide and amino acid sequences are numbered according to RefSeq ID (NM_080875) and Ensembl protein ID (ENSP00000426103), respectively. The Ensembl protein sequence (ENSP00000426103) differs from the UniProt reference sequence (Q96AX9). Only a few 3D models of the human MIB2 are available at ModBase (https://modbase.compbio.ucsf.edu/), a repository that collects comparative protein structure models. The sequence identity between MIB2 (a.a. 715–829) and a designed ankyrin repeat protein whose crystal structure was used as template (PDB ID: 1n0r, chain A, 1.5Å resolution) is 46%, well above the threshold of 30% for reliable fold assignment. We also focused on the 3D modelling of the MIB2 region 945–987, which accommodates the missense variant p.V984L and a Zinc-finger RING type motif. We submitted the MIB2 sequence to the I-TASSER method (http://zhanglab.ccmb.med.umich.edu/I-TASSER) for 3D modelling. The ab initio 3D model calculated by I-TASSER shows a 33% sequence identity with the crystal structure of Baculoviral cIAP2 RING domain (PDB ID: 3eb5, chain A, 2.0Å resolution).

To gain insight into the putative functional effects of the MIB2 variants, we mapped p.V742G and p.V984L onto 3D models. We also used the dbNSFP v2.8 database (63) to obtain the damage prediction of identified missense variants by PolyPhen-2, SIFT, Mutation Assessor, SVM and VEST3. For the identified variants, protein stability calculations were performed using CUPSAT (http://cupsat.tu-bs.de), I-Mutant 2.0 (http://folding.biofold.org/i-mutant/i-mutant2.0.html), and PoPMuSiC (http://dezyme.com). Functional annotations (e.g. ligand-binding residues, protein domains, reported variants in COSMIC database, UniProt annotations for missense variants that overlap the mutation position, residues located in close physical proximity or that affect protein interaction surfaces) were retrieved with the Structure-PPi method available at http://structureppi.bioinfo.cnio.es/Structure (15).

In vitro ubiquitination assay

Full-length human MIB2 cDNA in pCMV-Sport6 backbone was purchased from Open Biosystems (GE Healthcare). The MIB2 mutant constructs (MIB2^V742G and MIB2^V984L) were generated using QuikChange II XL-site-directed mutagenesis (Agilent Technologies) according to manufacturer’s instructions. The cDNAs, which encoded full length MIB2 were subcloned into a pGEX-4T-3 GST expression vector (GE Healthcare). Cloned cDNA full-length sequences were confirmed by Sanger sequencing (Primm). GST-tagged MIB2 constructs were expressed in Rosetta (DE3) cells using isopropyl-1-thio-β-D-galactopyranoside (IPTG) (Sigma-Aldrich). Recombinant proteins were affinity-purified using glutathione Sepharose 4B resin (GE Healthcare). Purified proteins were then analysed by sodium dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis (PAGE) and Coomassie Brilliant Blue.

In vitro ubiquitination assay was performed according to a modified version of a method previously reported (18). Recombinant ubiquitin-activating enzyme E1 (Sigma-Aldrich) and GST-fused MIB2 proteins were mixed into a reaction mixture containing 50 mmol/L Tris-HCl pH 7.4, 5 mmol/L MgCl₂, 4 mmol/L ATP, 0.5 mmol/L dithiothreitol, 15 µg of FLAG-ubiquitin (Sigma-Aldrich) and incubated in equimolar amounts with human His-tagged Ubiquitin Conjugating Enzyme 5b (UbcH5b) (E2) (BioVision) at 30°C for 4 h. Proteins were then subjected to SDS-PAGE under reducing conditions and detected by western blotting using primary antibodies against the FLAG tag (Sigma-Aldrich) or MIB2 (Novus Biological). ECL anti-mouse and anti-rabbit HRP secondary antibodies (GE Healthcare) were used for FLAG and MIB2, respectively. Analysis of band intensities was performed using Quantity One basic software (Bio-Rad Laboratories).

Transfections of MIB2 mutants

MIB2^WT-FLAG, MIB2^V742G-FLAG and MIB2^V984L-FLAG were transfected into HEK293 cells using a standard calcium phosphate protocol. Twenty-four hours after transfection, the medium was supplemented with 10 µg/ml cycloheximide (CHX; Calbiochem) for 1, 2, and 4 h prior to cell lysis in RIPA Buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 0.5% NaDOC, 1% Triton X-100 and 0.1% SDS) supplemented with 90 μg/ml Phenylmethylsulfonyl fluoride (PMSF), 1 mM NaVO₄ (all from Sigma-Aldrich) and proteases inhibitors (Roche). The proteasome inhibitor MG132 (Calbiochem) was supplemented at a concentration of 10 µM for 4 h. After sonication and pre-clearing, protein lysate concentration was determined by Bradford Assay (Biorad). Equal amounts of protein were resolved in 8% SDS-PAGE mini-gels and transferred to nitrocellulose membranes (GE Healthcare). Western Blotting was performed as described above, using a monoclonal FLAG tag antibody (Pierce, Thermo Scientific).

HeLa cells were co-transfected with pCMV-GFP and plasmid expressing wild-type, MIB2 mutants, or p3xFLAG-CMV-14 (Sigma-Aldrich) with Lipofectamine2000 (Life Technologies) according to manufacturer’s instructions. Eighteen hours post-transfection, proteins and RNA were extracted in either RIPA buffer or with the RNeasy kit (Qiagen). Proteins were analysed by western blotting using rabbit anti-MIB2 (Novus Biological), mouse anti-MIB2 (Abcam) and anti-GFP (Life Technologies). Anti-GAPDH (Santa Cruz Biotechnology) was used for normalization. ECL anti-rabbit or anti-mouse HRP were used as secondary antibodies (GE Healthcare).

RNA was reverse transcribed using a first-strand complementary deoxyribonucleic acid kit with random primers according to the manufacturer’s protocol (Life Technologies). The qPCR reactions were performed using Roche Light Cycler 480 system (Roche). PCR reactions were performed with SYBR Green Master Mix (Roche). PCR conditions were as follows: pre-heating, 5 min at 95°C; 40 cycles of 15 s at 95°C, 15 s at 60°C and 25 s at 72°C. Quantification results were expressed in terms of cycle threshold (Ct). For the expression analysis GUSB and HPRT1 housekeeping genes were used as endogenous controls (reference markers) using LightCycler 480 software v1.5. Differences between mean C_t values of tested genes and those of the reference gene were calculated as ΔC_t gene = C_t gene - C_t reference. WT1 sample was used as calibrator and relative fold increase in expression levels was determined as E^-ΔΔCt, E being primer efficiency. The Ct values were averaged for each technical duplicate. Primers are reported in Supplementary Material, Table S4.

NOTCH gene expression in white blood cells

White blood cells were isolated from peripheral blood using Lymphocyte Separation Medium (Lonza) from subjects III.6 and IV.5 of pedigree A, and from age-matched healthy controls. Total RNA was extracted from primary cultured fibroblasts using the RNeasy kit (Qiagen) according to the supplier’s instructions. RNA was reverse transcribed and amplified by qPCR reaction as reported above. Primers are in the Supplementary Material, Table S4.

rAAV vectors

Recombinant AAV vectors were prepared by the AAV Vector Unit at the International Centre for Genetic Engineering and Biotechnology in Trieste, as described previously (64). In brief, infectious rAAV6 vector particles were generated in HEK293 cells by co-transfecting each vector plasmid (pAAV-MIB2^WT, -MIB2^V742G, -MIB2^V984L and pAAV-MCS) together with the packaging plasmid pAAV2/6 (65) and helper plasmid (pHELPER; Stratagene) expressing AAV and adenovirus helper functions, respectively. Viral stocks were obtained by CsCl₂ gradient centrifugation; rAAV titers were determined by measuring the copy number of viral genomes (vg) in pooled, dialyzed gradient fractions, as described previously (66), and they were in the range of 1x10¹⁰ to 1x10¹³ vg/ml.

Rat primary cardiomyocytes

Animal care and treatment were performed following the Institutional Guidelines in compliance to National and International laws and policies. Wistar rats were purchased from Charles River Laboratories, Italia and maintained under controlled environmental conditions. Ventricular cardiomyocytes from neonatal rats were isolated as previously described (22). In brief, ventricles from neonatal rats (post-natal day 1) were separated from the atria, cut into pieces and dissociated in CBFHH (calcium and bicarbonate-free Hanks with Hepes) buffer containing 2 mg/ml of trypsin (BD Difco) and 20 µg/ml of DNase II (Sigma-Aldrich) under constant stirring. Digestion was performed at room temperature in eight-to-ten 10-min steps, collecting the supernatant after each step. The supernatants were centrifuged to separate the cells, which were then re-suspended in Dulbecco’s modified Eagle medium 4.5 g/L glucose (DMEM, Life Technologies) supplemented with 5% FBS, 20 µg/ml vitamin B12 (Sigma-Aldrich), 100 U/ml penicillin and 100 µg/ml of streptomycin (Sigma-Aldrich). The collected cells were filtered through a cell strainer (40 µm, BD Falcon), and then seeded onto uncoated 100-mm plastic dishes for 2 h at 37°C in 5% CO₂ and humidified atmosphere. The supernatant composed mostly of cardiomyocytes was then collected and plated. Cultures of neonatal rat ventricular cardiomyocytes prepared by this procedure yielded consistently a purity >90%. In the experiments with rAAV-mediated gene transfer, neonatal rat cardiomyocytes were infected immediately after isolation with 1 × 10⁵ vg/cell. Twelve hours later, the culture medium was changed and cells were subjected to the different treatments and subsequent analyses.

Functional studies in primary cardiomyocytes

To detect NOTCH1 activity under different experimental conditions, neonatal cardiomyocytes were seeded onto 96-well primary cell culture plates (1.5 ×10⁴ cells per well) and transduced with rAAV6-MIB2^WT, rAAV6-MIB2^V742G, rAAV6-MIB2^V984L, or an empty rAAV6-MCS vector [multiplicity of infection (moi) =1×10⁵ vg/cell]. The day after the infection, cardiomyocytes were transfected with 0.5 μg of 4XCBF1-Luc reporter plasmid and 0.01 μg pRL-Renilla (which was used as an efficiency control) using Lipofectamine 2000 transfection reagent (Life Technologies). After 6 h of incubation, the medium was removed and cells were fed in complete medium for 8 h in DMSO (control) or γ-secretase inhibitor containing medium (DAPT 20 µM) prior to lysis in PLB buffer (Promega). Firefly luciferase activity was corrected for the transfection efficiency using the control Renilla luciferase activity in each sample. Firefly and Renilla luciferase activities were assayed with the Dual-luciferase assay kit (Promega).

Cardiomyocyte proliferation was analysed 4 days after their isolation. Cells transduced with either rAAV6-MIB2^WT, or rAAV6- MIB2^V742G, or rAAV6- MIB2^V984L vector were grown in culture medium supplemented with 5 µM 5-ethynyl-2’-deoxyuridine (EdU, Life Technologies) for 8 h. Cells were fixed with 4% paraformaldehyde for 10 min, permeabilized with 0.1% Triton X-100 in phosphate buffered saline (PBS) for 10 min, followed by 1 h blocking in 2% Bovine Serum Albumin (BSA, Roche) in PBS. Cells were then stained overnight at 4°C with the mouse anti α-sarcomeric actinin (EA-53) (Abcam). Cells were washed with PBS and incubated for 1 h with the secondary antibodies goat anti-mouse conjugated to Alexa Fluor 488. All washes were in PBS 0.2% Tween 20. Cells were further processed using the Click-IT EdU 555 Imaging kit to reveal EdU incorporation, according to the manufacturer’s instructions, and stained with Hoechst 33342 (Life Technologies).

Image acquisition was performed using an ImageXpress Micro automated high-content screening fluorescence microscope at 10X magnification; a total of 16 images were acquired per wavelength, well and replicate, corresponding to approximately 2,500 cells analysed per condition and replicate. Image analysis was performed using the Multi-Wavelength Cell Scoring application module implemented in MetaXpress software (Molecular Devices). In all quantifications, cardiomyocytes were distinguished from other cells present in the primary cultures for their sarcomeric α-actinin positive signals. Images were acquired at room temperature with a DMLB upright fluorescence microscope (Leica) equipped with a charge-coupled device camera (CoolSNAP CF; Roper Scientific) using MetaView 4.6 quantitative analysis software (MDS Analytical Technologies). Within each experiment, instrument settings were kept fixed.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We thank the patients for participation to the study. We are grateful to Hamed Jafar-Nejad for fruitful discussions. We thank Michele Santoro and Antonella De Matteis for technical support with recombinant protein production and purification.

Conflict of Interest statement. None declared.

Funding

This work was supported by a grant from the Italian Association for Cancer Research (AIRC) to N.B.-P (MFAG11938), by the Indiana University Health-Indiana University School of Medicine Strategic Research Initiative (T.C.L., P.C.S., K.G.S. and M.V.) and by the EU FP7 project ASSET (grant agreement 259348 to AV).