Co-Occurring Heterozygous CNOT3 and SMAD6 Truncating Variants: Unusual Presentation and Refinement of the IDDSADF Phenotype

Objective, the application of genomic sequencing in clinical practice has allowed us to appreciate the contribution of co-occurring pathogenic variants to complex and unclassified clinical phenotypes. Besides the clinical relevance, these findings have provided evidence of previously unrecognized functional links between genes in the context of developmental processes and physiology. Patients and Methods, a 5-year-old patient showing an unclassified phenotype characterized by developmental delay, speech delay, peculiar behavioral features, facial dysmorphism and severe cardiopathy was analyzed by trio-based whole exome sequencing (WES) analysis to identify the genomic events underlying the condition. Results, two co-occurring heterozygous truncating variants in CNOT3 and SMAD6 were identified. Heterozygous loss-of-function variants in CNOT3, encoding a subunit of the CCR4-NOT protein complex, have recently been reported to cause a syndromic condition known as intellectual developmental disorder with speech delay, autism and dysmorphic facies (IDDSADF). Enrichment of rare/private variants in the SMAD6 gene, encoding a protein negatively controlling transforming growth factor β/bone morphogenetic protein (TGFB/BMP) signaling, has been described in association with a wide spectrum of congenital heart defects. We dissected the contribution of individual variants to the complex clinical manifestations and profiled a previously unappreciated set of facial features and signs characterizing IDDSADF. Conclusions, two concomitant truncating variants in CNOT3 and SMAD6 are the cause of the combination of features documented in the patient resulting in the unique multisystem neurodevelopmental condition. These findings provide evidence for a functional link between the CCR4-NOT complex and TGFB/BMP signaling in processes controlling cardiac development. Finally, the present revision provides evidence that IDDSADF is characterized by a distinctive facial gestalt.


Introduction
Proband-parents whole exome sequencing (trio-WES) analysis has substantially improved the chance of obtaining a genetic diagnosis in rare and ultra-rare diseases. In approximately 5% of cases, WES also allows dissection of complex phenotypes as the result of concomitant pathogenic variants involving two or multiple genes (dual molecular diagnoses) [1,2]. Patients with dual diagnoses can be characterized by either qualitatively distinct phenotypes or attenuated/strengthened clinical presentation with respect to the individual diseases depending on the functional link and the role of the genes involved in terms of developmental processes and physiology, eventually resulting in a modulated expression of one or several features [3]. In both cases, a priori discrimination of the contribution of individual pathogenic variants to the complexity of the clinical phenotype is generally a hard task [1]. This is traditionally due to the "common use" of the intuitive principle of parsimony (Occam's razor), in which the justification based on the simplest assumption (i.e., single genetic cause) is generally preferred in explaining a rare clinical presentation [4]. However, the massive use of WES in selected cohorts of "partially solved" patients (i.e., presentations that only in part are explained by the identified genetic variant, based on the available clinical information for the disorder) has provided evidence that at least one third of cases reported as 'phenotypic expansion' actually represents blended phenotypes due to concomitant co-occurring pathogenic variants [5]. The use of genomic sequencing has eventually changed the classical concept of Mendelian inheritance, since alleles with simple Mendelian behavior seem to be an exception rather than the rule in patients with complex phenotypes [3]. Similarly, we appreciate that even in apparently "monogenic diseases" a higher burden of functionally relevant rare variants increases the likelihood of involving modifier alleles able to modulate the phenotypic expression of the driver event [6], either exacerbating or smoothening it [7][8][9]. To add complexity to this landscape, the presence of multiple genetic diseases within a family, each affecting different members with variable modes of inheritance and age-of-onset, might represent an additional confounding element.
Heterozygous variants in CNOT3, encoding a subunit of the CCR4-NOT protein complex, a multimeric regulator with variegated function in mRNA metabolism and gene expression, have recently been reported to cause a developmental delay (DD)/intellectual disability (ID) condition (intellectual developmental disorder with speech delay, autism, and dysmorphic facies (IDDSADF), Mendelian Inheritance in Man (MIM): 618672) [10][11][12]. CNOT3 variants are largely truncating, which supports haploinsufficiency as the mechanism of disease.
As a paradigmatic example of complex dual diagnosis requiring accurate dissection of clinical features after WES to explain the whole presentation, here we describe a 5-year-old boy affected with a syndromic ID disorder with speech delay, peculiar behavioral phenotype, facial dysmorphism and a severe cardiopathy, who carried two co-occurring pathogenic variants in CNOT3 and SMAD6. We dissect the contribution of individual variants to the complex clinical manifestations of the patient and more accurately characterize the clinical features of IDDSADF by reviewing the previous reports to assess consistent common dysmorphic features and clinical signs. Finally, we also provide evidence for a functional link between the CCR4-NOT complex and TGFB/BMP signaling in processes controlling cardiac development.

Clinical Case
The proband was a 5-year-old boy born at term from apparently non-consanguineous parents after an uneventful pregnancy. Caesarian section was required due to failure in engagement and progression. No exposure to teratogens during pregnancy was reported. Birth weight was 2650 g (5th centile), length 47 cm (10th centile), and head circumference 33 cm (3rd centile). Developmental milestones were normally acquired during the first year, but he presented with language regression soon after 12 months. Language was limited to few words at 4 years. He was diagnosed with aortic stenotic coarctaction (CoA), hypertrophic left ventriculum and bicuspid aortic valve (BAV), surgically treated at 3 years and then pharmacologically controlled afterwards with combined therapy of angiotensinconverting-enzyme (ACE) inhibitor (enalapril) and β blocker (atenolol). In the same period, he was also diagnosed with DD, opposite behavior with emotional outbursts hardly controlled and short span of attention. Suspect of autism spectrum disorder (ASD) was considered, and he was treated with logopedy with limited benefits. At 4 years he presented with facial dysmorphisms, including a prominent forehead, upslanted palpebral fissures, long and thin eyebrows, flat nasal bridge, deep set eyes with periorbital puffiness, low-set anteverted ears with upper folded helix, flabby cheeks with nasolabial sulci prominence, anteverted nares with broad nasal tip, long and smooth philtrum, thin upper lip, and inferior lip prominence with prominent labiomental groove. He also showed a short neck, mild short stature, soft skin and hypermobility of joints. Spontaneous rectal prolapse occurred at 4 years. At our last examination (5 years), he had a weight of 21 Kg (75th centile), height of 108 cm (25th centile) and head circumference of 52 cm (60th centile). The cardiological follow up (5 years), revealed a slight re-coarctation at the former CoA site. A cerebral brain magnetic resonance imaging (MRI) showed hypoplasia of the posterior corpus callosum (Figure 1). In the same period, he was admitted to hospital because of a suspect critical episode that spontaneously resolved. The proband's mother is a normally intellectually developed 32-year-old woman. Numerous echocardiograms, thoracic aorta ultrasound evaluations and electrocardiogram (ECG) showed no evidence of cardiopathy, thoracic-abdominal aneurism (TAA) nor anomalies in cardiac conduction. She is currently under cardiac and thoracic aorta follow up with annually repeated ultrasounds. The proband's mother is a normally intellectually developed 32-year-old woman. Numerous echocardiograms, thoracic aorta ultrasound evaluations and electrocardiogram (ECG) showed no evidence of cardiopathy, thoracic-abdominal aneurism (TAA) nor anomalies in cardiac conduction. She is currently under cardiac and thoracic aorta follow up with annually repeated ultrasounds.

Methods
The proband was evaluated in the frame of a research program dedicated to undiagnosed patients. The study was conducted in accordance with the declaration of Helsinki principles and after a signed written informed consent was secured. Genomic DNA was extracted from circulating leukocytes. Target regions were captured using the SureSelec-tAllExon v.7 kit (Agilent, Santa Clara, CA, USA), and sequenced on a NextSeq550 platform (Illumina, San Diego, CA, USA). WES raw data were processed and analyzed using an in-house implemented pipeline previously described [23][24][25] mainly based on the Genome Analysis Toolkit (GATK) Best Practices. Briefly, the GRCh37/hg19 genome assembly was used for reads alignment by means of the Burrows-Wheeler Alignment-Maximal Exact Match (BWA-MEM) tool [26], and HaplotypeCaller (GATK v3.7) for variant calling [27]. Variants/genes functional annotation was made by SnpEff v.4.3 [28] and dbNSFP v.3.5 [29], and various clinical databases (OMIM, ClinVar, HGMD). In silico prediction of the impact of variants was performed by Combined Annotation Dependent Depletion (CADD) v.1.4 [30], Mendelian Clinically Applicable Pathogenicity (M-CAP) v.1.0 [31], and Intervar v.2.0.1 [32]. By filtering against an in-house population-matched database (~2500 exomes) and public databases (dbSNP150 and gnomAD V.2.0.1), the analysis was focused on high-quality rare variants affecting coding sequences and adjacent intronic regions. For variant prioritization, we took advantage of an in-house developed scoring system integrating various variant functional annotation (by means of SnpEff), variant clinical impact according to American College of Medical Genetics and Genomics (ACMG) criteria (by means of InterVar), variant functional impact (by means of CADD, M-CAP, S-CAP [33], Spidex [34]), population frequency for the variant in different databases (gnomAD and internal one), gene-tolerance to mutations (Gene Damage Index [GDI] and Residual Variation Intolerance Score [RVIS] tools from dbNSFP v3.5), and genotype-phenotype correlation (by means of Phenolyzer [35]). Sanger sequencing was performed for variant validation and segregation analysis (primers available on request).
Clinical data review of published IDDSADF patients and assessment of available clinical pictures were double blinded performed by three experienced clinical geneticists (P.M., R.F.C. and D.B.).

Results
Trio-based WES analysis revealed two clinically relevant variants as the molecular causes of the complex phenotype in the proband. A single base insertion in CNOT3 (c.732dup, p.Ser245GlnfsTer8, NM_014516.4) was identified as a de novo event. A second truncating 19-nucleotide-long deletion in SMAD6 (c.232_250del, Gln78GlyfsTer41, NM_005585.5) was inherited from the apparently healthy mother. The two variants had previously been reported as pathogenic, causing IDDSADF (rs753475896, ClinVar ID: VCV000694669) [11] and isolated craniosynostosis [20], respectively. No other additional functionally or clinically relevant variant compatible with dominant or recessive (autosomal or X-linked) inheritance model was identified (Supplementary Materials Table S1). Proper coverage excluded occurrence of other variants within the coding exons of the two genes (Supplementary Materials Figure S1).
By reviewing the clinical features associated with pathogenic CNOT3 and SMAD6 variants, we suspected a causal role of CNOT3 haploinsufficiency in the pathogenesis of the peculiar facial features, ID and behavioral anomalies, based on two recently published papers reporting 16 sporadic cases and two families with multiple affected members sharing pathogenic variants in this gene [11,12]. On the other hand, we could ascribe the severe congenital heart defect to the heterozygous truncating variant in SMAD6. Loss-of-function SMAD6 variants have been variably associated with different clinical phenotypes, including cardiac, craniosynostosis and RUS, in the absence of any strict genotype-phenotype correlation. Among the cardiac features, an association between BAV/aortic coartaction and SMAD6 haploinsufficiency has been established [16][17][18][19].

Discussion
We identified two co-occurring frameshift variants in CNOT3 and SMAD6 as the molecular events likely underlying a complex multisystem phenotype. p.Gln78GlyfsTer41 in SMAD6 is predicted to result in a truncated protein lacking both the MH1 and MH2 domains, which are required for the inhibitory activity of SMAD6 on TGFβ/BMP signaling [13]. Previous studies documented that even more 3 -terminal frameshifts cause a reduced SMAD6 protein level, suggesting either nonsense-mediated decay (NMD) or accelerated protein degradation [21]. Similarly, p.Ser245GlnfsTer8 in CNOT3 is predicted to generate a truncated protein lacking the functional domains downstream the linker region CNOT3-M [36], including the C-terminal region of CNOT3 containing the functionally relevant NOT1 anchor region (NAR) and NOT-box domain. NMD or accelerated protein degradation cannot be ruled out, as well.
IDDSADF is caused by de novo and inherited variants in CNOT3 [11,12], which encodes a subunit of a multimeric versatile complex with role in multiple processes controlling gene expression [37][38][39][40][41][42][43]. The spectrum of pathogenic CNOT3 variants (10 missense and 12 truncating, the present representing the second recurring mutation) and the intolerance for loss-of-function variation (pLI = 1) [44], supports haploinsufficiency as the mechanism of disease. The main IDDSADF clinical features and signs are summarized in Table 1. By reviewing the previous series and comparing them with the present case, we basically confirmed the presence of a recognizable developmental disorder (100%) with speech delay (95%), autism and/or behavioral problems (59%), dysmorphic face and relatively short stature with a median presentation around ≤25th centile (18/22 pts; 82%). The type and severity of DD and learning difficulties is highly variable but invariably present. Speech delay is a constant feature with the exception of one patient [11]. Behavioral problems ranging from pervasive developmental disorder (7/22 pts, 32%) to autism (7/22 pts, 32%) also occur. Muscular hypotonia has been reported in up to 65% of patients. Consistently, our proband showed muscular hypotonia and developed a rectal prolapse at 4 years without any apparently causative trigger. We might speculate on possible pelvic muscle floor hypotonia in the pathogenesis of this unusual finding. Vision impairment, mostly strabismus, is quite frequent (62%). Brain MRI anomalies are common (64%), in particular corpus callosum hypoplasia (5/14 pts; 55% of total brain anomalies), which was also present in our patient. Seizures/electroencephalography (EEG) abnormalities occur in 24% of cases. Our patient manifested a single critical episode without EEG abnormalities. Due to his young age, he is currently under follow up. Based on published data, some recurrent distinctive facial features are recognized, including prominent forehead, flat/high nasal bridge, upslanting palpebral fissures, flabby cheeks with nasolabial sulci prominence, broad nasal tip with anteverted nares, smooth long philtrum, thin upper lip and inferior lip prominence with prominent labiomental groove.
More rarely, CNOT3 variants have been associated with unilateral/bilateral moyamoya angiopathy (MMA), a vascular trait conferring increased risk to strokes starting in childhood through the fourth decade of life [45]. Among the four reported subjects, the patient with bilateral MMA carried a previously described truncating IDDSADF-causing CNOT3 variant (p.Gln215*) and presented with ID/ASD, supporting the diagnosis of ID-DSADF, even though the lack of detailed clinical description did not allow us to include in our revision. A re-evaluation of our patient's MRI brain scans excluded signs of MMA. Of note, the remaining three MMA cases did not show ID and/or ASD and carried missense CNOT3 changes not reported in IDDSADF with a variably functional impact by in silico prediction tools or inherited from a normal parent without signs of MMA, thus not supporting their role in this disorder. While available data do not suggest MMA as a major complication of IDDSADF, further studies are required to exclude any link between MMA and IDDSADF.
In the two published IDDSADF series, 2 patients out 21 (9%) had a cardiac anomaly (i.e., perimembranous VSD and abnormal right pulmonary venous return) [11], suggesting that CHD is not a recurrent feature in this disorder. However, our proband presented with a severe heart defect, which was explained by the co-occurring heterozygous SMAD6 frameshift variant. Smad proteins are intracellular mediators required for regulation of intracellular signal transduction by members of the TGFβ/BMP superfamily [13]. Truncating and missense SMAD6 variants have been associated with AVD type 2. Incomplete penetrance for cardiovascular features has been reported in SMAD6 variant carriers (up to 82%) [18]. Variable clinical expressivity within families has also been evidenced, with CoA being a common comorbidity [15,18]. Both cardiac features were present in our patient. While SMAD6 variants, including the frameshift here reported, have been proposed to contribute to craniosynostosis, in association with a common polymorphism within the BMP2 locus (rs1884302) [20], this association has not been confirmed in additional craniosynostosis cohorts nor in cardiovascular/RUS patients [15,18,19,21,22]. Available data, however, support a model in which loss-of-function SMAD6 variants are likely to result in variable phenotypes depending on concomitant genetic or environmental factors, as proved by the recurrence of the same variants in patients affected with either BAV/TAA or craniosynostosis [17,20]. Consistently, our patient had a heterozygous truncating SMAD6 variant also reported in a patient with isolated craniosynostosis [20], in which the proband's parent transmitting the SMAD6 variant was apparently unaffected, as in the present family. Loss-of-function SMAD6 variants have been found in subjects with late-onset cardiopathies and TAAs and evolutive cardiovascular complications, such as postoperative aneurysm at the former CoA site. In one family, the proband's father, who was heterozygous for the SMAD6 variant, deceased at age 80 from diffuse TAA, and had disseminated aortic calcifications [18], which are quite well represented in SMAD6 variant-positive individuals [19]. In the present family, the proband showed early-onset CoA, which relapsed at the former CoA site 2 years after surgical correction. Although additional studies are required to confirm late-onset complications in these subjects, long-life cardiac monitoring would be highly be recommended. Accordingly, the proband's mother is currently monitored for cardiac and thoracic aorta complications.
Previous reports in SMAD6-related AVD support the idea that pathogenic SMAD6 variants require a concomitant genetic event to promote adverse clinical outcome. This can be a second hit involving the other SMAD6 allele, a variant affecting another functionally related gene (TGFB/BMP pathways), or a co-occurring genomic rearrangement (e.g., 15q11.2 BP1-BP2 microdeletion) [18]. In this context, although no direct link between the two proteins is apparent in terms of functional pathways and physical interactions (Supplementary Materials Figure S2), CNOT3 has been identified as a critical regulator in cardiomyocyte differentiation and proliferation in human embryonic stem cells [46]. Silencing of the components of the CCR4-Not complex in Drosophila results in myofibrillar disarray and dilated cardiomyopathy [47]. Consistently, heterozygous Cnot3 knockout (KO) mice display impaired cardiac contractility and increased susceptibility to heart failure [47], while conditional heart and muscle KO Cnot3 mice are characterized by severe cardiac contractility defects accompanied by long QT intervals and various arrhythmic changes [38]. Thus, it is reasonable to speculate on a possible role of CNOT3 haploinsufficiency as the "second hit" contributing to the cardiac phenotype associated with the inactivating SMAD6 variant in the present patient.

Conclusions
In conclusion, we describe a patient affected with a unique dual molecular diagnosis involving inactivating CNOT3 and SMAD6 variants, whose presentation is fully explained by either one (IDDSADF due to CNOT3 haploinsufficiency) or concomitant events (AVD type 2/CoA due to CNOT3 and SMAD6 variants). We also document the occurrence of a distinctive facial gestalt in IDDSADF by defining the recurrent facial features and signs. The present finding suggests a functional link between CCR4-NOT and TGFB/BMP signaling in processes controlling cardiac development. Eventually, we further prove the need for accurate expertise in both clinical and genomic aspects as a key to properly managing patients receiving complex molecular diagnoses. From a clinical perspective, recognizing a multiple molecular cause has several implications for both treatment and the estimation of familial recurrence risk. Multiple variants require careful interpretation of clinical validity by the referring clinician in terms of clinical relevance and proper patient management (i.e., monitoring and follow up), particularly in apparently healthy subjects.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and was approved by the local Institutional Ethical Committee of the Ospedale Pediatrico Bambino Gesù (1702_OPBG_2018, 11 February 2019).

Informed Consent Statement:
Signed informed consents were obtained from the participating family.

Data Availability Statement:
Data that support the findings of this study are available from the corresponding authors upon request.