1 INTRODUCTION

Cornelia de Lange syndrome (CdLS, MIM#s 122470, 300590, 610759, 614701, 300882) is a rare developmental disease characterized by pre-natal and post-natal growth retardation, facial dysmorphism, cognitive impairment, and structural defects in multiple organs (Mannini, Cucco, Quarantotti, Krantz, & Musio, 2013). This wide phenotypic variability of CdLS depends on its wide genetic heterogeneity. In fact, seven CdLS-causative genes (NIPBL, SMC1A, SMC3, HDAC8, RAD21, BRD4, and ANKRD11) have been identified up to now (Kline et al., 2018). All these genes encode core cohesin components or associated factors. The cohesin complex plays multiple roles in biological processes such as sister chromatid cohesion, gene expression regulation, and DNA repair (Cucco & Musio, 2016; Merkenschlager & Odom, 2013; Musio, 2020). The importance of cohesin in maintaining genome stability is proven by the identification of somatic cohesin variants in many human cancers (Kon et al., 2013; Sarogni et al., 2019).

CdLS cell lines do not display sister chromatid cohesion abnormalities (Castronovo et al., 2009; Revenkova et al., 2009) indicating that the molecular mechanism of CdLS is not directly linked to the disruption of sister chromatid cohesion. Instead, considerable evidence suggests that CdLS etiopathology is due to the impaired ability of the cohesin to mediate other more dosage-sensitive cellular functions such as transcriptional regulation. NIPBL, HDAC8, and SMC1A mutant CdLS cell lines display transcriptional dysregulation; being conservative, the changes in gene expression are typically less than 1.5-fold (Cukrov et al., 2018; Deardorff et al., 2012; Liu et al., 2009; Mannini et al., 2015). Interestingly, this modest transcription dysregulation is in agreement with proteomic published data (Gimigliano et al., 2012). Furthermore, hundreds of genes are weakly transcriptionally dysregulated in the mouse, Drosophila and zebrafish CdLS model organisms (Kawauchi et al., 2009; Muto, Calof, Lander, & Schilling, 2011; Newkirk et al., 2017; Schaaf et al., 2009). These observations further support the notion that CdLS phenotype results from the collective action of many mild perturbations (Sarogni, Pallotta, & Musio, 2020). The molecular basis underlying gene dysregulation in CdLS is not clearly defined. It has been shown that mutant cohesin makes the binding chromatin/cohesin more stable and affects the recruitment of RNA polymerase II (Pol II) at both promoter and gene body of transcriptional active genes (Mannini et al., 2015; Revenkova et al., 2009).

The seven CdLS genes account for about 80% of cases, so the etiopathology of the remaining 20% of the patients with a clinical diagnosis of CdLS is unknown (Kline et al., 2018; Ramos et al., 2015). However, a correct diagnosis of CdLS is not readily made. Several observations indicate a high percentage of mosaicism in known CdLS genes that were previously excluded by Sanger sequencing using DNA obtained from peripheral blood samples (Ansari et al., 2014; Baquero-Montoya et al., 2014; Castronovo et al., 2010; Huisman, Redeker, Maas, Mannens, & Hennekam, 2013; Pozojevic et al., 2018). Furthermore, patients presenting CdLS or CdLS-like phenotype such as KBG (OMIM 148050), Wiedemann–Steiner (OMIM 605130), or Rubinstein–Taybi (RSTS2, OMIM 613684) have been described, notably, to harbor variants in chromatin-associated factors other than cohesin genes (Ansari et al., 2014; Aoi et al., 2019; Parenti et al., 2017; Woods et al., 2014; Yuan et al., 2015).

Here we report on the clinical and molecular characterization of two patients with features overlapping CdLS, for whom the analysis CdLS-causative genes did not identify any pathogenic variant. The subsequent use of exome sequencing allowed us to identify heterozygous variants in the ANKRD11 and EP300 genes. In addition, pathogenic variants led to low levels of both ANKRD11 and p300 proteins suggesting that this could contribute to the atypical form of CdLS. ANKRD11 and p300 are involved in gene expression regulation by chromatin remodeling (Chen et al., 2008; Gallagher et al., 2015). This report, presenting phenotypic overlap between CdLS, KBG, and RSTS, provides more evidence for the possibility that three distinct disorders share a common pathogenetic molecular mechanism.

2 MATERIALS AND METHODS

3 RESULTS

3.1 Patient ascertainment

Patient A (Figure 1a) is the second child of an unrelated healthy couple. During the first months of life she showed feeding difficulties, with persistent crying, regurgitation, and poor weight gain. Gastroesophageal reflux symptoms manifested at night. She showed micrognathia, hirsutism, and synophrys. Her hands and feet were normal with the absence of broad thumb/hallux and large distal phalanges. In addition, she displayed profound hearing loss, both sensorineural and conductive. At the age of 6 years, neuropsychiatric evaluation showed moderate–severe cognitive impairment with an IQ of 49 (based on the Leiter-R scale) corresponding to a mental age of 36 months. Her growth measurements were small for her age. Her language was almost completely absent (producing about 20 words), and her personal autonomy was very limited. Finally, oppositional behaviors and low threshold of frustration tolerance were also present.

Patient B (Figure 1b) is the first child of an unrelated healthy couple. At birth, the proband had facial dysmorphism, long philtrum, thin lips, micrognathia, low anterior hairline, arched eyebrows, long eyelashes, synophrys, small hands, and hirsutism. At age 4 years, he was operated on for atrial septal defect. He also showed bilateral hypoacusis. His growth measurements were normal for his age. At the age of 7 years, neuropsychiatric evaluation showed moderate intellectual disability, specific behavioral issues such as aggression, defiance, extreme shyness, perseveration, and obsessive–compulsive behaviors. Both patients (A and B) were classified as non-classic CdLS (Table 1) when their clinical features were reevaluated based on clinical diagnostic criteria (Kline et al., 2018).

3.2 Exome sequencing

Since Sanger sequencing could not identify a disease-causing variant in CdLS-causative genes in any of the two patients, we used exome sequencing to sequence the patients' DNA.

Patient A was found to carry the c.4408_4409insA truncating frameshift variant in heterozygous status in exon 27 of the EP300 gene (RefSeq NM_001429.4) predicting a truncated protein sequence of 1,471 amino acids (aa), lacking part of the lysine (K) acetyltransferase (KAT) domain, instead of the full-length 2,414-aa p300 protein, p.(Met1470AsnfsTer3). This pathogenic variant was confirmed as de novo by Sanger sequencing (Figure S1a).

Patient B carried the pathogenic variant c.3224_3227del in heterozygous status in the ANKRD11 gene (RefSeq NM_013275.6) leading to p.(Glu1075GlyfsTer242) amino acid change. Neither parent had the variant indicating a de novo event (Figure S1b).

We next investigated the effect of the identified pathogenic variants on protein expression. Western blotting experiments showed that both c.4408_4409insA and c.3224_3227del pathogenic variants led to low levels of p300 and ANKRD11, respectively (Figure 2a,b, Figure S2a,b). RT-qPCR results revealed excellent agreement with western blotting experiments (Figure S3).

4 DISCUSSION

Exome sequencing has quickly become the method of choice for identifying non-high-frequency variations such as rare and de novo pathogenic variants in protein-coding exomes. In this study, exome sequencing was employed for the mutational screening of two probands with a clinical diagnosis of CdLS spectrum. By using this approach, we were able to identify the c.4408_4409insA and c.3224_3227del pathogenic variants in the EP300 and ANKRD11 genes that are associated with RSTS and KBG syndromes, respectively. To our knowledge, the variant in the EP300 has been described for the first time while that in the ANKRD11 gene has been recently reported in a patient with Coffin–Siris overlapping syndrome (Miyatake et al., 2017), adding more evidence that distinct disorders share a common genetic mechanism.

RSTS is a rare genetically heterogeneous multiple anomaly syndrome, and it is caused by pathogenic variants in the CREBBP (RSTS1) and EP300 (RSTS2) genes. According to HGMD (updated to October 2019), 21.3% (84 out of 394 variants) of clinically recognized RSTS cases carry EP300 pathogenic variants. EP300 variants have been associated with a milder phenotype, especially regarding the skeletal defects (Bartholdi et al., 2007; Bartsch et al., 2010; Fergelot et al., 2016; Foley, Bunyan, Stratton, Dillon, & Lynch, 2009; Tsai et al., 2011). KBG syndrome, a rare genetic disorder, is associated with the loss of function variants in the ANKRD11 gene, which encodes a chromatin regulatory protein (Gallagher et al., 2015). Our finding that ANKRD11 and EP300 inactivating pathogenic variants could be leading to low levels of the respective proteins suggests that this phenomenon results in an atypical form of CdLS. In this context, the EP300 and ANKRD11 pathogenic variants lead to premature termination codons, mRNA could be degraded by a nonsense-mediated decay mechanism to avoid the translation and accumulation of truncated proteins.

Although classic CdLS, KBG, and RSTS are unique conditions that are unlikely to be confused with one another, the differential diagnosis is more complicated in borderline-mild phenotypes (Avagliano et al., 2020; Kline et al., 2018). Until now, nine CdLS-like patients carrying variants in the EP300 or ANKRD11 genes have been described (Ansari et al., 2014; Aoi et al., 2019; Parenti et al., 2016; Woods et al., 2014), and clinical features were reported for six of them (Aoi et al., 2019; Parenti et al., 2016; Woods et al., 2014). Detailed findings of these CdLS-like patients, including our two, are shown in Tables S2 and S3. Most of the pathogenic variants are deletions, with the exception of two single nucleotide changes and a nucleotide insertion. Clinical reevaluation of our patients showed that some clinical signs, such as cognitive impairment, synophrys, microcephaly, and thin upper lip, and the absence of macrodontia or fused incisors suggested the clinical diagnosis of CdLS. This diagnosis was confirmed for the patient carrying the ANKRD11 variant, also in accordance with the recent CdLS consensus statement (Kline et al., 2018). Instead, the patient harboring EP300 variant was diagnosed as RSTS2. Since increasing data show that CdLS, KBG, and RSTS share these phenotypic features (Ansari et al., 2014; Aoi et al., 2019; Parenti et al., 2016; Woods et al., 2014), the testing for both EP300 and ANKRD11 variants in probands with features of CdLS should be warranted and included on panels for CdLS testing.

The finding that ANKRD11 and EP300 pathogenic variants cause a phenotype that mimics CdLS is intriguing. p300 functions as histone acetyltransferase that regulates transcription via chromatin remodeling and as a transcription co-factor by bridging transcription factors and intergenic RNA polymerase II (Chen et al., 2008; Roelfsema et al., 2005). Furthermore, it has been shown that p300, NIPBL, and the core cohesin proteins co-occupy enhancer regions in different cell types (Chen, Morris, & Mitchell, 2012; Kagey et al., 2010). Also ANKRD11 is a chromatin-associated/modifying protein that is particularly involved in the inhibition of the transcriptional activation of nuclear receptor targets genes via the recruitment of histone deacetylase proteins (Gallagher et al., 2015; Sirmaci et al., 2011; Zhang et al., 2004). Altogether, these observations indicate that perturbations in chromatin remodeling pathways, which include cohesin, and the EP300 and ANKRD11 genes, lead to human diseases and that their phenotypic effects might mimic one another. In this view, overlapping disorders can develop from a common genetic mechanism of disrupting the transcriptional machinery. This notion is supported by the observation that CdLS and CdLS-like cell lines display gene expression dysregulation (Deardorff et al., 2012; Izumi et al., 2015; Liu et al., 2009; Mannini et al., 2015; Olley et al., 2018). In summary, pathogenic variants in the EP300 and ANKRD11 can result in a CdLS-like phenotype overlapping that of other chromatin-dysregulated disorders.

CONFLICT OF INTEREST

Authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

F.C., P.S., and A.P.: performed both exome and Sanger sequencing, and western blotting experiments; S.R., M.A., and C.M.: evaluated patients and provided clinical data; A.L., B.P., F.J.R., and J.P.: interpreted patient's data and contributed to the manuscript writing; A.M.: contributed to the concept, design, and manuscript writing. All authors commented on the draft and approved the final manuscript.