The expanding phenotypes of cohesinopathies: one ring to rule them all!

ABSTRACT

Preservation and development of life depend on the adequate segregation of sister chromatids during mitosis and meiosis. This process is ensured by the cohesin multi-subunit complex. Mutations in this complex have been associated with an increasing number of diseases, termed cohesinopathies. The best characterized cohesinopathy is Cornelia de Lange syndrome (CdLS), in which intellectual and growth retardations are the main phenotypic manifestations. Despite some overlap, the clinical manifestations of cohesinopathies vary considerably. Novel roles of the cohesin complex have emerged during the past decades, suggesting that important cell cycle regulators exert important biological effects through non-cohesion-related functions and broadening the potential pathomechanisms involved in cohesinopathies. This review focuses on non-cohesion-related functions of the cohesin complex, gene dosage effect, epigenetic regulation and TGF-β in cohesinopathy context, especially in comparison to Chronic Atrial and Intestinal Dysrhythmia (CAID) syndrome, a very distinct cohesinopathy caused by a homozygous Shugoshin-1 (SGO1) mutation (K23E) and characterized by pacemaker failure in both heart (sick sinus syndrome followed by atrial flutter) and gut (chronic intestinal pseudo-obstruction) with no intellectual or growth delay. We discuss the possible impact of SGO1 alterations in human pathologies and the potential impact of the SGO1 K23E mutation in the sinus node and gut development and functions. We suggest that the human phenotypes observed in CdLS, CAID syndrome and other cohesinopathies can inform future studies into the less well-known non-cohesion-related functions of cohesin complex genes.

Abbreviations: AD: Alzheimer Disease; AFF4: AF4/FMR2 Family Member 4; ANKRD11: Ankyrin Repeat Domain 11; APC: Anaphase Promoter Complex; ASD: Atrial Septal Defect; ATRX: ATRX Chromatin Remodeler; ATRX: Alpha Thalassemia X-linked intellectual disability syndrome; BIRC5: Baculoviral IAP Repeat Containing 5; BMP: Bone Morphogenetic Protein; BRD4: Bromodomain Containing 4; BUB1: BUB1 Mitotic Checkpoint Serine/Threonine Kinase; CAID: Chronic Atrial and Intestinal Dysrhythmia; CDK1: Cyclin Dependent Kinase 1; CdLS: Cornelia de Lange Syndrome; CHD: Congenital Heart Disease; CHOPS: Cognitive impairment, coarse facies, Heart defects, Obesity, Pulmonary involvement, Short stature, and skeletal dysplasia; CIPO: Chronic Intestinal Pseudo-Obstruction; c-kit: KIT Proto-Oncogene Receptor Tyrosine Kinase; CoATs: Cohesin Acetyltransferases; CTCF: CCCTC-Binding Factor; DDX11: DEAD/H-Box Helicase 11; ERG: Transcriptional Regulator ERG; ESCO2: Establishment of Sister Chromatid Cohesion N-Acetyltransferase 2; GJC1: Gap Junction Protein Gamma 1; H2A: Histone H2A; H3K4: Histone H3 Lysine 4; H3K9: Histone H3 Lysine 9; HCN4: Hyperpolarization Activated Cyclic Nucleotide Gated Potassium and Sodium Channel 4;p HDAC8: Histone deacetylases 8; HP1: Heterochromatin Protein 1; ICC: Interstitial Cells of Cajal; ICC-MP: Myenteric Plexus Interstitial cells of Cajal; ICC-DMP: Deep Muscular Plexus Interstitial cells of Cajal; I_f: Pacemaker Funny Current; IP3: Inositol trisphosphate; JNK: C-Jun N-Terminal Kinase; LDS: Loeys-Dietz Syndrome; LOAD: Late-Onset Alzheimer Disease; MAPK: Mitogen-Activated Protein Kinase; MAU: MAU Sister Chromatid Cohesion Factor; MFS: Marfan Syndrome; NIPBL: NIPBL, Cohesin Loading Factor; OCT4: Octamer-Binding Protein 4; P38: P38 MAP Kinase; PDA: Patent Ductus Arteriosus; PDS5: PDS5 Cohesin Associated Factor; P-H3: Phospho Histone H3; PLK1: Polo Like Kinase 1; POPDC1: Popeye Domain Containing 1; POPDC2: Popeye Domain Containing 2; PP2A: Protein Phosphatase 2; RAD21: RAD21 Cohesin Complex Component; RBS: Roberts Syndrome; REC8: REC8 Meiotic Recombination Protein; RNAP2: RNA polymerase II; SAN: Sinoatrial node; SCN5A: Sodium Voltage-Gated Channel Alpha Subunit 5; SEC: Super Elongation Complex; SGO1: Shogoshin-1; SMAD: SMAD Family Member; SMC1A: Structural Maintenance of Chromosomes 1A; SMC3: Structural Maintenance of Chromosomes 3; SNV: Single Nucleotide Variant; SOX2: SRY-Box 2; SOX17: SRY-Box 17; SSS: Sick Sinus Syndrome; STAG2: Cohesin Subunit SA-2; TADs: Topology Associated Domains; TBX: T-box transcription factors; TGF-β: Transforming Growth Factor β; TGFBR: Transforming Growth Factor β receptor; TOF: Tetralogy of Fallot; TREK1: TREK-1 K(+) Channel Subunit; VSD: Ventricular Septal Defect; WABS: Warsaw Breakage Syndrome; WAPL: WAPL Cohesin Release Factor.

Introduction

Mitotic or meiotic cell division are essential for the functioning and the reproduction of the living. The cell cycle is divided into four phases, i.e. the G1, S, G2 and M phases. The M phase is itself divided into several stages termed prophase, metaphase, anaphase, and telophase, which is followed by the cytokinesis where the cytoplasm splits to form two daughter cells. Until the sister chromatids segregate at the anaphase, they are held together by a multiprotein complex, the cohesin. The functions of the cohesin complex in mitotic and meiotic division need to be precisely orchestrated in time and space (reviewed in more detail in [1,2]). Additionally, new biological roles of the cohesin complex and its partner proteins have emerged, including DNA damage checkpoint and repair, DNA replication, gene transcription regulation, and chromatin structure [3–6]. Such functions are compatible with the observation that human patients with mutations in components of the cohesin complex or its regulators, a condition named cohesinopathy, exhibit predominantly developmental phenotypes with mild or no defects in cell cycle, which are not sufficient to cause a deleterious phenotype [7–9]. A misshaping of the chromatin configuration in the proximity of genes may affect their transcription at the time of a cell fate decision or alter cellular functions [4,10]. The expanding number of genes and proteins involved in the clinical spectrum of cohesinopathies generates novel hypotheses on how the cohesin complex and its regulators impact organ development and maintenance before and after birth.

The purpose of this review is to discuss the recent findings about the molecular pathways underlying cohesinopathies, specifically exploiting contrasting genotype-phenotype correlations of CAID syndrome, a very distinct cohesinopathy characterized by sick sinus syndrome (SSS) and chronic intestinal pseudo-obstruction (CIPO), versus other cohesinopathies. We exploit the fact that the phenotype of CAID syndrome, a monogenic disease caused by the recessive founder mutation SGO1 K23E, cannot be fully reconciled with a perturbation of the canonical function of this gene in cell cycling alone. We review commonalities and differences in cohesinopathies and CAID syndrome with the goal to inform future directions of research, with an emphasis on non-cohesion-related functions of the cohesin complex.

Structure and role of the cohesin complex and its regulators in chromosome segregation

The mitotic cohesin complex consists of four subunits: Structural Maintenance of Chromosomes 1 (SMC1) and Structural Maintenance of Chromosomes 3 (SMC3) (which form a heterodimer), RAD Cohesin Complex Component (RAD21 in mitosis, REC8 Meiotic Recombination Protein in meiosis) and Cohesin Subunit SA (STAG) (STAG1/STAG2 in mitosis and STAG3 in meiosis) which together establish the cohesin ring structure (Figure 1). Loading and unloading of the cohesin complex on chromosomes is ensured by a “dual gate” mechanism. The NIPBL Cohesin Loading Factor/MAU Sister Chromatid Cohesion Factor (NIPBL/MAU2) complex loads the cohesin onto chromatids during the G1 phase of mitotic/meiotic cell cycle. To do so, NIPBL first needs to bind MAU2, which is located on the closed cohesin ring with the ATP-bound heads engaged [11]. Then, NIPBL activates cohesin ATPase activity to separate the heads and activate conformational change to open the cohesin ring and allow DNA entry [11]. During S phase, cohesion establishment is intimately linked to DNA replication and requires that both sister chromatids are entrapped within the cohesin ring [11]. This ring must also stay closed to prevent premature release of sister chromatids. This function is ensured by the acetylation of SMC3 by Establishment of Sister Chromatid Cohesion N-Acetyltransferase 2 (ESCO2), making the cohesin complex insensitive to WAPL Cohesin Release Factor (WAPL) [11–13]. In vertebrates, the cohesin complex is divided into two pools that are removed in two stages. The first pool, namely the WAPL sensitive pool, is removed from chromosome arms in mitotic prophase by the PDS5 Cohesin Associated Factor (PDS5)/WAPL complex through phosphorylation of SA and sororin by Polo Like Kinase 1 (PLK1) and Cyclin Dependent Kinase 1 (CDK1), respectively [11]. Sororin phosphorylation disrupts its interaction with PDS5, which is now available to interact with WAPL and release cohesin [11,14]. At the centromeres, the cohesin complex is stabilized by SGO1, which recruits the protein phosphatase Protein Phosphatase 2 (PP2A) to prevent SA subunit phosphorylation [15]. PP2A dephosphorylates sororin, maintaining is association with PDS5 [11]. Cohesin is also maintained in the centromeric region by the inhibition of separase by securin [16]. Once the alignment of sister chromatids is complete, the anaphase promoter complex (APC) ubiquitinates securin, thereby releasing separase that will cleave RAD21 and dissociate the rest of the cohesin complex to the centromeres to allow segregation of sister chromatids [16]. At the end of mitosis, ESCO2-induced acetylation of SMC3 during S phase is reversed by Histone Deacetylase 8 (HDAC8), SMCs are recycled and re-loaded onto chromatin [17,18].

Figure 1.

Components (SMC1, SMC3, RAD21, and STAG1/2) and partners of the cohesin complex. Colors indicate proteins whose mutated forms are involved in cohesinopathies. Blue: CdLS (NIPBL, HDAC8, SMC1A, SMC3, RAD21, BRD4, ANKRD11), Orange: RBS (ESCO2), Pink: WBS (DDX11), Red: ATRX (ATRX), Yellow: CHOPS (AFF4), Green: STAG2-related X-linked Intellectual Deficiency (STAG2) and Purple: CAID syndrome (SGO1) .

Cohesinopathies: common and distinct phenotypes

Pathologies arising from mutations in the cohesin complex or its regulators are collectively called cohesinopathies, presenting both phenotypic similarities and differences.

Among these syndromes, the best defined and common is the Cornelia de Lange syndrome (CdLS, OMIM #122,470, # 300,590, # 610,759, # 614,701 and # 300,882) with a prevalence of 1/100 000 to 1/10 000 live births [19,20]. CdLS is caused by dominant mutations in genes encoding for the cohesin regulators NIPBL (~50–70%) and HDAC8 (~4%) or structural components such as SMC1A (3–5%), SMC3 (3%) and RAD21 (~1%) [21,22]. More recently, mutations in Bromodomain Containing 4 (BRD4) (1 individual), a NIPBL interactor, and Ankyrin Repeat Domain 11 (ANKRD11) (5 individuals), an inhibitor of ligand-dependent activation of transcription, were also found in CdLS [21]. The functional interaction of ANKRD11 with the cohesin complex is currently unknown [21]. Patients with CdLS mainly display developmental delay and short stature with particular phenotypic characteristics that are listed in Table 1. Mutations in genes other than NIPBL tend to be less severe. Mosaicism in NIPBL variants has been found in 15–20% of patients with classic CdLS features and is thought to be main mechanism responsible for variable expressivity [22,23]. The presence of cardiovascular and gastrointestinal disorders is also particularly frequent in CdLS. Indeed, tetralogy of Fallot (TOF), the most common form of cyanotic congenital heart disease consisting of four common heart defects (ventricular septal defect (VSD), pulmonary stenosis, aortic displacement, and right ventricular hypertrophy), accounts for about 50% of cardiac malformations present in CdLS patients. They may also have VSD or atrial septal defect (ASD), valvular abnormalities, and patent ductus arteriosus (PDA) [24]. In contrast to CAID syndrome, sick sinus syndrome has never been observed in CdLS. At gastrointestinal level, some patients with CdLS also suffer from gastroesophageal reflux disease, constipation, and feeding difficulties [21,25].

Table 1.

Summary of clinical characteristics of Cohesinopathies.

					Cohesinopathies
	CdLS[1–30]							CIPO[31]	RBS[32–35]	WABS[36–39]	ATRX[40–43]	CHOPS[44]	STAG2[45–49]		CAID[50]
	Dominant Mutation NIPBL	X-linked Dominant Mutation HDAC8		X-linked Dominant Mutation SMC1A	Dominant Mutation SMC3	Dominant Mutation RAD21	Recessive Mutation RAD21		Recessive Mutation ESCO2	Recessive Mutation DDX11	X-linked Recessive Mutation ATRX	De novo Dominant Mutation AFF4	X-linked Recessive Mutation STAG2	De novo Dominant Mutation STAG2	Recessive Mutation SGO1
Pathomecanism
Function	Loss	Loss		Loss	Loss	Loss	Loss		Loss	Loss	Loss	Gain	Loss	Loss/Missence	Gain/Change?
Mental retardation	+	+		+	+	+	-		+	+	+	+	+	+	-
Intellectual disabilities	Severe/ Moderate/ Mild	Severe/ Moderate/ Mild		Severe/ Moderate	Severe/ Moderate	Severe/Mild	-		Severe/Mild	Severe/ Mild	Severe/ Moderate/ Mild	+	Severe/ Moderate	Severe/ Moderate	-
Speech and language	+	+		±	-	±	-		±	+	+	-	-	+	-
Psychomotor retardations	+	+		+	-	±	-		±	+	+	-	-	-	-
Behavior and neurogical problems	+	±		±	-	±	±		-	-	±	-	-	+	-
Self-injurious behavior	+	±		±	±	-	-		-	-	-	-	-	-	-
Attention deficit disorder	+	-		-	±	±	-		-	-	-	-	-	+	-
Epilepsy	±	-		±	-	±	±		-	-	-	-	-	-	-
Hyperactivity	+	±		-	±	±	-		-	-	-	-	-	+	-
Autism	+	±		±	±	+	-		-	-	+	-	-	-	-
Aneurysm	-	-		-	-	-	-		-	-	-	-	-	-	±
Growth/Developmental anomalies (pre- and postnatal)	+	±		+	+	+	-		+	+	+	+	+	+	-
Growth retardation	+	±		±	+	+	-		+	+	+	-	+	+	-
Short stature	+	±		±	+	+	-		+	+	+	+	+	±	-
Skeletal anomalies	+	±		±	-	±	±		+	-	+	+	-	±	-
Limb malformations	+	+		+	+	±	-		+	-	+	+	-	+*	-
Upper limb truncations	±	-		-	-	-	-		+	-	-	-	-	-	-
Lower limb truncations	-	-		-	-	-	-		+	-	-	-	-	-	-
Small extremities/Brachydactyly	+	+		+	+	±	-		+	-	+	+	-	-	-
Craniofacial anomalies	+	+		+	+	+	-		+	+	+	+	+	+	-
Microcephaly	+	±		±	+	±	-		+	+	+	±	-	+	-
Brachycephaly	-	+		±	+	±	-		±	-	-	-	-	-	-
Proeminent eyes	-	-		-	-	-	-		+	-	-	±	+	-	-
Spaced eyes	±	±		-	-	-	-		+	-	+	-	-	±	-
Arched eyebrows	+	+		+	+	±	-		-	-	-	±	-	-	-
Synophrys	+	+		+	+	+	-		-	-	-	+	-	-	-
Long eyelashes	+	±		+	+	±	-		-	-	-	±	-	±	-
Low-set ears	+	±		±	±	±	-		±	-	+	-	-	±	-
Upturned nose/nostrils	+	+		+	+	±	-		-	-	+	-	-	-	-
Nasal bridge anomalies	+	±		+	+	+	-		±	-	±	-	-	±	-
Micrognathia	+	+		+	±	±	-		+	-	-	-	-	-	-
Philtrum anomalies	+	+		+	+	+	-		±	±	-	±	-	±	-
Thin upper lip	+	±		+	+	+	-		-	-	+	-	-	±	-
Downturn mouth	+	+		+	+	-	-		±	-	+	±	-	-	-
Widely spaced teeth	+	+		±	±	-	-		-	-	±	-	-	-	-
Palate anomalies	+	±		±	±	±	-		+	+	±	±	+	+	-
Cleft lip	-	-			-	-	-		+	-	-	-	-	±	-
Low posterior or anterior hairline	+	+		+	+	±	-		-	-	-	-	-	±	-
Frontal baldness	-	-		-	-	-	-		-	-	-	-	+	-
Hirsutism	+	+		±	+	±	-		-	-	-	-	-	-	-
Vision defects	±	+		±	±	±	-		±	-	±	±	-	-	-
Cataracts	-	-		-	±	-	-		±	-	-	±	-	-	-
Myopia	±	±		±	±	±	-		-	-	±	±	-	-	-
Ptosis	±	±		±	±	±	-		-	-	-	-	-	-	-
Glaucoma	-	-		-	±	-	±		-	-	-	-	-	-	-
Ear defects	+	+		±	+	±	-		-	+	±	±	+	+	-
Hearing loss	+	+		±	+	±	-		-	+	±	±	+	±	-
Genitalia defects	±	±		-	±	±	-		±	-	±	-	-	-	-
Hypoplasia	±	±		-	±	-	-		-	-	±	-	-	-	-
Microphallus	±	±		-	-	-	-		-	-	±	-	-	-	-
Enlarged phallus	-	-		-	-	-	-		±	-	-	-	-	-	-
Undescended testes	±	±		-	-	±	-		±	-	±	-	-	-	-
Hypospadia	±	±		-	-	±	-		-	-	±	-	-	-	-
Gastrointestinal defects	+	+		+	+	±	+		-	-	+	+	-	-	+
Gastroesophageal reflux disease	+	+		+	+	±	-		-	-	±	+	-	-	-
Constipation	-	-		±	-	±	+		-	-	±	±	-	-	+
Chronic intestinal pseudo-obstruction	-	-		-	-	-	+		-	-	-	-	-	-	+
Feeding difficulties	+	+		+	+	-	+		-	-	-	-	-	-	+
Heart defects	+	±		±	±	±	±		±	±	±	+	-	±	+
Atrial septal defect	±	±		±	±	-	-		±	±	±	-	-	-	-
Ventricular septal defect	±	±		±	±	-	±		±	±	±	+	-	±	-
Atrial dysrhythmia	-	-		-	-	-	-		-	-	-	-	-	-	+
Sick sinus syndrome	-	-		-	-	-	-		-	-	-	-	-	-	+
Valves anomalies	±	±		±	-	-	±		-	-	±	-	-	-	+
Patent ductus arteriosus	±	±		±	±	-	-		±	±	±	+	-	-	-
Tetralogy of Fallot (TOF)	+	±		±	±	±	-		-	±	±	-	-	-	-
Congenital heart disease	+	±		±	±	±	-		-	-	±	-	-	±	-
Blood disorders	-	-		-	-	-	-		-	+	+	-	-	-	-
α-Thalassemia	-	-		-	-	-	-		-	-	+	-	-	-	-
Fanconi Anemia	-	-		-	-	-	-		-	+	-	-	-	-	-

Legend: +: present in most cases, ±: present in certain cases, -: not present or not specified.

Robert’s syndrome (RBS, OMIM # 268,300) is a rare autosomal recessive disorder resulting from mutations in the gene encoding ESCO2, a cohesin complex regulator [26,27]. Some RBS clinical features overlap with CdLS, including mild to severe growth deficiency, limb malformations, craniofacial anomalies, and mental retardation, but there are obvious differences between both syndromes (Table 1). RBS patients also exhibit heart defects (VSD, ASD, and PDA) but no gastrointestinal defects have been reported [28]. At the cellular level, RBS patients display mitotic defects, including lagging chromosomes, aneuploidy and micronuclei formation (nucleoli fragmentation) due to premature sister chromatids segregation, especially in heterochromatin regions [9,29].

Less well characterized cohesinopathies include Warsaw breakage syndrome (WABS, OMIM #613,398), resulting from recessive mutations in the DEAD/H-Box Helicase 11 (DDX11) gene encoding an Iron-Sulfur DNA helicase essential for chromatid cohesion, and the X-linked α-thalassemia mental retardation syndrome (ATRX, OMIM # 300,032) which is caused by dominant mutations in ATRX Chromatin Remodeler (ATRX) gene, a chromatin remodeler that contributes to chromosome dynamics during mitosis [30–32]. WABS patients share phenotypic characteristics with CdLS and RBS (Table 1). Like RBS, they also display sister chromatin cohesion defects with typical railroad track appearance of chromosomes during metaphase, due to lack of cohesion in centromeric regions [30]. Such a centromeric cohesion defect is considered to be pathognomonic for cohesinopathies at cellular level. At clinical level, some patients with WABS also have TOF and VSD, but no intestinal involvement has been described to date [30,33]. WABS also present a unique phenotype, namely Fanconi anemia, which is not found in any other cohesinopathy [30]. ATRX syndrome also exhibits a unique phenotype, namely the α-Thalassemia syndrome [34]. At cardiac level, ATRX patients may also have VSD, ASD, TOF, congenital heart disease (CHD), valvular abnormalities and PDA [34]. Some patients with ATRX also suffer from gastroesophageal reflux disease and constipation [35].

Gain-of-function mutations in AF4/FMR2 Family Member 4 (AFF4) has been found in patients with clinical features similar to CdLS. This syndrome named Cognitive impairment and coarse facies, Heart defects, Obesity, Pulmonary involvement, Short stature and skeletal dysplasia (CHOPS, OMIM # 616,368) [36]. The main heart phenotypes include VSD and PDA [36]. Like the ATRX syndrome, the main gastrointestinal manifestations of CHOPS syndrome are gastroesophageal reflux and constipation [36]. Of note, AFF4 is not a part of the cohesin complex but rather a factor of the super elongation complex (SEC), which regulates the mobilization of the paused RNA polymerase II (RNAP2) machinery. However, the cohesin complex and the SEC complex interact together to promote transition of paused RNAP2 to transcriptional elongation. It has been shown that the cohesin complex promotes this transition, possibly through MAU2/NIPBL complex which is mediated by AFF4 and SEC [37,38]. Thus, the mutation of the AFF4 gene disrupts RNAP2 elongation, due to altered genome-wide binding of AFF4 and cohesin [36].

More recently, a familial germline mutation of the cohesin complex subunit STAG2 has been identified in five patients [7]. Since this mutation also co-segregates with syndromic mental retardation in a characteristic X-linked recessive pattern, this syndrome was named STAG2-related X-linked Intellectual Deficiency (OMIM # 301,022) [7]. Clinical features include moderate intellectual deficiency, short stature, sensory hearing deficiency, large nose, prominent ears, and frontal baldness (Table 1). No defect in sister chromatids cohesion was observed in patient fibroblasts [7]. However, they exhibit higher percentage of G2/M cells and upregulation of genes involved in cell division, mitotic regulators, and DNA replication factors [7]. De novo heterozygous and hemizygous mutations of STAG2 in patients with similar phenotype have also been identified, including one patient with VSD (Table 1) [39–42].

We have described a syndrome termed CAID for Chronic Atrial and Intestinal Dysrhythmia (OMIM # 616,201) [8]. All affected patients are homozygous carriers of a recessive founder mutation in SGO1 (K23E), a modulator of the cohesin complex. Clinically, this disorder can be viewed as a progressive failure of pacemaking tissues, the sinoatrial node (SAN) in the heart and the interstitial network of Cajal (ICC) in the gut. CAID patients share a combination of two distinct disorders, the sick sinus syndrome and the chronic intestinal pseudo-obstruction. Consistent with the known roles of SGO1, chromosomes in cultured dermal fibroblasts from affected individuals show the typical railroad appearance of a centromeric cohesion defect [8]. Interestingly, the percentage of cells in mitosis is significantly higher in cells from CAID patients and we also identified upregulation of genes involved in cell cycle regulation, as is the case in STAG2-related X-linked Intellectual Deficiency [7,8,43]. Additionally, we found enhanced activation of canonical Transforming Growth Factor β (TGF-β) signaling, DNA methylation, chromatin compaction and disruption of potassium currents [43]. Intestinal tissues derived from affected individuals display pathological changes in both the enteric nervous system and smooth muscles including disruption of fiber architecture of smooth muscles and extensive fibrosis, which is characteristic of enhanced TGF-β signaling [8]. Of interest, a recessive mutation of RAD21 has recently been identified in a consanguineous family of patients with CIPO (OMIM # 611,376) [44]. Within that family, megaduodenum, Barrett’s long-segmented esophagus, and cardiac abnormalities of varying severity including valvular abnormalities and VSD have been observed [45]. In zebrafish, this mutation leads to disturbance of intestinal transit and enteric neuron development, suggesting that this loss of function of RAD21 causes neurogenic CIPO through disruption of the structure and function of enteric innervation [44].

Even though there is some clinical overlap between CAID and other cohesinopathies (valve defects and gastrointestinal defects), CAID patients do not exhibit intellectual and growth delays. Furthermore, whereas the majority of cohesinopathies are manifest at birth with major developmental defects, this is not the case for the syndromes associated with recessive mutations in SGO1 (CAID) and RAD21 (CIPO). Patients are healthy at birth and the earliest phenotype occurs about at 6 years of age [8]. The lack of major mitotic consequences in CAID syndrome and other cohesinopathies – such as cancer or premature aging – indicates that the cohesion-related function of the cohesin complex, which is to bundle DNA together during all cell cycle phases, is maintained and suggests that non-cohesion-related mechanisms underly the observed phenotypes.

Non-cohesion-related functions of the cohesin complex and involvement in cohesinopathies

In addition to its stabilizing role during cell cycling, the cohesin complex is also involved in double-strand DNA break repair, DNA replication, centrosome duplication, chromatin architecture and transcriptional regulation (Figure 2) [46]. Several models have emerged to explain these novel functions and their roles in cohesinopathies.

Figure 2.

Functions of the cohesin complex in cohesion and in non-cohesion-related mechanisms.

First, studies in Drosophila and humans indicate that cohesin can play a dual role both in activation and repression of gene transcription. On the one hand, it can act as a physical barrier to prevent interaction between regulatory regions and inhibit gene transcription [47,48]. The cohesin complex performs this function via the formation of chromosomal loops or topology-associated domains (TADs) that inhibit the interaction between the enhancer and promoter of target genes [48–50]. On the other hand, opposite effects can also occur by inhibiting the interaction between an inactivator and a promoter to activate gene transcription [49]. This regulation requires the insulator protein CCCTC-Binding Factor (CTCF), which recruits the cohesin complex and defines the loop boundaries [48–51]. In order to block the enhancer/promoter (or inactivator/promoter) interaction, the CTCF binding site must be in between the two [52]. In mammals, around 90% of cohesin binding loci are at CTCF binding sites [48,50]. However, cohesin can also regulate gene transcription independently of CTCF by forming a complex with the transcriptional coactivator Mediator to activate gene transcription [4]. This complex brings enhancer and promoter close together by forming a loop connecting two DNA segments [4]. Of note, loops and TADs can also activate gene transcription in the same way [49].

In another non-cohesion-related role, cohesin and NIPBL also mediate interchromosomal and intrachromosomal interactions by forming long-distance DNA loops to regulate gene transcription [4,53–55]. Also, NIPBL can bind DNA independently of the cohesin complex. This function may be involved in the observed developmental phenotype of CdLS patients since the cohesin-independent binding of NIPBL is reduced in their cells [56]. Genes bound by NIPBL include transcription factors important for development and have been reported previously to be differentially expressed in CdLS [56]. In pluripotent stem cells, cohesin forms a bridge between two enhancers of SRY-Box 2 (SOX2) and SRY-Box 17 (SOX17) genes in the absence or presence of CTCF, respectively [57]. When cells differentiate into endodermal and mesodermal cells, this bridge is disrupted by the reprogramming factor Octamer-Binding Protein 4 (OCT4). The enhancer of SOX17 is free to fold onto the promoter to interact in the presence of cohesin allowing transcription of SOX17, which is required for mesendodermal lineage commitment and subsequent induction of cardiac differentiation [58]. A down-regulation of NIPBL prevents these bridge changes and impairs cardiogenesis, suggesting that NIPBL mutations involved in CdLS play a direct role in heart phenotypes identified [57]. The cohesin complex is dynamic in its association with chromatin through its different subunits [59]. Therefore, mutations of this complex can prevent cohesin binding to specific effectors or disrupt loops and TADs, leading to deficient gene regulation and transcription.

In addition, cohesin regulates sub-nuclear localization of certain genes leading to a subsequent impact on their expression [60–62]. Considering the cyclic association of the cohesin complex with the chromosomes, it is quite conceivable that it plays a role in DNA organization in the nucleus and nucleoli. Cohesinopathy-associated mutations of NIPBL and ESCO2 homologs in yeast disrupt subnuclear organization of the chromatin resulting in chromosomal decondensation and aberrant nucleolar morphology [63]. Therefore, if the cohesin complex regulates subnuclear localization of DNA sequences involved in embryogenesis, the mutations in the cohesin complex might disrupt this process and thus, explain the observed developmental dysfunctions in cohesinopathies [64]. It is however likely that some genes may be more sensitive than others to the 3D configuration of chromatin depending upon their localization on chromosomes. A recent study reported that the cohesin binding sites within the genome differ in two investigated tissues of the same individual [65]. Cohesin also regulates ribosomal RNA synthesis in nucleoli, promotes protein translation, and may affect chromosome arrangement throughout nucleoli [66]. Cohesinopathy mutations of ESCO2 and SMC1A homologs result in lower production of ribosomal RNA, which is expected to restrain ribosome biogenesis and protein translation [66]. Most, but not all cohesinopathy patients are of short stature and weight already at birth (CdLS, RBS, WABS, STAG2 de novo mutations) [21,28,30,33,39,40]. No such anomalies were reported in ATRX, CHOPS, STAG2 germline mutation, and CAID patients.

Gene dosage effects in cohesinopathies

In Cornelia de Lange syndrome, gene dosage effects play an important role and may indeed underlie the severity of the syndrome. Most mutations responsible for CdLS are point mutations, frame shift or single-nucleotide variants (SNVs) in NIPBL gene. However, rare genomic rearrangements such as large deletions have also been seen in some CdLS patients [44,67,68]. In increasing order of severity, patients with missense mutations have a milder phenotype than those with nonsense mutations or large deletions, suggesting that NIPBL is sensitive to gene dosage [69].

CAID syndrome also seems to be sensitive to gene dosage, but in a different way. Since CAID follows an autosomal recessive inheritance mode, heterozygous individuals for the SGO1 K23E mutation do not show characteristic disease phenotypic manifestations. However, SGO1 dosage appears to be of fundamental importance in CAID syndrome pathology. Indeed, RNAseq analysis performed on CAID patient dermal fibroblasts reveals overexpression of mutated SGO1 and several genes involved in cell cycle regulation and cohesin complex. Whether this upregulation is causal or secondary remains to be determined [43]. It can be speculated that overexpression of mutated allele coupled to the expression of the wild type allele seems to be sufficient in heterozygous carriers to attenuate direct effects of the SGO1 K23E mutation and maintain a normal phenotype. On the other hand, it is also possible that this overexpression is causal and leads to CAID phenotype. This hypothesis is further supported since copy number variants (CNVs) of STAG2 (microduplications and triplication) in individuals with intellectual disability lead to gain of STAG2 mRNA and protein, dysregulating neuronal gene networks relevant for the disease in patient cells [70]. Thus, a similar mechanism may be involved in CAID pathology since we identified upregulation of SGO1 mRNA and correlated impact on phenotype-compatible pathways [43].

Epigenetic regulation in cohesinopathies

Through life, organisms must orchestrate the expression of genes in time and space to perform certain functions. This depends on transcription regulation, which itself results from epigenetic regulation. Euchromatin, or open chromatin, is associated with gene expression, whereas heterochromatin, or closed chromatin, is associated with gene repression. Several modifications regulate chromatin state, including DNA methylation, post-translational modifications of histones, nucleosomes positioning, and higher chromatin organization. Over the recent years, epigenetic regulation arose as a new pathomechanism in several human diseases and a promising therapeutic avenue. The implication of epigenetic regulation in cohesinopathies have also started to emerge. NIPBL haploinsufficient cardiomyocytes cell lines show dysregulation of many chromatin modifiers, including histone modifiers, nucleosome, and DNA bending complexes, providing evidence for NIPBL’s direct or indirect role in epigenetic regulation in CdLS [71]. Additionally, ATRX syndrome exhibits specific epigenetic signatures. ATRX regulates histone H3 positioning at the telomeric and pericentromeric heterochromatic regions, histones Histone H3 Lysine 4 (H3K4) and Histone H3 Lysine 9 (H3K9) methylation detection, and DNA methylation [72–74]. Methylome characterization of ATRX syndrome patients show a unique enrichment of differentially methylated genes in or close to pericentromeric or telomeric chromosomal regions, suggesting that ATRX mutation disrupts its function in heterochromatin dynamic regulation[75]. CAID patient-derived fibroblasts also exhibit a global hypermethylation and chromatin closing signatures, suggesting that epigenetic dysregulation is also part of CAID pathogenesis. This clearly demonstrates the involvement of epigenetics in cohesinopathy pathomechanisms.

TGF-β and cohesinopathies

The TGF-β family includes a variety of cytokines involved in development, differentiation, tissue repair and tumorigenesis. The regulation of TGF-β signaling is based on the transmission of intracellular signals via the TGF-β receptors (TGFBR) TGFBRI and TGFBRII, two transmembrane threonine/serine kinase receptors that form heteromeric complexes. The canonical pathway of TGF-β signaling depends on SMAD Family Member (SMAD) proteins. Activation by TGF-βs, activins and Nodal results in phosphorylation of SMAD2/3, while activation by Bone Morphogenetic Proteins (BMPs) induces phosphorylation of SMAD1/5/8[76]. TGFBRI is responsible for SMAD2/3 phosphorylation which then form a complex with SMAD4. The SMAD2/3/4 complex enters the nucleus where it regulates transcription of target genes by interaction with different cofactors[76]. SMAD1/5/8 also form complexes with SMAD4 to regulate different target genes[76]. Noncanonical TGF-β signaling is activated in a SMADs-independent manner and includes Mitogen-Activated Protein Kinase (MAPK), P38 MAP Kinase (p38) and C-Jun N-Terminal Kinase (JNK) signaling. Main effectors of those pathways are Extracellular Signal-Regulated Kinase (ERK) ERK1/2, p38, and JNK1/2/3, respectively,[77].

It is well known that TGF-β signaling is a key cascade involved in cardiomyopathy, arrhythmias and valvular diseases [78,79]. Among the TGF-β-opathies, we find the Marfan syndrome (MFS) and the Loeys-Dietz Syndrome (LDS) in which increased TGF-β signaling causes dilation and dissection of the aorta [80–82]. Increased TGF-β signaling plays a role in a large number of cardiac conditions, such as hypertrophic and dilated cardiomyopathy [79,83]. In mouse models, enhanced TGF-β signaling induces fibrosis in the ventricle, impairs electrical propagation, and leads to arrhythmia development through pathological deposition of extracellular matrix proteins [84–86]. The study of TGF-β signaling in CAID syndrome revealed an up-regulation of the canonical pathway and its uncoupling from the noncanonical one, suggesting that the SGO1 K23E mutation favors the canonical TGF-β pathway through an unknown mechanism[43]. The serine/threonine kinase BUB1 Mitotic Checkpoint Serine/Threonine Kinase (BUB1), previously known to be responsible for SGO1 localization at the centromeres, has been found to be essential for the recruitment of TGFBRI and subsequent regulation of canonical and noncanonical TGF-β signaling[87]. Since we previously showed up-regulation of BUB1 expression in CAID syndrome, it is therefore conceivable that the enhanced canonical TGF-β signaling in CAID syndrome is related to BUB1 overexpression, which may provide a clinically relevant crosstalk between TGF-β signaling and the cohesin complex[43]. Since BUB1 is required for SGO1 localization at centromere, it seems plausible that BUB1 can also recruit SGO1 in the cytoplasm. Therefore, we hypothesize that cytoplasmic SGO1 might play a role in canonical TGF-β pathway by sequestrating BUB1 to the canonical pathway only, possibly by steric interference. There is little data on the implication of the TGF-β pathway in other cohesinopathies. Interestingly, the BMP pathway was among the most significantly affected cascades in a Nipbl^± CdLS mouse model[88]. Compound effects in the Nipbl^±, Bmp4^± double heterozygous mouse cause skeletal abnormalities reminiscent of CdLS, but no overt limb truncations[89]. This implies that cohesin-BMP crosstalk explains some, but not all, aspects of certain CdLS phenotypes.

SGO1 functions

First identified through a screen for genes required for centromeric cohesion in yeast, the main function of SGO1 is to protect the cohesin complex from cleavage by separase during meiotic divisions[90]. Separase is a protease that triggers anaphase by cleaving cohesin within the RAD21/REC8 subunit. Related to this task, SGO1 also plays a similar role during mitosis by ensuring proper chromosome segregation[15]. There are different isoforms of the SGO1 protein in humans, namely the A1, A2, B1, B2, C1, C2, D1 and P1, which differ mainly by the exclusion, in whole or in part, of the exon 6 and/or exon 9 (Figure 3(a)). The isoforms A2 and C2 are localized at the centrosomes while the isoforms A1 and C1 locate at the centromere to protect the centromeric cohesion[91]. Centrosome localization has been shown to be dependent on a small C-terminal peptide encoded by exon 9 [91]. Our clinical observations in CAID syndrome do not provide any evidence of a progeria-like phenotype, or of premature development of cancer. In addition, we have described two homozygous females, who successfully carried spontaneous pregnancies to term[8]. Thus, mitotic or meiotic dysfunctions do not seem to be in the foreground in CAID syndrome.

Figure 3.

(a) Schematic representation of SGO1 isoforms. (b) Schematic structure of SGO1 (isoform A2) and binding sites. Numbers refer to SGO1 exons. In sillico 3D reconstruction and electronic cloud of (c) the native SGO1 protein (d) the K23E mutated SGO1 protein. I-TASSER server was used for protein structure and function prediction. The K23E mutation alters the 3D configuration and the electronic cloud in the N-terminal region.

Additional roles for SGO1 have emerged recently, such as the protection of centrioles cohesion, and its interaction with Heterochromatin Protein 1 (HP1) (Figure 3(b)) [92,93]. Furthermore, SGO1 can also interact with Baculoviral IAP Repeat Containing 5 (BIRC5), an anti-apoptotic gene, through a phospho-mimic sequence at its amino-terminus (Figure 3(b))[94]. We have modeled the SGO1 K23E mutation and found it has a profound effect on this region, possibly perturbing the binding to a selected set of amino-terminal interaction partners (PP2A, BIRC5) and leaving intact interactions dependent on SGO1 carboxy-terminal domains, such as Histone H2A (H2A) and HP1 (Figure 3(b–d)) [93,95].

The participation of SGO1 in biological functions of the cohesin complex could also entail consequences at transcriptional level. As discussed previously, it has been shown over the recent years that components of the cohesin complex play important roles in long-range transcriptional regulation [96–98]. A non-cohesion-related role of SGO1 and the cohesin complex – possibly through chromatin remodeling and epigenetic marks, given the already identified interaction partners of SGO1 – in maintaining shared transcriptional modules of the SAN and the ICC observed in CAID syndrome, therefore is an intriguing possibility. This begs the question what the commonalities are in SAN and ICC transcriptional regulation both in development and during adult life, and how the cohesin complex is involved in them.

Effects of SGO1 alterations in human pathologies

Loss of SGO1 in yeast leads to poor viability, and loss of its homolog in mice results in embryonic lethality [99,100]. Haploinsufficiency of Sgo1^± in mice leads to genomic instability and accelerated development of tumors after exposure to colon carcinogens but does not affect viability or fertility[99]. In humans, mutations and alterations of SGO1 expression are found in hepatocellular carcinoma, and colorectal and gastric cancers [101–103]. In addition, reduction of SGO1 expression by siRNA in cultured cells results in premature chromosome separation and multipolar spindles mitosis, which is characteristic of cohesinopathy [92,104,105]. Interestingly, although CAID patients show centromeric cohesion defects[8], they do not exhibit mitotic defects and do not have any type of cancer. This is in line with observations in other cohesinopathies, despite the stark differences in CAID and other cohesinopathy phenotypes. However, since all cohesinopathies, including CAID syndrome, are monogenic diseases with strong effects of the mutant alleles, we put forward that the associated phenotypes (as well as the differences among them) highlight divergent biological roles of the causative genes and their mutations in humans. Based on the autosomal recessive mode of inheritance of CAID syndrome, with a single causative gene and no known modifiers, we speculate that the effects of the SGO1 K23E mutation could be due to one or more of the following mechanisms; (a) a gain of function, (b) a change of function, (c) a noncanonical (not related to protection of the cohesin ring during mitosis and centrosome integrity) role for SGO1 (Figure 4). Unfortunately, the SGO1 K23E point mutant is lethal in two different mouse backgrounds (129Sv and C57Bl6N), limiting the use of mouse models available for study (data not shown).

Figure 4.

Comparison of the canonical and noncanonical findings of SGO1 K23E mutation with SGO1 haploinsufficiency in mice and other cohesinopathies to highlight the differences between each condition. Blue: Cohesinopathies. Green: CAID syndrome (SGO1 K23E mutation). Purple: Sgo1^± mice. Direct findings are indicated by arrows.

Mice haploinsufficient for Sgo1^± show proneness to cancer, chromosome cohesion defect and centrosome number defect leading to chromosome instability and prolonged mitosis and accumulation of β-amyloid in the brain (Figure 4) [99,106–108]. Specifically, brain pathology has a late onset in Sgo1^± mice[108]. However, while amyloid deposition is one of several hallmarks of Alzheimer’s disease (AD), it is not the only driver. As an example, mouse models overexpressing amyloid, such as 5xFAD and APPSWE/PSEN1deltaE9, do not fully recapitulate the late-onset neurological phenotype [109,110]. In human patients, cognitive impairment and memory loss do not closely correlate with the extent of amyloid deposition, and no association between SGO1 variants and large Late-onset Alzheimer’s disease (LOAD) human cohort has been reported to date [111,112]. Moreover, microglia pathology or cognitive deficits would need to be investigated in Sgo1^± mice to better understand the potential role of Sgo1 in brain pathologies. Interestingly, the Sgo1-lacZ reporter mouse and immunohistochemical data provide evidence for cytosolic localization of Sgo1 in the brain [113,114]. Additional imaging and functional studies of the brain in Sgo1^± mice and CAID patients will yield insight into the potential roles of SGO1 in neurological diseases.

Brains from AD patients exhibit a diffused cytosolic pattern of p-H3 localization, which has been confirmed in brain from Sgo1± mouse. However, phospho-Histone H3 (p-H3) localization is generally restricted to the nucleus in the majority of organs [115,116]. It was believed that the SGO1 localization was restricted to the nucleus as well, but we have shown its cytosolic localization in several organs[113]. Mutated SGO1 K23E is delocalized to the cytosol in CAID patient fibroblasts, in a pattern reminiscent of aberrant p-H3 localization in AD brains [8,115,116]. Since SGO1 and H3 are known to compete with each other for Survivin binding in the nucleus[94], it is conceivable that they might also compete for binding of some unknown partners in the cytoplasm and possess noncanonical roles depending on this cytoplasmic localization. Identification of these roles would have enormous potential for understanding the molecular noncanonical mechanisms involved in CAID syndrome, as well as in AD and cognitive pathologies involved in cohesinopathies.

Possible impact of SGO1 mutation on sinus node and gut development and functions

Since CAID patients exhibit phenotypes associated with pacemaker cell dysfunctions (SSS and CIPO), it is likely that the SGO1 mutation may play a role in sinus node and gut development and/or homeostasis. The function of the SAN is to initiate the action potentials responsible for heart rhythm. The sinoatrial cells depolarize spontaneously and thus trigger the cardiac action potential. Cardiomyocytes are all connected to each other via gap junctions and a specialized conduction system, which allows action potential propagation throughout the entire organ. In mice, development and physiological functions of the sinus node have been dissected in great detail (for review, see [117,118]). The transcriptional cascades required for proper embryonic development of the sinus node require T-box transcription factors (TBX) and other finely orchestrated cues to repress a genetic program that otherwise would lead to the development of working myocardium [118–120]. Lineage tracing experiments have shown that mesenchymal precursors are recruited to the myocardial lineage to form the SAN. Growth of the SAN is achieved by proliferation of a subpopulation of specified SAN primordial cells. The developmental default program for all cardiomyocytes is that of working myocardium; where it is repressed by transcription factors Tbx3 and Tbx18, the SAN will develop [119,120].

A distinct set of molecular markers has been found to be characteristic for adult sinoatrial cells, including Hyperpolarization Activated Cyclic Nucleotide Gated Potassium and Sodium Channel 4 (HCN4), which is a channel generating the pacemaker funny current (I_f), and Gap Junction Protein Gamma 1 (GJC1 or CX45), which is a gap junction protein [121,122]. Sodium Voltage-Gated Channel Alpha Subunit 5 (SCN5A), the cardiac sodium channel responsible for rapid depolarization, is highly expressed in the SAN, but also present in cardiomyocytes[123]. Mutations in SCN5A and HCN4 are linked to SSS and sinus bradycardia, respectively, in man [124,125]. Of note, gastrointestinal motility disorders have been identified in Brugada and long QT syndromes caused by SCN5A mutations [126,127]. SCN5A is also expressed in ICC and mutations in SCN5A have been reported in irritable bowel syndrome (~2%), reinforcing the notion that both Cajal cells and the SAN share selected functional similarities at electrophysiological level [128,129].

In the gut, ICC networks contribute to a large number of functions that are important for gastrointestinal motility, such as the production, coordination, and propagation of slow electric waves that control intestinal contractions, the coordination of the stimulation and propagation of slow electric wave, the signal transduction from the enteric nervous system to motor neurons to activate contraction of smooth muscle and the sensing to mechanical stretching of gastrointestinal smooth muscle[130]. The ICC associated with the myenteric plexus (ICC-MP) generates pacemaker activity via slow waves that are transmitted to the circular muscle layer [131,132]. Initiation of slow waves occurs in the proximal intestine and control the peristaltic contractions[133]. Several mechanisms have been developed to explain the generation of electrical slow waves by the interstitial cells of Cajal. First, a chloride channel would be responsible for rhythmic depolarization leading to the rise of slow electric waves[134]. The second, meanwhile, suggested that potassium channels encoded by the gene Transcriptional Regulator ERG (ERG) ensure the function of stimulator channels[135]. However, the most plausible hypothesis is that the generation of slow electric waves is due to an increase in intracellular Inositol trisphosphate (IP3) induced-calcium release via the smooth endoplasmic reticulum through a mechanism reminding the one observed in the earliest cardiac pacemaker cells located in the left inflow tract [136–138]. In fact, the ICC can generate low-frequency rhythmic transient depolarizations in response to rhythmic changes in intracellular calcium[139]. During the transmission of depolarizations to muscle layers, phase–amplitude interaction with the slow-wave occurs[139]. This causes an electrical slow-wave activity pattern of waxing and waning via coupling of ICC-MP and deep muscular plexus interstitial cells of Cajal (ICC-DMP) phase amplitude and then, a checked pattern of non-propagating motor activity occurs[139]. Thus, ICC-MP induction of rhythmic transient depolarizations is a key factor to change the motor pattern from propulsion to segmentation[139].

On the other hand, the embryologic origin and molecular orchestration of ICC in the gut is much less understood. ICC clearly share commonalities with smooth muscle and mesenchymal cells and does not arise from the neural crest (for reviews, see [140,141]). KIT Proto-Oncogene Receptor Tyrosine Kinase (C-kit) is a marker of Cajal cells throughout the intestine. Interestingly, mice with a hypomorphic allele of c-kit (W/Wv) have a severely underdeveloped ICC network and disorder of gastrointestinal motility[132]. The same mouse strain is prone to develop atrial fibrillation due to a dysregulation of mast cells, but these mice do not have sick sinus syndrome[142]. However, in a different mouse model in which a lacZ reporter disrupts the c-kit locus, c-kit is not required for the embryonic differentiation or survival of ICC, but for postnatal proliferation[143].

There are several principle modes by which the mutation of SGO1 could exert its deleterious effects on the human pacemakers. First, SGO1 may guard pacemaker functions through maintaining proper expansion of SAN and ICC precursors during embryogenesis. BIRC5, a known interactor of SGO1, is involved in the control of cardiomyocyte numbers and its cardiac-specific knock-out leads to conduction defects and SAN dysfunction [144,145]. Interestingly, we found upregulation of BIRC5 expression in CAID patients skin fibroblast at young and late passages[43]. How the upregulation of BIRC5 is associated to SSS in CAID syndrome remain unclear. Since the BIRC5 interaction site on SGO1 is near the K23E mutation (Figure 3), it is conceivable that the mutation alters BIRC5 binding. Thus, BIRC5 overexpression might be compensatory, but insufficient. It could also reflect a phenomenon often observed in human genetics, namely that opposite alterations of gene dosage have a similar, and not necessarily opposed, phenotypic outcome. On the other hand, since increased BIRC5 expression has been associated with aging in human skin fibroblasts, it could be speculated that premature aging of more vulnerable cell populations, such as the SAN or its precursors, plays a role in CAID syndrome and SAN dysfunction through a mechanism different from loss of BIRC5[146]. Since gastrointestinal motility and sinus rhythm are universally present at birth in CAID, we conclude that the initial formation of the SAN and ICC network must have occurred in sufficient fashion. Also, it has been well documented that even a large reduction of SAN size is clinically tolerated without loss of sinus rhythm [147,148].

Second, SGO1 may guard pacemaker functions by preventing premature senescence and fibrosis in the SAN, the left atrium and the ICC network. The observation that CAID replicates all features of bradycardia-tachycardia syndrome suggests that processes leading to this condition in the elderly may be accelerated in our patients. Premature aging of SAN and ICC subpopulations therefore remains a possibility as a cause for SSS. However, we have not observed other phenotypes overlapping with progeria syndromes in CAID, and it is unclear how and why the SGO1 founder mutation affects only a very specific subset of cells in the human body. A genetic basis for age-related sinus node dysfunction under stress has recently been demonstrated in mice deficient for the popeye domain containing proteins Popeye Domain Containing 1 (POPDC1) and Popeye Domain Containing 2 (POPDC2). In this model, POPDC1 and POPDC2 are required for the integrity of a macromolecular complex at the cell membrane to ensure proper gating of the 2-pore domain potassium channel TREK-1 K(+) Channel Subunit (TREK1)[149]. Senescence induction by SMC1A deficiency has also been identified, providing a plausible explanation for the accelerated aging phenotype found in CdLS[150]. Therefore, senescence may be a common mechanism among cohesinopathies. Outstanding questions include whether SGO1 K23E impairs cohesin loading or chromosomal configuration, and whether this could occur in a cell-type specific context.

Finally, SGO1 may guard pacemaker functions by direct effects on transcriptional regulation of target genes conferring pacemaking properties to the SAN and the ICC network. Numerous studies have addressed in detail the transcriptional profile and electrophysiologic properties of SAN cells, and to a lesser extent those of the ICC network. It is conceivable that SGO1 contributes to maintaining cell-type specific expression signatures of those specialized cells postnatally through yet unidentified mechanisms.

Conclusion

Human cohesinopathies are genetically heterogeneous, monogenic disease caused by strong effects of the disease-causing alleles. In vitro studies of cohesin complex genes have been mostly informed by the role these genes in cell cycling. Here, we compare different human cohesinopathy phenotypes and propose that non-cohesion related roles of the cohesin complex are likely to be of crucial importance in the pathomechanisms of these disorders. In this review, we used these comparisons to further describe mechanisms associated with gene transcription regulation, chromatin 3D architecture, gene dosage effects, epigenetic regulation, and TGF-β signaling. Increasingly, the study of single-gene disorders indicates that associated phenotypes can result from perturbation of multiple pathways, rather than single cascades. Since the cohesin complex regulates the expression of genes involved in several pathways, including genes involved in development, mutations in cohesin complex components are likely to alter chromatin structure, resulting in developmental defects. Gene dosage also seems to be a main component of cohesinopathy severity, especially in CdLS, STAG2 intellectual deficiency, and CAID syndrome. Epigenomic studies of cohesinopathies may be particularly promising, since numerous drugs targeting specific epigenetic modifications are already approved or under development. As a result, understanding cohesinopathies promises to also yield insights into potential treatments for more common cardiac and intestinal conditions.

Funding Statement

This work was supported by the Fonds de Recherche en Santé du Québec (FRSQ), the Canadian Institutes of Health Research (CIHR), the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (ZonMW), the Fonds for Wetenschappelijk Onderzoek (FWO), and the Agence Nationale de Recherche (ANR) for the E-RARE CoHEART network.

Acknowledgments

We are profoundly thankful to all participating families and patients with CAID syndrome. We acknowledge de CoHEART Consortium for the long-standing support of Jeroen Bakkers, Bart Loeys, Michel Pucéat and Gregor Andelfinger.

Author Contributions

J.P. and G.A. wrote the paper. P.V.V and M.P. provided critical revision.

Disclosure statement

No potential conflict of interest was reported by the authors.