Mutations in Cohesin Complex Members SMC3 and SMC1A Cause a Mild Variant of Cornelia de Lange Syndrome with Predominant Mental Retardation

Abstract

The cohesin proteins compose an evolutionarily conserved complex whose fundamental role in chromosome cohesion and coordinated segregation of sister chromatids has been well characterized across species.^1,2 Recently, regulators of cohesin and a structural component of the complex have surprisingly been found to cause phenotypically specific human developmental disorders when mutated. Mutations in NIPBL, the vertebrate homolog of the yeast Sister chromatid cohesion 2 (Scc2) protein, a regulator of cohesin loading and unloading, are responsible for ∼50% of cases of Cornelia de Lange syndrome (CdLS [MIM #122470 and #300590]).^3–5 Mutations in another cohesin regulator, ESCO2, have been found to result in Roberts syndrome and SC phocomelia.^6,7 Two mutations in the cohesin structural component SMC1A (for structural maintenance of chromosomes 1A based on revised ^HUGO nomenclature; also called “SMC1L1”) were recently found to result in an X-linked form of CdLS.⁸ The conserved developmental perturbations seen in these disorders are likely the result of disruption of the cohesin complex’s role in facilitating long-range enhancer promoter interactions and subsequent transcriptional dysregulation.^9,10

CdLS is a dominantly inherited genetic multisystem developmental disorder. The clinical features consist of craniofacial dysmorphia, hirsutism, malformations of the upper extremities, gastroesophageal dysfunction, growth retardation, neurodevelopmental delay, and other structural anomalies (see facies and limbs of patient 1P in fig. 1).^11,12 The mental retardation seen in CdLS, although typically moderate to severe, displays a wide range of variability.¹²

Open in a separate window

Figure 1. 

Facies and hands of classic CdLS in SMC3- and SMC1A-mutation–positive individuals. Proband numbers are indicated. A “P” following the number indicates proband, and an “S” indicates an affected sister. Facial features and upper extremities are shown for 1P, a patient with classic CdLS and a truncating NIPBL mutation; 2P, a male with a sporadic SMC3 E488del mutation; 3P, a female with a sporadic V58-R62del mutation; 4P, a male with a sporadic F133V mutation; 6P, a male with a sporadic R496C mutation; 7P and 7S, two sisters of family 2 with the R496H mutation and mosaicism in the unaffected parent; 8P and 8S, two sisters of family 1 who share a R496H mutation; 9P, a female with a sporadic R496H mutation; 10P, a male with a sporadic R711W mutation; 11P, a female with a sporadic R790Q mutation; and 12P, a female with a sporadic F1122L mutation.

The facial features of an individual with classical CdLS are unique, are easily recognizable, and may be among the most useful diagnostic signs. A milder CdLS phenotype has been reported consistently,^11,13–15 characterized by less significant psychomotor and growth retardation, a lower incidence of major malformations, and milder limb anomalies¹⁵ and accounting for ∼20%–30% of the CdLS population.¹⁵

Mutations in NIPBL account for ∼50% of CdLS cases and have been shown to cause both mild and severe forms.^3,16,17 The recent discovery that mutations in the X-linked SMC1A gene also result in a variant of CdLS⁸ suggests that other cohesin complex members may contribute to the etiology of CdLS and related disorders. Here, we report the screening of 115 NIPBL-mutation–negative individuals with sporadic and familial CdLS and probands with CdLS-variant phenotypes for mutations in the cohesin complex components SMC3 and SMC1A.

Mutation analysis.—All probands were suspected to have CdLS or CdLS variant phenotypes by clinical geneticists experienced in the diagnosis of CdLS, and all were enrolled under an institutional review board–approved protocol of informed consent. All were prescreened for mutations in the 46 coding exons of NIPBL by use of conformation-sensitive gel electrophoresis and/or direct sequencing, as reported elsewhere.³ No NIPBL mutations were previously identified in these 115 probands. SMC3 and SMC1A (^GenBank accession numbers {"type":"entrez-nucleotide","attrs":{"text":"NM_005445","term_id":"1780222575","term_text":"NM_005445"}}NM_005445 and {"type":"entrez-nucleotide","attrs":{"text":"NM_006306","term_id":"1732746274","term_text":"NM_006306"}}NM_006306) were analyzed by PCR amplification and direct sequencing of coding exons 1–29 and 2–25, respectively. Oligonucleotide primer sequences and PCR conditions for SMC3 and SMC1A are available on request.

On the basis of examination, clinical information, and/or photographs, patients for whom detailed information was available were categorized as indicated in table 1, which also shows the percentage of SMC3 and SMC1A mutations in each subgroup. A single mutation was identified in SMC3 in 1 of the 96 probands screened. Eight unique SMC1A mutations were identified in 10 of 115 probands, giving a prevalence of 9% among this cohort of NIPBL-mutation–negative probands and an overall prevalence of ∼5% among unscreened patients with CdLS.

One male (2P) (fig. 1 and tables ​tables22 and ​and3)3) was found to have a unique SMC3 mutation that comprised a 3-nt deletion (c.1464_1466delAGA) and that resulted in the deletion of a single amino acid (p.E488del). Paternity was confirmed, and neither parent carried the mutation, which indicates a de novo event. Furthermore, this change was not observed in >350 control alleles.

Nine probands had missense mutations in SMC1A, and one had an in-frame deletion of 5 aa. One amino acid residue (R496) was mutated in 4 unrelated probands, 2 of which involved familial cases, to account for 7 of the 14 mutation-positive patients. One R496H familial case (involving patients 8P and 8S) (fig. 1 and tables ​tables22 and ​and3)3) resulted from germline mosaicism, and the other (involving patients 7P and 7S) resulted from a somatic and germline mosaic parent with no clinical features of CdLS. The mother in the third family (5P) was found to carry the SMC1A mutation (R496C) and is thought to have mild mental retardation and small hands. A mutation-positive, mentally retarded, affected half-brother developed speech but is institutionalized. All other mutations (V58_R62del, F133V, R196H, R711W, R790Q, and F1122L) were de novo and were identified in patients with unaffected parents who were negative for mutations. In addition, all mutations were absent in >220 normal control alleles. Analysis of protein sequences for human, Fugu, Saccharomyces cerevisiae, and Thermotoga maritime, aligned by the ClustalW method¹⁸ (MacVector [Accelrys]), demonstrated that all mutated residues affect evolutionarily conserved amino acids (fig. 2).

Open in a separate window

Figure 2. 

Conservation of SMC3 and SMC1A and location of mutations. Alignment of human, mouse, Drosophila melanogaster, S. cerevisiae, and T. maritima SMC3- and SMC1-related sequences. Identical residues are indicated in bold with dark shading. Similar residues are indicated in normal font with light shading. Amino acids for which human mutations have been identified are outlined by boxes, with the position indicated above.

SMC3 and SMC1A mutations result in a mild variant of CdLS.—Tables ​Tables22 and ​and33 list the mutations and clinical characteristics of all probands and available affected family members. Notably, the SMC3- and SMC1A-mutation–positive patients demonstrated very mild facial features, no absence or reduction of limbs or digits, and no other major structural anomalies. This is in contrast to classical CdLS (see patient 1P in fig. 1), as summarized in table 4. As noted for the patients studied by Musio et al.,⁸ several of our patients had a more prominent nasal bridge (4P, 8S, 10P, and 12P) (fig. 1) than is typically seen in CdLS. It should be emphasized that, unlike in classic CdLS, in this cohort (1) 80% had birth weights that were normal, and several probands (6P, 9P, and 12P) had growth and head circumferences within the normal range; (2) all individuals walked, and all but one acquired speech; and (3) all had cognitive delays—however, some participated in mainstream classes (3P and 4P), and one was employed in a supervised position in a greenhouse (2P). Of six individuals with the R496H and R496C mutations, considerable variability was noted for growth and development, although all were considered to have mild CdLS. Polymorphisms identified in this work are shown in table 5.

Mapping of mutations to the cohesin crystal structure.—To understand the molecular implications of SMC3 and SMC1A mutations, we capitalized on extensive previous SMC protein biochemistry, electron microscopy, and x-ray crystallography.^1,2 SMC proteins consist of globular N- and C-terminal domains that contain ATP-binding Walker A and Walker B motifs, respectively. These motifs are juxtaposed in the globular head domain of SMCs as a result of an intramolecular antiparallel coiled coil (figs. ​(figs.33 and ​and4).4). A central domain of SMC1 (SMC1A in humans) forms a globular hinge domain that interacts with a similar motif of SMC3 to form a heterodimer. The C-terminal domains of SMC1 and SMC3 also contain ATP binding cassette (ABC) signature (or C) motifs. In addition to their interactions at the hinge domain, the globular head domains of SMC1 and SMC3 interact to complete two tripartite ATP-binding pockets formed by the Walker A and B motifs of one molecule and the signature/C motif of the other (figs. ​(figs.33 and ​and4).4). It has been shown that ATP binding promotes association of the two intermolecular heads and that ATP hydrolysis drives them apart²²; both events facilitate the ability of cohesin to encircle chromatin.

Open in a separate window

Figure 3. 

Mapping of SMC1A and SMC3 mutations to known SMC crystal structure data. A, Mutations in the hinge domain are represented on the Thermotoga SMC dimeric hinge crystal structure. Monomers are colored either red or blue. The darker red and blue regions indicate the N-terminal portions of each monomer. Arrows indicate the N- and C-termini, which are at the end of the coiled-coil domain nearest the hinge region. Mutated residues are shaded bright red. An asterisk (*) indicates the E493 altered residue reported by Musio et al.⁸ The R496 residue in humans is a glycine in Thermotoga (see fig. 2). B, Mutations placed on the yeast SMC1 head domain dimer. These regions are key in the binding and hydrolysis of ATP, indicated by ball and stick molecules. Magnesium molecules at the active sites are indicated by gray balls. Three SMC1A mutations occur in the head domain—two in the N-terminus and one in the C-terminus. The V58_R62del overlaps a loop that is unstructured in the yeast SMC1 crystal structure but is positioned near the Walker A and B motifs (yellow residues A and B) and thus could affect ATP binding or hydrolysis. The F133V mutation is adjacent to the Walker B motif that contains a serine, which contacts ATP directly at the active site. The F1122L mutation is in the signature/C motif (yellow residues C) of the C-terminal head domain. It is positioned at the SMC1/SMC3 head-domain intermolecular interface adjoining the adenine ring of ATP. Structural analysis was performed with Cn3D (NCBI Structure Group).

Open in a separate window

Figure 4. 

Schematic indicating the presumed organization of the SMC1/SMC3 heterodimer and the locations of SMC1A and SMC3 mutations on the cohesin ring. SMC1 is shaded red, and SMC3 is shaded blue, with darker red and blue regions indicating the N-terminal portions of each monomer. Yellow circles indicate ATP molecules bound in proximity to Walker A, B, and C motifs (a, b, and c, respectively). The N-and C-termini are indicated. Mutation locations are indicated by black circles. Altered residues reported by Musio et al.⁸ are indicated by an asterisk (*).

In figure 3, SMC3 and SMC1A mutations are mapped onto crystal data derived from the Thermotoga SMC hinge domain dimer²³ and a yeast Smc1 head domain dimer²⁴ (NCBI Protein Database). The SMC3 E488del and the SMC1A R496H, R496C, and E493A⁸ mutations all map to the junction of the N-terminal coiled-coil and the hinge (fig. 3A). This region is remarkably conserved from bacteria to humans (fig. 2).

The process of loading the cohesin onto DNA is dependent on factors that include NIPBL (also known as “Scc2”)^25,26 and the hydrolysis of ATP.^26,27 It has been shown that mutations of the Bacillus SMC hinge domain disrupt both DNA binding²² and ATP hydrolysis.² Thus, it is possible that SMC3 and SMC1A mutations at the boundary of the hinge domain disrupt DNA binding or ATP hydrolysis kinetics, leading to similar phenotypes as well as to those seen for some mutations in NIPBL. Three SMC1A mutations (V58_62del, F133V, and F1122L) occur in the head domain (fig. 3B) and are positioned near the Walker A, B, and signature/C motifs that could affect ATP binding, ATP hydrolysis, and/or SMC1/SMC3 head-domain dimerization, an ATP-dependent process.

The SMC1A mutations R196H, R711W, and R790Q and the familial D831_Q832delinsE mutation⁸ all reside in the coiled-coil domain (fig. 4). By use of the Coils program,²⁸ which predicts the probability of a protein to form a coiled-coil, these mutations had a small likelihood of disrupting the coiled-coil arms. However, the alterations caused by these mutations may affect the angulation of the coiled-coil,²⁹ resulting in impaired intra- or intermolecular approximation of the SMC head domains, or disrupt binding of accessory proteins to the cohesin ring.

SMC1A mutations in females.—In contrast to previous work on SMC1A,⁸ 10 of 14 total SMC1A-mutation–positive individuals were female. Furthermore, we describe similarly affected male and female probands, implying an X-linked dominant mode of expression. Interestingly, several males were rather mildly affected (4P and 6P) and no more severely affected than many of the SMC1A-mutation–positive females. Since SMC1A escapes X inactivation,³⁰ it is likely that the mechanism in affected females is due to a dominant negative effect of the altered protein and less likely that is is due to decreased protein levels⁸ or skewed X inactivation. Consistent with this dominant negative effect on cohesin, the SMC3 mutation is also a single amino acid deletion.

The cohesin complex and mental retardation.—Of note, all the patients described here have disease in the mild-to-moderate range of CdLS and were ascertained as having mild facial features reminiscent of CdLS, but none has major structural anomalies typically seen in classic CdLS (table 4). However, without exception, all had varying degrees of mental retardation (fig. 1 and tables ​tables33 and ​and4).4). Although the facial features in many of these individuals may be appreciated by dysmorphologists experienced in this diagnosis, for the most part, they would present to clinical attention as individuals with mild-to-moderate mental retardation, often without the hallmark short stature and/or microcephaly of CdLS (table 4). This strongly suggests that brain development is the process most sensitive to perturbation of these proteins. The relatively small contribution of SMC3 (∼1%) and SMC1A (∼5%) mutations in our study group may be the result of the selection bias of our study subjects, who were ascertained as having CdLS or CdLS-variant phenotypes. There may be a subset of individuals among the larger diagnostic category of mental retardation spectrum disorders with an apparently nonsyndromic etiology who may be carrying the bulk of mutations in cohesin pathway genes. Furthermore, additional “cohesinopathies” may result from perturbation of the >15 additional components of this complex that have yet to be associated with human disorders.

Acknowledgments

Web Resources

References