Heterozygous loss-of-function SMC3 variants are associated with variable and incompletely penetrant growth and developmental features

Abstract

Heterozygous missense variants and in-frame indels in SMC3 are a cause of Cornelia de Lange syndrome (CdLS), marked by intellectual disability, growth deficiency, and dysmorphism, via an apparent dominant-negative mechanism. However, the spectrum of manifestations associated with SMC3 loss-of-function variants has not been reported, leading to hypotheses of alternative phenotypes or even developmental lethality. We used matchmaking servers, patient registries, and other resources to identify individuals with heterozygous, predicted loss-of-function (pLoF) variants in SMC3, and analyzed population databases to characterize mutational intolerance in this gene. Here, we show that SMC3 behaves as an archetypal haploinsufficient gene: it is highly constrained against pLoF variants, strongly depleted for missense variants, and pLoF variants are associated with a range of developmental phenotypes. Among 13 individuals with SMC3 pLoF variants, phenotypes were variable but coalesced on low growth parameters, developmental delay/intellectual disability, and dysmorphism reminiscent of atypical CdLS. Comparisons to individuals with SMC3 missense/in-frame indel variants demonstrated a milder presentation in pLoF carriers. Furthermore, several individuals harboring pLoF variants in SMC3 were nonpenetrant for growth, developmental, and/or dysmorphic features, some instead having intriguing symptomatologies with rational biological links to SMC3 including bone marrow failure, acute myeloid leukemia, and Coats retinal vasculopathy. Analyses of transcriptomic and epigenetic data suggest that SMC3 pLoF variants reduce SMC3 expression but do not result in a blood DNA methylation signature clustering with that of CdLS, and that the global transcriptional signature of SMC3 loss is model-dependent. Our finding of substantial population-scale LoF intolerance in concert with variable penetrance in subjects with SMC3 pLoF variants expands the scope of cohesinopathies, informs on their allelic architecture, and suggests the existence of additional clearly LoF-constrained genes whose disease links will be confirmed only by multi-layered genomic data paired with careful phenotyping.

Keywords: CdLS3; Cornelia de Lange syndrome; LoF; SMC3; cohesin; loss-of-function.

Figure 1.. SMC3 predicted loss-of-function (pLoF) variants.

a. pLoF simple nucleotide variants (SNVs) mapped to the SMC3 protein sequence. Subject cases are in black, top. The c.430–1G>T splice variant immediately precedes exon 8, which if skipped would result in a shift of reading frame, however splicing may be rescued (see text). The p.(Gly1217MetfsTer39) variant is in the final exon and is predicted to escape nonsense mediated decay. gnomAD pLoF variants passing quality filters are in grey, bottom. UKBB pLoF variants are in brown, bottom. Domains via UniProt. Arrowhead denotes the C-terminal end (equivalent to codon 211) of the minor transcript described in Fig. 3. Diamond denotes the point after which nonsense mediated decay may be escaped (codon 1176). The 3^′ region homologous to the SMC3P1 pseudogene (codons 974–1217) is in orange. b. Somatic SMC3 pLoF SNVs among 13,714 tumors from the NCI GDC collection (https://portal.gdc.cancer.gov/genes/ENSG00000108055). c. Regional missense constraint over SMC3, via an analysis of gnomAD. The transition point is in codon 390 (exon 13 of 29). d. Case deletions (based on genome build GRCh37/hg19). Of the OMIM phenotype-associated genes in this region (green), none are dominant haploinsufficient disease genes for disorders that include developmental delay or dysmorphic features.

Figure 2.. Growth and developmental milestones associated with SMC3 pLoF compared with pathogenic missense/in-frame indels.

a. Growth measurements adjusted for age and sex (Z-score) are shown for SMC3 predicted loss-of-function (LoF) and missense/in-frame indel cases. ‘Postnatal’ measurements are at the time of study enrollment/sampling. b. Age at which developmental milestones were reached for the two categories of SMC3 variants (LoF vs pathogenic missense/in-frame indel). 25^th/50^th/75^th centiles of population data (Denver II, via DECIPHER) are shown at the right of each figure in (b). Missense/in-frame indel data were obtained from . Plots are standard boxplots. Statistical comparisons are the Mann-Whitney U test (Wilcoxon).

Figure. 3.. Gene expression in wild-type and SMC3^+/− tissues.

a. GTEx data demonstrate that only a single, full-length SMC3 isoform (ENST00000361804.4) is expressed at an appreciable level in any adult human tissue represented (https://gtexportal.org/home/gene/SMC3). b. pLoF SMC3 variants (LoF; nonsense, frameshift indel, splice site) in hematopoietic and lymphoid tumors correlate with decreased SMC3 gene expression, to a mean of 71.9% of wild-type (wt) (p=0.019, Mann-Whitney U test). Data derived from the Cancer Cell Line Encyclopedia (CCLE) (http://xenabrowser.net). c-d. Comparison of genes with altered expression in early postnatal Smc3^+/− mouse cortex and those altered in a late prenatal Nipbl^+/− mouse whole brain. c. 50 of 406 (12.3%) differentially expressed genes (DEGs) any age (P1, P3, P7, P14, P21) in Smc3^+/− mice overlap with 3,507 DEGs in E17.5 Nipbl^+/− mice (top). 9 of 58 (15.5%) DEGs in 2+ ages in Smc3^+/− mice overlap with those same 3,507 genes (bottom). d. Neither of these is significantly different from chance (3,507/30,686 total genes; 11.4%) (any age p=0.632, chi-squared test of proportions; 2+ ages p=0.441). See Methods for derivation of denominator of 30,686 genes and Discussion for a commentary on methodological differences between the two models. e. Minimal displacement embedding of pseudobulk-averaged, z-normalized expression profiles from a genome-wide Perturb-seq experiment in K562 CML cells (plot generated via https://gwps.wi.mit.edu/). Each dot is one of 1,973 genes with strong transcriptomic signatures. The SMC3 knockdown transcriptional disturbances group tightly with those of other cohesin genes (pink) and even an epigenetic regulator occasionally mutated in CdLS-like patients (EP300).

Figure 4.. Global DNA methylation pattern of SMC3 pLoF compared with that of CdLS.

Multidimensional scaling (MDS) plot of global methylation signature (episignature) of blood-derived DNA. A SMC3 pLoF subject (p.Arg360Ter, purple) does not plot with individuals clinically diagnosed with CdLS (red), instead plotting among the control population (green).

Figure 5.. Hypothesized phenotypic spectra of SMC3 variant types.

Missense (ms) /in-frame (i.f.) indel variants, via an apparent dominant negative mechanism as supported by prior literature, carry the most severe phenotype, with many patients appearing to have atypical Cornelia de Lange syndrome (CdLS). Loss-of-function (LoF) variants carry a milder phenotype, with only some individuals being recognized to have CdLS and some individuals lacking key phenotypic features (growth retardation, developmental delay, characteristic facial dysmorphism). We hypothesize that nullimorphic ms and hypomorphic ms variants also exist.