A Novel Variant in CDKN1C Is Associated With Intrauterine Growth Restriction, Short Stature, and Early-Adulthood-Onset Diabetes

Abstract

Context:

CDKN1C, a cyclin-dependent kinase inhibitor and negative regulator of cellular proliferation, is paternally imprinted and has been shown to regulate β-cell proliferation. CDKN1C mutations are associated with growth disorders, including Beckwith-Wiedemann syndrome and IMAGe syndrome.

Objective:

To investigate the genetic basis for a familial disorder characterized by intrauterine growth restriction, short stature, and early-adulthood-onset diabetes.

Design, Setting, and Participants:

Genomic DNA samples (15 affected and 26 unaffected from a six-generation pedigree) were analyzed by genome-wide single nucleotide polymorphism arrays, whole exome and Sanger sequencing, and multiplex ligation-dependent probe amplification.

Main Outcome Measure(s):

Subjects were assessed for height, weight, adrenal gland size, ACTH, diabetes status, and testis volume. Linkage and sequence analyses were performed, and the identified genetic variant was functionally evaluated in reconstitution studies.

Results:

The pedigree followed a paternally imprinted pattern of inheritance, and genetic linkage analysis identified a single significant 2.6-megabase locus on chromosome 11p15, within the imprinting center region 2. Multiplex ligation-dependent probe amplification did not detect copy number variants or methylation abnormalities. Whole exome sequencing revealed a single novel variant in the proliferating cell nuclear antigen-binding region of CDKN1C (c.842G>T, p.R281I) that co-segregated with affected status and, unlike variants found in IMAGe, did not entirely abrogate proliferating cell nuclear antigen binding. Clinical assessments revealed that affected individuals had low testicular volume but normal adrenal function.

Conclusions:

We report a novel CDKN1C mutation associated with features of IMAGe syndrome, but without adrenal insufficiency or metaphyseal dysplasia, and characterized by early-adulthood-onset diabetes. Our data expand the range of phenotypes observed with CDKN1C defects and suggest that CDKN1C mutations may represent a novel monogenic form of diabetes.

We previously reported an extended pedigree from Ecuador whose family members exhibited intrauterine growth retardation (IUGR), failure of an adolescent growth spurt with proportional adult short stature, minimal subluxation of the fifth metacarpal-phalangeal joint, and adult-onset diabetes unrelated to obesity or other manifestations of metabolic syndrome (1). Targeted genomic sequencing failed to identify a shared mutation among five affected family members in candidate genes of the GH/IGF-1 pathway. Here, we extend this pedigree to six generations, and in doing so reveal strong evidence of a paternally imprinted inheritance pattern. Using a combination of linkage analysis and whole exome sequencing, we have uncovered the genetic basis for this syndrome and provide evidence that it is related molecularly to other growth disorders, including Beckwith-Wiedemann syndrome (BWS; OMIM 130650), a single familial case of Russell Silver syndrome (RSS; OMIM 180860) (2), and IMAGe syndrome (IMAGE; OMIM 614732) (3–5), but with several key phenotypic differences.

Subjects and Methods

This study was approved by the institutional review board of the Institute of Endocrinology, Metabolism, and Reproduction in Quito, Ecuador. Informed consent was obtained for all subjects. Subjects under the age of 18 provided assent, and parents provided consent. Details of all methods are supplied in the Supplemental Data.

Height, weight, and ACTH were measured as previously reported (1).

Nine affected individuals and four unaffected family members were genotyped using Affymetrix Genome-Wide Human SNP Arrays version 6.0 (Affymetrix, Inc). Linkage analysis was performed using Superlink-Online SNP 1.1 (6). Detection of copy number alterations and methylation profiling of the 11p15 region were performed using the SALSA MS-MLPA BWS/RSS ME030-C3 MLPA assay from MRC-Holland. Whole exome sequencing was performed at the Broad Institute using genomic DNA from five affected family members. Exon 2 (NM_000076.2) of CDKN1C was sequenced using Sanger sequencing for 15 affected individuals, 23 unaffected family members, and 66 unrelated individuals of Ecuadorian ancestry living in New York City.

N-terminally FLAG-tagged human CDKN1C cDNA, wild-type (F-CDKN1C) or IMAGe variants p.Phe276Val (F-276V) and p.Lys278Glu (F-K278E), in the pcDNA3.1 expression vector were generous gifts from Dr Valerie Arboleda and Dr Eric Vilain (3). Missense CDKN1C p.Arg281Ile was regenerated employing the QuikChange II site-directed mutagenesis kit (Agilent Technologies). HEK293 cells were transiently transfected with 2 μg per well of either empty vector, F-CDKN1C wild-type or F-CDKN1C variants. After 24 hours of transfection, the cells were serum-starved in DMEM supplemented with 0.1% BSA for 16 hours before treatment with cycloheximide (CHX), as indicated. Immunoblot experiments for FLAG, proliferating cell nuclear antigen (PCNA), and p38 were performed. In some cases, before immunoblot analysis, FLAG-tagged proteins from cell lysates were first immunoprecipitated with anti-FLAG-M2-agarose beads.

Results

We have expanded the pedigree from our previous report (1) to six generations by identifying and phenotyping a total of 41 family members, of whom 15 show characteristics of the previously described phenotype (Figure 1). The extended pedigree shows that parent-to-child transmission only occurs maternally, suggesting that this disorder is paternally imprinted, such that expression of the mutated gene is silenced when inherited from the father (Figure 1). Single nucleotide polymorphism (SNP)-based linkage analysis identified a single region on chr11p15 that had a significant LOD score of 3.4 (Supplemental Figures 1 and 2). Manual haplotype analysis confirmed an approximately 2.6-megabase shared region with break-points between base pairs 2,578,539 and 2,773,471 (start) and 5,139,027 and 5,144,407 (end) (coordinates based on Reference Genome Build 37). This region contains 45 genes and is controlled by the imprinting center region 2 (ICR2), which is consistent with our assumption that this disorder is inherited in a dominant manner with paternal imprinting. Multiplex ligation-dependent probe amplification (MLPA) assays revealed no methylation abnormalities or copy number abnormalities in any of the five affected individuals tested.

Figure 1.

Pedigree showing evidence of paternal imprinting. GIV-11 has short stature and later onset diabetes but does not carry the variant. She has additional dysmorphic features, including a cleft palate, and has a higher height SD score than all of the variant carriers except for one individual. She was not considered to be affected with this syndrome by other members of the family or by her physician.

Whole exome sequencing of five affected individuals identified a single novel nonsynonymous variant within the chr11p15 linkage interval: CDKN1C c.842G>T, p.R281I. Sanger sequencing confirmed that this variant co-segregates with affected or carrier status (Figure 1) and was not present in 66 unrelated Ecuadorian controls currently living in New York, indicating that it is not a common Ecuadorian allele.

Mutations in CDKN1C have been found to be causal for the undergrowth disorder IMAGe syndrome and in a single familial case of RSS, so we assessed affected family members for other features of these syndromes. Features of RSS including body asymmetry, frontal bossing, café au lait spots, downturned corners of the mouth, and triangular face shape were rare and were seen in both affected and unaffected individuals (Supplemental Table 1). The only features of RSS specific to affected individuals in this pedigree are IUGR and severe short stature. All seven affected males had low testicular volume, <20 cc as an adult (Supplemental Tables 1 and 2), a feature seen in IMAGe syndrome. Although fertility tests were not performed, eight of the 11 affected subjects who are married have produced children, suggesting that there is no evidence of reproductive difficulty in this population. In contrast to IMAGe syndrome, affected individuals did not demonstrate any evidence of primary adrenal insufficiency, as evidenced by normal ACTH levels (Supplemental Tables 1 and 2), although subtle adrenal defects cannot be ruled out completely because data from stimulation testing are not available. Affected individuals had smaller adrenal glands on ultrasound, but the reduced adrenal gland size may be secondary to smaller body size (Supplemental Tables 1 and 2, comparison to kidney volume). Affected individuals did not have any radiological sign of metaphyseal dysplasia.

Of the 15 affected family members, eight have previously been clinically diagnosed with diabetes, all prior to age 40, based on classic symptoms or an elevated fasting glucose or hemoglobin A1c. Details of current treatments and age of diagnoses are provided in Supplemental Table 2. Subject GVI-1 has not been evaluated for diabetes, but he developed slightly elevated fasting glucose and acanthosis nigricans when placed on GH therapy. The remaining six affected family members have not yet been diagnosed with diabetes, although some of them have instituted intensive lifestyle changes to prevent the onset of diabetes. Two of these six individuals are still under age 20. Of the 26 unaffected family members, only one has been diagnosed with diabetes (P = .0005 for comparison with affected individuals). The clinical presentation of IUGR, short stature, and diabetes thus appears to be due to a single inherited syndrome.

The CDKN1C gene encodes a cyclin-dependent kinase inhibitor that negatively regulates cellular proliferation (7–9). The CDKN1C p.R281I mutation identified in our patients is located in the PCNA binding domain, within a 10-amino acid residue sequence clustered with missense mutants associated with IMAGe syndrome and one familial RSS report (2, 3) (Figure 2A). In HEK293 reconstitution studies, stability of protein expression (CHX treatment) was comparable between mutant F-R281I and F-CDKN1C by immunoblot analysis (Figure 2B, upper panel). F-R281I coimmunoprecipitated with endogenous PCNA (Figure 2B, lower panel), but in contrast to F-CDKN1C, coimmunoprecipitation with PCNA was markedly reduced 48 hours after CHX treatment (Figure 2B, lower panel). This reduction in PCNA interaction was specific because F-R281I remained capable of coimmunoprecipitating with endogenous p38/SAPK, a stress-activated protein kinase believed to interact with the N terminus of CDKN1C (10) (Figure 2B, lower panel). IMAGe mutants F-F276V and F-K278E (p.Phe276Val and p.Lys278Glu, respectively; Figure 2A) did not coimmunoprecipitate with PCNA (Figure 2C, bottom panel), whereas F-R281I exhibited significantly reduced interactions with PCNA only at 48 hours after CHX treatment (Figure 2C, bottom panel). Interaction with p38/SAPK, in contrast, was retained by all CDKN1C variants. These results support impaired functions for CDKN1C p.R281I that is distinctly less severe than those observed for IMAGe-associated CDKN1C mutants, correlating with the less severe phenotype exhibited by our subjects.

Figure 2.

CDKN1C p.R281I demonstrates reduced PCNA binding functions. A, Schematic of the human CDKN1C protein and the three encoding exons (NM_000076.2, transcript variant 1). Amino acid residue numbering, the carboxy terminus (COOH), and known protein domains are indicated. CDKi, cyclin dependent kinase inhibitor; PAPA, Pro-Ala-Pro-Ala repeat; QT, carboxyterminal domain. Residues of the PCNA domain (20), Ala263 to Glu285, are shown; residues in red, highly conserved among species. Black asterisks represent residues that are mutated in IMAGe syndrome or RSS; green, the variant identified in this report. Missense mutations associated with IMAGe are listed above residue 263 – 285 sequence (3, 11, 12), and for RSS, listed below (2). B, F-R281I expression and ability to interact with PCNA, compared to wild-type F-CDKN1C, in HEK293 reconstitution studies. Upper panel, Cell lysates, immunoblot (IB) analysis, 40 μg total protein per lane. Lower panel, 280 μg total protein per condition was first immunoprecipitated (IP) with anti-FLAG antibody before IB analysis. C, F-R281I interaction with PCNA compared to IMAGe associated variants. Upper panel, Cell lysates, IB analysis, 25 μg protein per lane. Two replicates of the mutated protein were assessed (F-R281I-1 and F-R281I-2). Lower panel, 150 μg protein per condition were IP with anti-FLAG antibody before IB analysis.

Discussion

Our study has uncovered the genetic basis for a novel syndrome that expands the spectrum of conditions characterized by disordered growth and abnormal CDKN1C activity. Gain-of-function missense variants in the PCNA binding domain of CDKN1C were initially described as the cause of IMAGe syndrome, which is characterized by IUGR, short stature, metaphyseal dysplasia, genitourinary anomalies in males, and adrenal hypoplasia congenita (3, 11). A recent study of a single family identified additional phenotypes of IMAGe, including relative macrocephaly, milder adrenal insufficiency, history of oligohydramnios, and arachnodactyly (12). Our subjects' phenotype overlaps considerably with IMAGe syndrome, but does not appear to include adrenal insufficiency or metaphyseal dysplasia. Recently, Brioude et al (2) described a familial case of RSS due to a novel missense variant in the same region of CDKN1C as the IMAGe variants and the c.842G>T variant reported here. Affected family members were characterized by IUGR, short stature, and normal adrenal function. All affected individuals were female, so effect on testicular volume could not be assessed. These subjects met criteria for RSS due to relative macrocephaly at age 2, low body mass index at age 2, or feeding difficulties during infancy. Information regarding these clinical parameters was unavailable for subjects in our pedigree. Regardless of terminology, there is clear phenotypic overlap among subjects with classic IMAGe syndrome, the family reported by Brioude et al (2), and our family—all associated with missense variants in the PCNA binding region of CDKN1C.

To our knowledge, diabetes has not been reported as a feature of subjects with defects in CDKN1C. There is strong evidence in the literature supporting a role for CDKN1C in β-cell proliferation, and this could be a mechanistic explanation for the diabetes seen in these subjects. Multiple lines of evidence support the hypothesis that loss of CDKN1C function leads to increased β-cell proliferation. For example, one study found that pancreatic CDKN1C expression was highest in β-cells, but expression was lost within focal areas of β-cell hyperplasia in pancreases from subjects with hyperinsulinism of infancy who have somatic deletions of maternal 11p15 leading to silencing of CDKN1C (13). Another study showed that short hairpin RNA-induced silencing of CDKN1C in human islet cells derived from deceased adult donors and transplanted into mice resulted in proliferation of the transplanted β-cells (14). Additionally, approximately 50% of individuals with BWS develop hypoglycemia at birth, frequently due to hyperinsulinism (15, 16). A pathological study of pancreases removed from four subjects with BWS and hyperinsulinism revealed marked hyperplasia of the endocrine cells throughout the pancreas (17). The loss-of-function mutations in CDKN1C associated with BWS are likely major precipitants of the increased β-cell proliferation seen in those subjects. We hypothesize that overactivity of CDKN1C p.R281I results in the opposite phenotype: decreased proliferation of β-cells, leading to reduced insulin production and onset of diabetes.

Although diabetes has not been reported in the subjects with classic IMAGe syndrome, none of the IMAGe subjects reported thus far have reached the age of diabetes onset seen in the Ecuadorian family. It will be critical to reassess individuals with IMAGe as they get older for evidence of diabetes. In the single family with RSS due to a CDKN1C variant, diabetes was not reported, but detailed studies of glucose metabolism have not been performed to date.

Although it is possible that the diabetes is a consequence of a second variant present in the same linkage interval as CDKN1C p.R281I, no other novel shared variants within this region were identified. Interestingly, this region contains the KCNQ1 gene, in which diabetes-associated SNPs have been identified via genome-wide association studies (18, 19). No shared rare nonsynonymous variants were found by exome sequencing within this gene. It is possible that the genome-wide association study SNP is mediating its effect on diabetes risk via CDKN1C.

Finally, it is of note that the variant present in our family is just two amino acids distal to the cluster of variants described in the families with IMAGe syndrome (Figure 2A). Our reconstitution studies of CDKN1C p.R281I support a functional aberrancy in its PCNA binding domain less severe than the nearby IMAGe associated mutations, p.F274V and p.K278E. Because PCNA binding activity by CDKN1C is required for arrest of the cell cycle (20), a reduction or loss of this activity appears contradictory to the observed clinical phenotypes described above and in previous reports (2, 3, 5). Hamajima et al (11) suggested that these missense variants increase stability of the CDKN1C mutants, a phenomenon we did not observe. We did demonstrate, however, that the PCNA binding irregularities of p.R281I and IMAGe variants did not interfere with the ability of these CDKN1C mutants to associate with other proteins, such as p38/SAPK. Further investigations are necessary to fully elucidate how CDKN1C variants defective only in the PCNA binding domain confer such a wide spectrum of clinical presentations.

In conclusion, we describe a family with a syndrome of IUGR, short stature, and early-adulthood-onset diabetes due to a mutation in CDKN1C. This family expands the phenotypic spectrum of diseases caused by mutations in CDKN1C and may represent a novel form of monogenic diabetes.