Background
Abdominal obesity-metabolic syndrome 3 (AOMS3 [
OMIM:615812]) is a rare autosomal dominant disorder caused by pathogenic variants in the dual-specificity tyrosine phosphorylation-regulated kinase 1B gene (
DYRK1B) located on chromosome 19q13.2 [
1]. This monogenic form of metabolic syndrome (MetS) is characterized by abdominal obesity, type 2 diabetes, hypertension, and early-onset coronary artery disease [
2]. DYRK1B inhibits the Sonic Hedgehog and WNT pathways, increasing the expression of master adipogenic transcription factors CCAAT/enhancer binding protein (C/EBP-alpha) and peroxisome proliferator-activated receptor gamma (PPAR-gamma) [
2]. Moreover, DYRK1B induces the expression of glucose-6-phosphatase, a key enzyme in hepatic gluconeogenesis [
3].
The identification of carriers of pathogenic variants in genes such as
DYRK1B could be useful for establishing early diagnostic strategies and medical interventions in a reasonable number of affected individuals even before symptoms or complications appear. Until now, only two different mutations (p.Arg102Cys and p.His90Pro) in
DYRK1B have been described as being causative of AOMS3 in three Iranian families and five unrelated Caucasian individuals [
2]. However, families with rare monogenic forms of metabolic diseases could be misdiagnosed or even overlooked if causative variants are not directly explored, especially in populations with a high prevalence of these entities.
Therefore, we searched for
DYRK1B variants in the exome sequencing data derived from 968 unrelated individuals (509 with type 2 diabetes) belonging to the DMS1 SIGMA-cohort (ExAC) [
4], focusing on variants classified by Variant Effect Predictor (VEP) as deleterious and damaging to confirm their co-segregation with AOMS3. Here, we describe two novel
DYRK1B mutations as causative of AOMS3 in two families previously misdiagnosed with type 2 diabetes.
Discussion
Over the past few years, there have been many attempts to gain more insight into the genetic factors involved in metabolic diseases [
7‐
9]. Multiple genetic variants have been shown to participate in the pathogenesis of each of the traits of metabolic syndrome. The polygenic nature of these conditions implies that the effect of the majority of genetic variants in these disorders is small [
7]. However, families in which autosomal dominant inheritance is present have been used to search for rare mutations in genes with a strong contribution [
10,
11]. The availability of exome sequencing is leading to the rapid identification of new players in the pathogenesis of metabolic diseases. This is the case for
DYRK1B mutations, which cause a rare monogenic form of MetS known as AOMS3 [
2]. This syndrome has been described as the presence of abdominal obesity, type 2 diabetes, hypertension, and early-onset coronary artery disease [
2]. Pathological
DYRK1B variants result in the enhanced expression of transcription factors C/EBPalpha and PPARgamma, leading to increased adipogenesis. In addition,
DYRK1B increases glucose-6-phosphatase, which is strongly associated with insulin resistance, explaining the metabolic phenotypes characterizing AOMS3 [
2,
3].
The participation of
DYRK1B in MetS is poorly studied. Six years after
DYRK1B was associated with AOMS3, only a few carriers have been reported [
1]. Furthermore, only two missense
DYRK1B mutations (p.Arg102Cys and p.His90Pro) have been identified in these individuals [
2]. It is possible for rare diseases that mimic symptoms of common diseases to be confused with them. The prevalence of metabolic diseases in Mexico is one of the highest in the world [
12,
13], and rare metabolic diseases are often hidden behind them; therefore, families with rare monogenic forms can remain unnoticed.
Next generation sequencing technologies have greatly improved the possibility of identifying rare pathogenic variants involved in monogenic diseases [
4,
11]. In this study, after analyzing the sequence of all
DYRK1B exons in a sample of 968 adult, including 509 with type 2 diabetes (SIGMA-ExAC) [
14], we found 29 variants. SIFT and PolyPhen predicted that four of them (p.Leu28Pro, p.Asp436Asn, p.Arg252His, and p.Lys68Gln) have a deleterious and damaging effect. The p.Leu28Pro variant was described previously as having a protective effect against type 2 diabetes in a phenome-wide association study [
15]. In this study, we found five heterozygotes individuals with neither type 2 diabetes nor AOMS3, but we were not able to recruit the familial relatives to confirm its protective effect. The p.Asp436Asn variant was found in a male heterozygote, who was 98 years old and metabolically healthy. However, p.Arg252His and p.Lys68Gln exhibited co-segregation with the AOMS3 phenotype with classic dominant autosomal inheritance and full penetrance. Both variants were absent in the 1000 Genomes database [
16], and the gnomAD database reports eight individuals with p.Arg252His variant and five with p.Lys68Gln variant, including the ones we report here. In Latinos, these variants were found with a frequency of 0.00008677 and 0.00005783, respectively [
17].
MD structural analyses predicted that, when DYRK1B-252His was present, the formation of three hydrogen bonds was impaired, with instability in the N- and C-terminal regions. In contrast, the p.Lys68Gln variant did not produce any significant changes in the protein structure, although the NLS motif could be affected, in accordance with Kosugi et al., who showed that the N-terminal basic pattern “Lys
68Arg
69” is required for a strong NLS activity [
18]. More over, the effect of these variants could be similar to those documenting that DYRK1B-102Cys and DYRK1B-90Pro variants cause changes to the structure and perinuclear aggregation but barely affect the kinase activity [
6]. We classified p.Arg252His and p.Lys68Gln as causative of AOMS3, in agreement with the American College of Medical Genetics and Genomics standards and guidelines [
19].
Compared to non-carriers of pathogenic variants, we found that carriers had higher BMI, and FG and triglyceride levels. Furthermore, pulse wave velocity was increased in all carriers, even in the absence of arterial hypertension, explaining the high cardiovascular risk found in this condition. Insulin action was decreased in all but one case, but the insulinogenic index was significantly decreased in all carriers, even in normoglycemic individuals, suggesting that the remarkable severity of hyperglycemia found in this condition results from a combination of moderate insulin resistance and a moderate to severe defect in insulin secretion. This could be supported by recent findings showing that DYRK1A and DYRK1B play an important role in human β cells proliferation [
20].
Notably, we are describing additional features of the disease and manifestations that develop throughout life. Our findings exhibit age-dependent variance in expressivity in all patients, with some clinical features apparent at a very early age and other manifestations appearing later in life. Central obesity and insulin resistance started during childhood and then progressed rapidly to morbid obesity and labile type 2 diabetes. They also had severe hypertriglyceridemia with onset as a teenager. Similarly, AOMS3 patients developed hypertension in the fifth decade of life, and both families had a history of premature death due to cardiovascular events. Another interesting finding was that p.Lys68Gln carriers had a higher BMI, hypoinsulinemia, worse diabetes control and an increased urine albumin/creatinine ratio compared to the p.Arg252His carriers, suggesting an allelic heterogeneity.
Remarkably, none of the non-carriers had diabetes; only a 49-year-old woman (II.2 in family 2) had glucose serum levels of 109 mg/dl, along with obesity and hypertension. Other non-carrier individuals also had low levels of HDL or obesity, however, none of them met all the AOMS3 clinical features. These manifestations could be a reflection of the high prevalence of metabolic diseases in our population [
12,
13].
Material and methods
Study participants
We included 968 unrelated adult Mexican Mestizos belonging to the DMS1 SIGMA-cohort [
14] who were previously sequenced by Sure-Select Human All Exon v2.0 (Illumina) and included in the ExAC project [
4]. A peripheral blood sample was collected after fasting for at least 8 h. The following clinical and biochemical data were obtained for all participants of the DMS1 cohort using the Synchron CX5 Analyzer System (Beckman Coulter Fullerton, CA, USA): FG (mg/dL), HDL (mg/dL), and serum triglycerides (mg/dL). HbA1c levels were measured using the IN2it analyzer (Bio-Rad, Hercules, CA, USA). Blood pressure was measured using a digital blood pressure monitor (HEM-907XL, OMRON). Weight and height were measured using a body composition monitor (HBF-500 INT, OMRON) and electronic stadiometer (ADE Germany). Waist circumference was measured midway between the inferior margin of the ribs and the border of the iliac crest using a flexible clinical measuring tape.
The study was carried out according to the Declaration of Helsinki and was approved by the Research, Ethics, and Biosafety Human Committees of the Instituto Nacional de Medicina Genómica (INMEGEN) in Mexico City. All participants provided written informed consent. They were recruited from August 2017 to December 2018, all of them inhabited the Valley of Mexico.
Identification of pathogenic variants
We identified
DYRK1B variants by analyzing the exon sequences in each individual. To find the deleterious and damaging variants, we annotated them using the VEP toolset [
21]. Missense variants that were predicted as deleterious or affecting the protein structure were used to perform genotype–phenotype linkages. Individuals identified as carriers of these variants and had clinical manifestations suggesting AOMS3 were re-contacted and invited, along with their family members, to participate in a familial co-segregation study. All individuals who participated in the family segregation study provided written informed consent; in the case of children, the parents provided written consent and the children assented.
Genotyping was performed by Sanger sequencing using specific primers: p.Arg252His forward primer, CCTTTCTTCTCTGGCCAT; p.Arg252His reverse primer, ACCCAAACTACTAGCCGTGC; p.Lys69Gln forward primer, TGCCAGCAGCCTTACAGTT; p.Lys69Gln reverse primer, CCACTGCGCAACGATGTAGTC. The obtained sequences were analyzed by 4Peaks V1.8 [
22].
The same clinical, demographic, and biochemical data described for index cases were obtained for each family participant. In addition, seven carriers (four p.Arg252His and three p.Lys69Gln) and three non-carriers were clinically and biochemically re-evaluated at the Unidad de Investigación de Enfermedades Metabólicas, Instituto Nacional de Ciencias Médicas y Nutrición. The assessment included fasting biochemical measurements (i.e., clinical chemistry, liver panel, and lipid profile), estimation of body composition, pulse wave velocity measurements, and albumin/creatinine ratio in a spot urine sample. An oral glucose tolerance test was performed using a 75 g oral glucose charge. Serum glucose and insulin were measured at fasting, 30-, 60-, 90-, 120-, and 180-min after oral glucose intake. The glucose concentration was measured by an automated glucose analyzer (Yellow Springs Instruments Co.). The serum insulin concentration was measured by a chemiluminescent immunoassay (Beckman Coulter Access 2) and HbA1c levels by HPLC (Variant II Turbo, BIORAD). Cholesterol, triglycerides, HDL, apolipoprotein B, uric acid, creatinine, and hepatic enzymes levels were measured using colorimetric assays (Unicel DxC 600 Synchron Clinical System Beckman Coulter). LDL was calculated by the Friedewald equation when the triglyceride concentration was < 250 mg/dL.
Molecular dynamics
The wild-type amino acid sequence of DYRK1B (Q9Y463) from the UniProt database [
23] was modeled to obtain the 3D protein structure using the I-TASSER server [
24]. The structures of mutated proteins were predicted using the predicted wild-type protein and VMD v1.9.3 software [
25]. Next, we carried out an atomistic MD simulation with explicit atom representation for proteins, water, and ions under force-field using the CHARMM package and NAMD v2.3 software [
26]. Periodic boundary conditions, particle mesh Ewald, and a non-bonded cut-off of 14 Å and 2 fs time step were used. The isothermal-isobaric conditions were maintained with a Langevin thermostat (310 K) and Langevin piston barostat (1 atm). For each model, the system was subjected to energy minimization for 1000 steps, followed by equilibration for 5 ns, and the simulation continued for 50 ns without restraints. In the simulations analyses, we used the VEGAZZ v3.1.2 [
27], Carma [
28], and R cran project v3.4 [
29] programs. The MD analysis included the RMSD and the RMSF.
Statistical analysis
Clinical data are reported as mean ± standard deviation. Data were analyzed using R cran project v3.4 [
29] and a general linear model test for comparing metabolite levels between carriers and non-carriers. Gender- and age-adjusted
P < 0.05 was considered significant.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.