Increased dosage of DYRK1A leads to congenital heart defects in a mouse model of Down syndrome

Abstract

Down syndrome (DS) is caused by trisomy of human chromosome 21 (Hsa21). DS is a gene dosage disorder that results in multiple phenotypes including congenital heart defects. This clinically important cardiac pathology is the result of a third copy of one or more of the approximately 230 genes on Hsa21, but the identity of the causative dosage-sensitive genes and hence mechanisms underlying this cardiac pathology remain unclear. Here, we show that hearts from human fetuses with DS and embryonic hearts from the Dp1Tyb mouse model of DS show reduced expression of mitochondrial respiration genes and cell proliferation genes. Using systematic genetic mapping, we determined that three copies of the dual-specificity tyrosine phosphorylation-regulated kinase 1A (Dyrk1a) gene, encoding a serine/threonine protein kinase, are associated with congenital heart disease pathology. In embryos from Dp1Tyb mice, reducing Dyrk1a gene copy number from three to two reversed defects in cellular proliferation and mitochondrial respiration in cardiomyocytes and rescued heart septation defects. Increased dosage of DYRK1A protein resulted in impairment of mitochondrial function and congenital heart disease pathology in mice with DS, suggesting that DYRK1A may be a useful therapeutic target for treating this common human condition.

Figure 1. Transcriptomic similarities in embryonic hearts from human fetuses with DS and mouse models of DS.

(A, B) RNAseq analysis of human DS and euploid embryonic hearts (13-14 post-conception weeks, n=5), showing (A) a volcano plot of fold-change in gene expression (DS versus euploid) against adjusted P-value for significance of the difference, showing Hsa21 genes (red), differentially expressed genes (blue), and DYRK1A (black) (B) unsupervised hierarchical clustering of the 10 samples showing heatmap of differentially expressed genes (C, D) RNAseq analysis of E13.5 hearts from WT and Dp1Tyb embryos (n=5) showing (C) a volcano plot as in A, indicating genes in 3 copies in Dp1Tyb mice (red), differentially expressed genes (blue), and Dyrk1a (black) and (D) hierarchical clustering of the samples as in B. (E) Hallmark gene sets from the Molecular Signatures Database that are significantly enriched or depleted (GSEA, ≤5% FDR) in both human DS and Dp1Tyb mouse hearts; NES, normalized enrichment scores. Gene sets showing the same direction of change in human and mouse data are indicated in bold. (F) Map of Hsa21 (length in Mb) showing cytogenetic bands and regions of orthology to Mmu10, Mmu17 and Mmu16 (grey) and indicating regions of Mmu16 that are duplicated in mouse strains (bold) that show CHD (yellow) and that do not (black); numbers of coding genes indicated below duplicated regions. (G) Comparison of dysregulated gene sets determined by GSEA of RNAseq data from human DS versus euploid embryonic hearts and in hearts from Dp1Tyb, Dp3Tyb and Ts1Rhr mouse embryos compared to WT controls. All show heart defects except Ts1Rhr mice. Colors and sizes of circles indicate NES and FDR q-value, respectively.

Figure 2. scRNAseq reveals similar gene expression changes across different cell types in Dp1Tyb mouse embryonic hearts.

(A) Uniform Manifold Approximation and Projection (UMAP) clustering of scRNAseq data pooled from WT, Dp1Tyb and Dp1TybDyrk1a^+/+/- E13.5 mouse hearts. Each dot represents a single cell; colors indicate 14 clusters whose identity was inferred based on expression of markers genes. V, ventricular; CM, cardiomyocytes; AVC, atrioventricular canal; OFT, outflow tract. (B) Violin plots showing the expression of representative marker genes across the 14 clusters; Y-axis shows the log-scale normalized read count. (C) Stacked column plot showing the percentage of cells in each of the 14 cell populations, colored according to cluster designation. Clusters with significantly altered percentages in Dp1Tyb hearts compared to WT are indicated; Fisher's exact test, * 0.01 < P < 0.05, ** 0.001 < P < 0.01, *** P < 0.001. (D) GSEA of Dp1Tyb versus WT (a) or Dp1Tyb versus Dp1TybDyrk1a^+/+/- (b) scRNAseq data from the 11 most abundant clusters analyzed individually and pooled (pseudo-bulk), showing key pathways and their NES. Colors and sizes of circles indicate NES and FDR q-value, respectively. Sample numbers: n=2 WT, 1 Dp1Tyb, 2 Dp1TybDyrk1a^+/+/- embryonic hearts.

Figure 3. Proliferative defects in Dp1Tyb embryonic hearts.

(A) Flow cytometric analysis of freshly isolated embryonic mouse hearts pulsed with EdU. EdU and DNA content of cardiomyocytes (CD106⁺CD31^-) and endocardial cells (CD106^- CD31⁺) were used to distinguish cells in G1, S and G2/M phases. Numbers indicate percentage of cells in gates. (B) Mean (±SEM) percentage of cells in each cell cycle phase taken from data as in A. n=38 WT, 14 Dp1Tyb, 12 Dp1TybDyrk1a^+/+/- embryonic mouse hearts. (C) Diagram showing how DYRK1A may regulate the cell cycle. Cyclin D2 (CCND2) in complex with CDK4/6 phosphorylates RB1 promoting cell cycle progression. DYRK1A phosphorylates CCND2 leading to its degradation thereby causing reduced CDK4/6 activity, reduced RB1 phosphorylation and impaired cell cycle progression. (D) Representative immunoblot analysis of lysates from WT, Dp1Tyb and Dp1TybDyrk1a^+/+/- E13.5 embryonic hearts probed with antibodies to phospho-RB1 (p-RB1) and GAPDH. Each lane represents an individual embryonic heart. (E) Mean p-RB1 abundance determined by immunoblots such as those in D, normalized to GAPDH and to the mean of WT samples which was set to 1. Dots represent individual embryos. n=27 WT, 14 Dp1Tyb, 12 Dp1TybDyrk1a^+/+/- embryonic hearts. Statistical significance was calculated using a Kruskal-Wallis test; * 0.01 < P < 0.05, ** 0.001 < P < 0.01; ns, not significant.

Figure 4. Mitochondrial defects in Dp1Tyb mouse embryonic cardiomyocytes.

(A) Flow cytometric analysis showing gating strategy used to measure mitochondrial mass (MTG) and mitochondrial potential (TMRM) in cardiomyocytes (CD106⁺CD31^-) and endocardial cells (CD106^-CD31⁺) from E13.5 mouse embryonic hearts of the indicated genotypes. Cardiomyocytes were subdivided into cells that have a high (TMRM-H) and medium (TMRM-M) potential. Numbers indicate percentage of cells in gates. (B) Mean fluorescence intensity (MFI) of MTG and TMRM in endocardial cells and of MTG in cardiomyocytes normalized to the average of WT samples which was set to 1. For cardiomyocytes mitochondrial potential was measured using a TMRM-H/TMRM-M ratio. Dots represent individual embryos. n=17 WT, 13 Dp1Tyb, 15 Dp1TybDyrk1a^+/+/- embryonic hearts. (C) Representative confocal microscopy images of cells from E13.5 mouse hearts of the indicated genotypes showing staining with MitoTracker Deep Red (MTDR – mitochondria, red), anti-CD106 (cardiomyocytes, green) and DAPI (blue). Images show a maximum projection of Z-stacks from 0 to 3 μm with a step size of 1 μm. Insets are enlarged images of the region in the dashed square showing the mitochondrial network. Scale bar 50 μm. (D) Violin plots of mitochondrial aspect ratio and form factor in cardiomyocytes (CD106⁺) determined from images such as those in C (fig. S3A, B). Aspect ratio and form factor are measures of distortion from circularity and degree of branching, respectively (54). Black lines indicate median, dotted lines indicate 25th and 75th centiles. n=25 WT, 10 Dp1Tyb, 17 Dp1TybDyrk1a^+/+/- mouse embryonic hearts. (E) Mean±SEM oxygen consumption rate (OCR) in E13.5 mouse heart cells from embryos of the indicated genotypes analyzed using a Seahorse analyzer with oligomycin (ATP synthase inhibitor), FCCP (depolarizes mitochondrial membrane potential), and rotenone and antimycin (complex I and III inhibitors) added at the indicated times. Basal respiration rate was calculated from the mean of the first three measurements, maximal respiration rate from the three time points after addition of FCCP. (F) Mean basal and maximal respiration rates of E13.5 mouse heart cells normalized to the mean rates in WT hearts. Dots represent individual embryos. n=14 WT, 6 Dp1Tyb, 8 Dp1TybDyrk1a^+/+/- embryonic hearts. (G) Mean±SEM extracellular acidification rate (ECAR) in E13.5 mouse heart cells from embryos of the indicated genotypes analyzed using a Seahorse analyzer with glucose, oligomycin (ATP synthase inhibitor) and 2 deoxy-glucose (2-DG, competitive inhibitor of glucose) added at the indicated times. Glycolysis rate was calculated as the difference between the mean ECAR of the three measurements before and after glucose injection. (H) Mean glycolysis rates of E13.5 mouse heart cells normalized to the mean rates in WT hearts. Dots represent individual embryos. n=35 WT, 8 Dp1Tyb, 10 Dp1TybDyrk1a^+/+/- embryonic hearts. (I) Representative images of sections of E13.5 mouse hearts of the indicated genotypes showing a 4-chamber view stained with anti-Endomucin (endothelial cells, green), anti-Hypoxyprobe (hypoxia, red) and DAPI (blue). Dashed line indicates a region of interest (ROI) encompassing the ventricles and the atrioventricular cushions. Scale bar 200 μm. (J) MFI of anti-Hypoxyprobe in ROI determined from images such as those in I. n=4 WT, 6 Dp1Tyb mouse embryonic hearts. Dots represent individual embryos. LV, left ventricle; RV, right ventricle; LA, left atrium; RA, right atrium; AVC, atrioventricular cushion; VS, ventricular septum. Statistical significance was calculated using a Kruskal-Wallis (B, D, F, H) or Mann Whitney test (J), * 0.01 < P < 0.05, ** 0.001 < P < 0.01, **** P < 0.0001; ns, not significant.

Figure 5. Three copies of Dyrk1a are necessary to cause heart defects.

(A) Map of Hsa21 showing regions of orthology to Mmu10, Mmu17 and Mmu16 (grey) and indicating regions of Mmu16 that are duplicated in mouse strains that show CHD (yellow) and in those that do not (black); genes at boundaries of these duplications are indicated next to the Mmu16 map; numbers of coding genes indicated below duplicated regions. Two genetic intervals containing 3 and 2 candidate genes for CHD are indicated in orange and blue, respectively. (B-G) Graphs of percentage of CHD in E14.5 mouse embryonic hearts from the indicated mouse models, indicating the frequency of AVSD and other CHD, which are predominantly VSDs and occasionally outflow tract defects such as overriding aorta. The orange and blue boxes highlight the genes found in the 2 candidate regions. The cross of Dp1Tyb to Del4Tyb was used to determine if 3 copies of Cbr1 are required for heart defects. Numbers of embryonic hearts analyzed: (B) WT (n=25), Dp1Tyb (n=20), Dp1TybMir802^+/+/- (n=27) hearts with two copies of Mir802; (C) WT (n=24), Dp1Tyb (n=23), Dp1TybSetd4^+/+/- (n=26) hearts with two copies of Setd4; (D) WT (n=7), Dp1Tyb (n=13), Dp1TybDel4Tyb (n=15) hearts with two copies of the region deleted in Del4Tyb; (E) WT (n=34), Dp1Tyb (n=41), Dp1TybKcnj6^+/+/- (n=30) hearts with two copies of Kcnj6; (F) WT (n=91), Dp1Tyb (n=86), Dp1TybDyrk1a^+/+/- (n=23) hearts with two copies of Dyrk1a; (G) WT (n=30), Dp1Tyb (n=23), Dp1TybDyrk1a^+/+/K188R (n=44) hearts with two WT and one kinase-inactive allele of Dyrk1a. Fisher's exact test, * 0.01 < P < 0.05, ** 0.001 < P < 0.01, *** P < 0.001 for difference in number of total CHD, except for Dp1TybDyrk1a^+/+/K188R cohort, where statistics were calculated for number of AVSD; ns, not significant. (H) 3 Dimensional high resolution episcopic microscopy (HREM) rendering of WT, Dp1Tyb, Dp1TybDyrk1a^+/+/- and Dp1TybDyrk1a^+/+/K188R mouse hearts, eroded to show an anterior four-chamber view (top and middle) and a two-chamber view (bottom) seen from the atria at the level of the atrioventricular canal. Top and bottom rows show eroded 3D views, middle row shows 2D sections at the same level as shown in the top row. Red arrowheads indicate VSD (top, middle) or AVSD (bottom). iAVC, inferior atrio-ventricular cushion; LV, left ventricle; MV, mitral valve; RV, right ventricle; sAVC, superior atrio-ventricular cushion; TV, tricuspid valve; VS, ventricular septum.

Figure 6. Increased dosage of Dyrk1a causes key transcriptional changes in Dp1Tyb mouse embryonic hearts.

(A) Volcano plots showing fold-change in gene expression in E13.5 mouse embryonic hearts, Dp1TybDyrk1a^+/+/- versus WT (left) and Dp1Tyb versus Dp1TybDyrk1a^+/+/- (right), plotted against adjusted P-value for significance of the difference. Genes present in three copies in Dp1Tyb mice (red) and differentially expressed genes (blue) and Dyrk1a (black) are indicated. (B) Volcano plots showing fold-change in abundance of proteins and phosphorylated sites in Dp1Tyb versus Dp1TybDyrk1a^+/+/- E13.5 hearts. DYRK1A, CCND1, CCND2 and CCND3 are indicated in green on the proteome plot (left); phosphorylated sites known to be DYRK1A targets and an autophosphorylation site on DYRK1A are indicated in green on the phosphoproteome plot (right). (C) Comparison of dysregulated pathways determined by GSEA of RNAseq and proteomic experiments. Colors and sizes of circles indicate NES and FDR q-value, respectively. Sample numbers: n=5 embryonic hearts.

Figure 7. Dyrk1a expression in a broad range of cell types in the developing mouse heart.

(A) Violin plots showing expression of Dyrk1a in mouse E13.5 hearts across 14 cell clusters identified in Fig. 2B. Dots indicate single cells. Sample numbers: n=5 embryonic hearts. (B) Left, schematic illustrations of a sagittal section (top) of an E12.5 mouse heart and a 4-chamber view (bottom) of an E14.5 heart. Right, RNAscope analysis of Dyrk1a expression (brown dots) in sections of the right ventricle myocardial wall (1) and atrioventricular cushions (2) at E12.5 and right ventricular myocardial wall (3) and ventricular septum (4) at E14.5; sections were counterstained with hematoxylin (blue). Scale bar 50 μm. iAVC, inferior atrioventricular cushion; LA, left atrium; LV, left ventricle; OFT, outflow tract; PT, pulmonary trunk; RA, right atrium; RV, right ventricle; sAVC, superior atrioventricular cushion; VS, ventricular septum.

Figure 8. Pharmacological inhibition of Dyrk1a partially rescues CHD in the Dp1Tyb mouse model of DS.

(A) C57BL/6J females that had been mated with Dp1Tyb males, were treated daily (vertical arrows) by oral gavage with Leucettinib-21 (LCTB-21) or iso-Leucettinib-21 (iso-LCTB-21) from 5 days after vaginal plug (VP) was found (embryonic day 0.5, E0.5). Embryos were collected at E13.5 for RNAseq, or at E14.5 for HREM. (B) Scatter plots comparing the log₂(fold-change) (Log2FC) of mRNA expression for all genes (black) for Dp1Tyb versus WT embryos both treated with iso-LCTB-21 or Dp1Tyb embryos treated with LCTB-21 versus WT embryos treated with iso-LCTB-21. Genes from the Hallmark genesets for Oxidative Phosphorylation (left), Myc targets V1 (middle) and inflammatory response (right), are highlighted in red. n=5 for each condition. (C) Violin plots showing the Log2FC of expression of the same genesets as in B, Oxidative Phosphorylation (left), Myc targets V1 (middle), and inflammatory response (right), for the following comparisons: untreated Dp1Tyb versus WT, untreated Dp1TybDyrk1a^+/+/- versus WT, Dp1Tyb treated with iso-LCTB-21 versus WT treated with iso-LCTB-21 and Dp1Tyb treated with LCTB-21 versus WT treated with iso-LCTB-21. Black lines indicate median, dotted lines indicate 25th and 75th centiles. Dotted line at 0 indicates no change. n=5 for each condition. (D) Graph of percentage of CHD in E14.5 mouse embryonic hearts from the indicated models. Number of hearts analyzed: WT (n=34) and Dp1Tyb (n=26) treated with iso-LCTB-21, and Dp1Tyb treated with LCTB-21 (n=32). (E) Three copies of Dyrk1a and a second unknown gene (GeneX) lead to impaired proliferation and mitochondrial respiration in cardiomyocytes which is required for correct septation of the heart. Created in Biorender. Statistical tests were carried out with a Kruskal-Wallis (C) or Fisher's exact (D) test; * 0.01 < P < 0.05, **** P < 0.0001; ns, not significant.