DYRK1A is overexpressed in human T21 cells. To assess the role of DYRK1A in the pathogenesis of ML-DS arising from fetal progenitors, we measured its expression in euploid and T21 fetal liver–derived megakaryocytes, as well as in undifferentiated euploid and T21 iPSCs and iPSC-derived megakaryocytes with and without the GATA1s mutation. Compared with euploid cells, DYRK1A was overexpressed in T21 cells in all 3 tissue types (Figure 1, A–C, and Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.172851DS1). Notably, in iPSC-derived megakaryocytes, DYRK1A expression was further increased in cells with GATA1s (T21/GATA1s) compared with wtGATA1 (Figure 1C and Supplemental Figure 1C). While the levels of DYRK1A expression do vary over the course of differentiation at all time points, T21/GATA1s megakaryocytes expressed increased DYRK1A compared with T21/wtGATA1 (Supplemental Figure 10B). Although T21 typically results in small increases in HSA21 gene expression, this marked increase in DYRK1A in T21 tissues, and specifically in megakaryocytes expressing GATA1s, prompted our studies to focus on DYRK1A. CRISPR/Cas9 gene editing was used to knock out all isoforms of human DYRK1A in isogenic T21 iPSC lines with wtGATA1 or GATA1s that previously had been generated by our laboratory (32, 33) (Supplemental Figure 2 and Supplemental Table 1). DYRK1A loss (DYRK1A–/–/–) was confirmed by Western blotting (Figure 1D) and clones analyzed are listed in Supplemental Table 2.

DYRK1A loss results in aberrant hematopoietic progenitor potential. (A) Western blot for DYRK1A expression in CD61-selected ex vivo fetal liver–derived megakaryocytes, (B) undifferentiated iPSCs, and (C) iPSC-derived megakaryocytes on day 6 of liquid culture from euploid or T21 iPSCs, with wild-type GATA1 (G1) or GATA1 short (G1s). Percentage CD41+CD42+ cells from lanes 1 to 4: 70%, 68%, 76%, and 78%, respectively. (D) Western blot for DYRK1A expression in DYRK1A-untargeted and -targeted T21/wtGATA1 and T21/GATA1s iPSCs. Each lane represents an individual iPSC clone. (E) Flow cytometric analysis of hematopoietic progenitors on day 7 from EB differentiation of T21/wtGATA1 or T21/GATA1s iPSCs, with DYRK1A WT (+/+/+) or knockout (–/–/–). (F) Absolute CD43+ progenitor yield on day 7 of EB differentiation, normalized to starting DYRK1A+/+/+ iPSC number. Each column represents an individual iPSC clone. n = 14–15 independent experiments per clone. Data represent the mean ± SEM. Statistical significance was determined by ordinary 1-way ANOVA. ***P ≤ 0.001, ****P ≤ 0.0001.

DYRK1A loss alters hematopoietic progenitor cells. To evaluate the effects of DYRK1A deficiency on the blood-forming capacities of T21/wtGATA1 and T21/GATA1s iPSCs, we performed hematopoietic differentiation by embryoid body (EB) formation (34, 35). On day 7 of EB differentiation, both CD43+CD41+CD235+ and CD43+CD41–CD235– hematopoietic progenitor populations were produced from DYRK1A+/+/+ and DYRK1A–/–/– lines (Figure 1E). Consistent with prior studies, both CD43+CD41+CD235+ and CD43+CD41–CD235– populations were multipotent (36). Of note, T21/GATA1s progenitors did not generate erythroid cells (32), and consistently showed lower CD235 mean fluorescence intensity (MFI) compared with T21/wtGATA1 progenitors (Supplemental Figure 3). DYRK1A–/–/– resulted in lower CD41 and CD235 MFI regardless of GATA1 status (Supplemental Figure 3), and in the T21/GATA1s/DYRK1A–/–/– clones, the CD43+ day 7 progenitor population was skewed toward CD41–CD235– (Figure 1E). The absolute number of day 7 CD43+ progenitors generated per iPSC undergoing differentiation was decreased in both T21/wtGATA1 and T21/GATA1s lines when DYRK1A was disrupted (Figure 1F). Together, these data suggest that loss of DYRK1A alters hematopoietic progenitor cell production in the setting of T21.

DYRK1A loss enhances megakaryocyte and myeloid expansion in T21 progenitors with GATA1s. To determine the effects of DYRK1A loss on erythroid, megakaryocyte, and myeloid lineages, CD43+ progenitors from day 7 of hematopoietic differentiation of T21/wtGATA1 and T21/GATA1s iPSCs with DYRK1A+/+/+ or DYRK1A–/–/– were assayed by lineage-specific liquid cultures. Consistent with our previous findings (32), T21/GATA1s/DYRK1A+/+/+ progenitors showed enhanced megakaryocytic and myeloid growth capacity compared with T21/wtGATA1/DYRK1A+/+/+ progenitors, as measured by fold-change growth (gray dots; Figure 2, A and B). In the context of T21/wtGATA1, DYRK1A+/+/+ and DYRK1A–/–/– progenitors differentiated into megakaryocytic (CD41+CD42+), myeloid (CD45+CD18+), or erythroid (CD71+CD235+) lineages with similar frequency and proliferative capacity (Figure 2, A–C). In contrast, loss of DYRK1A in T21/GATA1s enhanced the fold-change growth of megakaryocyte and myeloid cells from total day 7 CD43+ progenitors (Figure 2, A and B), as well as flow cytometry–purified CD43+CD41+CD235+ and CD43+CD41–CD235– cells (data not shown). Independently generated iPSC clones from another T21 individual with TAM confirmed the decrease in absolute number of CD43+ progenitors and increased megakaryocyte fold-change growth with DYRK1A loss (Supplemental Figure 4, A and B). While generating clones with 1- or 2-allele DYRK1A disruption was challenging given the high efficiency of the CRISPR/Cas9 guide RNAs (gRNAs), analysis of a clone with 2 disrupted alleles demonstrated an intermediate phenotype with decreased progenitor production and increased megakaryocyte expansion, but to a lesser extent compared with 3-allele DYRK1A–knockout clones (Supplemental Figure 5).

DYRK1A loss selectively enhances megakaryocyte and myeloid potential in the context of T21/GATA1s. Fold change of indicated lineages from T21/wtGATA1 or T21/GATA1s iPSC–derived CD43+ hematopoietic progenitors with DYRK1A WT (+/+/+) or knockout (–/–/–), differentiated in liquid culture for 6 days to support (A) megakaryocyte (CD41+CD42+), (B) myeloid (CD45+CD18+), or (C) erythroid (CD235+CD71+) cell growth. Each column represents an individual iPSC clone. n = 14–17, 5–7, and 6–7 independent experiments per clone for panels A, B, and C, respectively. (D) Megakaryocyte (CD41+CD42+) fold-change growth from euploid/GATA1s/DYRK1A–/– CD43+ progenitors on day 6 of megakaryocyte liquid culture, compared with T21/GATA1s with DYRK1A WT (+/+/+) or knockout (–/–/–). n = 3 independent experiments per clone. (E) Megakaryocyte fold-change growth from T21/GATA1s CD43+ progenitors with DYRK1A WT (+/+/+) or knockout (–/–/–) transduced with lentivirus containing the HMD vector to express enhanced green fluorescence protein (GFP) or HMD-DYRK1A to express GFP and DYRK1A, and subsequently cultured for 6 days. Fold change normalized to day 2 GFP+. n = 3 independent experiments. Data represent the mean ± SEM. Statistical significance was determined by ordinary 1-way ANOVA. NS, not significant. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

We next used colony-forming assays to determine lineage bias of CD43+ progenitors from T21/wtGATA1 and T21/GATA1s lines with DYRK1A+/+/+ or DYRK1A–/–/– to produce megakaryocyte, myeloid, and erythroid colonies. Day 7 CD43+ (CD43+CD41+CD235+ and CD43+CD41–CD235– that are CD45–CD18–) progenitor cells were seeded into colony-forming megakaryocyte (CFU-Mk) assays containing thrombopoietin (TPO), interleukin-3 (IL-3), and IL-6, or methylcellulose assays with stem cell factor (SCF), IL-3, erythropoietin (EPO), and granulocyte-macrophage colony–stimulating factor (GM-CSF) to support myeloid and erythroid colony formation. We observed no difference in megakaryocyte, myeloid, or erythroid colony formation among day 7 CD43+ progenitors from T21/wtGATA1 or T21/GATA1s with DYRK1A+/+/+ or DYRK1A–/–/– (Supplemental Figure 6, A and B). Taken together with the fold-change growth of T21/GATA1s/DYRK1A–/–/– CD43+ progenitors in megakaryocyte and myeloid liquid cultures, DYRK1A loss combined with T21/GATA1s may enhance megakaryocyte and myeloid cell proliferation but does not alter their cell fate or lineage potential.

While gene targeting the T21/GATA1s iPSCs, we identified a clone that lost the redundant copy of HSA21. Interestingly, DYRK1A knockout in euploid/GATA1s iPSC–derived progenitors demonstrated enhanced megakaryocyte proliferation similar to the DYRK1A loss in the context of T21, suggesting the interplay between DYRK1A and GATA1s may not require the extra copy of HSA21 (Figure 2D). Conversely, DYRK1A overexpression by lentiviral transduction decreased megakaryocyte growth from both T21/GATA1s/DYRK1A+/+/+ and T21/GATA1s/DYRK1A–/–/– CD43+ hematopoietic progenitors (Figure 2E). The restricted growth with DYRK1A overexpression contrasts with murine studies where Dyrk1a overexpression in euploid wtGATA1 or GATA1s bone marrow cells induced expansion of megakaryocytes (16). Our results suggest a species-specific role of DYRK1A in human progenitor cells, where DYRK1A restrains megakaryocyte expansion and its loss combined with GATA1s further exacerbates the enhanced megakaryocyte growth observed with GATA1s mutations.

DYRK1A loss enhances megakaryocyte proliferation in the context of T21/GATA1s. DYRK1A has diverse roles in the cell cycle by regulating the balance between cell cycle entry and quiescence, which is dependent on cellular context (37). We assayed cell cycle stages in T21/wtGATA1 and T21/GATA1s CD41+CD42b+ megakaryocytes with DYRK1A+/+/+ or DYRK1A–/–/– by measuring EdU DNA incorporation on days 5 to 6 of culture (Figure 3A). We previously demonstrated that GATA1s enhances megakaryocyte expansion compared with wtGATA1 irrespective of HSA21 status (32). On average, T21/GATA1s/DYRK1A+/+/+ megakaryocytes incorporated EdU in 22% of cells, representing those in S phase, compared with 7.5% in T21/wtGATA1/DYRK1A+/+/+ megakaryocytes (Figure 3B). The number of megakaryocytes in S phase increased further to an average of 32% with T21/GATA1s/DYRK1A–/–/–, while the percentage of T21/wtGATA1 megakaryocytes in S phase was not significantly changed with DYRK1A loss (Figure 3, A and B). Consistent with EdU assays, T21/GATA1s/DYRK1A+/+/+ megakaryocytes showed an increased proportion of Ki67+ mitotically active cells compared with T21/wtGATA1/DYRK1A+/+/+, which was further increased with DYRK1A loss in T21/GATA1s but not T21/wtGATA1 cells (Figure 3C). DYRK1A loss had no effect on apoptosis in either T21/wtGATA1 or T21/GATA1s megakaryocytes, as measured by flow cytometry for Annexin V (Figure 3D). These findings support a model where DYRK1A loss specifically increases megakaryocyte proliferation in the context of T21/GATA1s.

DYRK1A loss selectively enhances megakaryocyte proliferation in the context of T21/GATA1s. (A) Representative EdU flow cytometric analysis of CD41+CD42b+ megakaryocytes differentiated from T21/wtGATA1 and T21/GATA1s iPSCs with DYRK1A WT (+/+/+) or knockout (–/–/–). (B) Percentage EdU-, (C) percentage Ki67-, and (D) percentage Annexin V–positive CD41+CD42b+ megakaryocytes on day 5 to 6 of megakaryocyte culture of T21/wtGATA1 and T21/GATA1s CD43+ progenitors with DYRK1A WT (+/+/+) or knockout (–/–/–). n = 3–5 independent experiments per genotype. Data represent the mean ± SEM. Statistical significance was determined by ordinary 1-way ANOVA. NS, not significant. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.

DYRK1A loss impairs maturation of T21/GATA1s megakaryocytes. Since proliferation and maturation are often uncoupled, we performed characterization of T21/wtGATA1 and T21/GATA1s megakaryocytes with DYRK1A+/+/+ or DYRK1A–/–/– to determine whether DYRK1A loss impairs maturation. We observed several features in T21/GATA1s/DYRK1A–/–/– megakaryocytes consistent with an impaired capacity to mature. First, we found a significant decrease in CD41 and CD42b MFI in T21/GATA1s/DYRK1A+/+/+ megakaryocytes that was further reduced by DYRK1A loss, consistent with an immature phenotype (Figure 4, A and B). DYRK1A loss in T21/GATA1s megakaryocytes was associated with decreased forward scatter (FSC) and side scatter (SSC), as measured by flow cytometry (Supplemental Figure 7A); smaller size and less granularity are consistent with less mature megakaryocytes. Lentiviral overexpression of DYRK1A rescued this phenotype, with increased CD41 MFI in the megakaryocytes differentiated from transduced T21/GATA1s/DYRK1A+/+/+ and T21/GATA1s/DYRK1A–/–/– CD43+ progenitors (Supplemental Figure 7B).

DYRK1A loss selectively impairs megakaryocyte activation and maturation in the context of T21/GATA1s. (A) Representative and (B) composite CD41 mean fluorescence intensity (MFI) of megakaryocytes from T21/wtGATA1 and T21/GATA1s iPSCs with DYRK1A WT (+/+/+) or knockout (–/–/–), assayed on day 5 or 6 of megakaryocyte liquid culture. n = 11 independent experiments per genotype. (C) Representative and (D) composite flow cytometric analyses for PAC-1 after thrombin stimulation of megakaryocytes. n = 3–5 independent experiments per genotype. Data represent the mean ± SEM. Statistical significance was determined by ordinary 1-way ANOVA. NS, not significant. *P ≤ 0.05, **P ≤ 0.01, ****P ≤ 0.0001.

To test the effect of DYRK1A loss on megakaryocyte function, we analyzed their responsiveness to stimulation with thrombin. Activation with thrombin results in a conformational change in the surface integrin αIIbβ3 receptors that enables the binding of the conformation-specific monoclonal antibody, PAC-1 (38). All megakaryocytes showed minimal basal PAC-1 binding, demonstrating no spontaneous preactivation of these cells in culture. Upon thrombin stimulation, T21/GATA1s/DYRK1A+/+/+ megakaryocytes demonstrated an approximately 30% lower level of PAC-1 binding following thrombin activation compared with T21/wtGATA1/DYRK1A+/+/+ (average 38% vs. 66%, P ≤ 0.05). DYRK1A loss resulted in a significant further decrease in PAC-1 binding in T21/GATA1s megakaryocytes (average 15%, P ≤ 0.05), providing functional evidence of impaired megakaryocyte maturation, while it had no effect on T21/wtGATA1 megakaryocytes (Figure 4, C and D).

T21/GATA1s/DYRK1A–/–/– megakaryocytes have a proliferative gene signature. To investigate the mechanism underlying DYRK1A’s effect on T21/GATA1s hematopoietic cells, we performed RNA sequencing (RNA-seq) of flow cytometry–purified CD43+CD41+CD235+ progenitors and CD41+CD42b+ megakaryocytes from T21/GATA1s iPSCs with DYRK1A+/+/+ or DYRK1A–/–/–. Among the 21,451 genes examined, 15,159 were expressed in at least 1 cell type. Overall, 1,380 genes were differentially expressed in hematopoietic progenitors, with 654 genes upregulated and 726 downregulated with DYRK1A–/–/–. In megakaryocytes, 1,187 genes were differentially expressed, with 444 genes upregulated and 743 downregulated with DYRK1A–/–/– (Benjamini-Hochberg FDR < 0.25) (Figure 5A and Supplemental Figure 8).

T21/GATA1s/DYRK1A–/–/– megakaryocytes demonstrate enhanced cell proliferation and decreased platelet signaling. Flow cytometry–sorted T21/GATA1s/DYRK1A+/+/+ and DYRK1A–/–/– CD41+CD43+CD235+ progenitors or CD41+CD42b+ megakaryocytes on day 4 of megakaryocyte liquid culture for RNA analyses. (A) Volcano plots showing differential gene expression for T21/GATA1s/DYRK1A–/–/– compared with DYRK1A+/+/+. Each dot represents 1 gene, with gating of the dot reflecting the clustering information for each gene; dots gated in red squares are genes that are upregulated compared with DYRK1A+/+/+; dots gated in blue square are genes that are downregulated compared with DYRK1A+/+/+. (B) Up- and downregulated pathways from gene set enrichment analysis (GSEA) of RNA-seq data. (C) GSEA for indicated pathways comparing DYRK1A–/–/– to DYRK1A+/+/+. NES, normalized enrichment score; FDR, false discovery rate. (D) Relative mRNA expression levels for CCND1, CCND2, GATA2, and E2F target genes. (E) Relative mRNA expression levels for megakaryocyte-related genes. n = 5–7 independent experiments per genotype. Data represent the mean ± SEM. Statistical significance was determined by 2-tailed, unpaired t test. *P ≤ 0.05, **P ≤ 0.01, ****P ≤ 0.0001.

Gene set enrichment analysis (GSEA) of T21/GATA1s progenitors with DYRK1A–/–/– compared with DYRK1A+/+/+ showed increased enrichment of pathways related to endothelial and CD34+ cells consistent with the abnormal progenitor production we observed, and decreased enrichment of interferon signaling genes, which have an important role in the self-renewal, quiescence, and differentiation of HSCs (Supplemental Figure 9) (39). Consistent with our observed phenotype of enhanced proliferation in T21/GATA1s/DYRK1A–/–/– megakaryocytes, GSEA in megakaryocytes but not progenitors identified enrichment of genes related to cell cycle, cell proliferation, and cell division (Figure 5B). E2F targets were enriched in T21/GATA1s/DYRK1A–/–/– megakaryocytes (Figure 5C), suggesting DYRK1A loss in addition to GATA1s derepresses E2F (9–11). DYRK1A is also important for the formation of the DREAM complex, for repression of DREAM target genes, and for induction of quiescence (28, 40). Genes typically repressed by the DREAM complex were enriched in T21/GATA1s/DYRK1A–/–/– megakaryocytes (Figure 5C), suggesting DYRK1A loss results in DREAM complex dissociation. The DREAM complex typically represses activator E2Fs (E2F1, -2, and -3), but with its dissociation these bind to downstream targets to promote cell cycle entry. DNA damage repair responses that are required for highly proliferating cancer cells to maintain chromosome stability were also enriched (Figure 5, B and C). Consistent with impaired maturation, T21/GATA1s/DYRK1A–/–/– megakaryocytes showed decreased enrichment for genes involved in platelet activation and aggregation (Figure 5, B and C).

Among all genes, CCND2 (cyclin D2) and the hematopoietic transcription factor GATA2 were among the top 5 most upregulated in T21/GATA1s/DYRK1A–/–/– megakaryocytes besides replication-dependent histones. Quantitative reverse transcriptase PCR (qRT-PCR) confirmed upregulation of these genes (Figure 5D). Other E2F-related or target genes, such as CCND1 (cyclin D1), CCNA2 (cyclin A2), CDC6, CDC25A, E2F8, MCM2, and CHK1 were upregulated in T21/GATA1s/DYRK1A–/–/– megakaryocytes, while platelet and megakaryocyte pathway–related genes, including VWF, PF4, GP9, and SELP were downregulated (Figure 5E). Overall, these data are consistent with our findings that T21/GATA1s/DYRK1A–/–/– megakaryocytes show enhanced proliferation but reduced maturation and activation potential.

DYRK1A loss increases expression of cell cycle and proliferative proteins in T21/GATA1s megakaryocytes. DYRK1A has been shown to regulate the cell cycle and inhibit proliferation through the cyclin D/CDK-Rb/E2F pathway (19, 37). Since E2F targets, D-type cyclins, and GATA2 mRNA levels were upregulated in T21/GATA1s/DYRK1A–/–/– megakaryocytes, we compared protein expression of cyclin, Rb, and GATA in T21/wtGATA1 and T21/GATA1s megakaryocytes with DYRK1A+/+/+ or DYRK1A–/–/–. Without DYRK1A manipulation, many cell cycle–related genes were already expressed at a higher level in T21/GATA1s megakaryocytes compared with T21/wtGATA1, including cyclin D2, cyclin B1, CDK1, and cyclin A2 (Figure 6A), consistent with a proliferative state. Comparable levels of cyclin D1, cyclin D3, and CDK4 were observed. In both T21/wtGATA1 and T21/GATA1s megakaryocytes, DYRK1A loss resulted in upregulation of cell cycle–related genes, including all 3 D-type cyclins (D1, D2, and D3), cyclin B1, cyclin A2, and CDK4 (Figure 6A). Western blot analysis of megakaryocytes from an independent patient-derived iPSC line confirmed the upregulation of cell cycle–related genes in association with T21/GATA1s and DYRK1A–/–/– (Supplemental Figure 10, A and B).

Increased expression of proliferative genes in T21/GATA1s/DYRK1A–/–/– megakaryocytes. (A) Representative Western blot analysis of day 4 T21/wtGATA1 or T21/GATA1s megakaryocytes with DYRK1A WT (+/+/+) or knockout (–/–/–) for DYRK1A and cell cycle–related proteins. (B and D) Phosphorylation of cyclin D2 (pT280) and (C) cyclin D3 (pT283) quantified by Western blot band intensity and normalized to total cyclin D2 or D3 expression. n = 3–4 independent experiments per genotype. Data represent the mean ± SEM. Statistical significance was determined by 2-tailed, unpaired t test. *P ≤ 0.05, **P ≤ 0.01. (E) Representative Western blot analysis of day 4 T21/wtGATA1 or T21/GATA1s megakaryocytes with DYRK1A WT (+/+/+) or knockout (–/–/–) for phosphorylated (pRb), hypophosphorylated (hypo-pRb), and total Rb, and (F) GATA1 and GATA2.

In cardiomyocytes and lymphocytes, DYRK1A promotes quiescence through phosphorylation of cyclin D2 at Thr280 and cyclin D3 at Thr283, respectively, leading to their ubiquitination and degradation (19, 22). While total cyclin D2 and D3 expression was increased in T21/wtGATA1 and T21/GATA1s megakaryocytes with DYRK1A–/–/–, the proportion that was phosphorylated, i.e., cyclin D2 pT280 and cyclin D3 pT283, was decreased, which would facilitate cell cycle entry (Figure 6, B and C, and Supplemental Figure 11). Of note, the proportion of phosphorylated cyclin D2 (pT280) to total cyclin D2 was lower in T21/GATA1s/DYRK1A+/+/+ compared with T21/wtGATA1/DYRK1A+/+/+ megakaryocytes (Figure 6D and Supplemental Figure 11), consistent with cyclin D2 stabilization and a proliferative state.

When cells are stimulated to enter the cell cycle, cyclin D–CDK4/6 complexes phosphorylate Rb (41). In T21/GATA1s megakaryocytes, higher levels of phosphorylated Rb were observed compared with T21/wtGATA1 (Figure 6E). DYRK1A loss resulted in a further increase in Rb phosphorylation and a decrease in hypophosphorylated Rb (Figure 6E), which lead to release of activator E2Fs and increased cell cycle entry.

While GATA1 and GATA2 expression overlap in megakaryocytes, these well-described GATA switches can alter transcription and cell fate (42, 43). DYRK1A knockout did not alter full-length GATA1 or GATA1s protein levels in iPSC-derived megakaryocytes (Figure 6F). However, GATA2, which is associated with megakaryocyte expansion and is overexpressed in ML-DS samples (12), was among the top upregulated genes in the RNA-seq analysis in T21/GATA1s/DYRK1A–/–/– megakaryocytes. GATA2 expression was increased in T21/GATA1s/DYRK1A+/+/+ megakaryocytes compared with T21/wtGATA1/DYRK1A+/+/+ (Figure 6F and Supplemental Figure 10B). We observed a striking increase in GATA2 expression with DYRK1A loss in T21/GATA1s megakaryocytes (Figure 6F), consistent with enhanced proliferation of immature megakaryocytes and dysregulated GATA2 expression that has been linked to myeloid leukemogenesis (43) and ML-DS (12).