Vascular defects of DYRK1A knockouts are ameliorated by modulating calcium signaling ... · knockout mutants of dyrk1aa, a mammalian DYRK1A homolog, named dyrk1aakrb1, generated by

RESEARCH ARTICLE

Vascular defects of DYRK1A knockouts are ameliorated bymodulating calcium signaling in zebrafishHyun-Ju Cho1,2,3, Jae-Geun Lee1,2, Jong-Hwan Kim2,4, Seon-Young Kim2,4, Yang Hoon Huh5, Hyo-Jeong Kim5,Kyu-Sun Lee2,6, Kweon Yu1,2,3 and Jeong-Soo Lee1,3,*

ABSTRACTDYRK1A is a major causative gene in Down syndrome (DS). Reducedincidence of solid tumors such as neuroblastoma in DS patients andincreased vascular anomalies in DS fetuses suggest a potential role ofDYRK1A in angiogenic processes, but in vivo evidence is still scarce.Here, we used zebrafish dyrk1aa mutant embryos to understandDYRK1A function in cerebral vasculature formation. Zebrafishdyrk1aa mutants exhibited cerebral hemorrhage and defects inangiogenesis of central arteries in the developing hindbrain. Suchphenotypes were rescued by wild-type dyrk1aa mRNA, but not by akinase-dead form, indicating the importance of DYRK1A kinaseactivity. Chemical screening using a bioactive small molecule libraryidentified a calcium chelator, EGTA, as one of the hits that mostrobustly rescued the hemorrhage. Vascular defects of mutants werealso rescued by independent modulation of calcium signaling byFK506. Furthermore, the transcriptomic analyses supported thealterations of calcium signaling networks in dyrk1aa mutants.Together, our results suggest that DYRK1A plays an essential rolein angiogenesis and in maintenance of the developing cerebralvasculature via regulation of calcium signaling, which may havetherapeutic potential for DYRK1A-related vascular diseases.

KEY WORDS: DYRK1A, Vascular development, Hemorrhage,Zebrafish embryo

INTRODUCTIONThe cerebral vasculature plays an essential role in maintaining thehomeostasis of the brain by providing oxygen and nutrients andremoving waste products. During development, new branches of thecerebral vasculature are formed through angiogenesis by endothelialcells, the primary cell component of the vasculature, via complexcell-cell interactions and signaling pathways from existing vessels

(Udan et al., 2013). The cerebral vasculature also contributesto the formation of the neurovascular unit (NVU), which comprisespericytes, astrocytes, microglia and neurons in addition toendothelial cells. Compromise of the normal development orfunction of the NVU has been implicated in childhood braindevelopment disorders and adult neurological dysfunction(Quaegebeur et al., 2011; Zlokovic, 2008). In the NVU, cerebralendothelial cells, connected mainly by tight junction proteins, buildthe blood-brain barrier that acts as a primary semipermeable barrierand confers a high selectivity for molecular exchanges between theblood and the brain parenchyma (Dejana et al., 2009). Therefore,the inappropriate development of cerebral endothelial cells maylead to defects in angiogenesis and/or endothelial permeability,which are closely linked to vascular pathologies such as vascularmalformations and stroke (Folkman, 1995; Pandya et al., 2006).

The development of the brain vasculature is coordinated byvarious extra- and intracellular signals. Among them, calciumsignaling is one of the major regulators of vascular developmentand related pathogenesis. Vascular endothelial growth factorsignals and various stimuli trigger the change of intracellular Ca2+

levels that act as second messengers in endothelial cells, which inturn affect the activity of transcription factors for angiogenesis,such as nuclear factor of activated T-cells, via Ca2+-dependentcalmodulin/calcineurin activity (Hogan et al., 2003; Loh et al.,1996). In addition, an overload of Ca2+ can cause endothelialbarrier dysfunction; intracellular Ca2+ release via activationof IP3 receptors (IP3R) or ryanodine receptors (RyRs) of theendoplasmic reticulum into the cytoplasm can increase vascularpermeability through disorganization of VE-cadherin (Cdh5) orcytoskeletal rearrangement (Gao et al., 2000; Shen et al., 2010;Tiruppathi et al., 2002). Intriguingly, it has been reported thatcalcium supplementation for bone health can unexpectedly inducestroke and cardiovascular diseases (Reid and Bolland, 2008), andcalcium channel blockers and calcium antagonists have beenclinically used as therapeutic agents for strokes and blood vesseldysfunction (Inzitari and Poggesi, 2005). Thus, understandingthe detailed underlying molecular mechanisms of Ca2+ signalingin angiogenesis and vascular permeability along with theidentification of key players may provide an importanttherapeutic means for treating vascular diseases.

Dual-specificity tyrosine phosphorylation-regulated kinase 1A(DYRK1A) is a serine-threonine kinase that has a dual kinaseactivity capable of autophosphorylating its own tyrosine residues andphosphorylating other substrates. The DYRK1A gene was firstidentified in a Drosophila screening as a mutant minibrain, namedfor thebrainmorphologydefectswith reduced brain size (Tejedor et al.,1995).DYRK1A is located in aDown syndrome critical region (DSCR)and is best known as a major causative gene that is implicated in brainfunction, neurological defects and neurofibrillary tangle formation inDown syndrome (DS) (Liu et al., 2008; Wegiel et al., 2011).Received 3 September 2018; Accepted 11 April 2019

1Disease Target Structure Research Center, Korea Research Institute of BioscienceandBiotechnology, 125Gwahak-ro, Yuseong-gu,Daejeon, 34141,Republic of Korea.2KRIBB School, University of Science and Technology, 125 Gwahak-ro, Yuseong-gu,Daejeon, 34141, Republic of Korea. 3Dementia DTC R&D Convergence Program,Korea Institute of Science and Technology, Hwarang-ro 14-gil 5, Seongbuk-gu, Seoul,02792, Republic of Korea. 4Genome Editing Research Center, Korea ResearchInstitute of Bioscience and Biotechnology, 125 Gwahak-ro, Yuseong-gu, Daejeon,34141, Republic of Korea. 5Electron Microscopy Research Center, Korea BasicScience Institute, 162 Yeongudanji-ro, Ochang-eup, Cheongwon-gu, Cheongju-si,Chungcheongbuk-do, 28119, Republic of Korea. 6HazardsMonitoringBNTResearchCenter, Korea Research Institute of Bioscience and Biotechnology, 125 Gwahak-ro,Yuseong-gu, Daejeon, 34141, Republic of Korea.

*Author for correspondence ( [email protected])

J.-S.L., 0000-0002-8491-4429

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2019. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

Interestingly, epidemiological studies suggested that DS patientshave a reduced incidence of angiogenesis-related solid tumors(Hasle et al., 2000; Nižetic and Groet, 2012) and carry numerousvascular anomalies such as umbilico-portal system anomalies,vertebral and right subclavian artery defects, and pulmonary veinstenosis (Gowda et al., 2014; Pipitone et al., 2003; Rathore andSreenivasan, 1989; Stewart et al., 1992). Furthermore, they alsohave an increased incidence of Moyamoya disease and cerebralamyloid angiopathy, which are associated with a cerebrovasculardysfunction and intracerebral hemorrhage (de Borchgrave et al.,2002; Donahue et al., 1998; Jastrzebski et al., 2015; Mito andBecker, 1992; Sabde et al., 2005). Consistent with the potential roleof DYRK1A in angiogenesis, the TS65Dn mouse model of DS,trisomic for the Dyrk1a gene, exhibited reduced tumor growth,presumably by suppressing tumor angiogenesis (Baek et al., 2009).Although vascular defects had not been directly associated withhuman DYRK1A haploinsufficiency syndrome, it has been reportedthat nearly 75% of children with autism, some of which may haveDYRK1A mutations or its reduced activity (Kim et al., 2017),exhibited hypoperfusion in the brain detected by neuroimaging(Bjorklund et al., 2018; Zilbovicius et al., 2000). Also, retinalangiogenesis is disrupted in Dyrk1a heterozygote mice (Rozenet al., 2018), suggesting a role of DYRK1A loss-of-function inregulating angiogenesis in the brain. Taken together, DYRK1A maybe implicated in vascular formation and/or function, and this couldprovide a new perspective to understanding DYRK1A-relatedpathogenesis; however, in vivo evidence of DYRK1A function invascular pathology is scarce.To investigate the role of DYRK1A in vascular formation, we

adopted developing zebrafish as a model organism. Zebrafish is avertebrate animal model used for genetic studies of human diseasesexhibiting a high similarity to humans at anatomical and molecularlevels, especially in the vascular and nervous system (Isogai et al.,2001; Schmidt et al., 2013). Zebrafish embryos can be readilymanipulated for genetic gain-of-function studies with transgenesisor mRNA overexpression, and loss-of-function studies with geneknockouts or morpholino use (Clark et al., 2011; Hogan et al., 2008;Timme-Laragy et al., 2012; Varshney et al., 2015; Zu et al., 2013).Large clutch sizes and various inexpensive and fast experimentaltechniques allow the use of zebrafish for unique high-throughput invivo small molecule screening, which enables the identification ofhit compounds and gives insights into potential underlyingmechanisms (MacRae and Peterson, 2015).We have recently reported autistic behavioral phenotypes of

knockout mutants of dyrk1aa, a mammalian DYRK1A homolog,named dyrk1aakrb1, generated by transcription activator-likeeffector nucleases (Kim et al., 2017). In the current study, weinvestigated a role of DYRK1A in cerebrovascular developmentduring embryogenesis using the dyrk1aa loss-of-function mutants.The dyrk1aakrb1 mutants exhibited cerebral hemorrhage andangiogenic defects in the developing hindbrain, as analyzedat high resolution by confocal fluorescent microscopy usingtransgenic animals and transmission electron microscopy. Thesevascular abnormalities were rescued by expression of wild-type(WT) dyrk1aa mRNA, but not a kinase-dead form, indicating anessential role of its kinase activity. Chemical screening using a USFood and Drug Administration (FDA)-approved chemical libraryidentified the calcium chelator ethylene glycol-bis(β-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA) that efficiently rescuedcerebral hemorrhage as well as abnormal cerebrovascular defects indyrk1aa mutants. Another calcium signaling modulator, FK506,rescued the hemorrhagic and cerebrovascular defects of dyrk1aa

mutants in a similar manner as EGTA, and transcriptomic analysesidentified changes in calcium signaling as the main pathwayaffected in dyrk1aamutants. Together, the cerebral hemorrhage andcerebrovascular defects of zebrafish dyrk1aa mutants and thechemical screening revealed an important but less-known role ofDYRK1A in in vivo vascular formation, which involves amechanismthat is mediated by calcium signaling, providing a potentialtherapeutic target for DYRK1A-related vascular disorders.

RESULTSCerebral hemorrhage and a vascular phenotype ofdyrk1aakrb1 mutant embryosWe recently reported the generation of the dyrk1aakrb1 mutants thatdisplayed microcephaly and autistic behavioral phenotypes inadults, whereas no distinct morphological defects were observedduring embryogenesis (Kim et al., 2017). Detailed inspection ofdyrk1aakrb1 homozygous mutant embryos, however, revealed acerebral hemorrhage phenotype as early as 52 h postfertilization(hpf ) (arrows in Fig. 1A). Of the offspring of dyrk1aakrb1 mutants,27.9% of a clutch displayed cerebral hemorrhage, whereas less than9.9% of WT offspring showed spontaneous hemorrhage at 52 hpfusing o-dianisidine staining, which detects hemoglobin activityand allows observation of cerebral hemorrhage more closely(Fig. 1B). Spontaneous hemorrhage of WT offspring at 52 hpfwas detected in less than 5% of the clutch under the brightfieldmicroscope (without o-dianisidine staining, data not shown).Cerebral hemorrhage was detected as patches in the forebrain,midbrain, hindbrain and retina, with some embryos exhibitingmultiple hemorrhages simultaneously (Fig. 1A). To obtain morepronounced hemorrhagic effects and observe a more explicitphenotype, we challenged the mutant embryos with heat stress byincubation at 35°C for 2.5 h from 48 hpf to 50.5 hpf, followedby keeping them at 28.5°C until 52 hpf. The following experimentsfor hemorrhagic phenotype were performed in the heat stresscondition. It has been previously reported that cardiovascular stressaggravates the hemorrhagic phenotype in the brain, presumably byincreasing the heart rate (Barrionuevo and Burggren, 1999; Wanget al., 2014). Consistently, the cerebral hemorrhage of WT anddyrk1aakrb1 mutants under the heat-stressed condition increased by16.3% and 40.5%, respectively (Fig. 1B, Heat-stressed).

Because cerebral hemorrhage is sometimes accompanied bydefective cerebrovascular formation (Arnold et al., 2014), weexamined the formation of central arteries (CtAs) in the hindbrain, awell-characterized stereotypical developing cerebrovascularstructure, using Tg(kdrl:EGFP) transgenic animals (Ulrich et al.,2011) in the WT or dyrk1aakrb1 mutants. CtAs sprouted fromthe primordial hindbrain channels, which contained the pool ofendothelial cells required for CtA formation, and invaded thehindbrain between 32 and 36 hpf with a stereotypical morphologywithin rhombomeres, formed with over 50% ipsilateral CtAconnectivity at 48 hpf (Bussmann et al., 2011; Ulrich et al.,2011). A detailed, high-resolution confocal imaging analysis tocompare the CtA formation of dyrk1aakrb1 mutants and WT at52 hpf revealed that the stereotypical structure of CtAs wasabolished in dyrk1aakrb1 mutant embryos (Fig. 1C). To quantitatethe CtA vascular defects, the length and branching points of CtAswere measured, representing the migration/proliferation andsprouting activities of CtA endothelial cells, respectively(AlMalki et al., 2014; Phng et al., 2009; Tímár et al., 2001)(Fig. 1D). In dyrk1aakrb1 mutants, the total lengths and branchingpoints of CtAs were reduced to 70.2% and 73.3% compared to theWT control, respectively (Fig. 1D). In the following experiments,

2

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

we used the length and branching points of CtAs as quantitativemeasures of defects in vascular development.CtA angiogenic defects were also confirmed by examining RNA

expression of vascular markers kdrl (vegfr2) and dll4 by whole-mount RNA in situ hybridization (WISH) at various developmentalstages including 30, 36 and 52 hpf (Fig. S1, kdrl and dll4).Consistent with the angiogenic defects revealed by Tg(kdrl:EGFP)transgenic animals, expression of kdrl and dll4 in the vasculature ofthe hindbrain was reduced in dyrk1aakrb1 mutant embryos at allstages examined (red arrows in Fig. S1, kdrl and dll4). Thesevascular defects in mutants appeared not to be due to gross defectsof brain development because the expression of krox20, whichmarks rhombomere boundaries (Moens and Prince, 2002; Oxtobyand Jowett, 1993), and isl1, which labels primary motoneurons inthe hindbrain (Ericson et al., 1992; Inoue et al., 1994; Korzh et al.,1993), were grossly unaffected in mutant embryos (Fig. S1, krox20and isl1). These vascular defects appeared not to be due to defectiveheart development or reduced blood flow, based on the normalheartrate of mutants compared to the WT embryos (Fig. S2,Movies 1 and 2).

Zebrafish dyrk1aa and dyrk1ab, mammalian DYRK1Aorthologs, are expressed in endothelial cells and thenervoussystem during developmentWe examined transcripts of dyrk1aa in zebrafish embryos usingWISH during embryogenesis. dyrk1aa mRNA had a broadexpression in the forebrain (black arrowheads/brackets), midbrain(gray arrowheads/brackets), hindbrain (blue arrowheads/brackets),

spinal cord (orange arrowheads), heart (asterisks) and retina (redarrows) at 24, 48 and 72 hpf (Fig. 2Aa-Al). Transverse sections ofWISH embryos at the midbrain level showed that dyrk1aa wasbroadly expressed in the tectum (green arrows), tegmentum (blackarrow), the ganglionic cell layer (purple arrows) and the innernuclear layer of the retina (red arrows) (Fig. 2Am,An). We alsochecked the expression patterns of dyrk1abmRNA (ZFIN ID: ZDB-GENE-030131-5677), another zebrafish ortholog of humanDYRK1A, using WISH, which appeared to be significantlyoverlapped with those of dyrk1aa (Fig. S3).

The temporal expression patterns of dyrk1aa and dyrk1abmRNAs according to developmental stages by reversetranscription-PCR (RT-PCR) were also examined. Transcripts ofdyrk1aa and dyrk1ab mRNA were strongly detectable at the one-cell stage but decreased after 6 hpf, presumably because of thematernal effect (Harvey et al., 2013; Mathavan et al., 2005)(Fig. 2B). Zygotic dyrk1aa and dyrk1ab expression appeared to startat ∼15 hpf, increasing at 24 hpf, and strong expression wasmaintained until 5 days postfertilization (dpf ) (Fig. 2B).

Furthermore, whether dyrk1aa and dyrk1ab mRNA wasspecifically expressed in endothelial cells was confirmed byperforming RT-PCR analyses using GFP-positive Tg(kdrl:EGFP)endothelial cells isolated by fluorescence-activated cell sorting(FACS). The dyrk1aa and dyrk1ab mRNA was enriched in GFP-positive endothelial cells but was also expressed in GFP-negativecells, suggesting a role in endothelial as well as non-endothelialcells (Fig. 2C). These expression patterns were consistent with thereported mammalian Dyrk1a expression in the heart primordium

Fig. 1. dyrk1aakrb1 mutant embryos show cerebral hemorrhagic phenotype and abnormal development of CtAs in the brain. (A) Cerebral hemorrhagewas observed in dyrk1aakrb1 mutant embryos (dyrk1aakrb1) at 52 hpf (Ac-Af, arrows) compared to WT (Aa, Ab). Aa,Ac,Ae show the lateral view; Ab,Ad,Afshow the dorsal view. (B) Embryonic cerebral hemorrhage of WT occurred spontaneously in 9.9% of embryos, whereas dyrk1aakrb1 embryos showed cerebralhemorrhage of 27.9% penetrance (Normal), using o-dianisidine staining. The cerebral hemorrhage of WT and dyrk1aakrb1 mutants increased up to 16.3%and 40.5%, respectively, by inducing heat stress with 2.5 h incubation at 35°C from 48-50.5 hpf (Heat-stressed). The mean percentages for each genotype werepresented from four independent experiments with approximately 20 embryos for each repeat. (C) Confocal fluorescent images at 52 hpf showing thedevelopment of CtAs in WT and dyrk1aakrb1 mutant embryos in the Tg(kdrl:EGFP) background. (D) The lengths and branching points of CtAs in dyrk1aakrb1

mutants were reduced down to 70.2% and 73.3%, respectively, compared to the WT embryos as 100% at 52 hpf. *P<0.05, ***P<0.005 (Mann–Whitney U test).Data are mean±s.e.m. fb, forebrain; mb, midbrain; hb, hindbrain; e, eye; y, yolk. Scale bar: 250 µm in A; 50 µm in C.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

and the central nervous system of developing mouse embryos,including the inner (neural) layer of the optic cup (Hämmerle et al.,2008; Rahmani et al., 1998), and endothelial cells isolated fromDscr1 transgenic mice, as assessed using western blotting (Baeket al., 2009), suggesting a functional conservation across species.Recently, the expression of DYRK1A at the protein and RNA levelin human endothelial cells and in mouse lung endothelial cells hasalso been shown by Rozen et al. (2018).

Rescuing dyrk1aakrb1 mutant phenotypes by dyrk1aaexpressionTo confirm that loss of function of the dyrk1aa genewas responsiblefor dyrk1aakrb1 mutant phenotypes, we tested whether WT dyrk1aamRNA rescued the cerebral hemorrhage and aberrant vascularphenotype in dyrk1aakrb1 mutants by globally expressing full-length WT dyrk1aa mRNA. Using o-dianisidine staining, the highincidence of cerebral hemorrhage in dyrk1aakrb1mutants under heatstress (41.7% of the offspring) was shown to be significantlyreduced down to 25.4% by 0.1 ng dyrk1aa mRNA injection(Fig. 3A,B), whereas the same dose injected into the WTbackground had little effect (Fig. 3A,C). Similarly, the reducedmean percentages of lengths (66.5%) and branching points (61.2%)of CtAs in dyrk1aakrb1mutants relative toWT controls (100%) wererescued (89.9% and 122%, respectively) with the expression of0.1 ng dyrk1aamRNA (Fig. 3D,E). The rescue of CtA angiogenesisdefects were effective only within specific dose ranges of dyrk1aamRNA expression in dyrk1aakrb1 mutants, and overexpression of

dyrk1aa mRNA in WT background increased CtA formation in adose-dependent manner (Fig. 3E), suggesting the importance ofspatial and temporal contexts for effects of dyrk1aa expression aswell as its dose-sensitive nature.

DYRK1A regulates its target substrate proteins viaphosphorylation (Hämmerle et al., 2003). To verify whether thecerebral hemorrhage and CtA defects in dyrk1aakrb1 mutantswere dependent on the kinase activity of Dyrk1aa, we performedrescue experiments with K193R-dyrk1aa mRNA, the predictedkinase-dead form ofDyrk1aa (Himpel et al., 2001). The expression ofK193R-dyrk1aa mRNA or control mCherryRed mRNA failed torescue the defects of hemorrhage and CtA formationwith comparabledoses of WT-dyrk1aa mRNA (Fig. 3B; Fig. S4A and S4B). Theseresults suggest that the kinase activity of Dyrk1aa is critical fornormal CtA development and prevention of hemorrhage.

Ultrastructural analyses of cerebral vessels in dyrk1aakrb1

mutants by transmission electron microscopyTo examine whether the dyrk1aamutation caused an ultrastructuralchange in brain vessels, we analyzed the cytoarchitecture of bloodvessels in the WT and dyrk1aakrb1 embryos at 52 hpf usingtransmission electron microscopy. In the WT group, blood vesselscomposed of endothelial cells and lumens were well formed andtightly arranged at this stage (Fig. 4A,A′), although smooth musclecells, pericytes and astrocytes were in the process of differentiationand not yet clearly identified (Liu et al., 2007). Characteristically,brain tissues in dyrk1aakrb1 mutants exhibited enlarged interstitial

Fig. 2. Zebrafish dyrk1aa and dyrk1ab are expressed in the developing brain during embryogenesis and detected in endothelial cells of thevasculature. (A) WISH showed that dyrk1aa was expressed in the forebrain (black arrowheads, Aa-Af; black brackets, Ag-Al), the midbrain (gray arrowheads,Aa-Af; gray brackets, Ag-Al), the hindbrain (blue arrowheads, Aa-Af; blue brackets, Ag-Al) at 24, 48 and 72 hpf, and the spinal cord (orange arrowheads,Aa and Ab) at 24 hpf. It was also detected in the heart (black asterisks, Ad,Af,Aj,Al) and in the retina (red arrows, Ai-Al) at 48 and 72 hpf. Am and An are sectionedimages of WISH embryos showing the expression of dyrk1aa in the tectum (green arrows), tegmentum (black arrow), the ganglionic cell layer (purplearrows) and the inner nuclear layer of the retina (red arrows) at 48 hpf and 72 hpf. See Fig. S3 for WISH of dyrk1ab. (B) Semi-quantitative RT-PCR analysis ofdyrk1aa and dyrk1ab mRNA expression in whole embryos at indicated developmental stages. dyrk1aa and dyrk1ab expression was observed at theone-cell stage followed by diminishing at 6 hpf and resuming after 24 hpf. (C) Semi-quantitative RT-PCR analysis using endothelial cells isolated by FACS forsorting GFP-positive and -negative cells of Tg(kdrl:EGFP) embryos at 48 hpf revealed that dyrk1aa and dyrk1ab mRNAs were expressed in GFP-positiveendothelial cells as well as in GFP-negative embryonic cells. Scale bars: 200 µm in Aa-Al; 50 µm in Am,An.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

spaces, presumably because of loose connections between layers ofvessel walls (arrowheads in Fig. 4B,B′), suggesting that thisabnormal formation of vessel walls was one of the causes of thehemorrhagic phenotype in dyrk1aakrb1 mutant embryos.

DYRK1A inhibition by harmine and chemical screeningCerebral angiogenic defects and hemorrhage in dyrk1aakrb1

mutants were also recapitulated by harmine (7-methoxy-1-methyl-9H-pyrido[3,4-b]-indole), a well-known DYRK1A inhibitor (Bainet al., 2007; Göckler et al., 2009). WT zebrafish embryos exposed todifferent concentrations of harmine (10, 25 and 50 µM) starting at24 hpf for 28 h showed the hemorrhage and CtA formation defectsat 52 hpf, similar to those of the dyrk1aakrb1 mutants, and thenumber of CtA sprouts was reduced down to 15.2% of the controlnumber at a concentration of 50 µM harmine (Fig. 5A,B). Theharmine-induced hemorrhagic phenotype increased up to 65.7%that of the offspring, compared to 2.3% of the DMSO-treatedcontrols (Fig. 5C). This chemical suppression of DYRK1A usingharmine further suggested that the vascular phenotypes ofdyrk1aakrb1 mutants were owing to the loss of DYRK1A function.

Based on the hemorrhagic phenotype induced by harminetreatment, we developed an embryonic screening assay using achemical library that consisted of 1280 FDA-approved andpharmacologically active compounds (LOPAC 1280; Sigma-Aldrich), which allowed us to identify small molecules thatmodulated the harmine-induced hemorrhagic phenotype. Five WTzebrafish embryos at 24 hpf were placed in each well of a 48-wellplate, exposed to 30 µM harmine together with the individualchemicals of the chemical library at 10 µM as a final concentration

Fig. 3. The cerebral hemorrhage and CtA angiogenic defects indyrk1aakrb1 mutants can be rescued by dyrk1aa expression, with thekinase activity of Dyrk1aa required for its phenotypic rescues. (A) Theo-dianisidine staining images at 52 hpf showed that the cerebral hemorrhage(arrows) of dyrk1aakrb1 embryos was rescued by WT dyrk1aa mRNAexpression, but not in no-injection control, K193R-dyrk1aa or mCherryRedmRNA. (B) The frequency of cerebral hemorrhage of dyrk1aakrb1 was reducedfrom 41.7% to 25.4% by injecting WT dyrk1aa mRNA of 0.1 ng, but notsignificantly changed by injecting K193R-dyrk1aa or mCherryRed mRNAat 52 hpf. (C) Injection of mRNAs of WT dyrk1aa, K193R-dyrk1aa ormCherryRed control did not affect the cerebral hemorrhage in WT embryos at52 hpf. The mean percentages for each genotype were presented from threeindependent experiments with approximately 20 embryos in each repeat.(D) The compiled images of CtAs of Tg(kdrl:EGFP) at 52 hpf by confocalmicroscopy showed that the angiogenic defects of the CtAs of dyrk1aakrb1

embryos were rescued by WT dyrk1aa mRNA injection. (E) The meanpercentages of length and branching points of CtAs in dyrk1aakrb1 mutantswere rescued from 66.5% to 83.4% and from 61.2% to 108.3%, respectively,with 0.05 ng of dyrk1aamRNA injection, relative to WT as 100%, and rescuedto 89.9% and 122.0%, respectively, with 0.1 ng of dyrk1aa mRNA injection.Expression with 0.2 ng of dyrk1aa mRNA did not exhibit the rescue effect.Expression of dyrk1aa mRNA with the same doses in WT embryos alsoincreased the length and branching points of CtAs in a dose-dependentmanner (113.8% and 139.6% with 0.2 ng of dyrk1aa mRNA, respectively).*P<0.05, **P<0.01, ***P<0.005 (one-way ANOVA). n.s., not significant.Data are mean±s.e.m. Scale bar: 100 µm in A; 50 µm in D.

Fig. 4. The transmission electron microscopy revealed that dyrk1aakrb1

embryos had abnormal vessel walls in the brain at 52 hpf. (A,B) Brainvessels in WT and dyrk1aakrb1. A′ and B′ show enlarged images of the boxedareas in A and B, respectively. (A′,B′) Arrows in A′ indicate the compactstructure of vessel walls, whereas arrowheads in B′ designate the looseconnection of vessel walls. Dashed lines demarcate the border between thevessel walls and the lumen. EC, endothelial cell; LM, vessel lumen; RBC, redblood cell; VW (dotted arrows), vessel wall. Scale bars: 1 µm.

5

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

for 28 h, and analyzed for the increased or decreased hemorrhagicphenotype (Fig. 5D). As a result, 171 of 1280 compounds testedwere found to be phenotype modifiers, which were categorizedaccording to their common features as ‘class’ based on their knownfunctions (Tables S1-S3). Some chemicals have already beenreported to cause hemorrhage. For example, atorvastatin, whichincreased the hemorrhage in our screening, has been previouslyreported to induce hemorrhagic stroke as a side effect in zebrafishembryos (Gjini et al., 2011). Of the 171 compounds, EGTA, aspecific calcium chelator, was identified as one of the most efficientsuppressors of the hemorrhagic phenotype induced by harminetreatment, in our chemical screening.

EGTA effectively suppressed the vascular defects ofdyrk1aakrb1 mutantsTo determine whether our findings from the chemical screeningwere applicable to the genetic model, we added EGTA to thedyrk1aakrb1 mutants and examined the hemorrhagic and CtAdevelopment. Under heat-stressed conditions, the EGTA treatmentsignificantly reduced the hemorrhagic phenotype of dyrk1aakrb1

mutants at a specific concentration of 10 nM (from 37.5% to 23.6%,Fig. 6A,B; P<0.05). In addition, a reduced CtA development ofmutants in lengths and branching points (76% and 57.4% reductioncompared to WT, respectively) was also rescued up to 86.6% and85.9% of the normal levels, respectively, by treatment with the sameconcentrations of EGTA (Fig. 6C,D; P<0.01). EGTA treatmentappeared to be effective only within a narrow range, because 1 nMor 100 nM EGTA treatment failed to rescue the vascular defects ofdyrk1aakrb1 mutants, except for the rescue of CtA branching pointswith 1 nM EGTA (Fig. 6B,D). In contrast, EGTA treatment of theWT embryos did not induce any vascular defects (Fig. S5),suggesting a specific role of EGTA on dyrk1aakrb1 mutants.To identify the temporal requirement of EGTA for the

suppressive effects, developing embryos were incubated withvarious doses of EGTA during an early period of 8 h

(24∼32 hpf), followed by washing, or during a late period of 16 h(32∼48 hpf) (Fig. 6E). Interestingly, only the early treatment with10 nM EGTA was effective in significantly preventing the cerebralhemorrhage (from 37.7% to 26.3%), whereas the early treatmentwith other doses and the late treatment was not (Fig. 6F,G). Becausesprouting and elongation of cerebral vessels for angiogenesis occursactively during day 1 postfertilization (Fujita et al., 2011; Isogaiet al., 2001), EGTAmay exert its suppressive effects on the vasculardefects of dyrk1aakrb1 mutants by regulating early angiogenicprocesses.

Transcriptomic analyses of dyrk1aakrb1 mutantsIdentification of EGTA as a suppressor of vascular defects indyrk1aakrb1 mutants by chemical screening implies that calciumsignaling may be compromised in the dyrk1aakrb1 mutants. Tocorroborate this finding, we examined transcriptomic changes ofdyrk1aakrb1 mutants compared to WT embryos at 48 hpf usingRNA-seq analysis (Fig. 7). This analysis identified 222 transcriptsas differentially regulated genes (DEG), of which 101 wereupregulated and 121 were downregulated in the dyrk1aakrb1

mutants (more than 2-fold and less than 0.5-fold, respectively;P<0.05). When DEGs were analyzed for enriched biological geneontology (GO) categories by the functional annotation tools in theDatabase for Annotation, Visualization and Integrated Discovery(DAVID; https://david.ncifcrf.gov), the calcium ion bindingcategory was the most enriched GO molecular function (MF)(Fig. 7B). This GO category includes genes encoding severalcalcium-dependent adhesion proteins, protocadherin (Pcdh) familymembers ( pcdh1g1, pcdh2ab10, pcdh1gc6, pcdh2g17, pcdh1g18and pcdh1g30) and calcium-dependent calpains (capn8 and capn2l)(Khorchid and Ikura, 2002) (Fig. 7A). Other genes encodingmyosin light chain 4 (myl4), low-density lipoprotein receptor b(ldlrb), and mannan-binding lectin serine protease 2 (masp2) in thisGO term are also regulated by calcium signaling, directly orindirectly (Kang et al., 1999; Orr et al., 2016; Zhao and Michaely,

Fig. 5. Inhibition of DYRK1A by harmine inducesbrain hemorrhage and vasculature defects in thehindbrain. (A) WT Tg(kdrl:EGFP) embryos weretreated with increasing concentrations of DYRK1Ainhibitor harmine, from 24 hpf until 52 hpf: Aa, DMSO;Ab, 10 µM harmine; Ac, 25 µM harmine; Ad, 50 µMharmine. Vascular patterning defects of CtAs areshown (red brackets) by confocal imaging at 52 hpf(lateral view). (B) Quantification of the effect onnumber of developing CtAs by harmine treatment.The numbers of CtAs were dramatically reducedby treating harmine in a dose-dependent manner.n=11 each group. (C) Quantification of the brainhemorrhage penetrance. The cerebral hemorrhagicphenotype of 2.3% in DMSO-treated embryos wasincreased from 14.8% to 65.7% by harmine treatmentfrom 10 µM to 50 µM. The mean percentage for eachtreatment was shown from three independentexperiments with approximately 40 embryos in eachrepeat. (D) Schematic showing the strategy of thein vivo chemical library screening to identify smallmolecule modifiers for cerebral hemorrhagicphenotype upon DYRK1A inhibition. Data aremean±s.e.m. Scale bar: 100 µm.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

2009) (Fig. 7A), although their roles in regulating vascularformation are not understood at present. In addition, the‘oxidation-reduction process’, the top-ranked GO biological

process (BP), is also well known to affect calcium signalingnetworks (Görlach et al., 2015; Lounsbury et al., 2000) (Fig. 7C),and the genes belonging to the ‘homophilic cell adhesion via plasma

Fig. 6. EGTA identified by the chemical library screening effectively rescues the cerebral hemorrhage and abnormal CtA development of dyrk1aakrb1

mutants. (A) o-dianisidine staining images at 52 hpf showing that the cerebral hemorrhage (arrow) of dyrk1aakrb1 embryos were rescued by treating with10 nM EGTA. WTembryos were not affected by the treatment at the same concentration. (B) Quantitation of the rescue of the cerebral hemorrhagic phenotype byEGTA. The cerebral hemorrhage in dyrk1aakrb1 embryos is reduced from 37.5% to 23.6% by 10 nM EGTA treatment. The data is presented with five independentexperiments with approximately 20 embryos in each repeat. (C) Compiled confocal microscopy images of the rescue effect on angiogenic defects of CtAs indyrk1aakrb1 embryos by EGTA treatment. (D) Quantitative data showing that the reduced mean percentages of length and branching points of CtAs in dyrk1aakrb1

embryos at 52 hpf (76% and 57.4%, respectively, compared to WT embryos as 100%) were increased up to 86.6% and 85.9%, respectively, by 10 nM EGTAtreatment. n≥11 each group. (E) Schematic depicts the treatment scheme of EGTA at developmental stages from 24 to 52 hpf in dyrk1aakrb1 embryos. (F) Thecerebral hemorrhage of dyrk1aakrb1 embryos at 52 hpf (arrows) were rescued by 10 nM EGTA treatment for 24-32 hpf but not by the treatment for 32-48 hpf.(G) Quantitative data showing that cerebral hemorrhage of dyrk1aakrb1 embryos reduced from 37.7% to 26.3%by treatingwith 10 nMEGTA for 24-32 hpf. *P<0.05,**P<0.01, ***P<0.005 (one-way ANOVA). n.s., not significant. Data are mean±s.e.m. Scale bars: 100 µm in A,F; 50 µm in C.

7

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

membrane adhesion molecules’ BP category consist of the six Pcdhfamily genes ‘calcium ion binding’ MF category (Fig. 7C).We checked the validity of DEG identification by RNA-seq

analysis with RT-PCR using RNAs from thewhole embryos or fromFACS-isolated endothelial cells at 48 hpf (Fig. S6). As a result, nineout of 15 candidate DEGs ( pcdh1g1, pcdh2ab10, pcdh1gc6, myl4,pcdh1g18, f7i, capn8, pcdh1g30 and capn2l) were clearly changedin the whole dyrk1aakrb1 mutant embryos (Fig. S6A). Those nineDEGs were further tested for expression changes in endothelial cellsby performing another RT-PCR using RNAs prepared from FACS-isolated endothelial cells of 48 hpf Tg(kdrl:EGFP) WT and

dyrk1aakrb1 mutants. Seven DEGs ( pcdh1g1, pcdh2ab10, myl4,pcdh1g18, f7i, pcdh1g30 and capn2l) were found to be correlativewith the similar patterns as in the whole embryos (Fig. S6B),suggesting their potential roles in dyrk1aa-mediated calciumsignaling in endothelial cells.

In order to check whether calcium signaling changes occur evenearlier than 48 hpf, especially when rescue by EGTA treatment iseffective (Fig. 6), we also investigated the transcriptomic changes indyrk1aakrb1 mutants at 32 hpf using RNA-seq analysis (Fig. S7).Similar to the findings of 48 hpf transcriptomic analysis, DEGs at32 hpf were already highly enriched in the GO categoriesof ‘calcium ion binding’ and ‘oxidoreductase activity’ in MF(Fig. S7A,B). Furthermore, expression of six genes in the calciumion binding category ( pcdh1g1, pcdh1g18, f7i, capn8, pcdh1g30and capn2l) turned out to be concurrently changed in both 32 hpfand 48 hpf transcriptomes (Fig. S7C). Interestingly, five genes( pcdh1g1, pcdh1g18, f7i, pcdh1g30 and capn2l) out of these sixDEGs were also changed in the FACS-isolated endothelial cells(Fig. S6B), potentially highlighting their indispensability indyrk1aa-mediated calcium signaling.

Collectively, alterations of genes in ‘calcium ion binding’ andother GO categories may reflect dysregulation of calciumhomeostasis in dyrk1aakrb1 mutants, consistent with the rescueeffects by EGTA on vascular defects of mutants.

Modulation of intracellular calcium signaling also rescuedvascular defects of dyrk1aakrb1 mutantsAs noted previously, EGTA is a chelating agent that has strongselectivity for calcium ions, primarily by lowering the level ofextracellular calcium, eventually affecting the calcium homeostasisinside a cell, which is an essential process for regulatingangiogenesis as well as other cellular processes including musclecontraction and neurogenesis (Berridge et al., 2000; Munaron,2006). To assess whether the modulation of intracellular calciumsignaling can recapitulate the effects of EGTA treatment, weattempted to rescue the cerebral hemorrhage and defective CtAformation of dyrk1aakrb1 mutants by inhibiting calcineurin proteinwith FK506, a well-known specific calcineurin inhibitor (Li et al.,2011). The signaling pathway mediated by calcineurin, a serine/threonine protein phosphatase, is one of the major signalingpathways that is regulated by calcium (Crabtree, 2001; Hogan et al.,2003). Interestingly, the hemorrhagic phenotype of dyrk1aakrb1wasrescued by treatment with 50 and 100 ng/ml FK506 in a dose-dependent manner, whereas no significant change was observed inthe WT control when treated (Fig. 8A,B). The same concentrationsof FK506 also rescued defects in the CtA branching points (100 ng/ml) and length (50 ng/ml) of mutants, although FK506 alsoincreased those of CtAs in the WT control (Fig. 8C,D). These datawere consistent with the notion that the dysregulation of calciumhomeostasis was responsible for the vascular defects of thedyrk1aakrb1 mutants, and such defects could be rescued, at leastpartially, by manipulating a calcium-dependent signaling pathway.

Modulation of calcium signaling by EGTA or FK506, however,did not reverse the transcriptional changes of DEGs found intranscriptomic analysis of dyrk1aakrb1 mutants (Fig. S8), implyingthat they may not be appropriate readouts for the phenotypicreversal. In addition, expression of potential zebrafish homologs ofRCAN1.4, a known target gene in Dyrk1a/Calcineurin pathway(Minami et al., 2004; Rozen et al., 2018), was not changed indyrk1aakrb1 mutants (Fig. S9). These additional data suggest aninvolvement of more complex mechanisms in vivo with regard tocalcium signaling regulated by dyrk1aa.

Fig. 7. Whole transcriptomic analysis of dyrk1aakrb1 compared withWT using RNA-seq. (A) Volcano plot showing the comparison of the wholetranscriptomes of the pools of ∼20 WT and dyrk1aakrb1 embryos, set at48 hpf (two biological replicates for each group). The colored dots show thedifferentially upregulated (red) or downregulated (blue) genes in dyrk1aakrb1

embryos compared to WT embryos (more than 2-fold, P<0.05). The coloreddots with black circles indicate the upregulated (red dot with black circle) anddownregulated (blue dot with black circle) genes in the group of ‘calcium ionbinding’ (see text for details). (B) Bar graph showing the list of the groups ofDEGs in the classification of the molecular functions. DEGs of the ‘calcium ionbinding’ are most abundant. (C) Bar graph showing the list of the groups ofDEGs in the classification of the biological process. DEGs of the ‘oxidation-reduction process’ are most abundant. Bar graph x-axis (B,C) represents thenumber of DEG counts in respective annotations.

8

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

DISCUSSIONProper cerebrovascular development and maintenance are essentialprocesses for normal brain development and function, and aregoverned by diverse and coordinated signaling pathways (Hogan andSchulte-Merker, 2017). In this report, we established a less-knownfunction of DYRK1A, one of the critical genes contributing to someDS phenotypes, in regulating angiogenesis and preventing

hemorrhage in the brain, using zebrafish dyrk1aa knockoutmutants. The in vivo chemical screening using zebrafish embryosidentified small molecules that were able to modify the cerebralhemorrhage upon DYRK1A inhibition. Among them, EGTA, aknown specific calcium chelator, was identified as one of the mosteffective small molecules that rescued the vascular defects of dyrk1aamutants. Changes in calcium-related signaling pathways revealed byRNA-seq analyses and the rescuing activity by chemical inhibition ofcalcineurin, a major component of the calcium-dependent pathway,on the vascular defects corroborated the notion that the vasculardefects of dyrk1aa mutants were primarily due to calciumdysregulation, which could be reversed by inhibition of excessivecalcium-dependent processes.

Zebrafish dyrk1aa knockoutmutants exhibited compromised vesselintegrity, with cerebral hemorrhage and angiogenesis simultaneously(Figs 1 and 4), compared to WT controls. Although the relationshipbetween angiogenesis, vessel permeability and cerebral hemorrhageduring developmental processes is not clearly defined, the interactionsof these processes may be important for development of the functionalcerebral vasculature. For example, a targeted deletion of miR-126 inmice showed that the cerebral hemorrhage due to defective vascularintegrity accompanied a severe reduction in cranial vessel formation(Wang et al., 2008). In addition, excessive angiogenesis was shown toprecede the cerebral hemorrhage in the developing brain ofneuroendothelium-specific Itgb8 knockout mice (Arnold et al.,2014). Based on the mutant phenotypes described in our report, it ispossible that DYRK1A may be a novel dual-function player thatregulates both cerebral angiogenesis and the maintenance of vascularintegrity simultaneously to prevent hemorrhage in the developingbrain. Consistent with this idea, it has been reported that DS fetusesdisplayed several developmental vascular defects revealed byultrasound scanning (Chaoui, 2005), whereas DS adults suffered asignificantly higher risk of hemorrhagic stroke, as shown in a largecohort study (Sobey et al., 2015), suggesting potential dual roles ofDYRK1A in these vascular disorders.

DYRK1A function in angiogenesis and hemorrhage preventionis likely to be a calcium-dependent process, based on the comparativetranscriptome analyses of the WT and dyrk1aamutants (Fig. 7). TheDEG analysis identified ‘calcium ion binding’ as the top-ranked GOcategory, which may reflect dysregulation of calcium homeostasis indyrk1aa mutants. Of interest, several genes in the ‘calcium ionbinding’ category have been associated with endothelial cellpermeability and vascular dysfunction. For example, protocadheringenes such as the Pcdh-gamma cluster, one of the prominent groupsof DEGs from our analysis, were recently reported to be highlyexpressed in endothelial cells of the brain microvasculature and maycontribute to junctional stability of the blood-brain barrier (Dillinget al., 2017). Altered DEGs in the ‘oxidoreductase activity’ MFcategory and ‘oxidation-reduction process’ BP category in dyrk1aamutants are also consistent with the disruption of Ca2+ signaling,because redox homeostasis is one of the major factors in regulatingintracellular Ca2+ signaling events as well as vascular development(Lounsbury et al., 2000).

The calcium signal-related function of DYRK1A incerebrovascular formation and maintenance is also supportedby results from the in vivo chemical screening and calcineurininhibition studies. The specific calcium chelator EGTA wasidentified as one of the most potent small molecules that rescuedthe cerebral hemorrhage elicited by DYRK1A inhibition (Fig. 6).The Ca2+ ion level inside the cytosol is normally maintained at a lowconcentration (approximately 100 nM), compared to a significantlyhigher (more than 20,000-fold) concentration in the extracellular

Fig. 8. FK506 rescues the cerebral hemorrhage and CtA angiogenicdefects in dyrk1aakrb1 mutants. (A) The cerebral hemorrhage of dyrk1aakrb1

mutant embryos (arrows) was rescued by the treatment of 50 ng/ml FK506.(B) The cerebral hemorrhage in dyrk1aakrb1mutant embryos at 52 hpf (40.5%)was reduced to 17.8% and 13.4% by the treatment of 50 ng/ml and 100 ng/mlFK506, respectively. No differences were seen in WT with the sametreatments. (C) The compiled confocal microscopy images of CtAs in theTg(kdrl:EGFP) background showed that CtA defects in dyrk1aakrb1 embryoswere rescued by FK506 treatment in a dose-dependent manner, and CtAangiogenesis in WT was also affected. (D) The reduced mean percentageof CtA length of dyrk1aakrb1 mutants (75.5%) was significantly rescued up to84.6% by 50 ng/ml FK506, whereas that of mutant branching points (74.3%)was significantly rescued up to 96.4% by 100 ng/ml FK506 at 52 hpf,compared to WT embryos as 100%. 50 ng/ml FK506 treatment increasedthe WT CtA length up to 110.3%, whereas 100 ng/ml FK506 treatmentincreased both length and branching points of WT CtAs (109.5% and 133.8%,respectively). *P<0.05, **P<0.01, ***P<0.005 (one-way ANOVA). n.s., notsignificant. Data are mean±s.e.m. Scale bars: 100 µm in A; 50 µm in C.

9

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

environment (Clapham, 2007), illustrating the importance of thetight regulation of calcium homeostasis across the membrane inmaintaining normal cellular functions. As EGTA is an extracellularcalcium chelator, it may affect the calcium function eitherextracellularly by directly attenuating the activities of calcium-dependent extracellular vascular ligands and cell adhesionmolecules (Deli, 2009; Tomita et al., 1996), or intracellularly bylowering the amount of calcium entry into the cells. Our datasuggested that both scenarios would be possible based on the factsthat cerebral vessels in the mutants were disrupted at theultrastructural level, presumably because of the altered calcium-dependent vascular integrity (Fig. 4), and rescue of the mutants’vascular phenotypes were mimicked by inhibiting a calcium-dependent intracellular signaling pathway using FK506, a specificcalcineurin inhibitor (Fig. 8). Consistent with these findings,dantrolene, a RyR antagonist that decreases the intracellularcalcium level (Fruen et al., 1997), was also identified as amodestly effective small molecule in our chemical screening,although it did not reach statistical significance in our detailedanalyses (Table S3 and data not shown).An underlying mechanism by which DYRK1A regulates the

calcium signaling in vascular formation is not yet clear. DYRK1Amay affect Ca2+ flux directly, as in the case of GluN2A-containingN-methyl-D-aspartate glutamate receptor phosphorylation byDYRK1A, leading to the elevation of their density on themembrane, and eventually to neurological dysfunction (Grauet al., 2014). Alternatively, DYRK1A may indirectly influenceCa2+ signaling by phosphorylating mediators that determine theexpression or activity of calcium-dependent effectors, similar tomyocardial pathology, in which phosphorylation of alternativesplicing factor by DYRK1A increases the expression of Ca2+/calmodulin-dependent protein kinase II δ (He et al., 2015). BecauseDYRK1A is likely to be functional both in the nucleus and in thecytoplasm, based on its ubiquitous localization at the cellular level(Marti et al., 2003), it may directly or indirectly regulate membrane-bound ion channels near the cell membrane (similar to GluN2Aphosphorylation), intracellular signaling molecules in thecytoplasm or transcription factors in the nucleus, eventuallymaintaining the Ca2+ homeostasis of cells.Together, our characterization of zebrafish dyrk1aa knockout

mutants showed that DYRK1A is implicated in cerebral angiogenicactivity and the maintenance of vascular integrity duringdevelopment. The combination of detailed transcriptomic analysesand chemical screening results strongly suggested that the calcium-dependent signaling regulated by DYRK1A is one of the majorsignaling pathways responsible for such vascular phenotypes.However, detailed signaling molecules and pathways affected indyrk1aa mutants remain to be further studied. Our results alsoillustrate the usefulness of zebrafish dyrk1aa mutants in providingan in vivo animal model to understand the pathophysiology ofhuman vascular diseases related to DYRK1A function and insuggesting potential therapeutic approaches for effective treatments.

MATERIALS AND METHODSMaintaining zebrafish embryosZebrafish (Danio rerio) embryos of AB (WT) strain, transgenic zebrafishTg(kdrl:EGFP) and dyrk1aa knockout mutant zebrafish (dyrk1aakrb1) weremaintained in E3 egg water (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2,0.33 mM MgSO4) in a petri dish at 28.5°C. In order to generate transparentzebrafish embryos for imaging confocal microscopy, performingWISH andexamining hemorrhagic phenotype, embryos were incubated in 1× PTU(0.003% 1-phenyl 2-thiourea, Sigma-Aldrich)-E3 egg water after 6 hpf.Zebrafish husbandry and animal care were performed in accordance with

guidelines from the Korea Research Institute of Bioscience andBiotechnology (KRIBB) and approved by KRIBB-IACUC (approvalnumber: KRIBB-AEC-17126).

o-Dianisidine staining and quantification of the hemorrhagicphenotypeEmbryos at 52 hpf were fixed with 4% paraformaldehyde in 1× phosphate-buffered saline (PBS) for 4 h at room temperature (RT) and washed with 1×PBS containing 0.1% Tween 20 (1× PBST). The embryos were placed in o-dianisidine stain solution [0.6 mg/ml o-dianisidine (Sigma-Aldrich),0.01 M sodium acetate (pH 5.5), 0.65% hydrogen peroxide and 40%ethanol] in the dark for several minutes at RT to detect hemoglobin activity.o-Dianisidine is a peroxidase substrate, and hemoglobin catalyzes the H2O2-mediated oxidation of o-dianisidine. Stained embryos were washed severaltimes with 1× PBST and stored in 70% glycerol for imaging using anOlympus SZX16 microscope equipped with TUCSEN Dhyana 400DCcamera. The cerebral hemorrhagic phenotype was calculated as the meanpercentages by counting the number of embryos with hemorrhagicphenotype in the brain and the retina.

FACS analysis of endothelial cellsTo isolate GFP-positive endothelial cells from Tg(kdrl:EGFP) embryos, weadopted the protocol developed by Manoli and Driever (2012). Briefly,150 embryos at 48 hpf were dechorionated with 1 mg/ml protease (P6911,Sigma-Aldrich) in E3 egg water for 5 min at RT, and washed in0.5× Danieau’s solution [29 mM NaCl, 0.35 mM KCl, 0.2 mMMgSO4•7H2O, 0.3 mM Ca(NO3)2, 2.5 mM HEPES (pH 7.6)]. The yolksof those embryos were removed using the deyolking buffer (55 mM NaCl,1.8 mM KCl, 1.25 mM NaHCO3), followed by washing in 0.5× Danieau’ssolution, and embryonic cells were dissociated using FACS max celldissociation solution (T200100, AMS Biotechnology) and the cellsuspension passed through a 40 µm strainer (93040, SPL). FACS wasperformed at RT under sterile conditions using a FACSAria or FACSAria-Fusion (BD Biosciences).

RNA preparation and RT-PCR analysisZebrafish embryos in each developmental stage or cells isolated by FACSwere harvested with TRI reagent solution (Ambion), followed by purifyingtotal RNA with Direct-zol RNA miniprep kit (Zymo Research) andsynthesizing cDNA with SuperScript III First-Strand Synthesis System(Invitrogen). The synthesized cDNA was amplified by PCR using: theforward primer 5′-TCAGTGATGCTCACCCACAG-3′ and reverse primer5′-CGTCATAGCCGTCGTTGTAA-3′ for dyrk1aa; the forward primer5′-GAAACGACGGCATCAACAGG-3′ and reverse primer 5′-CAGCTC-GGTCGTAGGCTTTT-3′ for dyrk1ab; the forward primer 5′-GGCAAG-CTGACCCTGAAGTT-3′ and reverse primer 5′-TTCTGCTTGTCGG-CCATGAT-3′ for EGFP; the forward primer 5′-CCCTTACCCTGGCTT-ACACA-3′ and reverse primer 5′-TCTTGTTGGTTCCGTTCTCC-3′ forkdrl; and the forward primer 5′-CTGGTTCAAGGGATGGAAGA-3′ andreverse primer 5′-ATGTGAGCAGTGTGGCAATC-3′ for eef1a1l1. TheRT-PCR primers used for supplementary figures are shown in Table S4.

Chemical treatment and small molecule library screeningHarmine (Sigma-Aldrich) was dissolved in DMSO and added to 1× PTU-E3egg water to earn final concentrations of 10-50 µM with 0.1% DMSO, andthen applied to ∼40 embryos (dechorionated) on a 90 mm plate from 24to 52 hpf.

For chemical screening, each of 1280 small molecules in the Library ofPharmacologically Active Compounds (LOPAC1280, Sigma-Aldrich) with10 µM as a final concentration was individually applied into each well of48-well plates containing 1× PTU-E3 egg water with 30 µM harmine andfive embryos from 24 to 52 hpf. As a negative control, 1× PTU-E3 egg watercontaining 0.4% DMSO was used. Upon 30 µM harmine treatment, onaverage three embryos displayed the hemorrhage in the brain regions. Basedon this criterion, the increased hemorrhage was defined by four or fiveembryos showing the hemorrhage in brain regions, whereas the reducedphenotype was defined by 0 to two embryos with such defect.

10

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

For EGTA and FK506 treatment, EGTA at final concentrations of1-100 nM and FK506 at final concentrations of 5-100 ng/ml was appliedinto each well of 6-well plates containing ∼20 dechorionated embryos ofWT and dyrk1aa mutants from 24 to 52 hpf, with 1× PTU-E3 egg watercontaining 0.1% DMSO alone used as a negative control.

Confocal microscopic analysis for cerebrovascular phenotypesof zebrafish embryosTo analyze the brain vasculature phenotypes of Tg(kdrl:EGFP) embryos withhigh resolution, embryos were grown up to 52 hpf and fixed with 1× stainingsolution (4% paraformaldehyde, 4% sucrose, 0.15 mM CaCl2, 1× PBS)overnight at 4°C. Fixed embryos were washed briefly with 1× PBST andembedded on glass-bottomed imaging dishes with 1% low melting pointagarose (Promega). The embryos were imaged using Olympus FV1000confocal microscopy and the CtA development in the hindbrain wasquantified using the length and branching points, by measuring total lengthswith ImageJ and manually counting the junctional sites of the CtAs.

WISH and section of the hybridized embryosWISH in zebrafish embryos was performed as previously reported (Thisseand Thisse, 2008). The DNA templates for zebrafish dyrk1aa, dyrk1ab anddll4 (GeneBank accession numbers BC129212.1, NM_001347831.1, andNM_001079835.1, respectively) were amplified from cDNAofWTembryosat 52 hpf. Prof. Cheol-HeeKim (ChungnamNational University, Republic ofKorea) donated kdrl, krox20 and isl1 DNA. Dig-labeled anti-sense probeswere in vitro transcribed using SP6 or T7 RNA polymerase kits (Roche) andpurified with NucAway spin columns (Invitrogen). Embryos for WISH wereprepared by fixing with 4% paraformaldehyde in 1× PBS, dehydratingusing methanol, stored at−20°C for 30 min, and serially rehydrated using 1×PBST. The rehydrated embryos were treated with proteinase K in 1× PBS andpost-fixed with 4% paraformaldehyde. The antisense probes were hybridizedwith the fixed embryos at each developmental stage in hybridizing solution(5 mg/ml torula yeast RNA type VI, 50 µg/ml heparin, 50% formamide,5× SSC, 0.1% Tween-20, 1 M citric acid used to adjust to pH 6.0) at 70°Covernight. The probes were washed serially using 2× SSCT-F (2× SSCT,50% formamide, 0.1% Tween-20), 2× SSCT (2× SSCT, 0.1% Tween-20),0.2× SSCT (0.2× SSCT, 0.1% Tween-20) at 70°C and 1× PBST at RT. Theembryos were blocked with blocking solution (5% horse serum, 1× PBST) atRT, and the alkaline phosphatase-conjugated anti-digoxigenin antibody(11 093 274 910, Roche) was added into the blocking solution at 4°Covernight. To detect the expression signal of transcripts, NBT/BCIP solution(11 681 451 001, Roche) was used as alkaline phosphatase substrate. Theexpression patterns of transcripts were observed using an Olympus SZX16microscope and imaged with a TUCSEN Dhyana 400DC or Olympus XC10camera. To observe detailed expression patterns, whole-mount RNA in situhybridized embryos were prepared for cryosectioning by embedding in anagar-sucrose solution (1.5% agar, 5% sucrose). After the agar blockscontaining the embryos were kept in 30% sucrose solution, they wereprocessed for transverse cryosectioning using a LEICACM1860 cryostat at athickness of 25-35 µm.

Microinjection of RNAIn order to prepare mRNAs of dyrk1aa, dyrk1aa-K193R and mCherryRedfor rescue experiments, the pCS2+ vectors inserted with each DNA werelinearized, in vitro transcribed using a mMESSAGE mMACHINEkit (Invitrogen) and purified with NucAway spin columns (Invitrogen).One-cell-stage eggs were collected and microinjected with each mRNAconstruct containing 0.05% Phenol Red solution as a visible indicator usinga PV380 Pneumatic picopump (World Precision Instruments).

Transmission electron microscopyTissue samples from embryos of WT and dyrk1aakrb1 at 48 hpf were fixedimmediately with 2% glutaraldehyde and 2% paraformaldehyde in 0.1 Mphosphate buffer (pH 7.4) for 2 h at 4°C. Following three washes in thephosphate buffer, tissues were post-fixed with 1% osmium tetroxide on icefor 2 h andwashed three times in the phosphate buffer. The tissues were thenembedded in pure Epon 812 mixture after dehydration in ethanol seriesand followed by infiltration in propylene oxide:epon mixture series.

Polymerization was conducted with pure resin at 70°C for 24 h. Ultrathinsections (∼70 nm) were obtained with a model MT-X ultramicrotome(RMC Boeckeler) and then collected on 100 mesh copper grids. Afterstaining with 2% uranyl acetate (7 min) and lead citrate (2 min), the sectionswere visualized using the Bio-HVEM system (JEM-1400Plus at 120 kV andJEM-1000BEF at 1000 kV, JEOL).

Isolation, library preparation and sequencing for RNA-seqTotal RNAwas isolated using Trizol reagent (Invitrogen). RNA quality wasassessed by an Agilent 2100 bioanalyzer using the RNA 6000 Nano Chip(Agilent Technologies), and RNA quantification was performed using anND-2000 Spectrophotometer (Thermo Fisher Scientific). For control andtest RNAs, the construction of the library was performed using the SENSEmRNA-Seq Library Prep Kit (Lexogen) according to the manufacturer’sinstructions. Briefly, 2 μg total RNA are prepared and incubated withmagnetic beads decorated with oligo-dT, and other RNAs except mRNAwere removed by washing, and the mRNA was isolated from the oligo-dTbead of the poly(A) RNA selection kit (Lexogen). Library production wasinitiated by the random hybridization of starter/stopper heterodimerscontaining Illumina-compatible linker sequences to the poly (A) RNAbound to the magnetic beads. A single-tube reverse transcription andligation reaction extends the starter to the next hybridized heterodimer,where the newly synthesized cDNA insert is ligated to the stopper. Secondstrand synthesis was performed to release the library from the beads, and thelibrary was then amplified. Barcodes were introduced when the library wasamplified. High-throughput sequencing was performed as paired-end 100sequencing using HiSeq 2500 (Illumina). The sequenced reads weremapped to the University of California Santa Cruz zebrafish genome(danRer10) using STAR (v.2.5.1) (Dobin et al., 2013), and the geneexpression levels were quantified using the count module in STAR. TheedgeR (v.3.12.1) (Robinson et al., 2010) package was used to selectdifferentially expressed genes from the RNA-seq count data. Meanwhile,the trimmed mean of M-values-normalized counts per million value of eachgene was set to a baseline of 1 and log2-transformed for volcano plotdrawing (Figs 7A and S7A).

Statistical analysesStatistical analyses of the datawere performed using aMann–WhitneyU testor one-way ANOVAwith Dunnett’s multiple comparisons test using Prismsoftware (Ver.7). Data are mean±s.e.m. with *P<0.05, **P<0.01 and***P<0.005.

AcknowledgementsThe authors thank members of the Dr Lee and Dr Yu laboratories for helpfuldiscussion on the manuscript.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: K.-S.L., K.Y., J.-S.L.; Methodology: H.-J.C., J.-G.L, J.-H.K.,S.-Y.K., Y.H.H., H.-J.K., K.Y., J.-S.L.; Software: J.-H.K., S.-Y.K., J.-S.L.; Validation:J.-G.L., J.-S.L.; Formal analysis: J.-G.L., J.-H.K., S.-Y.K., Y.H.H., H.-J.K., K.Y.;Investigation: H.-J.C., J.-G.L., Y.H.H., K.-S.L., J.-S.L.; Resources: J.-S.L.; Datacuration: H.-J.C., Y.H.H., K.-S.L., K.Y., J.-S.L.; Writing - original draft: H.-J.C.,J.-H.K., S.-Y.K., Y.H.H., J.-S.L.; Writing - review & editing: H.-J.C., J.-S.L.;Supervision: J.-S.L.; Project administration: J.-S.L.; Funding acquisition: J.-S.L.

FundingThis work was supported by theNational Research Foundation of Korea (NRF-2011-0023507) and the National Research Council of Science & Technology (CRC-15-04-KIST), the Ministry of Science, ICT and Future Planning (MSIP) and the KoreaResearch Institute of Bioscience and Biotechnology Research Initiative Program.

Data availabilityRNA-seq data have been deposited in GEO under accession number GSE111280.

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.037044.supplemental

11

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

ReferencesAlMalki, W. H., Shahid, I., Mehdi, A. Y. and Hafeez, M. H. (2014). Assessmentmethods for angiogenesis and current approaches for its quantification. IndianJ. Pharmacol. 46, 251-256. doi:10.4103/0253-7613.132152

Arnold, T. D., Niaudet, C., Pang, M.-F., Siegenthaler, J., Gaengel, K., Jung, B.,Ferrero, G. M., Mukouyama, Y. S., Fuxe, J., Akhurst, R. et al. (2014). Excessivevascular sprouting underlies cerebral hemorrhage in mice lacking alphaVbeta8-TGFbeta signaling in the brain. Development 141, 4489-4499. doi:10.1242/dev.107193

Baek, K.-H., Zaslavsky, A., Lynch, R. C., Britt, C., Okada, Y., Siarey, R. J.,Lensch, M. W., Park, I.-H., Yoon, S. S., Minami, T. et al. (2009). Down’ssyndrome suppression of tumour growth and the role of the calcineurin inhibitorDSCR1. Nature 459, 1126-1130. doi:10.1038/nature08062

Bain, J., Plater, L., Elliott, M., Shpiro, N., Hastie, C. J., McLauchlan, H.,Klevernic, I., Arthur, J. S. C., Alessi, D. R. and Cohen, P. (2007). The selectivityof protein kinase inhibitors: a further update. Biochem. J. 408, 297-315. doi:10.1042/BJ20070797

Barrionuevo,W. R. and Burggren,W.W. (1999). O2 consumption and heart rate indeveloping zebrafish (Danio rerio): influence of temperature and ambient O2.Am. J. Physiol. 276, R505-R513. doi:10.1152/ajpregu.1999.276.2.r505

Berridge, M. J., Lipp, P. and Bootman, M. D. (2000). The versatility anduniversality of calcium signalling. Nat. Rev. Mol. Cell Biol. 1, 11-21. doi:10.1038/35036035

Bjorklund, G., Kern, J. K., Urbina, M. A., Saad, K., El-Houfey, A. A., Geier, D. A.,Chirumbolo, S., Geier, M. R., Mehta, J. A. and Aaseth, J. (2018). Cerebralhypoperfusion in autism spectrum disorder. Acta Neurobiol. Exp. (Wars.) 78,21-29. doi:10.21307/ane-2018-005

Bussmann, J., Wolfe, S. A. and Siekmann, A. F. (2011). Arterial-venous networkformation during brain vascularization involves hemodynamic regulation ofchemokine signaling. Development 138, 1717-1726. doi:10.1242/dev.059881

Chaoui, R. (2005). Prenatal ultrasound diagnosis of Down syndrome. After majormalformations, soft markers, nuchal translucency and skeletal signs, a newvascular sign? Ultrasound Obstet. Gynecol. 26, 214-217. doi:10.1002/uog.1985

Clapham, D. E. (2007). Calcium signaling. Cell 131, 1047-1058. doi:10.1016/j.cell.2007.11.028

Clark, K. J., Urban, M. D., Skuster, K. J. and Ekker, S. C. (2011). Transgeniczebrafish using transposable elements. Methods Cell Biol. 104, 137-149. doi:10.1016/B978-0-12-374814-0.00008-2

Crabtree, G. R. (2001). Calcium, calcineurin, and the control of transcription. J. Biol.Chem. 276, 2313-2316. doi:10.1074/jbc.R000024200

de Borchgrave, V., Saussu, F., Depre, A. and de Barsy, T. (2002). Moyamoyadisease and Down syndrome: case report and review of the literature.Acta Neurol.Belg. 102, 63-66.

Dejana, E., Tournier-Lasserve, E. and Weinstein, B. M. (2009). The control ofvascular integrity by endothelial cell junctions: molecular basis and pathologicalimplications. Dev. Cell 16, 209-221. doi:10.1016/j.devcel.2009.01.004

Deli, M. A. (2009). Potential use of tight junction modulators to reversibly openmembranous barriers and improve drug delivery. Biochim. Biophys. Acta 1788,892-910. doi:10.1016/j.bbamem.2008.09.016

Dilling, C., Roewer, N., Forster, C. Y. and Burek, M. (2017). Multipleprotocadherins are expressed in brain microvascular endothelial cells and mightplay a role in tight junction protein regulation. J. Cereb. Blood Flow Metab. 37,3391-3400. doi:10.1177/0271678X16688706

Dobin, A., Davis, C. A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut,P., Chaisson, M. and Gingeras, T. R. (2013). STAR: ultrafast universal RNA-seqaligner. Bioinformatics 29, 15-21. doi:10.1093/bioinformatics/bts635

Donahue, J. E., Khurana, J. S. and Adelman, L. S. (1998). Intracerebralhemorrhage in two patients with Down’s syndrome and cerebral amyloidangiopathy. Acta Neuropathol. 95, 213-216. doi:10.1007/s004010050789

Ericson, J., Thor, S., Edlund, T., Jessell, T. M. and Yamada, T. (1992). Earlystages of motor neuron differentiation revealed by expression of homeobox geneIslet-1. Science 256, 1555-1560. doi:10.1126/science.1350865

Folkman, J. (1995). Angiogenesis in cancer, vascular, rheumatoid and otherdisease. Nat. Med. 1, 27-31. doi:10.1038/nm0195-27

Fruen, B. R., Mickelson, J. R. and Louis, C. F. (1997). Dantrolene inhibition ofsarcoplasmic reticulum Ca2+ release by direct and specific action at skeletalmuscle ryanodine receptors. J. Biol. Chem. 272, 26965-26971. doi:10.1074/jbc.272.43.26965

Fujita, M., Cha, Y. R., Pham, V. N., Sakurai, A., Roman, B. L., Gutkind, J. S. andWeinstein, B. M. (2011). Assembly and patterning of the vascular network of thevertebrate hindbrain. Development 138, 1705-1715. doi:10.1242/dev.058776

Gao, X., Kouklis, P., Xu, N., Minshall, R. D., Sandoval, R., Vogel, S. M. andMalik,A. B. (2000). Reversibility of increased microvessel permeability in responseto VE-cadherin disassembly. Am. J. Physiol. Lung Cell. Mol. Physiol. 279,L1218-L1225. doi:10.1152/ajplung.2000.279.6.L1218

Gjini, E., Hekking, L. H., Kuchler, A., Saharinen, P., Wienholds, E., Post, J.-A.,Alitalo, K. and Schulte-Merker, S. (2011). Zebrafish Tie-2 shares a redundantrole with Tie-1 in heart development and regulates vessel integrity. Dis. ModelMech. 4, 57-66. doi:10.1242/dmm.005033

Gockler, N., Jofre, G., Papadopoulos, C., Soppa, U., Tejedor, F. J. and Becker, W.(2009). Harmine specifically inhibits protein kinase DYRK1A and interferes withneurite formation.FEBS J. 276, 6324-6337. doi:10.1111/j.1742-4658.2009.07346.x

Gorlach, A., Bertram, K., Hudecova, S. and Krizanova, O. (2015). Calciumand ROS: a mutual interplay. Redox Biol. 6, 260-271. doi:10.1016/j.redox.2015.08.010

Gowda, S., Bhat, D., Feng, Z., Chang, C.-H. and Ross, R. D. (2014). Pulmonaryvein stenosis with Down syndrome: a rare and frequently fatal cause of pulmonaryhypertension in infants and children. Congenit. Heart Dis. 9, E90-E97. doi:10.1111/chd.12088

Grau, C., Arato, K., Fernandez-Fernandez, J. M., Valderrama, A., Sindreu, C.,Fillat, C., Ferrer, I., de la Luna, S. and Altafaj, X. (2014). DYRK1A-mediatedphosphorylation of GluN2A at Ser(1048) regulates the surface expression andchannel activity of GluN1/GluN2A receptors. Front Cell Neurosci. 8, 331. doi:10.3389/fncel.2014.00331

Hammerle, B., Elizalde, C., Galceran, J., Becker, W. and Tejedor, F. J. (2003).TheMNB/DYRK1A protein kinase: neurobiological functions and Down syndromeimplications. J. Neural. Transm. Suppl. 67, 129-137. doi:10.1007/978-3-7091-6721-2_11

Hammerle, B., Elizalde, C. and Tejedor, F. J. (2008). The spatio-temporal andsubcellular expression of the candidate Down syndrome gene Mnb/Dyrk1A in thedeveloping mouse brain suggests distinct sequential roles in neuronaldevelopment. Eur. J. Neurosci. 27, 1061-1074. doi:10.1111/j.1460-9568.2008.06092.x

Harvey, S. A., Sealy, I., Kettleborough, R., Fenyes, F.,White, R., Stemple, D. andSmith, J. C. (2013). Identification of the zebrafish maternal and paternaltranscriptomes. Development 140, 2703-2710. doi:10.1242/dev.095091

Hasle, H., Clemmensen, I. H. and Mikkelsen, M. (2000). Risks of leukaemia andsolid tumours in individuals with Down’s syndrome. Lancet 355, 165-169. doi:10.1016/S0140-6736(99)05264-2

He, J., Yao, J., Sheng, H. and Zhu, J. (2015). Involvement of the dual-specificitytyrosine phosphorylation-regulated kinase 1A-alternative splicing factor-calcium/calmodulin-dependent protein kinase IIdelta signaling pathway in myocardialinfarction-induced heart failure of rats. J. Card. Fail. 21, 751-760. doi:10.1016/j.cardfail.2015.05.015

Himpel, S., Panzer, P., Eirmbter, K., Czajkowska, H., Sayed, M., Packman, L. C.,Blundell, T., Kentrup, H., Grotzinger, J., Joost, H. G. et al. (2001). Identificationof the autophosphorylation sites and characterization of their effects in the proteinkinase DYRK1A. Biochem. J. 359, 497-505. doi:10.1042/bj3590497

Hogan, B. M. and Schulte-Merker, S. (2017). How to plumb a pisces:understanding vascular development and disease using zebrafish embryos.Dev. Cell 42, 567-583. doi:10.1016/j.devcel.2017.08.015

Hogan, P. G., Chen, L., Nardone, J. and Rao, A. (2003). Transcriptional regulationby calcium, calcineurin, and NFAT. Genes Dev. 17, 2205-2232. doi:10.1101/gad.1102703

Hogan, B. M., Verkade, H., Lieschke, G. J. andHeath, J. K. (2008). Manipulation ofgene expression during zebrafish embryonic development using transientapproaches.MethodsMol. Biol. 469, 273-300. doi:10.1007/978-1-60327-469-2_19

Inoue, A., Takahashi, M., Hatta, K., Hotta, Y. and Okamoto, H. (1994).Developmental regulation of islet-1 mRNA expression during neuronaldifferentiation in embryonic zebrafish. Dev. Dyn. 199, 1-11. doi:10.1002/aja.1001990102

Inzitari, D. and Poggesi, A. (2005). Calcium channel blockers and stroke. AgingClin. Exp. Res. 17, 16-30.

Isogai, S., Horiguchi, M. andWeinstein, B. M. (2001). The vascular anatomy of thedeveloping zebrafish: an atlas of embryonic and early larval development. Dev.Biol. 230, 278-301. doi:10.1006/dbio.2000.9995

Jastrzebski, K., Kacperska, M. J., Majos, A., Grodzka, M. and Glabinski, A.(2015). Hemorrhagic stroke, cerebral amyloid angiopathy, Down syndrome andthe Boston criteria.Neurol. Neurochir. Pol. 49, 193-196. doi:10.1016/j.pjnns.2015.04.006

Kang, I., Kim, J.-I., Chang, S.-G., Lee, S.-J., Choi, S.-L., Ha, J. and Kim, S. S.(1999). Mannan-binding lectin (MBL)-associated plasma protein present inhuman urine inhibits calcium oxalate crystal growth. FEBS Lett. 462, 89-93.doi:10.1016/S0014-5793(99)01509-4

Khorchid, A. and Ikura, M. (2002). How calpain is activated by calcium.Nat. Struct.Biol. 9, 239-241. doi:10.1038/nsb0402-239

Kim, O.-H., Cho, H.-J., Han, E., Hong, T. I., Ariyasiri, K., Choi, J.-H., Hwang,K.-S., Jeong, Y.-M., Yang, S.-Y., Yu, K. et al. (2017). Zebrafish knockout of Downsyndrome gene, DYRK1A, shows social impairments relevant to autism. Mol.Autism 8, 50. doi:10.1186/s13229-017-0168-2

Korzh, V., Edlund, T. and Thor, S. (1993). Zebrafish primary neurons initiateexpression of the LIM homeodomain protein Isl-1 at the end of gastrulation.Development 118, 417-425.

Li, H., Rao, A. and Hogan, P. G. (2011). Interaction of calcineurin with substratesand targeting proteins. Trends Cell Biol. 21, 91-103. doi:10.1016/j.tcb.2010.09.011

Liu, J., Fraser, S. D., Faloon, P.W., Rollins, E. L., VomBerg, J., Starovic-Subota,O., Laliberte, A. L., Chen, J.-N., Serluca, F. C. and Childs, S. J. (2007). A

12

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms

betaPix Pak2a signaling pathway regulates cerebral vascular stability in zebrafish.Proc. Natl. Acad. Sci. USA 104, 13990-13995. doi:10.1073/pnas.0700825104

Liu, F., Liang, Z., Wegiel, J., Hwang, Y. W., Iqbal, K., Grundke-Iqbal, I.,Ramakrishna, N. andGong, C. X. (2008). Overexpression of Dyrk1A contributesto neurofibrillary degeneration in Down syndrome. FASEB J. 22, 3224-3233.doi:10.1096/fj.07-104539

Loh,C., Shaw,K. T.-Y.,Carew, J., Viola, J. P. B., Luo,C., Perrino,B.A. andRao,A.(1996). Calcineurin binds the transcription factorNFAT1and reversibly regulates itsactivity. J. Biol. Chem. 271, 10884-10891. doi:10.1074/jbc.271.18.10884

Lounsbury, K. M., Hu, Q. and Ziegelstein, R. C. (2000). Calcium signaling andoxidant stress in the vasculature. Free Radic. Biol. Med. 28, 1362-1369. doi:10.1016/S0891-5849(00)00222-7

MacRae, C. A. and Peterson, R. T. (2015). Zebrafish as tools for drug discovery.Nat. Rev. Drug Discov. 14, 721-731. doi:10.1038/nrd4627

Manoli, M. and Driever, W. (2012). Fluorescence-activated cell sorting (FACS) offluorescently tagged cells from zebrafish larvae for RNA isolation. Cold SpringHarb. Protoc. 2012, pdb.prot069633. doi:10.1101/pdb.prot069633

Marti, E., Altafaj, X., Dierssen, M., de la Luna, S., Fotaki, V., Alvarez, M., Perez-Riba, M., Ferrer, I. and Estivill, X. (2003). Dyrk1A expression pattern supportsspecific roles of this kinase in the adult central nervous system. Brain Res. 964,250-263. doi:10.1016/S0006-8993(02)04069-6

Mathavan, S., Lee, S. G. P., Mak, A., Miller, L. D., Murthy, K. R. K., Govindarajan,K. R., Tong, Y., Wu, Y. L., Lam, S. H., Yang, H. et al. (2005). Transcriptomeanalysis of zebrafish embryogenesis using microarrays. PLoS Genet. 1, 260-276.doi:10.1371/journal.pgen.0010029

Minami, T., Horiuchi, K., Miura, M., Abid, M. R., Takabe, W., Noguchi, N., Kohro,T., Ge, X., Aburatani, H., Hamakubo, T. et al. (2004). Vascular endothelialgrowth factor- and thrombin-induced termination factor, Down syndrome criticalregion-1, attenuates endothelial cell proliferation and angiogenesis. J. Biol. Chem.279, 50537-50554. doi:10.1074/jbc.M406454200

Mito, T. and Becker, L. E. (1992). Vascular dysplasia in Down syndrome: a possiblerelationship to moyamoya disease. Brain Dev. 14, 248-251. doi:10.1016/S0387-7604(12)80239-7

Moens, C. B. and Prince, V. E. (2002). Constructing the hindbrain: insights from thezebrafish. Dev. Dyn. 224, 1-17. doi:10.1002/dvdy.10086

Munaron, L. (2006). Intracellular calcium, endothelial cells andangiogenesis.RecentPat. Anticancer Drug. Discov. 1, 105-119. doi:10.2174/157489206775246502

Nizetic, D. and Groet, J. (2012). Tumorigenesis in Down’s syndrome: big lessonsfrom a small chromosome. Nat. Rev. Cancer 12, 721-732. doi:10.1038/nrc3355

Orr, N., Arnaout, R., Gula, L. J., Spears, D. A., Leong-Sit, P., Li, Q., Tarhuni, W.,Reischauer, S., Chauhan, V. S., Borkovich, M. et al. (2016). A mutation in theatrial-specific myosin light chain gene (MYL4) causes familial atrial fibrillation.Nat.Commun. 7, 11303. doi:10.1038/ncomms11303

Oxtoby, E. and Jowett, T. (1993). Cloning of the zebrafish krox-20 gene (krx-20)and its expression during hindbrain development. Nucleic Acids Res. 21,1087-1095. doi:10.1093/nar/21.5.1087

Pandya, N. M., Dhalla, N. S. and Santani, D. D. (2006). Angiogenesis–a new targetfor future therapy.Vascul. Pharmacol. 44, 265-274. doi:10.1016/j.vph.2006.01.005

Phng, L.-K., Potente, M., Leslie, J. D., Babbage, J., Nyqvist, D., Lobov, I., Ondr,J. K., Rao, S., Lang, R. A., Thurston, G. et al. (2009). Nrarp coordinatesendothelial Notch and Wnt signaling to control vessel density in angiogenesis.Dev. Cell 16, 70-82. doi:10.1016/j.devcel.2008.12.009

Pipitone, S., Garofalo, C., Corsello, G., Mongiovì, M., Piccione, M., Maresi, E.and Sperandeo, V. (2003). Abnormalities of the umbilico-portal venous system inDown syndrome: a report of two new patients. Am. J. Med. Genet. A 120A,528-532. doi:10.1002/ajmg.a.20081

Quaegebeur, A., Lange, C. and Carmeliet, P. (2011). The neurovascular link inhealth and disease: molecular mechanisms and therapeutic implications. Neuron71, 406-424. doi:10.1016/j.neuron.2011.07.013

Rahmani, Z., Lopes, C., Rachidi, M. and Delabar, J.-M. (1998). Expression of themnb (dyrk) protein in adult and embryonic mouse tissues.Biochem. Biophys. Res.Commun. 253, 514-518. doi:10.1006/bbrc.1998.9803

Rathore, M. H. and Sreenivasan, V. V. (1989). Vertebral and right subclavian arteryabnormalities in the Down syndrome. Am. J. Cardiol. 63, 1528-1529. doi:10.1016/0002-9149(89)90023-4

Reid, I. R. and Bolland, M. J. (2008). Calcium supplementation and vasculardisease. Climacteric 11, 280-286. doi:10.1080/13697130802229639

Robinson, M. D., McCarthy, D. J. and Smyth, G. K. (2010). edgeR: a Bioconductorpackage for differential expression analysis of digital gene expression data.Bioinformatics 26, 139-140. doi:10.1093/bioinformatics/btp616

Rozen, E. J., Roewenstrunk, J., Barallobre, M. J., Di Vona, C., Jung, C.,Figueiredo, A. F., Luna, J., Fillat, C., Arbones, M. L., Graupera, M. et al. (2018).DYRK1A kinase positively regulates angiogenic responses in endothelial cells.Cell Rep. 23, 1867-1878. doi:10.1016/j.celrep.2018.04.008

Sabde, S., Tullu, M. S., Lahiri, K. R. and Vaswani, R. K. (2005). Moyamoyasyndrome in a child with Down syndrome. Arch. Med. Res. 36, 421-424. doi:10.1016/j.arcmed.2005.03.003

Schmidt, R., Strahle, U. and Scholpp, S. (2013). Neurogenesis in zebrafish - fromembryo to adult. Neural Dev. 8, 3. doi:10.1186/1749-8104-8-3

Shen, Q., Rigor, R. R., Pivetti, C. D., Wu,M. H. andYuan, S. Y. (2010). Myosin lightchain kinase in microvascular endothelial barrier function. Cardiovasc. Res. 87,272-280. doi:10.1093/cvr/cvq144

Sobey, C. G., Judkins, C. P., Sundararajan, V., Phan, T. G., Drummond, G. R.and Srikanth, V. K. (2015). Risk of major cardiovascular events in people withdown syndrome. PLoS One 10, e0137093. doi:10.1371/journal.pone.0137093

Stewart, A. D., Calder, A. L., Neutze, J. M., James, A. H. and Brandt, P. W. T.(1992). Stenosis of pulmonary veins in Down syndrome. J. Paediatr. Child Health28, 192-195. doi:10.1111/j.1440-1754.1992.tb02641.x

Tejedor, F., Zhu, X. R., Kaltenbach, E., Ackermann, A., Baumann, A., Canal, I.,Heisenberg, M., Fischbach, K. F. and Pongs, O. (1995). minibrain: a newprotein kinase family involved in postembryonic neurogenesis in Drosophila.Neuron 14, 287-301. doi:10.1016/0896-6273(95)90286-4

Thisse, C. and Thisse, B. (2008). High-resolution in situ hybridization to whole-mount zebrafish embryos. Nat. Protoc. 3, 59-69. doi:10.1038/nprot.2007.514

Tımar, J., Dome, B., Fazekas, K., Janovics, A. and Paku, S. (2001).Angiogenesis-dependent diseases and angiogenesis therapy. Pathol. Oncol.Res. 7, 85-94. doi:10.1007/BF03032573

Timme-Laragy, A. R., Karchner, S. I. and Hahn, M. E. (2012). Gene knockdown bymorpholino-modified oligonucleotides in the zebrafish (Danio rerio) model:applications for developmental toxicology. Methods Mol. Biol. 889, 51-71.doi:10.1007/978-1-61779-867-2_5

Tiruppathi, C., Minshall, R. D., Paria, B. C., Vogel, S. M. and Malik, A. B. (2002).Role of Ca2+ signaling in the regulation of endothelial permeability. Vascul.Pharmacol. 39, 173-185. doi:10.1016/S1537-1891(03)00007-7

Tomita, M., Hayashi, M. and Awazu, S. (1996). Absorption-enhancing mechanismof EDTA, caprate, and decanoylcarnitine in Caco-2 cells. J. Pharm. Sci. 85,608-611. doi:10.1021/js9504604

Udan, R. S., Culver, J. C. and Dickinson, M. E. (2013). Understanding vasculardevelopment. Wiley Interdiscip. Rev. Dev. Biol. 2, 327-346. doi:10.1002/wdev.91

Ulrich, F., Ma, L.-H., Baker, R. G. and Torres-Vazquez, J. (2011). Neurovasculardevelopment in the embryonic zebrafish hindbrain. Dev. Biol. 357, 134-151.doi:10.1016/j.ydbio.2011.06.037

Varshney, G. K., Sood, R. and Burgess, S. M. (2015). Understanding and editingthe zebrafish genome. Adv. Genet. 92, 1-52. doi:10.1016/bs.adgen.2015.09.002

Wang, S., Aurora, A. B., Johnson, B. A., Qi, X., McAnally, J., Hill, J. A.,Richardson, J. A., Bassel-Duby, R. and Olson, E. N. (2008). The endothelial-specific microRNA miR-126 governs vascular integrity and angiogenesis. Dev.Cell 15, 261-271. doi:10.1016/j.devcel.2008.07.002

Wang, Y., Pan, L., Moens, C. B. and Appel, B. (2014). Notch3 establishes brainvascular integrity by regulating pericyte number. Development 141, 307-317.doi:10.1242/dev.096107

Wegiel, J., Gong, C.-X. and Hwang, Y.-W. (2011). The role of DYRK1A inneurodegenerative diseases. FEBS J. 278, 236-245. doi:10.1111/j.1742-4658.2010.07955.x

Zhao, Z. and Michaely, P. (2009). The role of calcium in lipoprotein release by thelow-density lipoprotein receptor. Biochemistry 48, 7313-7324. doi:10.1021/bi900214u

Zilbovicius, M., Boddaert, N., Belin, P., Poline, J. B., Remy, P., Mangin, J. F.,Thivard, L., Barthelemy, C. andSamson, Y. (2000). Temporal lobe dysfunction inchildhood autism: a PET study. Positron emission tomography. Am. J. Psychiatry157, 1988-1993. doi:10.1176/appi.ajp.157.12.1988

Zlokovic,B.V. (2008). Theblood-brain barrier in healthand chronic neurodegenerativedisorders. Neuron 57, 178-201. doi:10.1016/j.neuron.2008.01.003

Zu, Y., Tong, X., Wang, Z., Liu, D., Pan, R., Li, Z., Hu, Y., Luo, Z., Huang, P., Wu,Q. et al. (2013). TALEN-mediated precise genome modification by homologousrecombination in zebrafish. Nat. Methods 10, 329-331. doi:10.1038/nmeth.2374

13

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm037044. doi:10.1242/dmm.037044

Disea

seModels&Mechan

isms