RESEARCH ARTICLE

Functional testing of a human PBX3 variant in zebrafish reveals apotential modifier role in congenital heart defectsGist H. Farr, III1, Kimia Imani1,2, Darren Pouv1,2 and Lisa Maves1,3,*

ABSTRACTWhole-genome and exome sequencing efforts are increasinglyidentifying candidate genetic variants associated with humandisease. However, predicting and testing the pathogenicity of agenetic variant remains challenging. Genome editing allows for therigorous functional testing of human genetic variants in animalmodels. Congenital heart defects (CHDs) are a prominent example ofa human disorder with complex genetics. An inherited sequencevariant in the human PBX3 gene (PBX3 p.A136V) has previouslybeen shown to be enriched in a CHD patient cohort, indicating thatthe PBX3 p.A136V variant could be a modifier allele for CHDs.Pbx genes encode three-amino-acid loop extension (TALE)-classhomeodomain-containing DNA-binding proteins with diverse roles indevelopment and disease, and are required for heart development inmouse and zebrafish. Here, we used CRISPR-Cas9 genome editingto directly test whether this Pbx gene variant acts as a geneticmodifier in zebrafish heart development. We used a single-strandedoligodeoxynucleotide to precisely introduce the human PBX3p.A136V variant in the homologous zebrafish pbx4 gene ( pbx4p.A131V). We observed that zebrafish that are homozygous forpbx4 p.A131V are viable as adults. However, the pbx4 p.A131Vvariant enhances the embryonic cardiac morphogenesis phenotypecaused by loss of the known cardiac specification factor, Hand2.Our study is the first example of using precision genome editing inzebrafish to demonstrate a function for a human disease-associatedsingle nucleotide variant of unknown significance. Our workunderscores the importance of testing the roles of inheritedvariants, not just de novo variants, as genetic modifiers of CHDs.Our study provides a novel approach toward advancing ourunderstanding of the complex genetics of CHDs.

KEY WORDS: CRISPR-Cas9, Genetic variant, Heart, Modifier, Pbx,Zebrafish

INTRODUCTIONWhole-genome and exome sequencing efforts are increasinglyidentifying genetic variation in the general human population as wellas candidate genetic variants associated with disease (Lek et al., 2016;Wright et al., 2015). There are several approaches for predicting

and testing the pathogenicity of a genetic variant (Cox, 2015;Richards et al., 2015; Starita et al., 2017). However, demonstratinga function of a particular sequence variant remains challenging.Genome editing now allows for the precise engineering of humangenetic variants in animal models for rigorous functional testing(Doudna and Charpentier, 2014; Peng et al., 2014). There areexamples of the use of CRISPR-Cas9 engineering in mouse andCaenorhabditis elegans animal models to demonstrate functionaleffects of human disease-associated sequence variants (Arno et al.,2016; DiStasio et al., 2017; Lin et al., 2016; Prior et al., 2017).However, it is not yet clear how effectively human variants ofunknown significance can be functionally tested through genomeediting in animal models.

Congenital heart defects (CHDs) are a prominent example of ahuman disorder with complex genetics (Fahed et al., 2013; Gelb andChung, 2014; Zaidi and Brueckner, 2017). CHDs occur in ∼1% oflive births and are the leading cause of infant death owing to birthdefects. Intensive studies have uncovered prominent roles fortranscription and chromatin factors in heart development and CHDs(Chang and Bruneau, 2011; Fahed et al., 2013; Gelb and Chung,2014). Cardiac transcription factors of the GATA, HAND, MEF2,NKX, SRF and TBX families are required for heart developmentin mouse and zebrafish animal models, and mutations in genesencoding these factors can cause human CHDs (Evans et al., 2010;McCulley and Black, 2012; Olson, 2006). Large-scale whole-exome sequencing studies find that CHD cases show an excess of denovo mutations for many transcription and chromatin factors(Homsy et al., 2015; Zaidi et al., 2013). In spite of these efforts,these de novo mutations likely account for only ∼10% of CHDs(Gelb and Chung, 2014; Homsy et al., 2015; Zaidi et al., 2013).Although additional studies are identifying potential contributionsof inherited mutations in CHDs (Jin et al., 2017), our understandingof the genetics of CHDs is still incomplete.

The genetics of CHDs is complex, in part, because the samecandidate gene, and even the same sequence variant, can be associatedwith a spectrum of heart malformations and can even be present incontrol cases (Fahed et al., 2013; Gelb and Chung, 2014; Zaidi andBrueckner, 2017). This is exemplified in studies ofNKX2.5mutationsin CHD patients and families (Elliott et al., 2003; McElhinney et al.,2003; Stallmeyer et al., 2010). Thus, genetic risk factors, or modifiergenes, likely influence the phenotype of CHDs, but modifier genes aredifficult to identify and characterize (Fahed et al., 2013; Gelb andChung, 2014; Zaidi and Brueckner, 2017). In order to understand theetiology of CHDs and the roles of genetic risk factors in influencingthe phenotypic spectrum of CHDs, it is imperative that we increaseour understanding of how genetic variants and modifier allelesregulate heart development and contribute to CHDs.

Sequence variants in human PBX genes have been identified inpatients with CHDs (Arrington et al., 2012; Slavotinek et al., 2017),indicating that these PBX variants could contribute to CHDs.One of these variants, an inherited missense variant in the codingReceived 4 June 2018; Accepted 3 September 2018

1Center for Developmental Biology and Regenerative Medicine, Seattle Children’sResearch Institute, Seattle, WA 98101, USA. 2University of Washington, Seattle,WA 98195, USA. 3Department of Pediatrics, University of Washington, Seattle,WA 98195, USA.

*Author for correspondence (lmaves@u.washington.edu)

L.M., 0000-0002-9798-790X

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

region of the PBX3 gene (p.A136V; 9:128678097 C>T), occurredat a frequency of 2.6% in a cohort of CHD patients (0.66% incontrols; Arrington et al., 2012). These patients exhibited aspectrum of CHDs, particularly outflow tract malformations(Arrington et al., 2012). Pbx genes encode three-amino-acid loopextension (TALE)-class homeodomain-containing DNA-bindingproteins, which have diverse roles in development and disease(Cerdá-Esteban and Spagnoli, 2014; Moens and Selleri, 2006).In mouse embryos, Pbx1 is required for heart development, and lossof different combinations of null alleles of Pbx1, Pbx2 and Pbx3leads to a spectrum of cardiac outflow tract defects (Chang et al.,2008; Stankunas et al., 2008). Our previous studies have shownthat zebrafish Pbx proteins are also needed for outflow tractdevelopment, and for early myocardial differentiation andmorphogenesis (Kao et al., 2015; Maves et al., 2009). The PBX3p.A136V variant lies in a highly conserved polyalanine tract, whichhas been implicated in Pbx binding to histone deacetylase (HDAC)chromatin proteins (Saleh et al., 2000). Although in silico programspredict this variant to be deleterious (Arrington et al., 2012), it is notknown whether it affects the function of PBX3 or contributes toCHD. Because this variant is present in controls and has thepotential to be inherited, it might represent a modifier or risk factorfor CHDs. The PBX3 p.A131V variant is present in the humanpopulation with an allele frequency of >0.6% [Arrington et al.,2012; Exome Aggregation Consortium (ExAC), http://exac.broadinstitute.org/], and so it is likely to be excluded fromstudies of de novo or rare inherited variants associated with CHDs(Homsy et al., 2015; Jin et al., 2017; Sifrim et al., 2016; Zaidiet al., 2013).Here, we use CRISPR-Cas9 genome editing in zebrafish to test

whether the PBX3 p.A131V variant can function as a modifier allelein CHDs. In particular, we test whether this variant enhances thephenotype caused by loss of a cardiac specification factor, Hand2, inzebrafish heart development. Our study is the first example, ofwhich we are aware, of using precision genome editing in zebrafishto demonstrate a function for a human disease-associated singlenucleotide variant of unknown significance. Our work provides aproof of principle for using genome editing in zebrafish to test thefunctions of human DNA variants in complex genetic disease. Ourwork also underscores the importance of testing the roles ofinherited variants as genetic modifiers of CHDs.

RESULTSZebrafish pbx4, but not pbx3b, is required for earlycardiac morphogenesisAn inherited heterozygous variant in PBX3, c.407 C>T, predictingp.Arg136>Val, was previously identified as enriched in a cohort ofpatients with CHDs (Arrington et al., 2012). The allelic frequency ofthis variant was significantly less frequent in the study’s controlpopulation (P=0.047; Arrington et al., 2012), and is also significantlyless frequent in the current ExAC database population (P=0.0047,Chi-square test; ExAC, http://exac.broadinstitute.org/). This PBX3p.A136V variant is present in a highly conserved polyalanine tract,which lies between the two PBC domains that interact with HDACandMeis proteins (Fig. 1A; Choe et al., 2009; Saleh et al., 2000). Thehuman and zebrafish Pbx genes that do not show 100% conservationof the polyalanine tract (human PBX4 and zebrafish pbx2 and pbx3a;Fig. 1A) appear to have reduced functional roles in development, ashuman PBX4 variants in the general population are observed at aboutthe same frequencies as expected by chance (ExAC, http://exac.broadinstitute.org/), and zebrafish pbx2 null mutants are homozygousviable as adults (G.H.F. and L.M., unpublished). However,

loss-of-function variants in PBX3 are highly underrepresented inthe ExAC database, suggesting that human PBX3 is required forviability. The enrichment of the PBX3 p.A136V variant in a CHDpatient cohort suggests that it might contribute a modifier role inthe complex genetics of CHDs, and led us to investigate a potentialfunction of this allele in zebrafish.

In zebrafish and in mice, multiple Pbx genes are expressedbroadly and have high functional redundancy (Capellini et al., 2006;Ferretti et al., 2011; Moens and Selleri, 2006; Pöpperl et al., 2000;Ruzicka et al., 2015). Zebrafish have six Pbx genes, and pbx3b is theclosest ortholog of human PBX3 (Fig. 1B). Neither pbx3b nor pbx3aare detectably expressed during early zebrafish development priorto ∼48 h postfertilization (hpf) (Fig. 1C; Ruzicka et al., 2015;Waskiewicz et al., 2002). Because pbx3b has a conservedpolyalanine tract (Fig. 1A) and shows upregulated expression by48 hpf (Fig. 1C) (after heart tube formation in zebrafish but aroundthe stage of early outflow tract development) (Grimes et al., 2006),we tested the function of pbx3b using CRISPR-Cas9 to generatezebrafish pbx3bmutants. The pbx3bscm8mutation that we generatedcreates an early stop codon (Fig. 1D). We found that zebrafishpbx3bscm8/scm8 larvae are viable and appear normal (Fig. 1E).We confirmed that the pbx3bscm8 mutation leads to downregulationof pbx3b expression in pbx3bscm8/scm8 embryos, likely due tononsense-mediated RNA degradation, as observed for pbx4 inpbx4b557/b557 embryos (Pöpperl et al., 2000). We did not observeany upregulation of other Pbx genes that could compensate for lossof pbx3b (Fig. 1F). We also determined that pbx3bscm8/scm8 embryosshow reduced levels of Pbx3b protein but, again, show no evidencefor upregulation of other Pbx proteins (Fig. 1G). pbx3b does notshow any detectable requirements during early heart development,as heart tube formation, heart chamber formation and outflow tractdevelopment appear grossly normal in pbx3bscm8/scm8 embryos(Fig. 1F,G). pbx4, however, is the main zebrafish Pbx gene expressedduring early zebrafish development (Fig. 1C; Waskiewicz et al.,2002). We have previously shown that zebrafish pbx4 is neededfor proper heart and outflow tract development (Kao et al., 2015;Maves et al., 2009). In particular, pbx4 mutant embryos showdisrupted myocardial morphogenesis and variable defects in outflowtract development (Kao et al., 2015; Fig. 1F,G). Because Pbxgenes have been shown to function redundantly (Capellini et al.,2006; Ferretti et al., 2011; Maves et al., 2007; Waskiewicz et al.,2002), we tested whether pbx3b and pbx4 function redundantlyduring early myocardial morphogenesis and outflow tractdevelopment. We crossed our pbx3bscm8 strain with thepbx4b557 null allele strain (Kao et al., 2015; Pöpperl et al.,2000) to generate pbx3bscm8/scm8;pbx4b557/b557 embryos.pbx4b557/b557 and pbx3bscm8/scm8;pbx4b557/b557 embryos bothshow variably disrupted formation of the myocardium andoutflow tract (Fig. 1F,G), but pbx3bscm8/scm8;pbx4b557/b557

embryos do not appear to exhibit any more severe myocardialdifferentiation and outflow tract defects than those that wepreviously described for pbx4b557/b557 embryos (Fig. 1F,G; Kaoet al., 2015). Thus, in zebrafish, pbx4, but not pbx3b, plays acritical function in early myocardial and outflow tract formation.Studies of gene expression in early human embryos find thatPBX3 is expressed at higher levels than other PBX genes (Yanet al., 2013), similar to pbx4 in zebrafish (Fig. 1C). Functionaldifferences between Pbx genes have been argued to be more likelycaused by differences in expression than differences inbiochemical activity (Moens and Selleri, 2006; Pöpperl et al.,2000). Therefore, we decided to address the function of the humanPBX3 p.A136V variant in the zebrafish pbx4 gene.

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Fig. 1. Zebrafish pbx4, but not pbx3b, is required for early cardiac morphogenesis. (A) Alignment of human (Hs) and zebrafish (Dr) Pbx proteins in theregion of the polyalanine tract. Numbers indicate amino acid positions. Partial PBC-A and PBC-B domains are underlined. The arrow marks the position ofamino acid 136 in human PBX3. (B) Phylogenetic analysis of human (Hs), mouse (Mm) and zebrafish (Dr) Pbx genes. DmExd is the Drosophila Pbxgene ortholog extradenticle. (C) qRT-PCR analysis of Pbx gene expression in wild-type zebrafish embryos at four embryonic stages: eight-cell (∼1.25 hpf),14 somites (s; ∼16 hpf), 24 hpf and 48 hpf. Levels of expression of each Pbx gene are shown relative to the expression of odc1. Error bars represent standarddeviations for three technical replicates. (D) Schematic of zebrafish Pbx3b protein domains and inferred domains encoded by the CRISPR-Cas9-generatedpbx3bscm8 allele. (E) Images of live pbx3bscm8/+ and pbx3bscm8/scm8 larvae at 5 dpf. pbx3bscm8/scm8 larvae show no obvious heart or other defects at least up to7 dpf (n=15, pbx3bscm8/scm8; n=14, pbx3bscm8/+; n=7, pbx3b+/+). Scale bar: 300 μm. (F) qRT-PCR analysis of Pbx gene expression in pbx3bscm8/scm8 embryosrelative to sibling pbx3b+/+ embryos at 48 hpf. Levels of expression of each Pbx gene are normalized to the expression of eef1a1l1. Error bars represent standarddeviations for four biological replicates. *P=0.0005, Student’s t-test using Welch’s correction for unequal standard deviations. (G) Western blot analysis of Pbxprotein expression. pbx3scm8/scm8 and pbx4b557/b557 embryos were used to document identities of the proteins recognized by the anti-pan-Pbx antibody.The upper band is Pbx2, as previously described (Maves et al., 2007; Waskiewicz et al., 2002; G.H.F. and L.M., unpublished). Quantification of the middleband, normalized to Actin levels, shows that pbx3scm8/scm8 embryos have 65% of wild-type levels, pbx4b557/b557 embryos have 54% of wild-type levels andpbx3scm8/scm8;pbx4b557/b557 embryos have 10% of wild-type levels, demonstrating that the middle band consists of both Pbx3b and Pbx4. (H) Myocardial markermyl7 expression at 24 hpf appears normal in pbx3+/+;pbx4+/+ (n=11), pbx3scm8/scm8;pbx4+/+ (n=2) and pbx3scm8/scm8;pbx4b557/+ (n=15) embryos. pbx3+/+;pbx4b557/b557 (n=8) and pbx3scm8/scm8;pbx4b557/b557 (n=7) embryos show similarly disrupted early heart tube morphogenesis, as we previously described forpbx4b557/b557 embryos (Kao et al., 2015). Dorsal views; anterior is up. Scale bar: 50 μm. (I) Expression of myocardial markermyl7 (red) and outflow tract markerelnb (green) (Miao et al., 2007) at 60 hpf appears normal in pbx3+/+;pbx4+/+ (n=8) and in pbx3scm8/scm8;pbx4+/+ (n=9) embryos. pbx3+/+;pbx4b557/b557 (n=11)and pbx3scm8/scm8;pbx4b557/b557 (n=10) embryos show variably disruptedmyocardial and outflow tract morphogenesis, similar to what we previously described forpbx4b557/b557 embryos (Kao et al., 2015). V, ventricle; A, atrium. Ventral views; anterior is up. Scale bar: 50 μm.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Engineering zebrafish pbx4 p.A131V variant strainsCompared with the human PBX3 p.A136 site, the zebrafishpbx4 gene has an orthologous site of A131 (nucleotide C392;Figs 1A and 2A). To engineer zebrafish carrying the pbx4 p.A131Vvariant, we turned to the CRISPR-Cas9 system (Blackburn et al.,2013; Hwang et al., 2013a, 2015). Although there have been recentimprovements, most studies report a low efficiency of introducingprecise base changes using CRISPR-Cas9 in zebrafish (Hoshijimaet al., 2016; Shah and Moens, 2016; Zhang et al., 2017, 2018).Therefore, to optimize our efforts, we first wanted to ensure that wewere using efficient CRISPR-component reagents.We began by designing and testing synthetic guide RNAs.

We designed guide RNAs that utilized two adjacent potentialprotospacer adjacent motif (PAM) sequences (NGG) and overlappedthe C392 nucleotide (Fig. 2B,C). Synthetic single-guide RNAs(sgRNAs) were produced in vitro, as previously described (Hwang

et al., 2013a), by annealing and ligating pairs of complementaryoligonucleotides into a plasmid vector containing a T7 promoter(Table 1, Fig. 2C). To test the ability of each sgRNA to induceinsertions or deletions (indels) at the target locus, we injected one-cell-stage zebrafish embryos with a mixture of Cas9 mRNA andone of the three sgRNAs, allowed the embryos to develop for 24 hand then collected individual embryos for DNA analysis. Indelswere detected by PCR amplifying the region around the target siteand digesting the amplicons with PstI, which has a recognitionsite within the target site (Fig. 2B,D). We observed a strikingdifference in the efficacy of the three sgRNAs. The first twosgRNAs produced no or extremely low levels of indels (Fig. 2Dand data not shown). In contrast, 24 of 24 embryos assayed afterinjection of Cas9+sgRNA-3 had clearly detectable levels ofnondigested PCR product, indicating disruption of the PstIrestriction site by CRISPR-induced indels (Fig. 2D).

Next, we designed and tested synthetic single-strandedoligodeoxynucleotide donor templates (ssODNs) for introducingthe pbx4 p.A131V variant through homology directed repair (HDR)(Hwang et al., 2015). We designed five ssODNs for which thesequences overlap the C392 target site and match the genomicsequence, except at the C392 position, where the ssODNs contain aT residue (corresponding to *C in Fig. 2B; ssODNs listed inTable 1). At the time this work was being done, there was not a clearconsensus regarding the design of oligodeoxynucleotides for HDR,with some studies having success with ssODNs with homologyarms in the 18-26 nucleotide range (Bedell et al., 2012; Hwanget al., 2013b), and others using ssODNs with homology arms in the40-49 nucleotide range (Chen et al., 2011; Hruscha et al., 2013). Wetherefore tested ssODNs with a variety of homology arm lengths. OnessODN had 40 nucleotides matching the genomic sequence on eitherside of the mismatched T and corresponded to the sense strand(ssODN-1). Two oligodeoxynucleotides had 20-nucleotide homologyarms on either side of the base to be changed, with oneoligodeoxynucleotide corresponding to the sense strand (ssODN-2)

Fig. 2. CRISPR engineering enables the generation of a zebrafish pbx4p.A131V variant strain. (A) Schematic of zebrafish Pbx4 protein domains.HD, homeodomain; PBC A/B, conserved Pbx domains. The asterisk indicatesthe polyalanine tract. (B) pbx4 locus for targeting the pbx4 C392/A131 site.The *C nucleotide is nucleotide 392 in the A131 codon. Two adjacent PAMsequences (TGG and AGG) are underlined and the target site for sgRNA-3 isin uppercase. Surrounding sequence is in lowercase. (C) pbx4 guide RNAstested. The middle column shows the genomic region for each guide RNA.The target site is in uppercase. The PAM is in bold underlined uppercase.Surrounding sequence is in lowercase. The right column shows thesequence of the variable region of the sgRNA produced after transcriptionfrom pDR274, which begins with a 5′ GG dinucleotide. A red ‘G’ indicates amismatch with the genomic sequence; a green ‘G’ indicates a match.(D) DNA restriction fragment length polymorphism (RFLP) analysis of thepbx4 target region after injection of sgRNA-2 or sgRNA-3. The wild-typePCR amplicon is 684 bp and is cleaved to 241 bp and 443 bp fragments byPstI. Embryos were injected with either a low dose (300 pg Cas9+12.5 pgsgRNA) or a high dose (600 pg Cas9+25 pg sgRNA) of RNA. In embryosinjected with Cas9+sgRNA-2, only background levels of nondigested PCRproduct are present, similar to those seen in noninjected embryos.Injection of Cas9+sgRNA-3 results in a large proportion of nondigested PCRproduct in all embryos assayed (arrow). (E) DNA RFLP analysis of F1embryos, from CRISPR-Cas9-engineered F0 fish, shows incorporation ofthe C>T mutation in one embryo (asterisk), owing to F0 germline mosaicism.Digestion of the PCR product is dependent on the C>T base change.(F) DNA sequencing chromatogram from heterozygous F1 embryo (asteriskby C) showing precise incorporation of the C>T change. (G) Wild-typesequence around the pbx4 C392 site, with the pbx4 sgRNA target siteunderlined and C392 highlighted. Sequencing of scm14 allele showsprecise C392T/A131V editing.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

and the other to the antisense strand (ssODN-3). Finally, twooligodeoxynucleotides had 25-nucleotide homology arms andcorresponded to the sense (ssODN-4) and antisense (ssODN-5)strands. We injected zebrafish embryos with Cas9 mRNA+sgRNA-3+each ssODN, allowed the embryos to develop for24 h, and then assayed them for the presence of indels (using thePstI digest, as above for the sgRNA assays) and for theintroduction of the C>T change. To detect the specific basechange, a derived cleaved amplified polymorphic sequences(dCAPS) assay (Neff et al., 1998) was designed, such that a newAccI restriction site is createdwhenC392 is changed to a T. Of the fivessODNs tested, two produced detectable levels of the restrictionfragment expected from introduction of the C>T base change insome injected embryos (ssODN-4, 5/12 embryos; ssODN-5, 2/12embryos). The remaining three oligodeoxynucleotides induced nodetectable levels of C>T base change. None of the ssODNs appearedto affect the frequency of induction of indels, with all injectionconditions yielding ∼90% of embryos with some level of indels(similar to that shown in Fig. 2D; data not shown).Based on these results, we then co-injected the pbx4 guide RNA

(sgRNA-3), Cas9 mRNA and the ssODN-4 oligodeoxynucleotideinto one-cell zebrafish embryos and raised these F0 animals toadulthood. We screened a total of 60 adult F0 fish for thosetransmitting germline pbx4 p.A131Vmutations by assaying their F1embryos at 24 hpf for incorporation of the C392T mutation, usingthe same PCR and restriction fragment length polymorphism(RFLP) analysis used in screening the ssODNs (Fig. 2E). Eleven of60 F0 fish yielded clutches of F1 embryos positive for the C392Tchange, irrespective of other indels induced at the pbx4 locus. Fromthese 11 F0 fish, we observed that a range of one to six out of 12 F1embryos was positive for the C392T change (Fig. 2E), yielding an8-50% rate (average 14%) of germline transmission of the mutation.For these positive F1 embryos, we sequenced the genomic regionaround the CRISPR target site and found that two of the 11 F0 fishwere transmitting the C392T change alonewithout any indels in pbx4(9/11 had additional indels). Additional F1 embryos from the two F0fish that were positive for the precise C392T change were then raisedto adulthood and screened for the mutation, using the RFLP assay onF1 adult fin biopsies. The genomic region around the CRISPR targetsite was again sequenced to identify F1 adult fish heterozygous forthe precise desired pbx4 p.A131V variant (Fig. 2F,G).In addition to identifying F0 and F1 fish carrying the precise

C392T change, we identified fish carrying a variety of indels atthe pbx4 target site. The mutations we found, consisting of small(1-39 nucleotides) deletions at the target site, are similar to theCRISPR-Cas9-induced indels previously reported in zebrafish

(Chang et al., 2013; Gagnon et al., 2014; Hwang et al., 2013a,b;Varshney et al., 2015). However, out of 13 different indels weidentified in F1 fish (from 10 different F0 founders) none wereinsertions (even in combination with a deletion). In addition, adisproportionate number were in-frame deletions (9/13 indels).As we have only examined in detail the mutations resulting fromtargeting a single gene with CRISPR-Cas9 and an HDR ssODN, wedo not know if the ssODN affected the type of indels produced, or ifthe indels induced were affected by the particular sgRNA used or bythe target site sequence.

With this approach, we identified F1 fish, from two independentF0 founder fish, transmitting the precise C392T change. F1 fishcarrying the precise change with no other mutations were then bredto wild-type fish to generate a zebrafish strain, pbx4scm14, carryingthe pbx4 p.A131V variant (Fig. 2G). We sequenced the pbx4 genein the pbx4scm14 strain to confirm that there were no other changesintroduced in pbx4 (data not shown).

Upon genotyping adult fish derived from a pbx4scm14/+×pbx4scm14/+

cross, we obtained the expected frequency of pbx4scm14/scm14 fish(7/32 fish; Chi-square test including all genotypes of the cross,P=0.7788), showing that we obtain viable homozygous adults forthe pbx4 p.A131V variant, whereas homozygous null pbx4b557/b557

animals die at ∼5 days postfertilization (dpf) (Pöpperl et al., 2000).The adult viability suggests that the pbx4 p.A131V variant does not,on its own, lead to a significant defect in heart development (as wefurther confirm below).

To test the strength of the pbx4 p.A131V allele, we crossedthe pbx4scm14 strain with the null pbx4b557 strain. From a crossof pbx4scm14/+ fish with pbx4b557/+ fish, we observed thattransheterozygous pbx4scm14/b557 fish survive to adulthood andare present at the expected frequency (10/43 fish; Chi-square testincluding all genotypes of the cross, P=0.2640). Thus, this testdoes not detect any effect of the pbx4 p.A131V allele. We thenused the new pbx4 variant allele strain to test whether the pbx4p.A131V variant functions as a genetic modifier of otherCHD genes.

The pbx4 p.A131V variant enhances loss of the CHDgene hand2To further test the function of the pbx4 p.A131V allele, we nextwanted to additionally perturb the genetic burden of fish carryingthe pbx4 p.A131V allele. In particular, we decided to directly testwhether the pbx4 p.A131V variant functions as a genetic modifierand enhances the phenotype caused by loss of a known CHD gene,hand2. hand2 encodes a basic helix-loop-helix factor that hascritical requirements for embryonic heart development in zebrafish

Table 1. Oligonucleotides used for CRISPR-Cas9 mutagenesis and screening

Gene Forward primer (5′-3′) Reverse primer (5′-3′)

pbx3b sgRNA [phos]TAGGTGTAGCCGGGCCAGAGAA [phos]AAACTTCTCTGGCCCGGCTACApbx4 sgRNA-1 [phos]TAGGGCAGCAGCTGCAGCGGCAGC [phos]AAACGCTGCCGCTGCAGCTGCTGCpbx4 sgRNA-2 [phos]TAGGGCAGCTGCAGCGGCAGCTGG [phos]AAACCCAGCTGCCGCTGCAGCTGCpbx4 sgRNA-3 [phos]TAGGAGCTGCAGCGGCAGCTGG [phos]AAACCCAGCTGCCGCTGCAGCTpbx3b indel assay TGTGAGGCATAACATCACGATTCG GGGAAGAAAGCGCAGAGGAGTpbx4 indel assay TGGCTCTGGTTTCCTTTGCTT CAGCAAACACGACGTATAGCCpbx4 dCAPS assay CTGAGCATCAGAGGCGTGCAG GATGCTGCCATCATTAGGTGATCCTCCAGCTGCGTCTpbx4 ssODN-1 CCAGAGAAGGGTGGAGGATCTGCGGCGGCAGCAGCAGCTGTAGCGG

CAGCTGGAGGATCACCTAATGATGGCAGCATCGAApbx4 ssODN-2 TGCGGCGGCAGCAGCAGCTGTAGCGGCAGCTGGAGGATCACpbx4 ssODN-3 GTGATCCTCCAGCTGCCGCTACAGCTGCTGCTGCCGCCGCApbx4 ssODN-4 GGATCTGCGGCGGCAGCAGCAGCTGTAGCGGCAGCTGGAGGATCACCTAATpbx4 ssODN-5 ATTAGGTGATCCTCCAGCTGCCGCTACAGCTGCTGCTGCCGCCGCAGATCC

5

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

and mice (Srivastava et al., 1997; Yelon et al., 2000), and mutationsin the human HAND2 gene have been associated with CHD(Lu et al., 2016; Shen et al., 2010; Sun et al., 2016; Töpf et al., 2014).In zebrafish, hand2 is required for both cardiomyocyte differentiationand early myocardial morphogenesis (Garavito-Aguilar et al., 2010;Schoenebeck et al., 2007; Trinh et al., 2005; Yelon et al., 2000).Furthermore, our previous work used morpholino knockdowns toshow that hand2 and pbx4 act together in zebrafish early myocardialmorphogenesis (Maves et al., 2009). We therefore crossed our pbx4p.A131V allele strain (pbx4scm14) with the established null mutantstrain, hand2s6 (Yelon et al., 2000). We additionally included thepbx4b557 null allele in these crosses. Fig. 3A shows the genetic

crosses used to obtain embryos of all possible homozygous andheterozygous mutant combinations needed for our analyses.

We first examined early cardiac development in embryos fromthese crosses at 24 hpf, using RNA in situ expression of myl7 toexamine myocardial precursors. For these experiments, embryosfrom all crosses were genotyped for pbx4b557, pbx4scm14 andhand2s6, using tail tissue from post-in situ-hybridized embryos asdescribed in the Materials and Methods. At 24 hpf, wild-typeembryos have formed a heart tube (Fig. 3B; Staudt and Stainier,2012), and pbx4scm14/scm14 embryos also appear to have anormal heart tube (data not shown). As we previously described,pbx4b557/b557 embryos show an abnormally shaped, mediallypositioned myocardium (Fig. 3C; Kao et al., 2015), and hand2s6/s6

embryos show reduced, medial, crescent-shaped myocardialdomains that have not assembled together properly, withdefective fusion of the myocardial precursors at the midline(Fig. 3D; Yelon et al., 2000). We quantified the myocardial fusiondefects in these embryos by measuring the distance between themyl7 domains (Fig. 3H). The hand2s6/s6 myocardial fusion defectis significantly more severe in pbx4b557/b557;hand2s6/s6 embryos,in that the myl7 domains have greater bilateral separation(Fig. 3D,E,H). We found that pbx4scm14/b557;hand2s6/s6 embryosalso show a more severe effect on myocardial fusion than that seenin hand2s6/s6 embryos, and, critically, the defect is also moresevere than that seen in pbx4b557/+;hand2s6/s6 embryos, or inpbx4scm14/scm14;hand2s6/s6 embryos (Fig. 3D-H). These resultsshow that the pbx4 p.A131V variant, combined with a null pbx4allele, increases the severity of, or enhances, the early myocardialmorphogenesis phenotype caused by loss of hand2 function. Weanalyzed the frequencies of the myocardial phenotypes in embryosfrom these crosses (Table 2). We classified embryos as having awild-type heart tube, a medial heart cone (the pbx4b557/b557

phenotype), a medial crescent (the hand2s6/s6 phenotype) or abilateral domain phenotype (the pbx4b557/b557;hand2s6/s6 phenotype).This analysis also shows that the myocardial fusion defect is moresevere in pbx4scm14/b557;hand2s6/s6 embryos than in hand2s6/s6

Fig. 3. The pbx4 p.A131V variant enhances myocardial morphogenesisdefects caused by loss of hand2. (A) Genetic crosses of zebrafish strainsused to obtain embryos for the analyses of pbx4;hand2 mutant embryos. Alladult breeder fish used in crosses 1-3 were ‘siblings’, derived from the sameclutch from a group cross. (B-G) Myocardial markermyl7 expression at 24 hpf.Dorsal views; anterior is up. Animal numbers for phenotypic classes areprovided in Table 2. (B) The heart tube appears normal in pbx4+/+;hand2+/+

embryos. (C) pbx4b557/b557;hand2+/+ embryos have a medial heart cone myl7domain. (D) pbx4+/+;hand2s6/s6 embryos show a crescent-shaped myocardialfusion defect of the myl7 domains. (E) A similar phenotype is observed inpbx4b557/+;hand2s6/s6. (F,G) The myl7 fusion defect is more severe inpbx4b557/b557;hand2s6/s6 (F) and pbx4scm14/b557;hand2s6/s6 (G). (H) Quantitationof fusion defect of myl7 domains in different genetic combinations. myl7distance measurements were made blind to embryo genotypes. The averagesfor the myl7 distances among the different genotypes were compared usingone-way ANOVA, and P-values were corrected for multiple comparisons usingTukey’s test. The boxes extend from the 25th to 75th percentiles, the whiskersare at the minimum and maximum, and the bar within the box representsthe median. (I-N) Myocardial marker myl7 expression at 60 hpf. In I-J, ventralviews; anterior is up. In K-N, anterior views; dorsal is up. Animal numbers forphenotypic classes are provided in Table 3. (I) The heart appears normal inpbx4+/+;hand2+/+ embryos. V, ventricle; A, atrium. (J) pbx4b557/b557;hand2+/+

embryos have dysmorphic hearts with bulges of the ventricular myocardium(arrow). (K) pbx4+/+;hand2s6/s6 embryos show an abnormally shaped, medialmyocardium positioned more caudally between the eyes (asterisks).(L) A similar phenotype is observed in pbx4b557/+;hand2s6/s6. (M,N) Moresevere myl7 bilateral domain phenotypes are observed in pbx4b557/b557;hand2s6/s6 (M) and pbx4scm14/b557;hand2s6/s6 (N) embryos. Scale bars: 50 μm.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

embryos, pbx4b557/+;hand2s6/s6 embryos or pbx4scm14/scm14;hand2s6/s6

embryos (Table 2).We then examined cardiac morphogenesis in embryos from

these crosses at 60 hpf, again using RNA in situ expression ofmyl7.As above, post-in situ-hybridized embryos from all crosses weregenotyped for pbx4b557, pbx4scm14 and hand2s6. At 60 hpf,wild-type embryos have formed both cardiac chambers (Fig. 3I;Staudt and Stainier, 2012), and pbx4scm14/scm14 embryos similarlyshow normal cardiac morphology (data not shown). As wepreviously described, pbx4b557/b557 embryos show an abnormallyshaped myocardium with bulges of the ventricle (Fig. 3J; Kao et al.,2015). hand2s6/s6 embryos also show an abnormally shaped,medial myocardium that is positioned more caudally (Fig. 3K). Inpbx4b557/b557;hand2s6/s6 embryos, themyl7 domains are bilaterallyseparated, as at 24 hpf (Fig. 3L). We analyzed the frequenciesof these 60 hpf myocardial phenotypes in embryos from allgenotypes, classifying embryos as having a wild-type heart, adysmorphic ventricle (the pbx4b557/b557 phenotype), a medial/caudalmyocardium (the hand2s6/s6 phenotype) or a bilateral domainphenotype (the pbx4b557/b557;hand2s6/s6 phenotype; Table 3). Wefound that pbx4scm14/b557;hand2s6/s6 embryos show a more severemyocardial morphogenesis phenotype than that seen in hand2s6/s6

embryos, and, as at 24 hpf, the defects aremore severe than those seenin pbx4b557/+;hand2s6/s6 embryos, or in pbx4scm14/scm14;hand2s6/s6

embryos. (Fig. 3K-N, Table 3). These results show that the pbx4p.A131V variant, combined with a null pbx4 allele, increasesthe severity of, or enhances, the later embryonic myocardialmorphogenesis phenotype caused by loss of hand2 function. Weattempted to address whether the pbx4 p.A131V variant enhancedoutflow tract or later-stage cardiac phenotypes of hand2s6/s6

mutants, but we found that hand2s6/s6 mutant embryos lackedexpression of the outflow tract marker elnb and that hand2s6/s6

mutants showed lethality starting at ∼3 dpf (data not shown), thusprecluding these further analyses.

Taken together, these results demonstrate that the pbx4 p.A131Vvariant functions as a genetic modifier and enhances the myocardialmorphogenesis phenotypes caused by loss of hand2 function.Therefore, even though the pbx4 p.A131V fish are viable anddo not have obvious heart defects in an otherwise wild-typegenetic background, our results support our hypothesis that thepbx4 p.A131V allele can function as a genetic modifier inheart development.

DISCUSSIONHere, we used CRISPR-Cas9 precision genome editing tosuccessfully engineer a zebrafish strain with a PBX3 p.A131Vvariant of unknown significance that was previously identified in acohort of patients with CHDs. We engineered the human variantinto the homologous pbx4 p.A131V site in zebrafish. We used theengineered zebrafish to demonstrate that the pbx4 p.A131V alleleacts as a genetic enhancer of the known CHD gene hand2. Our workunderscores the importance of testing the roles of inherited variantsas genetic modifiers of CHDs. Our study is the first example, ofwhich we are aware, of using precision genome editing in zebrafishto demonstrate a function for a human disease-associated variant ofunknown significance. Our work helps advance our understandingof the complex genetics of CHDs, and also provides an example ofusing genome editing in zebrafish to test the causal roles and geneticinteractions of human disease-associated DNA variants.

Many approaches in zebrafish and mammalian models havepreviously been used to characterize the functions of human heartdisease-associated DNA sequence variants. One study showed thatTa

ble2.G

enotyp

e-ph

enotyp

ean

alys

isof

pbx4

;han

d2mutan

tembryo

sat

24hp

f

hand

2+/+

hand

2s6/+

hand

2s6/s6

Myo

cardial

phen

otyp

epb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557pb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557pb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557

Wild-typ

ehe

arttub

e14

2214

019

2035

3535

017

290

01

00

0

Med

ial

heartc

one

00

017

01

00

043

15

22

30

20

Med

ial

cres

cent

00

00

00

00

00

00

99

121

112

Bilateral

00

00

00

00

00

00

614

615

817

Num

bers

indica

tethetotaln

umbe

rof

embryo

sof

each

geno

type

with

theindica

tedmyo

cardialp

heno

type

clas

s.Sha

dedce

llsindica

tege

notype

s/ph

enotyp

esthat

wereco

mpa

redby

Fishe

r’sex

acttes

t(two-tailed):p

bx4s

cm14/b557;han

d2s6

/s6ve

rsus

pbx4

+/+;han

d2s6

/s6,P

=0.00

17;p

bx4s

cm14/b557;han

d2s6

/s6ve

rsus

pbx4

b557/+;han

d2s6

/s6,P

=0.03

5;pb

x4sc

m14/b557;han

d2s6

/s6ve

rsus

pbx4

scm14/scm

14;han

d2s6

/s6,P

=0.00

19.

7

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

zebrafish embryos could be used to characterize heart functiondefects caused by a human sodium channel gene SCN5A variant,which is strongly associated with human heart disorders such asarrhythmias (Huttner et al., 2013). However, this study employedoverexpression of the human gene in transgenic zebrafish. Zebrafishembryos have often been used to test the functions of potentialhuman disease variants through complementation, in which theorthologous zebrafish gene is typically knocked out throughgenetic mutation or knocked down with antisense morpholinos,and then mRNA injections of a human wild-type or mutant formare used to attempt rescue of a phenotype (Davis et al., 2014).However, owing to the inherent transient and mosaic nature ofthe mRNA injection and overexpression approach, thesecomplementation assays have challenges for examining geneticinteractions. A CHD-associated GATA4 variant has beenfunctionally characterized in a mouse model (Misra et al.,2012), but this was done with conventional embryonic stem celltargeting, which incorporates a targeting vector into the genome.This same GATA4 variant was also functionally characterized inpatient-derived cardiomyocytes induced from pluripotent stemcells, and this analysis provided a deep, systems-level mechanisticunderstanding of how this particular variant disrupts cardiomyocytegene expression and function (Ang et al., 2016). However, suchcell culture models also have challenges for examining geneticinteractions and might not reveal the full effects of a variant on agene’s function in vivo.

Although the development of CRISPR-Cas9 technology hasmade possible the creation of targeted single-nucleotide changesin zebrafish, the process is still inefficient. Our studies supportoptimization of the component reagents for successful single-nucleotide editing. We found that different sgRNAs and ssODNshad very different efficacies. Of the three sgRNAs tested, two failedto induce indels at a level we could detect with an RFLP assay,whereas the third resulted in readily detectable indels in nearly allinjected embryos. The highly effective sgRNA has a 20-nucleotidetargeting region with a single mismatch at the second position,whereas the ineffective sgRNAs had 20-nucleotide targetingregions with two additional mismatched nucleotides at their5′ ends. Although sgRNAs similar to these (two free Gs 5′ to a20-nucleotide protospacer) have been used in zebrafish effectively(Hwang et al., 2013b), most published studies have used sgRNAswith 20-nucleotide protospacers and no additional 5′ mismatchedbases. One- or two-base mismatches at or near the 5′ end of theprotospacer have been shown to be tolerated (Cong et al., 2013;Gagnon et al., 2014; Hwang et al., 2013b), and a guanineimmediately 5′ to the PAM sequence has been shown to correlatepositively with indel frequencies (Farboud and Meyer, 2015;Gagnon et al., 2014). Thus, the high incidence of indels seenwith pbx4 sgRNA-3 might result from a combination of optimalprotospacer length, the presence of only a single mismatch near the5′ end of the protospacer and a G nucleotide in the final position ofthe protospacer before the PAM. Less is known about the design ofhomology-directed repair templates to maximize precise editing,and there is conflicting evidence as to whether ssODNs are moreefficient at introducing single-nucleotide changes than long double-stranded DNA repair templates (Bedell et al., 2012; Boel et al.,2016; Chen et al., 2011; Hoshijima et al., 2016; Hruscha et al.,2013; Hwang et al., 2013b; Irion et al., 2014; Liang et al., 2017;Zhang et al., 2018). Previous studies have achieved template-mediated repair using ssODNs of a similar length to, or longer than,our ssODN-4 (Boel et al., 2016; Hruscha et al., 2013; Hwang et al.,2013b; Liang et al., 2017).Ta

ble3.

Gen

otyp

e-ph

enotyp

ean

alys

isof

pbx4

;han

d2mutan

tembryo

sat

60hp

f

hand

2+/+

hand

2s6/+

hand

2s6/s6

Myo

cardial

phen

otyp

epb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557pb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557pb

x4+/+

pbx4

b557/+

pbx4

scm14/+

pbx4

b557/b557pb

x4sc

m14/scm

14pb

x4sc

m14/b557

Wild-typ

ehe

art

2241

303

3422

5372

471

5253

10

00

00

Dysmorph

icve

ntric

le0

01

360

43

21

528

50

00

00

2

Med

ial/

caud

al0

00

00

00

10

30

026

2729

636

33

Bilateral

00

00

00

00

00

00

11

228

215

Num

bers

indica

tethetotaln

umbe

rof

embryo

sof

each

geno

type

with

theindica

tedmyo

cardialp

heno

type

clas

s.Sha

dedce

llsindica

tege

notype

s/ph

enotyp

esthat

wereco

mpa

redby

Fishe

r’sex

acttes

t(two-tailed):p

bx4s

cm14/b557;han

d2s6

/s6ve

rsus

pbx4

+/+;han

d2s6

/s6,P

=0.00

22;p

bx4s

cm14/b557;han

d2s6

/s6ve

rsus

pbx4

b557/+;han

d2s6

/s6,P

=0.00

48;p

bx4s

cm14/b557;han

d2s6

/s6ve

rsus

pbx4

scm14/scm

14;han

d2s6

/s6,P

=0.00

15.

8

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Our study provides further support for crucial roles for PBX-related TALE-class homeodomain transcription factors in heartdevelopment and CHDs. Pbx1, Pbx2 and Pbx3 proteins and thePbx-related factor Meis1 all contribute to outflow tract developmentin mouse embryos (Chang et al., 2008; Stankunas et al., 2008).Zebrafish Pbx and Meis genes are also needed for proper heartdevelopment (Guerra et al., 2018; Kao et al., 2015; Maves et al.,2009; Paige et al., 2012). In human cardiomyocyte cell culturemodels, DNA binding sites for PBX and MEIS factors have beenidentified as enriched in open chromatin and occurring nearby sitesfor other cardiac transcription factors (He et al., 2011; Paige et al.,2012; Stergachis et al., 2013; Wamstad et al., 2012). Humansequencing studies have found PBX and MEIS gene variantsassociated with CHDs (Arrington et al., 2012; Crowley et al., 2010;Louw et al., 2015; Slavotinek et al., 2017). Notably, a recent studydescribed five patients, each with a CHD and other congenitalanomalies, that all had de novo missense or nonsense sequencevariants in PBX1 (Slavotinek et al., 2017). The missense PBX1variants occurred in or near the homeodomain and likely affectthe transcriptional capabilities of PBX1 (Slavotinek et al., 2017).The PBX3 p.A136V variant that we addressed here is in the moreN-terminal polyalanine tract, which might bind HDAC proteins(Saleh et al., 2000; Choe et al., 2009). The polyalanine tract isadjacent to the PBC domains, which interact with Meis and HDACproteins (Choe et al., 2009; Mann et al., 2009; Saleh et al., 2000).Even though one caveat of our study is that we model the humanPBX3 variant in the zebrafish pbx4 locus, the polyalanine tract ishighly conserved in both PBX3 and Pbx4. As future studies identifyadditional inherited and de novo variants in PBX genes associatedwith CHDs, we will gain a better understanding of which humanPBX genes, and which PBX protein domains, are needed for humanheart development.Our study also provides further support for the oligogenic basis of

CHDs. Previous studies of Pbx genes in mouse heart developmentsupported a multigenic basis for CHDs (Stankunas et al., 2008).Family members carrying the same HAND2 mutation can exhibitdifferent cardiac defects (Sun et al., 2016), supporting a role foradditional genetic changes influencing the severity of CHDsassociated with HAND2 mutation. In order to reveal a contributionof the pbx4 p.A131V variant in zebrafish heart development, we hadto employ an additional null allele of pbx4 as well as homozygousloss of hand2, conditions that are likely to not be present in humanCHD cases. Our efforts underscore the challenges of modeling thefunctions of human disease-associated genetic variants in animalmodels. Mutations that are haploinsufficient in humans can often betoleratedwhen heterozygous inmice (Cassa et al., 2017; Georgi et al.,2013). There are increasing numbers of examples showing that toproduce disease-related phenotypic effects of a specific locus in miceor zebrafish animal models, additional or more severe alleles need tobe utilized (Bradford et al., 2017; Cassa et al., 2017; Distasio et al.,2017; Varga et al., 2018; Zhang et al., 2014). For example, humanpatients homozygous for a variant inCOPB2 exhibitedmicrocephaly,whereas a mouse model required the Copb2 variant allele to becombined with a Copb2 null allele to cause brain malformations(Distasio et al., 2017). Furthermore, we do not yet understand the fullgenetic burden of variants in human CHD cases. A recent study foundevidence for rare inherited variants, in geneswith a known associationwith cardiac malformations and in parent-child trios affected byatrioventricular septal defects, and these particular rare variants weregenerally not observed in control trios (Priest et al., 2016). This andour study support the idea that inherited variants across multiplegenetic loci might contribute to the penetrance and expressivity of

CHDs, and provide further support to the oligogenic inheritance ofCHDs (Fahed et al., 2013; Gelb and Chung, 2014).

Forward genetics in mice was used recently to identify aninteraction between two genes, Sap130 and Pcdha9, in causinghypoplastic left heart syndrome (HLHS) (Liu et al., 2017).Intriguingly, this study also identified one human HLHS patientwith variants in both SAP130 and PCDHA13, a Pcdha9 homolog(Liu et al., 2017). In mice, Sap130 and Pcdha9 might workcombinatorially, each regulating different pathways importantfor proper heart development (Liu et al., 2017). Although wepreviously described pbx4 and hand2 as working together inzebrafish heart development (Maves et al., 2009), it is possible thatthey are also working through parallel pathways. Future studies thatcontinue to take advantage of animal models, as well as humangenetic sequence data, will further enhance our understanding of thecomplex genetics of human diseases such as CHD.

MATERIALS AND METHODSZebrafish husbandryAll experiments involving live zebrafish (Danio rerio) were carried out incompliance with Seattle Children’s Research Institute Institutional AnimalCare and Use Committee guidelines. Zebrafish were raised and staged aspreviously described (Westerfield, 2000). Staging time refers to hpf at28.5°C. The wild-type stock and genetic background used was AB. Thepbx4b557 mutant strain was previously described and is likely a null allele(Pöpperl et al., 2000). pbx4b557 genotyping was performed as previouslydescribed (Kao et al., 2015) using forward primer 5′-ACTCGGCGGACT-CTCGCAAGC-3′ and reverse primer 5′-GGCTCTCGTCGGTGATGG-CCATGATCT-3′ primers. The genotyping PCR product is 128bp;digesting with XbaI yields a 98bp product from the mutant allele.In some cases, pbx4b557 animals were genotyped using a KASP assay(LGC Genomics). Reactions were run and fluorescence was measured in aBio-Rad CFX96, and genotypes were assigned using the Bio-Rad CFXManager software, according to instructions provided by LGC Genomics.Details for ordering the pbx4b557 KASP assay are available upon request.The hand2s6 mutant strain was previously described and is a deletion ofthe hand2 locus (Yelon et al., 2000). For hand2s6 genotyping, we deviseda quantitative PCR assay for the number of copies of the hand2 gene.hand2 was amplified with forward primer 5′-ACCAAAGCGTACTCC-GTCTG-3′ and reverse primer 5′-CAGCGAAGGAATAGCCGTCA-3′.pbx4 was also amplified as a normalization control using forward primer5′-GCCGTTAAAACAGCCGTGG-3′ and reverse primer 5′-GTGTTGC-TGGAGAGTTTGCC-3′. Reactions were run in duplicate for both genesusing the KAPA SYBR FAST kit (KAPA Biosystems KK4600) on aBio-Rad CFX96 machine. The resulting Ct values were averaged, and aΔCt was calculated to distinguish hand2 homozygous, heterozygous andwild-type embryos.

Table 4. Pbx accession numbers

Pbx gene Protein sequence

Human PBX1 NP_002576.1Mouse Pbx1 NP_001278437.1Zebrafish pbx1a NP_571689.1Zebrafish pbx1b NP_001077322.2Human PBX2 NP_002577.2Mouse Pbx2 NP_059491.1Zebrafish pbx2 NP_571687.1Human PBX3 NP_006186.1Mouse Pbx3 NP_058048.1Zebrafish pbx3a XP_005171794.1Zebrafish pbx3b CAQ13247.1Human PBX4 NP_079521.1Mouse Pbx4 NP_001020125.1Zebrafish pbx4 NP_571522.1Fruit fly exd NP_523360.1

9

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Pbx phylogenetic analysisAccession numbers for the sequences aligned in Fig. 1A are provided inTable 4. Sequences were aligned using Clustal Omega (www.EMBL.org).The phylogeny was constructed with PhyML using the following settings:Substitution Model LG; Tree Improvement SPR & NNI; Bootstrapping500 (www.ATGC-montpellier.fr). The phylogenetic tree (Fig. 1B) wasdrawn with TreeDyn (http://phylogeny.lirmm.fr/phylo_cgi/one_task.cgi?task_type=treedyn).

Quantitative reverse transcription PCRTotal RNA was isolated using TriZol (Ambion; Thermo Fisher Scientific)and reverse-transcribed with a SensiFAST cDNA Synthesis kit (BiolineBIO-65053). Primers were designed using Primer-BLAST such that theyeither span an intron or one of the pair spans an exon-exon boundary.Primers are listed in Table 5. Quantitative reverse transcription PCR (qRT-PCR) was carried out using a KAPA SYBR FAST kit (KAPA BiosystemsKK4600) on a Bio-Rad CFX96 machine. For the analysis of Pbx expressionlevels in wild-type embryos at different stages, Ct values for the zebrafishPbx genes were corrected for observed primer efficiencies and normalized toodc1. For the analysis of Pbx gene expression in pbx3bscm8 mutants, Pbxgenes were normalized to eef1a1l1 and ΔΔCt values were calculated for fourwild-type and four mutant replicates, each consisting of pools of eight48 hpf embryos. P-values were calculated with an unpaired Student’s t-testwith Welch’s correction for unequal population standard deviations, usingthe ΔΔCt values. The graph was constructed by log transforming the ΔΔCtvalues. Statistical analysis was performed and the box plot was made inGraphPad Prism 7. The boxes extend from the 25th to 75th percentiles, thewhiskers are at the minimum and maximum, and the bar within the boxrepresents the median.

Generation of mutant zebrafish strains with CRISPR-Cas9Single-guide constructs were made by annealing pairs of oligonucleotides(listed in Table 1) and ligating them into BsaI-digested pDR274 (Hwanget al., 2013a). The single-guide plasmids were digested withDraI, and guideRNA was transcribed with the T7 Maxiscript kit (Ambion; Thermo FisherScientific). Cas9 mRNA was made by transcribing PmeI-digestedpMLM3613 (Hwang et al., 2013a) with the T7 Ultra kit (Ambion;Thermo Fisher Scientific). One-cell-stage zebrafish embryos were injectedwith 600 pg (for pbx3b) or 400 pg (for pbx4) of Cas9 mRNA and 25 pg ofan sgRNA in a volume of 2 nl. To generate the desired single-nucleotidechange in pbx4, 50 pg of an ssODN (Table 1) was co-injected with Cas9mRNA and the sgRNA. Zebrafish were screened for Cas9-generatedmutations (insertions or deletions, ‘indels’) by amplifying with primersflanking the target site (indel assay primers, Table 1) followed by a digest totest for loss of a restriction site just upstream of or overlapping the PAM site(HpyAV for pbx3b, PstI for pbx4). To detect the single-nucleotide change inpbx4, a dCAPS assay was designed (http://helix.wustl.edu/dcaps/; Neffet al., 1998) such that PCR amplification of the mutant allele generated anAccI site ( pbx4 dCAPS assay primers in Table 1). Mutant alleles wereidentified in heterozygous F1 or F2 fish by Sanger sequencing anddeconvolving the chromatograms with Poly Peak Parser (http://yosttools.genetics.utah.edu/PolyPeakParser/).

ImmunoblottingEmbryos were obtained from an incross of pbx3bscm8/+;pbx4b557/+ fish.At 48 hpf, tail tips were cut from the embryos and arrayed in 96-well plates

for genotyping. The remaining portions of the embryos were arrayed inseparate 96-well plates, to which sodium dodecyl sulfate (SDS) samplebuffer was added and stored at −20°C until genotyping was completed, atwhich point the embryo lysates were pooled by genotype. Approximatelyone embryo equivalent was separated by reducing SDS-polyacrylamide gelelectrophoresis and blotted as previously described (Maves et al., 2007).Blots were probed with anti-pan-Pbx (rabbit serum, 1:500; Pöpperl et al.,2000) and with anti-Actin as a loading control (1:500; Clone C4, Millipore).The anti-pan-Pbx antibody was raised and validated against multiplezebrafish Pbx proteins (Maves et al., 2007; Pöpperl et al., 2000; Waskiewiczet al., 2002). Infrared dye-labeled secondary antibodies (Rockland) werevisualized using a LI-COR Odyssey infrared scanner.

Whole-mount RNA in situ hybridization andmeasurement analysisThe following cDNA probes were used: myl7 (Yelon et al., 1999) and elnb(Miao et al., 2007). Whole-mount in situ hybridization colorimetric andfluorescent in situ staining was performed as previously described (Maveset al., 2007; Talbot et al., 2010), with the following modifications. Embryosat 60 hpf were depigmented in 1 part 0.1% KOH (vol.): 1 part 1× PBS-0.1%Tween (vol.): 0.1 part 30% hydrogen peroxide (vol.) for 2 h at roomtemperature with gentle agitation. Hybridizations for both colorimetric andfluorescent in situ experiments were performed in hybridization bufferwith 5% dextran sulfate. Following staining, tail clips from post-in situ-hybridized embryos were lysed and genotyped for pbx3bscm8, pbx4b557,pbx4scm14 and hand2s6, as above. Imaging was performed on a Leica SP5confocal microscope with a 40× water immersion objective.

Formyl7measurements at 24 hpf, we imaged genotyped embryos. Samplesizes were not pre-determined, and no animals were excluded from theanalysis. Embryos stained for myl7 were individually imaged at a consistentmagnification on a stereomicroscope. ImageJ (https://imagej.nih.gov/ij/) wasused tomeasure the area ofmyl7 expression and the distance between bilateraldomains of expression, expressed in pixels. The investigator making themeasurements was blinded to the sample genotype. For embryos withseparate bilateral myl7 domains, three distance measurements were made foreach embryo: between the medial edges of expression at the anterior extent ofexpression, between the medial edges of expression at the posterior extent ofexpression, and at the approximate midpoint between the anterior andposterior measurements. These three values were then averaged. For embryoswith a single midline domain of myl7 expression, a distance value of 0 wasassigned. For embryos with a crescent-shaped domain, the posterior distancemeasurement was 0 and the middle and anterior measurements were made asfor separate bilateral domains. The averages for themyl7 distances among thedifferent genotypes were compared using one-way ANOVA, and P-valueswere corrected for multiple comparisons using Tukey’s test. ANOVA wasperformed and the box plots were made in GraphPad Prism 7. The boxesextend from the 25th to 75th percentiles, the whiskers are at the minimum andmaximum, and the bar within the box represents the median.

AcknowledgementsWe thank Fred Keeley, Heike Popperl, and Debbie Yelon for generously providingreagents; Stephen Ekker and Cecilia Moens for advice on CRISPR editing inzebrafish; and Jerry Ament, Dante D’India and the SCRI Aquatics Facility staff forexpert zebrafish maintenance.

Competing interestsThe authors declare no competing or financial interests.

Table 5. Oligonucleotides for qRT-PCR

Gene Forward primer (5′-3′) Reverse primer (5′-3′)

pbx1a ACGAAAAAGGAGAAACTTCAACAAG AACCAGTTGGATACCTGTGAGpbx1b GAAAACATGCGCTCAACTGCC GAGCTCCACGGATACTCAACApbx2 AAACCGGCCTTATTCAGCGT CAATCGTACGAGCTGTGGGTpbx3a GACGCAACTTCAGCAAGCAG GTTCGACACCTGTGAAACCGpbx3b TATCCCAGGGGGTGGGAAC TGCAGTCACGGCTGTCTTAGpbx4 CCGTTAAAACAGCCGTGGAC GTGTAGCACGGTGCAGGATAodc1 TCAGCCTCCGATGAAGAAGAGG TACATGCGCTCATCGGGCTTeef1a1l1 AAAGGGGAGCAGCAGCTGAG TGTGGGTCTTTTCCTTTCCCATGA

10

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Author contributionsConceptualization: G.H.F., L.M.; Methodology: G.H.F., K.I., D.P., L.M.; Validation:G.H.F., L.M.; Formal analysis: G.H.F., L.M.; Investigation: G.H.F., K.I., D.P., L.M.;Writing - original draft: G.H.F., L.M.;Writing - review & editing: G.H.F., K.I., D.P., L.M.;Visualization: G.H.F., L.M.; Supervision: L.M.; Project administration: L.M.; Fundingacquisition: K.I., L.M.

FundingThis work was supported by the American Heart Association (14BGIA18190004to L.M.), the Saving Tiny Hearts Society (L.M.), the Myocardial RegenerationInitiative at Seattle Children’s Research Institute, and the University of WashingtonMary Gates Research Scholarship Program (K.I.).

ReferencesAng, Y.-S., Rivas, R. N., Ribeiro, A. J. S., Srivas, R., Rivera, J., Stone, N. R.,Pratt, K., Mohamed, T. M. A., Fu, J.-D., Spencer, C. I. et al. (2016). Diseasemodel of GATA4 mutation reveals transcription factor cooperativity in humancardiogenesis. Cell 167, 1734-1749.e22.

Arno, G., Agrawal, S. A., Eblimit, A., Bellingham, J., Xu, M., Wang, F.,Chakarova, C., Parfitt, D. A., Lane, A., Burgoyne, T. et al. (2016). Mutations inREEP6 cause autosomal-recessive retinitis pigmentosa. Am. J. Hum. Genet.99, 1305-1315.

Arrington, C. B., Dowse, B. R., Bleyl, S. B. and Bowles, N. E. (2012).Non-synonymous variants in pre-B cell leukemia homeobox (PBX) genes areassociated with congenital heart defects. Eur. J. Med. Genet. 55, 235-237.

Bedell, V. M., Wang, Y., Campbell, J. M., Poshusta, T. L., Starker, C. G., Krug , II,R. G., Tan, W., Penheiter, S. G., Ma, A. C., Leung, A. Y. H. et al. (2012). In vivogenome editing using a high-efficiency TALEN system. Nature 491, 114-118.

Blackburn, P. R., Campbell, J. M., Clark, K. J. and Ekker, S. C. (2013). TheCRISPR system–keeping zebrafish gene targeting fresh. Zebrafish 10, 116-118.

Boel, A., Steyaert, W., De Rocker, N., Menten, B., Callewaert, B., De Paepe, A.,Coucke, P. and Willaert, A. (2016). BATCH-GE: batch analysis of Next-Generation Sequencing data for genome editing assessment. Sci. Rep. 6, 30330.

Bradford, Y. M., Toro, S., Ramachandran, S., Ruzicka, L., Howe, D. G., Eagle, A.,Kalita, P., Martin, R., Taylor Moxon, S. A., Schaper, K. et al. (2017). Zebrafishmodels of human disease: gaining insight into human disease at ZFIN. ILAR J.58, 4-16.

Capellini, T. D., Di Giacomo, G., Salsi, V., Brendolan, A., Ferretti, E.,Srivastava, D., Zappavigna, V. and Selleri, L. (2006). Pbx1/Pbx2 requirementfor distal limb patterning is mediated by the hierarchical control of Hox gene spatialdistribution and Shh expression. Development 133, 2263-2273.

Cassa, C. A., Weghorn, D., Balick, D. J., Jordan, D. M., Nusinow, D.,Samocha, K. E., O’Donnell-Luria, A., MacArthur, D. G., Daly, M. J.,Beier, D. R. et al. (2017). Estimating the selective effects of heterozygousprotein-truncating variants from human exome data. Nat. Genet. 49, 806-810.

Cerda-Esteban, N. and Spagnoli, F. M. (2014). Glimpse into Hox and taleregulation of cell differentiation and reprogramming. Dev. Dyn. 243, 76-87.

Chang, C. P. and Bruneau, B. (2011). Epigenetics and cardiovasculardevelopment. Annu. Rev. Physiol. 74, 13.1-13.28.

Chang, C.-P., Stankunas, K., Shang, C., Kao, S.-C., Twu, K. Y. and Cleary, M. L.(2008). Pbx1 functions in distinct regulatory networks to pattern the great arteriesand cardiac outflow tract. Development. 135, 3577-3586.

Chang, N., Sun, C., Gao, L., Zhu, D., Xu, X., Zhu, X., Xiong, J.-W. and Xi, J. J.(2013). Genome editing with RNA-guided Cas9 nuclease in Zebrafish embryos.Cell Res. 23, 465-472.

Chen, F., Pruett-Miller, S. M., Huang, Y., Gjoka, M., Duda, K., Taunton, J.,Collingwood, T. N., Frodin, M. and Davis, G. D. (2011). High-frequency genomeediting using ssDNA oligonucleotides with zinc-finger nucleases. Nat. Methods8, 753-755.

Choe, S.-K., Lu, P., Nakamura, M., Lee, J. and Sagerstrom, C. G. (2009).Meis cofactors control HDAC and CBP accessibility at Hox-regulated promotersduring zebrafish embryogenesis. Dev. Cell 17, 561-567.

Cong, L., Ran, F. A., Cox, D., Lin, S., Barretto, R., Habib, N., Hsu, P. D., Wu, X.,Jiang, W., Marraffini, L. A. et al. (2013). Multiplex genome engineering usingCRISPR/Cas systems. Science 339, 819-823.

Cox, T. C. (2015). Utility and limitations of animal models for the functional validationof human sequence variants. Mol. Genet. Genomic Med. 3, 375-382.

Crowley, M. A., Conlin, L. K., Zackai, E. H., Deardorff, M. A., Thiel, B. D. andSpinner, N. B. (2010). Further evidence for the possible role of MEIS2 in thedevelopment of cleft palate and cardiac septum. Am. J. Med. Genet. A 152A,1326-1327.

Davis, E. E., Frangakis, S. and Katsanis, N. (2014). Interpreting human geneticvariation with in vivo zebrafish assays. Biochim. Biophys. Acta 1842, 1960-1970.

DiStasio, A., Driver, A., Sund, K., Donlin, M., Muraleedharan, R. M., Pooya, S.,Kline-Fath, B., Kaufman, K. M., Prows, C. A., Schorry, E. et al. (2017). Copb2 isessential for embryogenesis and hypomorphic mutations cause humanmicrocephaly. Hum. Mol. Genet. 26, 4836-4848.

Doudna, J. A. and Charpentier, E. (2014). Genome editing. The new frontier ofgenome engineering with CRISPR-Cas9. Science 346, 1258096.

Elliott, D. A., Kirk, E. P., Yeoh, T., Chandar, S., McKenzie, F., Taylor, P.,Grossfeld, P., Fatkin, D., Jones, O., Hayes, P. et al. (2003). Cardiac homeoboxgene NKX2-5 mutations and congenital heart disease: associations with atrialseptal defect and hypoplastic left heart syndrome. J. Am. Coll. Cardiol. 41,2072-2076.

Evans, S. M., Yelon, D., Conlon, F. L. and Kirby, M. L. (2010). Myocardial lineagedevelopment. Circ. Res. 107, 1428-1444.

Fahed, A. C., Gelb, B. D., Seidman, J. G. and Seidman, C. E. (2013). Genetics ofcongenital heart disease: the glass half empty. Circ. Res. 112, 707-720.

Farboud, B. and Meyer, B. J. (2015). Dramatic enhancement of genome editing byCRISPR/Cas9 through improved guide RNA design. Genetics 199, 959-971.

Ferretti, E., Li, B., Zewdu, R., Wells, V., Hebert, J. M., Karner, C.,Anderson, M. J., Williams, T., Dixon, J., Dixon, M. J. et al. (2011).A conserved Pbx-Wnt-p63-Irf6 regulatory module controls face morphogenesisby promoting epithelial apoptosis. Dev. Cell 21, 627-641.

Gagnon, J. A., Valen, E., Thyme, S. B., Huang, P., Ahkmetova, L., Pauli, A.,Montague, T. G., Zimmerman, S., Richter, C. and Schier, A. F. (2014). Efficientmutagenesis by Cas9 protein-mediated oligonucleotide insertion and large-scaleassessment of single-guide RNAs. PLoS ONE 9, e98186.

Garavito-Aguilar, Z. V., Riley, H. E. and Yelon, D. (2010). Hand2 ensures anappropriate environment for cardiac fusion by limiting Fibronectin function.Development 137, 3215-3220.

Gelb, B. D. and Chung, W. K. (2014). Complex genetics and the etiology of humancongenital heart disease. Cold Spring Harb. Perspect. Med. 4, a013953.

Georgi, B., Voight, B. F. and Bucan, M. (2013). From mouse to human:evolutionary genomics analysis of human orthologs of essential genes. PLoSGenet. 9, e1003484.

Grimes, A. C., Stadt, H. A., Shepherd, I. T. and Kirby, M. L. (2006). Solving anenigma: arterial pole development in the zebrafish heart. Dev. Biol. 290, 265-276.

Guerra, A., Germano, R. F., Stone, O., Arnaout, R., Guenther, S., Ahuja, S.,Uribe, V., Vanhollebeke, B., Stainier, D. Y. and Reischauer, S. (2018). Distinctmyocardial lineages break atrial symmetry during cardiogenesis in zebrafish.eLife 7, e32833.

He, A., Kong, S.W., Ma, Q. and Pu,W. T. (2011). Co-occupancy bymultiple cardiactranscription factors identifies transcriptional enhancers active in heart. Proc. Natl.Acad. Sci. USA 108, 5632-5637.

Homsy, J., Zaidi, S., Shen, Y., Ware, J. S., Samocha, K. E., Karczewski, K. J.,DePalma, S. R., McKean, D., Wakimoto, H., Gorham, J. et al. (2015). De novomutations in congenital heart disease with neurodevelopmental and othercongenital anomalies. Science 350, 1262-1266.

Hoshijima, K., Jurynec, M. J. and Grunwald, D. J. (2016). Precise editing of thezebrafish genome made simple and efficient. Dev. Cell 36, 654-667.

Hruscha, A., Krawitz, P., Rechenberg, A., Heinrich, V., Hecht, J., Haass, C. andSchmid, B. (2013). Efficient CRISPR/Cas9 genome editing with low off-targeteffects in zebrafish. Development 140, 4982-4987.

Huttner, I. G., Trivedi, G., Jacoby, A., Mann, S. A., Vandenberg, J. I. andFatkin, D. (2013). A transgenic zebrafish model of a human cardiac sodiumchannel mutation exhibits bradycardia, conduction-system abnormalities andearly death. J. Mol. Cell. Cardiol. 61, 123-132.

Hwang, W. Y., Fu, Y., Reyon, D., Maeder, M. L., Tsai, S. Q., Sander, J. D.,Peterson, R. T., Yeh, J.-R. J. and Joung, J. K. (2013a). Efficient genome editingin zebrafish using a CRISPR-Cas system. Nat. Biotech. 31, 227-229.

Hwang, W. Y., Fu, Y., Reyon, D., Maeder, M. L., Kaini, P., Sander, J. D.,Joung, J. K., Peterson, R. T. and Yeh, J.-R. J. (2013b). Heritable and precisezebrafish genome editing using a CRISPR-Cas system. PLoS ONE 8, e68708.

Hwang,W. Y., Fu, Y., Reyon, D., Gonzales, A. P.W., Joung, J. K. and Yeh, J.-R. J.(2015). Targeted mutagenesis in zebrafish using CRISPR RNA-guidednucleases. Methods Mol. Biol. 1311, 317-334.

Irion, U., Krauss, J. and Nusslein-Volhard, C. (2014). Precise and efficientgenome editing in zebrafish using the CRISPR/Cas9 system. Development141, 4827-4830.

Jin, S. C., Homsy, J., Zaidi, S., Lu, Q., Morton, S., DePalma, S. R., Zeng, X.,Qi, H., Chang, W., Sierant, M. C. et al. (2017). Contribution of rare inherited andde novo variants in 2,871 congenital heart disease probands. Nat. Genet.49, 1593-1601.

Kao, R. M., Rurik, J. G., Farr, G. H., III, Dong, X. R., Majesky, M. W. and Maves, L.(2015). Pbx4 is required for the temporal onset of zebrafish myocardialdifferentiation. J. Dev. Biol. 3, 93-111.

Lek, M., Karczewski, K. J., Minikel, E. V., Samocha, K. E., Banks, E., Fennell, T.,O’Donnell-Luria, A. H., Ware, J. S., Hill, A. J., Cummings, B. B. et al. (2016).Analysis of protein-coding genetic variation in 60,706 humans. Nature536, 285-291.

Liang, X., Potter, J., Kumar, S., Ravinder, N. and Chesnut, J. D. (2017).Enhanced CRISPR/Cas9-mediated precise genome editing by improveddesign and delivery of gRNA, Cas9 nuclease, and donor DNA. J. Biotechnol.241, 136-146.

Lin,X., Pelletier, S.,Gingras,S., Rigaud,S.,Maine,C. J.,Marquardt, K., Dai, Y. D.,Sauer, K., Rodriguez, A. R., Martin, G. et al. (2016). CRISPR-Cas9-mediatedmodification of the NOD mouse genome with Ptpn22R619W mutation increasesautoimmune diabetes. Diabetes 65, 2134-2138.

11

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms

Liu, X., Yagi, H., Saeed, S., Bais, A. S., Gabriel, G. C., Chen, Z., Peterson, K. A.,Li, Y., Schwartz, M. C., Reynolds, W. T. et al. (2017). The complex genetics ofhypoplastic left heart syndrome. Nat. Genet. 49, 1152-1159.

Louw, J. J., Corveleyn, A., Jia, Y., Hens, G., Gewillig, M. and Devriendt, K.(2015). MEIS2 involvement in cardiac development, cleft palate, and intellectualdisability. Am. J. Med. Genet. A 167A, 1142-1146.

Lu, C.-X., Gong, H.-R., Liu, X.-Y., Wang, J., Zhao, C.-M., Huang, R.-T., Xue, S.and Yang, Y.-Q. (2016). A novel HAND2 loss-of-function mutation responsible fortetralogy of Fallot. Int. J. Mol. Med. 37, 445-451.

Mann, R. S., Lelli, K. M. and Joshi, R. (2009). Hox specificity unique roles forcofactors and collaborators. Curr. Top. Dev. Biol. 88, 63-101.

Maves, L., Waskiewicz, A. J., Paul, B., Cao, Y., Tyler, A., Moens, C. B. andTapscott, S. J. (2007). Pbx homeodomain proteins direct Myod activity topromote fast-muscle differentiation. Development 134, 3371-3382.

Maves, L., Tyler, A., Moens, C. B. and Tapscott, S. J. (2009). Pbx acts with Hand2in early myocardial differentiation. Dev. Biol. 333, 409-418.

McCulley, D. J. and Black, B. L. (2012). Transcription factor pathways andcongenital heart disease. Curr. Top. Dev. Biol. 100, 253-277.

McElhinney, D. B., Geiger, E., Blinder, J., Benson, D. W. and Goldmuntz, E.(2003). NKX2.5 mutations in patients with congenital heart disease. J. Am. Coll.Cardiol. 42, 1650-1655.

Miao, M., Bruce, A. E. E., Bhanji, T., Davis, E. C. and Keeley, F. W. (2007).Differential expression of two tropoelastin genes in zebrafish. Matrix Biol. 26,115-124.

Misra, C., Sachan, N., McNally, C. R., Koenig, S. N., Nichols, H. A., Guggilam, A.,Lucchesi, P. A., Pu, W. T., Srivastava, D. and Garg, V. (2012). Congenital heartdisease-causing Gata4 mutation displays functional deficits in vivo. PLoS Genet.8, e1002690.

Moens, C. B. and Selleri, L. (2006). Hox cofactors in vertebrate development. Dev.Biol. 291, 193-206.

Neff, M. M., Neff, J. D., Chory, J. and Pepper, A. E. (1998). dCAPS, a simpletechnique for the genetic analysis of single nucleotide polymorphisms:experimental applications in Arabidopsis thaliana genetics. Plant J. 14, 387-392.

Olson, E. N. (2006). Gene regulatory networks in the evolution and development ofthe heart. Science 313, 1922-1927.

Paige, S. L., Thomas, S., Stoick-Cooper, C. L., Wang, H., Maves, L., Sandstrom,R., Pabon, L., Reinecke, H., Pratt, G., Keller, G. et al. (2012). A temporalchromatin signature in human embryonic stem cells identifies regulators of cardiacdevelopment. Cell 151, 221-232.

Peng, Y., Clark, K. J., Campbell, J. M., Panetta, M. R., Guo, Y. and Ekker, S. C.(2014). Making designer mutants in model organisms. Development 141,4042-4054.

Popperl, H., Rikhof, H., Cheng, H., Haffter, P., Kimmel, C. B. and Moens, C. B.(2000). Lazarus is a novel pbx gene that globally mediates hox gene function inzebrafish. Mol. Cell 6, 255-267.

Priest, J. R., Osoegawa, K., Mohammed, N., Nanda, V., Kundu, R., Schultz, K.,Lammer, E. J, Girirajan, S., Scheetz, T., Waggott, D. et al. (2016). De novo andrare variants at multiple loci support the oligogenic origins of atrioventricular septalheart defects. PLoS Genet. 12, e1005963.

Prior, H., Jawad, A. K., MacConnachie, L. and Beg, A. A. (2017). Highly efficient,rapid and Co-CRISPR-independent genome editing in Caenorhabditis elegans.G3 (Bethesda) 7, 3693-3698.

Richards, S., Aziz, N., Bale, S., Bick, D., Das, S., Gastier-Foster, J., Grody,W. W., Hegde, M., Lyon, E., Spector, E. et al. (2015). Standards and guidelinesfor the interpretation of sequence variants: a joint consensus recommendation ofthe American College of Medical Genetics and Genomics and the Association forMolecular Pathology. Genet. Med. 17, 405-423.

Ruzicka, L., Bradford, Y. M., Frazer, K., Howe, D. G., Paddock, H.,Ramachandran, S., Singer, A., Toro, S., Van Slyke, C. E., Eagle, A. E. et al.(2015). ZFIN, The zebrafish model organism database: updates and newdirections. Genesis 53, 498-509.

Saleh, M., Rambaldi, I., Yang, X.-J. and Featherstone, M. S. (2000). Cell signalingswitches HOX-PBX complexes from repressors to activators of transcriptionmediated by histone deacetylases and histone acetyltransferases.Mol. Cell. Biol.20, 8623-8633.

Schoenebeck, J. J., Keegan, B. R. and Yelon, D. (2007). Vessel and bloodspecification override cardiac potential in anterior mesoderm. Dev. Cell 13,254-267.

Shah, A. N. and Moens, C. B. (2016). Approaching perfection: new developmentsin zebrafish genome engineering. Dev. Cell 36, 595-596.

Shen, L., Li, X. F., Shen, A. D., Wang, Q., Liu, C. X., Guo, Y. J., Song, Z. J. and Li,Z. Z. (2010). Transcription factor HAND2 mutations in sporadic Chinese patientswith congenital heart disease. Chin. Med. J. 123, 1623-1627.

Sifrim, A., Hitz, M.-P., Wilsdon, A., Breckpot, J., Turki, S. H. A., Thienpont, B.,McRae, J., Fitzgerald, T. W., Singh, T., Swaminathan, G. J. et al. (2016).Distinct genetic architectures for syndromic and nonsyndromic congenital heartdefects identified by exome sequencing. Nat. Genet. 48, 1060-1065.

Slavotinek, A., Risolino, M., Losa, M., Cho, M. T., Monaghan, K. G.,Schneidman-Duhovny, D., Parisotto, S., Herkert, J. C., Stegmann, A. P. A.,

Miller, K. et al. (2017). De novo, deleterious sequence variants that alter thetranscriptional activity of the homeoprotein PBX1 are associated with intellectualdisability and pleiotropic developmental defects. Hum. Mol. Genet. 26,4849-4860.

Srivastava, D., Thomas, T., Lin, Q., Kirby, M. L., Brown, D. and Olson, E. N.(1997). Regulation of cardiac mesodermal and neural crest development by thebHLH transcription factor, dHAND. Nat. Genet. 16, 154-160.

Stallmeyer, B., Fenge, H., Nowak-Gottl, U. and Schulze-Bahr, E. (2010).Mutational spectrum in the cardiac transcription factor gene NKX2.5 (CSX)associated with congenital heart disease. Clin. Genet. 78, 533-540.

Stankunas, K., Shang, C., Twu, K. Y., Kao, S.-C., Jenkins, N. A., Copeland, N. G.,Sanyal, M., Selleri, L., Cleary, M. L. and Chang, C.-P. (2008). Pbx/Meisdeficiencies demonstrate multigenetic origins of congenital heart disease. Circ.Res. 103, 702-709.

Starita, L. M., Ahituv, N., Dunham, M. J., Kitzman, J. O., Roth, F. P., Seelig, G.,Shendure, J. and Fowler, D. M. (2017). Variant interpretation: functional assaysto the rescue. Am. J. Hum. Genet. 101, 315-325.

Staudt, D. and Stainier, D. (2012). Uncovering the molecular and cellularmechanisms of heart development using the zebrafish. Annu. Rev. Genet. 46,397-418.

Stergachis, A. B., Neph, S., Reynolds, A., Humbert, R., Miller, B., Paige, S. L.,Vernot, B., Cheng, J. B., Thurman, R. E., Sandstrom, R. et al. (2013).Developmental fate and cellular maturity encoded in human regulatory DNAlandscapes. Cell 154, 888-903.

Sun, Y.-M., Wang, J., Qiu, X.-B., Yuan, F., Li, R.-G., Xu, Y.-J., Qu, X.-K., Shi, H.-Y.,Hou, X.-M., Huang, R.-T. et al. (2016). A HAND2 loss-of-function mutationcauses familial ventricular septal defect and pulmonary stenosis. G3 (Bethesda)6, 987-992.

Talbot, J. C., Johnson, S. L. and Kimmel, C. B. (2010). hand2 and Dlx genesspecify dorsal, intermediate and ventral domains within zebrafish pharyngealarches. Development 137, 2507-2517.

Topf, A., Griffin, H. R., Glen, E., Soemedi, R., Brown, D. L., Hall, D., Rahman,T. J., Eloranta, J. J., Jungst, C., Stuart, A. G. et al. (2014). Functionallysignificant, rare transcription factor variants in tetralogy of Fallot. PLoS ONE 9,e95453.

Trinh, L. A., Yelon, D. and Stainier, D. Y. R. (2005). Hand2 regulates epithelialformation during myocardial differentiation. Curr. Biol. 15, 441-446.

Varga, M., Ralbovszki, D., Balogh, E., Hamar, R., Keszthelyi, M. and Tory, K.(2018). Zebrafish models of rare hereditary pediatric diseases. Diseases 6, 43.

Varshney, G. K., Pei, W., LaFave, M. C., Idol, J., Xu, L., Gallardo, V., Carrington,B., Bishop, K., Jones, M. P., Li, M. et al. (2015). High-throughput genetargeting and phenotyping in zebrafish using CRISPR/Cas9. Genome Res. 25,1030-1042.

Wamstad, J. A., Alexander, J. M., Truty, R. M., Shrikumar, A., Li, F., Eilertson,K. E., Ding, H., Wylie, J. N., Pico, A. R., Capra, J. A. et al. (2012). Dynamic andcoordinated epigenetic regulation of developmental transitions in the cardiaclineage. Cell 151, 206-220.

Waskiewicz, A. J., Rikhof, H. A. and Moens, C. B. (2002). Eliminating zebrafishpbx proteins reveals a hindbrain ground state. Dev. Cell 3, 723-733.

Westerfield, M. (2000). The Zebrafish Book. A Guide for the Laboratory Use ofZebrafish (Danio rerio), 4th edn. Eugene: Univ. of Oregon Press.

Wright, C. F., Fitzgerald, T. W., Jones, W. D., Clayton, S., McRae, J. F., vanKogelenberg, M., King, D. A., Ambridge, K., Barrett, D. M., Bayzetinova, T.et al. (2015). Genetic diagnosis of developmental disorders in the DDD study: ascalable analysis of genome-wide research data. Lancet 385, 1305-1314.

Yan, L., Yang, M., Guo, H., Yang, L., Wu, J., Li, R., Liu, P., Lian, Y., Zheng, X.,Yan, J. et al. (2013). Single-cell RNA-Seq profiling of human preimplantationembryos and embryonic stem cells. Nat. Struct. Mol. Biol. 20, 1131-1139.

Yelon, D., Horne, S. A. and Stainier, D. Y. (1999). Restricted expression of cardiacmyosin genes reveals regulated aspects of heart tube assembly in zebrafish.Dev.Biol. 214, 23-37.

Yelon, D., Ticho, B., Halpern, M. E., Ruvinsky, I., Ho, R. K., Silver, L. M. andStainier, D. Y. (2000). The bHLH transcription factor hand2 plays parallel roles inzebrafish heart and pectoral fin development. Development. 127, 2573-2582.

Zaidi, S. and Brueckner, M. (2017). Genetics and genomics of congenital heartdisease. Circ. Res. 120, 923-940.

Zaidi, S., Choi, M., Wakimoto, H., Ma, L., Jiang, J., Overton, J. D., Romano-Adesman, A., Bjornson, R. D., Breitbart, R. E., Brown, K. K. et al. (2013). Denovo mutations in histone-modifying genes in congenital heart disease. Nature498, 220-223.

Zhang, Y., Ear, J., Yang, Z., Morimoto, K., Zhang, B. and Lin, S. (2014). Defects ofprotein production in erythroid cells revealed in a zebrafish Diamond-Blackfananemia model for mutation in RPS19. Cell Death Dis. 5, e1352.

Zhang, Y., Qin, W., Lu, X., Xu, J., Huang, H., Bai, H., Li, S. and Lin, S. (2017).Programmable base editing of zebrafish genome using a modified CRISPR-Cas9system. Nat. Commun. 8, 118.

Zhang, Y., Zhang, Z. andGe,W. (2018). An efficient platform for generating somaticpoint mutations with germline transmission in the zebrafish by CRISPR/Cas9-mediated gene editing. J. Biol. Chem. 293, 6611-6622.

12

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035972. doi:10.1242/dmm.035972

Disea

seModels&Mechan

isms