Setd5 is required in cardiopharyngeal mesoderm for heart development and its haploinsufficiency is associated with outflow tract defects in mouse

Summary Congenital heart defects are a feature of several genetic haploinsufficiency syndromes, often involving transcriptional regulators. One property of haploinsufficient genes is their propensity for network interactions at the gene or protein level. In this article we took advantage of an online dataset of high throughput screening of mutations that are embryonic lethal in mice. Our aim was to identify new genes where the loss of function caused cardiovascular phenotypes resembling the 22q11.2 deletion syndrome models, that is, heterozygous and homozygous loss of Tbx1. One gene with a potentially haploinsufficient phenotype was identified, Setd5, thought to be involved in chromatin modification. We found murine Setd5 haploinsufficiency to be associated with double outlet right ventricle and perimembranous ventricular septal defect, although no genetic interaction with Tbx1 was detected. Conditional mutagenesis revealed that Setd5 was required in cardiopharyngeal mesoderm for progression of the heart tube through the ballooning stage to create a four‐chambered heart.

genes is their propensity for network interactions at the gene or protein level. In this article we took advantage of an online dataset of high throughput screening of mutations that are embryonic lethal in mice. Our aim was to identify new genes where the loss of function caused cardiovascular phenotypes resembling the 22q11.2 deletion syndrome models, that is, heterozygous and homozygous loss of Tbx1. One gene with a potentially haploinsufficient phenotype was identified, Setd5, thought to be involved in chromatin modification. We found murine Setd5 haploinsufficiency to be associated with double outlet right ventricle and perimembranous ventricular septal defect, although no genetic interaction with Tbx1 was detected. Conditional mutagenesis revealed that Setd5 was required in cardiopharyngeal mesoderm for progression of the heart tube through the ballooning stage to create a four-chambered heart.

K E Y W O R D S
birth defects, early development, heart, mesoderm 1 | INTRODUCTION Syndromic congenital heart disease (CHD) encompasses several haploinsufficiency conditions. Haploinsufficient genes may be present at a "hub" of interacting networks (Huang, Lee, Marcotte, & Hurles, 2010), and phenotypic similarity has been used in human genetics to cluster syndromes based on the premise that these groupings inform biological relationships of the genes involved that is, the underlying genes may function within the same developmental "module" (Oti & Brunner, 2007). The candidate gene for 22q11.2 deletion syndrome (22q11DS) is TBX1, and is haploinsufficient in mice (Jerome & Papaioannou, 2001;Merscher et al., 2001). In mice, Tbx1 heterozygotes present with hypo-or aplasia of the fourth PAA at embryonic day (E)10.5 which gives rise to great vessel defects such as interrupted aortic arch type B and aberrant right subclavian artery at later stages. Other cardiovascular defects include double outlet right ventricle (DORV) and ventricular septal defect (VSD) (Lindsay et al., 2001;Merscher et al., 2001).
An initiative entitled Deciphering the Mechanisms of Developmental Disorders (DMDD) aimed to take advantage of embryonic lethal mouse lines containing insertions that initially offer a gene-trap-like reporter allele with heterozygous or hypomorphic effect (https://dmdd.org.uk/; Mohun et al., 2013). Upon Flpe recombination, this converts to a conditional allele offering the potential for tissue-specific studies of heart and brain development. Thus, the DMDD browser provides an online database of mutants where standard phenotype ontology terms can be used to search for genes with phenotypes related to known birth defect syndromes. The database contains high-resolution episcopic microscopic (HREM) images of embryos at E9.5 and E14.5 allowing local rephenotyping.
Querying the online browser with key phenotype ontology terms that define the Tbx1 haploinsufficient phenotype such as "DORV" and "VSD" generated a list of genes that caused these defects when mutated. Out of the 209 total genes, a mutation in only a single gene caused DORV and VSD when heterozygous: Setd5 (MGI allele: 4432631). We concluded Setd5 was a strong candidate for causing haploinsufficient cardiac defects that are similar to those observed due to Tbx1 haploinsufficiency. Moreover, there was a severe loss of function, lethal phenotype in Setd5 nulls (https:// dmdd.org.uk/). Setd5, a ubiquitously expressed gene (Osipovich, Gangula, Vianna, & Magnuson, 2016), is thought to act as a histone modifier. Unlike other SET-domain proteins, SETD5 appears to lack histone methyltransferase activity (Deliu et al., 2018;Osipovich et al., 2016). Immunoprecipitation and mass spectrometry studies indicate SETD5 associated with HDAC3, a protein complex containing histone-deacetylating activity; however, there is conflicting evidence on whether SETD5 is required for HDAC3 recruitment (Deliu et al., 2018;Nakagawa et al., 2020;Osipovich et al., 2016). Clinical reports reveal the implications of SETD5 haploinsufficiency in human intellectual disability, with some patients presenting with cardiac abnormalities (Fernandes et al., 2018;Grozeva et al., 2014;Szczałuba et al., 2016). Furthermore, CHD is present in up to 50% of patients with deleterious mutations in chromatin modification proteins underlying neurodevelopmental or psychiatric syndromes (Zaidi & Brueckner, 2017). Despite this, the role of Setd5 in cardiac development has not been extensively investigated.
Osipovich et al showed that mice lacking Setd5 (MGI allele: 5576778) die before E10.5 and were developmentally delayed; embryos were severely underdeveloped with many presenting with hemorrhaging compared to littermate heterozygous controls (Osipovich et al., 2016). Interestingly, unlike the heterozygous embryos presented in the DMDD browser, Osipovich et al reported heterozygous mice as viable and indistinguishable from the wild-type embryos. Furthermore, the embryonic lethality of Setd5 nulls by E10.5 limits the full assessment of the cardiac phenotype. The purpose of our study was, primarily, to more fully delineate the cardiac phenotype caused by the disruption of Setd5 expression, and secondarily to explore whether there was any genetic interaction between Setd5 and Tbx1. Currently, attempts to understand the Setd5 function are focused on neuronal contexts; here we show the importance of this protein during heart development. Cre (Lewandoski, Meyers, & Martin, 1997). F1 mice harboring the Actin-Cre were backcrossed with wild-type mice in order to generate mice without the Actin-Cre for use in downstream experiments. Generation of double heterozygotes involved breeding Setd5 +/À with Tbx1 lacZ/+ heterozygotes, described previously (Lindsay et al., 2001). To generate Tbx1-cKOs, Setd5 Fl/Fl mice were crossed with Tbx1 Cre/+ heterozygous mice, described previously (Huynh, Chen, Terrell, & Baldini, 2007). To delete Setd5 in the cardiopharyngeal mesoderm, Setd5 Fl/Fl mice were bred with mice expressing Mesp1-Cre, described previously (Saga et al., 1999). Timed matings for embryo collection involved pairing sexually mature mice overnight and checking the presence of a copulation plug the following morning, which indicated 0.5 days post copulation. The PCR strategies for genotyping are available upon request.

| Tissue processing, whole-mount analysis, and histological analysis
Whole embryos were fixed in 4% paraformaldehyde overnight at 4 C and then stored in PBS until required. For histological analyzes, whole embryos or hearts were dehydrated in ethanol, and then processed in Histoclear and several washes in wax. Samples were embedded in paraffin and cut into 10 μm sections, air-dried overnight and then rehydrated through a decreasing gradient of ethanol and stained with hemotoxylin and eosin. Whole-mount embryo pictures were taken using the Zeiss Lumar V12 Stereoscope. Histological sections were imaged using the Zeiss Axioplan or the slide scanning facility at UCL IQPath. Stage-matching was performed by assessment of the limb buds and pharyngeal arches (Boehm et al., 2011;Musy et al., 2018).

| Calculation of double heterozygote expected frequencies
In the absence of a synergistic genetic interaction, the expected frequency of defects in a double heterozygote in which two genes of interest are haploinsufficient is calculated as follows: A = percentage of defects observed in one single heterozygote. B = percentage of defects observed in alternate single heterozygote. n = total number of double heterozygotes collected.
Taking the example of VSDs presented in Table 1

| Statistical analysis
To determine whether genetically modified mice or embryos presented at Mendelian ratios, Chi-squared analysis was performed. To determine the association of genotype with the frequency of defects observed, Fisher's exact test was performed. A p value less than .05 was taken to be significant. to align with the two future ventricles at the looping stage. Correct looping of the heart tube is necessary to align the two definitive outflow vessels with their corresponding ventricular chambers (Christoffels et al., 2000;Moorman & Christoffels, 2003). If the OFT is misaligned such that an abnormal amount of aortic blood originates from the right ventricle, then this is known as an overriding aorta. If more than 50% of the aortic blood arises from the right ventricle, the result is DORV (Creazzo, Godt, Leatherbury, Conway, & Kirby, 1998).
We found that wild-type hearts displayed correctly OFT septation with the outflow vessels originating from separate ventricles ( Figure 1a-a 000 ). In contrast, despite correct OFT septation, 50% of Setd5 heterozygotes presented with rotational defects which included DORV or overriding aorta (Figure 1b-b 000 ) (p < .05, Fisher's exact test, Table 1). DORV is always accompanied by a VSD, since the fusion of the interventricular septum with the OFT cushions is necessary to complete ventricular septation. If the outflow vessels of the OFT are misaligned with the ventricular chambers, as in DORV, then this fusion cannot take place which leads to a perimembranous VSD (Anderson, 2003;Lin, Lin, Chen, Zhou, & Chang, 2012). Figure 1b 0000 shows a perimembranous VSD, which was observed in 75% of Setd5 heterozygotes (p < .05, Fisher's exact test, Table 1).
Therefore, in alignment with the DMDD browser, our results show that Setd5 haploinsufficiency causes OFT rotation defects and VSDs.

| No detectable genetic interaction between
Setd5 and Tbx1, a well-established haploinsufficient gene Because some of the phenotypes observed in Setd5 heterozygotes overlap with those observed in Tbx1 heterozygotes, the genetic interaction between Setd5 and Tbx1 was investigated by generating double heterozygotes, in which both genes of interest are haploinsufficient. Setd5 heterozygotes were mated with Tbx1 lacZ/+ heterozygotes (Tbx1 tm1Bld ) and embryos were harvested at E14.5. An exacerbation of defects, that is, an increased frequency of defects over that expected by additive effects, or an increase in severity of defects, would be observed in the double heterozygotes in the case of synergistic genetic interaction.
There was no difference in the number of embryos at E14.5 based on the genotypes (Table STable 1, Chi-squared, p > .05). Gross morphological analysis during embryo dissection and internal assessment by histological methods revealed no exacerbation or increased frequency of defects in double heterozygotes, suggesting that there was no detectable synergistic genetic interaction between Setd5 and Tbx1 and that these two genes are not functionally interdependent (Table 1).

| Conditional mutagenesis of Setd5 reveals a role in cardiogenic mesoderm
Our results show that the haploinsufficiency of Setd5 led to rotational defects and VSD. These defects are reminiscent of the second heart field (SHF)-specific defects seen in Tbx1 mutants. To investigate whether Setd5 has a specific role in the SHF, and/or pharyngeal epithelia, Setd5 was deleted using a Tbx1-Cre driver known to be active in these tissues (Huynh et al., 2007). Figure 2 presents the gross morphology of control and several Tbx1 conditional knockout (Tbx1-cKO) F I G U R E 1 Rotational defects, including DORV was observed in 50% of Setd5 heterozygotes (Setd5 +/À ) and VSD were observed in 75% of Setd5 +/À embryos at E14.5. In the WT heart, the pulmonary trunk and aorta open into separate ventricles: the right and left ventricle, respectively (a-a 000 ). In Setd5 +/À hearts, DORV was observed. In panel B 0 the aorta and pulmonary trunk are juxtaposed, and in posterior sections b 000 , the aorta eventually opens into the right ventricle; both the pulmonary trunk and aorta open into the right ventricle. A perimembranous VSD is observed in Setd5 +/À (panel b 0000  Figure S1). Therefore, we concluded Setd5 likely exerts its role earlier during cardiac development, or in other lineages, or both.
Indeed, the DMDD browser revealed an embryonic lethality of homozygous Setd5 mutation (no surviving embryos at E14.5), and Osipovich et al., showed that mice constitutively null for Setd5 die before E10.5, and are severely underdeveloped. Hearts at E9.5 were swollen with a single ventricular chamber (Osipovich et al., 2016). The embryonic lethality of Setd5 constitutive null embryos by E10.5 limits analysis of the importance of Setd5 in cardiac development. Lineage tracing experiments show that the majority of cardiac cells are derived from Mesp1-expressing progenitor cells (Saga et al., 1999). Therefore, we used the Mesp1-Cre driver to delete Setd5 in mesoderm encompassing both the first heart field (FHF) and SHF, as well as the   Table 2). 90% of Mesp1-cKOs displayed abnormal cardiac chamber ballooning, with 25% of these exhibiting a dumbbell-shaped heart (Figure 3b), while others displayed right ventricular hypoplasia and abnormal atrial ballooning (p < .05, Figure 3d). We assessed the internal heart morphology by histological techniques and found that 100% of Mesp1-cKO embryos presented with a short OFT, indistinct ventricular chambers, and poorly developed atria, revealing the importance of Setd5 in OFT elongation and cardiac chamber ballooning ( Figure 4). These defects were accompanied by a statistically significant growth delay, determined by the crown-to-rump length (p < .05, Figure S2). The OFT length was further assessed by quantifying the number of paraffin sections containing the OFT and expressing this as a ratio to the crown-to-rump length, confirming the shorter OFT in Mesp1-cKO compared to control embryos ( Figure S3). Morphological analysis of Mesp1-Cre conditional heterozygotes (n = 6) did not reveal any OFT phenotype, and histological analysis revealed only a single embryo as having poorly developed cardiac chambers.

| DISCUSSION
Our results indicate that haploinsufficiency of Setd5, a gene implicated in human intellectual disability (Fernandes et al., 2018;Grozeva et al., 2014;Szczałuba et al., 2016) with reported histonemodifying activity (Deliu et al., 2018;Nakagawa et al., 2020;Osipovich et al., 2016), leads to cardiac defects in mice. It is worth noting that only one Mesp1-Cre conditional heterozygote presented with poorly developed cardiac chambers. The remaining embryos exhibited normal cardiac morphology, highlighting the fact that deleting one allele of Setd5 in the Mesp1 lineage does not fully recapitulate the global heterozygotes and that haploinsufficiency of Setd5 requires constitutive heterozygosity. Since haploinsufficient genes exist in hubs or networks and are more likely to interact with other haploinsufficient genes (Huang et al., 2010), and because haploinsufficiency of Setd5, and of Tbx1 lead to similar cardiac defects, one of our initial aims were to investigate their genetic interaction. However, our findings did not reveal a detectable synergistic genetic interaction, suggesting that Setd5 and Tbx1 are not functionally interdependent. Nevertheless, homozygous deletion of Setd5 in the cardiopharyngeal mesoderm revealed a novel role of Setd5 in cardiac development: its importance in OFT elongation and cardiac chamber ballooning.
In humans, loss of function mutations in SETD5 lead to moderate or severe intellectual disability (Fernandes et al., 2018;Grozeva et al., 2014;Powis et al., 2018;Rawlins, Stals, Eason, & Turnpenny, 2017;Szczałuba et al., 2016). It is also one of three genes within the candidate region of 3p25 microdeletion syndrome which is characterized by developmental delay, intellectual disability, low birth weight, microcephaly and craniofacial abnormalities, and some observations of CHD (Grozeva et al., 2014;Kuechler et al., 2015;Szczałuba et al., 2016). Mutations of SETD5 have been reported in other conditions. For instance, one deletion and two mutations were described in patients clinically diagnosed with KBG syndrome without a mutation in ANKRD11, which is usually observed in these cases (OMIM 148050). One such patient presented with mitral stenosis (Crippa et al., 2020). Interestingly, ANKRD11 recruits HDACs and is mutated in autism (Gallagher et al., 2015). Recently, SETD5 has been implicated in moyamoya angiopathy, which is the occlusion of large cerebral arteries leading to childhood stroke (Pinard et al., 2020). Our results T A B L E 2 Summary of defects observed in Setd5 Fl/Fl , Setd5 Fl/WT , Setd5 Fl/WT ; Mesp1 Cre/+ , and Setd5 Fl/Fl ; Mesp1 Cre/+ embryos at E10.5 Abnormal external morphology

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.