Sall1 and Sall4 cooperatively interact with Myocd and SRF to promote cardiomyocyte proliferation by regulating CDK and cyclin genes

In vertebrates, the heart is formed by mesodermal progenitors in the first and second heart fields (FHF and SHF, respectively). FHF cells, which express Tbx5 and Hcn4, are defined as the progenitors in the cardiac crescent that give rise to the early heart tube and then form the left ventricle (LV) and atria (Bruneau et al., 1999, 2001; Devine et al., 2014; Liang et al., 2013; Später et al., 2013). The SHF cells, expressing islet 1 (Isl1) and Tbx1, contribute to the right ventricle (RV), outflow tract (OFT) and parts of the atria by progressive addition of progenitors from the splanchnic pharyngeal mesoderm to both arterial and venous poles (Brown et al., 2004; Cai et al., 2003; Xu et al., 2004). Developmental defects in FHF- and SHF-derived tissues caused by mutations in specific genes are linked to congenital heart malformations (Buckingham et al., 2005; Cai et al., 2003; Katano et al., 2019). Previous studies have shown that haploinsufficiency of FHF- or SHF-specific genes causes congenital heart diseases, as in the case of Holt-Oram syndrome (HOS), the responsible gene for which is TBX5, which is strongly expressed in the FHF (Bruneau et al., 1999, 2001). In 22q11.2 deletion syndrome (also known as DiGeorge syndrome), the responsible gene is TBX1, which is specifically expressed in the SHF (Lindsay et al., 2001; Xu et al., 2004). Thus, clarifying the mechanisms of the development and maintenance of both the FHF and SHF is important for understanding heart formation.

During heart formation, the proliferation of multipotent cardiac progenitors in the SHF and differentiated cardiomyocytes in the FHF is important (Buckingham et al., 2005; Günthel et al., 2018). Previous studies have reported that the proliferation capacity of SHF cells is mediated by the canonical WNT/β-catenin signaling (Kwon et al., 2007), whereas the proliferation of embryonic cardiomyocytes is promoted by growth signaling molecules, e.g. Hippo, Notch, Nrg1 and Erbb2 (Del Monte-Nieto et al., 2018; Grego-Bessa et al., 2007; Mia and Singh, 2019). In addition, recent studies have revealed that cyclin-dependent kinase 1 (CDK1)/cyclin B1 (CCNB) and CDK4/cyclin D1 (CCND1) complexes are highly expressed in embryonic cardiomyocytes, and the expression of these genes in adult mature cardiomyocytes can promote cell proliferation (Mohamed et al., 2018). However, the upstream regulators of these growth signals and of CDK and cyclin genes in SHF cardiac progenitors and cardiomyocytes are poorly understood.

Spalt-like (Sall) genes, homologs of the Drosophila spalt gene, encode zinc-finger transcription factors. Sall genes are known to play diverse roles in embryonic development through the regulation of stem cell proliferation (Sakaki-Yumoto et al., 2006; Tatetsu et al., 2016) and are heavily involved in heart development (Kiefer et al., 2008; Koshiba-Takeuchi et al., 2006; Morita et al., 2016; Sakaki-Yumoto et al., 2006; Zhao et al., 2021). Four genes in the Sall family (Sall1, Sall2, Sall3 and Sall4) are found in vertebrates. Among these, Sall1 and Sall4 (Sall1/4) are mainly expressed in the heart (Koshiba-Takeuchi et al., 2006; Morita et al., 2016; Sakaki-Yumoto et al., 2006). During heart development, Sall4 is mostly expressed in the LV derived from the FHF (Koshiba-Takeuchi et al., 2006), whereas Sall1 is mostly expressed in the SHF (Morita et al., 2016). Mutations in SALL4 cause Okihiro syndrome (OS), also called Duane-radial ray syndrome (OMIM# 607323), an autosomal-dominant disorder characterized by upper limb, eye and heart defects (Kohlhase, 1993; Kohlhase et al., 2002). Heart defects in individuals with OS occur in the ventricular and atrial septa, and the cardiac conduction system (Kohlhase, 1993). Previous studies have reported that Sall4-deficient mice exhibit ventricular and atrial septal defects (Koshiba-Takeuchi et al., 2006; Sakaki-Yumoto et al., 2006). As malformations of the upper limbs and heart in individuals with OS are similar to those in individuals with HOS, the interaction between Sall4 and Tbx5 has been examined. The results revealed that Sall4 interacts with Tbx5 to regulate downstream genes in a positive or negative manner, and plays important roles in heart and limb development (Koshiba-Takeuchi et al., 2006).

Mutations in SALL1 cause Townes-Brocks syndrome (TBS; OMIM# 107480), which is characterized by limb, ear, anal and heart defects (Kohlhase et al., 1998). The heart anomalies in individuals with TBS include truncus arteriosus, ventricular septal defect, pulmonary atresia and atrial septal defects (Surka et al., 2001). Previous studies have shown that Sall1 is highly expressed in undifferentiated cardiac progenitors and promotes cardiac differentiation (Morita et al., 2016). Mice lacking Sall1 exhibit no dominant TBS-like phenotypes except for renal agenesis (Nishinakamura et al., 2001), whereas mice expressing a truncated form of Sall1 protein exhibit TBS-like phenotypes, including cardiac malformation (Kiefer et al., 2003, 2008). In vitro immunoprecipitation studies have shown that Sall1 and Sall4 can form a heterodimer (Sakaki-Yumoto et al., 2006). These data suggest that Sall1 and Sall4 can substitute mutual functions by sharing target regulatory regions. Therefore, to understand the detailed functions of Sall1 and Sall4 genes in heart development, a new mouse model is required in which both Sall1 and Sall4 functions are interfered with specifically in the heart.

In this study, to clarify the detailed roles of Sall1/4 during heart development, we generated heart-specific Sall1/4 functional inhibition mice by forced expression of ΔSall4, which lacks the C-terminal domain of Sall4, can interfere with both Sall1 and Sall4, and functions in a dominant-negative manner. Our ΔSall4-overexpressing (OE) mice exhibited hypoplasia of the RV and OFT, both of which are derived from SHF progenitors, a hypoplastic trabecular layer and a thinner compact layer. These morphological defects are due to the reduction in proliferative SHF progenitors and left ventricular cardiomyocytes, respectively. RNA sequencing (RNA-seq) analysis of ventricles showed that the expression levels of key regulators of cell cycle progression, including Cdk1, Ccnb1 and cyclin E2 (Ccne2), were downregulated in ΔSall4-OE mice. Moreover, the expression of myocardin (Myocd), a key regulator of cardiomyocyte proliferation and maintenance, was significantly downregulated in the ΔSall4-OE mice. Myocd is a transcriptional co-activator that does not bind to DNA but induces gene expression by interactions with a co-factor, serum response factor (SRF), which binds to DNA through the CArG motif (Wang et al., 2001). The regulatory regions of Cdk1, Ccnb1 and Ccne2 contain the binding sequences of SRF near the Sall4-binding sequences. We examined whether Sall4 directly regulates the genes interacting with Myocd and SRF in ventricular cardiomyocytes. Our findings indicate that Sall1/4 is crucial for cardiac morphogenesis through the promotion of the cell cycle in both SHF progenitors and cardiomyocytes.

Previous reports have shown that Sall1 is mainly expressed in SHF progenitor cells and Sall4 is mainly expressed in the LV (Koshiba-Takeuchi et al., 2006; Morita et al., 2016), but the detailed expression patterns of Sall1 and Sall4 in mouse heart development remain unclear. We analyzed the Sall1 and Sall4 expression patterns by in situ hybridization and immunohistochemistry at embryonic day (E) 9.5 and E10.5. Consistent with the previous report, we confirmed strong Sall1 expression in the SHF at E9.5 (Fig. S1A). In addition, Sall4 was also expressed in the SHF at E9.5 (Fig. S1A). Furthermore, the expression of both Sall1 and Sall4 was observed in cardiomyocytes of the ventricle at E10.5, especially in the LV (Fig. S1B,C). These data indicate that the expression patterns of Sall1 and Sall4 overlap during heart development and that both Sall proteins may function in the SHF as well as in cardiomyocytes.

To study the role of Sall1/4 during heart development, we generated CAG-CAT-Myc-ΔSall4-EGFP; Nkx2.5^Cre/+ (hereafter, ΔS4; NC) conditional mutant mice in which ΔSall4 is specifically induced in the heart together with EGFP (Fig. 1A). ΔSall4 (also known as Sall4-R2) (Koshiba-Takeuchi et al., 2006) lacks the C-terminal zinc-finger motif (Fig. 1B), which is essential for transcriptional activity (Koshiba-Takeuchi et al., 2006), and has a dominant-negative effect on both Sall1 and Sall4. In fact, a luciferase assay using the promoter region of Gja5, which is a known downstream target gene of Sall1/4 (Kiefer et al., 2008; Koshiba-Takeuchi et al., 2006), in HEK293T cells showed that the Gja5 promoter was activated by Sall1 or Sall4 transfection but that these transcriptional activations were inhibited when ΔSall4 was transfected together with Sall1 or Sall4 (Fig. 1C). In addition, ΔSall4 single transfection was not affected by Gja5-promoter activation (Fig. S2). To confirm the expression of ΔSall4 in ΔS4; NC mice, we performed western blotting with anti-Sall4 and anti-Myc antibodies, and found that Myc-ΔSall4 was specifically expressed in ΔS4; NC mouse hearts (Fig. 1D, Fig. S3). Heart-specific expression of EGFP was observed in ΔS4; NC mice (Fig. 1E), and immunohistochemistry with anti-EGFP and anti-cTnT, anti-CD31 or anti-Isl1 antibodies showed that EGFP was expressed in cardiomyocytes, endocardium and SHF progenitors (Fig. 1F,G, and Fig. S4). To test for the function of Sall1 protein in ΔSall4, we performed a proximity ligation assay (PLA) using anti-Myc-tag and anti-Sall1 antibodies. ΔSall4 interacted with Sall1 in the LV of the ΔS4; NC heart (Fig. S5A,B). Subsequently, we examined the mRNA expression level of Gja5 in the ΔS4; NC mice. The results of qPCR and in situ hybridization showed that Gja5 was significantly downregulated in the ΔS4; NC mice compared with control mice (Fig. 1H,I). The expression of Nppa, which is repressed by Sall4, was expanded from the LV to RV (Fig. 1I). These data showed that the functions of both Sall1 and Sall4 were disrupted in the hearts of ΔS4; NC mice.

Morphological defects in ΔS4; NC mice were clearly observed from E10.5, whereas the identifiable phenotypic differences were not observed between wild-type and ΔS4; NC mice at E8.5 and E9.5 (Fig. S6). The ΔS4; NC mice exhibited RV hypoplasia at E10.5 and malformation of the aorta at E12.5 (Fig. 2A,B), and were lethal around E14.5. The 3D morphological analysis of the heart was performed using correlative microscopy and block–face imaging (CoMBI) system, and clearly showed a hypoplastic RV and OFT in ΔS4; NC mice at E10.5 (Fig. 2C). Histological and 3D imaging analyses showed that the formations of the ventricular septum and dorsal mesenchymal protrusion (DMP), which is required for the separation of right and left atrioventricular junctions, were disturbed at E12.5 in ΔS4; NC mice (Fig. 2D,E). We also found that the compact layer of the myocardium was very thin, and the trabecular myocardium was completely absent or attenuated, especially in the LV, at E12.5 in ΔS4; NC mice (Fig. 2F). Thus, these observations indicated that Sall1/4 are required for not only ventricular septal formation but also for the growth of the trabecular and compact layer, and for the formation of the OFT, RV and DMP.

To clarify the reason for the thinner ventricular wall in ΔS4; NC mice, we examined the proliferation and cell death states of cardiomyocytes at E9.5. We counted the number of Ki67-positive cells (proliferative cells) or cleaved-caspase 3-positive cells (apoptotic cells) among the Nkx2.5-positive cardiomyocytes in the LV and RV. The population of Nkx2.5- and Ki67-double-positive cells decreased significantly in the LV; however, the population of those cells was unaffected in the RV of ΔS4; NC mice (Fig. 3A,C and Fig. S7A,B), and the population of Nkx2.5- and cleaved-caspase 3 double-positive cells increased significantly in the LV; however, the population of those cells was unaffected in the RV of ΔS4; NC mice (Fig. 3B,D and Fig. S7C,D). Furthermore, quantification of the number of Ki67- or pHH3 (mitosis marker)-positive cells in the primary cultured cardiomyocytes also showed that a dramatic reduction of proliferative cardiomyocytes occurred in the ΔS4; NC mice (Fig. 3E-H). We next performed qPCR to examine the expression of cardiac myosin genes (Myh6 and Myh7), and the results indicated that Myh6 expression levels did not change; however, levels of Myh7 were significantly decreased in ΔS4; NC mice at E12.5 (Fig. S8). These data indicate that Sall1/4 are crucial for the proliferation, maintenance and differentiation of cardiomyocytes; in addition, the thinned ventricular wall with attenuated trabeculae in the LV of ΔS4; NC hearts may be caused by the drastic decrease in proliferative cardiomyocytes and the increase in apoptotic cells in the LV.

As both RV and OFT are derived from the SHF, we examined the proliferation and cell death states of the SHF progenitors in the ΔS4; NC mice. We counted Ki67-positive cells or cleaved-caspase 3-positive cells among Isl1-positive SHF progenitor cells at E9.5. A significant reduction in proliferative progenitors was observed in the anterior SHF of ΔS4; NC mice (Fig. 3I,K) and the posterior SHF (Fig. S9A,B). Apoptotic progenitors were increased in both SHFs of ΔS4; NC mice (Fig. 3J,L), especially in the posterior region (Fig. S9C,D). These results indicate that the severe hypoplasia in the SHF-derived structures in the ΔS4; NC mice could be due to the marked reduction in proliferative progenitor cells and the increase in apoptotic cells in the SHF.

To clarify the downstream genes affected by Sall1/4 inhibition, we performed RNA-seq analysis using the dissected ventricles, including OFT from wild-type, CAG-CAT-ΔSall4 (ΔS4), Nkx2.5^Cre/+ (NC) and ΔS4; NC mice at E10.5 (Fig. 4A). We then performed Gene Ontology (GO) enrichment analysis using DAVID (https://david.ncifcrf.gov/) to classify the functions of genes that showed differential expression of at least a 1.2-fold change between NC and ΔS4; NC mice. This analysis revealed that deregulated genes were significantly enriched in the GO terms ‘cell cycle’, ‘cell division’ and ‘mitotic nuclear division’, which are related to cell proliferation, and were also enriched in the GO terms linked to ‘sarcomere organization’ and ‘cardiac muscle contraction’ (Fig. 4B). Sall1/4 may be important regulators of cell cycle progression and cardiomyocyte differentiation in the developing heart.

Because the ΔS4; NC mice exhibited a significant reduction in the number of proliferative cardiomyocytes and cardiac progenitors in the SHF, we focused on the expression of genes involved in the cell cycle. To identify the cell cycle genes regulated by Sall1/4, we compared the expression levels of the positive and negative regulators of G1-S and G2-M phase transitions in control (wild type, ΔS4 and NC) and ΔS4; NC mice. The heatmap and qPCR data indicated that Ccne2, a positive regulator of the G1- to S-phase transition, and Cdk1 and Ccnb1, positive regulators for the G2- to M-phase transition, were downregulated in ΔS4; NC mice (Fig. 4C,D,E), whereas cyclin-dependent kinase inhibitor 1 (Cdkn1a), a negative regulator for both the G1- to S- and G2- to M-phase transitions, and cyclin G1 (Ccng1), a negative regulator for the G2- to M-phase transition, were upregulated (Fig. 4C,D,F). These results demonstrate that Sall1/4 may progress the cell cycle by regulating cell cycle gene expression during heart development.

Transient expression of Isl1 in the splanchnic mesoderm at early stages is required for the proliferation of SHF cells (Cai et al., 2003). As the expression of Sall1 overlaps that of the Isl1-positive region (Morita et al., 2016), we examined the interaction between Sall1/4 and Isl1. The immunohistochemistry showed that the expression of Isl1 was dramatically decreased in both anterior and posterior SHFs at E9.5 in ΔS4; NC mice (Fig. S10A,B). A previous study reported two Sall1-binding sequences (Isl1-3.1 and -3.2) in the Isl1-enhancer region (Morita et al., 2016). Motif analyses using Sall4-binding sequences in the JASPAR database (Castro-Mondragon et al., 2022) and Sall1-binding sequences (Yamashita et al., 2007) revealed Sall1- and Sall4-binding sequences in the Isl1-3.1 and -3.2 regions (Fig. S10C,D). To test whether Sall1 regulates Isl1 transcription via Isl1-3.1 and -3.2, we performed a luciferase assay using HEK293T cells. Although the Isl1-3.1 region was not activated by Sall1 transfection (Fig. S10E), the Isl1-3.2 region was activated by Sall1 transfection in a concentration-dependent manner (Fig. S10F) and was also promoted by Sall4 transfection (Fig. S10G). Transfection of Sall1 or Sall4 with ΔSall4 repressed the transcriptional activities of Sall1 and Sall4 (Fig. S10G,H). These results indicate that Sall1/4 can control the proliferation of SHF progenitors by activating Isl1 expression.

To clarify how Sall1/4 control ventricular myocardial development, we examined the expression of genes that are related to trabecular and compact layer formation. RNA-seq data showed that Myocd, Srf and Nkx2.5 were significantly downregulated in ΔS4; NC mice (Fig. 5A). We performed qPCR and in situ hybridization, and found that Myocd expression level was decreased in the ventricle of ΔS4; NC mice (Fig. 5B,C). We also found that the expression levels of Srf, which encodes a co-factor of Myocd, and Nkx2.5 and Mef2c, which are the downstream targets of Myocd and SRF (Huang et al., 2012; Small et al., 2005), were decreased in ΔS4; NC mice (Fig. 5B,D,E,F). These data suggest that Sall1/4 act upstream of Srf and Myocd to regulate trabecular and compact layer formation.

In ΔS4; NC mice, the number of proliferative cardiomyocytes was severely decreased (Fig. 3A,E,F), and the expression levels of CDK and cyclin genes were dramatically changed (Fig. 4B-F). To clarify whether Sall4 directly regulates CDK and cyclin genes to promote cell cycle progression, we performed motif analyses using FIMO (Grant et al., 2011) with the JASPAR database (Castro-Mondragon et al., 2022), which are tools for identifying the motifs of transcription factors, for the promoter regions of Cdk1, Ccnb1 and Ccne2, which expression levels were significantly decreased in ΔS4; NC mice (Fig. 4C,E). We found that each promoter contained multiple Sall4-binding sequences (Fig. S11). Myocd- and SRF-binding sequences (CArG DNA motifs) were also enriched near Sall4-binding sequences in the promoter regions of Cdk1, Ccnb1 and Ccne2 (Fig. S11). Thus, we hypothesized that Sall1/4 regulate the expression of Srf and Myocd (Fig. 5A-D). Subsequently, Sall4 could interact with Myocd and SRF to activate the expression of cell cycle-related genes. Myocd is a strong transcriptional co-activator in cardiomyocytes interacting with SRF (Huang et al., 2012; Wang et al., 2001), and Myocd-deficient mice (Huang et al., 2012) exhibit repression of cardiomyocyte proliferation, similar to ΔS4; NC mice, suggesting an interaction with Sall4. To test this hypothesis, we first performed co-immunoprecipitation (Co-IP) experiments, and the results showed that Sall4 physically interacted with both Myocd and SRF (Fig. 6A,B and Fig. S12). In addition, the results of PLA indicated that the endogenous interactions of Sall4-Myocd and Sall4-SRF were formed in trabecular and compact layer cardiomyocytes (Fig. 6C,D and Fig. S13A) and ΔSall4 also interacted with Myocd and SRF (Fig. S13B,C), whereas PLA signals of Sall1-Myocd and Sall1-SRF were not detected in cardiomyocytes (Fig. S13D,E). Subsequently, we performed ChIP-seq (chromatin immunoprecipitation followed by sequencing) and Cleavage Under Targets & Release Using Nuclease (CUT & RUN)-qPCR using anti-Sall4 and anti-Myocd antibodies to examine whether Sall4 and Myocd could directly regulate the cell cycle-related genes. GO analyses based on ChIP-seq data for target genes of Sall4 and Myocd revealed significant enrichment in the ‘regulation of cell cycle’ and ‘cell cycle process’, and the ‘regulation of cell cycle’, respectively (Fig. S14A,B). In addition, both Sall4 and Myocd bound to the regulatory regions of Cdk1 and Ccnb1 (Fig. 6E-G). In the Ccne2 regulatory region, the ChIP-peaks of both Sall4 and Myocd were observed but CUT & RUN-qPCR data indicated that the enrichment of Sall4 was weak (Fig. S15). In the regulatory regions of Cdkn1a and Ccng1, which are negative regulators of the cell cycle and upregulated in ΔS4; NC mice, Sall4- and Myocd-ChIP peaks were not observed (Fig. S16A,B). In both Ccnb1 and Ccne2 regulatory regions, H3K27ac (a marker of active enhancers and promoters) peaks were decreased in ΔS4; NC mice; in the Cdk1 and Ccnb1 regulatory regions, there was a drastic increase in H3K27me3 (a marker of transcriptional repression) peaks in ΔS4; NC mice. These results suggest that the regulatory regions of all three genes were silenced when ΔSall4 was overexpressed (Fig. 6E and Fig. S15A). We further analyzed the regulatory regions of Cdk1 and Ccnb1 around Sall4- and Myocd-ChIP peaks using FIMO with the JASPAR database to find the motifs of other transcription factors. Tbx5-binding motifs exist in the Cdk1 and Ccnb1 regulatory regions (Fig. 6H,I), suggesting the potential of Sall4 in forming a complex with Tbx5 to regulate these genes.

Finally, we performed Sall4 or Sall1 and Myocd double knockdown experiments using siRNAs that were transfected into the primary culture of cardiomyocytes to test whether the cooperative interaction between Sall1/4 and Myocd is crucial in promoting cardiomyocyte proliferation. We confirmed that Sall1, Sall4 and Myocd expression levels were successfully decreased by treatment with siRNAs (Fig. S17A-C). We then performed immunohistochemistry using anti-Ki67 and anti-cTnT antibodies, and found that the ratio of proliferative cardiomyocytes was unchanged between Sall1 and Myocd double-knockdown and Sall1 or Myocd single-knockdown cardiomyocytes (Fig. S18A,B), whereas the same ratio was considerably reduced in the Sall4 and Myocd double-knockdown cardiomyocytes compared with the Sall4 or Myocd single-knockdown cardiomyocytes (Fig. 7A,B).

Sall1 and Sall4 are known to have functional redundancies with each other (Böhm et al., 2008; Sakaki-Yumoto et al., 2006; Tsubooka et al., 2009). Therefore, we examined whether the expression of Sall1 or Sall4 is increased in Sall4- or Sall1-knockdown cardiomyocytes, respectively. qPCR analyses showed that Sall1 or Sall4 expression levels were unchanged in siSall4- or siSall1-treated cardiomyocytes, respectively (Fig. S17D,E), indicating Sall1 and Sall4 have no functional redundancy when either one is lacking in cardiomyocytes. Taken together, these results suggest that Sall4 regulates Cdk1 and Ccnb1 expression by interacting with Myocd and SRF, and contributes to cardiomyocyte proliferation (Fig. 7C).

Our data demonstrate that Sall1/4 act as key regulators of cell cycle progression by regulating Isl1 in the SHF and regulating cell cycle genes in cardiomyocytes. Functional inhibition of Sall1/4 leads to hypoplasia of the OFT and the RV, which are derived from the SHF, and leads to a thinner compact layer with trabeculation defects. The most important thing that we found is that Sall1/4 promote both Myocd and Srf expression in the heart, and that Sall4, Myocd and SRF cooperatively promote cardiomyocyte proliferation by regulating Cdk1 and Ccnb1 (Fig. 7C), suggesting that Sall4, SRF and Myocd form a feed-forward loop for stable regulation of cell cycle genes for embryonic cardiomyocyte proliferation.

Sall proteins can function by forming homodimers and heterodimers with each other (Rao et al., 2010; Sakaki-Yumoto et al., 2006; Sweetman et al., 2003), and Sall4 regulates the gene expression in a partner protein-dependent manner (Koshiba-Takeuchi et al., 2006; Miller et al., 2016). The results of the luciferase assay revealed that ΔSall4 has dominant-negative effects on transcriptional activation of Sall1 and Sall4, and the PLA experiments demonstrated that ΔSall4 interacts with Sall1, SRF and Myocd. Thus, ΔSall4 can inhibit the interaction of Sall4 with its transcriptional partner. SALL proteins contain a glutamine (Q)-rich region at the N terminal (Farrell et al., 2001; Koshiba-Takeuchi et al., 2006; Sweetman et al., 2003), which is required for the protein-protein interaction with all SALL family members and other partner proteins (e.g. Sall4-Tbx5 interaction) in vitro (Koshiba-Takeuchi et al., 2006; Sweetman et al., 2003). Thus, the Q-rich domain is still active in ΔSall4, and ΔSall4 can compete with Sall proteins to interact with partner proteins and repress their transcriptional abilities. ΔSall1, in which the C-terminal zinc-finger motif is truncated as in ΔSall4, can interact with all SALL family members (Kiefer et al., 2003). The mutant mice encoding ΔSall1 show more severe phenotypes in multiple organs, including the heart compared with Sall1-null mice (Kiefer et al., 2003, 2008). Therefore, ΔSall1 inhibits the function of other Sall proteins in a dominant-negative manner. The predicted protein products of ΔSall4 and ΔSall1 resemble those in individuals with OS and TBS, respectively (Al-Baradie et al., 2002; Kiefer et al., 2003, 2008; Kohlhase et al., 2002; Koshiba-Takeuchi et al., 2006). Therefore, dominant-negative models expressing ΔSall4 and ΔSall1 would be useful for understanding the pathology of OS and TBS.

We also tested the redundancy of Sall1 and Sall4 through knockdown experiments using siRNA; the results showed no upregulations of Sall1 and Sall4 in Sall4- and Sall1-knockdown cardiomyocytes, respectively. Thus, no redundant mechanism between Sall1 and Sall4 exists when either one is lacking in the developing heart.

ΔS4; NC mice showed hypoplasia of both the RV and OFT with a lack of proliferation of progenitor cells in the SHF. As the proliferative capacity of cardiomyocytes in RV did not differ between control and ΔS4; NC mice, and the proliferative anterior SHF progenitors significantly decreased in ΔS4; NC mice, the defects in the RV and OFT may be due to the lack of contribution of SHF progenitors. A previous study has shown that Sall1 is transiently expressed in SHF during the early stages of mouse embryos and that Sall1-positive cells contribute to OFT and RV formation (Morita et al., 2016). In addition, in vitro Sall1 overexpression experiments have demonstrated that Sall1 can induce Isl1 expression (Morita et al., 2016). Isl1, a LIM homeodomain transcription factor, is essential for the proliferation and maintenance of SHF cells. Lack of Isl1 in SHF progenitors causes hypoplasia of both the OFT and RV, as well as ventricular and atrial septal defects (Cai et al., 2003; Gao et al., 2019). In this study, we found that Isl1 expression was markedly reduced in the SHF of ΔS4; NC mice. A mouse Isl1 enhancer that is sufficient to induce endogenous Isl1 expression in the SHF has been reported (Kang et al., 2009), and it contains two Sall1-binding sequences (Morita et al., 2016). Luciferase assay analyses showed that one of the Isl1-regulatory regions (Isl1-3.2) was activated by Sall1 or Sall4 transfection. Taken together, these results suggest that Sall1/4 contribute to the development of SHF derivatives through transcriptional activation of Isl1. Motif analysis revealed that the Isl1-3.2 region contained the SRF-binding sequences near those of Sall1 and Sall4. Thus, Sall1/4 may regulate Isl1 expression and development of SHF by interacting with SRF and Myocd but further experiments are needed to validate this speculation.

Our data show that Sall1/4 are expressed in both anterior and posterior SHFs, and are important for cell proliferation and maintenance in both regions. The proliferation of cardiac progenitors in the posterior SHF is crucial for DMP formation (Hoffmann et al., 2009; Stefanovic et al., 2020), and our 3D imaging and histological analyses demonstrated that ΔS4; NC mice caused severe defects in DMP formation. Thus, Sall1/4 are important factors for the separation of right and left atrioventricular junctions through the regulation of posterior SHF development.

Our RNA-seq data identified multiple genes downstream of Sall1/4, and further analyses revealed that some of the downstream genes were involved in ventricular myocardial proliferation during mouse heart development. Among them, we focused on Myocd, as it is known to be an important factor for cardiomyocyte differentiation, proliferation and maintenance during heart development (Huang et al., 2012, 2009). The phenotypes of Myocd-deficient mice are similar to those of ΔS4; NC mice, with a thinner compact layer of cardiomyocytes with a hypoplastic trabecular myocardium (Huang et al., 2012; Wang et al., 2001). Myocd physically associates with SRF, a MADs box transcription factor, and synergistically regulates the expression of target genes in the developing heart (Huang et al., 2012; Wang et al., 2001). We found that Srf and several target genes of Myocd were downregulated in the ventricles of ΔS4; NC mice. The co-immunoprecipitation and PLA results demonstrate that Sall4 physically interacts with both Myocd and SRF in vivo and in vitro. These data suggest that Sall1/4 promote the transcription of Srf and Myocd, and then Sall4 forms a transcriptional complex with SRF and Myocd to play an important role in the development of compact and trabecular myocardium. These data also suggest that Sall1 and Sall4 might use different transcriptional partners in cardiomyocytes. A previous study reported that Sall4-hetero knockout mice exhibited ventricular septal defect at a rate of 20%, while Sall1-hetero knockout mice did not exhibit ventricular septal defects (Sakaki-Yumoto et al., 2006). An underlying reason for the different phenotypes between Sall4- and Sall1-hetero knockout mice may be the partner use of Sall1 and Sall4. Thus, the functions of Sall4 are not completely compensated for by the functions of Sall1 in cardiomyocytes.

ΔS4; NC mice exhibited hypoplastic trabecular and thinner compact layer accompanied by a decrease in cardiomyocyte proliferation. Previous studies have reported that defects in trabecular and compact layer formation are caused by abnormal regulation of growth signals, including transforming growth factor β (TGFβ), Erk, Notch, Nrg1 and Erbb2 (Del Monte-Nieto et al., 2018; DiMichele et al., 2009; Grego-Bessa et al., 2007; Kodo et al., 2016). The gain- and loss-of-function experiments of these growth signaling molecules resulted in alterations in cell cycle regulation in developing cardiomyocytes (DiMichele et al., 2009; Kodo et al., 2016). Therefore, regulation of cell cycle checkpoints is the most important aspect for cardiomyocyte proliferation. The transcriptions of cyclin and CDK genes are highly upregulated, whereas CDK inhibitor genes are repressed during heart development (Ponnusamy et al., 2017). Meis1, Tbx20, Mlf1 and Rest have previously been reported as upstream activators and repressors of cell cycle genes (Muralidhar and Sadek, 2016; Rangrez et al., 2017; Xiang et al., 2016; Zhang et al., 2017). Our heatmap analyses about cell cycle-related genes identified multiple genes whose expression was specifically altered in ΔS4; NC mice, but the cell cycle-related transcription factor genes (Meis1, Tbx20, Mlf1 and Rest) and the genes related to growth signals (Wnt/β-catenin, Notch, Nrg1 and Erbb2) were not affected. These results indicate that Sall1/4 can regulate the expression of cyclin, CDK and CDK inhibitors in a different way from previously reported transcription factors. This means that Sall1/4 are previously unreported upstream regulators of multiple cell cycle-associated genes in mouse heart development. GO analysis of Sall4 ChIP-seq data indicated the potential of Sall4 in directly regulating multiple cell cycle-associated genes. In this study, we identified Cdk1, Ccnb1 and Ccne2 as the downstream targeted genes of Sall4. The CDK1/CCNB1 complex promotes mitosis through the regulation of the G2-M phase transition (Fang et al., 2014; Krek and Nigg, 1991; Morgan, 1995), and recent studies have reported that CDK1 and CCNB1 are important components for the induction of cardiomyocyte proliferation in adult post-mitotic cells (Bicknell et al., 2004; Mohamed et al., 2018). CCNE2 is crucial for G1- to S-phase progression through its interaction with CDK2 to activate CDK2 kinase activity (Hwang and Clurman, 2005; Siu et al., 2012). Our data suggest that Sall1/4 enhance the activities of the CDK1/CCNB1 and CDK2/CCNE2 complexes through the transcriptional activation of these genes to progress the cell cycle.

It has been shown that Sall4 can act as either a transcriptional activator or a repressor, depending on the partner protein (Tatetsu et al., 2016; Yang, 2018). Our ChIP-seq and co-immunoprecipitation data indicate that Sall4 physically interacts with Myocd and SRF, and activates the expression of Cdk1 and Ccnb1. We found that Tbx5-binding sequences were located near the Sall4- and SRF-binding sequences in the Cdk1 and Ccnb1 regulatory regions. Tbx5 is a major positive regulator of the embryonic cardiomyocyte proliferation (Goetz et al., 2006; Misra et al., 2014), and physically interacts with Sall4 to regulate the heart development (Koshiba-Takeuchi et al., 2006). Sall4 can also interact with NurD complexes, which are transcriptional repressors that modulate chromatin conformation (Lu et al., 2009; Yang et al., 2012; Yuri et al., 2009), suggesting that Sall4 with Myocd, SRF, Tbx5 and/or NurD complexes may synergistically activate or repress the expression of cell cycle-related genes to provide efficient proliferative ability to cardiomyocytes during embryonic development. Thus, the identification of Sall4 partners in vivo is important for understanding the regulatory mechanisms of cardiomyocyte proliferation during the transition from the active embryonic state to the silent adult state.

In this study, we revealed that Sall1/4 play important roles in the maintenance and proliferation of cardiomyocytes in embryonic hearts, but the function of Sall1/4 in adult is still unclear. Whether the forced expression of Sall1/4 with and/or without Myocd and SRF in adult mature cardiomyocytes, followed by injury promotes heart regeneration by activation of CDK and cyclin genes necessitates further investigation.

In conclusion, our study improves the understanding of the mechanisms underlying the proliferation and maintenance of cardiac progenitors in anterior and posterior SHFs and cardiomyocytes. Importantly, our findings present a working model for the promotion of cell cycle progression in cardiomyocytes, indicating a previously unreported function for Sall1/4 and Myocd and SRF. These observations provide a new insight into the molecular basis of cardiomyocyte proliferation and contribute to a better understanding of cardiac regeneration after myocardial infarction.

All animal experimental procedures and protocols were approved by the Animal Care and Use Committee of Toyo University (approval numbers 2017-018, 2018-27, 2019-26, 2020-19, 2021-31 and 2022-25). To construct the CAG-CAT-Myc-ΔSall4-EGFP transgene, a truncated form of the Sall4 protein (Sall4-R2), which lacks C-terminal zinc-finger motifs as previously described (Koshiba-Takeuchi et al., 2006) was inserted into the CAG-CAT vector (Kokubo et al., 2007; Sakai and Miyazaki, 1997) together with the IRES-EGFP region derived from pIRES2-EGFP (Clontech). This construct was injected into fertilized eggs to generate permanent transgenic lines using standard methods. Nkx2.5^Cre/+ mice have been previously reported (Moses et al., 2001). We crossed CAG-CAT-Myc-ΔSall4-EGFP and Nkx2.5^Cre/+ mice to generate conditional mutant mice expressing ΔSall4 specifically in the heart. Mice were bred in a C57BL/6J background.

Hearts were dissected from three mouse embryos at E12.5 and pooled for each sample. The minced heart tissues were treated with lysis buffer [50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 1 mM EDTA, 1% Triton-X-100 and 0.1% Proteinase Inhibitor Cocktail; Nacalai Tesque], and sonicated. SDS-PAGE was performed as the standard method. The lysate of the hearts was transferred to the membranes electrically. The membranes were then blocked with 10% Blocking One (Nacalai Tesque) in TBST for 30 min at room temperature and incubated for 1 h 30 min at room temperature with primary antibodies: anti-Sall4 antibody (1:1000, PP-PPZ0601-00, Perseus Proteomics), anti-β-actin antibody (1:2500, PM053, MBL), anti-Myc-tag antibody (1:1000, 562, MBL), anti-DDDDK (Flag)-tag antibody (1:1000, PM020, MBL), and anti-SRF antibody (1:100, sc-2590, Santa Cruz). After treatment with the primary antibodies, the membranes were washed with TBST. The membranes were incubated for 1 h at room temperature with anti-mouse IgG HRP antibody (1:5000, 715-035-151, Jackson ImmunoResearch) or anti-rabbit IgG HRP antibody (1:5000, 711-035-152, Jackson ImmunoResearch). After washing in TBST, the signal was detected using an ECL detection kit (Chemi-Lumi One Super, Nacalai Tesque). Images were acquired using ChemiDoc (BioRad).

Mouse hearts were fixed in 4% w/v paraformaldehyde (PFA) in phosphate-buffered saline (PBS) and, after dehydration, they were embedded in paraffin wax and sectioned at 10 µm. The sections were stained with Hematoxylin and Eosin (H&E). Images were acquired using a BZ-9000 microscope (KEYENCE).

In situ hybridization was performed as previously described (Koshiba-Takeuchi, 2018). DIG-labeled antisense probes were hybridized overnight at 65°C with whole embryos or sections. After hybridization, the samples were treated with anti-digoxin-AP Fab fragments (Roche, Sigma-Aldrich) and visualized using a BM purple AP substrate (Roche, Sigma-Aldrich) or NBT/BCIP (Nacalai Tesque).

Serial images were obtained using the CoMBI system (Ishii et al., 2021; Sutrisno et al., 2023; Tajika et al., 2023, 2017) or paraffin wax-embedded sections. For the CoMBI, we performed whole-mount in situ hybridization with a Myl7-RNA probe for marking the heart region in E10.5 mouse embryos and counterstained with 1% tannin acid/PBS for 1 h. The paraffin sections were stained with Hematoxylin and Eosin. Reconstruction of serial images was performed using AMIRA 5.4.1 (Thermo Fisher Scientific).

Immunohistochemistry was performed as previously described (Nakamura et al., 2016). The following antibodies were used at the indicated concentrations: monoclonal anti-cTnT antibody (1:200, MA5-12960, Invitrogen), polyclonal anti-cTnT antibody (1:50, 15513-AP, Proteintech), anti-CD31 antibody (1:200, 550274, BD Pharmingen), anti-GFP antibody (1:200, A-11122, Invitrogen), monoclonal anti-Sall1 antibody (1:200, PP-K9814-00, Perseus Proteomics), monoclonal anti-Sall4 antibody (1:200, PP-PPZ0601-00, Perseus Proteomics), anti-Nkx2.5 antibody (1:200, sc-8697, Santa Cruz), monoclonal anti-Isl1 antibody (1:200, 39.4D5, Developmental Studies Hybridoma Bank), polyclonal anti-Ki67 antibody (1:100, ab15580, Abcam), polyclonal anti-pHH3 antibody (1:200, 06-570, Merck), polyclonal anti-Cleaved Caspase-3 antibody (1:100, 9661, Cell Signaling Technology) and anti-Mef2 antibody (1:200, sc-313, Santa Cruz). The following corresponding secondary antibodies were used: Alexa Fluor 594 donkey anti-goat IgG (H+L) antibody (1:200, A-11058, Thermo Fisher Scientific), Alexa Fluor 594 donkey anti-rat (H+L) antibody (1:200, A-21209, Thermo Fisher Scientific), Alexa Fluor 488 donkey anti-rabbit IgG (H+L) antibody (1:200, A-21206, Thermo Fisher Scientific), Alexa Fluor 488 donkey anti-mouse IgG (H+L) antibody (1:200, A-21202, Thermo Fisher Scientific), Alexa Fluor 594 donkey anti-rabbit IgG (H+L) antibody (1:200, A-21207, Thermo Fisher Scientific), Alexa Fluor 594 donkey anti-mouse IgG (H+L) antibody (1:200, A-21203, Thermo Fisher Scientific) and Alexa Fluor 488 goat anti-rabbit IgG (H+L) antibody (1:200, A-11034, Thermo Fisher Scientific). Images were acquired using a BZ-9000 microscope (KEYENCE).

Transfection of HEK293T cells was performed using polyethylenimine (PEI). Total cell lysates were prepared 48 h after transfection, luciferase activity was assessed using the Promega Dual Luciferase Reporter kit (Promega) and transfection efficiency was normalized to β-gal activity using the Beta-Glo Assay System (Promega). The data were acquired using Mithras LB940 (Berthold Technologies).

Cells were collected from microdissected mouse hearts at E11.5 or E12.5, using the Pierce Primary Cardiomyocyte Isolation Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. The cells were cultured in DMEM (Nacalai Tesque) containing 10% FBS (BioWest) and 1% penicillin-streptomycin mixed solution (Nacalai Tesque).

Total RNA for RNA-sequencing was collected from microdissected ventricles, including the outflow tract in E10.5 mouse embryo using Sepasol-RNAISuper G (Nacalai Tesque), and the RNA from five ventricles was pooled. A DNA library was prepared using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina). Paired-end fastq sequence reads from each sample were obtained using the Nova-seq 6000 (Illumina) and the adapter sequence reads were trimmed using trimmomatic (ver. 0.38), and then mapped to the GRCm38.p6 genome using STAR (ver. 2.7). The count data were calculated and normalized with transcripts per million (TPM) for comparing the gene expression levels using RSEM. Data analysis was performed by Genble (Fukuoka, Japan) (https://genble.co.jp).

To extract total RNA, ventricles including the outflow tract from three mouse embryos at E10.5 for each sample, whole ventricles at E12.5 from a mouse embryo and primary cultured cardiomyocytes were used. Total RNA was extracted using Sepasol-RNA I Super G (Nacalai Tesque) and reverse-transcribed using ReverTra ACE qPCR RT Master Mix with gDNA Remover (TOYOBO). qPCR was performed using TB Green Premix Ex Taq II (Takara) or Taq Pro Universal SYBR qPCR Master Mix (NIPPON Genetics) on a QuantStudio 3 real-time PCR system (Applied Biosystems). Gene expression levels were normalized to β-actin and analyzed using the ΔΔCt method. The primer sequences are listed in Table S1.

Lipofectamine RNAiMAX transfection reagent (Thermo Fisher Scientific) was used for the transfection of primary cultured cardiomyocytes according to the manufacturer's instructions. Cardiomyocytes were incubated with siRNA for 48 h and then fixed with 4% PFA for immunohistochemistry. The following siRNAs were used at the indicated concentrations: Sall4 (5 nM per well, Ambion Silencer Select pre-designed s97424, Life Technologies), Myocd (5 nM per well, Ambion Silencer Select, pre-designed s103073, Life Technologies), Sall1 (5 nM per well, Ambion Silencer Select pre-designed s81539, Life Technologies) and negative control (Silence Negative Control No. 1, AM4611, Life Technologies).

HEK293T cells were transfected with the Myc-Sall4 expression construct together with Flag-Myocd or HA-SRF expression constructs using PEI. Total cell lysates were prepared 48 h after transfection. Co-immunoprecipitation was performed using the Dynabeads Co-Immunoprecipitation Kit (VERITAS) according to the manufacturer's instructions. The following antibodies were used for immunoprecipitation: anti-Myc-tag (562, MBL), anti-DDDDK (Flag)-tag (PM020, MBL) and anti-SRF (sc-2590, Santa Cruz). The purified protein was detected using western blotting.

Before PLA, cryosections of embryonic mouse hearts at E10.5 underwent immunohistochemistry with anti-Nkx2.5 antibody (1:200, sc-8697, Santa Cruz) and corresponding secondary antibody conjugated to Alexa Fluor 488 (Invitrogen) for labeling the cardiomyocytes. Duolink kit (Merck) was used for PLA according to the manufacturer's instructions. The following antibodies were used at the indicated concentrations: anti-Sall1 antibody (1:200, PP-K9814-00, Perseus Proteomics), anti-Sall4 antibody (1:200, PP-PPZ0601-00, Perseus Proteomics), anti-Myocd antibody (1:200, SAB4200539, Merck), anti-SRF antibody (1:200, 16821-AP, Proteintech), anti-Myc-tag antibody (1:200, 562, MBL), anti-Myc tag antibody (1:200, 60003-2-Ig, MBL) and rabbit IgG (1:200, 66362, Cell Signaling Technology).

Tissue extracts from the embryonic heart (E9.5) were crosslinked with 1% formaldehyde for 10 min, quenched with 0.125 M glycine for 5 min and stored at −80°C after washing with PBS. The chromatin solution was incubated with anti-H3K27ac-mouse antibody (MAB10309), anti-H3K27me3-mouse antibody (MAB10323), anti-Sall4-mouse antibody (PP-PPZ0601-00, Perseus Proteomics), anti-myocardin-mouse antibody (MAB4028, R&D) and conjugated Magnet-Dynabeads M-280 (Life Technologies 11202D) at 4°C. The immunoprecipitated samples (Mahoro) were eluted from the beads, incubated to reverse crosslinking and purified for DNA analysis. DNA libraries were generated using NEBNext ChIP-Seq (Illumina) from 5 ng of ChIP-DNA and sequenced on NovaSeq (Illumina). Sequence data were trimmed using Trimmomatic (ver.0.38) and mapped using Bowtie2 (ver. 2.3.4.2). Peak calling was performed using settings of MACS2 (ver. 2.2.6) callpeak with a P-value set to 0.005 and data were visualized using igv (2.7.2). GO analyses were performed using GREAT (ver. 4.0.4) with the following association rule: basal+extension: 5000 bp upstream, 1000 bp downstream, 10,000 bp max extension. Preparation of DNA library and raw fastq data were carried out by Rhelixa. (https://www.rhelixa.com).

Ventricles of mouse hearts at E10.5 were collected and dissociated by using Accumax (Innovative Cell Technologies). After dissociation, cells were crosslinked with 0.1% formaldehyde (Cell Signaling Technology) for 10 min, quenched with 10x glycine (Cell Signaling Technology) for 5 min and washed with PBS containing protease inhibitor cocktail (Cell Signaling Technology). A CUT & RUN assay kit (Cell Signaling Technology) was used for extraction of the cleaved DNA fragments associated with transcription factor-binding regions of interest according to the manufacturer's instructions. The following antibodies were used for the primary antibody incubation: anti-Sall4 antibody (1:20; PP-PPZ0601-00, Perseus Proteomics), anti-Myocd antibody (1:10; MAB4028, R&D) and rabbit IgG (1:20; 66362, Cell Signaling Technology). qPCR was performed using Taq Pro Universal SYBR qPCR Master Mix (NIPPON Genetics) on a QuantStudio 3 real-time PCR system (Applied Biosystems). Enrichment levels of Sall4 and Myocd were compared with rabbit IgG (negative control). Samples were normalized with Spike-In DNA (Cell Signaling Technology). The primer sequences are listed in Table S1.

The P-value for the data between two groups was determined using an unpaired two-tailed t-test. Multiple comparison analysis was performed using ordinary one-way ANOVA followed by Šídák's multiple comparisons test. Statistical significance was set at P<0.05. GraphPad Prizm 9.5.1 (GraphPad Software) was used for statistical analysis and creating graphs.

We thank members of the Koshiba lab and the colleagues of the Faculty of Life Sciences at Toyo University for their critical discussions. ChIP was performed with the automated eqipment MAHORO at the Research Core of Tokyo Medical and Dental University (TMDU). We also thank Yumiko Saga and Hiroki Kokubo for providing us with the CAG-CAT plasmid, Manabu Shirai for Nkx2.5-Cre mice and the plasmids for in situ hybridization, Chulan Kwon for the plasmids for in situ hybridization, Akiyasu Iwase and Hiroki Kurihara for help with RNA-seq data analysis, Ryuichi Nishinakamura for Sall1-expression plasmid, Taku Nedachi for instructions on protein assays, and Yusuke Watanabe for instructions on luciferase assay and providing HEK293T cells. We thank the reviewers and editor for critical reading of the manuscript.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201913.reviewer-comments.pdf