Glypicans are heparan sulphate proteoglycans (HSPGs) attached to the cell membrane by a glycosylphosphatidylinositol (GPI) anchor, and interact with various extracellular growth factors and receptors. The Drosophila division abnormal delayed (dally) was the first glypican loss-of-function mutant described that displays disrupted cell divisions in the eye and morphological defects in the wing. In human, as in most vertebrates, six glypican-encoding genes have been identified (GPC1-6), and mutations in several glypican genes cause multiple malformations including congenital heart defects. To understand better the role of glypicans during heart development, we studied the zebrafish knypek mutant, which is deficient for Gpc4. Our results demonstrate that knypek/gpc4 mutant embryos display severe cardiac defects, most apparent by a strong reduction in cardiomyocyte numbers. Cell-tracing experiments, using photoconvertable fluorescent proteins and genetic labeling, demonstrate that Gpc4 ‘Knypek’ is required for specification of cardiac progenitor cells and their differentiation into cardiomyocytes. Mechanistically, we show that Bmp signaling is enhanced in the anterior lateral plate mesoderm of knypek/gpc4 mutants and that genetic inhibition of Bmp signaling rescues the cardiomyocyte differentiation defect observed in knypek/gpc4 embryos. In addition, canonical Wnt signaling is upregulated in knypek/gpc4 embryos, and inhibiting canonical Wnt signaling in knypek/gpc4 embryos by overexpression of the Wnt inhibitor Dkk1 restores normal cardiomyocyte numbers. Therefore, we conclude that Gpc4 is required to attenuate both canonical Wnt and Bmp signaling in the anterior lateral plate mesoderm to allow cardiac progenitor cells to specify and differentiate into cardiomyocytes. This provides a possible explanation for how congenital heart defects arise in glypican-deficient patients.
Heparan sulphate proteoglycans (HSPGs) such as glypican can interact with various growth factors, including Wnt, Bmp, Fgf and Hh, and modulate their activity (Fico et al., 2011). In most vertebrate genomes, six GPC-encoding genes have been identified (GPC1-6). The Drosophila mutant division abnormal delayed (dally) was the first glypican loss-of-function mutant described that displays disrupted cell divisions in the eye and morphological defects in the wing (Nakato et al., 1995), by affecting the signaling capacity and range of Wingless (Wnt related) and Decapentaplegic (Bmp related) (Jackson et al., 1997; Tsuda et al., 1999; Lin and Perrimon, 1999). Thus far, the only described vertebrate genetic loss-of-function models for a glypican are the Gpc3 knock-out mouse and the zebrafish (Danio rerio) gpc4 mutant. Loss of murine Gpc3 results in a variety of embryonic defects, including developmental overgrowth and congenital heart defects (Cano-Gauci et al., 1999; Ng et al., 2009). The zebrafish knypek/glypican4 (kny/gpc4) model was the first gpc4 loss-of-function model to be described, and it is characterized by a shortened anterior-posterior body axis, craniofacial abnormalities and laterality defects (Solnica-Krezel et al., 1996; Topczewski et al., 2001). Thus far, no cardiac phenotypes in kny/gpc4 mutants have been described.
In vertebrates, the multi-chambered heart develops from different pools of cardiac progenitor cells, termed the first and second heart field (FHF and SHF, respectively) [reviewed by Hutson and Kirby (2007); Vincent and Buckingham (2010)]. The FHF refers to a group of mesodermal cells that are the first to differentiate into cardiomyocytes and that form the early linear heart tube. The SHF refers to a region of progenitor cells located in the splanchnic mesoderm that add to the arterial and venous poles of the linear heart tube, leading to growth of the heart (Mjaatvedt et al., 2001; Waldo et al., 2001; Zaffran and Kelly, 2012). The SHF can be subdivided into different domains that are distinguished by gene expression and by their contribution to the heart [reviewed by Vincent and Buckingham (2010); Kelly (2012)]. The anterior part of the SHF expresses Fgf8, Fgf10 (Kelly et al., 2001) and Tbx1 (Xu et al., 2004) and contributes cardiomyocytes to the arterial pole of the heart. Defects in proliferation, migration or differentiation of cells from the anterior SHF result in severe congenital outflow tract defects (Goddeeris et al., 2007). The posterior SHF expresses Hoxb1, Hoxa1, Hoxa3 and Wnt2 (Vincent and Buckingham, 2010; Bertrand et al., 2011). Anterior and posterior SHF cells express Islet1 (Isl1) and Nkx2.5 (Vincent and Buckingham, 2010; Cai et al., 2003; Galli et al., 2008).
In zebrafish, cardiac progenitor cells are specified as two bilateral populations in the anterior lateral plate mesoderm (LPM), where they express nkx2.5, hand2 and gata4 at 12 h post fertilization (hpf) [reviewed by Bakkers (2011)]. Cell-labeling experiments in the anterior LPM at 13 to 14 hpf revealed that medially located nkx2.5+hand2+ progenitors predominantly form ventricle cardiomyocytes, and that the more laterally located nkx2.5−hand2+ progenitors predominantly form atrial cardiomyocytes (Schoenebeck et al., 2007). Consistent with these cell-labeling experiments, lineage-tracing experiments demonstrated that cardiac progenitor cells that constitute the ventricle and the OFT are specified and express gata4 and nkx2.5 in the anterior LPM as early as 15 hpf (Schoenebeck et al., 2007; Guner-Ataman et al., 2013). After fusion of the bilateral progenitor populations at the midline, the linear heart tube is formed around 22-24 hpf. We previously showed that the number of myl7-expressing cardiomyocytes in the heart tube increases rapidly from 24 to 48 hpf, whereas the rate of cardiomyocyte proliferation in the heart tube is very low (de Pater et al., 2009). Using a developmental timing assay and expression of a photoconvertible protein, we demonstrated that the observed increase in myl7+ cardiomyocytes is due to the continuous differentiation of cardiomyocytes at the poles of the heart tube, thus explaining the increase in both atrial and ventricular cardiomyocytes (de Pater et al., 2009). A distinct population of progenitor cells that express nkx2.5, mef2c and latent TGF-β-binding protein 3 (ltbp3) is located anterior to the arterial pole of the linear heart tube (Lazic and Scott, 2011; Zhou et al., 2011; Hinits et al., 2012). Moreover, lineage tracing of ltbp3-expressing cells demonstrated that these progenitors give rise to three cardiovascular lineages (myocardium, endocardium and smooth muscle cells) at the arterial pole (Zhou et al., 2011). Thus far, it remains unclear from which progenitor population the cardiomyocytes accreted to the venous pole of the zebrafish heart are derived. Despite growing evidence that the various parts of the zebrafish heart tube are formed by distinct progenitor populations located in the anterior LPM, there is no clear definition of FHF and SHF, due to a lack of markers that could specifically label these populations. Nevertheless, several signaling pathways (e.g. FGF, Hh and TGF-β) have been identified that regulate the late differentiation of progenitors at the arterial pole in zebrafish (de Pater et al., 2009,, 2012; Hami et al., 2011; Zhou et al., 2011).
Defects in human genes encoding for several members of the glypican protein family have been linked to syndromes that include congenital heart defects. Congenital heart defects are common in Simpson–Golabi–Behmel syndrome (SGBS), an X-linked syndrome caused by mutations in GPC3 or a duplication of GPC4 (Lin et al., 1999; Waterson et al., 2010; Fico et al., 2011). Congenital heart defects have been reported in patients with autosomal-recessive omodysplasia, which is caused by mutations in GPC6. A recent whole-genome association study identified a single nucleotide polymorphism in the GPC5 gene located on chromosome 13q that was associated with Tetralogy of Fallot (TOF), a severe congenital heart defect (Cordell et al., 2013). Several individual cases with 13q deletions that include the GPC5 gene have been reported in patients with TOF (Quelin et al., 2009).
To investigate the role of glypicans during heart development we studied the zebrafish kny/gpc4 mutant. Whereas cardiac tissue is formed in kny/gpc4 mutants, both the atrium and the ventricle are reduced in size. We found that the reduction in chamber size is due to compromised specification and differentiation of cardiomyocytes. Mechanistically, we found that both canonical Wnt and Bmp signaling pathways are overactivated in kny/gpc4 mutants. Whereas inhibiting Bmp signaling in kny/gpc4 mutants restored cardiomyocyte differentiation, overexpression of the Wnt antagonist Dkk1 restored both cardiomyocyte specification and differentiation in kny/gpc4 mutants. Thus, our results provide a new role for Glypican4 as a regulator of canonical Wnt and Bmp signaling during cardiac differentiation. Furthermore, our study gives a potential explanation for how congenital heart defects in glypican-deficient patients can occur.
Loss of knypek/glypican4 results in reduced cardiomyocyte numbers
In kny/gpc4 mutant embryos, the embryonic anterior-posterior axis is reduced and they develop cardiac edema at 2 days post fertilization (dpf) (Fig. 1A,B). To investigate heart development in kny/gpc4 mutants we carried out in situ hybridizations for various myosin genes and observed reduced heart looping at 58 hpf, combined with a reduction in atrial and ventricular chamber size (supplementary material Fig. S1A-F). Atrial and ventricular markers were expressed in kny/gpc4 mutant embryos, indicating that the cardiac chambers were specified. To determine the possible cause for the reduced cardiac chamber size in kny/gpc4 mutants, we quantified the number of myl7-expressing cardiomyocytes by generating a Tg(myl7:galFF/UAS:h2a-gfp) line in which cardiomyocytes express a nuclear green fluorescent protein (GFP) (Fig. 1D-F). Quantification of myl7+ cardiomyocyte numbers revealed significantly reduced cell numbers in both the ventricle and atrium of kny/gpc4 mutant hearts. Mutant hearts displayed an average number of 75±6 myl7+ cardiomyocytes in the atrium and 71±10 cells in the ventricle, whereas wild-type sibling hearts contained an average of 129±10 myl7+ cells in the atrium and 134±14 cells in the ventricle (Fig. 1G,H). In kny/gpc4 mutant embryos, the early gastrulation defects were rescued by injection of synthetic gpc4 mRNA (Fig. 1C). As synthetic mRNA is rapidly degraded in the embryo, late phenotypes are more difficult to rescue by this method. Indeed, we observed that the reduction in myl7+ cardiomyocyte numbers was not rescued by gpc4 mRNA injection (Fig. 1I). This result suggests that the reduced number of cardiomyocytes in kny/gpc4 mutant embryos is not secondary to the gastrulation defects also observed in the mutant embryos. Together, these results demonstrate that Glypican4 is required to achieve the correct number of cardiomyocytes and normal cardiac size, and they suggest that this function is independent from its role in regulating cell movements during gastrulation.
Reduced cardiomyocyte accretion in kny/gpc4 mutants
To determine the cause for the reduced cardiomyocyte numbers in kny/gpc4 mutants, we first analyzed when and where kny/gpc4 is expressed in the embryo. As previously reported, we observed a ubiquitous expression, including in the anterior LPM, where cardiac progenitors reside (supplementary material Fig. S2) (Topczewski et al., 2001). At 24-48 hpf, kny/gpc4 expression was more restricted to the eyes, brain and pharyngeal arch region, but was absent from the heart tube. To address whether there is elevated apoptosis in the heart of kny/gpc4 mutants, we incubated the embryos with Acridine Orange staining (supplementary material Fig. S3). We did not detect Acridine Orange+ cells in the hearts of kny/gpc4 mutants, suggesting that the smaller hearts of kny/gpc4 mutants are not caused by an elevated rate of apoptosis. Next, we investigated at which stage of development the reduction in cardiomyocyte numbers was most apparent. At 19 hpf, the cardiac field forms a disk-like epithelial structure that gives rise to the linear heart tube. We observed a moderate but significant reduction of myl7+ cardiomyocytes at 19 hpf in kny/gpc4 mutants (141±6 cells in wild type versus 123±4 cells in kny/gpc4 mutants), consistent with a mild reduction in the expression of nkx2.5, gata4 and hand2 at 13 hpf (Fig. 2A; supplementary material Fig. S4). From 19 hpf, the linear heart tube is formed, and we previously showed that cardiomyocyte numbers increase rapidly due to the accretion of myl7+ cells to the arterial and venous poles (de Pater et al., 2009). Interestingly, we observed a fourfold reduction in the accretion of myl7+ cardiomyocytes in kny/gpc4 mutants compared with their wild-type siblings (Fig. 2A). We observed an increase of 144±3 myl7+ cardiomyocytes in wild-type embryos between 19 and 48 hpf, whereas the observed increase in myl7+ cells was only 31±11 in kny/gpc4 mutants. Next, we analyzed whether the accretion of myl7+ cells is affected at both the venous and arterial pole by photoconversion of a fluorescent protein (Kaede or nlsKikGR) expressed in myl7+ cardiomyocytes (de Pater et al., 2009; Lazic and Scott, 2011). Using tg(myl7:gal4;UAS:Kaede) or tg(myl7:KikGr) embryos, myl7+ cardiomyocytes at a specific stage can be visualized by photoconversion of the fluorescent protein from green to red. Due to the continuous expression of the fluorescent protein in myl7-expressing cardiomyocytes, both the green and the photoconverted red fluorescent proteins are observed. Cardiomyocytes that initiate myl7 expression after the photoconversion event only have the green fluorescent protein. We observed that when the Kaede photoconversion was performed at the linear heart tube stage (24-26 hpf) and analyzed at 48 hpf, green-only cardiomyocytes were detected in both the arterial pole and the venous pole of the wild-type looped heart (Fig. 2B). Interestingly, in kny/gpc4 mutant hearts the regions of green-only myocardium at both the arterial and venous poles are reduced (Fig. 2C). To quantify the number of green-only cells, we repeated the experiment with the nuclear KikGr and observed a nearly threefold reduction in the number of green-only cardiomyocytes at the arterial pole (Fig. 2D). The weakness of the KikGr expression at the venous pole allowed only a qualitative observation (Fig. 2C). Together, these results demonstrate that Glypican4 is required for growth of the heart tube by regulating the accretion of myl7-expressing cardiomyocytes to both the venous and arterial poles.
Knypek/Glypican4 regulates the nkx2.5+ progenitor pool and their differentiation towards myl7+ cardiomyocytes
Previous work by various research groups has demonstrated that cells which are located in a region posterior to the arterial pole and express nkx2.5, ltbp3 and mef2cb contribute to cardiac cells of the arterial pole (Lazic and Scott, 2011; Zhou et al., 2011; Hinits et al., 2012). We therefore analyzed the expression of these genes and found that they are still expressed in kny/gpc4 mutant embryos (supplementary material Fig. S5). To visualize cardiac progenitor cells in living embryos, we generated a Tg(nkx2.5BAC:galFF/ UAS:h2a-gfp) line in which all nkx2.5-expressing cells are labeled with a nuclear eGFP (further referred to as nkx2.5:GFP) (supplementary material Fig. S6A,B). To analyze the differentiation of progenitor cells into cardiomyocytes, we combined Tg(myl7:dsred, kny+/−) and Tg(nkx2.5BAC:galFF/UAS:h2a-gfp, kny+/−), resulting in cardiomyocytes that are labeled in green (nkx2.5:GFP) and red (myl7:dsred) in the ventricle of 48-hpf embryos (Fig. 3A,B). To distinguish cardiomyocytes located in the ventricle from those in the atrium, an antibody recognizing an atrial myosin was used. We observed that at 48 hpf, the number of both nkx2.5+ and myl7+ cells were decreased in kny/gpc 4 mutants (88±9 nkx2.5+ cells and 64±13 myl7+ cells versus 126±10 nkx2.5+ cells and 135±6 myl7+ cells in wild-type siblings, respectively; Fig. 3C). This is consistent with the reduced nkx2.5 expression in kny/gpc4 mutants at 13 hpf (supplementary material Fig. S4) and suggests that Glypican4 is required for the efficient specification and/or proliferation of cardiac progenitor cells. In addition, we observed that whereas in wild-type embryos the number of GFP+ to DsRed+ cells was almost equal (ratio of 0.93±0.06), this ratio was significantly increased to 1.4±0.13 in kny/gpc4 mutants (Fig. 3D). nkx2.5+ cells in the ventricle that fail to differentiate and initiate myl7 expression might explain this increase in GFP+/DsRed+. Indeed, in ventricles of kny/gpc4 mutants, we observed clusters of cells that expressed nkx2.5:GFP and not myl7:dsred, and these GFP-only clusters were surrounded by GFP/DsRed double-labeled cells (Fig. 3E-E″,F-F″). Such nkx2.5:GFP-only cells surrounded by GFP/DsRed double-labeled cells were only sparsely observed in the ventricle of wild-type siblings. Corroborating a model in which cardiomyocyte differentiation of nkx2.5+ cells is compromised in kny/gpc4 mutants, quantitative RT-PCR revealed a strong downregulation of myl7 expression, whereas only a modest downregulation of nkx2.5 expression was observed in kny/gpc4 mutants at the heart-looping stage (supplementary material Fig. S6D). From these results we conclude that Glypican4 is required for efficient specification/proliferation of nkx2.5+ progenitor cells and their differentiation into myl7+ cardiomyocytes.
Elevated BMP signaling in kny/gpc4 mutants compromises cardiomyocyte differentiation
To address the molecular mechanism by which Glypican4 regulates cardiac specification and differentiation, we first investigated whether BMP activity is affected in kny/gpc4 mutants. HSPGs such as glypicans can interact with extracellular BMP, and BMP signaling is required for the induction of nkx2.5 expression, whereas inhibition of BMP signaling promotes the differentiation of cardiac progenitor cells (Schultheiss et al., 1997; Fujise et al., 2003; Belenkaya et al., 2004; Andrée et al., 1998; Schlange et al., 2000; Yuasa et al., 2005; Matsui et al., 2005; Taha et al., 2007; Hao et al., 2008; de Pater et al., 2012; Cai et al., 2013). We analyzed Bmp signaling activity in the LPM, using a Bmp reporter line Tg(bre:gfp) crossed into the kny/gpc4 mutant. We observed specifically in mutants that individual cells within the cardiac field at 19 hpf expressed high levels of GFP, which was not seen in the cardiac field of wild-type siblings (Fig. 4A-D,H). This observation indicates that Bmp signaling is ectopically activated in the cardiac field of kny/gpc4 mutants. In addition, we observed ectopic GFP+ cells in the lateral plate mesoderm (Fig. 4E′-G′).
As inhibition of Bmp signaling in cardiac progenitor cells is necessary to promote cardiomyocyte differentiation (Yuasa et al., 2005; Matsui et al., 2005; Taha et al., 2007; Hao et al., 2008; de Pater et al., 2012; Cai et al., 2013), we hypothesized that elevated Bmp signaling in the anterior LPM is responsible for the reduction of cardiomyocyte numbers in kny/gpc4 mutants. To test this hypothesis, we crossed the Tg(hsp70:nog), in which the Bmp antagonist Noggin3 is induced upon heat shock, into the kny/gpc4 mutant line. Induction of Noggin3 at 16 hpf in wild-type embryos results in a moderate reduction in atrial cardiomyocytes, whereas the number of ventricular cardiomyocytes remains unaffected, probably reflecting the temporal differences in the specification of atrial versus ventricular progenitor cells (de Pater et al., 2012). Corroborating our hypothesis, induction of Noggin3 at 16 hpf in kny/gpc4 mutants resulted in a modest but significant increase in myl7+ ventricular cardiomyocytes (Fig. 5A-C). Moreover, when combining the Tg(hsp70:nog) with the Tg(nkx2.5:galFF;UAS:h2a-gfp) and the Tg(myl7:dsred), we observed that although the number of myl7+ cardiomyocytes in kny/gpc4 mutants increased, there was no effect on the number of nkx2.5+ cells as a consequence of Noggin3 overexpression (Fig. 5D). As a result, Noggin3 overexpression restored the abnormally high ratio of GFP+/DsRed+ cells in kny/gpc4 mutants from 1.4±0.12 to 0.8±0.02 (Fig. 5E). Together, these results suggest that blockage of excessive Bmp signaling in kny/gpc4 mutants does not affect cardiomyocyte specification but leads to a restoration of cardiomyocyte differentiation.
Inhibiting canonical Wnt signaling restores cardiomyocyte numbers in kny/gpc4 mutants
HSPGs can also bind different Wnt ligands and the Wnt-promoting factor R-spondin, and affect both canonical as well as non-canonical Wnt signaling (Topczewski et al., 2001; Ohkawara et al., 2011; Sakane et al., 2012). To investigate whether canonical Wnt signaling in kny/gpc4 mutants was affected, we performed in situ hybridizations for axin2, a direct target gene of canonical Wnt signaling. Interestingly, we observed an upregulation of axin2 expression in the anterior lateral plate mesoderm of kny/gpc4 mutants (Fig. 6A-D).
We have previously shown that Wnt signaling is upstream of BMP signaling in the heart of 48 hpf embryos (Verhoeven et al., 2011). To test whether, in kny/gpc4 mutants, the ectopic BMP activation observed at 20 hpf could be the consequence of enhanced canonical Wnt signaling, we crossed the Tg(hsp70:wnt8a-gfp) line with a BMP reporter line expressing an orange fluorescent protein, Tg(bre:dmKO2), and heat-shocked embryos at 16 hpf (Collery and Link, 2011). In contrast to later stages of heart development, we did not detect enhanced bre:dmKO2 activity in the cardiac disk or LPM of heat-shocked wnt8-gfp embryos (supplementary material Fig. S7), indicating that at 16 hpf, ectopic Wnt signaling does not induce Bmp signaling.
Canonical Wnt signaling can have both positive and negative effects on cardiac specification and differentiation, dependent on the developmental stage when being induced (Manisastry et al., 2006; Verma and Lenka, 2010; Dohn and Waxman, 2012; Willems et al., 2011). Using the heat shock-inducible Wnt8 line Tg(hsp70:wnt8-gfp), we indeed observed that induction of Wnt8 at 16 hpf reduced the number of ventricular cardiomyocytes (supplementary material Fig. S8C-E). To address whether this was due to compromised cardiac specification or differentiation, we combined the Tg(hsp70:wnt8-gfp) with the Tg(nkx2.5:galFF;UAS:h2a-gfp) and the Tg(myl7:dsred), and observed that both nkx2.5+ cells and myl7+ cells were reduced as a consequence of ectopic Wnt8 expression (supplementary material Fig. S8F). As the ratio of nkx2.5+/myl7+ cells in the ventricle was also affected by ectopic Wnt8 expression, these results indicate that both cardiomyocyte specification and differentiation are affected by ectopic canonical Wnt signaling at 16 hpf (supplementary material Fig. S8G). To test the hypothesis that in kny/gpc4 mutants, the observed reduction in cardiomyocyte specification and differentiation is due to an upregulation of canonical Wnt signaling, we aimed to restore cardiomyocyte numbers in kny/gpc4 mutants by blocking canonical Wnt signaling. Temporally controlled inhibition of canonical Wnt signaling was achieved using the previously described heat shock-inducible Tg(hsp70:dkk1-gfp) line (Stoick-Cooper et al., 2007). Corroborating our hypothesis, inhibition of canonical Wnt signaling in kny/gpc4 mutants by overexpression of the Wnt inhibitor Dkk1 at 16 hpf resulted in a significant increase in the number of myl7+ cardiomyocytes in both atrium and ventricle (Fig. 6E-G). Overexpression of Dkk1 in kny/gpc4 mutants at 24 hpf had no effect on the number of myl7+ cardiomyocytes (Fig. 6G). From these results, we conclude that Glypican4 is required to attenuate canonical Wnt signaling at 16 hpf, which is required to allow sufficient cardiomyocytes to be specified. Whereas inhibiting canonical Wnt signaling in kny/gpc4 mutant embryos rescued cardiomyocyte numbers, cardiac morphology was still affected. Heart looping was still disturbed in kny/gpc4 mutants after inhibition of Wnt signaling, and the atrial-ventricular canal was still broader than in wild-type siblings (Fig. 6F), indicating that Glypican4 has additional roles in cardiac development independent from canonical Wnt signaling.
Whereas defects in human genes encoding for several members of the glypican protein family have been linked to syndromes that include congenital heart defects, very little was known about their role during heart development. Our genetic and embryological analysis of Glypican4-deficient zebrafish embryos demonstrates that Glypican4 is required to regulate cardiomyocyte formation in the anterior LPM. As a result of Glypican4 deficiency there are fewer nkx2.5+cells in the embryonic heart and even fewer differentiated myl7+ cardiomyocytes, providing an explanation for the small hearts of kny/gpc4 mutants. Mechanistically, we propose that Glypican4 is required to attenuate both canonical Wnt and Bmp signaling in the anterior LPM and that this attenuation is required to ensure the formation of sufficient nkx2.5+ cardiac progenitor cells and their differentiation into cardiomyocytes (Fig. 6I). Glypicans, as other HSPGs, are decorated with long heparin chains that interact extracellularly with various secreted growth factors, including Wnt and BMP (Desbordes and Sanson, 2003; Williams et al., 2010). Such interactions can control growth factor signaling by regulating growth factor availability or facilitating interactions with (co)-receptors (see below) (Fig. 6H). Our results suggest that Glypican4 attenuates both Wnt and BMP signaling independently, possibly by preventing their release from the ECM and interaction with their receptor complex during cardiac specification and differentiation.
Here, we show that loss of Glypican4 results in a local enhancement of Bmp signaling and that inhibiting Bmp signaling in kny/gpc4 mutants can rescue the cardiomyocyte differentiation defects by restoring the normal ratio of nkx2.5+ and myl7+ cells in the ventricle. Interestingly, inhibition of BMP signaling in kny/gpc4 mutants did not restore the number of nkx2.5+ cells, suggesting that other functions of Glypican4 (e.g. attenuating Wnt signaling) regulate this process. Indeed, BMP signaling and Wnt signaling have distinct roles in regulating SHF specification and differentiation in the anterior LPM of mouse embryos (Klaus et al., 2012). Our observation that Glypican4 is required to attenuate BMP signaling to allow proper differentiation of myl7+ cardiomyocytes is consistent with a model in which BMP signaling is required to induce nkx2.5 expression in the anterior LPM, but that BMP signaling in cardiac progenitor cells needs to be downregulated to allow efficient cardiomyocyte differentiation (Hao et al., 2008; de Pater et al., 2012; Cai et al., 2013). Glypicans have been implicated in the control of TGF-β/Bmp activity (Jackson et al., 1997; Paine-Saunders et al., 2000; Li et al., 2004). The Drosophila glypican Dally is required for the modulation of the signaling strength of the TGF-β/Bmp-superfamily member Decapentaplegic (Dpp) (Fujise et al., 2003; Belenkaya et al., 2004). It could thus be hypothesized that Glypican4 attenuates BMP signaling through physical interaction of the heparin chains with the BMP growth factor in the extracellular space, and subsequently prevents it from activating Smad protein phosphorylation in the cardiac progenitor cells.
Besides a role in attenuating BMP signaling, our results also indicate that Glypican4 is required to attenuate canonical Wnt signaling. A role for glypicans in regulating canonical Wnt signaling was shown in Drosophila, demonstrating that gpc4 homolog dally-like is essential for Wg gradient formation (Han et al., 2005). As proposed for BMP signaling, one possible hypothesis is that Glypican4 affects the extracellular distribution of Wnt growth factors. Alternatively, Glypican4 might interact with specific components of the Wnt signaling pathway. Corroborating such a model is the observation that Glypican4 can modulate Wnt/β-catenin and Wnt/PCP signaling pathways through interactions with Dkk1 and LRP5/6 (Caneparo et al., 2007). Moreover, in more recent studies it was shown that glypicans are required for the Wnt-inhibiting activity of Wnt Inhibitory Factor-1 (WIF-1) (Avanesov et al., 2012). In vitro, the Wnt-promoting factor R-spondin 3 was shown to interact with Syndecans and Glypican3 (Ohkawara et al., 2011), suggesting a complex interaction of heparan sulphate proteoglycans and Wnt signaling components at the cell membrane. Interestingly, human SGBS patients have a tendency to develop pre- and postnatal tumors, such as Wilm's tumor, hepatoblastomas, neuroblastomas and gonadoblastomas (Golabi et al., 1993; Rodríguez-Criado et al., 2005; Thomas et al., 2012; Mateos et al., 2013). The development of these tumors has been brought into correlation with a misregulation of canonical Wnt signaling (Honecker et al., 2004; Lapunzina, 2005; Liu et al., 2008; Armengol et al., 2011), indicating that the lack of glypicans might result in a misregulation of canonical Wnt signaling in SGBS patients. It remains to be addressed how Glypican4 attenuates Wnt signaling in the anterior LPM during cardiac specification and differentiation.
Although we observed that cardiomyocyte numbers in kny/gpc4 mutants were fully restored by inhibiting canonical Wnt signaling (Fig. 6G), we also observed that heart morphology was still affected in these rescued embryos (Fig. 6F). The hearts of the rescued mutants displayed reduced looping and an enlarged atrial-ventricular canal, indicating that Glypican4 has other functions during heart development independent of canonical Wnt signaling. Besides the canonical Wnt pathway, Glypican4 also affects non-canonical Wnt signaling (Topczewski et al., 2001). Canonical and non-canonical Wnt signaling pathways can be simultaneously controlled by Syndecan-4, which inhibits canonical Wnt signaling through interaction with LRP6 and R-spondin 3 and activates the non-canonical Wnt pathway (Ohkawara et al., 2011; Astudillo et al., 2014). The importance of non-canonical Wnt signaling for cardiac development has been shown in numerous studies (Phillips et al., 2005; Schleiffarth et al., 2007; Zhou et al., 2007; Ramsbottom et al., 2014). Whether Glypican4 also affects non-canonical Wnt signaling in the anterior LPM is not addressed here.
Various studies have linked genetic alterations in glypican-coding genes with congenital heart defects. Heart defects in SGBS patients are very diverse and include defects of the outflow tract (transposition of the great arteries, pulmonary artery stenosis, aortic stenosis and patent ductus arteriosus) (Lin et al., 1999). The outflow tract derives from the SHF and defects in SHF differentiation lead to severe congenital outflow tract defects (Goddeeris et al., 2007). Nucleotide polymorphisms in GPC5 have been brought into correlation with the occurrence of TOF (Quelin et al., 2009; Cordell et al., 2013). TOF patients display defects in SHF-derived structures (pulmonary stenosis, overriding aorta, ventricular septal defects) [reviewed by Page (1986)]. Together with our results presented here, this suggests that glypicans also have an important role in attenuating BMP and Wnt signaling to regulate cardiac outflow tract development in the anterior LPM of human embryos.
MATERIALS AND METHODS
All animal experiments were conducted under the guidelines of the animal welfare committee of the Royal Netherlands Academy of Arts and Sciences (KNAW).
Generation of transgenic lines
The I-SceI-myl7:galFF-SV40 construct was obtained by ClaI-XbaI cohesive-end cloning of the galFF sequence from pT2KhspGFF (Asakawa and Kawakami, 2008) into I-SceI-myl7-SV40 (a kind gift from S. Abdelilah-Seyfried, University of Potsdam, Germany), containing about 900 bp of the myl7 promoter. Plasmid DNA was injected at 20 ng/μl in the presence of 0.75U I-SceI meganuclease (New England Biolabs) into Tg(UAS:rfp) embryos (Asakawa and Kawakami, 2008). Healthy embryos displaying robust myl7-specific RFP fluorescence were grown to adulthood and screened for germline transmission, resulting in establishment of the Tg(myl7:galFF/UAS:rfp) transgenic line.
The Tg(myl7:galFF/UAS:h2a-gfp) line was generated by injecting plasmid plm262 at 20 ng/μl, a h2a-gfp in-frame version derived from plm74 (Koole and Tijsterman, 2014), in the presence of 25 ng/μl tol2 mRNA into Tg(myl7:galFF/UAS:rfp) embryos. Healthy embryos displaying robust myl7-specific h2a-gfp fluorescence were grown to adulthood, resulting in establishment of the Tg(myl7:galFF/UAS:h2A-gfp) transgenic line.
Recombineering of BAC clone DKEY-9I15 was performed following the manufacturer's protocol with minor modifications, as described in Bussmann and Schulte-Merker (2011). Primers used were nkx2.5_Gal4FF_F: 5′-CATAGGTACCTCTGATAAATCCCAACCGGATTATCCAAGTGG-ACCATCATACCATGAAGCTACTGTCTTCTATCG-AAC-3′ and nkx2.5_KanR_R: 5′-ATGTCCCGCACTGAGAAAGGAGTGGAAGTCATTTGGCTAGAGAACATTGCTCAGAAGAACTCGTCAAGA-AGGCG-3′.
BAC DNA was injected at 130 ng/μl in the presence of 25 ng/μl tol2 mRNA into Tg(UAS:gfp) embryos (Asakawa and Kawakami, 2008). Healthy embryos displaying robust nkx2.5-specific fluorescence were grown to adulthood and screened for germline transmission, resulting in the establishment of the Tg(nkx2.5BAC:galFF/UAS:gfp) line. Subsequent outcross to Tg(myl7:galFF/UAS:h2a-gfp) carriers and selection of embryos with nuclear-specific GFP fluorescence in the nkx2.5 expression domain resulted in the establishment of the Tg(nkx2.5BAC:galFF/UAS:h2a-gfp) transgenic line.
In this study, the following zebrafish lines were used: kny/gpc4 mutants that carried the knyfr6 allele (Topczewski et al., 2001). Tg(myl7:EGFP)twu26 (Huang et al., 2003), hereafter referred to as Tg(myl7:gfp), Tg(nkx2.5:galFF/UAS:h2a-gfp) as described above, Tg(myl7:galFF, UAS:Kaede) and Tg(myl7:nlsKikGR)hsc6 (Lazic and Scott, 2011), hereafter referred to as Tg(myl:nlsKikGR), Tg(hsp70:dkk1-gfp) (Stoick-Cooper et al., 2007), Tg(-5.1myl7:nDsRed2)f2 (Mably et al., 2003), referred to as [(Tg(myl7:dsred)], Tg(hsp70:wnt8a-gfp) (Weidinger et al., 2005), Tg(Bre:GFP)p77 (Alexander et al., 2011), Tg(bre:dmKO2) fish (Collery and Link, 2011) and Tg(hsp70I:Nog3)fr14 (Chocron et al., 2007).
Wild-type embryos and kny/gpc4 mutants were injected with 5 ng/μl of kny/gpc4 mRNA into the yolk at the one-cell stage. Embryos were grown until 48 hpf and genotyped for the kny/gpc4 mutation.
Acridine Orange staining
Embryos were incubated with a 2 μg/ml Acridine Orange solution in E3 medium for 45 min in the dark, rinsed three times in E3 medium and photographed live.
In situ hybridization
In situ hybridizations were carried out as described in Thisse and Thisse (2008). For double in situ hybridizations, the antibody was removed by washing embryos in 0.1 M glycin/HCl supplemented with 0.1% Tween, and afterwards in PBST. Embryos were incubated with anti-fluorescein antibody (Roche, 11426320001; 1:100) overnight. Following extensive washing with PBST and TBST, embryos were subjected to 2,4-iodophenyl-3-4-nitrophenyl-5-phenyl-tetrazolium chloride (INT) with BCIP (Roche), diluted in TBST until staining had developed. The embryos were fixed overnight in 4% PFA. Embryos were washed in PBST and mounted in glycerol before imaging. Riboprobes were transcribed from linearized template in the presence of digoxigenin- or fluorescein-11-UTP.
Cells were counted by using the Tg(myl7:galFF/UAS:h2a-gfp), the Tg(-5.1myl7:nDsRed2)f2 (Mably et al., 2003) or the Tg(nkx2.5BAC:galFF/UAS:h2a-gfp) lines. Embryos were stained with anti-DsRed antibody (Clontech, 632496; 1:200) or anti-GFP antibody (rabbit anti-GFP, Torrey Pines Biolabs, TP401; 1:200; or anti-chicken GFP, Abcam, ab13970; 1:200) and anti-Myh6 antibody (S46 from the Hybridoma bank, 1:200). To count nkx2.5+ cells expressing gfp in Tg(hsp70:wnt8a-gfp) fish, embryos were counterstained with DAPI after the antibody staining. The nuclear-specific DAPI staining was used to facilitate the counting of the nkx2.5+ cells. Images were acquired by confocal microscopy. Three-dimensional (3D) reconstruction and cell counting were performed using Volocity software. Each cell-counting experiment was carried out on at least three biological repeats.
Total RNA was extracted from wild-type and kny/gpc4 mutant embryos. cDNA was generated using the RTIII First Strand kit (Invitrogen). Quantitative PCRs were performed using SYBR Green (BioRad) in a BioRad MyIQ PCR machine. Gene expression levels were normalized to gapdh and ef1a levels and the ratio between wild-type and kny/gpc4 mutant expression levels was calculated. Three biological repeats were used for each genotype and reactions were performed in triplicates. Primers used in this study were nkx2.5_F set 1: 5′-ACCTACAACACCTACCCTGCG-3′ and nkx2.5_R set 1: 5′-ACTGTGAAGGTTGGATGCTGGACA-3′ for qPCRs at 48 hpf and for the first biological experiment at 10 somite stage. For the second and third biological repeats at 10 somite stage, the following set was used: nkx2.5_F set 2: 5′-CAACACCTACCCTGCGTTTA-3′, nkx2.5_R set 2: 5′-GTTGGACTGTGAAGGTTGGA-3′, gata4_F: 5′-TCTCGCCATGAAGAAAGAGGG-3′, gata4_R: 5′-TGTGTGTGTAGAGCTGGTGTG-3′, hand2_F: 5′-CCAGTCGCTGTCATGAAGAA-3′, hand2_R: 5′-GTACGAAGGTGCCATGGTATAG-3′, cmlc2_F: 5′-CTGTCTTCCTCACCCTCTTTG-3′ and cmlc2_R: 5′-TGACAACTCCTGTGGCATTAG-3′.
Photoconversion of Kaede or nlsKikGR fluorescence from green to red was achieved by exposing transgenic embryos to a 405 nm laser, using a confocal microscope (Leica, TCS SPE) that allowed measuring the development of red fluorescence. Exposure was stopped when the red fluorescence had reached a maximum peak. Images were acquired by confocal microscopy. 3D reconstruction and cell counting were performed using Volocity software. The cell counting was performed on three biological repeats.
Vibratome sections were used to visualize bre:gfp+ cells. Embryos were embedded in 3% agarose and 1% gelatin in PBS. Sections (50 μm) were stained overnight with primary antibody and secondary antibody overnight. Agarose slices were counterstained for 1 h with DAPI (Invitrogen, 1:2500). Antibody against tropomyosin (Sigma, T9283; 1:200) was used to visualize cardiac tissue. In order to visualize axin2-expressing cells, Tg(nkx2.5:galFF/UAS:gfp) embryos underwent in situ hybridization for axin2 and subsequent staining with anti-gfp antibody (anti-rabbit GFP, Torrey Pines Biolabs, TP401; 1:200), followed by incubation with an anti-rabbit biotinylated secondary antibody (Vector Laboratories, BA-1000; 1:200). For sectioning, embryos were mounted in Technovit 8100 (Heraeus Kulzer) and sectioned at 7 μm.
Pixel intensity analysis
Mean gray values of GFP signal were measured within cardiac tissue, using histological cross-sections with the help of ImageJ (NIH). Mean gray values were measured in two wild-type and two mutant samples. The analysis was carried out on 14 cardiomyocytes in wild type and kny/gpc4 mutants each. The amount of cells that displayed mean gray values above the 1.5-fold average value for wild-type cells were counted.
Embryos were heat-shocked by transferring them to E3 medium preheated to 37°C and incubation at 37°C for 30 min to induce the hsp70:noggin transgene, and for 45 min in case of hsp70:wnt8a-gfp and hsp70:dkk1-gfp transgenes. Cell counts were performed at 48 hpf, when the GFP signal from hsp70:wnt8a-gfp and hsp70:dkk1-gfp transgenes had ceased.
We would like to thank S. Chocron for technical assistance and E. Noël for critical reading of the manuscript.
I.S. and J.B. conceived the research. F.T. generated the zebrafish transgenic lines. I.S. performed the experiments and analyzed the data together with J.B. I.S., F.T. and J.B. wrote and/or revised the manuscript.
I.S. was funded by the Deutsche Forschungsgemeinschaft (DFG) [grant: STR1225/1-1]. The work in J.B.'s laboratory was supported by the Netherlands Cardiovascular Research Initiative and the Dutch Heart Foundation [grant: Hustcare CVON 2011-2] and ZonMW [grant: 912.12.086].
The authors declare no competing or financial interests.