DiGeorge Syndrome Gene tbx1 Functions through wnt11r to Regulate Heart Looping and Differentiation

DiGeorge syndrome (DGS) is the most common microdeletion syndrome, and is characterized by congenital cardiac, craniofacial and immune system abnormalities. The cardiac defects in DGS patients include conotruncal and ventricular septal defects. Although the etiology of DGS is critically regulated by TBX1 gene, the molecular pathways underpinning TBX1's role in heart development are not fully understood. In this study, we characterized heart defects and downstream signaling in the zebrafish tbx1−/− mutant, which has craniofacial and immune defects similar to DGS patients. We show that tbx1−/− mutants have defective heart looping, morphology and function. Defective heart looping is accompanied by failure of cardiomyocytes to differentiate normally and failure to change shape from isotropic to anisotropic morphology in the outer curvatures of the heart. This is the first demonstration of tbx1's role in regulating heart looping, cardiomyocyte shape and differentiation, and may explain how Tbx1 regulates conotruncal development in humans. Next we elucidated tbx1's molecular signaling pathway guided by the cardiac phenotype of tbx1−/− mutants. We show for the first time that wnt11r (wnt11 related), a member of the non-canonical Wnt pathway, and its downstream effector gene alcama (activated leukocyte cell adhesion molecule a) regulate heart looping and differentiation similarly to tbx1. Expression of both wnt11r and alcama are downregulated in tbx1−/− mutants. In addition, both wnt11r −/− mutants and alcama morphants have heart looping and differentiation defects similar to tbx1−/− mutants. Strikingly, heart looping and differentiation in tbx1−/− mutants can be partially rescued by ectopic expression of wnt11r or alcama, supporting a model whereby heart looping and differentiation are regulated by tbx1 in a linear pathway through wnt11r and alcama. This is the first study linking tbx1 and non-canonical Wnt signaling and extends our understanding of DGS and heart development.


Introduction
DiGeorge syndrome (DGS) is the most common microdeletion syndrome occurring in 1/4000 live births [1]. Approximately 75-80% of patients have congenital heart disease with conotruncal defects (Tetralogy of Fallot, aortic arch defects, and truncus arteriosus) and ventricular septal defects. Other defects include thymic hypoplasia, palate defects and thyroid and parathyroid abnormalities. Cardiac defects are the leading cause of mortality in DGS, but the underlying molecular pathobiology is not well understood.
Most DGS patients have a 3 Mb or a nested 1.5 Mb deletion of chromosome 22q11.2 that includes the TBX1 gene. Genetically engineered mouse mutants have led to the identification of Tbx1 as the gene responsible for cardiovascular and thymic defects [2,3,4]. Several patients without the chromosomal deletion harbor frameshift and missense mutations in TBX1, strongly suggesting that DGS is caused by haploinsufficiency of TBX1 [5,6]. TBX1 is a member of a group of transcription factors that are characterized by the presence of a T-box, a highly conserved DNA-binding region that also has a conserved interaction domain for other transcription factors [7]. T-box genes mediate transcriptional activation and/or repression and are important for embryonic development and differentiation of all three germ layers [8,9,10]. In addition, T-box genes are extremely dose-sensitive and often act in a combinatorial or hierarchical fashion [7].
Dose-sensitivity of mouse Tbx1 gene has been tested using various hypomorphic and null alleles and transgenic models [3,11,12]. While Tbx1+/2 mice have a milder phenotype, Tbx12/2 mice have a severe phenotype with single cardiac outflow tract (OFT), aortic and pharyngeal arch defects, thymus and parathyroid gland aplasia [2]. Tissue-specific knockdown of Tbx1 in the pharyngeal endoderm or mesoderm recapitulates the cardiac defects observed in Tbx12/2 mice and DGS patients, making it difficult to study the role of Tbx1 separately in different tissues [13,14]. Previous studies have identified several genes, including Fgf10, as part of the Tbx1 pathway [12]. However, validation and detailed description of their roles downstream of Tbx1 during development is lacking.
While non-canonical Wnt signaling has no known link to Tbx1, it inhibits b-catenin signaling and promotes its own signaling through protein kinase C and c-jun terminal kinase during normal heart morphogenesis [15,16,17,18]. Wnt11, a non-canonical Wnt member, is required for heart specification and can induce expression of cardiac genes in Xenopus explants and non-cardiac cells from humans and mouse [15,19,20,21,22,23,24]. Further-more, mutations in Wnt11 cause cardiac OFT defects such as truncus arteriosus, similar to those observed in Tbx12/2 mutants [25]. Knockdown of wnt11r, an ortholog of Wnt11 present in Xenopus, resulted in heart morphology defects and cardia bifida in some cases [15]. In spite of the similarity in phenotypes caused by loss of Tbx1 and Wnt11 signaling, no link between these pathways has been established to date.
The pattern of heart development is conserved through evolution. Development begins with specification of the bilateral Primary Heart Fields (PHF) in lateral plate mesoderm. Subsequent migration towards the midline results in the cardiac crescent and later the linear heart tube, at which point the heart starts beating. The tube then undergoes asymmetric looping and morphogenetic movements to result in the multi-chambered heart. This complex process involves ballooning of the chamber walls, addition of cardiomyocytes derived from the Secondary Heart Field (SHF) at the arterial pole and formation of septa and valves. Formation of heart chambers is accompanied by differential gene expression within the heart [18,26]. Cell tracing experiments in mice and chick have established that heart myocardium derives from PHF, while SHF contributes to the OFT [27]. Interestingly, previous studies suggest that Tbx1 is expressed in SHF and that Tbx1 positively regulates SHF cell proliferation and contribution to the muscle layer of OFT [12,28,29].
This vertebrate pattern of heart development is largely conserved in zebrafish. Similar to mice and chick, zebrafish PHF cells migrate towards the midline to form a linear heart tube [30]. Moreover, presence of SHF and its contribution to OFT has been recently confirmed in zebrafish [31,32]. Zebrafish have a simple two-chambered heart with a single inflow and outflow tract and looping results in an S-shaped heart instead of the 4-chambered heart with multiple outlets in higher vertebrates [30]. The OFT retains the bulbous arteriosus (BA), an accessory chamber equivalent to the conus arteriosus of amphibians and reptiles. This structure has been replaced by the pulmonary and aortic trunk in mammals and birds [33,34]. The simpler structure of the heart combined with transparency of embryos and availability of a tbx1 mutant, make zebrafish the ideal system to study cardiac defects associated with DGS. Here we use the zebrafish tbx1 2/2 mutant [35] to study cardiac defects associated with DGS and identify tbx1 target genes. We show that tbx1 is required for normal heart looping and differentiation of myocardium derived from the PHF and SHF. In addition we provide evidence that tbx1 mediates its function at least in part via wnt11r, and one of its downstream mediators, alcama.

Fish stocks and maintenance
Fish were maintained at 28.5uC under standard conditions [36] and were staged as described [37]. The vgo tm208 (tbx1 2/2 ) line and wnt11r fh224/+ (wnt11r 2/2 ) mutant lines were a kind gift from Dr Tatjana  Identification and genotyping of mutants tbx1 2/2 mutants have an A-to-T missense mutation [35]. Mutant embryos were identified by high resolution melting analysis [39]. DNA was extracted from the tails of stained embryos and PCR was conducted using the primers 59-CACAACTGAAAATCGCCAGCAATC-39 (forward, 0.2 mM), 59-AATATGGTAAAACTCACCAGTCCT -39 (reverse, 1 mM) and 59-TTTACCAAAGGCTTCAGAGACTGTAATCCC -39 (unlabelled probe complementary to reverse strand, 1 mM). After PCR, the samples were heated to 94uC for 2 minutes and then cooled to 10uC before melting. LCGreen dye (Idaho Technology) was included in the PCR mix and melting analysis was done on LightScanner TM (Idaho Technology). The wnt11r 2/2 mutant was generated by tilling, and has a G-to-T nonsense mutation that generates a stop codon at amino acid 94. wnt11r 2/2 mutants were identified in a similar fashion using the primers 59-GGTCTGCCAAAAGACCTTCACAG-39 (forward, 0.2 mM), 59-TTGGAAATAAATGCGTTTAGACACGGTT-39 (reverse, 1 mM) and 59-TGTTCCCCTATTGATGGACC-GAAACTCCT-39 (unlabeled probe complementary to reverse strand, 1 mM). All the identified WT and mutants were included in the analysis.

Cloning and RNA transcription
The tbx1 gene cloned in pCS2+, and wnt11r gene cloned in pCMVSport6.1, were obtained from ZIRC (Zebrafish International Resource Center). To make sense RNA for injection, these plasmids were cut with NotI and XhoI respectively, and in vitro transcribed from Sp6 promoter using mMessage Machine kit (Ambion). RNA for alcama was made as previously described [40].

Morpholino antisense oligonucleotide and RNA injections
Morpholino and RNA were dissolved in molecular biology grade water and pressure injected into 1-4 cell zebrafish embryos. For RNA rescue experiments, 26 pg of tbx1, wnt11r or alcama RNA was injected per embryo. The previously described translation blocking morpholino for alcama was used at 1.1 ng per embryo [40].

Tissue labeling procedures
Whole mount RNA in situ hybridization (ISH) with digoxigenin was performed as previously described [41]. The plasmids for versican and eln2 were a gift from Dr Joseph Yost (Department of Neurobiology, University of Utah) and for amhc and vmhc from Dr Dean Li (Department of Human Genetics, University of Utah). Alcama protein was stained using Zn-5 antibody from ZIRC at 1:500 dilution. A goat anti-mouse secondary antibody conjugated with Alexa 555 (Invitrogen) was used for fluorescence labeling of cell boundaries. For non-fluorescence detection, secondary antibody conjugated to alkaline phosphatase (Bio-Rad) was used with NBT-BCIP (Roche). The Isl-1 antibody was a gift from Dr Richard Dorsky (Department of Neurobiology and Anatomy, University of Utah) and was used at 1:100 dilution. Labeling of bulbous arteriosus with DAF-2DA (VWR) was done as previously described [42].

Imaging and quantification
Head-on photographs of 48 hours post fertilization (hpf) larvae were taken on Nikon SMZ1000. Dissected hearts and ISH embryos were photographed on a Nikon Y-IDP microscope at 206 zoom using Spot software. Confocal images of dissected hearts were taken on Olympus FV1000 microscope at 406 zoom. Images of all larvae from the same experiment were taken using the same settings and exposure. Cell numbers were counted using Imaris software. Measurements for circularity, length and volume were done using ImageJ and angles were measured on Adobe Photoshop. Live embryos were mounted in 1% low melt agarose and movies of the beating heart were taken with the Nikon Y-IDP microscope.

Morphometric measurements
Head-on pictures of cmlc2 stained 48 hpf embryos were used for measuring looping angles. To quantitate the looping angle, we measured the angle between the longitudinal axes of ventricle and atrium as shown in schematic Figure S1D. The ventricular axis was drawn by connecting the midpoint at the dorsal end of the ventricle (m, where it meets the OFT) and the midpoint at the widest part of the ventricle in a parallel plane to the dorsal end (n). Similarly the atrial axis was determined by joining the midpoint at the ventral end of the atrium (p, where it meets the IFT) and the midpoint at the widest part of the atrium in a parallel plane to the ventral end (q). The length of the chambers was determined by measuring along the longitudinal axes (m-n-o for the ventricle, and p-q-r for the atrium) with the end point at the bisection of the arc formed by the boundary of cmlc2 staining near the AVC (o and r, respectively).
The BA is shaped like a circle sliced horizontally at dorsal and ventral ends. Width of the BA was measured at the widest part of the small circular chamber (diameter of the circle). The length was measured from the ventral end of the cut where it meets the ventricle, to the dorsal end where it continues as an artery.

Analysis of cardiac performance
Movies of beating hearts were imaged as above and analyzed as follows. Heart rate was calculated by counting the number of sequential contractions. The widths of ventricles/atria at diastole and systole were measured. The width was measured at the widest part along the anterior-posterior axis. Since the chambers are prolate spheroids, width measurement along the anterior-posterior axes in lateral view is equivalent to left-right measurement in a head-on view. Shortening fractions were calculated as width [(diastole -systole)/diastole]. Stroke volume was determined as previously described [43]. The perimeter of the ventricle during diastole and systole was outlined on ImageJ and analyzed with a ''fit-to-ellipse'' algorithm, giving the major and minor axes. Volume was calculated using V = 4/3 * p * a * b 2 , where ''a'' is the major axis and ''b'' the minor axis. The stroke volume was obtained by subtracting the volume at systole from volume at diastole.
Paraffin embedding and sectioning 72 hpf larvae were fixed in 4% paraformaldehyde for 2 days, dehydrated in ethanol series and transferred directly to xylene. They were allowed to equilibrate for 90 min and then placed in paraffin. Embryos were embedded in disposable plastic molds and cooled before sectioning at 5 mm on a Leica RM2155 microtome. Glass slides were heated to 60uC overnight, placed in xylenes for 5 min, rehydrated through an ethanol series to water, stained with 0.1% toluidine blue, and coverslipped using Cytoseal 60 (Fisher Scientific).

Ethics statement
This work has been approved by the University of Utah IACUC (#11-07006). All the fish are housed at the campus fish facility where, all laboratory personnel are trained for the maintenance and care of zebrafish. I am trained and certified for handling zebrafish.

tbx1 is required for normal cardiac looping in zebrafish
In zebrafish, tbx1 expression initiates between 12-14 hpf in the lateral plate mesoderm containing the cardiac progenitors [35], and is maintained in the myocardium as the heart develops (Fig. S1A-C). By 48 hpf, expression is stronger in the ventricle as compared to the atrium and particularly pronounced in the atrioventricular canal (AVC) (Fig. S1C). We investigated whether tbx1 is necessary for correct cardiac morphogenesis in zebrafish, similar to humans and mice. To that end, we visualized cardiomyocytes in the tbx1 2/2 null mutant [35] using cmlc2 in situ hybridization (ISH). The cardiac progenitors are correctly specified at 14 somites, migrate towards the midline to fuse at 21 somites and form the linear heart tube at 24 hpf in tbx1 2/2 mutants. However, jogging of ventricle (30-36 hpf) and looping of the heart (36-48 hpf), is defective in tbx1 2/2 mutants resulting in a straight heart (Fig. 1B) rather than the WT S-shaped heart (Fig. 1A). The straight heart phenotype was variable between larvae, so we measured the angle between the longitudinal axes of ventricle and atrium to quantify the defect (see Experimental Procedures and Fig. S1D). There is a statistically significant decrease in the angle between WT (22u) and tbx1 2/2 mutants (10u) (Fig. 1C). While the atrioventricular angle and left-right positioning of the heart is affected in tbx1 2/2 mutants, the antero-posterior positioning of the heart is unaffected. Our data demonstrate that tbx1 is necessary for heart looping. tbx1 regulates cardiomyocyte shape Cardiomyocyte shape changes are a major contributor to normal heart looping [44,45]. During the process of looping, the cells transition from small and rounded (isotropic) morphology in the linear heart tube (24-28 hpf) to flattened and elongated anisotropic cell shapes in the outer curvature at the expanded chamber stage (48-58 hpf) [46]. We investigated the possibility that tbx1 regulates the shape of cardiomyocytes, hence affecting heart looping. To that end, we stained 48 hpf Tg(cmlc2:EGFP) larvae with Alcama antibody to visualize the cell boundary, and analyze cell shape. Larvae were co-stained with DAPI to visualize the nuclei, and hearts were dissected and imaged using a confocal microscope ( Fig. 2A, B). In WT siblings, the cells in the outer curvature are elongated with their longitudinal axes pointing towards the AVC (white outline), while cells in the inner curvature continue to be small and rounded (yellow outline) ( Fig. 2A). However, tbx1 2/2 mutant cells in the outer and inner curvatures retain the small and rounded morphology at 48 hpf (Fig. 2B). We quantified this defect by calculating the circularity of 10 cells from the inner and outer chambers of 7 mutant and WT embryos. Cells from tbx1 2/2 mutants in the outer curvature have 1.2-1.3 fold higher circularity (are rounder) as compared to WT siblings ( Fig. 2C), while cells in the inner curvature are equally round between the mutant and WT (Fig. 2D). In summary, our data suggest that tbx1 is required for regulation of cardiomyocyte shape, and heart looping defect in tbx1 2/2 mutants.

tbx1 regulates cardiac morphology and cardiomyocyte number
In addition to the looping defect, tbx1 2/2 mutants have other cardiac morphological defects. Mutant ventricles are significantly shorter at 48 hpf (Fig. S1E), while atrial length is similar between mutants and WT. The ventricle at its widest point is significantly wider in the mutant when compared to WT at 48 hpf, (Fig. S1F) and 72 hpf (data not shown). Atrium width differences between mutants and WT are significant at 72 hpf (Fig. S1G), but not at 48 hpf (data not shown). In addition, sections of the heart at 72 hpf, reveal that mutant hearts have thinner ventricular walls and larger intercellular spaces (Fig. 1D).
The smaller ventricle size in tbx1 2/2 mutants may be due either to fewer or to smaller cardiomyocytes. To distinguish between these possibilities, we stained tbx1 2/2 ; Tg(cmcl2:EGFP) larvae with DAPI to count total cell numbers in hearts (Fig. S2A,B). Total cardiomyocyte number was unchanged at 33 hpf, but was significantly decreased 1.3 fold at 48 hpf (post-looping and morphological defects) (Fig. S2C). This correlates with the lower cell density and larger intercellular spaces observed in sections of tbx1 2/2 mutant hearts (Fig. 1D). Our data suggests that tbx1 2/2 mutants have proper specification and differentiation of cardiomyocytes during early stages marked by cmlc2 expression (Fig. 1A), but have decreased number of cardiomyocytes during later stages (after 33 hpf, Fig. S2C).
The late decrease in cardiomyocyte number observed in tbx1 2/2 mutants may be caused by decrease in proliferation of the PHF cells. We tested this possibility by analyzing proliferation of cardiomyocytes using phosphohistone 3 (PH3) staining to determine the percentage of cells undergoing mitosis (Fig. S2A, B). Analysis of PH3 staining reveals that the mean number of proliferating cells per heart is significantly reduced 1.6 times at 33 hpf (15 in WT siblings versus 9 in tbx1 2/2 mutants, Fig. S2D). However, the low percentage of cardiomyocytes undergoing mitosis (5-8%) combined with the physiologic decrease in cell proliferation from 24 to 48 hpf [47], suggests that decreased proliferation in tbx1 2/2 mutants might be insufficient to drive the dramatic difference in cardiomyocyte number at 48 hpf (235 in WT siblings versus 194 in tbx1 2/2 mutants).
An alternative or additional explanation of reduced cardiomyocyte number in tbx1 2/2 mutants is reduced contribution of SHFderived cells. Recently several groups have reported the existence of a SHF in zebrafish that contributes cardiomyocytes to the arterial pole of the heart tube between 24 and 48 hpf [31,32,48]. During the course of our analysis, Hami et al. reported that tbx1 2/2 mutants have reduced incorporation of dye-labeled cells from the pericardial wall to the arterial pole of the heart. Thus the reduced cardiomyocyte number in tbx1 2/2 mutants could result from a combination of decrease in PHF cell proliferation and decrease in incorporation of SHF cells. Taken together, these data suggest that tbx1 regulates proliferation of PHF cells and incorporation of SHF cells into the developing heart. Whether SHF cell incorporation plays a role in heart looping is an intriguing possibility and will be a part of future investigations. tbx1 expression is necessary for regional differentiation of heart At 48 hpf the heart becomes differentiated to form the various regions: inflow tract (IFT), atrium, atrioventricular canal (AVC), ventricle and OFT. In zebrafish, heart looping defects are highly correlated with defects in regional differentiation of the heart, suggesting that these regional changes in expression pattern are important for the looping process [46,49,50,51,52,53]. We performed ISH analysis in tbx1 2/2 mutants at 52 hpf to evaluate if regional differentiation occurs normally. The ventricular marker vmhc and atrial marker amhc are expressed normally in tbx1 2/2 mutants, suggesting that atrial and ventricular specification occurs normally in tbx1 2/2 mutants (Fig. S3).
We next analyzed expression of markers specific to different regions of the myocardium. At 52 hpf, bone morphogenetic protein 4 (bmp4) is expressed in the IFT, AVC and OFT in WT siblings (Fig. 3A), but expression is absent in tbx1 2/2 mutants (Fig. 3E). Similarly, WT siblings express versican and notch1b in the AVC myocardium and endocardium, respectively (Fig. 3B, C). However, tbx1 2/2 mutants fail to express these markers in the AVC at this stage (Fig. 3F, G). Next, we analyzed atrial natriuretic factor (anf), whose expression is restricted to the outer curvature of the ventricle and atrium in WT siblings (Fig. 3D). However, in tbx1 2/2 mutants, anf expression is expanded to the inner curvature in the ventricle and to a lesser degree in the atrium. In order to determine if these defects are heart specific, we examined versican expression in the otoliths. tbx1 2/2 mutant otoliths express versican (inset Fig. 3G), indicating that tbx1 is required for regional expression of these markers specifically in the heart.
The abovementioned markers are expressed broadly along the linear heart tube at 30 hpf, but become restricted to specific regions by 48 hpf. These markers are expressed normally in the tbx1 2/2 mutants at 30 hpf (data not shown), indicating that the differentiation defect arises later in development, concomitant with the heart looping and cardiomyocyte shape defects. The differentiation defect may be an effect or a cause of the looping defect. In addition, penetrance of these defects is not uniform in tbx1 2/2 mutant larvae ( Fig. 3M-P). Alterations in notch1b expression showed the highest penetrance and we will henceforth use it as a marker for regional differentiation. Failure of tbx1 2/2 mutants to restrict anf expression, taken together with absence of bmp4, notch1b, and anf, suggests that tbx1 is necessary for regional differentiation of the heart. tbx1 is necessary for normal OFT formation and differentiation DGS patients and Tbx12/2 mutant mice have defects in OFT such as persistent truncus arteriosus and interrupted aortic arch. Hence we studied the OFT in the zebrafish tbx1 2/2 mutants. The zebrafish ventricle ends in a constriction leading to the BA, an accessory chamber present in lower vertebrates that is composed of smooth muscle. The BA is shaped like a circle sliced horizontally at the dorsal and ventral ends, where it meets the artery and ventricle, respectively (Fig. 4A, D; for details of measurement parameters, please refer to the ''morphometric measurements'' in the Materials and Methods section). While BA width is unaffected (Fig. 4 G), its length is reduced in tbx1 2/2 mutants versus WT siblings (Fig. 4H).
In addition to morphology, we analyzed the tbx1 2/2 mutant for proper differentiation of the BA. To that end we used the fluorescent nitric oxide sensor DAF-2DA, a marker for smooth muscle [42]. Our analysis revealed that BA smooth muscle in tbx1 2/2 mutants is defective as evidenced by negative staining with DAF-2DA (Fig. 4B, E). tbx1 2/2 mutants also fail to express the BA-specific marker tropoelastin2 (eln2) at 72 hpf (Fig. 4F). Taken together, our findings suggest that in the absence of tbx1 signal, the BA is underdeveloped and undifferentiated. In keeping with our data, a recent report demonstrated that tbx1 2/2 mutants have reduced incorporation of SHF cells at the arterial pole [32], which may contribute to decreased size of BA.

tbx1 2/2 mutants have defects in cardiac performance
To determine how the heart looping and differentiation defects in tbx1 2/2 mutants affected cardiac function, we analyzed movies of beating hearts in live embryos at 31, 48 and 72 hpf. Cardiac contractility was assessed using the ventricular and atrial shortening fractions. Mean ventricular shortening fraction was unaffected at 48 hpf but was significantly lower at 72 hpf in tbx1 2/2 mutants (0.16060.012) compared to WT (0.19760.014) (Fig. S4A). In contrast, mean atrial shortening fraction was unaffected at 48 hpf and 72 hpf (Fig. S4B). As shortening fraction is a function of width, the unchanged atrial contractility was in keeping with the weaker morphological defect observed in tbx1 2/2 mutant atria (Fig. S1E, G). Considering the decreased ventricular contractility in tbx1 2/2 mutants, we next assessed the cardiac output as stroke volume of the ventricle. While the stroke volume was 10-20% lower in tbx1 2/2 mutants at both 48 and 72 hpf, this difference is not statistically significant (p-value .0.05) (Fig. S4C). The unchanged stroke volume suggests that in spite of decrease in ventricular length, the mutant heart probably compensates by increased ventricular width to maintain cardiac output.
In addition to contractility and stroke volume, another important factor to measure for cardiac performance is the heart rate. Heart rate was lower in tbx1 2/2 mutants at both 48 and 72 hpf. At 48 hpf the mean heart rate decreased to 75% in tbx1 2/2 mutants (10862 beats per minute) as compared to WT siblings (14362 beats per minute). At 72 hpf the mutant heart rate (16762) partially recovered to 86% of WT (19564) (Fig. S4D). This is in direct contrast to DGS patients where the heart beats faster to compensate for OFT obstructions. Since zebrafish do not have any obstruction in the BA, they probably do not need this compensatory mechanism. In conclusion, the ventricle in tbx1 2/2 zebrafish mutants has defects in contractility and heart rate, while stroke volume is unaffected.
wnt11r regulates heart looping and regional differentiation similar to tbx1 While several genes have been implicated in the tbx1 pathway, detailed characterization of the tbx1 pathway regulating DGS phenotype is lacking. We sought to utilize the zebrafish heart looping defect to identify new genes that may function downstream of tbx1 during this process. To that end we utilized a candidate gene approach, and analyzed all genes known to regulate heart looping and whose misregulation cause cardiac diseases in humans. Interestingly, T-Box proteins frequently interact and several among them, such as tbx2a, tbx3b [50], tbx5 [53], and tbx20 [54], have been implicated in zebrafish heart looping. Furthermore, mutations in TBX5 and TBX20 are associated with congenital heart disease in humans. To test whether these genes function downstream of tbx1 to regulate zebrafish heart development, we analyzed the expression of these tbx genes in the tbx1 2/2 mutant. By ISH analysis, all these genes are more highly expressed in the ventricle as compared to the atrium in WT embryos at 52 hpf. However, these genes are Another important gene that regulates heart development and looping is Wnt11. In spite of similarity in cardiac phenotype of Tbx12/2 and Wnt112/2 mouse mutants, no link between these genes has been established to date. While zebrafish wnt11r is homologous to Xenopus wnt11r, and Xenopus wnt11r has been implicated in heart looping [15], the role of zebrafish wnt11r in heart development and looping has not been studied to date. We sought to test whether wnt11 may function in the tbx1 pathway using zebrafish mutants. Hence we first studied heart development in the zebrafish wnt11r 2/2 mutant [55]. cmcl2 ISH revealed that the wnt11r 2/2 mutants also have straight hearts, similar to tbx12/2 mutants (Fig. 5A). Quantification of angle between the ventricular and atrial axes revealed that the looping angle is significantly reduced to 10u in wnt11r 2/2 mutants from 19u in WT siblings.
Next, we analyzed the regional differentiation in wnt11r2/2 hearts. Similar to tbx12/2 mutants (Fig. 3), wnt11r 2/2 mutants have unchanged ventricular and atrial differentiation (data not shown). However, similar to tbx12/2 mutants (Fig. 3), wnt11r 2/2 mutants have down-regulated bmp4, versican and notch1b (Fig. 5C). Additionally, expression of anf is expanded to the inner curvature especially in the ventricle (Fig. 5C). Staining with DAF-2DA and eln2 ISH reveal that wnt11r 2/2 mutants do not have BA defects (Fig. S6), suggesting that the heart looping and regional differentiation defects are independent of BA defects. Together, the heart looping and regional differentiation defects observed in wnt11r 2/2 mutants indicate that wnt11r may function downstream of tbx1 during these processes. tbx1 and wnt11r function in the same pathway to regulate heart looping and regional differentiation The expression of wnt11r in zebrafish heart development in zebrafish is unknown. Our ISH analysis revealed that wnt11r is first expressed in cardiac mesoderm just prior to fusion (21 somites, Fig. 6A) and after tbx1 expression in the heart begins. wnt11r expression is maintained in cardiomyocytes in both the atrium and the ventricle as the heart forms and loops (Fig. 6B). Over time, wnt11r expression weakens throughout the heart, although it is still present as late as 48 hpf (Fig. 6C). The expression of wnt11r in the heart correlates with the straight heart defect in wnt11r 2/2 mutants. Furthermore, the initiation of wnt11r expression after the onset of tbx1 expression in cardiac progenitors supports the possibility that that tbx1 initiates heart expression of wnt11r.
To assess whether tbx1 regulates wnt11r during heart development we analyzed wnt11r expression in tbx1 knockdowns. If tbx1 activates wnt11r expression, we would expect a decrease in wnt11r signal in the absence of tbx1. Indeed, we observed that wnt11r expression is down-regulated in tbx1 2/2 mutants (Fig. S5E, J and 6D, E). This result was corroborated by quantitative RT-PCR, which indicates that wnt11r RNA is down-regulated by 37% in hearts of tbx1 2/2 mutants (not shown). In contrast, tbx1 expression is unchanged in wnt11r 2/2 mutants as compared to WT siblings (Fig. 7A, C). The defective heart looping and regional differentiation defects of wnt11r 2/2 mutants, the timing and location of wnt11r expression with respect to tbx1 expression, and the reduction of wnt11r in tbx1 2/2 mutant larvae all support our hypothesis that tbx1 regulates heart looping and regional differentiation through wnt11r.
To further explore this hypothesis we conducted non-allelic non-complementation assays. tbx1 +/2 and wnt11r +/2 heterozygotes have unaffected cardiac looping and regional differentiation (Fig. 6F, G). We crossed tbx1 +/2 and wnt11r +/2 mutants and analyzed the double heterozygous larvae. Our analysis reveals that tbx1 +/2 ;wnt11r +/2 larvae have incomplete looping and fail to express notch1b in significantly more larvae as compared to tbx1 +/2 or wnt11r +/2 larvae (Fig. 6F, G). The penetrance of defects in tbx1 +/2 ;wnt11r +/2 is slightly lower than tbx1 2/2 or wnt11r 2/2 mutants, but the resulting phenotype is the same. Thus the double heterozygote carrying recessive mutations in tbx1 and wnt11r exhibits a phenotype similar to homozygous mutant of either gene, while single heterozygotes do not. These double-heterozygote analyses indicate a case of non-allelic non-complementation and suggest that tbx1 and wnt11r function in the same pathway to regulate heart looping and regional differentiation. tbx1 regulates heart looping and regional differentiation via wnt11r and alcama in a linear pathway Previous work identified Xenopus alcama as a downstream target of wnt11r during differentiation of cardiomyocytes and heart looping [16]. This led us to propose tbx1 regulates heart looping and differentiation by activating wnt11r, which in turn activates alcama. We tested our hypothesis by analyzing Alcama expression during heart development. alcama expression starts in the heart progenitors at 21 somites and continues on until 4 dpf [56,57]. Similar to tbx1 expression, Alcama protein expression is stronger in the ventricle than atrium and is prominent in the AVC in WT embryos (Fig. 7B, E). However in both tbx1 2/2 and wnt11r 2/2 mutants, Alcama expression is strongly down-regulated in ventricles and weakly down-regulated in atria at 48 hpf (Fig. 7D,  F), supporting our hypothesis that alcama is downstream of tbx1 and wnt11r.
To test the specific function of alcama during zebrafish heart morphogenesis, we injected a morpholino directed against alcama [40] into zebrafish at the 1-cell stage and assessed heart looping and regional differentiation at 48 hpf. alcama morphants have defective heart looping of varying severity (Fig. 7H), a finding that may be explained by dosage differences in morpholino injections. In addition to looping defects, alcama morphants have defects in regional differentiation of the heart as assessed by notch1b ISH (Fig. 7J). Furthermore, in accordance with our hypothesis that alcama functions downstream of tbx1 and wnt11r, their expression is unaffected in alcama morphants (data not shown). Hence, we have shown that Alcama is down-regulated in tbx1 2/2 and wnt11r 2/2 mutants, and that alcama is needed for normal heart looping and differentiation. We further tested whether wnt11r and alcama function downstream of tbx1 by rescue experiments. If cardiac defects in tbx1 2/2 mutants are caused specifically by suppression of the wnt11r-alcama pathway, ectopic expression of wnt11r or alcama should rescue cardiac defects in tbx1 2/2 mutants. We injected wnt11r or alcama RNA into 1-cell stage tbx1 2/2 mutants at concentrations that do not produce a phenotype in WT embryos (see methods). Injected mutants were assessed for heart looping (cmlc2 ISH, Fig. 8B) at 52 hpf. tbx1 RNA was injected as a positive control for rescue. Injection of either wnt11r RNA or alcama RNA increased the angle between the ventricular and atrial axes to nearly WT levels (18u in RNA injected tbx1 2/2 mutants versus 9u in uninjected tbx1 2/2 mutants and 23u in WT) (Fig. 8C). This data demonstrates that wnt11r and alcam do indeed function downstream of tbx1 to regulate heart looping in zebrafish. Similarly, mutants were assessed for regional differentiation defects (notch1b ISH, Fig. 8E) following injection of wnt11r or alcama RNA. Injection of either RNA decreased the percentage of larvae with absent notch1b staining (Fig. 8F). Partial rescue of tbx1 2/2 mutants by RNA injection substantiates our model whereby tbx1, wnt11r and alcama function in a linear pathway regulating zebrafish cardiac development. Partial rescue can be explained by differences in dosage and timing of expression, degradation of RNA by late stages, or other downstream targets of tbx1 that remain unexplored to date. These data taken together with our expression analysis support our hypothesis whereby tbx1 activates wnt11r, which then activates alcama to regulate heart looping and differentiation (Fig. 8G).

Discussion
tbx1 is required for heart formation, cardiomyocyte shape and differentiation Tbx1 has been identified as the gene critical for regulating DGS etiology However, the molecular and cellular defects caused by loss of tbx1 have not been well characterized. Using the zebrafish tbx1 2/2 mutant, we have demonstrated that similar to mice, zebrafish tbx1 is required for normal heart development. Zebrafish tbx1 2/2 mutants have improperly looped hearts, undifferentiated BA and other morphological defects. These defects may be a milder presentation of the conotruncal defects and ventricular septal defects in DGS patients and mouse Tbx12/2 mutants. Indeed, genes regulating zebrafish heart looping such as Bmp4, Tbx2, Tbx3, Tbx5, Tbx20, and Wnt11r, are associated with conotruncal defects in humans or mice, but not looping defects [58,59,60,61,62].
Furthermore, data indicates that OFT defects result from a failure of OFT myocardial wall to rotate [63], suggesting that myocardial wall rotation might be the final stage of the looping process that is affected in mouse mutants and DGS patients. This would explain the intersection of genes involved in the two processes. Hence, improper looping and BA morphology in zebrafish may be a milder presentation of the conotruncal defects in DGS patients. Our data reveals a new role for tbx1 in regulating heart looping thereby contributing to the conotruncal and septal development of the heart. This may provide a new mechanism by which loss of Tbx1 contributes to conotruncal cardiac defects in DGS. The looping defect, taken together with the weaker ventricular contractions and slower heart rate in tbx1 2/2 mutants, suggests that tbx1 is required for proper heart formation. Despite the defects in heart formation, the early stages of cardiogenesis proceed normally in tbx1 mutants. Atrial and ventricular fates are assigned normally in the tbx1 2/2 mutant as indicated by normal cmcl2, vmhc and amhc ISH staining (Fig. 1, S3), and normal antibody staining for S46 and MF20 (data not shown). Given the expression of tbx1 in cardiac mesodermal progenitors from the fusion stage (Fig.  S1A), this finding was surprising. However, it is consistent with mouse data [12] and suggests that tbx1 is not required for heart specification and early development.
While cells are specified properly our data show that cell shape is disrupted in tbx1 mutants. In tbx1 2/2 mutants cells fail to change from isotropic to anisotropic shape in the outer curvature. Other zebrafish mutants/morphants with looping defects share the observed defects in cell shape [46,50,52]. This cell shape defect combined with reduction in cell number may contribute to looping defects by changing cell-cell interaction. Indeed, cross-sections of tbx1 2/2 mutant hearts reveal an increase in extracellular spaces compared to WT siblings. This observation is important, as increased extracellular spaces have been observed in wnt11r and alcama morphants in Xenopus [15,16]. Decreased cell adhesion due to down-regulation of the adhesion molecule alcama, may be causing the increase in extracellular spaces in tbx1 2/2 mutants. Alternatively, the increased extracellular spaces may lead to altered cell polarity or may preclude changes in cell shape or differentiation. These data indicate that tbx1 is required for induction of morphological, proliferative and shape changes in cardiomyocytes.
An additional factor contributing to the heart defects observed in tbx1 2/2 is a loss of tissue differentiation. Cardiomyocytes fail to differentiate into the OFT and AVC as indicated by down regulation of bmp4, versican and notch1b. In addition mis-expression of anf also corroborates this conclusion. Absent expression of bmp4, versican and notch1b in tbx1 2/2 mutants is in striking contrast to other looping mutants, where these markers fail to be restricted to their respective domains [46,49,50,52]. Hence we propose that tbx1 regulation of heart looping is distinct from genes such as tbx2a, tbx3b and nkx2.5, which also regulate this process. An unresolved question is whether differentiation of the heart regions contributes to looping or vice versa. tbx1 regulates differentiation of the SHF and BA DGS patients and Tbx12/2 mice have OFT defects such as persistent truncus arteriosus [2]. Moreover, Tbx1 is expressed in the mouse SHF and regulates SHF contribution to the OFT [12]. The zebrafish SHF has been recently described [31,32] but tbx1 expression in SHF cells has not been described so far. However, similarly to data in mouse, zebrafish tbx1 2/2 mutants have a smaller BA, indicating decreased contribution of SHF to the arterial pole of the heart. In addition, absent eln2 expression and DAF-2DA staining indicates that smooth muscle in the BA is undifferentiated. In addition to the BA, the ventricle in tbx1 2/2 mutants is also small and dysmorphic, with a severe reduction in total cardiomyocyte number. While our investigation of the BA defect in tbx1 2/2 mutants was ongoing, absence of tbx1 was reported to reduce differentiation and incorporation of SHFderived cells to the arterial pole of the heart [32]. These results are complementary to our data, indicating that tbx1 is required for SHF cell incorporation into the BA and subsequent differentiation of smooth muscle cells.
Reduced incorporation of cardiomyocytes from the SHF in zebrafish tbx1 2/2 mutant hearts suggests that the processes involved in heart development are conserved through evolution. Furthermore, the BA defects in zebrafish are consistent with conotruncal and septal defects observed in DGS patients and Tbx12/2 mice. However, as opposed to humans and mice, zebrafish do not develop ventricular hypertrophy as a compensatory mechanism to outflow tract defects. The defects in looping, differentiation, and heart rate emerge at 48 hpf, before the first detectable time point for outflow tract defect (72 hpf), suggesting that the ventricular defect is not secondary to the outflow tract defect.
While Tbx1's role in regulating proliferation and contribution of SHF cells has been studied [12], our data reveals a role for tbx1 in regulating heart looping. We postulate that the conotruncal and septal defects in DGS may be a more severe manifestation of the defects in SHF cell incorporation and the defects in morphological movements of looping during development.
wnt11r functions downstream of tbx1 Although Wnt11 signaling has been shown to be important for specification of cardiac fate and Wnt112/2 mutants have OFT defects [19,20,25], Tbx1 and Wnt11 signaling pathways have not been linked to date. Unlike mouse and chicken embryos, Xenopus and zebrafish embryos do not express wnt11 in the developing heart [15]. wnt11r, a second wnt11 gene in Xenopus and zebrafish, with high homology to human and chicken Wnt11, mediates noncanonical Wnt signaling and is necessary for normal heart morphogenesis. In Xenopus, wnt11r starts to be expressed just prior to fusion of cardiac progenitors and subsequently continues to be expressed in the heart tissue. It was later discovered that alcama regulates cardiac looping and functions downstream of wnt11r in Xenopus [16].
The role of wnt11r in zebrafish heart morphogenesis has not been previously characterized. We show that similarly to Xenopus, zebrafish wnt11r is first expressed at 21somites just prior to cardiac progenitor fusion and continues to be expressed as the heart develops. Our data shows that wnt11r 2/2 mutants have similar looping and regional differentiation defects as observed in tbx1 2/2 mutants, and that wnt11r is down-regulated in tbx1 2/2 mutant hearts. Furthermore, our non-allelic non-complementation assay indicates that tbx1 and wnt11r function in the same pathway. Knockdown of alcama, a gene demonstrated to be downstream of wnt11r, also presents with looping and regional differentiation defects. Importantly, we were able to rescue the looping and differentiation defects in tbx1 2/2 mutants by injection of wnt11r and/or alcama RNA. All these data lead us to our working model that tbx1 regulates wnt11r, wnt11r regulates alcama, which in turn regulates heart looping and differentiation (Fig. 8G). Elucidation of new players in the pathway may help in developing future therapies for patients with DGS and other cardiac defects.

Conclusions
In summary, this study demonstrates that zebrafish tbx1 2/2 mutants have defects in heart looping and function. This is the first demonstration that tbx1 regulates differentiation and shape of cardiac cells derived from PHF. In addition, we describe tbx1's role in regulating total cardiomyocyte number via PHF cell proliferation that may be compounded by lack of contribution of SHF cells in tbx1 2/2 mutants. Importantly, we have identified wnt11r and alcama as novel mediators of the tbx1 pathway. We show for the first time that in zebrafish wnt11r 2/2 mutants and alcama morphants have heart looping and differentiation defects similar to tbx1 2/2 mutants, and our expression and non-complementation assay confirms that wnt11r and tbx1 function in the same pathway. Moreover, these defects can be rescued by over-expression of wnt11r and/or alcama in tbx1 2/2 mutants, suggesting that they function downstream of tbx1. Our data support a model whereby tbx1 regulates heart looping and differentiation via wnt11r and alcama. These findings are an important contribution to our understanding of tbx1 signaling and heart development. showing rescue of looping (B) and regional differentiation defect (E) at 48 hpf after injection of 26 pg wnt11r RNA at the 1-cell stage. Plots in C and F quantify the rescue of looping and regional differentiation defect after the indicated injections. WT siblings are represented in green and uninjected tbx1 2/2 mutants in pink, with injected mutants in dark pink or red. Scale bars: 50 mm. N indicates the total number of embryos from 3 experiments. (G) Outline of the working model, where tbx1 regulates wnt11r; wnt11r regulates alcama, which in turn regulates heart looping and regional differentiation. tbx1 regulates migration/differentiation of SHF cells by an independent mechanism. doi:10.1371/journal.pone.0058145.g008

Supporting Information
Figure S1 tbx1 expression and heart shape defect of tbx1 2/2 1arvae. ISH for tbx1 in WT embryos showing expression in the fusing heart fields (21 somites, A), linear heart tube (24 hpf, B) and the looped heart (48 hpf, C). A and B are dorsal views and C is a head-on view. Arrowheads in A and B point towards pharyngeal pouches and the cardiac cells are outlined in A-C. (D) Schematic of head-on view of 48 hpf heart. Arrow indicates the AVC; a, atrium; v, ventricle. Dotted red lines indicate the widths at the ends and at the widest part of the ventricle and atrium. The longitudinal axes (shown in green) were drawn by joining the midpoints m and n in ventricle and midpoints p and q in atrium. mo and pr were measured for length of ventricle and atrium, respectively (see Materials for details).