Wnt and Nodal asymmetries stratify mouse laterality phenotypes in the absence of node flow Authors: Amaia Ochandorena-Saa, Emeline Perthame, Zoé Oulerich, Alexander Chamolly, Thierry Blisnick, Johanna Lokmer, Cécile Rouillon, Philippe Bastin, Sigolène M. Meilhac Published: 2026 ABSTRACT Left-right symmetry breaking in mice is considered to occur via node leftward fluid flow. The molecular cascade within the node has been reconstructed at high resolution; however, its importance for organogenesis remains poorly understood. Here, we show in mutants for the motile cilium component CCDC40 that 70% of mice develop normal situs at birth despite abrogation of node flow. The discrete morphospace output, including situs inversus totalis and heterotaxy with left isomerism, supports an innovative model of symmetry breaking, in which node flow only biases asymmetry orientation, while a distinct mechanism, potentially self-amplifying, generates it. Longitudinal, quantitative, and paired transcriptomic analyses uncover the molecular signature of laterality clusters, highlighting WNT in addition to NODAL pathways. We identify asymmetry of cardiopulmonary progenitors, the disruption of which is associated with combined heart and lung defects in mutants. Functional pharmacological perturbation reinforces the importance of WNT, besides the node, in the establishment of asymmetry. INTRODUCTION Left-right patterning of visceral organs is not just an anatomical curiosity but a functional requirement. Defective coordination of left-right patterning leads to the severe syndrome of heterotaxy (or situs ambiguus), characterized by abnormal organ shapes or discordant asymmetries between organs or organ compartments. Heterotaxy affects 1 of 10,000 live births and is frequently associated with complex congenital heart defects, which determine the prognosis of patients (1, 2). Heterotaxy can cosegregate in families with completely normal (situs solitus) or completely mirror-image (situs inversus totalis) organ laterality, which is usually asymptomatic (1). Heterotaxy can be categorized as right or left isomerism, if organ compartments that ought to be asymmetric are instead similar (isomeric) on both sides, for example, lung lobes, spleen, and atria. However, heterotaxy encompasses a broad spectrum of defects, with interindividual variations. Despite phenotypic variations, a monogenetic origin has been identified in about 20% of heterotaxy cases (3). Patient genome sequencing efforts have now uncovered 42 distinct Online Mendelian Inheritance in Man (OMIM) genes associated with heterotaxy. Many were identified on the basis of knowledge of asymmetry establishing mechanisms acquired in the mouse embryo. The secreted transforming growth factor–β (TGFβ) NODAL was first discovered as a major left determinant (4–6), and its inactivation impairs the morphogenesis of visceral organs, including the heart, lungs, spleen, stomach, liver, and intestine, thus modeling heterotaxy with right isomerism (7–9). The laterality of Nodal expression was found abnormal in a spontaneous mouse mutant (Dnah11iv) and a mutant arising from a transgenic insertion (Invsinv), showing that the underlying ciliary genes are upstream determinants of laterality (6). A breakthrough came with the discovery of motile cilia in a pit of cells referred to as the node, establishing a leftward fluid flow as the earliest event of symmetry breaking in mammals (10, 11). The node is now considered as a left-right organizer (12). Forcing node flow in a rightward direction by microfluidics can reverse the expression laterality of the NODAL target Pitx2 (13). The flow may carry chemical cues (14, 15) and can be sensed mechanically by primary cilia of node crown cells (16, 17). Sensing of flow on the left triggers the left-sided degradation of Dand5 mRNA, which encodes a NODAL antagonist (18, 19). These observations have thus reconstructed a linear cascade establishing left-right patterning, from motile cilia in the node pit to node flow to elevated NODAL signaling in the left node crown, and lastly activation of NODAL signaling in organ precursor cells in the left lateral plate mesoderm (LPM) (20–22). Cilia are organelles built on a scaffold of microtubule doublets, called axoneme. Motility is provided by DYNEIN motor proteins arranged as inner and outer arms. These arms are assembled along the axoneme with a 96-nm periodicity, set by a molecular ruler complex, composed of CCDC39 and CCDC40 (23, 24). Thus, Ccdc40 is required for ciliary beat. In humans, it is the third most frequent gene associated with the respiratory disease primary ciliary dyskinesia, and it is also associated with infertility and situs inversus totalis (25–27). In the mouse, the lnks allele of Ccdc40 has an N-Ethyl-N-nitrosourea (ENU)-induced nonsense point mutation, which introduces an early stop codon, and models motile ciliopathy, primary ciliary dyskinesia, and laterality defects (25). Within the framework of a linear cascade of left-right patterning, paralyzed cilia are predicted to abrogate flow and maintain bilateral Dand5 expression. However, in mutants for motile cilium genes Dnah11, Dnah5, and Ccdc40, all possible patterns of laterality have been observed, including absent, bilateral, or unilateral (left or right) NODAL signaling, overall reported as randomization of left-right asymmetry (6, 28, 29). The origin of this variability is poorly understood. Does it reflect truly random mechanisms of symmetry breaking, which are normally overridden by the node flow? Could variability arise from experimental pitfalls, which do not provide a complete picture of the mutant phenotype? It has been proposed that two cilia are sufficient to generate enough flow and trigger the asymmetry cascade (30), suggesting that variability in flow intensity could underlie variable laterality outputs in mutants. Alternatively, asymmetric gene expression is very transient (9, 31), so that absent expression could reflect an inappropriate stage of observation. In keeping with this, NODAL signaling has been reported to be delayed in motile cilium mutants (28, 29, 32). Until now, it has not been possible to assess the phenotypic output of a given pattern of Nodal expression in the early embryo, because it requires observations at different stages and partial penetrance precludes correlation. When laterality was diagnosed at fetal stages based on visceral organs, 18 to 20% of mutants for Dnah5 or Dnai1 displayed normal situs solitus, whereas the rest have situs inversus totalis or heterotaxy. However, the developmental trajectory of these fetuses is unknown. Thus, the role of the left-right organizer has remained unclear. Asymmetric molecular cues that have been identified, other than the NODAL pathway, include Wnt3 in the left node crown (33, 34), and bone morphogenetic protein (BMP) signaling in the right LPM (35). However, within the framework of a linear cascade of left-right patterning, only NODAL signaling is routinely assessed in laterality mutants, leaving the question open of whether these alternative asymmetric factors are mainly required in feedback loops to reinforce NODAL signaling or for independent functions. The heart provides an interesting readout of laterality mechanisms. It is asymmetric in its function, driving a double blood circulation. Heart morphogenesis is sensitive to left-right patterning (36). This is first detected by the rightward looping of the embryonic heart tube. However, heart looping is more than just a direction that is usually scored in laterality mutants; it corresponds to the transformation of an initial straight tube into a helical shape (37). During heart looping, the right ventricle, which was lying cranially to the left ventricle, is repositioned on its right side. We have previously shown that heart looping is centered on a buckling mechanism, when the heart tube grows between fixed poles (38). Tube elongation is fed by different populations of precursor cells located dorsally, ventrally or posteriorly (39). Nodal is transiently expressed in left cardiac precursors and controls independent asymmetries at the arterial and venous poles. Nodal is not required to initiate asymmetric morphogenesis, but to bias and amplify buckling (9, 40). Our analysis of heart morphogenesis thus shows that left-right patterning is not a simple linear cascade but rather entails several processes contributing to asymmetry. It is a current challenge to unravel the contribution of individual asymmetric cues to organogenesis. We have developed a framework for a comprehensive analysis of laterality defects, including three-dimensional (3D) imaging and quantification, at anatomical and molecular levels (fig. S1A) (9, 40, 41). We have also established a paired transcriptomic approach, able to uncover asymmetric gene expression in the LPM and avoiding noise from interindividual and stage variations (40). This has opened the possibility to investigate with higher precision how cells perceive asymmetry during organogenesis. Here, we analyze the role of the node, using Ccdc40lnks mutants as a model of motile ciliopathy. We perform in-depth analysis of laterality defects, using quantitative 3D imaging, showing that the left-right organizer is dispensable in 70% of cases to establish organ laterality. We tackle the challenge of correlating molecular asymmetries with measure of the node flow on one hand and with morphological phenotype on the other hand. Unsupervised clustering of samples identifies key geometrical parameters and domains of NODAL signaling as signatures of heterotaxy compared to situs solitus or situs inversus totalis. Last, we provide broader profiling of molecular asymmetries based on paired transcriptomics. This reveals the WNT pathway as an important determinant to stratify laterality phenotypes. In embryo culture, we show that WNT is functionally important before Nodal expression onset to modulate its patterning. Overall, we demonstrate that the node flow is required as a bias of symmetry breaking, controlling the robustness of left-right asymmetry, but that laterality can be established without it. Our work provides original insight into the origin of the spectrum of laterality defects. RESULTS Ccdc40 mutants are a model of motile ciliopathy with abrogated node flow To understand how the motile cilium component CCDC40 affects left-right patterning, we first investigated its expression pattern. We detected high Ccdc40 expression at E8.5b/c in the pit of the node, where motile cilia are localized (Fig. 1, A to D), but not in the notochord (fig. S1D). Colabeling shows that Ccdc40 is not expressed in heart progenitors in the LPM (fig. S1, A to E). Published single-cell transcriptomics further support specific expression of Ccdc40 in the node (fig. S1F). We also wondered which cell types express a toolkit of motile cilia and interrogated single-cell datasets. We found that the node/notochord was the only tissue between embryonic day 6.5 (E6.5) and E8.5 to express a toolkit of 82 motile cilium genes (fig. S2). We then analyzed how the node is affected in Ccdc40lnks/lnks mutants (Fig. 1E) (25). By scanning electron microscopy, we found that mutant nodes had many cilia with an abnormal ballooned morphology (Fig. 1, F and G), mainly localized in the node pit (fig. S1, G to I). We evaluated their function by live imaging of fluorescent bead movements. Compared to controls, mutant nodes were unable to generate a detectable flow (Fig. 1, H to L, and movie S1). We conclude that Ccdc40 mutants are a model of motile ciliopathy with fully penetrant abrogated node flow. Ccdc40 mutation disrupts the shape and function of motile cilia in the node pit. (A and B) Images by scanning electron microscopy of an E8.5c mouse embryo seen on a ventral view. The node is outlined in red and shown at higher magnification in (B). (C) Transverse optical section of the node, counterstained with Hoechst, showing Ccdc40 expression (white) in the pit (white arrowhead) by whole-mount in situ hybridization in a wild-type embryo. (D) Schema of the node. (E) Predicted CCDC40 protein domains produced in wild-type (+) and mutant (lnks) alleles. (F and G) High magnification of cilia in the node pit of control [(F); n = 3] or Ccdc40lnks/lnks mutant [(G); n = 3] embryos, imaged by scanning electron microscopy. (H and I) Map of node flow velocity intensity by particle image velocimetry of fluorescent bead movements in control (H) or Ccdc40lnks/lnks (I) live embryos at E8.5b. (J and K) Map of the direction of flow velocity vectors in the same embryos. (L) Corresponding quantification of leftward flow velocity at E8.5b-d. Means and SDs are shown. ***P < 0.001; n.s., not significant (Mann-Whitney test, n = 16 Ccdc40+/+ or Ccdc40lnks/+ control live embryos, n = 10 Ccdc40lnks/lnks mutant live embryos, and n = 4 control fixed embryos). Cran, cranial; Cau, caudal; D, dorsal; R, right; L, left; V, ventral. See also figs. S1 and S2, table S7, and movie S1. The node is required for the robustness of visceral organ laterality We analyzed laterality defects upon loss of Ccdc40 by 3D imaging of visceral organs in situ at birth. On the basis of the clinical nomenclature, we found three categories of laterality phenotypes. In contrast to situs solitus in controls (Fig. 2A and movie S2), 15% of Ccdc40 mutants had situs inversus totalis, in which visceral organs are in mirror image, as well as the heart structure (Fig. 2C and movie S2). We did not detect congenital heart defects associated with situs inversus totalis, but some mild anomalies in lung or liver lobation or in the heart apex position (table S1). Another 15% of mutants had heterotaxy with left isomerism of the lungs and bronchi (Fig. 2, D and E, and movie S2). The heart always had complex defects, in the venous return, atrial appendages (100% left isomerism), ventricle laterality (50% D-loop and 50% L-loop), septation, ventriculo-arterial connections, position of the great arteries, or aortic arch (table S1). Two cases also had laterality defects in the stomach, spleen, liver, and colon. Unexpectedly, the majority of mutants (70%) had no laterality defects, with either complete situs solitus (Fig. 2B and movie S2) or mild asymptomatic anomalies in lung or liver lobation, in the trajectory of the inferior caval vein or in the colon flexure (table S1). Thus, despite the full penetrance of node disruption, Ccdc40 mutants display partial penetrance of laterality defects at birth. The distribution of the three groups of laterality phenotypes significantly differs from a uniform randomization hypothesis (P = 0.0002, chi-square test). Mutants were obtained at the expected Mendelian ratios at E17.5 or postnatal day 7 (P7); thus, there was no sign of in utero lethality (Fig. 2, F and G). Mortality of mutants occurred by the second week of life (Fig. 2G). At this stage, we detected hydrocephalus in all mutants, which displayed situs solitus (Fig. 2, H and I). Hydrocephalus is a typical phenotype of motile ciliopathy, caused by the lack of cilia-driven motion of the cerebrospinal fluid (42), further highlighting the lack of cilium motility in situs solitus mutants. These results overall show that, while Ccdc40 mutation causes fully penetrant ciliopathy, laterality defects are only partially penetrant, indicating that node flow is dispensable to initiate asymmetric organogenesis and only required for the robustness of laterality patterning. Ccdc40 mutants display partial penetrance of laterality defects at birth. Examples of controls (A) and Ccdc40 mutants (B to E) at E17.5 or P0. [(A1) to (E1), (A2) to (E2)] Frontal sections from 3D micro–computed tomography (micro-CT) scans showing the position of visceral organs in the thoracic and abdominal cavities. Red arrowheads point to the heart apex. Frontal [(A3) to (E3)] and transverse [(A4) to (E4)] sections of explanted hearts imaged in 3D by high-resolution episcopic microscopy (HREM). The asterisk indicates complete atrioventricular septal defect. White arrowheads point to the presence (filled arrowhead) or absence (empty arrowhead) of atrial septation. Ccdc40 mutants have partial penetrance of laterality defects: situs inversus totalis (C) or heterotaxy with left isomerism of the airways [(D) and (E)]. (F and G) At E17.5 and P7, Ccdc40 mutants were found at the expected Mendelian ratio (P = 0.703, chi-square test; n = 115 at E17.5), whereas lethality was observed at P14 (black). (H and I) Frontal sections from 3D micro-CT scans showing the brain (H) and visceral organs (I) at P14. Examples of a control (H1) and a Ccdc40 mutant (H2); the asterisk indicates hydrocephalus in mutants. Visceral organ configuration shows situs solitus of heart (I1) and lungs (I2) in P14 mutants. Scale bars, 2000 μm [(A1) to (E1), and (A2) to (E2)], 500 μm [(A3) to (E3), and (A4) to (E4)], and 1000 μm [(H) and (I)]. Sample size (n) refers to number of fetuses/neonates in each category. R1 to R4, right lung lobes; L, left lung lobe; St, stomach; LAA, left atrial appendages; RAA, right atrial appendages; LV, left ventricle; RV, right ventricle; Ao, aorta; PA, pulmonary artery. See also tables S1 and S7 and movie S2. Embryonic heart looping can be defective in Ccdc40 mutants The heart is the first organ to undergo asymmetric morphogenesis. We investigated the phenotype of Ccdc40 mutants at E9.5, when looping of the embryonic heart tube is complete. In controls, the embryonic heart forms a rightward helix. In most mutants, heart looping was also rightward. However, it was inverted, i.e., leftward, in 27% of cases (Fig. 3, A to C). Because heart looping can be affected in its shape and not just in its direction (9, 38), we segmented and quantified it (Fig. 3, D to H, and movie S3). The length of the heart tube in mutants was not significantly changed, indicating that heart growth is normal (Fig. 3F). However, we found occasional defects in the orientation of the right ventricle-left ventricle axis, which failed to align perpendicular to the notochord (4 of 50 mutants; Fig. 3, D and G). Mutant hearts also showed anomalies in the position of the atrioventricular canal (AVC), which was significantly more medial (7 of 50; Fig. 3, E and H). Overall, our quantifications at E9.5 indicate three categories of heart looping in Ccdc40 mutants: normal, complete mirror image, and defective. Defective looping was observed in both rightward and leftward heart tubes. This parallels the laterality phenotypes at birth: situs solitus, situs inversus totalis, and heterotaxy with D-loop or L-loop. Quantification of heart looping defects in Ccdc40 mutants at E9.5. (A and B) Bright-field images [(A1) and (B1)] of Ccdc40 mutant embryos at E9.5, showing rightward or leftward (inverted) heart looping direction. [(A2) and (B2)] 3D segmentation of the embryonic heart tube with the looping direction outlined by a dashed arrow. (C) Bar diagram showing the frequency of heart looping direction (n = 106 controls and 101 Ccdc40 mutants). (D and E) 3D segmentation of the embryonic heart tube in Ccdc40lnks/lnks mutants, aligned in a ventral (D) and cranial (E) view, using the notochord as a reference axis (cyan). Examples are shown of mutants with a normal [(D1) and (E1)], mirror-image [(D2) and (E2)], or defective [(D3) and (E3)] geometry of the looped heart tube. The orientation of the RV (red)/LV (blue) axis (dashed white line) relative to the notochord (dashed blue line) is measured as an angle in (D). The lateral displacement of the atrioventricular canal (AVC; yellow) relative to the midline (dashed blue line) is measured as a distance (black double arrows) in (E). Cardiac segments are color coded as indicated in the legend. (F to H) Corresponding quantifications. The tube length, which is not significantly different between controls and mutants, indicates normal growth of the heart and homogenous staging. Absolute values of the parameters in (G) and (H) are shown, whereas the orientation is indicated as separate rightward and inverted groups in the x axis. n.s., not significant (t test, n = 17 Ccdc40+/+ embryos, n = 32 Ccdc40lnks/lnks mutant embryos with rightward looping, and n = 18 Ccdc40lnks/lnks mutant embryos with inverted looping direction). Dashed lines show the 95% distribution interval based on a Gaussian distribution of wild-type samples. Four (G) and seven (H) Ccdc40lnks/lnks mutants were outside this interval. Cran, cranial; Cau, caudal; D, dorsal; LV, left ventricle; R, right; RV, right ventricle; L, left; V, ventral. See also table S7 and movie S3. In the absence of node flow, variable laterality of Nodal in the LPM clusters phenotypes Given that the node triggers the asymmetric expression of Nodal (6) and that Nodal regulates asymmetric heart morphogenesis (9), we investigated Nodal patterning in Ccdc40 mutants at E8.5c. Compared to left-sided expression in the LPM of controls, mutants had variable Nodal patterns: left sided (two of eight), bilateral (three of eight), or right sided (one of eight) (Fig. 4). Whereas left-sided patterns were similar to controls, bilateral or right-sided patterns were less propagated along the anterior-posterior axis, failing to expand posterior to the node and reaching less anterior levels. Some embryos were collected immediately after imaging of the node flow, showing that pattern variability did not reflect detectable flow differences (Fig. 4, B2 to B4 and C2 to C4). This rather indicates that the laterality of Nodal is variable in the absence of node flow but still patterned in the LPM. Nodal patterning in Ccdc40 mutants. (A) Schematic representation of Nodal expression (white) at E8.5c. (B and C) Correlated analysis of embryos at E8.5c, showing Nodal expression by whole-mount in situ hybridization [white in (C)], after imaging of node flow (B) as in Fig. 1 (H and I). Genotypes are indicated above. The anterior limit of Nodal expression in the left and right lateral plate mesoderm (LPM) is highlighted with white and red arrowheads, respectively. Absent left-sided Nodal is indicated by an empty arrowhead. The embryo in C5 was not assessed for node flow and is shown in separate anterior and posterior views. Cran, cranial; Cau, caudal; L, left; R, right. In some embryos, both controls (one of five) and mutants (two of eight), we did not detect any Nodal expression in the LPM, whereas Nodal was expressed in the node. Because Nodal is very transiently expressed (9), this may reflect a too early stage of observation. To overcome the sensitive timing of Nodal expression and further correlate Nodal pattern with morphogenesis, we took advantage of the Nodal-ASE-lacZ transgenic line, in which β-galactosidase is produced under the control of the asymmetric enhancer of Nodal (43). Because of the perdurance of the lacZ mRNA and/or β-galactosidase protein, the transgenic pattern is still detectable after endogenous Nodal expression has been turned off (9). We crossed the transgene into Ccd40 mutants and dissected embryos at E9.5, to correlate the shape of heart looping with Nodal pattern. Eleven of the 23 Ccdc40 mutants were indistinguishable from wild types. They had a rightward looped heart tube with Nodal signal in the inferior-left outflow tract (OFT), superior-right AVC, and dorsal left atrium. Nodal signal was also seen in heart precursors in the left dorsal pericardial wall (DPW) and left cardiopulmonary progenitors (CPPs; Fig. 5, A1 and A2; tables S2 and S3; and movie S4). Nodal signal further expands posterior to the heart in the left LPM. In 2 of the 23 mutants with an inverted direction of heart looping, Nodal signal appeared as a mirror image (Fig. 5A3 and tables S2 and S3). However, the posterior LPM had bilateral Nodal signal. Last, 9 of the 23 mutants, with either a rightward or inverted looping direction, had more complex patterns of Nodal signal, abnormally inferior or absent in the AVC, absent in the DPW, and bilateral in the atria or CPPs (Fig. 5, A4 and A5, and tables S2 and S3). Overall, our longitudinal tracking of Nodal pattern at E9.5 supports the existence of three categories of Ccdc40 mutants: normal, complete mirror image, and defective. Clustering of Ccdc40 mutants at E9.5 based on Nodal pattern and heart shape. (A) Longitudinal analysis of Nodal expression by genetic tracing with Tg Nodal-ASE-lacZ at E9.5, shown in transverse sections of Ccdc40+/+ (A1) or Ccdc40lnks/lnks [(A2) to (A5)] embryos. β-Galactosidase staining in the dorsal pericardial wall (DPW) and atria (A) is in red and histology in gray, in mutants with a normal (A2), mirror-image (A3), or defective [(A4) and (A5)] distribution of cells which have expressed Nodal. (B to E) Unsupervised clustering of 67 E9.5 embryos, based on 43 geometrical parameters of the segmented looped heart tube, and five domains of Nodal-ASE-lacZ expression. Axes correspond to the partial least squares (PLS) regression components. Genotype (B), heart looping direction (C), alignment between the AVC and atria (D), and cluster identity (E) of embryos are mapped. (F) Corresponding clustering tree, including 17 controls and 50 mutants, of which 28 situs solitus (SS), 14 situs inversus totalis (SIT), and 8 heterotaxy (HTX) with rightward (n = 4) or inverted (n = 4) looping. (G) Parameters significantly associated with heterotaxy ordered by effect size, in pairwise comparisons between SS (controls and mutants) and heterotaxy with rightward looping (G1) or between SIT and heterotaxy with inverted looping (G2). Common parameters are in bold. (H) Illustration of abnormal medial connection of the AVC to both atria in a Ccdc40 mutant in the heterotaxy cluster (H3), compared to a single left atrium connection in SS (H1) and SIT (H2). (I) Illustration of abnormal outflow tract (OFT) shape, longer and straighter, in heterotaxy (I2), compared to SS (I1). (J) Representation of centroids. CPP, cardiopulmonary progenitors; D, dorsal; dOFT, distal OFT; IVS, interventricular septum; L, left; LA, left atrium; LV, left ventricle; R, right; RA, right atrium; RV, right ventricle; V, ventral. See also table S3 and movie S4. We noticed that the defects in geometrical parameters of the looped heart tube and in the laterality of specific domains of Nodal signal did not always correlate. We thus decided to take a global multivariate approach and performed unsupervised clustering of embryos, based on their variable parameters (table S3), resulting in a morphospace with four groups (Fig. 5, B to F). One cluster associates together wild types and Ccdc40 mutants with rightward heart looping, thus corresponding to the situs solitus phenotype. One cluster contains mutants with inverted heart looping and mainly normal AVC, corresponding to situs inversus totalis. The remaining clusters of mutants have consistent abnormal AVC, corresponding to heterotaxy, subdivided into either rightward or inverted heart looping direction. To identify the parameters that are most important to define heterotaxy, we performed pairwise comparisons of clusters with the same heart looping direction. We found that Nodal signal in the DPW, CPPs, atria and posterior LPM, as well as the alignment of the AVC and atria, the OFT shape, and the ventricle position were the most significant parameters associated with heterotaxy (Fig. 5, G to J). These features are reminiscent of heterotaxy defects at birth in the great arteries, complete atrioventricular septal defects, and left atrial isomerism (Fig. 2, D and E, and table S1). Together, our analysis of heart looping at E9.5 shows the origin of the categories of laterality phenotypes at birth: situs solitus, situs inversus totalis, and heterotaxy. In addition to the direction of heart looping, the pattern of mesoderm cells that have expressed Nodal as well as the geometry of the OFT, AVC, and position of ventricles appear as key factors to cluster phenotypes. Laterality of the NODAL pathway in the LPM does not correlate with node crown laterality in Ccdc40 mutants Lateralization of Nodal in the LPM was shown to depend on that in the node crown (8, 20). We thus wondered whether the variable laterality of Nodal in the LPM of Ccdc40 mutants reflects laterality in the node crown. We quantified Nodal asymmetry in the node crown at E8.5c. In controls, Nodal is 1.7 ± 0.3 fold more expressed on the left side, whereas it becomes symmetric on average in the absence of flow (Fig. 6A). In addition, the slight expression bias of Nodal in mutant nodes did not correlate with the expression side in the LPM, which was opposite or bilateral in most cases (five of six), except in one case of situs inversus totalis (yellow, Fig. 6A). Thus, LPM patterning did not generally meet expectations based on the node asymmetry. Uncoupling between asymmetry in the node and LPM in Ccdc40 mutants. (A) Whole-mount fluorescent in situ hybridization of Nodal (white) in the node (left panels) and LPM (right panels) of controls and Ccdc40 mutants at E8.5b-c. Its relative expression in the left (L) versus right (R) node crown is quantified, with data points color coded according to expression in the LPM of the same embryo. The pattern expected to be induced by asymmetry in the node is schematized. **P = 0.006 between controls and mutants (Mann-Whitney test, n = 5 control Ccdc40+/+ or Ccdc40lnks/+ embryos and n = 8 Ccdc40lnks/lnks mutant embryos). n.s., nonsignificant difference between the mutant mean and a symmetry hypothesis (P = 0.67, one-sample t test). Panels from Fig. 4 (C1 and C5) are reused. (B) Molecular cascade in the NODAL pathway. (C) Double whole-mount fluorescent in situ hybridization of Dand5 (pink, left panels) in the node and Lefty1 (white, right panels, shown in transverse sections) in the neural tube floor plate and LPM of controls and Ccdc40 mutants at E8.5c-e. Dand5 relative expression in the left versus right node crown is quantified, with data points color coded according to expression of Lefty1 in the same embryo. Note that Lefty1 is expressed at much lower levels in the LPM, so that a different signal intensity is required to assess asymmetry in the floor plate. Heterotaxy indicates bilateral expression or incongruent laterality between floor plate and LPM. The pattern expected to be induced by asymmetry in the node is schematized. The 95% distribution interval of control asymmetry, based on a Gaussian distribution, is outlined with dashed lines. **P = 0.0043 between controls and mutants (Mann-Whitney test, n = 5 control Ccdc40+/+ or Ccdc40lnks/+ embryos and n = 6 Ccdc40lnks/lnks mutant embryos). Means and SDs are shown. See also tables S4 and S7. Because Nodal is an indirect target of node flow (19), we investigated patterning of the direct target Dand5 in Ccdc40 mutants, together with a readout of NODAL signaling, Lefty1 (Fig. 6B). In wild types, Dand5, which encodes a NODAL antagonist, is 3.7 ± 1.5-fold more expressed on the right side of the node crown. In contrast, Ccdc40 mutants showed mainly symmetric Dand5 pattern, with a slight default left-sided bias (Fig. 6C). This was associated either with situs solitus patterning of Lefty1 (green in Fig. 6C and table S4) or nonlateralized patterns because of bilaterality or incongruence between the LPM and floor plate (red in Fig. 6C and table S4). In one mutant, Dand5 was significantly left predominant, with a fold-difference similar to wild types. Such inversion was associated with congruent inversion of Lefty1, both in the floor plate and LPM (yellow in Fig. 6C and table S4), indicative of situs inversus totalis. We conclude that the laterality of Nodal patterning in the LPM is uncoupled from the node in most Ccdc40 mutants, except in situs inversus totalis. Asymmetry of the WNT pathway and CPPs stratify laterality phenotypes To identify additional asymmetries in Ccdc40 mutants, we adopted a transcriptomic approach. We microdissected the field of heart progenitors at a specific stage (E8.5i) when looping direction can be determined. Paired left and right samples were collected for comparison of gene expression (Fig. 7A). Unsupervised clustering of gene expression asymmetry identifies three groups of embryos (Fig. 7B). As seen in our analysis at E9.5, one cluster associates together wild types and mutants with rightward heart looping, thus corresponding to the situs solitus phenotype. Another cluster contains mutants with inverted heart looping direction, corresponding to situs inversus totalis. The remaining cluster of mutants corresponds to heterotaxy with rightward heart looping. To extract the main transcriptomic signature of clusters, we performed a differential analysis comparing left and right expression within each laterality cluster. We thus identified a set of eight genes, which are common and significantly changed both between situs solitus and situs inversus totalis, as well as between situs solitus and heterotaxy (Fig. 7C). One gene, Ptgfr, is right sided in situs solitus (fig. S3A), whereas the others, Pitx1, Postn, Trabd2b, Retreg1, Ccn4, Misp3, and Hspb8 are left sided. Left-right transcriptomic signature of Ccdc40 mutant phenotypes at E8.5i. (A) Schematic representation of heart field microdissection at E8.5i for paired left-right transcriptomics. (B) Unsupervised clustering of 21 embryos, based on the 25 most consistent asymmetric genes in wild-type embryos. The genotype, sex, and heart looping direction of embryos are color coded. Situs solitus includes five wild-types and five mutants, situs inversus totalis includes five mutants, and heterotaxy includes six mutants. (C) Transcriptomic signature of clusters. Heatmap of the asymmetric expression of genes that are common and significantly changed in differential analyses between situs solitus and situs inversus totalis clusters and between situs solitus and heterotaxy clusters. Genes associated with WNT pathway GO terms are in bold. Asterisks indicate significance of asymmetry per cluster (adjusted P value). (D) Schematic representation of the six pairwise comparisons between embryo clusters (D1) used to perform a GO analysis of signaling pathway enrichment. Pathways are ordered by decreasing order of significance level (D2). The most significantly changed pathway is WNT (three GO terms in bold), representing a total of 56 distinct genes. BMP, bone morphogenetic protein signaling; VEGF, vascular endothelial growth factor. (E) Plot of asymmetric expression of WNT pathway genes (n = 56) in the indicated embryo clusters. The WNT pathway is fully left predominant in situs solitus, largely devoid of asymmetry in heterotaxy and several genes are right-enriched in situs inversus totalis. FDR, false discovery rate. (F) Plot of Ccn4 expression levels by transcriptomics. The adjusted P values of right over left expression levels are shown. FC, fold change; KO, knockout (Ccdc40lnks/lnks); L, left; n.s., not significant; R, right; WT, wild type. See also figs. S3 to S6 and table S5. Unexpectedly, Pitx2, a NODAL-target gene was not recovered as a main transcriptomic signature, because it was not differentially sequenced between situs solitus and heterotaxy (fig. S3B). Although Nodal patterning is a key determinant of laterality groups at E9.5, Pitx2 laterality is difficult to extract by sequencing, because it is entangled with symmetric expression domains (e.g., yolk sac) and several isoforms (fig. S3, C and D). However, transcriptomic changes in the NODAL pathway were detected (fig. S3E), notably in a right-predominant inhibitor Ism1. Overall, in situs inversus totalis, we identified 42 genes with an inverted laterality and a high number of gained asymmetries compared to situs solitus (fig. S3F). One example is Lefty2, a NODAL target, which is normally no longer expressed at E8.5i in wild types. However, it was significantly right sided in situs inversus totalis (fig. S3, E and F). This suggests that embryos with situs inversus totalis have abnormal kinetics of the NODAL pathway, potentially delayed, as also suggested by the lower anteroposterior propagation of Nodal expression in the LPM at E8.5c (Fig. 4C5). Among genes that gained asymmetry in situs inversus totalis are markers of the AVC (Bmp2 and Gal) or the sinoatrial node (Shox2). In heterotaxy, 84 genes became symmetric and several gained asymmetry compared to situs solitus (fig. S3G). Gained asymmetry again probably reflects a delayed developmental program, given the lower anteroposterior propagation of bilateral Nodal expression in the heterotaxy group at E8.5c (Fig. 4, C3 and C4). By in situ hybridization at sequential stages, we confirmed delayed expression of Lefty1 in Ccdc40 mutants, which is still high at E8.5e, when controls have down-regulated it (fig. S4). As a user-friendly community resource, we have created a web interface to interrogate this unique transcriptomic dataset, accessible at https://ccdc40.pasteur.cloud/. We then performed a Gene Ontology (GO) term analysis of genes identified in a broader differential analysis, including also differences in expression levels between embryo clusters (Fig. 7D1). Among the terms reflecting pathways, the WNT pathway emerged as the most significantly affected, representing 56 distinct genes (Fig. 7D2 and table S5). Whereas this set of genes was left sided in wild types, it was largely symmetric in the heterotaxy cluster and showed either reversed laterality (e.g., Ccn4, Wnt11, and Pitx2) or gained asymmetry (e.g., Sfrp2, Wnt16, and Fzd10) in the situs inversus totalis cluster (Fig. 7, C and E). WNT pathway genes were mainly affected for their laterality (Fig. 7E) rather than their level of expression (fig. S5E). Ccn4, also known as Wnt-inducible signaling pathway protein 1 (Wisp1), stood out as the most representative of embryo clusters among genes of the WNT pathway (Fig. 7F): left sided in situs solitus, right sided in situs inversus totalis, and symmetric in heterotaxy. We further explored the spatial expression pattern of Ccn4. We found high expression in a specific domain caudal to the heart tube at E8.5i, covering the sinus venosus and CPPs, which face the infolding of the lung bud in the foregut endoderm (Fig. 8, A to E). These Ccn4-positive cells do not express the myocardial marker Acta1 (Fig. 8C) but are positive for the CPP marker Wnt2 (Fig. 8B). Ccn4 expression is enriched on the left side, overlapping with the left marker Pitx2 (Fig. 8, A and D). Ccn4 is also expressed in the yolk sac, but symmetrically, and at low levels in the heart tube (fig. S5, A to C). To validate our transcriptomic analysis, we mapped Ccn4 expression in Ccdc40 mutants (Fig. 8, F and G). We found 4 of the 11 mutants with left-predominant Ccn4 expression, similar to wild types. These mutants had a rightward heart looping direction, reflecting the situs solitus cluster. In 3 of the 11 mutants, Ccn4 expression was reversed, as well as heart looping, reflecting the situs inversus totalis cluster. The remaining 4 of the 11 mutants displayed symmetric Ccn4 expression. These mutants had either a rightward or inverted heart looping direction, reflecting the heterotaxy cluster. Ccn4 thus identifies a previously unknown asymmetry in CPPs. Additional markers of CPPs, such as Osr1, can become asymmetric in situs inversus totalis or heterotaxy (fig. S5D). Asymmetric expression of Ccn4 in CPPs. (A to D) Whole-mount in situ hybridization of Ccn4 (white) in wild-type embryos at E8.5i, relative to markers of CPPs (Wnt2, red in B, n = 6), cardiomyocytes [Acta1, red in (C), n = 4], and left [Pitx2, red in (D), n = 2]. The white arrowhead points to Ccn4 expression in undifferentiated cardiac cells, the orange to overlapping expression of Ccn4 and Pitx2 in CPPs and the red to Pitx2 only expression in the DPW. Transverse [(A1) and (B)], frontal (A2), and sagittal [(C) and (D)] sections are shown. (E) 3D segmentation of the Ccn4-positive domain relative to the heart tube (yellow) and the foregut endoderm (gray), divided in right (red) and left (blue) sides. Segmentations are shown overlying the raw in situ hybridization (E1), in ventral [(E1) and (E2)] or dorsal (E3) views. (F) Whole-mount Ccn4 expression (white) in Ccdc40 mutants at E8.5i, seen in transverse sections, showing three patterns: left-predominance in embryos with a rightward heart looping [(F1); n = 4], right-predominance in embryos with inverted heart looping [(F3); n = 3] and symmetric expression [(F2); n = 4] in the sinus venosus and CPPs. (G) Corresponding quantification in left/right domains shown in (E) in embryos with the indicated genotype and heart looping direction. Abnormal values (red) are outside the 95% distribution interval of control samples based on a Gaussian distribution (dashed lines). (H and I) Ccn4 expression in mutants, in which Nodal is inactivated throughout the LPM (H) and corresponding quantification of its asymmetry (I). CPPs, cardiopulmonary progenitors; Cran, cranial; Cau, caudal; D, dorsal; DPW, dorsal pericardial wall; E, endoderm; L, left; LA, left atrium; LB, lung bud; LV, left ventricle; NT, neural tube; OE, oral ectoderm; PM, posterior mesoderm; R, right; RV, right ventricle; SV, sinus venosus; V, ventral. See also fig. S5 and table S7. Overall, we demonstrate a strong WNT pathway signature of the laterality phenotypes. We identify for the first time an asymmetry in CPPs, marked by the WNT pathway gene Ccn4, which stratifies phenotypes and is concordant with combined heart and lung defects in heterotaxy and situs inversus totalis. Cross-talk between WNT and NODAL pathways is required to establish LPM asymmetry WNT and NODAL pathways are known to regulate each other during mesoderm formation or node asymmetry (33, 44, 45). To understand whether asymmetry of the WNT pathway gene Ccn4 depends on NODAL signaling, we analyzed it in Nodal mutants. In the absence of Nodal in the LPM, Ccn4 expression became symmetric (Fig. 8, H and I). This indicates that NODAL signaling can modulate the WNT pathway to bias CPP asymmetry. Given our observation that situs solitus can be established in Ccdc40 mutants without node flow and node asymmetry, we wondered whether WNT signaling could activate Nodal in the LPM. By culturing precisely staged embryos and applying the WNT pathway activator CHIR99021 (46), we monitored cumulative patterning with the Nodal-ASE-lacZ transgene. Pharmacological WNT activation applied after the onset of Nodal expression at E8.5c did not affect Nodal patterning (Fig. 9, G to J). However, treatment applied before Nodal onset, impaired its expression in most cases (six of eight; Fig. 9, A to F): Nodal was not detected (n = 2; Fig. 9C), asymmetric in the anterior LPM only (n = 1; Fig. 9D) or asymmetric posteriorly but bilateral anteriorly (n = 3; Fig. 9, E and F). This was seen regardless of the genotype, in control Ccdc40+/lnks or mutant Ccdc40lnks/lnks embryos. Impaired Nodal patterning was occasionally associated with abnormal heart looping direction (Fig. 9, C and F). We thus provide functional evidence of a cross-talk between WNT and NODAL pathways that is important to establish LPM asymmetry. Pharmacological WNT activation modulates Nodal patterning in the LPM. (A) Example of an embryo dissected before expression of asymmetric Nodal, i.e. between the neural plate (NP) and late headfold (LHF) stages, at the beginning of the culture. (B) Example of a Ccdc40+/lnks;Tg Nodal-ASE-lacZ embryo (n = 3) at the end of the culture, treated 4 hours with the adjuvant [dimethyl sulfoxide (DMSO)]. β-Galactosidase staining is in blue in bright-field images (left) and red in 3D HREM images, shown in transverse (middle) and frontal (right) sections. The filled arrowhead indicates normal staining in the left DPW, and the gray arrow indicates normal rightward heart looping. (C to E) Tg Nodal-ASE-lacZ embryos (n = 7 Ccdc40+/lnks controls and n = 1 Ccdc4lnks/lnks mutants), at the end of the culture, treated 4 hours with the WNT pathway activator CHIR. The empty and red arrowheads indicate absent and bilateral staining in the left DPW, respectively. The yellow arrow indicates abnormal heart looping direction. The forming heart is outlined in white in (D). Two Ccdc40+/lnks embryos had normal Nodal patterning. (G) Example of an embryo dissected after expression of asymmetric Nodal, i.e., at E8.5c-g stages, at the beginning of the culture. (H) Example of a Tg Nodal-ASE-lacZ embryo (n = 1 Ccdc40+/+, n = 3 Ccdc40+/lnks controls, and n = 2 Ccdc4lnks/lnks mutants) at the end of the culture, treated 4 hours with the adjuvant (DMSO). (I and J) Tg Nodal-ASE-lacZ embryos (n = 1 Ccdc40+/+, n = 8 Ccdc40+/lnks controls, and n = 3 Ccdc4lnks/lnks mutants) at the end of the culture, treated 4 hours with the WNT pathway activator CHIR. Scale bars, 100 μm. h, hours. DISCUSSION We find that Ccdc40 mutants have fully penetrant motile ciliopathy that abrogates node flow and causes hydrocephalus. We thus demonstrate that node flow is not the only event of symmetry breaking in mice. It is required as a bias for the robustness of left-right patterning, but asymmetry can be established independently of the node. With quantitative and longitudinal analyses, we reconstructed the developmental trajectories of laterality defects in motile ciliopathy (Fig. 10). We show the existence of a situs solitus developmental trajectory, independent of the node, with left-sided Nodal patterning, normal asymmetric transcriptomics in the heart field, overall normal embryonic morphogenesis, and organ laterality at birth, but occasionally mild asymptomatic anomalies. In another developmental trajectory, situs inversus totalis, fully mirror image embryonic morphogenesis, and organ laterality at birth arise from right-sided Nodal patterning and an inverted asymmetric transcriptomic signature including in the WNT pathway. This can be traced back to baseline left-sided Dand5 asymmetry in the node crown. However, this trajectory shows anomalies in Nodal patterning, with temporal delay and posterior bilaterality. Last, the heterotaxy trajectory manifests in the absence of anterior Nodal patterning in the DPW, bilateral Nodal patterning in atria and CPPs. It is associated with abnormal symmetry of the WNT pathway, incongruent asymmetric cues, variable direction, and defective geometry of heart looping. At birth, this trajectory results in a typical heterotaxy syndrome with left isomerism of the airways and atrial appendages, as well as complex congenital heart defects. Overall, we demonstrate that laterality phenotypes not only correlate with Nodal, but also with WNT pathway patterning. Our transcriptomics identifies a number of asymmetric genes that characterize these trajectories. We also uncover asymmetry in a cell population relevant to laterality defects, the CPPs. Working model. Node flow is not required as an asymmetry generator, but instead as a bias for asymmetry. In the presence of flow, left-right patterning is robust with fully penetrant situs solitus of the heart and lungs (top line). In the absence of flow, phenotypic output is variable, with NODAL and WNT pathways as the main asymmetric signatures that stratify phenotypes. The node crown is generally symmetric (middle line) but left-right patterning of the LPM can develop independently of the node, giving rise to a situs solitus trajectory in 70% of cases. Anomalies in LPM patterning, symmetry or incongruence, give rise to a heterotaxy trajectory in 15% of cases. Rarely, baseline left-sided Dand5 asymmetry in the node crown can reach a threshold sufficient to trigger reversed patterning of the LPM, giving rise to a situs inversus totalis trajectory in 15% of cases. The three categories of laterality phenotypes at birth seen in Ccdc40 mutants are in line with the phenotype of other mice or patients with motile cilium gene variants (29, 47). Although the node flow is upstream of NODAL signaling (6, 13), abrogating node flow leads to a completely different spectrum of defects compared to Nodal inactivation, which leads to full penetrance of heterotaxy (9). Heart looping direction in Nodal mutants is uniformly randomized, yet their inverted (leftward) looping is not a mirror image and no situs inversus totalis is detected. All Nodal mutant fetuses show heterotaxy with right isomerism of the airways and atrial appendages, while heterotaxy occurs with left isomerism in Ccdc40 mutants. In the clinics, heterotaxy with right or left isomerism tends to be associated with specific congenital heart and spleen defects (1, 48). Our quantitative analyses of Ccdc40 and Nodal mutants now outline a mechanistic stratification of laterality defects. Human patients with variants in motile cilium genes display laterality defects similar to Ccdc40 mouse mutants, as well as the respiratory disease primary ciliary dyskinesia and infertility (26, 27, 49). Situs inversus totalis in patients and mice is associated with ciliary genes (3), with one of them, Inversin, regulating the WNT pathway (50). Using innovative tools, we have assessed Nodal patterning with higher resolution. Nodal is transiently expressed and dynamically propagates spatially (6, 9, 51), the timing of which can be affected in mutants (28, 32, 52). Our genetic tracing of Nodal with a transgenic reporter opens the possibility to analyze cumulative NODAL signaling, better catching its overall expression pattern and correlation with embryonic morphogenesis. The sensitive fluorescent in situ hybridization that we have used provides quantitative insight into asymmetries, compared to previous chromogenic substrates. Our paired transcriptomic analysis shows that situs inversus totalis is not only a phenotypic mirror image of situs solitus but also a different developmental trajectory, with delayed NODAL signaling. On the basis of 3D imaging, we have uncovered a cardiac cell population with left-right asymmetry, the CPPs. This population had been previously characterized as a pool of common precursors to the heart and lung, marked by WNT signaling components (53, 54). We now report asymmetry in Ccn4 and Nodal genetic tracing in CPPs, which provides a molecular basis to combined heart and lung defects in heterotaxy. Nodal mesoderm patterning in Ccdc40 mutant embryos is channeled in three types—left sided, anteriorly right sided, and abnormal—in agreement with other mutants disrupting the node flow or flow sensing (6, 28, 29). Their frequency is not uniformly randomized in Ccdc40 mutants, and Nodal is spatially not noisy but still expressed in stripes of LPM. This raises the question of the origin of the variations. Partial disruption of the node flow can explain phenotypic variability (30). However, Ccdc40 mutants have a fully penetrant undetectable node flow, and situs solitus mutants still manifest hydrocephalus, in agreement with the reported immotility of cilia devoid of CCDC40 (55). Within the current model of a linear cascade of left-right patterning, a single bilateral Nodal pattern would be expected in the absence of node flow, because this prevents the left-sided decay of Dand5. Dand5 is symmetric in most Ccdc40 mutants, as shown by us and others (28), and in other models disrupting flow sensing by Pkd2 or Dand5 (19, 56, 57). In the opposite situation, upon Dand5 inactivation or excessive bilateral decay, Nodal displays similar left-sided, right-sided, or bilateral variations as in Ccdc40 mutants (18, 58). Thus, the node and its readout of Dand5 asymmetry do not appear to be strictly required to set Nodal asymmetry but rather to bias it and make it robustly left or right sided. We show uncoupling between asymmetry in the node and in the LPM in the absence of flow. Together, these observations support the conclusion that, although Nodal asymmetry in the mesoderm is sensitive to the node, it can also be established in the LPM by a different mechanism. General principles of symmetry breaking have been proposed, including two main ingredients: a bias that is local and able to orient the asymmetry and an amplification mechanism that can spontaneously generate asymmetry but with no specific orientation (59, 60). The bias can rely on molecular chirality, as shown for tubulin, which orients the rotation of cilia and thus determines the direction of flow (13, 61). Brown and Wolpert (59) interpreted situs variations in Dnah11iv mutant mice as a loss of function in biasing, which is in line with our conclusion for Ccdc40 mutants. In these motile ciliary mutants, spontaneous generation of asymmetry seems to rely only on the second ingredient and occurs with a variable orientation. The gold standard of spontaneous amplification mechanisms is the reaction-diffusion process, depending on self-amplifying feedback and long-range inhibition (62, 63). This mechanism was modeled in the LPM on the basis of the activator NODAL and its long-range inhibitor LEFTY (64). This was considered as a way to amplify the low asymmetry generated by the node, rather than independent of it. However, in the absence of Nodal in the LPM, asymmetric morphogenesis is not lost, some heart looping occurs (9) and gut looping is largely normal (65). This suggests that there are factors other than NODAL participating in symmetry breaking. As a paradigm shift in left-right patterning, we now propose that the node is not the primary driver of symmetry breaking. We instead suggest an amplification mechanism in the LPM, independent of the node and involving NODAL and WNT pathways. Such randomly oriented mechanism is then biased by the node flow. Our paired transcriptomic analyses of the heart field uncover a strong asymmetric signature in the WNT pathway, which stratifies laterality phenotypes. Among the 56 genes significantly changed between laterality groups are Ccn4, Trabd2b, and Wnt11. TRABD2B is a metalloprotease that acts as a negative regulator of the Wnt signaling pathway (66). The WNT pathway has previously been shown to be asymmetric, with higher Wnt3 levels in the left node (33, 34). DAND5 can also antagonize WNT apart from NODAL (67). Wnt3 and Dand5 act in feedback loops in the node to reinforce asymmetry (33). Whether WNT in the LPM could also induce NODAL signaling in the absence of flow is an attractive possibility. Wnt3a in the posterior node and caudal mesoderm is required for Nodal patterning and asymmetric organogenesis (44). In the epiblast, WNT signaling is important to induce Nodal expression (45). Our embryo culture experiments now show that WNT activation before asymmetric Nodal expression can modulate its patterning, which could be mediated by direct activation of Lefty1 (68). Similarly in transgenic medaka, ubiquitous activation of WNT1 at two to four somites induces bilateral Nodal expression in the LPM (69). Our transcriptomics analyses uncovered other asymmetric markers, such as the NODAL inhibitor Ism1, which is broadly expressed in the LPM (70, 71). How Ism1 asymmetry is regulated is unknown. Other candidate pathways in our transcriptomics include prostaglandin signaling or PITX1, but whether they can induce Nodal expression remains to be investigated. Similarly to the mouse, ablation of the left-right organizer or abrogation of flow in other vertebrates is still compatible with normal laterality, with a lower frequency (72–75). Additional pathways other than NODAL signaling have been found. In the fish brain, WNT signaling in addition to NODAL is required for the asymmetry of the habenula (76, 77). In the frog, chick, and rabbit, fibroblast growth factor (FGF) signaling is an important regulator of left-right patterning, upstream of NODAL signaling (15, 78, 79). In the fish and frog, asymmetric cues before the left-right organizer have been observed, including maternal determinants, such as the TGFβ growth and differentiation factor 1 (GDF1) (80, 81). Such early developmental cues have not yet been reported in mice. If the node has a more restricted and biasing role, then the long-standing concept of a left-right organizer may be questioned. Blum et al. (12) introduced the term of “laterality organizing center” to avoid confusion in the use of the term “node” and distinguish the Spemann-Mangold organizer activity from later laterality patterning activity. An organizer is formally defined as a tissue that can both induce and pattern adjacent cells (82, 83), but the node is now shown to lack the inducing activity in this definition. Essner et al. (72) named the orthologous Kupffer’s vesicle in the fish as a transient embryonic “organ of asymmetry,” a term that would be more relevant to the finding that the node is required for asymmetry robustness rather than induction. Overall, our in-depth quantitative analysis of Ccdc40 mutants provides original insights into the mechanisms of symmetry breaking in the mouse and the developmental trajectories of laterality defects. Limitations of the study Our study stratifies laterality defects and identifies anatomical and molecular signatures. Functional investigations will be important to clarify the causal links between them. Given the low penetrance of heterotaxy and situs inversus totalis in Ccdc40 mutants, it has been challenging to collect and quantify a cohort at E9.5 with the transgenic marker of Nodal. We have enriched the study with nontransgenic samples, using an integration method based on partial least squares (PLS) regression and thus explored how phenotypes could be clustered from 48 parameters. Further increments of the pipeline with more samples and better knowledge of the essential asymmetric parameters will reinforce the precise boundaries between the laterality groups of embryos. For example, the frequency of cases with incongruent asymmetry between different markers exceeds that of heterotaxy cases at birth, suggesting that some are compatible with situs solitus. At birth, the mild anomalies that we detect in liver or lung lobations are now not used as markers of laterality in patients, although they are detected in many laterality mouse mutants. Knowledge of their developmental origins is now lacking to understand whether they reflect direct or indirect asymmetry mechanisms. MATERIALS AND METHODS Animal models Ccdc40lnks/+ males and females were crossed to generate Ccdc40lnks/lnks mutant embryos (25). The line was kept in a C56Bl6J background. The Tg Nodal-ASE-lacZ transgenic line in a mixed genetic background (43) was crossed with Ccdc40lnks/+ animals to generate Ccdc40 mutants carrying the transgene. For generation of conditional Nodal mutants, Nodalnull/+;Hoxb1Cre/+ males of a mixed genetic background were crossed with Nodalflox/flox females (9). The Ccdc40lnks/+ mouse line develops faster, so that collection of fetuses the day before birth corresponds to stage E17.5. Whenever possible, littermate wild types were used as controls. In some low-yield experiments, they were pooled with littermate Ccdc40lnks/+ heterozygotes, which have no phenotype (referred to as controls). Male and female embryos were collected and used randomly. E0.5 was defined as noon on the day of vaginal plug detection. At E8.5, embryos were staged on the basis of the morphology of the heart (fig. S1A) (9, 38). All embryos and fetuses were genotyped by polymerase chain reaction (PCR) of the yolk sac or tail tip, using primers listed in table S6. Mice were housed in individually ventilated cages containing tube shelters and nesting material, at 21°C and 50% humidity, under a 12-hour light/dark cycle, with food and water ad libitum, in the Laboratory of Animal Experimentation and Transgenesis of the Structure Fédérative de Recherche (SFR) Necker, Imagine Campus, Paris. Animal procedures were approved by the ethical committees of the Institut Pasteur, Université Paris Cité, and the French Ministry of Research (nos. 18049-201707201335745 v9, 31733 2020120811176379_v6, dha230022, and 60334-2025021116457052_v16). Ccdc40 mutant genotyping For genotyping of Ccdc40 alleles, forward and reverse primers spanning a region containing the lnks C>A point mutation site were designed (table S6). After PCR amplification, PCR products were subjected to enzymatic restriction by Hpy188I, which differentially finds one and two restriction sites in the mutant and wild-type amplicons, respectively. The expected products are shown in fig. S6A. Scanning electron microscopy Embryos were collected at E8.5b-d and fixed overnight in 2.5% glutaraldehyde. After three washes in 0.1 M cacodylate buffer (pH 7.4), samples were postfixed in 1% osmium tetroxide in 0.1 M cacodylate buffer (pH 7.4) for 1 hour. After three washes in water, samples were dehydrated in ethanol. Samples were dried in an automated critical point dryer (Leica EM CPD300) and coated with gold/palladium in a Gun ionic evaporator (PEC 681/1). Images were acquired on a JEOL IT700HR scanning electron microscope. β-Galactosidase staining Tg Nodal-ASE-lacZ embryos were collected at E9.5. They were incubated in cold 250 mM KCl cardioplegia solution for 1 min; fixed for 10 min in 4% paraformaldehyde (PFA), 5 mM EGTA, and 2 mM MgCl2; then permeabilized for 30 min in 0.2% NP-40 and 0.1% sodium deoxycholate; and lastly stained whole mount with X-gal at 37°C in the dark, overnight. After rinsing in PBS, the staining was postfixed in 4% PFA overnight before 3D imaging by high-resolution episcopic microscopy (HREM). The transgenic pattern was recorded independently in different domains (tables S2 and S3). Four regions of the heart tube were considered: the OFT, AVC, atrium facing the right ventricle, and atrium facing the left ventricle. Two regions correspond to cardiac precursors: DPW and CPPs and sinus venosus. Last, the LPM posterior to the heart (posterior LPM) reflects noncardiac cells. Chromogenic RNA in situ hybridization Embryos were collected at E9.5. After cardioplegia in cold 250 mM KCl for 1 min, embryos were fixed in 4% PFA overnight at 4°C. Fixed embryos were gradually dehydrated in methanol and stored at −20°C until whole-mount in situ hybridization. Wnt11 and Bmp2 antisense riboprobes (gifts from S. Evans and C. Vesque, respectively) were transcribed from plasmids using digoxigenin-11-UTP. Signals were detected by alkaline phosphatase–conjugated anti-digoxigenin antibodies (1/2000), which were revealed with the chromogenic BM purple (magenta) substrate. Samples were postfixed and imaged in 3D by HREM. Fluorescent RNA in situ hybridization Embryos were collected at E8.5, fixed in 4% PFA overnight at 4°C, gradually dehydrated in methanol, and stored at −20°C. Sensitive whole-mount RNAscope in situ hybridization was performed using the Mutliplex Fluorescent 426 v2 Assay (Bio-Techne) and a published protocol (84). Nodal (436321-C1), Dand5 (852201-C3), Lefty1 (506381-C2), Ptgfr (501841-C1), Ccn4/Wisp1 (501921-C1), Ccdc40 (1054111-C2), Isl1 (451931-C1), Pitx2 (412841-C2), Acta1 (808831-C3), and Wnt2 (313601-C2) probes were used. Amplification steps were performed using the TSA cyanine5 and cyanine3 amplification kit. Hoechst (1/1000) was used as a nuclear counterstain. Samples were then cleared in R2 CUBIC reagent and embedded in R2 containing agarose. Multichannel multisection 16-bit images were acquired with a Z.1 light-sheet microscope and a 20×/1.0 objective. Bulk RNA sequencing Embryos were collected at E8.5, and a ventral bright-field image of each embryo was acquired to record the heart looping stage. Then, embryos were microdissected, removing the head, back, trunk below the second somite, and heart tube (40). The remaining tissue, largely consisting of the heart field in the DPW and down to the second somite, was cut along the midline in two left and right halves, separately collected, and flash-frozen in liquid nitrogen. Samples were then stored at −80°C. Embryos were selected for bulk RNA extraction on the basis of stage (E8.5i), genotype, direction of heart looping, and sex (mixed). E8.5i is defined by the tilted position of ventricles, at an angle of about 40° to the anteroposterior axis, with a visible OFT and AVC (38). Given the low penetrance of inverted heart looping (Fig. 3C) and the difficulty to catch a specific transient stage, we had to collect 276 embryos at E8.5 to find five samples with inverted heart looping at E8.5i. We selected blindly a twice higher number of mutant samples with rightward heart looping (n = 11) and an equivalent group of controls (n = 5) for a total cohort of 21 embryos. Total RNA was extracted in TRIzol-Chloroform and purified using the RNeasy Micro Elute Columns kit (QIAGEN) including deoxyribonuclease treatment. RNA quality and quantity were assessed by capillary electrophoresis using High Sensitivity RNA reagents with the Fragment Analyzer (Agilent Technologies). All RNA quality numbers were >9.1. Only paired left/right samples with minimum 10 ng of RNA were retained: 5 paired wild-type samples (4 males and 1 female), 11 paired mutant samples with a rightward looping direction (6 males and 5 females), and 5 paired mutant samples with an inverted looping direction (3 males and 2 females). Batches of RNA preparation and library preparation were randomized. RNA sequencing (RNA-seq) libraries were prepared starting from 10 ng of total RNA using the NEBNext Single Cell/Low Input RNA Library Prep Kit for Illumina. The cDNAs produced from the poly-A+ fraction were PCR amplified in two steps (9 cycles for the first amplification and 8 for the second amplification). An equimolar pool of the final indexed RNA-seq libraries was sequenced on an Illumina NovaSeq6000, with paired-end reads of 100 bases and a mean sequencing depth of 80 million reads per sample. Embryo culture Embryos from Ccdc40lnks/+; Tg Nodal-ASE-lacZ intercrosses were dissected in warm Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum, 8 days after fertilization, corresponding to neural plate–E8.5g stages. With an intact ectoplacental cone, embryos were cultured in rolling bottles as previously published (85). A gas mixture containing 5% CO2 and 13% O2 was pumped from a mixer box at 6.5 psi (positive pressure over ambient external atmospheric pressure) into a water bottle; then, this humidified gas was circuited at 0.1 psi into the rotating drum. Embryos were kept in individual vials with 1 ml of culture medium [DMEM, GlutaMAX, penicillin/streptomycin, Hepes, nonessential amino acids, sodium pyruvate, glucose, 50% inactivated rat serum (Janvier Labs), and 25% inactivated human serum (HumanCells Biosciences, FP-006-C200)] (85). After 30 min in culture for postdissection recovery, 20 to 40 μM CHIR99021 (Sigma-Aldrich), or an equivalent volume of the adjuvant (dimethyl sulfoxide) alone, were added into the vials. After 4 hours of treatment, embryos were transferred into fresh culture medium without drug and kept in culture for a total of 24 hours. Yolk sacs were collected for genotyping, and dissected embryos were processed for β-galactosidase staining and imaging. Live imaging of node flow Embryos were collected at E8.5b-d in phenol red-free DMEM with 10% fetal bovine serum (culture medium) at 37°C, as established previously (30, 86). Preheated medium and a microscope heating pad were used to control the temperature. Embryos were imaged individually, while the rest of the litter was cultured in the culture medium at 37°C in rolling bottles in a precision incubator (BTC Engineering, Milton, Cambridge, UK). For live imaging, embryos were mounted on a microscopy glass with a thin layer of silicone glue, flattened with ventral side up, and covered by a ∼20-μl droplet of culture medium. Green-fluorescent beads (0.2 μm) were vortexed and diluted 1:10 in culture medium, and 2 μl was added on the embryo. Silicone vacuum grease was applied surrounding the embryo, and mounting was closed with a coverslip. Single-channel 16-bit images were acquired using a confocal microscope equipped with Yokogawa CSU-X1 Spinning Disk technology on an inverted Axio Observer.Z1, with an incubating chamber at 37°C, and an ORCA FLASH4 high-speed camera (Hamamatsu). Bright-field images with 10× and 63×/1.46 oil immersion objectives were acquired to record embryo orientation. Then, movies on the green fluorescent channel with the 63×/1.46 oil immersion objective were acquired, for three focal planes per embryo, at 100 fps over 3 s. After imaging, seven embryos were detached by gentle pipetting and fixed in 4% PFA overnight at 4°C, dehydrated in methanol. They were stored at −20°C and Nodal expression was evaluated by fluorescent RNA in situ hybridization. Micro–computed tomography Fetuses at E17.5 (n = 40), P0 (n = 7), or P14 (n = 16) were collected and euthanized by decapitation. As described previously (41), the body or head was immerged in HBSS at 37°C 5 min to remove blood and then in cardioplegia solution (110 mM NaCl, 16 mM KCl, 16 mM MgCl2, 1.5 mM Cacl2, and 10 mM NaHCO3) at 4°C. Fetuses were fixed in 4% PFA 24 hours at 4°C, washed, and stored in PBS azide. Before imaging, the left forelimb was cut as a lateral landmark, and skin around thorax was removed to facilitate penetration lugol as a contrast agent, incubated for 72 hours (E17.5 and P0) or 96 hours (P14). Images were acquired on a Micro–Computed Tomography Quantum FX (PerkinElmer). Field of exposure was 10-mm diameter, and xyz resolution was 20 μm. After imaging, fetuses were recovered and lugol was washed out. In E17.5 and P0 cohorts, as a follow-up analysis, the heart was explanted for HREM 3D imaging. High-resolution episcopic microscopy Whole embryos (E9.5) or explanted hearts (E17.5 to P0) were embedded in methacrylate resin (JB4) containing eosin and acridine orange as contrast agents (38, 41). The resin block was sectioned with a microtome, and images of its surface were acquired using the optical high-resolution episcopic microscope and a 1× Apo objective, repeatedly. The tissue architecture was imaged with a green fluorescent protein filter and chromogenic precipitates with a red fluorescent protein filter. E9.5 embryo datasets include 720 to 1765 images with of 1 to 2.1 μm resolution for all three (xyz) dimensions. E17.5-P0 heart datasets include 1000 to 1700 images, with 2.6- to 3.8-μm resolution for the three dimensions. Occasionally, one slice may be lost or overexposed, creating a minor variation in the overall 3D stack. Icy (87) and Fiji (ImageJ) (88) software were used to crop and scale the datasets. 3D reconstructions and analysis were performed using Imaris. Figure panels correspond to optical sections in the most appropriate plane, i.e., not necessarily in the plane of imaging. Phenotyping at perinatal stages Laterality defects in thoracic and abdominal organs were phenotyped on the basis of micro–computed tomography (micro-CT) scans, including bronchi, lungs, heart, venous return to the heart, aortic arch, stomach, spleen, liver, and colon (41). Cardiac anatomy was evaluated in HREM images on the basis of the segmental approach (89) and the clinical code from the International Paediatric and Congenital Cardiac Code (IPCCC) and the 11th iteration of the International Classification of Diseases (ICD-11). Phenotyping was conducted by two independent observers: one developmental biologist and one cardiac pediatrician. The position of the heart apex corresponds to that of the apex of the left ventricle within the thoracic cavity. The anatomical right and left ventricles are defined on the basis of the presence and absence of a septal attachment of the atrioventricular valve, respectively. The position of ventricles reflects the congruent (D-Loop) or not (L-Loop) position of the anatomic right ventricle on the right side of the left ventricle. Malposition of the great arteries was diagnosed in a transverse plane on the basis of the position of the aortic valve relative to the pulmonary valve. The ventriculo-arterial connection reflects the position of the great arteries, scored as abnormal in the following two cases: double-outlet right ventricle, when more than half of both arterial trunks arise from the anatomic right ventricle; transposition of the great arteries when the aorta arises from the anatomic right ventricle and the pulmonary trunk from the anatomic left ventricle. The anatomic right and left atria are defined based on the presence and absence of pectinate muscles in the posterior wall of the atrial chamber, respectively. The anatomic right and left lungs are defined on the basis of the number of lobes (four on the right and one on the left). The anatomic right and left bronchi are defined on the basis of their subdivision (early on the right and late one on the left). Heterotaxy was defined according to the criteria of Lin et al. (1). Quantification of heart looping at E9.5 HREM images were used to segment the different compartments of the heart tube using Imaris software. Segmentations were done following histological criteria (movie S3), such as cushion boundaries or the interventricular or interatrial sulcus, aided by the expression patterns of Wnt11 and Bmp2, as described previously (9). Nine landmarks along the tube were extracted and used for quantifications, of which three are two-dimensional polygonal cross sections of the tube and six are 3D segmented volumes. Polygons were placed at the distal exit of the OFT, at the interventricular sulcus, and at the point of bifurcation into the two atria. The segmented 3D structures are OFT, right ventricle, left ventricle, AVC, left atrium, and right atrium, of which a centroid and volume were extracted. The angles between three consecutive points along the tube axis were taken. Heart shapes were aligned in 3D using an in-house MATLAB code (84) (DOI: 10.5281/zenodo.8265324) so that the z axis corresponds to the notochord and the x axis to a perpendicular from the neural tube midline to a ventral point. The centroid of the distal OFT is set as the center of reference axes. The orientation of the axis between the left and right ventricles, the displacement of the AVC (taken here at the bifurcation point) from the notochord, and the tube length were measured as previously (9). Abnormal values were determined as falling outside the 95% distribution interval of control samples. Deviation in the shape of the OFT in mutants was evaluated qualitatively on the basis of length and curvature. The alignment of the AVC and atria was evaluated qualitatively: Alignment with the left atrium is the wild-type configuration, and partial alignment with the right, in addition to the left atrium, is defective. Quantification of fluorescent in situ hybridization signal 3D images were analyzed using Imaris software for the manual segmentation of expression domains. The presence of unspecific staining clumps was an exclusion criterion. Landmarks for segmentation of the node crown were set by Nodal or Dand5. For CPPs and sinus venosus, segmentation was set in the bilateral horns, from the posterior end of atria to the junction with the common cardinal veins. Differentiated myocardium (marked by Acta1 or histology), proepicardium, and endocardium were not included in the CPP region. After segmentation, the “Spots” tool in Imaris was used, with the “Different spot sizes” feature, to detect fluorescent signal. The volume of the detected spots was extracted and normalized by the total segmented volume and by Hoechst intensity. The resulting values were used for the calculation of the ratio between the left-side and right-side expression. Quantification of node cilia phenotype For quantification of the frequency of ciliary ballooning, the “Cell Counter” plugin in Fiji was used, and a manual marker was placed in each cilium, classifying them as the “normal” or “ballooned” type. For delimitation of concentric zones in the node, GIMP (GNU Image Manipulation Program) software was used. The number of ballooned cilia, divided by the total number of cilia, per zone, was calculated for each embryo. Particle image velocimetry analysis Live imaging of the node produced stacks of 220 to 280 images. Frames were individually processed in Fiji: Background illumination was subtracted by applying the “Gaussian blur” filter and subtracting, and, then, salt-and-pepper noise was removed with the “Despeckle” plugin. Image stacks were analyzed using the particle-image-velocimetry software (90) on interrogation windows with size 48 by 48 pixels. Flow maps were computed by averaging the flow over the duration of image acquisition. Node areas were selected manually on the basis of the visible distribution of fluorescent beads within a focal plane in the pit. Bioinformatics analyses of published single-cell RNA sequences A list of 114 genes associated with motile cilia was assembled on the basis of literature review (91–93). Two single-cell transcriptomic datasets were used for the analysis. The expression and annotation data (cluster and meta-cell identifiers) of whole embryos at E6.5 to E8.25 (94) were downloaded from Gene Expression Omnibus (GEO; GSE169210) according to https://github.com/tanaylab/embflow. The analysis was conducted at the meta-cell level. The expression and annotation data of microdissected heart regions (39) were accessed from https://marionilab.cruk.cam.ac.uk/heartAtlas/ and analyzed at the cluster level, focusing on cardiac clusters Me2 to Me7. Each dataset was normalized independently using voom transformation from limma R package (95), centered by gene using the mean of background genes defined as genes not in the motile cilia gene list, and scaled by gene, and the two datasets were merged. Motile cilia genes were then ranked by total median expression across meta-cells and cells, and k-means clustering was used to identify three distinct clusters. Expression levels were plotted by boxplots, showing the median, 25th (Q1) and 75th quantiles (Q3), and whiskers corresponding to Q1 − 1.5(Q3-Q1) and Q3 + 1.5(Q3-Q1). The number of meta-cells used is indicated in table S7. Bioinformatic analyses of paired bulk RNA sequences After demultiplexing, reads were trimmed to remove adapter sequences from the first amplification step. Sequence quality was assessed using FastQC and Picard metrics. Duplicates were removed. FASTQ files were mapped to the ENSEMBL Mouse GRCm39 reference genome using the Illumina STAR aligner and counted by featureCounts from the Subread package (v2.0.6). The resulting sequencing depth was 44 million reads in average. Genes with low expression levels were filtered out, using the filterByExpr function of edgeR R package (96) with parameters min.count = 5, min.total.count = 10, large.n = 50, and min.prop = 0.2. A principal components analysis was conducted on variance stabilizing transformation (VST)–transformed data, and, on the 500 most variant genes to detect potential outliers, no outlier was detected. All differential analysis were carried out using the DESeq2 R package (97), with local fit type for dispersion estimation and ashr type (10.32614/CRAN.package.ashr) to shrink the log2 fold changes. To infer embryo clusters, an initial differential analysis was performed on wild-type samples, considering both sexes together and males independently, to identify genes associated with asymmetry in wild type samples. The DESeq2 model was adjusted for the effect of the side (left/right) and embryo identifier to account for pairing. The resulting lists of differentially expressed genes (DEGs) were merged and filtered to remove genes not expressed in myocardial lineages [using published single-cell transcriptomics (39), we kept genes that belong to clusters Me3-Me7 and removed genes that exclusively belong to Me1-Me2, Ec1-Ec2, and En1-En2]. This list of markers was refined by retaining “congruent” genes, defined as those exhibiting the same left-right trend across all samples. This refinement resulted in 25 genes, which were used to perform hierarchical clustering of all samples of the dataset, based on Euclidean distance and the ward.D2 aggregation criterion. This retrieved three laterality clusters. A second differential analysis was then conducted to assess (i) asymmetry variations within clusters, using a DESeq2 model adjusted for the effect of the side and embryo identifier, and (ii) to evaluate expression changes within each side, using a DESeq2 model adjusted for the cluster and sex. The goodness of fit of these DEseq2 models was assessed by checking the distribution of the raw P values, ensuring, in particular, that the variability of embryos within clusters is lower than the variability between clusters. DEGs were filtered to exclude genes not expressed in myocardial lineages. For (i), genes were classified as gained asymmetry compared to the situs solitus cluster, when they had a significant right to left expression in a mutant laterality cluster, an absolute log2 fold change higher than 0.5 (for heterotaxy) or 1 (for situs inversus totalis), at least 200 reads in either side, whereas they had a nonsignificant right to left expression in situs solitus, with an absolute log2 fold change lower than 0.1. Genes were classified as conserved or reversed asymmetry compared to the situs solitus cluster, when they had a significant right to left expression in a mutant laterality cluster, an absolute log2 fold change higher than 0.5, and at least 200 reads in either side. Genes were classified as lost asymmetry compared to the situs solitus cluster, when they had a nonsignificant right to left expression in a mutant laterality cluster with an absolute log2 fold change lower than 0.1, whereas they had a significant right to left expression in situs solitus, with an absolute log2 fold change higher than 0.5 and at least 200 reads in either side. We selected as signature genes, the genes that were common and changed both between situs solitus and heterotaxy and between situs solitus and situs inversus totalis. For a global analysis, we compiled (i) and (ii), corresponding to 1842 genes exhibiting diverse variation patterns (Fig. 7D1), filtered on the basis of significance but not fold change or number of reads. The list includes genes that are symmetric in situs solitus but with altered overall expression levels in abnormal laterality clusters; genes that are asymmetric in situs solitus and show changes in overall expression levels in abnormal laterality clusters; and genes exhibiting an increase, decrease, loss, or gain of asymmetry in abnormal laterality clusters. Overrepresentation analysis was performed on these genes using the enricher function from the clusterProfiler R package (98). The analysis was applied to several genes set collections, including the full Kyoto Encyclopedia of Genes and Genomes (KEGG) collection and its subset filtered for “signaling pathways,” the full collection of GO biological process terms and its subset filtered for signaling pathways, and additional collections from Biocarta, Reactome, and WikiPathways. Heatmaps were generated using the VST-transformed count differences between the right and left side, with gene scaling to an SD of 1. Hierarchical clustering of genes was conducted with Euclidean distance and the Ward.D2 aggregation criterion. For Fig. 7B, a global clustering tree of all samples was performed. In other heatmaps, the order of laterality clusters was preserved, with sample clustering performed independently within each laterality cluster. In the analysis of Pitx2 isoforms, Kallisto (version 0.51.1) was used to quantify transcript abundances and identify expressed isoforms. The reference genome used for this analysis was Mus musculus mm10 (GRCm39, release 104). Analyses were conducted using the default parameters of Kallisto. Extraction of Ccdc40 sequence in bulk RNA sequences To identify variants in the specific Ccdc40 genomic region, we performed variant calling using SAMtools version 1.18. The command bcftools mpileup was applied on sequence alignment data stored in BAM format. These BAM files were obtained by cleaning fastq files of adapter sequences and low-quality sequences using cutadapt. Only sequences at least 25 nucleotides in length were considered for further analysis (options -O 6 –trim-n –max-n 1). STAR, with options (–outFilterMismatchNoverLmax 0.05 –outSAMunmapped Within –sjdbOverhang 250 and 2-pass mode by-sample), was used for alignment on the reference genome and get BAM files. The reference genome used for this analysis was the M. musculus mm10 (GRCm39, release 104). The analysis was restricted to chromosome 11, focusing on the genomic position range 119,141,000 to 119,143,000. Variant calling was then performed using bcftools call -mv, which applies a multiallelic variant detection model. This produced a Variant Call Format (VCF) file containing the identified variants in the specified region for subsequent analysis: No genetic variation was found in any wild-type samples, whereas the A genetic variation was detected in 57 mutant samples (fig. S6B). Genetic variation from five mutant samples could not be retrieved due to the absence of reads mapping to the position of interest, but, because the samples are paired, we were able to retrieve the genetic variation for each mutant embryo as mpileup detected variant for at least left or right sample. Clustering of E9.5 phenotypes The dataset was divided into two cohorts (table S3): one consisting of 28 embryos with complete data (Ccdc40lnks/lnks; Tg Nodal-ASE-lacZ mutants and littermate transgenic controls) and another of 67 embryos, including additional 39 samples with incomplete data because they were nontransgenic and, thus, the Nodal transgenic signal was missing. In a PLS analysis, we identified two blocks of variables: one comprising 25 geometrical parameters, computed from measures in 3D segmentations of the heart tube, and the other with five Nodal signal markers (excluding markers that were not variable or the OFT signal that was redundant with the looping direction). Given the mixed nature of the geometrical variables and the dataset structure, we carried out a multiple factorial analysis (MFA) as an appropriate multivariate method. The geometrical data were organized into five blocks of quantitative variables: coordinates (3D coordinates of eight reference points), volumes (six cardiac segments), angles (seven between three consecutive reference points, and the orientation of the left and right ventricles relative to the notochord), tube length, and absolute value of AVC lateral displacement; and nine blocks of qualitative variables, including looping direction, OFT shape, alignment of AVC and atria, and Nodal signal in different domains (DPW signal, AVC signal, signal in the atrium facing the right ventricle, posterior LPM signal, and CPP signal). To account for dependencies between the x, y, and z coordinates in the “coordinate” block, the data were summarized using six factors from the DISTATIS method, using the DistatisR R package (10.32614/CRAN.package.DistatisR). MFA for the 67 samples was performed on the geometrical variables, whereas a multiple correspondence analysis (MCA) for the 28 samples was applied to the Nodal signal variables due to their qualitative nature. Both analyses were performed using the FactoMineR R package (99). A PLS regression was then conducted to model the relationships between the geometrical and Nodal signal markers, enabling us to infer the MCA coordinates for the incomplete cohort of 39 samples. For both the MFA and MCA, we retained enough components to capture 80% of the variability, resulting in seven and five components, respectively. This methodology led to a 3D projection of all 67 samples, with contributions from all variables, including those with missing data, by leveraging the relationships between the geometrical and Nodal signal markers. This projection, representing the relations between the two sets of markers, is referred to as the “PLS-space.” PLS regression was performed using RGCAA R package (10.32614/CRAN.package.RGCCA). Hierarchical clustering was applied to the first two PLS components. Marker associations for each cluster were identified through pairwise statistical testing between clusters (e.g., heterotaxy with rightward looping versus situs solitus and heterotaxy with inverted looping versus situs inversus totalis) for all variables. For quantitative variables, the Mann-Whitney-Wilcoxon test was used, whereas qualitative variables were assessed with Fisher’s exact test. Effect sizes were also calculated: Eta-squared for quantitative variables and Cramer’s V for qualitative variables, both measures implemented in the effectsize R package (100). Statistical analyses Statistical tests and P values are described in the figure legends and table S7. P values less than 0.05 were considered statistically significant. Group allocation was based on PCR genotyping and heart looping direction. All sample numbers (n) indicated in the text refer to biological replicates, i.e., different embryos/fetuses or different cells. Investigators were blinded to allocation during imaging and phenotypic analysis, but not during quantifications. To compare whether two groups had equivalent means, a t test was used when a normal distribution could be assumed, and a Mann-Whitney test otherwise. To test whether the mean of a population was statistically different from a theoretical value, a one-sample t test was used. To test whether the distribution of classes followed an expected distribution (e.g., Mendelian distribution or random distribution), a chi-square test was used. Additional resources Web interface: Paired transcriptomic analysis of Ccdc40 mouse mutants (https://ccdc40.pasteur.cloud).