Fig 1.
morphoHeart image segmentation and volumetric tissue reconstructions.
(A) Schematic overview of the morphoHeart image processing and segmentation pipeline. (B–I) Generation of tissue contour libraries. Identification of outer (orange) and inner (blue) contours of the myocardium and endocardium in single z-slices (B, C). For each slice masks are generated representing filled outer (D, E) and inner (F, G) tissue contours. The exclusive disjunction operation of inner and outer contours per tissue results in a tissue mask for each slice (H, I). (J) Contour library generated by morphoHeart. (K) 3D mesh reconstructions of external (yellow) and internal (orange) myocardium, external (blue) and internal (pink) endocardium, and myocardial (green) and endocardial (magenta) tissue layers.
Fig 2.
The zebrafish heart grows and compacts during early cardiac morphogenesis.
(A) Flow diagram describing the phases involved in the process of acquiring comprehensive 3D morphometric data using morphoHeart. (B–D) Schematic depicting the early stages of heart development analysed, including looping of the early tube (B, 34–36 hpf), looping and ballooning (C, 48–60 hpf), and the looped heart (D, 72–74 hpf). (E–H) Reconstructions of myocardial and endocardial meshes during heart development. (I–L) Analysis of heart looping. Linear heart length (green line) and heart centreline or looped heart length (blue line) are extracted and measured (I). As the heart develops, the poles move closer together (J). During looping morphogenesis, the centreline’s looped distance elongates between 34 and 50 hpf, and subsequently shortens (K). Looping ratio also increases between 34 and 50 hpf, but then remains constant (L). (M) Cardiac chambers can be separated via placement of a user-defined disc. (N, O) Quantification of total heart volume reveals the heart increases in volume between 34 and 50 hpf and compacts again by 72 and 74 hpf (N). Lumen volume increases with heart volume and is then maintained (O). (P, Q) Analysis of chamber volume reveals while both chambers grow between 34 and 50 hpf, ventricle volume is maintained while the atrium shrinks (P). Lumen size in both chambers is maintained post-48 hpf (Q). One-way ANOVA with multiple comparisons.* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, ns = not significant; 34–36 hpf: n = 9; 48–50 hpf: n = 10; 58–60 hpf: n = 8; 72–74 hpf: n = 10. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 3.
Visualisation and quantification of chamber deformation reveals chamber-specific differences in growth.
(A–C) Ellipsoids are fitted to chambers to quantify chamber geometry (A). The atrium becomes more spherical (asphericity tends to 0) during development (B), while ventricle asphericity remains stable (C). One-way ANOVA with multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001, **** P < 0.0001. (D–F) Myocardial expansion/deformation can be quantified by measuring the distance between the myocardial centreline and the internal or external myocardial mesh (D). This value can then be mapped onto the internal or external myocardial mesh using a heatmap to visualise 3D cardiac ballooning (E). D and E depict this process using the internal myocardial mesh. 3D heatmaps can be unrolled into a standard 2D geometry for aggregation and comparison (F). (G–J) Visualisation of 3D myocardial ballooning heatmaps mapped to the external myocardial mesh identifies substantial deformation of the atrial outer curvature at 34–36 hpf (G). By 48–50 hpf, this outer curvature deformation is enhanced, and the atrium is more ballooned than the ventricle (H). The ventricular apex can be seen emerging (H–J). (K–N) Unrolled 2D ballooning heatmaps allows averaging of multiple hearts to identify conserved regions of deformation. By 74 hpf deformation of the atrium has become more uniform (N). Labels around the 2D heatmaps indicate cardiac region: D—dorsal, V—ventral, L—left, R—right, AOC—atrial outer curvature, AIC—atrial inner curvature, VOC—ventricular outer curvature, VIC—ventricular inner curvature, AV Canal—atrioventricular canal; 34–36 hpf: n = 9; 48–50 hpf: n = 10; 58–60 hpf: n = 8; 72–74 hpf: n = 10. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 4.
The atrium and ventricle exhibit different tissue dynamics during morphogenesis.
(A–C) Quantification of myocardial tissue volume from myocardial meshes (A). Total myocardial volume increases during looping and reduces at early stages of maturation (B). Chamber-specific analysis reveals a later reduction in atrial myocardium volume compared with an earlier increase and maintenance in ventricular myocardial volume (C). (D, E) Quantification of endocardial tissue volume from endocardial meshes. Total endocardial volume decreases after 58 hpf (D), driven by a reduction in endocardial tissue in both the atrium and ventricle (E); 34–36 hpf: n = 9; 48–50 hpf: n = 10; 58–60 hpf: n = 8; 72–74 hpf: n = 10. (F–H) Quantification of cardiomyocyte number, from live lightsheet z-stack images of Tg(myl7:BFP-CAAX);Tg(myl7:H2B-mScarlet) (F). The total number of cardiomyocytes increases between 48 and 60 hpf (G). Atrial cardiomyocyte number remains mostly constant, while ventricular cardiomyocyte number increases (H). One-way ANOVA with multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, ns = not significant; 34–36 hpf: n = 8; 48–50 hpf: n = 10; 58–60 hpf: n = 7; 72–74 hpf: n = 10. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 5.
Cardiac chambers undergo regionalised reduction in cell size.
(A–D) Quantification of internuclear cardiomyocyte distance as a proxy for cell size reveals an early increase in atrial cardiomyocyte size and a later reduction in ventricular cardiomyocyte size. Each chamber is subdivided into regions (distinguished by different colours) for more granular analysis (B–D). Growth and decrease in atrial cardiomyocyte size occurs predominantly in ventral and outer curvatures (C). Ventricular dorsal cardiomyocytes expand early, and all ventricular cardiomyocytes apart from those on the inner curvature subsequently decrease in size (D). Each dot represents the average internuclear distance, per region, in one heart. One-way ANOVA with multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001; 34–36 hpf: n = 10; 48–50 hpf: n = 12; 58–60 hpf: n = 10; 72–74 hpf: n = 10. (E–N) Myocardial wall thickness is quantified by measuring the distance between the inner and outer myocardial meshes (E) and mapped onto the outer myocardial mesh using a heatmap to visualise myocardial thickness in 3D (F). 3D myocardial thickness heatmaps (G–J) are unrolled to 2D, and average 2D heatmaps generated for each time point (K–N), illustrating that the atrial wall is consistently thinner than the ventricular, and that both chamber walls thin as development progresses. Labels around the outside indicate cardiac region: D—dorsal, V—ventral, L—left, R—right, AOC—atrial outer curvature, AIC—atrial inner curvature, VOC—ventricular outer curvature, VIC—ventricular inner curvature, AV Canal—atrioventricular canal. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 6.
The ECM undergoes chamber-specific regionalised expansion and reduction during heart morphogenesis.
(A) Schematic depicting the approach used to generate cardiac ECM meshes, by combining the filled external endocardial contour and the filled internal myocardial contour using AND and XOR logical operations. (B–D) Volumetric 3D reconstructions of the cardiac ECM during heart development (B), showing an apparent reduction in the ventricular ECM at 72–74 hpf. Quantification of total cardiac ECM volume reveals a significant increase in ECM volume between 34 hpf and 50 hpf, followed by a reduction between 58 hpf and 74 hpf (C). The majority of cardiac ECM is found in the atrium, and while both chambers expand their ECM during looping, ventricular ECM reduces first between 48 hpf and 60 hpf, while atrial ECM is reduced only after 58 hpf (D). (E–G) ECM thickness is quantified by measuring the distance between the outer endocardial mesh and inner myocardial mesh (E) and mapped onto the inner myocardial mesh (F) using a heatmap to visualise ECM thickness in 3D (G). (H–O) 3D heatmaps reveals the cardiac ECM is thicker in specific regions of the heart (H–K). Unrolled and averaged 2D ECM thickness heatmaps reveals the ECM is thicker in the atrium than the ventricle and in particular in the outer curvature of the atrium at 34–60 hpf (L–N). The atrial ECM is still regionalised at 72–74 hpf but the thickening is repositioned to the dorsal face of the atrium (O). Labels around the outside indicate cardiac region: D—dorsal, V—ventral, L—left, R—right, AOC—atrial outer curvature, AIC—atrial inner curvature, VOC—ventricular outer curvature, VIC—ventricular inner curvature, AV Canal—atrioventricular canal. (P) Schematic illustrating the cutting of the ECM mesh into left and right regions for both the atrium and ventricle. (Q, R) Quantification of ECM volume in outer and inner curvatures of the atrium (Q) and ventricle (R) reveal the regionalised dynamics that drive cardiac ECM expansion and reduction. One-way ANOVA with multiple comparisons. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, ns = not significant; 34–36 hpf: n = 9; 48–50 hpf: n = 10; 58–60 hpf: n = 8; 72–74 hpf: n = 10. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 7.
hapln1a mutants exhibit defects in atrial growth and ECM expansion.
(A, B) Myocardial (green) and endocardial (magenta) 3D reconstructions of wild-type sibling (A) and hapln1a mutant hearts (B) at 34–36 hpf, 48–50 hpf, and 72–74 hpf. (C–E) Quantification of heart size reveals that hapln1a mutant hearts (pink) are smaller at 48–50 hpf than wild-type siblings (blue, C), largely due to a failure of the atrium to balloon by 48–50 hpf (D). (F, G) Averaged 2D chamber ballooning heatmaps of wild type (F, n = 5) and hapln1a mutant embryos (G, n = 5) at 48–50 hpf demonstrates that hapln1a mutant atria fail to expand. (H–L) Quantification of lumen size and myocardial tissue volume shows that hapln1a mutants fail to expand the atrial lumen at 48–50 hpf (H), whereas the expansion dynamics of the ventricular lumen is unaffected (I). (M, N) Analysis of regional ECM volume in wild-type siblings and hapln1a mutants. ECM volume does not expand in either the atrium or ventricle of hapn1a mutants at 48–50 hpf compared to wild-type siblings (M, N). Asterisks indicate significant difference between time points for each genotype (blue indicates significance in siblings, pink indicates significance in hapln1a mutants). Grey boxes indicate significant difference between wild-type siblings and mutants at the indicated time point. Two-way ANOVA with multiple comparisons. * p < 0.05, ** p 0.01, *** p < 0.001, **** p < 0.0001, ns = not significant. Wild type 34–36 hpf: n = 6; all other genotypes and stages: n = 5. Labels around the heatmap indicate cardiac region: D—dorsal, V—ventral, L—left, R—right, AOC—atrial outer curvature, AIC—atrial inner curvature, VOC—ventricular outer curvature, VIC—ventricular inner curvature, AV Canal—atrioventricular canal. Plots display median and quartiles. The numerical data underlying this figure can be found in S1 Data.
Fig 8.
(A–C) Snapshots of morphoHeart”s GUI, showing the welcome window (A), segmentation tab (B), and the morphometric analysis tab (C).
Table 1.
Zebrafish strains and study resources.
Zebrafish strains, software resources, and general chemical reagents used in this study.