Transient cardiomyocyte fusion regulates cardiac development in zebrafish

Abstract

Cells can sacrifice their individuality by fusing, but the prevalence and significance of this process are poorly understood. To approach these questions, here we generate transgenic reporter lines in zebrafish to label and specifically ablate fused cells. In addition to skeletal muscle cells, the reporters label cardiomyocytes starting at an early developmental stage. Genetic mosaics generated by cell transplantation show cardiomyocytes expressing both donor- and host-derived transgenes, confirming the occurrence of fusion in larval hearts. These fusion events are transient and do not generate multinucleated cardiomyocytes. Functionally, cardiomyocyte fusion correlates with their mitotic activity during development as well as during regeneration in adult animals. By analyzing the cell fusion-compromised jam3b mutants, we propose a role for membrane fusion in cardiomyocyte proliferation and cardiac function. Together, our findings uncover the previously unrecognized process of transient cardiomyocyte fusion and identify its potential role in cardiac development and function.

Fig. 1

Establishment of the FATC transgenic line for in vivo labeling of fusion-derived cells. a Schematic illustration of the ubb:FATC construct and Cre recombination products. b, c Schematic illustrations of membrane fusion (b) and multiple transgene insertions (c) that could give rise to LIFEACT-GFP expression. b Cre recombination of Tg(ubb:FATC) animals harboring a single copy of the transgene generates two different cell populations, one carrying a GAL4 expression cassette and the other carrying a UAS:LIFEACT-GFP (LAGFP) cassette. Fusion between cells carrying different cassettes (green arrows), but not between cells carrying the same cassette (black arrows), leads to activation of LIFEACT-GFP expression (green). c Stochastic Cre recombination in cells harboring two copies of the FATC transgene will generate 50% of the time cells carrying both the GAL4 and UAS:LIFEACT-GFP cassettes (green). d, e LIFEACT-GFP expression is activated in fast twitch, but excluded from slow twitch, muscles. d Drawing of a zebrafish larva depicting the trunk area (red box), shown in e. e Tg(ubb:FATC);Tg(hsp:cre) embryos were heat-shocked at 24 hpf, and LIFEACT-GFP expression (green) was assessed at 5 dpf. Muscle fibers were visualized by immunostaining with EB165 and F59 (red) to label fast and slow twitch muscles, respectively. DAPI (blue) labeling shows fusion-derived multinucleation of fast twitch muscles. f heat induction of cre expression at 24 hpf activates LIFEACT-GFP (green) expression in skeletal and cardiac myocytes in 5 dpf Tg(ubb:FATC);Tg(hsp:cre) larvae, but not in Tg(ubb:FATC) animals (blue). g LIFEACT-GFP expression (green) labels sarcomeric structures of FATC-activated cardiomyocytes (identifiable by membrane expression of mKATE-CAAX (red)). Absence of LIFEACT-GFP expression without 4-OHT treatment confirms the dependency of the FATC reporter on Cre activity. e, f show maximum intensity projections of 10–50 μm thick confocal stacks. g 3D volume renderings of 90 μm thick confocal stacks of the entire cardiac ventricle. Representative images from a total of 8–9 (e), 32 (f), and 9 (g) animals are shown. Scale bars: 20 μm (e, f trunk, g), 50 μm (f heart)

Fig. 2

Cardiomyocyte proliferation does not mediate FATC reporter expression. a Schematic illustration of the ubb:NATC construct and Cre recombination products. b Schematic illustration of the cellular events that could occur in the experiment shown in c. c NTR-mCherry⁺ (red) cardiomyocytes (H2B-GFP⁺, green) were observed in Tg(ubb:NATC);Tg(myl7:creER);Tg(myl7:H2B-GFP) larvae following MTZ-mediated cell ablation, 7 days after transient CreER activation by 4-OHT treatment at 24 hpf. All images show the same heart. d NTR-mCherry expression does not succeed cardiomyocyte DNA synthesis. The 6 h EdU pulse-labeled cells undergoing DNA synthesis (green) in 54 hpf Tg(ubb:NATC);Tg(myl7:creER) embryos. Two hours prior to EdU removal, Cre activity was induced by an 18 h 4-OHT treatment. mCerulean (blue) and NTR-mCherry (red) expression was visualized by immunostaining. No EdU⁺NTR-mCherry⁺ cardiomyocytes were detectable in 20 hearts imaged. c, d Maximum intensity projections of 40–60 μm thick confocal stacks. Representative images from a total of 12 (c) and 20 (d) animals are shown. Scale bars: 10 μm (c), 20 μm (d)

Fig. 3

Membrane fusion occurs in the developing heart. Live imaging of Tg(myl7:MKATE-CAAX);Tg(myl7:H2B-GFP) embryos, in which heartbeats were blocked by morpholino-mediated tnnt2 knock-down (Supplementary Movie 1), showing establishment of a new membrane border (arrows) between cardiomyocytes 1 and 2, which initially exhibited cytoplasmic continuum. The plasma membrane border between cardiomyocytes 1 and 3 (arrowheads) dissolved and was subsequently re-established. mKATE-CAAX (red) and H2B-GFP (green) expression labeled cardiomyocyte membranes and nuclei, respectively. Timing of each still image is hour:minute. All images are 3D volume renderings of 60 μm thick confocal stacks of a representative heart showing the myocardial monolayer observed from the lumen. Examples of cytoplasmic continuum between cardiomyocytes were observed in all 6 hearts examined. Scale bar: 20 μm

Fig. 4

Blastula transplantations reveal cardiomyocytes expressing both donor and host transgenes. a Schematic drawing of transplantation experiment shown in b–d. Tg(myl7:EGFP) cells were transplanted into Tg(myl7:nDsred2) hosts at the blastula stage. b–d Cardiomyocytes expressing both host-derived myl7:EGFP (green cytoplasm) and donor-derived myl7:nDsred2 (red nucleus) transgenes (b) are evident from orthogonal sections (c) and Y axis rotation of a 3D volume rendering (d). EGFP⁺nDsred2⁺ cardiomyocytes (white circle) were detected in 9 out of 18 mosaic hearts at 3 dpf. e Schematic drawing of transplantation experiment shown in f–h. Tg(actb2:loxP-mCherry-loxP-EGFP) cells were transplanted into Tg(myl7:creER) hosts at the blastula stage. Transplanted animals were treated with 4-OHT starting at 24 hpf. New 4-OHT was added to the embryo medium daily. f A single focal plane of a confocal stack shows a dorsal view of a 3 dpf chimeric heart, anterior up. Donor-derived mCherry⁺ cells (red) and fusion-derived EGFP⁺ cardiomyocytes (green, arrows) were detected by live imaging in 5 out of 13 mosaic hearts (mixed genotypes, creER⁺ and creER⁻). g Maximum intensity projection of a 50 μm confocal stack shows a lateral view of a 7 dpf chimeric heart, anterior up, dorsal to the left. Tg(actb2:loxP-mCherry-loxP-EGFP) donor cells are shown in red. Host-derived Cre-mediated recombination of the donor transgene was detectable by EGFP immunofluorescence (green) in 11 out of 39 mosaic hearts (mixed genotypes, creER⁺, and creER⁻). h Magnified image showing area outlined by red box in g. Scale bars: 20 μm

Fig. 5

F/NATC-labeled cardiomyocytes are highly proliferative. a–c FATC-activated (LIFEACT-GFP⁺, green) cardiomyocytes (nDsRed⁺, red) in 3 and 5 dpf (a) and 6 mpf (b) Tg(ubb:FATC);Tg(myl7:creER);Tg(myl7:nDsRed2) fish treated with 4-OHT from 24 to 40 hpf were quantified as percentages of total ventricular cardiomyocytes (c). The same fish were analyzed at 3 and 5 dpf (a, c). d, e NTR-mCherry⁺ (red) cardiomyocytes (H2B-GFP⁺, green) of 3 and 5 dpf Tg(ubb:NATC);Tg(myl7:creER);Tg(myl7:H2B-GFP) fish treated with 4-OHT starting at 24 hpf (d) were quantified as percentage of total ventricular cardiomyocytes (e). f–h NTR-mCherry⁺ cardiomyocytes contribute substantially to the proliferating subset of cardiomyocytes. Tg(ubb:NATC);Tg(myl7:creER) embryos were treated with 4-OHT starting at 48 (f) or 24 (g) hpf to identify NATC-activated cardiomyocytes (mCerulean⁺,blue, and mCherry⁺, red). A 6 h EdU pulse (green) starting at 72 (f) or 120 (g) hpf labeled cells undergoing DNA synthesis. Arrowheads point to mCerulean⁺mCherry⁺EdU⁺ cardiomyocytes, which were quantified as percentages of total EdU⁺ ventricular cardiomyocytes (h). i Percentages of proliferating cardiomyocytes, assessed by a 6 h pulse of EdU, relative to total cardiomyocytes and relative to the NTR-mCherry⁺ cardiomyocyte population in 5 dpf Tg(myl7:nuDsRed2) and Tg(myl7:creER);Tg(ubb:NATC) ventricles, respectively. Three-dimensional volume renderings (a, d) and maximum or average intensity projections (b, f, g) of 90 (a, b), 84 (d), and 10–14 (f, g) μm thick confocal stacks are shown. In c, e, h and i, bars and error bars represent means ± S.E.M. Each circle (n = 12 in c, 20 in e, 10 in h, and 9 in i), triangle (n = 12 in c, 20 in e, 19 in h, and 19 in i), and square (n = 3) represents a cardiac ventricle. **p ≤ 0.01, ***p ≤ 0.001 (two-tailed student’s t-test). Representative images from a total of 12 (a), 3 (b), 18 (d), 11 (f), and 19 (g) animals are shown. Scale bars: 20 μm (a, d, f, g), 50 μm (b)

Fig. 6

Transgene coupling-mediated fluorescence expression in the adult heart increases after injury. a–c Cardiac injury induces NATC reporter expression. 6 mpf Tg(ubb:NATC);Tg(myl7:creER) fish were injected with 4-OHT intraperitoneally 30 days prior to cryoinjury. a Whole-mount immunostaining revealed that NTR-mCherry⁺ cardiomyocytes (red) were localized mainly adjacent to the damaged area (dashed line), identified by disorganized cardiac cells (DAPI⁺, blue). b Cytoplasmic NTR-mCherry (red) and nuclear DAPI (blue) labeling shows sham- and injury-induced NATC-activated cardiomyocyte morphologies; arrowheads point to double-labeled cardiomyocytes. c Ratios of mean gray values of NATC and DAPI labeling were calculated from maximum intensity projections of 200 μm thick confocal images of individual hearts. d Many NTR-mCherry⁺ cardiomyocytes (red) detected mainly at the lesion border zone (disorganized cardiac cells are revealed by DAPI staining, blue) were phospho-histone H3 (pH3, green) positive. e, f The majority of mitotic cardiomyocytes at the lesion border zone show NATC reporter expression. e NTR-mCherry (red) and myosin heavy chain (MHC, green) expression were detected by immunostaining of 6 mpf sham-operated and cryoinjured Tg(ubb:NATC); Tg(myl7:creER) hearts, in which Cre had been activated by 4-OHT treatment from 24 to 40 hpf. NTR-mCherry⁺ cardiomyocytes undergoing mitosis (white circles) were detected with an 8 h EdU pulse (EdU⁺ nuclei are in white) and their numbers were quantified as percentages of total numbers of mitotic cardiomyocytes (f). Cardiac cells were visualized by DAPI (blue) staining. Representative Tg(ubb:NATC);Tg(myl7:creER) hearts from four sham and six injured hearts (a–d), and three sham and three injured hearts (e, f) are shown/analyzed. Images are maximum or average intensity projections (a, e) and 3D volume renderings (d, e) of 200–300 (a), 20 (b, e), and 300 (d) μm thick confocal stacks. Bars and error bars in c, f represent mean ± S.E.M. Each circle (n = 4 in c and 3 in f) and triangle (n = 6 in c and 3 in f) represents a heart. *p ≤ 0.05, **p ≤ 0.01 (two-tailed student’s t-test). Scale bars: 5 μm (b), 20 μm (e), 30 μm (d), 50 μm (a)

Fig. 7

Cell fusion deficiency negatively affects cardiomyocyte proliferation and cardiac function. a–d jam3b mutants show a reduction in the number of LIFEACT-GFP⁺ skeletal and cardiac myocytes. Embryos from a cross of jam3b ^+/− ;Tg(ubb:FATC) and jam3b ⁺⁻ ;Tg(hsp:cre) fish were heat-shocked at 24 hpf. LIFEACT-GFP⁺ (green) skeletal (a) and cardiac (b) muscles were detectable at 48–52 hpf. DAPI-stained nuclei are shown in blue. c, d LIFEACT-GFP⁺ skeletal myocytes per 10,000 μm² of trunk surface area (c) and LIFEACT-GFP⁺ ventricular cardiomyocytes (d) in jam3b ^−/− embryos and jam3b ^+/? siblings. e–g Jam3b deficiency reduces cardiomyocyte proliferation. A 16 h EdU pulse (green) labeled mitotic cardiomyocytes (nDsRed⁺, red) (e) shown as percentage of total cardiomyocytes (f); total cardiomyocyte numbers (g) in ventricles of 6 dpf jam3b ^−/− ;Tg(myl7:nDsRed2) and jam3b ^+/? ;Tg(myl7:nDsRed2) siblings. h–j jam3b deficiency impairs cardiac function. h The 5 dpf jam3b mutants exhibit pericardial edema (28 out of 36 fish with edema were jam3b^−/−). i, j The 5 dpf jam3b mutants also display decreased fractional shortening (i) and blood flow velocity (j) compared to control siblings, as evidenced by a significant difference in average maximum flow velocity (k). a, b, e are maximum or average intensity projections of 20–65 μm thick confocal stacks. h are single focal planes of confocal images showing larvae in lateral views, anterior up and dorsal to the left. In all plots, bars and error bars represent means ± S.E.M. Each circle and triangle represents an embryo (c) or a heart (d, f, g, k, j). *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 (two-tailed student’s t-test); 10–12 (a–d), 16 (e–g), 36 (h), and 11–25 (i–k) −/− and +/? sibling animals were examined. Scale bars: 20 μm (a, b, e), 40 μm (h)

Fig. 8

Membrane fusion does not generate multinuclear or polyploid cardiomyocytes. a Mononucleated NATC-activated cardiomyocytes undergo mitosis. Tg(ubb:NATC);Tg(myl7:creER) embryos were treated with 4-OHT from 48 to 56 hpf. At 72 hpf, the larvae were given a 6 h EdU pulse. Membrane Cerulean (green) and cytoplasmic mCherry (red) expression were visualized by immunostaining. EdU (magenta) and DAPI (blue) staining shows a single nucleus in the EdU + mCherry⁺ cardiomyocyte (arrowhead). 7.2 μm thick confocal stacks are shown as maximum intensity projections. b Binucleation is rarely observed in embryonic cardiomyocytes. Images are 3D volume renderings of a 54 hpf heart. Cardiomyocyte membrane and nuclei were visualized by Tg(myl7:GAL4);Tg(UAS:EGFP-CAAX) (green) and Tg(myl7:nDsRed2) (red) expression, respectively. Y axis rotation of the ventricular area (small panels on the right), indicated by the white box shows the single-layered compact myocardium with a binuclear cardiomyocyte, outlined with white dashed lines. c Among the EdU⁻ populations, the distributions of DAPI intensity in NTR-mCherry⁺ cardiomyocytes (green bars) and NTR-mCherry⁻ cardiac cells (black bars) were similar, indicating that DNA content in fused cardiomyocytes is not different from that of non-fused cardiac cells. On the contrary, and as expected, DNA content in EdU⁺ cells (red bars) was significantly higher than that in EdU⁻ cells (either NTR-mCherry⁺ or NTR-mCherry⁻). Tg(ubb:NATC);Tg(myl7:creER) embryos were treated with 4-OHT from 24 to 40 hpf. NTR-mCherry expression and proliferating cells were labeled by immunostaining at 7 dpf after a 16 h EdU pulse. DAPI intensity measurements were performed on 3D volume-renderings obtained from confocal images. n = 311 NTR-mCherry⁻EdU⁻ cells, n = 227 NTR-mCherry⁺EdU⁻ cells, n = 173 EdU⁺ cells, from 5 larvae. ***p < 0.001, n.s., not significant, two-sample Kolmogorov–Smirnov test. Representative images from 11 larvae (a) and 10 embryos (b) are shown as average intensity projections (a) and 3D surface renderings (b) of 12 (a) and 60 (b) μm thick confocal stacks. Scale bars: 20 μm (a, b small panels), 30 μm (b, large panel on the left)