(A) Despite different exon numbering i.e. 6a/6b in human and 7/8 in zebrafish orthologues, exon structure is entirely conserved. (B) jnk1a, jnk1b and JNK1 lie in syntenic chromosomal regions. (C) Cartoon showing how the 8 jnk1a/b transcripts result from alternate exon 7/8 usage and differing C-terminal extension. (D) Translation of zebrafish jnk1a and jnk1b transcripts showing amino acids from alternatively spliced exons. Black text indicates identical amino acids, green text indicates favourable amino acid substitutions and red text divergent amino acid residues. Both jnk1a and jnk1b genes are capable of producing Ex7 and short C-terminus containing transcripts that fully match human JNK1 transcripts. However, whilst Ex8 derived from jnk1b matches the human peptide, the Ex8 from jnk1a is divergent and contains a serine rather than threonine residue (*). In contrast whilst the long C-terminal extension provided by jnk1a matches the human, the jnk1b long terminal extension is highly divergent and differs by 9/39 amino acids including insertion of an additional threonine residue. See S1 Fig for full peptide sequences.

(A) Schematic of assay for jnk1a (see methods for full details). RNA is extracted and a cDNA pool produced. PCR within linear phase performed using forward (fwd) primers specific for jnk1a/1b exons 7/8 and common reverse (rev) primers. The fraction with C-terminal extension is determined by restriction enzyme (RE) digestion. (B) Percentage of total jnk1a and jnk1b transcripts during embryonic development. 14 ss = 16 hpf. (C) Percentage of total Lg jnk1a and jnk1b transcripts during embryonic development. (D and D') Proportion of each individual alternatively spliced transcript during development in whole embryo. (E and E') Proportion of each individual transcript within isolated whole heart during development. jnk1a = blue, jnk1b = green, Ex7 = circle, Ex8 = square, Sh = outline black and Lg = no outline. n = 6–8 at all time points.

(A-D) Expression patterns of the four jnk1 transcripts most highly expressed in the heart (arrows) at 24hfp (i, iii), 48hpf (ii, iv) from lateral (i, ii) and left oblique views (iii, iv). (A) jnk1a Ex7Lg is expressed in the cardiac cone (arrow) at 24hpf. At 48hpf strong expression is seen in the proximal part of the cardiac ventricle (v; black arrow) and weak expression is seen in some cells in the outer curvature of the atrium (a; white arrow). (B) At 24hpf jnk1b Ex8Sh expression in the cardiac cone is difficult to determine due to expression in overlapping head structures (i, iii), but is clearly seen in the outflow tract (arrow), and lower level in the ventricle, by 48hpf (ii-iv). (C) jnk1a Ex8Lg is expressed at low level in both the atrium and ventricle at 24 and 48 hpf. (D) jnk1b Ex7Sh is expressed in the region of the heart but is also expressed in surrounding tissues. Scale bar in panels i and ii 1mm, panels iii, iv 0.5mm.

Sequences of genomic DNA for jnk1a(A) and jnk1b(B) showing targeting sequence for CRISPR-Cas9 guide RNA; deletions produced; translated peptide sequence (* = stop); and restriction enzymes that identify mutants. Both jnk1a and jnk1b mutations disrupt sequence within exon 6 and predicted to lead to loss of all alternatively spliced transcripts by nonsense-mediated decay. (C) Maternal zygotic (MZ) jnk1a, MZjnk1b and MZjnk1a/ MZjnk1b mutants all appear grossly normal and are fertile. (D) RT-PCR using primers specific for jnk1a and jnk1b exon 7 and 8 show nonsense-mediated decay in MZjnk1a and MZjnk1b mutants. * = p<0.05, ** = p<0.01, **** = p<0.0001. n = 3 (3 separate clutches of 40 embryos obtained from different pairs of fish). Scale bar 1mm.

(A) Live imaging of wildtype, MZjnk1a and MZjnk1a/ MZjnk1b mutants carrying myl7:gfp transgene at 28hpf and 50hpf. (B) overall primary heart tube length is reduced in MZjnk1b and MZjnk1a/ MZjnk1b embryos at 28hpf and (C) ventricular length is reduced at 50hpf. (D) Appearances of normal, reversed and no cardiac looping visualised by the myl7:gfp transgene in wildtype embryos. (E) Minimal levels of disordered cardiac looping in MZjnk1a and MZjnk1b mutants, alone and in combination. (F) Cardiac contractile function in MZjnk1a mutants and controls at 72 hpf. See S1 Movie, control, and S2 Movie, MZjnk1a mutant. External margin of the ventricle identified in diastole (yellow dashed line) and systole (red dashed line). i) Ventricular contraction is impaired in the FHF-derived ventricular segment (black arrowhead). ii) "M-mode" representations of the movies are obtained by resampling between the yellow arrows. Hypokinesis in the MZjnk1a heart (black arrowhead) can be seen, in comparison to the normal waveform of contraction in the control heart (black arrow). (G) Reduced fractional shortening and (H) heart rate in MZjnk1a mutants at 72 hpf. (I) Histological sections through heart of wildtype and MZjnk1a embryos in plane of imaging in (F). Red arrowheads indicate hypokinetic FHF ventricular segment, red asterix indicates atrioventricular valve. a = atrium, v = ventricle, L = left. * = p<0.05, ** = p<0.01, **** = p<0.0001, ns = not significant.

(A) Whole mount preparations of hearts from MZjnk1a/b mutants at completion of FHF addition (28hpf) and completion of SHF addition (50hpf). All cardiomyocytes are labelled wth mef2 antibody (red) and the atrial cardiomyocytes are labelled with atrial specific S46 antibody (green). Compared to wildtype (wt) embryos there is no reduction in ventricular size in MZjnk1b mutants at 28hpf or 50hpf. However the ventricular chamber is reduced in size in MZjnk1a null mutants at 28hpf and 50hpf. There is no further change in ventricular chamber size in MZjnk1a/MZjnk1b mutants. (B,D) Counting of ventricular cardiomyocytes confirms these findings. (C,E) There is no alteration in numbers of atrial cardiomyocytes. (F-J) Rescue of ventricular phenotype in (G,I) MZjnk1a and (H,J) MZjnk1a/MZjnk1b embryos with jnk1a transcripts at 28hpf (G,H) and 50hpf (I,J). Immunolabelling as in (A). The reduction in ventricular cardiomyocytes in MZjnk1a or MZjnk1a/MZjnk1b embryos is only reversed by injection of jnk1a Ex7 Lg transcript. ** = p<0.02, *** = p<0.001, **** = p<0.0001, ns = not significant.

(A) Wholemount preparations of control and MZjnk1a/MZjnk1b embryos expressing myl7:gfp transgene. Differentiated cardiomyocytes identified using anti-gfp antibody to detect myl7;gfp expression and DAPI to identify nuclei. Reduction in number of cardiomyocyte nuclei at 12ss (B) and 16ss (C) in MZjnk1a/MZjnk1b in comparison to control embryos. (D) Tracks of cardiomyocytes migrating to form the heart cone in control and MZjnk1a/MZjnk1b embryos. (E) speed, (F) velocity and (G) wander index. (H) Ventricular cardiomyocytes migrating through the anterior lateral plate mesoderm (ALPM) to form the heart are visualised by WISH for myh7. The distribution of cells is identical to the wildtype pattern in MZjnk1a null mutants between 14ss and 22ss stages. However at 27hpf the ventricular component of the extended primary heart tube is shorter in MZjnk1a null mutants than controls. (I) WISH showing expression patterns of nkx2.5, gata4 and hand2 transcription factors in MZjnk1a/MZjnk1b and control embryos at the 10ss. There are no differences between expression patterns of nkx2.5 or gata4 between MZjnk1a/MZjnk1b and control embryos. However, although the anterior exent of hand2 expression is unchanged, there is a reduction in the posterior expression of hand2 (arrowheads). (J) Area of nkx2.5, gata4 and hand2 gene expression at the 10ss. Only the area of the hand2 expression is significantly different between MZjnk1a/MZjnk1b and control embryos. *** = p<0.001. *** = p<0.001, ns = not significant.

(A) Wholemount preparations of control and MZjnk1a/MZjnk1b embryos labelled with anti-gfp antibody to detect myl7;gfp-positive cells and DAPI to identify nuclei at 12ss. The bar indicates the extent of the cardiomyocyte distribution, which is deficient in the posterior part of the field in MZjnk1a/MZjnk1b embryos but is rescued by hand2 mRNA. (B) Quantification indicates restoration of cardiomyocyte numbers in MZjnk1a/MZjnk1b embryos injected with hand2 mRNA. (C) At 28 hpf the deficit in FHF ventricular cardiomyocytes (red) in MZjnk1a/MZjnk1b embryos is rescued with hand2 mRNA. (D) Counting of cardiomyocytes reveals almost complete rescue of ventricular cardiomyocytes with no change in atrial numbers. (E) Crystal structure of JNK1 indicating position of C-terminal extension (yellow and yellow arrow) and alternatively spliced exon 6a/b (blue) threonine residue (T228) present in JNK1 Ex6b (red and red arrow). based on crystal structure at https://www.uniprot.org/uniprot/P45983. (F) Mechanism of FHF ventricular hypoplasia. (i) Although all 8 possible transcripts are produced by jnk1a and jnk1b, only jnk1a Lg and jnk1b Sh forms are expressed at high levels within the heart. (ii) Of these, jnk1a Ex7Lg modulates the posterior expression domain of hand2 (directly or indirectly). (iii) Overlaying the expression fields of transcription factors directly taken from WISH in Fig 7I reveals a reduction in the posterior cardiomyocyte field which gives rise to ventricular cardiomyocytes. (iv) This reduction in cardiomyocytes produces a smaller FHF ventricular segment, with normal SHF ventricular addition which does not compensate for the FHF hypoplasia. * = p<0.05, *** = p<0.001, **** = p<0.0001, ns = not significant.

Full peptide sequences of duplicated zebrafish jnk1a and jnk1b inferred from transcripts and compared with human JNK1. Black text indicates identical amino acids, green text indicates favourable amino acid substitutions and red text divergent amino acid residues. Exons 2–6 and 9–12 are common to all transcripts and highly conserved. Both jnk1a and jnk1b genes are capable of producing exon 7 and short C-terminus containing transcripts that fully match human JNK1 transcripts. However, whilst Ex8 derived from jnk1b matches the human peptide the Ex8 from jnk1a is divergent and contains a serine rather than threonine residue (*). In contrast whilst the long C-terminal extension provided by jnk1a matches the human, the jnk1b long terminal extension is highly divergent and differs by 9/39 amino acids including insertion of an additional threonine residue.

Validation of splicing assay. (A) specificity of primers used in RT-PCR splicing assay. Primers sets to identify Ex7 and Ex8 containing variants of both jnk1a and jnk1b transcripts were assessed against all 8 transcripts contained within plasmids, using primers for the plasmid backbone (M13) as loading control. Restriction enzyme digests showing original PCR product and complete digestion by enzyme to identify C-terminal extension in jnk1a and jnk1b transcripts. (B) Graphical representation of gel densitometry indicating the splicing assay was carried out during the linear phase of amplification for jnk1a and jnk1b Ex7 and Ex8 PCR (Arbitrary logarithmic Units). (C) PCR with plasmid containing transcripts were used as positive controls for PCR reactions and indicated efficiency of each reaction.

WISH for all jnk1a and jnk1b alternatively spliced transcripts. Chromogenic WISH is not quantitative but indicates the extent of expression. (A) Prior to heart formation all transcripts are present throughout the embryo at the start of gastrulation (6hpf) and the end of gastrulation (10hpf). The most abundant transcripts by RT-PCR splicing assay are indicated by dashed lines. (B) WISH for all jnk1a and jnk1b alternatively spliced transcripts at 24hpf and 48hpf. Transcripts with obvious expression within heart—jnk1a Ex7Lg and jnk1b Ex8Sh—are indicated by arrows. (C) Close up of heart in all jnk1a transcripts at 24 and 48hpf. Outline of heart indicated by dashed lines. Absent expression indicated by arrow heads and expression in heart indicated by arrows in jnk1a Ex7Lg, jnk1a Ex8Lg and jnk1b Ex8Sh. (D) WISH demonstrated expression of jnk2 and jnk3 in the developing brain (arrows) but not heart (arrowheads). (E) RT-PCR of extracted hearts indicates that jnk2 is expressed at low level at 24 and 48 hpf, whilst there is late onset of jnk3 gene expression by 48hpf. ef1alpha is used as a loading control.

RT-PCR using primers specific for jnk1a and jnk1b exon 7 and 8 show nonsense-mediated decay in MZjnk1a and MZjnk1b mutants, but no up-regulation of other jnk1 gene products * = p<0.05, ** = p<0.01, *** = p<0.001, **** = p<0.0001.

Ventricular hypoplasia in MZjnk1a/MZjnk1b mutants. (A) Cardiac morphology at 28 hpf. All ventricular cardiomyocytes are labelled by MF20 antibody (red) and atrial cells by both MF20 and S46 antibody (appear yellow). Normal appearances seen in wildtype (wt) embryos. Ventricular hypoplasia (indicated by size of white bar) is seen in MZjnk1a null mutants but is not seen in MZjnk1b null mutants. There is no additional ventricular hypoplasia in MZjnk1a/MZjnk1b double mutants. (B) MF20 antibody staining to identify somites/skeletal muscle blocks in wild type (wt) and MZjnk1a/MZjnk1b mutants (C) Somite counting excludes global somatic developmental delay in MZjnk1a/MZjnk1b null mutants. (D,E) TUNEL labelling at 12ss stage in wild type (wt) and MZjnk1a/MZjnk1b mutants. Myocytes in the ALPM are identified by Mef2 antibody. The percentage TUNEL positive cells in Mef2 population is unchanged in MZjnk1a/MZjnk1b null mutants. (F,G) BrdU incorporation in cardiomyocytes from 10ss to 12ss. Quantification shows that there is no difference in BrdU incorporation index in MZjnk1a/MZjnk1b null mutants compared to wild type (wt) controls. ns = nonsignificant.

Unillustrated author statements