A. Western blot indicating Rbm8a, Eif4a3, Magoh and HuC proteins detected in RNase I-treated zebrafish embryo total extract (TE, lane 1), depleted extract (DE, lanes 2 and 4) immunoprecipitated protein complexes (IP, lanes 3 and 5). Antigens detected in the blot are listed on the left and antibodies used to immunoprecipitate complexes are listed on top. The signal corresponding to the antibody light chain and heavy chain in the IP lanes is indicated by IgG_L and IgG_H, respectively. B. Boxplots showing the Rbm8a RIP-Seq normalized read densities (reads per kilobase per million, RPKM) in intronic versus exonic genomic regions. Asterisk at the top indicates Wilcoxon test p-value, which is < 10⁻⁶. C. Boxplots as in B showing the Rbm8a RIP-Seq normalized read densities (RPKM) in the indicated genomic regions (bottom). Exons with downstream introns include all but last exons. Asterisk at the top indicates Wilcoxon test p-values, which are < 10⁻⁶. D. Meta-exon plots showing Rbm8a RIP-Seq and RNA-Seq (indicated on the top left) normalized read depths in a 75 nt region starting from the exon 5′ (left of dashed black line) or 3′ ends (right of dashed black line). Vertical black line: expected canonical EJC binding site (-24 nt) based on human studies. A composite exon with the relative position of exon-exon junctions (EEJ) is diagrammed at the bottom. E. A meta-exon plot of start and end of Rbm8a RIP-Seq or RNA-Seq reads (indicated on the top left; 5′ ends, solid lines; 3′ ends, dotted lines). Vertical black line: canonical EJC site (-24 nt). Gray vertical dashed lines represent boundaries of the minimal EJC occupied site. F. Top: UCSC genome browser screenshots showing read coverage along the atp2a1 gene in the Rbm8a RIP-Seq or RNA-Seq replicates as labeled on the right. Bottom: A zoomed in view of the region between the two dotted lines on the top panel. The y-axis on the left of each track shows maximal read coverage in the shown interval.

A. Schematic illustrating the rbm8a^oz36 and magoh^oz37 alleles and the predicted proteins they encode. Full-length Rbm8a and Magoh proteins are also shown. RRM: RNA Recognition Motif. B. Whole mount images of live wild-type sibling, rbm8a mutant, and magoh mutant embryos at 24 hpf. Increased grayness in the head region of homozygous rbm8a and magoh mutant embryos indicates cell death. C. Top: Western blots showing EJC protein expression in wild type (WT) sibling and rbm8a^-/- mutant embryos. Antigens detected are listed on the right and embryo genotype is listed above the blot. Developmental time points (hpf) are indicated above each lane. Protein from five (0.75 hpf) or ten embryos (all other time points) was loaded in each lane. A longer exposure (L.E.) of the 0.75 hpf lane is on the left. Bottom: Line graphs showing the amount of protein (per embryo) in the mutant embryos compared to wild-type sibling as a percent of protein present at 0.75 hpf. Error bars represent standard error of means. D. Top: Western blots as in C showing EJC protein expression in wild type (WT) sibling and magoh^-/- mutant embryos. Bottom: Line graphs showing protein quantification as in C.

A. Boxplots showing the number of spontaneous contractions per minute measured for the EJC mutant embryos and WT siblings at 24 hpf as indicated on the x-axis. Welch’s t-test p-values are indicated at the top. B-D. Immunofluorescence images showing Myh1 expression in somites 10–14 of WT sibling (B) rbm8a mutant (C) and magoh (D) mutant embryos. Antibody used was anti-A4.1025 (see methods) (N = 10 embryos/genotype). E-G. Merged confocal images of somites 12–16 in WT siblings (E) rbm8a (F) and magoh (G) mutant embryos showing immunofluorescence detection of motor neurons (anti-SV2; red) and acetylcholine receptors (α-Bungarotoxin; green). Neuro-muscular junctions in the merged image appear yellow. White arrowheads point to the end of the motor neuron. Scale bar in G is 100 nm. H. Boxplots showing the quantification of motor axon length in somites 12–15 of wild-type sibling, rbm8a mutant, and magoh mutant embryos (N = 4 embryos/genotype and 4 neurons/embryo). Welch’s t-test p-values are at the top.

A-C. MA plots (M: log ratio; A: mean average) showing genes that are upregulated (fold change > 1.5 and FDR < 0.05) (red), downregulated (fold change < 1.5 and FDR < 0.05) (blue), or unchanged (gray) in rbm8a mutant embryos compared to WT siblings at 21 hpf (A), magoh mutant embryos compared to WT siblings at 21 hpf (B), and rbm8a mutant embryos compared to WT siblings at 27 hpf (C). rbm8a, magoh, and eif4a3 are labeled in each plot with label colors signifying no change (gray) or downregulation (blue). D. Venn diagrams showing the overlap between genes that are upregulated (top) and downregulated (bottom) in rbm8a mutant embryos at 21 hpf (left) and magoh mutant embryos at 21 hpf (right). Hypergeometric test p-values are below each comparison. E. Venn diagrams as in (D) comparing upregulated and downregulated genes in rbm8a at 21 (left) and 27 hpf (right). F. PANTHER14.0 [86] gene ontology (GO) term overrepresentation analysis of genes downregulated in rbm8a and magoh mutant embryos at indicated times. All significant terms (Benjamini-Hochberg corrected p-value < 0.05) are shown for each set. The number of genes in each term is indicated at the right of each bar.

A. Venn diagram showing the overlap of significantly upregulated genes in EJC mutant embryos and upf1 morphants. Each overlap and its corresponding hypergeometric test-based p-value are color-coded. B. Cumulative distribution frequency (CDF) plot shows the empirical cumulative distribution of the fold changes in upf1 morphants (12 hpf) for genes upregulated (blue) and unchanged (black) in magoh mutant embryos at 21 hpf. The empirical CDF is the proportion of all values less than or equal to the total number of observations in the group. The CDF function shown here and in all subsequent figures (except Fig 7D) is plotted as an increasing step function along the y-axis with a jump of 1/N at each value of gene number equal to an observed value of fold change (mutant/ WT) (x-axis). Thus, the group of genes that show a rightward shift along the x-axis when compared to control group of genes is a measure of their upregulation in mutants compared to WT siblings (x-axis label). Kolmogorov-Smirnov (KS) test p-value for differences in fold changes between the two groups is indicated on the bottom right. C. CDF plot as in B for genes upregulated in rbm8a mutant embryos at 27 hpf (red) compared to unchanged genes (black). D. Quantitative RT-PCR (qRT-PCR) analysis showing fold change of select NMD target transcripts (x-axis) compared to control (mob4) transcript in magoh mutant embryos at 21 hpf compared to wild-type siblings (dark gray bars) and in rbm8a mutant embryos at 27 hpf compared to wild-type siblings (light gray bars). The selected genes either contain a 3′UTR intron (eif4a2, srsf3a and srsf7a) and/or have orthologs that are known NMD targets (gadd45aa) or were previously shown to be zebrafish Upf1 targets (gtpbp1l, atxn1b) [55]. Red dots: the value of each individual replicate. Error bars: standard error of means. Horizontal black dashed line: fold change = 1. Welch’s t-test p-values are indicated by asterisks (** p-value < 0.05). E. Empirical CDF plot, as plotted in B, of fold changes in 21 hpf magoh mutant embryos for genes that contain 3′UTR introns (APPRIS 3′UTR intron, mauve), uORF (orange), defined in Ensembl as NMD-biotype (green) compared to intron-less genes (black). KS test p-value for differences in distribution of fold changes between intron-less genes and each of the particular groups is indicated on the bottom right. F. CDF plot as in E showing the fold changes in rbm8a mutant embryos at 27 hpf.

A. Top: Schematic illustrating genes with 3′UTR introns (3′UI) where the distance between the stop codon and the 3′UI is equal to or greater than 50 nts. Such 3′UI are classified as distal. Bottom: Schematic illustrating genes with 3′UI where the distance between the stop codon and 3′UI is less than 50 nts. Such 3′UI are classified as proximal. The ribosome (brown), direction of translation (black arrow), stop codon (‘UAA’ in white), EJC (green), exon-exon junction (EEJ), coding region of mRNA (black) and 3′UI of mRNA (gray) are labeled in the top panel. B. A scatter plot showing gene-level fold change (FC) for transcripts with proximal 3′UI (dark blue: FC > 1.5 and light blue: FC < 1.5) and distal 3′UI (black: FC > 1.5 and gray: FC < 1.5) in magoh mutant embryos at 21 hpf compared to wild-type siblings. Dots encircled in red represent genes that also contain an upstream open reading frame (see methods). Genes labelled on the plot also contain a proximal 3′UI in mouse and human, and are upregulated in both rbm8a mutant datasets (Fig 6C and S4C Fig) and the upf1 KD dataset (S4D Fig). These genes were independently validated in (D). C. A scatter plot as in B showing fold changes for rbm8a mutant embryos at 27 hpf compared to wild-type siblings. D. qRT-PCR analysis showing fold changes for proximal 3′UI-containing genes (blue bars), a distal 3′UI-containing gene (light gray bar), and a Upf1-regulated gene ([55], dark gray bar) compared to the control gene (black bar) in zebrafish embryos treated with NMDI14 from 3–24 hpf. Red dots: the value of each individual replicate. Error bars: standard error of means. Horizontal black dotted line: fold change = 1. Welch’s t-test p-values (** p-value < 0.05; * p-value < 0.1).

A. A major interaction cluster predicted by STRING network analysis of genes with a 3′UI in proximal position in zebrafish, mouse and human. Nodes are colored by gene/protein function: nervous system (red), presence of RNA recognition motif (RRM) (green), diseases of signal transduction (blue), FoxO signaling pathway (yellow). (167 nodes and 127 edges in total, PPI enrichment p-value = 0.02). B. Gene ontology enrichment analysis of all 167 genes with conserved 3′UI proximal positioning. The most significant GO term within the following functional categories are shown: Interpro domains, Biological process and Reactome pathways. C. A scatter plot showing fold changes for APPRIS-annotated proximal 3′UI transcripts (dark blue: FC > 1.5 and light blue: FC < 1.5) and distal 3′UI transcripts (black: FC > 1.5 and gray: FC < 1.5) in UPF1 knockdown HEK293 cells compared to control cells [57]. Dots encircled in red are transcripts that also contain an uORF as determined previously in [57]. Same analysis for Ensembl transcript annotations is shown in Fig S5G. D. Empirical CDF plot showing change in mRNA stability for different classes of NMD targets and intron-less genes upon UPF1 knockdown in HEK293 cells (data from [57]). The gene classes are as follows: proximal 3′UI-containing genes where distance is 20–50 nts (light green), 30–50 nts (olive green) and 36–50 nts (dark blue), Ensembl-annotated NMD-biotype genes (red) and intron-less genes (black). CDF function is plotted as in Fig 5B. KS test p-value for comparison of NMD targets to intron-less genes is indicated in the same color. E. qRT-PCR analysis showing fold changes for proximal 3′UI-containing genes CDKN1B, FOXO3 (upregulated in zebrafish); ULBP1; STX3 (highest change in stability upon UPF1 KD in HEK293 cells [57]); HNRNPD (encodes RRM-containing protein) and distal 3′UI-containing genes (ARC and SRSF4) upon UPF1 (left) and EIF4A3 (right) knockdown in HCT116 cells. The distance between stop codon and 3′UI for every 3′UI-containing gene is indicated below each bar. TBP is the normalizing gene used for qRT-PCR analysis. Welch’s t-test p-values are indicated using asterisks (** p-value < 0.05 and * p-value < 0.1). F. Empirical CDF plot of fold changes in levels of 3′UI-containing isoforms (dark blue) as compared to 3′UI-lacking isoforms (sky blue) encoded from same genes in UPF1-depleted HEK293 cells. KS test p-value for differences in the two distributions is indicated on the bottom right. G. CDF plot as in F showing the fold changes in levels of distal 3′UI-containing (black) versus 3′UI-lacking isoforms (gray). H. qRT-PCR analysis showing fold changes for 3′UI-containing and 3′UI-lacking isoforms encoded by proximal 3′UI-containing genes ALG8 and SLC30A7 and by a distal 3′UI-containing gene SRSF4 upon UPF1 knockdown in HCT116 cells. The distance of 3′UI when present, or its absence (-) is indicated below each bar. TBP is the normalizing gene used for qRT-PCR analysis. Welch’s t-test p-values for the three biological replicates of each transcript are < 0.05. The Welch’s paired t-test p-values for the comparison between fold changes of 3′UI-containing and 3′UI-lacking isoforms is indicated using red asterisks (** p-value < 0.05).

A. Illustration showing foxo3b gene structure in indicated vertebrates. The distance between the stop codon and the proximal 3′UTR intron is on the right. Open rectangles: UTRs, filled rectangles: coding region, gray lines: introns (hash marks denote shortened intron sequences). B. Top: Western blot showing protein levels in wild-type sibling (lane 1) and magoh mutant (lane 2) embryos at 21 hpf. Bottom: a dot plot showing Foxo3b levels normalized to tubulin levels in magoh mutant embryos and WT siblings at 21 hpf in three biological replicates. (N = 5 embryos per genotype per replicate). Error bars: standard error of means. C-F. Confocal images showing Myh1 immunofluorescence using anti-A4.1025 in somites 12–16 of WT sibling (C), magoh^-/- mutant (D), magoh^-/-; foxo3b^+/- mutant (E), and magoh^-/-; foxo3b^-/- mutant (F) embryos at 26 hpf (N = 13 embryos/genotype). G-J. Merged confocal images showing motor neurons (red; detected by anti-SV2 staining) and acetylcholine receptors (green; detected by alpha-bungarotoxin staining) in somites 12–16 of WT sibling (G), magoh^-/- mutant (H), magoh^-/-; foxo3b^+/- mutant (I), and magoh^-/-; foxo3b^-/- mutant (J) embryos. Neuromuscular junctions in the merged images are yellow. White arrowheads point to the distal end of the motor neuron. (N = 13 embryos per genotype). Scalebar in J (for panels C-J) is 100 nm. K. Boxplots showing quantification of motor axon length in embryos of genotypes indicated along the x-axis (4 motor neurons/embryo and 13 embryos/genotype). Welch’s t-test p-values for comparison between magoh^-/- mutant and magoh^-/-; foxo3b^-/- mutant embryos are at the top.

A. Top: EJCs can function to enhance NMD via translation stimulation. Ribosome (brown), direction of translation (straight arrow), start codon (‘AUG’ in white), EJC (green), exon-exon junction (EEJ), mRNA coding region (thick black line) and mRNA 5′ and 3′UTRs (thin gray lines) are labeled. Middle: a model for downstream EJC function in NMD of transcripts where the distance between stop codon (‘UAA’ in white) and downstream exon-exon junction is 36 nts or greater to accommodate both the EJC and the ribosome. Bottom: a model for NMD of transcripts where the distance between stop codon and downstream exon-exon junction is less than 36 nts. Displacement of stop codon-proximal EJC by the ribosome is shown. In this case other downstream factors such as a non-canonical EJC (ncEJC), EJC-interacting RBPs (e.g. SR proteins), or 3′UTR length and/or specific sequences may be responsible for NMD. B. A schematic depicting a model for EJC- and NMD-dependent regulation of foxo3b as a genetic pathway that is critical for zebrafish motor neuron development.

A. Multiple sequence alignments of Eif4a3, Rbm8a and Magoh protein sequences from organisms on the left. Consensus sequence is at the bottom with upper case letters indicating identity and lower case letters indicating similarity. Green indicates complete identity across all species, yellow and blue indicate the identical and unique amino acids in the regions with similarity. Identity between human and zebrafish EJC proteins: Eif4a3 (97%), Rbm8a (93%) and Magoh (100%). B. Western blot detecting proteins listed on the left in RNase I-treated zebrafish embryo total extract (TE, lane 1), depleted extract (DE, lanes 2, 4 and 6) and immunoprecipitates (IP, lanes 3, 5 and 7) with the Rbm8a antibody. Detergents supplemented to increase IP stringency are indicated on top of each lane. Optimized IP condition used in S1C is indicated by the dashed red box. C. Autoradiogram of γ32P 5′-end labeled RNAs from anti-Rbm8a RIP elution (lane 4) as well as indicated size-markers which include the low-molecular weight single-stranded DNA ladder (lane 1), 0.1 pmol 28 nt synthetic RNA (lane 2) and 100 bp DNA ladder (lane 3). D. Scatter plots comparing read counts for each gene in a pair of RIP-Seq replicates. The replicates (Rep1, Rep2, and Rep3) are indicated on the x- and y-axes. A pseudocount of 0.0001 was added to all genic read counts before log2 transformation. Pearson correlation coefficient (r) and p-value for the correlation test for each comparison is on the top left of each plot. E. Genome browser screenshots showing read coverage of Rbm8a RIP-Seq (only Rep 3, the deepest replicate is shown) in green and RNA-Seq in gray of select highly-expressed genes, krt4, eef2b, eif4g1a, and hist1h4l (intron-less gene).

A. Whole mount images of live 19 hpf EJC mutant embryos and WT sibling embryos stained with acridine orange. B. Whole mount images of live EJC mutant embryos and WT siblings at 21 hpf. C. Whole mount images of live EJC mutant embryos and WT siblings at 27 hpf. D. Whole mount images of live EJC mutant embryos and WT siblings at 32 hpf. Head necrosis is indicated by the dashed circle.

A. Venn diagram showing the overlap of genes that are significantly upregulated in EJC mutant embryos and upf1 morphant embryos at 24 hpf [55]. Hypergeometric test p-values for each comparison are also shown. B. MA plot showing the genes that are altered in expression (fold change > 1.5 and FDR < 0.05 in red and unchanged genes in gray) in upf1 morphant embryos compared to control embryos at 12 hpf. The number of significantly upregulated genes is at the top right and the number of downregulated genes is at the bottom right. C. Venn diagram showing the overlap of significantly upregulated genes in upf1 morphant embryos at 24 hpf [55] and upf1 morphant embryos at 12 hpf. Hypergeometric test p-value for the comparison is indicated. D. Empirical CDF plot showing the fold changes in upf1 morphant embryos (24 hpf) [55] of upregulated (blue) and unchanged genes (black) in 21 hpf magoh mutant embryos. Kolmogorov-Smirnov (KS) test p-value for differences in the two distributions are indicated at the bottom of the class descriptions. E. Empirical CDF plot as in S3D for upregulated (red) and unchanged genes (black) in rbm8a mutant embryos at 21 hpf. F. Empirical CDF plot as in S3D for upregulated (red) and unchanged genes (black) in rbm8a mutant embryos at 27 hpf. G. Empirical CDF plot showing the fold changes in upf1 morphants (12 hpf) of upregulated (red) and unchanged genes (black) in rbm8a mutant embryos at 21 hpf. H and I. Proportion of uORF-containing (H) or 3′UI-containing (I) genes in the total number of genes showing significant (FDR < 0.05) fold changes in rbm8a mutant, magoh mutant and upf1 morphant embryos. Genes are divided into four categories based on their log2 fold change: >1.5, 1.5 to 0, 0 to -1.5 and < 1.5.

A. Histogram depicting the frequency of all zebrafish 3′UI transcripts in Ensembl GRCz10 (with APPRIS annotation) as a measure of the distance of the 3′UI from the stop codon. Data are shown in 5 nts bins and bins beyond 500 nts are not shown. Bins of proximal 3′UI genes are in blue and distal 3′UI bins are in gray. Inset: Histogram of all zebrafish proximal 3′UI transcripts binned by 1 nt. B. PANTHER14.0 [87] gene ontology (GO) term enrichment analysis of proximal 3′UI-containing genes (top, shades of blue) and all 3′UI-containing genes (bottom, shades of gray). All significant terms (Benjamini-Hochberg corrected p-value < 0.05) are shown for each set. C. A scatter plot showing gene-level fold change (FC) for transcripts with proximal 3′UI (dark blue: FC > 1.5 and light blue: FC < 1.5) and distal 3′UI (black: FC > 1.5 and gray: FC < 1.5) in rbm8a mutant embryos at 21 hpf compared to wild-type siblings. Genes encircled in red also contain a uORF as determined from a previously published dataset (see Materials and Methods). D. A scatter plot as in C showing fold changes of 3′UI-containing genes for 12 hpf upf1 morphants compared to wild-type control embryos. E. Integrated genome browser (IGV) screenshots of Sashimi plots showing RNA-seq reads observed for foxo3b, phlda2 and cdkn1ba in four zebrafish 24 hpf whole embryo RNA-seq datasets (as labeled on figure in different colors) obtained from the DanioCode consortium. Range of the number of reads mapping to the genes are indicated to the left of each track in black. Number of junction reads are indicated at the spliced junction in the color corresponding to the specific track. In case of foxo3b, due to the length of the second intron the screenshots of the first two and the last two exons are shown separately. F. Empirical CDF plot showing the fold changes in magoh mutants (21 hpf) of genes upregulated that contain a proximal (green) or distal (light purple) 3′UI or intron-less genes (black). Genes that also contained an uORF were excluded from analysis. Kolmogorov-Smirnov (KS) test p-value for differences in fold changes between the two groups is indicated on the bottom right. G. CDF plot as in F showing the fold changes in rbm8a mutants (27 hpf) of genes upregulated that contain a proximal (green) or distal (light purple) 3′UI or intron-less genes (black). Kolmogorov-Smirnov (KS) test p-value for differences in fold changes between the two groups is indicated on the bottom right.

A. Histogram showing the frequency of all APPRIS 3′UI-containing transcripts in human GRCh38 as a measure of the distance of the 3′UI from the stop codon. Data are grouped in 5 nt bins from 1–500 nts. Proximal 3′UI-containing gene bins are indicated in blue; distal 3′UI-containing gene bins are indicated in gray. Red dotted line indicates distance from stop codon to farthest 3′UI = 50 nts. Histogram of transcripts from Ensembl annotation is shown in C. B. Histogram as in A of mouse proximal and distal 3′UI-containing transcripts in mouse GRCm38. Histogram of transcripts from Ensembl annotation is shown in D. C. Histogram as in A of proximal and distal 3′UI-containing human transcripts from Ensembl annotation of GRCh38. D. Histogram as in A of proximal and distal 3′UI-containing mouse transcripts from Ensembl annotation of GRCm38. E. PANTHER14.0 [86] gene ontology (GO) term enrichment analysis of human APPRIS proximal 3′UI-containing genes (shades of blue) and all human APPRIS 3′UI-containing genes (shades of gray). All significant terms (Benjamini-Hochberg corrected p-value < 0.05) are shown for each set. F. GO term enrichment analysis as in E of mouse APPRIS 3′UI-containing genes. G. A scatter plot showing gene-level fold changes for all Ensembl-annotated transcripts with proximal 3′UI (dark blue: FC > 1.5 and light blue: FC < 1.5) and distal 3′UI (black: FC > 1.5 and gray: FC < 1.5) in UPF1 knockdown human embryonic kidney cells (HEK293) compared to control cells using previously published data [57]. Genes encircled in red also contain a uORF as determined from a previously published dataset (see Methods). H. A scatter plot showing gene-level fold changes for all APPRIS-annotated transcripts with proximal 3′UI (dark blue: FC > 1.5 and light blue: FC < 1.5) and distal 3′UI (black: FC > 1.5 and gray: FC < 1.5) in UPF1 knockdown human embryonic stem cells (hESCs) compared to control cells using previously published data [62]. Genes encircled on red also contain a uORF as determined from a previously published dataset (see Materials and Methods). I. Scatter plot as in E showing gene-level fold changes of APPRIS-annotated mouse 3′UI-containing transcripts in Smg6-/- knockout mouse embryonic stem cells (mESCs) compared to wild-type cells [63]. J. Western blots showing representative protein knockdown in one replicate for Fig 7E.

A. Semi-quantitative RT-PCR shows transcript levels of foxo3b, eif4a2 and rpl13 (loading control) in rbm8a mutant and wild-type sibling embryos at 21 and 27 hpf. B. Western blots (on the left) show levels of Foxo3b, Rbm8a, Magoh, and Tubulin in rbm8a mutant embryos (lane 2) compared to WT siblings (lane 1) at 27 hpf (N = 20 embryos per genotype). Right: dot plot showing Foxo3b levels normalized to tubulin levels in rbm8a mutant embryos and WT siblings at 27 hpf in three biological replicates. Error bars: standard error of means. Welch’s t-test p-values are indicated at the top. C. Bar graph showing log2 fold changes of known Foxo3b transcriptional targets that show a significant upregulation (FDR < 0.05) in EJC mutant RNA-Seq datasets. Foxo3b targets are from Morris et al. 2015 [64]. A pound symbol indicates statistically non-significant log2 fold change with FDR > 0.05. D-G. Confocal images showing Myh1 immunofluorescence using anti-A4.1025 in somites 12–16 of WT sibling (D), rbm8a-/- mutant (E), rbm8a-/-; foxo3b+/- mutant (F), and rbm8a-/-; foxo3b-/- mutant (G) embryos. (N = 5 embryos/genotype). H-K. Merged confocal images showing motor neurons (red; detected by anti-SV2 staining) and acetylcholine receptors (green; detected by alpha-bungarotoxin staining) in somites 12–16 of WT sibling (H), rbm8a-/- mutant (I), rbm8a-/-; foxo3b+/- mutant (J), and rbm8a-/-; foxo3b-/- mutant (K) embryos. Neuromuscular junctions in the merged image are yellow. White arrowheads point to the distal end of the motor neuron. (N = 5 embryos/genotype). Scalebar in K (for panels D-K) is 100 nm. L. Boxplots showing quantification of motor axon length in embryos of genotypes indicated along the x-axis) (4 motor neurons/embryo and 5 embryos/genotype). Welch’s t-test p-values are indicated at the top.

Unillustrated author statements