February 3, 2015;
The alternative splicing regulator Tra2b is required for somitogenesis and regulates splicing of an inhibitory Wnt11b isoform.
Alternative splicing is pervasive in vertebrates, yet little is known about most isoforms or their regulation. transformer-2b (tra2b
) encodes a splicing regulator whose endogenous function is poorly understood. Tra2b
knockdown in Xenopus results in embryos with multiple defects, including defective somitogenesis. Using RNA sequencing, we identify 142 splice changes (mostly intron retention and exon skipping), 89% of which are not in current annotations. A previously undescribed isoform of wnt11b
retains the last intron, resulting in a truncated ligand (Wnt11b
-short). We show that this isoform acts as a dominant-negative ligand in cardiac gene induction and pronephric tubule
formation. To determine the contribution of Wnt11b
-short to the tra2b
phenotype, we induce retention of intron 4 in wnt11b
, which recapitulates the failure to form somites
but not other tra2b
morphant defects. This alternative splicing of a Wnt ligand adds intricacy to a complex signaling pathway and highlights intron retention as a regulatory mechanism.
[+] show captions
Tra2b Is Required for Somite Formation and Normal Embryogenesis
(A) A translation-blocking MO was used in X. laevis, and a splice-blocking MO was used in X. tropicalis.
(B and C) Delayed gastrulation in tra2b morphants at stage 14; red arrow indicates protruding mesendoderm in morphants (103/110 embryos).
(D and E) Neural tube closure defects at stage 18. White arrows indicate fused neural folds in control embryos; red arrows point to neural folds in the open neural plate of morphants (98/101 embryos).
(F–H) Axis elongation defects and endoderm detachment at stage 23. Red arrows in (G) and (H) indicate endoderm detaching from the embryo through the blastopore (87/99 embryos).
(B–E) Dorsal view with anterior up.
(F and G) Lateral view with anterior to the left.
(H) Posterior view with dorsal up.
(I–R) In situ hybridization (ISH) on control and tra2b morphants.
(I–L) Neural plate morphology (sox2) and mesoderm specification (t/bra) at stage 15.
(M and N) Paraxial mesoderm forms in tra2b morphants, but does not segregate into segmented muscle blocks. White arrow in (M) indicates segregated muscle block in control embryo.
(O and P) Presomitic mesoderm (PSM) is present in tra2b morphants (white brackets), but presomitic stripe formation is compromised. White arrow in (O) indicates normal stripe pattern, red arrow in (P) points to smaller and fewer stripes in tra2b morphants.
(Q and R) Mature somites marked by hey1 are almost completely absent in tra2b morphants.
(S and T) Quantification of hey1-positive somites and pcdh8-positive stripes in control and tra2b morphants. Bars show mean + SD; ∗∗∗ indicates that the difference compared with control is significant at p < 2.2 × 10−16 (t test). Number of embryos used for quantification: 63 (ctrl, hey1), 58 (tra2bMO, hey1), 68 (ctrl, pcdh8), and 59 (tra2bMO, pcdh8). Embryos shown are X. laevis.
See also Figure S1.
Analysis of Alternative Splicing in tra2b Morphants Shows Intron Retention as the Most Common Splice Change
(A) Outline of the RNA-seq analysis pipeline. Condition-specific transcript assemblies are merged with JGI annotation, resulting in an augmented transcriptome assembly that forms the basis for DEXSeq testing of differential exon expression.
(B) Table showing the number of novel transcripts found in this study and the number of significant splice changes in tra2b morphants.
(C) RNA-seq reads from X. tropicalis control and morphants in tra2b locus show MO-induced intron retention (RI, blue box) and an increase in variable exon 2 (black arrow).
(D) DEXSeq output showing fitted splicing (a proxy for the number of reads aligned) across all exonic regions. Control (black) and morphant (red) exon expression is similar except for the variable exon 2 and MO-induced retained intron, where the graphs diverge. Both events are significant (purple exons indicate adjusted p < 0.05).
(E) Alternative splicing changes in morphants grouped by category show RI (red) as the most common event, followed by skipped or included exons (ESI, yellow).
(F) A plot of individual alternative splicing events shows that retained introns are always included in morphants, whereas ESI events in all but two instances are included in normal embryos.
(G) Most of the alternative splicing events detected are novel and are not described in the annotation.
See also Figure S2 and Tables S1 and S2.
Differential Gene Expression in tra2b Morphants Confirms Reduction of Muscle Transcripts
(A) Comparison of Cuffdiff2 and DESeq2 programs in calling significant changes in gene expression (X. tropicalis).
(B) Heatmap showing muscle-related genes that are repressed in tra2b morphants.
(C) Table showing the top five enriched GO terms in differentially expressed genes, with muscle-related GO terms indicated in red.
See also Figure S3.
tra2b Knockdown Reveals a Novel Inhibitory wnt11b Isoform
(A) RNA-seq read profile and Cufflinks assembled transcripts on the wnt11b locus show retention of intron 4. The top panel shows the JGI gene model for wnt11b, and the middle and bottom panels show read profiles and Cufflinks-assembled transcripts from control and morphants.
(B) Intron 4 retention results in a truncated protein (red, Wnt11b-short) resembling a dominant-negative ligand (yellow) that lacks 57 C-terminal residues compared with normal (black).
(C) wnt11b is expressed in the presomitic mesoderm/circumblastoporal region and somites. Embryos are shown in dorsal view with anterior to the left.
(D) wnt11b-short mimics wnt11b-dn in a pronephric tubule inhibition assay. Embryos were injected unilaterally into the prospective lateral mesoderm and examined by ISH for atp1a1, which marks the developing pronephros. Arrows point to the proximal pronephric tubules on the injected side.
(E) Summary of the pronephros tubule inhibition assay. Number of embryos scored: 144 (control), 72 (tra2bMO), 87 (wnt11b-dn), and 69 (wnt11b-short).
(F) qRT-PCR for induced cardiac gene expression (gata4) or axial mesoderm (t/bra) on animal caps injected with combinations of activin, wnt11b, wnt11b-dn, and wnt11b-short, showing that wnt11b-short acts similarly to wnt11b-dn and counters the effect of wnt11b. Bar plots show the mean of three independent experiments + SEM of normalized fold induction compared with activin-injected embryos.
(A) shows data from X. tropicalis and (C)–(F) show data from X. laevis. See also Figure S4.
RI in wnt11b Is Responsible for Somite Defects in tra2b Morphants
(A) Diagram showing X. laevis wnt11b gene structure and the wnt11b-in4 MO (red).
(B) RT-PCR on stage 19 single embryos injected with wnt11b-in4 MO or tra2b MO shows efficient retention of intron 4. The top panel shows RT-PCR with primers in wnt11b exon 4 and intron 4, and the bottom panel shows the internal control odc.
(C) ISH for mesodermal gene expression in wnt11b-in4 MO- and tra2b MO-injected embryos. Black arrows in pcdh8-stained embryos indicate the presence of somitic stripes in control and wnt11b-in4 MO-injected embryos. Red arrows in hey1-stained embryos indicate mature somites in control embryos, which are absent in wnt11b-in4 and tra2b morphants. The asterisk (∗) in hey1 samples indicates nonsomitic midline hey1 expression, which is exposed because of a neural tube closure defect. All pictures show dorsal views with anterior up.
(D and E) Quantification of ISH results, showing the mean and SD of the number of hey1+ somites (D) and pcdh8+ stripes (E); ∗∗∗ indicates that difference is statistically significant from control at p < 2.2 × 10−15 (t test). Number of embryos used for quantification: 41 (ctrl, hey1), 30 (tra2bMO, hey1), 43 (wnt11bMO, hey1), 42 (ctrl, pcdh8), 32 (tra2bMO, pcdh8), and 43 (wnt11bMO, pcdh8). Data shown are from X. laevis.
Figure S1, Related to Figure 1. Expression of tra2b in Xenopus laevis. Whole mount in situ hybridization was performed at the indicated stages. Embryos at stages 7 and 11 are viewed from the animal pole, and the stage 18 embryo is viewed dorsally with anterior (A) down and posterior (P) up.
Figure S2, Related to Figure 2. RNA-seq Read Profiles from replicates showing
tra2b locus and confirmation of splice changes. (A) Top shows X. tropicalis tra2b
gene transcript models with splice junction targeted by tra2b-MO2 boxed in blue. Shown
beneath is read profiles from three biological replicates of control (black) and tra2b
morphants (red). Blue box in read profiles indicates intron retention in tra2b morphants
induced by tra2b-MO2. Arrows indicate variable exon 2, which shows increased
inclusion in tra2b morphants. (B) RT-PCR confirmation of ESI events. (C) RT-PCR
confirmation of RI events. Data shown are from X. tropicalis.
Figure S3, Related to Figure 3. Confirmation of differentially expressed musclerelated
genes in tra2b morphants. RT-qPCR on genes with associated muscle GO
terms that are differentially expressed in tra2b morphants. Bars show mean of three
biological replicates and SEM normalized to control embryo expression. Red horizontal
line indicates control level. Data shown are from X. tropicalis.
Figure S4, Related to Figure 4. Induced cardiac gene expression in animal caps
and wnt11b gene duplication in X. tropicalis. (A) Animal caps injected with
combinations of activin and wnt11b isoforms assayed for gata6b and sox17 by RTqPCR.
Coinjection of activin with wnt11b induces the cardiac mesoderm marker gata6b,
whereas coinjection of activin with wnt11-dn or wnt11-short does not. The endodermal
marker sox17 is induced by activin but not by either wnt11b isoform, as expected. Bar
plots show mean of three independent experiments + SEM of normalized fold-induction
compared to activin injected embryos. (B) JGI gene models for the annotated wnt11b
(right) and the adjacent duplication transcribed from the opposite strand (xetro.H00536;
left). RNA-seq read profiles show increased retention of intron4 in both genes in tra2b
morphants (blue dotted box).