Differential responses to Wnt and PCP disruption predict expression and developmental function of conserved and novel genes in a cnidarian

Abstract

We have used Digital Gene Expression analysis to identify, without bilaterian bias, regulators of cnidarian embryonic patterning. Transcriptome comparison between un-manipulated Clytia early gastrula embryos and ones in which the key polarity regulator Wnt3 was inhibited using morpholino antisense oligonucleotides (Wnt3-MO) identified a set of significantly over and under-expressed transcripts. These code for candidate Wnt signaling modulators, orthologs of other transcription factors, secreted and transmembrane proteins known as developmental regulators in bilaterian models or previously uncharacterized, and also many cnidarian-restricted proteins. Comparisons between embryos injected with morpholinos targeting Wnt3 and its receptor Fz1 defined four transcript classes showing remarkable correlation with spatiotemporal expression profiles. Class 1 and 3 transcripts tended to show sustained expression at "oral" and "aboral" poles respectively of the developing planula larva, class 2 transcripts in cells ingressing into the endodermal region during gastrulation, while class 4 gene expression was repressed at the early gastrula stage. The preferential effect of Fz1-MO on expression of class 2 and 4 transcripts can be attributed to Planar Cell Polarity (PCP) disruption, since it was closely matched by morpholino knockdown of the specific PCP protein Strabismus. We conclude that endoderm and post gastrula-specific gene expression is particularly sensitive to PCP disruption while Wnt-/β-catenin signaling dominates gene regulation along the oral-aboral axis. Phenotype analysis using morpholinos targeting a subset of transcripts indicated developmental roles consistent with expression profiles for both conserved and cnidarian-restricted genes. Overall our unbiased screen allowed systematic identification of regionally expressed genes and provided functional support for a shared eumetazoan developmental regulatory gene set with both predicted and previously unexplored members, but also demonstrated that fundamental developmental processes including axial patterning and endoderm formation in cnidarians can involve newly evolved (or highly diverged) genes.

Figure 1. Identifying Clytia early gastrula patterning genes by DGE.

A) Visualization of variation between two independent Wnt3-MO early gastrula replicate samples (“MA plot”). For each reference transcript (grey dots), M is the difference of read density between the two samples and A its average expression level in the combined samples (see Methods). Red and blue dotted lines give estimates of the distribution zone outside which values are statistically significant, based on the standard deviation (4xSD) of a theoretical distribution (red) calculated using the random sampling model, or real variation (blue) estimated by the comparison of technical replicates . B) Demonstration of the near-normal distribution of the log₂ counts of the mapped reads from one of the non-injected embryo samples by a QQ plot, a necessary condition to use the Random Sampling Model assumption in DGE analysis. C) MA plot of read data from Wnt3-MO versus non-injected embryo samples. Applying the 1% cut-off p-value for statistical significance corresponding to the threshold z-score of +/−3.3 identifies 148 transcript sequences as over-expressed (green dots) and 232 as under-expressed (orange dots) in Wnt3-MO embryos. The more stringent +/−5.0 threshold for z-score values eliminates a cluster of genes with expression characteristics very close to the overall population of non-differentially expressed genes, as demonstrated in the histogram (insert) and reduces the number of transcripts to 44 and 135 for Wnt3-MO embryo over- and under-expressed transcripts respectively. D, E) Z-scores for the Wnt3-MO embryo over- (green dots) and under- (orange dots) expressed transcripts plotted against those for two other morpholino-injected embryo groups harvested at the same developmental stage. Z-scores were calculated for experimental versus non-injected values in each case. D: Wnt3-MO versus Fz1-MO; E: Wnt3-MO versus Fz3 MO. Transcripts significantly under-expressed in all three MO groups (z-scores less than -5: probably from bacterial contaminants) are represented as blue dots and over-expressed (z-scores greater than 5; probably injection damage-related) as purple dots.

Figure 2. Preferential expression of Wnt3-MO-embryo under-expressed transcripts in oral and ingressing cells.

In situ hybridization analysis of Wnt3-MO-embryo under-expressed transcripts in early gastrula, 24hpf planula and 48hpf planula stages (from left to right in each panel). The expression of Wnt3 (A; Momose et al., 2008) is compared with that of the 20 down-regulated transcripts with the highest z-scores. All genes showed heavily oral-biased expression at the early gastrula stage, with two principle distribution patterns: Predominantly in the oral ectoderm at all stages (panels A-I; “Oral” expression pattern, outlined in red); Mainly in ingressing/ingressed cells during gastrulation and then in the endodermal region of the planula (panels J-U; ‘Ingressing/Endodermal” expression pattern, outlined in orange). Two of the “Oral pattern” transcripts (panels H and I) also showed expression in the endoderm at the planula stage. Representative images of the patterns observed in at least three experiments are shown. All embryos are oriented with the oral pole uppermost. Bottom right: gene name (see Table 1). Scale bar 50 µm.

Figure 3. Preferential expression of Wnt3-MO-embryo over-expressed transcripts in aboral domains and planulae.

In situ hybridization analysis of Wnt3-MO-embryo over-expressed transcripts in early gastrula, 24hpf planula and 48hpf planula stages (from left to right in each panel). The 18 transcripts with the highest z-scores were analyzed. Seven genes showed expression clearly restricted to the aboral territories at planula stages (A-G), already apparent at the early gastrula stage except in in the case of Dkk (C). Two additional transcripts showed an aboral expression pattern but with additional signal in endodermal region of planula larvae: ZpdA (H) and Botch2 (I). 9 transcripts (J-R) showed low ubiquitous or undetectable expression at the gastrula stage followed by a variety of patterns in planula larva including ubiquitous (Q-R), mainly endodermal (K-O) and/or with diverse dynamic distributions of scattered cells in the ectoderm and/or endoderm (J, M, N, O, P). We classified the expression profiles as “Aboral-type” (A-G, panels outlined in green), “Delayed” type (L-R, panels outlined in black) or as showing a mixture of these profiles (4 remaining panels). Representative images from at least three experiments are shown. All embryos are oriented with the oral pole uppermost. Bottom right: gene name (see Table 1). Scale bar 50 µm.

Figure 4. Summary of expression profiles observed for the analyzed transcripts.

The differentially expressed transcripts analyzed in this study showed four basic types of expression profile: Oral (oral pole ectoderm at all stages); Ingressing/Endodermal (cells moving into the endodermal region during gastrulation and persisting in this region); Aboral (aboral pole throughout) and Delayed (absent or very low expression at the beginning of gastrulation, then expression in a variety of regions/cell types at the planula stages). Within these four categories there were differences in the limits of detected expression as indicated.

Figure 5. Equivalent differential expression responses determined by DGE and Q-PCR.

For ten selected transcripts from different DGE classes and with different expression profiles (see Table 1, Figure 4), transcript levels at the early gastrula stage were determined by Q-PCR in Wnt3-MO and non-injected early gastrula embryos. The ratio of expression levels of selected genes between injected and control embryos, normalized with respect to EF-1α is compared to the DGE data represented in the same way by using the counts of reads mapped rather than the number of cycles of Q-PCR amplification. Transcript identities are shown beside each pair of bars.

Figure 6. Stereotyped modification of gene expression patterns in Wnt3-MO embryos.

In situ hybridization analysis of selected transcripts in untreated (left panels) and Wnt3-MO injected (right panels) early gastrulae. Each expression pattern type is represented in the transcripts analyzed, as indicated on the left. A-C: O-type pattern transcripts; D-F: IE-type pattern transcripts, G-I: A-type pattern transcripts; J-L: D-type pattern transcript. For Botch1 (J) and bZip (K), individual cells with high expression on the blastocoelar face of the ectoderm were discernible in the Wnt3-MO embryos. Representative images of the patterns observed in at least three experiments are shown. All control embryos are oriented with the oral pole uppermost. Gene identity is shown in the bottom right of each pair of panels. Scale bar 50 µm.

Figure 7. DGE responses correlate with spatial expression profiles.

A) Schematic representation of the 4 types of expression profile observed amongst characterized transcripts (see Figure 4) at early gastrula (left) and planula (right) stages, showing their mapping onto the differential responses of the transcripts to Wnt3-MO and Fz1-MO. indicated on the z-score plot in the center. Four DGE classes were defined on the basis of z-scores in Wnt3-MO and Fz1-MO embryos applying cutoffs of -5 for classes 1 and 2, and +5 for classes 3 and 4 as indicated respectively by the dark and light orange, green and gray zones on the graph. The numbers indicate how many of the transcripts for which expression patterns were determined for each class showed the corresponding expression profile. These transcripts are all listed in Table 1 and with expression patterns shown in Figures 2, 3 and 6. B) Equivalent Z-score plot mapping a transcriptome dataset for Stbm-MO early gastrula-stage embryos against the Fz1-MO dataset. Dot colors correspond to the 4 classes defined in A. There is a strong correlation between expression responses in these two conditions, especially for classes 3 and 4 (green and grey dots, ie transcripts over-expressed in Wnt3-MO embryos). C) Z-score plot mapping the Stbm-MO transcriptome dataset embryos against the Wnt3-MO dataset. This illustrates that the 4 DGE classes defined on the basis of Fz1-Mo and Wnt3-MO responses situate largely but not exclusively in the equivalent zones of the Stbm-MO vs Wnt3-Mo graph.

Figure 8. Fz1-MO and Stbm-MO have very similar effects on gene expression profiles.

Each set of 3 panels shows typical in situ hybridization patterns obtained for uninjected embryos (left), Fz1-MO embryos (center) and Stbm-MO embryos (right) injected, fixed at the early gastrula stage and processed in parallel. A-B: O-type pattern transcripts (Myb and ZnfO); C-D: IE-type pattern transcripts (FoxA and Znf845); E-F: A-type pattern transcripts (ZnfA and sFRP-A); G-H: D-type pattern transcripts (Botch1 and bZip). Cells with high Botch1 and bZip expression on the blastocoelar face of the ectoderm were discernible (narrows) in Fz1MO and StbmMO embryos. Gene identity is shown in the bottom left of each set of panels. For each transcript the two morpholinos have very similar effects on the expression patterns, confirming the z-score comparisons (File S1). Scale bar 50 µm.

Figure 9. Morpholinos targeting conserved and cnidarian-specific transcripts disrupt development.

DIC imaged larvae developed from morpholino-injected eggs after 24hpf and 48hpf development or at the early gastrula stage as indicated. Typical morphology observed at non-toxic morpholino doses are shown in each case, with the oral pole at the top. Similar phenotypes were obtained using a second morpholino in all cases, except FoxQ2c and WegA1, for which no second non-toxic morpholino could be designed. Uninjected embryos (A) and control-MO injected embryos (L) had completed gastrulation at 24hpf with endodermal epithelialisation starting at the aboral pole. By 48hpf uninjected planula larvae (G, Q) had formed with endodermal epithelial layers (black arrows) organized around a central stripe-like cavity and a well-defined lamina layer separating the endoderm and ectoderm (green arrows). WegO1-MO1 embryos showed minor disruption of gastrulation with embryo elongation slightly compromised at 24hpf (B) and the oral half, thin and tapered (asterisk) at 48hpf (H). Morpholinos targeting either Bra1(C, I) or Bra2 (D, J) showed severe delays in gastrulation. By 48 hpf some embryo elongation had occurred but the blastocoel contained only loose, disorganized material. WegA1-MO embryos (F) showed massive cell ingression at the onset of gastrulation, when cell ingression had barely initiated in uninjected embryos cultured in parallel (E). At 48 hpf (K) the endoderm was enlarged and the ectoderm layer was very thin and irregular (black arrowhead). FoxQ2c-MO (M, R) and WegIE2-MO (N, S), showed severely reduced endodermal layer at 48hpf. WegD1-MO and HD02-MO embryos showed severe defects in overall morphology (O, P, T, U), lacking a well defined basal lamina between endoderm and ectoderm. Green arrows indicate the position of this interface. Black arrows indicate the endodermal epithelial layers in G, Q, R and S. Red arrows indicate ingressing cells at the onset of gastrulation in C, D, E and F. Scale bar 50 µm for all panels.

Figure 10. Conserved and cnidarian-specific genes in all DGE classes.

Analysis performed on a sub-set of 126 transcript sequences in which predicted ORFs were complete at both 3′ and 5′ ends. Proportions of novel and conserved gene types, including transcription factors and signaling pathway regulators are similar in all groups except DGE class 4, which includes fewer cnidarian-specific genes.