Asymmetrically expressed axin required for anterior development in Tribolium

Abstract

Canonical Wnt signaling has been implicated in an AP axis polarizing mechanism in most animals, despite limited evidence from arthropods. In the long-germ insect, Drosophila, Wnt signaling is not required for global AP patterning, but in short-germ insects including Tribolium castaneum, loss of Wnt signaling affects development of segments in the growth zone but not those defined in the blastoderm. To determine the effects of ectopic Wnt signaling, we analyzed the expression and function of axin, which encodes a highly conserved negative regulator of the pathway. We found Tc-axin transcripts maternally localized to the anterior pole in freshly laid eggs. Expression spread toward the posterior pole during the early cleavage stages, becoming ubiquitous by the time the germ rudiment formed. Tc-axin RNAi produced progeny phenotypes that ranged from mildly affected embryos with cuticles displaying a graded loss of anterior structures, to defective embryos that condensed at the posterior pole in the absence of serosa. Altered expression domains of several blastodermal markers indicated anterior expansion of posterior fates. Analysis of other canonical Wnt pathway components and the expansion of Tc-caudal expression, a Wnt target, suggest that the effects of Tc-axin depletion are mediated through this pathway and that Wnt signaling must be inhibited for proper anterior development in Tribolium. These studies provide unique evidence that canonical Wnt signaling must be carefully regulated along the AP axis in an arthropod, and support an ancestral role for Wnt activity in defining AP polarity and patterning in metazoan development.

In vertebrate embryogenesis, Wnt signaling is required for posterior development and must be repressed for normal anterior development (1). Ectopic activation of Wnt signaling leads to posteriorization of the anterior neural plate anlagen in vertebrates (2), and mutations in axin, a negative regulator of the pathway, lead to reduced head and eyes in zebrafish (3). In Drosophila, Wnt signaling functions to define the boundaries of and polarity within segments but is not required for polarization along the entire AP axis (4). Ectopic activation of Wnt targets by loss of axin function produces segmental defects (5) but does not result in posteriorization or loss of anterior identities. In Tribolium and other short-germ arthropods, where segments are added sequentially, posterior Wnt signals are required for germ-band elongation and formation of abdominal segments (6–8). Segments that form in the Tribolium blastoderm develop normally, suggesting Wnt signaling is not necessary for anterior development. However, it is not clear whether Wnt activity is simply not relevant to anterior development, or must be carefully regulated. The molecular functions of canonical Wnt pathway components are highly conserved between Drosophila and vertebrates (9). Assuming they are similarly conserved in other insects, we examined expression and function of axin in Tribolium embryos.

Results

Asymmetric Distribution of Tc-axin Transcripts in Early Stages of Embryogenesis.

In situ hybridization revealed Tc-axin transcripts initially localized at the anterior pole of freshly laid eggs containing only the pronuclei (Fig. 1 A and A′). During cleavage stages the Tc-axin expression domain extended more posteriorly (Fig. 1 B, B′, C, and C′). At the differentiated blastoderm stage, Tc-axin was expressed ubiquitously, at somewhat higher levels in the embryo (posterior) than in the serosa (anterior) (Fig. 1 D and D′).

Fig. 1.

Expression of Tc-axin in wild-type embryos. (A–D) Brightfield images of Tc-axin expression, anterior to the left. (A′–D′) DAPI nuclear counterstained images of the same embryos. Maternal Tc-axin transcripts were localized at the anterior pole in freshly laid eggs (A and A′). During cleavage and early blastoderm stages, the expression domain expanded posteriorly (B, B′, C, and C′) until Tc-axin was expressed ubiquitously in the embryo in subsequent stages (D and D′). (Magnification: 70×.)

Tc-axin RNAi Affects Anterior Fates in Progeny.

Female adults injected with Tc-axin dsRNA (50 ng/μL) produced embryos displaying a wide range of phenotypes. Eggs collected during the first 13 d after injection failed to hatch and did not contain cuticles (Fig. S1A), suggesting loss of a function essential to early development. After 2 wk, as the RNAi effects began to wear off (10), embryos secreting partial cuticles were recovered. Careful examination revealed that they contained varying numbers of abdominal segments, but no head or thorax (Fig. 2 D and E). Ten days later, a wider range of phenotypes was observed; in addition to those containing abdomen only, others contained abdomen and thorax, or abdomen thorax and some head segments (Fig. 2 B and C), but no wild-type cuticles were found. When the amount of injected dsRNA was decreased to 10 ng/μL, the effects wore off more quickly (Fig. S1 B–D). Only empty eggs were recovered within the first 9 d, but partial cuticles displaying the same range of phenotypes described above were recovered thereafter. Close inspection of head segments revealed a graded sensitivity to RNAi in which the most anterior structures were most often missing (Fig. S1D). Thus, the severity of the response to Tc-axin RNAi was both dose-dependent and graded along the AP axis.

Fig. 2.

Comparison of wild-type and RNAi mutant cuticles. (A) Wild-type larvae (anterior to the left) contain head, thorax, and abdomen. (B–E) After Tc-axin RNAi, a graded loss of anterior structures was observed in cuticles from unhatched eggs. (B) Weak phenotype with reduced anterior head structures. (C) Intermediate phenotype with head and anterior thorax missing. (D) Stronger phenotype lacking head, thorax, and anterior abdomen. (E) Severe cuticular phenotype in which only gut (arrow) and posterior terminus formed. (F–H) Tc-pan RNAi results in truncated embryos. (F) Weak phenotype with loss of distal appendages and posterior abdominal segments. (G) Intermediate pheontype with small head and fewer abdominal segments. (H) Fully truncated embryo missing all abdominal segments. (I–K) Tc-pan, Tc-axin double RNAi. A range of phenotypes, (I) weak, (J) intermediate, and (K) strong, similar to Tc-pan single RNAi indicates that Tc-pan functions epistatic to, or downstream of, Tc-axin function. (Magnification: 25×.)

Anterior Shift of Blastodermal Fates.

To assess early fate-map changes in more detail, we analyzed the expression of early developmental markers including the serosal marker Tc-zerknüllt1 (Tc-zen1), the posterior marker Tc-caudal (Tc-cad), and the pair-rule protein Tc-Even-skipped (Tc-Eve) in Tc-axin RNAi embryos. At this stage in wild-type, Tc-zen1 is expressed in the large, widely spaced cells of the serosa, which extends dorsally (Fig. 3 A and A′) (see also ref. 11). In Tc-axin RNAi embryos, the few remaining serosal cells were restricted to the anterior tip of the egg and expressed Tc-zen faintly (Fig. 3 B and B′). At the blastoderm stage, Tc-cad is expressed in the posterior half of wild-type eggs (Fig. 3 C and C′) (see also ref. 12). In Tc-axin RNAi embryos, the Tc-cad expression domain was shifted anteriorly, and the anterior embryonic tissue free of Tc-cad staining, which is normally fated to develop into anterior head, was greatly reduced (compare bars in Fig. 3 C′ and D′). In addition, the nuclei of the few remaining serosal cells were farther apart than in wild-type embryos. In wild-type embryos, Tc-Eve is expressed in the posterior half of early blastoderm stage embryos (Fig. 3 E and E′) (see also ref. 13). In Tc-axin RNAi embryos, the Tc-Eve expression domain was shifted anteriorly (compare white arrowheads in Fig. 3 E′ and F′). Tc-cad and Tc-eve expression domains remained more anterior throughout the blastoderm stage (quantified in Fig. S2). Reduction of the Tc-zen1 expression domain and anterior shifts in the expression domains of Tc-cad and Tc-Eve, accompanied by the reduction of presumptive head regions, indicate that in the absence of Tc-axin function, posterior fates shifted at the expense of anterior fates.

Fig. 3.

Expression of Tc-zen, Tc-cad, and Tc-Eve in wild-type and Tc-axin RNAi embryos. (A–F) Brightfield images of blastoderm embryos (anterior to the left); (A′–F′) DAPI nuclear counterstained images of the same embryos. The axin RNAi image is shown immediately below its wild-type counterpart; gene is denoted in Lower Right. (A) In wild-type, Tc-zen1 expression marks the serosa, which forms an asymmetric, dorsally inclined boundary with the embryonic tissue. (B) After Tc-axin RNAi, Tc-zen1 expression was highly reduced and the border was no longer inclined. (C and D) Blastodermal Tc-cad expression was expanded anteriorly after Tc-axin RNAi, leaving only few embryonic cells unstained (compare bars in C′ with D′). (E and F) The anterior expression boundary of Tc-Eve (marked with arrowhead in E′) was shifted toward the anterior in Tc-axin RNAi embryos, often leaving only a few anterior embryonic cells unstained (arrowhead in F′). (Magnification: 45×.)

Analysis of Canonical Wnt Pathway Components.

Tc-axin RNAi embryos advanced normally through cleavage stages (Fig. 4, compare A and E), but at the differentiated blastoderm stage when serosa and embryonic cells are distinguishable, more nuclei were allocated to the embryonic rudiment at the expense of serosal tissue compared with wild-type (Fig. 4 B and F). During normal development, the embryonic germ rudiment forms on the ventral side of the egg and elongates over the poles as segments are added (Fig. 4 C and D). In Tc-axin RNAi embryos, the embryonic cells contracted toward the posterior pole of the egg, and loss of the serosal membrane left most of the anterior yolk exposed (Fig. 4 G and H). Similar effects are observed after Tc-shaggy/GSK-3 RNAi (Fig. S3), suggesting the effect is mediated through the canonical Wnt pathway.

Fig. 4.

Early development in wild-type and RNAi mutant embyros. DAPI-stained embryos, anterior to the left, dorsal up (if known). Early cleavage, differentiated blastoderm, early germ-band, and a later stage (equivalent to germ-band elongation in wild-type) are shown from left to right for each RNAi experiment. The total number of embryos examined and the number in each class are also shown: (A–D) wild-type, (E–H) Tc-axin RNAi, (I–L) Tc-pan/Tc-axin double RNAi, and (M–P) Tc-pan RNAi. (Magnification: 35×.)

Both axin and shaggy are members of the β-catenin destruction complex, which functions to inhibit the Wnt pathway in the absence of Wnt ligand (14). When the destruction complex is defective, β-catenin is stabilized and free to move into the nucleus, where it binds to TCF/Pangolin protein to activate Wnt targets. In the canonical Wnt pathway, Pangolin (Pan) represses transcription when not bound by β-catenin, but it is also required for activation when β-catenin is present (see Fig. 6). Depletion of pan transcripts derepresses Wnt pathway targets but at the same time does not allow full activation of them, producing mild wingless-like phenotypes (15). Tc-pan RNAi produced a range of cuticular embryonic phenotypes, including loss of distal appendages and posterior truncations (Fig. 2 F–H) that were not as severe as after depletion of Wnt ligands (6). Because loss of Tc-axin produces the opposite effect (loss of anterior fates), we performed double RNAi with Tc-axin and Tc-pan. The resulting embryos displayed the same range of cuticular phenotypes as Tc-pan RNAi embryos (Fig. 2 I–K), indicating that Tc-pan functions epistatic to, or downstream of, Tc-axin. We also examined early developmental stages in Tc-pan/Tc-axin double RNAi embryos (Fig. 4 I–L). These embryos developed similar to Tc-pan single RNAi embryos (Fig. 4 M–P), displaying transiently larger heads that sometimes cover the anterior pole of the egg (Fig. 4 K and O). During later development, posterior truncation and loss of distal appendage segments were observed (Fig. 4 L and P). Taken together, these experiments indicated that the Tc-axin acts via canonical Wnt signaling. Furthermore, the Tc-axin RNAi phenotype is not rescued by RNAi for Tc-arrow (Fig. S3), a Wnt receptor, indicating that Wnt ligand is not required for the effect. Thus, interfering with components downstream in the pathway perpetuates the effects.

Fig. 6.

Influence of Tc-axin and Tc-pangolin on canonical Wnt signaling and role of Wnt signaling in axis formation in the bilateria. (A) In the absence of Wnt ligands, the destruction complex marks β-catenin/Armadillo protein for degradation. TCF/Pangolin is bound to enhancers of Wnt-responsive genes and represses transcription by recruiting corepressors. (B) After binding of the Wnt ligand to the receptor, the destruction complex is disassembled leading to transport of β-catenin/Armadillo to the nucleus, where it activates Wnt targets by binding to TCF/pangolin. (C) Loss of TCF/pangolin leads to derepression of Wnt target genes. This effect is independent of β-catenin and other upstream factors (shaded in gray). (D) In axin RNAi, the destruction complex cannot form. Hence, canonical Wnt targets are activated by β-catenin/Armadillo binding to TCF/Pangolin; this is independent of upstream events (shaded in gray). (E) Throughout bilateria and their sister group, the hydrozoa, posterior activity, and anterior repression of canonical Wnt signaling have been shown to be essential for axis formation. Based on Drosophila data, insects have been regarded as an exception because Wnt activity is not required for axis formation. Previous data from Tribolium showed that it is required for posterior elongation. Here, we show that canonical Wnt signaling needs to be repressed for anterior patterning. Y, functional or expression data indicating involvement of the Wnt pathway; N, data not supporting involvement of Wnt pathway; ?, no data. Information in E is based on refs. 1 and 30 and references therein.

Maternal Contribution Is Required for Normal Early Development.

From the GEKU insertional mutagenesis screen (16) we identified a homozygous lethal mutant (E-12614) with an insertion in the first intron of Tc-axin. We confirmed that the first Tc-axin exon is joined to a splice acceptor within the mutator, resulting in a nonfunctional transcript. In matings of heterozygotes, 84.6% of the offspring hatched and produced wild-type cuticles (n = 188, sum of two replicates). Most of the unhatched eggs did not contain cuticles (10.1%), suggesting strong effects on embryogenesis. The remaining 5.5% did not display phenotypes like the ones described above for Tc-axin RNAi and probably represented background defects in the strain. We did not detect changes in Tc-axin expression in the mutant, suggesting that the insertion does not interfere with transcriptional regulation. Mutant offspring lack the zygotic contribution of Tc-axin, but retain a maternal contribution from the expression of a single wild-type maternal allele. In contrast, parental RNAi knocks down both maternal and zygotic contributions. Therefore, the comparison of the mutant and RNAi phenotypes allowed testing the relevance of the maternal contribution. Anaylsis of Tc-wg, Tc-cad, and Tc-zen in offspring of heterozygous parents, did not reveal altered blastodermal morphology or defects in young germ bands (the gnathal and first thoracic Tc-wg stripes formed normally (Fig. 5 A and B) (n = 19). Irregularities were observed in trunk Tc-wg stripes that formed thereafter (Fig. 5, compare D with wild-type sibling C). Older embryos displayed severely disturbed Tc-wg patterns (Fig. 5 E–I). Taken together, these experiments suggest that the maternal contribution of Tc-axin is required for proper development through the blastoderm stage.

Fig. 5.

Expression of Tc-wg in progeny of Tc-axin mutant animals revealed the onset of Tc-Axin depletion at the early germ-band stage when four Tc-wg stripes had formed. We did not detect altered patterns in blastoderm stage embryos (not shown). (A and B) No changes were detected in young germ bands up to a stage of three Tc-wg stripes in the trunk, although the genotype of the respective embryos cannot be confirmed. (C) Sixty-five percent of embryos containing four Tc-wg stripes were clearly wild-type (n = 20). (D) Four embryos were neither clearly wild-type nor mutant (20%). (E) Three embryos of a similar stage (15%) were clearly affected. (F–H) Clearly aberrant Tc-wg patterns were found at all subsequent stages, the severity of which increased with developmental time. (Magnification: 35×.)

Discussion

Analysis of genes expressed during the blastoderm stage revealed that interference with Tc-axin function results in an anterior shift of the blastodermal fate map. The similar effects of Tc-axin and Tc-shaggy RNAi, combined with the epistatic relationship of Tc-pan to Tc-axin, strongly implicate the canonical Wnt pathway in mediating the effects of Tc-axin RNAi. Furthermore, the expression domain of Tc-cad, a target of the canonical Wnt pathway in Gryllus (17) and Tribolium (Fig. S4), extends more anteriorly in Tc-axin RNAi embryos, suggesting increased levels of Wnt activity there.

Thus, anteriorly localized Tc-axin appears to be playing an important role in regulating canonical Wnt signaling. However, Wnt ligands are not expressed in freshly laid eggs; the first Wnt ligands to be expressed are found in the head lobes and at the posterior pole at the blastoderm stage, and depletion of Wnt ligands does not affect segmentation in the blastoderm (6). Thus, any Wnt pathway activity in the preblastoderm embryo must be ligand-independent. The expression of Tc-axin, maternally localized at the anterior pole, suggests a mechanism to produce graded levels of Wnt activity during early embryognesis. The early asymmetry in Tc-axin expression would inhibit Wnt activity anteriorly, producing a gradient with highest levels in the posterior region. Later, Wnt ligands expressed at the posterior pole maintain active Wnt-signaling there as Tc-axin becomes ubiquitous in the embryo. Despite apparent lack of Wnt function in Drosophila axis formation, a wg domain in the posterior blastoderm has recently been reconsidered as a legacy of such a function (18).

Axin and Shaggy are components of the Armadillo/β-catenin destruction complex. Localization of Tc-axin at the anterior pole suggests higher activity of the complex there, which would result in lower levels of β-catenin. There are two β-catenin homologs in Tribolium that function during the cleavage stages (19), obscuring the ability to test what roles, if any, they might play in determining the blastoderm fate map.

An additional level of Tc-cad regulation required for head development in Tribolium involves translational repression of Tc-cad activity by Tc-mex3 in the differentiated blastoderm (20). The earlier asymmetry of Tc-axin, combined with the more extensive phenotypes induced by Tc-axin RNAi, suggest that Wnt signaling acts upstream of Tc-mex3 or that both mechanisms cooperate to shape the anterior Tc-cad gradient. Importantly, neither system appears to function in Drosophila although they are clearly conserved in evolution; the Mex3/Pal1-Cad system is also found in Caenorhabditis elegans (21, 22), and the posteriorizing effect of Wnt signals is known from the vertebrate neural plate and planarian regeneration (23). Thus, in contrast to Drosophila, Tribolium has retained a more ancestral mode of AP axis formation.

In vertebrates, Wnt antagonists, such as DKK, are expressed anteriorly to repress Wnt signaling in the head (24). In insects, such antagonizing factors have yet to be found, and repression of Wnt signaling has not been implicated in anterior development. In contrast, Drosophila bicoid (bcd) mRNA is maternally localized to the anterior pole of the egg. Upon fertilization, the resulting protein gradient provides positional information in a concentration-dependent manner (25). However, bcd appears to be restricted to Dipterans and is not found in other insects, including Tribolium (26). It has been proposed that orthologs of gap genes including orthologs of Orthodenticle, a more conserved homeodomain protein with similar target specificity, provide anterior morphogen function in Tribolium and other short-germ insects (27, 28). However, Tc-orthodenticle1(Tc-otd1) is ubiquitously expressed in preblastoderm stages in Tribolium (27) and Tc-mex3 expression arises zygotically in the blastoderm (20), leaving in question the earliest asymmetric signal in the embryo. Our data indicate that maternal anterior localization of Tc-axin mRNA (similar to bicoid in Drosophila) is one of the first asymmetric signals in the Tribolium embryo. Furthermore, we show that inhibition of Wnt signaling is required for embryonic head development in arthropods, as it is in vertebrate embryogenesis and in planarian regeneration (summarized in Fig. 6E), supporting the hypothesis that Wnt signaling is an integral part of an ancestral metazoan mechanism that polarizes and patterns the AP axis early in development. Finally, anterior localization of β-catenin destruction complex activity is a previously undescribed mechanism by which to confer positional information.

Materials and Methods

Strains.

Ga-1 (KSU) and San Bernadino (GAU) strains of Tribolium castaneum were used for gene expression and functional analysis. Beetles were reared at 30 °C in whole-wheat flour supplemented with 5% dried yeast. Two independent inverse PCR reactions (16) revealed that in the E-12614 axin mutant, the insertion of the mutator construct (accession no. HE654700) is located in the first intron 5,761-bp downstream of the first exon of Tc-axin. By RT-PCR, we confirmed that the artificial splice acceptor site included in the mutator construct (a pangolin exon with artificially introduced stop codons in all frames) was spliced to the first exon of the Tc-axin gene in the mutant. This transcript is unlikely to produce functional Tc-Axin protein.

RNAi and in Situ Hybridization.

Double-stranded RNA (ds RNA) was synthesized using the T7 megascript kit (Ambion) and purified using the Megaclear kit (Ambion). dsRNA was mixed with injection buffer (5 mM KCl, 0.1 mM KPO₄, pH 6.8) before injection. Adult females were injected and embryos were analyzed as previously described (10). Eggs were collected and either fixed immediately for in situ hybridization or allowed to develop at 30 °C for several days before cuticle preparation. Off-target effects were excluded by injection of two nonoverlapping dsRNA fragments (base pairs 448–1091 and 1,168–1,617 of Tc-axin cds, respectively), which elicited the same phenotype as the injection of the long dsRNA fragments; 500- to 1,000-bp fragments were used as templates for digoxigenin-labeled riboprobes. In situ hybridization was performed as previously described (29).

Microscopy and Imaging.

Cuticles and stained embryos and were viewed with a Nikon Digital Camera DXM 1200F camera on an Olympus BX50 microscope and photographed using Nikon ACT-1 Version 2.62 software. Embryos were optically sectioned and images were assembled using MONTAGE software. Brightness and contrast were adjusted using Adobe Photoshop CS2 software.