tarsal-less is expressed as a gap gene but has no gap gene phenotype in the moth midge Clogmia albipunctata

Gap genes are involved in segment determination during early development of the vinegar fly Drosophila melanogaster and other dipteran insects (flies, midges and mosquitoes). They are expressed in overlapping domains along the antero-posterior (A–P) axis of the blastoderm embryo. While gap domains cover the entire length of the A–P axis in Drosophila, there is a region in the blastoderm of the moth midge Clogmia albipunctata, which lacks canonical gap gene expression. Is a non-canonical gap gene functioning in this area? Here, we characterize tarsal-less (tal) in C. albipunctata. The homologue of tal in the flour beetle Tribolium castaneum (called milles-pattes, mlpt) is a bona fide gap gene. We find that Ca-tal is expressed in the region previously reported as lacking gap gene expression. Using RNA interference, we study the interaction of Ca-tal with gap genes. We show that Ca-tal is regulated by gap genes, but only has a very subtle effect on tailless (Ca-tll), while not affecting other gap genes at all. Moreover, cuticle phenotypes of Ca-tal depleted embryos do not show any gap phenotype. We conclude that Ca-tal is expressed and regulated like a gap gene, but does not function as a gap gene in C. albipunctata.


Introduction
The gap gene network provides the first layer of zygotic regulation in the segmentation gene hierarchy of dipteran insects (flies, midges and mosquitos). (Kr), knirps (kni) and giant (gt), with additional inputs from the terminal gap genes tailless (tll) and huckebein (hkb) [1]. In other cyclorrhaphan flies, such as the hoverfly Episyrphus balteatus [2,3] and the scuttle fly Megaselia abdita [4][5][6], gap gene expression and regulation is strongly conserved. It leads to a set of virtually identical expression domains, comprising overlapping regions of blastoderm nuclei/cells, at the onset of gastrulation. Outside the cyclorrhaphan clade, among the nematoceran Diptera, there is little functional evidence on gap gene regulation although expression patterns have been described in the malaria mosquito Anopheles gambiae [7].
Here, we focus on another emerging nematoceran model system, the moth midge Clogmia albipunctata (Diptera, Psychodidae). In this species, we have a detailed description of the spatial arrangement [8,9] as well as the temporal dynamics [10] of gap gene expression. This descriptive evidence reveals a region of the C. albipunctata blastoderm embryo which is not covered by expression of any gap gene known from D. melanogaster [9]. This region lies between the abdominal domain of the C. albipunctata homologue of kni and knirps-related (called knirps-like, knl) and the posterior terminal domain of tll [9,10]. It suggests that we may be missing a posterior gap gene in this species.
One candidate for this missing gap gene in C. albipunctata is tarsal-less (tal) [11], also called polished rice (pri) [12]. tal/pri is a polycistronic gene encoding a long primary transcript from which several short peptides are produced that are required in different stages of embryonic development. It is part of a large class of polycistronic genes with small open reading frames (sORF/smORF), small encoded peptides or microproteins that play a wide range of roles in physiology, development and cell differentiation [13,14]. In D. melanogaster, tal/pri is first expressed in a stripe-like expression pattern at the late blastoderm stage [12] (by expression stripe, we mean a narrow expression domain, only a few nuclei wide). It is involved in epithelial morphogenesis and leg development [11,12,[15][16][17][18], but has no role in early embryonic patterning or segment determination.
Interestingly, a homologue of tal/pri was first described in the flour beetle Tribolium castaneum under the name of mille-pattes (mlpt) [19]. By contrast to tal/pri in D. melanogaster, mlpt in T. castaneum has a segmentation function acting as a bona fide gap gene [19]. mlpt is expressed in a gap-like fashion, with an anterior and a posterior terminal domain at blastoderm stage; subsequently, the anterior domain resolves into two stripes, and the terminal domain retracts from the pole and shifts anteriorly over time during germband extension; a third posterior domain appears at this stage; finally, mlpt is expressed in the peripheral nervous system and the forming appendage joints at later stages of development, which is similar to its expression pattern in D. melanogaster [19]. Knock-down of mlpt in T. castaneum by RNA interference (RNAi) leads to a gap-like phenotype with missing abdominal segments [19]. mlpt regulates trunk gap genes hb, Kr and gt, and is itself regulated by hb and Kr [19].
Here, we characterize expression of tal/pri in C. albipunctata, and examine its interactions with other segmentation genes using RNAi knock-down assays. We show that it exhibits a gap-gene-like expression pattern at the blastoderm stage. As in T. castaneum, it is expressed in an anterior and a posterior terminal domain, which later split into narrow stripes. In contrast to T. castaneum, however, tal/pri does not regulate gap genes in C. albipunctata, with the possible exception of its interaction with the posterior terminal tll domain. Even though it is regulated by gap genes hb, Kr and knl, it does not exhibit any gap-like phenotype when knocked down. This evidence suggests that although tal/pri is expressed and regulated in a gap-gene-like manner, it cannot be classified as a bona fide gap gene in C. albipunctata.

Results and discussion
2.1. Characterization of tarsal-less in the moth midge C. albipunctata We searched the early embryonic transcriptome of C. albipunctata [20] for a tal homologue using the D. melanogaster amino acid sequences for the small peptides encoded by tal. Our search identified a 2277 nt fragment that contained several short peptide repeats, probably corresponding to a primary transcript. Upon in silico translation, it was confirmed as a homologue of tal/pri/mlpt in C. albipunctata. We will call this fragment Ca-tal. Specific primers were generated to clone the gene from cDNA, and empirically confirm its sequence (see Material and methods). The Ca-tal sequence has been deposited in GenBank under accession number MG783326.
The polycistronic sequence of Ca-tal shows general structural similarities to tal genes in other organisms (figure 1a). tal genes exhibit variable numbers of repeats of N-terminal peptides containing a consensus region of LDPTGXY, and one C-terminal peptide with the consensus domain GREETSSCRRRR [19]. In Ca-tal, we find four short repeated N-terminal peptides of 11, 10, 11 and 29 amino acids separated by

Temporal expression profile of Ca-tal in the embryo
We have characterized the expression pattern of Ca-tal in the embryo of C. albipunctata from the blastoderm up to the extended germband stage [21] using enzymatic (colorimetric) in situ hybridization (ISH) (figure 2). The earliest pattern we detect is a posterior expression domain in the trunk region of the blastoderm embryo, covering 65 -80% antero-posterior (A -P) position (figure 2a). This domain shifts anteriorly over time (figure 2b). By the time it has reached 55 -75% A -P position, a second terminal domain becomes apparent at the posterior pole (figure 2c). Both domains continue to shift and expand anteriorly (figure 2d), consistent with shifts observed for posterior gap genes during the blastoderm stage [9,10]. Before gastrulation, the anterior border of the more anterior Ca-tal domain reaches 55% A-P position (figure 2d, arrowhead), and this domain starts to split into two stripes (figure 2d, asterisks). By the same time, the posterior terminal domain has expanded to 85% A-P position.
By the onset of gastrulation, the anterior domain has resolved completely into two stripes (figure 2e, asterisks). A weak third stripe appears shortly thereafter in a more anterior position (figure 2g, asterisk). This dynamic pattern is similar to what has been reported for the tal homologue mille-pattes (mlpt) in the flour beetle T. castaneum [19]. The terminal domain follows the morphogenetic movement of the posterior pole region during gastrulation [21], moving to the dorsal side of the embryo (figure 2f-h); at the same time, this domain clears from the pole (figure 2h, arrowhead) and divides into two sub-terminal stripes (figure 2i, arrowheads). During germband elongation, the first, and later the third, stripe of the anterior domain fade away (figure 2i,j, asterisks). Finally, an additional stripe appears anterior of the two sub-terminal stripes (figure 2j, arrowhead). Our results show that Ca-tal is expressed in a gap-gene-like manner during the blastoderm stage, partially overlapping with previously characterized gap domains in C. albipunctata [9]. Intriguingly, the terminal Ca-tal domain covers a region of the C. albipunctata blastoderm-between the abdominal Ca-knl domain and the terminal domain of Ca-tll-in which no gap gene expression has been detected before [9,10]. In contrast, tal is not expressed like a gap gene in D. melanogaster, where its transcripts appear directly in a stripe-like pattern during the late blastoderm stage [12]. Early Ca-tal expression shows much more resemblance to that of its homologue mlpt in T. castaneum, which acts as a bona fide gap gene in that species [19]. This suggests that Ca-tal may also play the role of a gap gene in C. albipunctata. In order to test this possibility, we performed knock-down by RNAi of Ca-tal, Ca-tll, and other trunk gap genes.

Ca-tal does not regulate, but is regulated by trunk gap genes
To assess the effect of Ca-tal on the gap genes in C. albipunctata, we performed RNAi knock-down against Ca-tal following a previously published protocol [22]. The resulting tal-depleted embryos were stained by colorimetric ISH for trunk gap genes Ca-hb, Ca-Kr, Ca-gt and Ca-knl, as well as the terminal gap gene Ca-tll. The other terminal gap gene, huckebein (hkb), is not expressed at the relevant stages in C. albipunctata [9]. We do not observe any clearly detectable differences in the expression patterns of the trunk gap genes in Ca-tal knock-down embryos (electronic supplementary material, figure S1). Quantitative assessment of domain boundary positions using our FlyGUI/FlyAGE image-processing pipeline [10,23] does not reveal any significant differences to the wild-type either (not shown). The only potential effect of Ca-tal on gap genes is the reduced expression in the posterior terminal domain of Ca-tll in a small percentage of Ca-tal knock-down embryos (4 out of 17; electronic supplementary material, figure S1e,f,k,l ). Target genes further downstream in the segmentation gene cascade, such as the pair-rule gene even-skipped (eve), and the segment polarity genes wingless (wg) and engrailed (en), also fail to show any clearly detectable defects in Ca-tal knock-down embryos (not shown). This suggests that Ca-tal does not play any essential role in segmentation gene regulation in C. albipunctata. Next, we investigated whether Ca-tal is regulated by gap genes. We assayed Ca-tal expression in embryos treated with RNAi against Ca-hb, Ca-Kr, Ca-gt, Ca-knl and Ca-tll using colorimetric ISH ( figure 4). The expression pattern of Ca-tal was affected by all gap genes with the exception of Ca-gt (not shown). In blastoderm embryos depleted of Ca-hb, the more anterior domain of Ca-tal is displaced anteriorly, extending past 45% A-P position (figure 4c; 10 out of 28 embryos). This suggests that Ca-hb positions the anterior border of expression of Ca-tal through repression. Alternatively, this repression could be indirect, mediated through repression of the activator encoded by Ca-Kr in this region (see below). In embryos depleted of Ca-Kr, we observe a loss of the more anterior Ca-tal domain, while its terminal domain appears to expand anteriorly (figure 4f, 17 out of 20 embryos). This is consistent with a dual influence of Ca-Kr, with an activating effect on the more anterior domain, and repression on the terminal domain of Ca-tal. However, it is not clear whether both of these effects are direct. Activation could be mediated through repression of repressor Ca-knl by Ca-Kr. This is unlikely, as Ca-knl is not affected in knock-downs of Ca-Kr (electronic supplementary material, figure S2, 17 out of 17 embryos). Still, we cannot exclude indirect activation mediated through repression of another unknown repressor. Finally, the effect of Ca-Kr could be interpreted as a deletion of the region between the two Ca-tal domains. This, however, seems unlikely, because Ca-Kr is not expressed near the potentially affected region of the embryo (cf. Figure 3b) and Ca-knl is still expressed there in Ca-Kr RNAi-treated embryos (electronic supplementary figure S2b). In embryos depleted of Ca-knl, we see strong ectopic expression of Ca-tal between its two domains of expression at the blastoderm stage (figure 4i, 25 out of 54 embryos). This suggests repression of Ca-tal by Ca-knl. The effect is probably weak, because the de-repression seen in figure 4i is incomplete, and the expression patterns of Ca-tal and Ca-knl show extensive overlap in the wild-type (figure 3d ). Just as in the case of Ca-hb knock-downs discussed above, this effect could be indirect, mediated through Ca-Kr. In late blastoderm embryos depleted of Ca-tll, the terminal domain of Ca-tal expression is either completely abolished or strongly reduced (figure 4l, 17 out of 44 embryos). Taken together, our evidence suggests that Ca-tal and Ca-tll activate each other in C. albipunctata.

Ca-tal and Ca-gt do not exhibit gap gene phenotypes in C. albipunctata
To further examine the function of Ca-tal and the gap genes in C. albipunctata, we obtained cuticle preparations of late-stage wild-type and RNAi embryos according to a previously published protocol ( figure 5) [22]. In cuticles of embryos treated with RNAi against Ca-hb (n ¼ 22), we observed a reduction in the number of thoracic segments in all specimens: seven embryos showed no, nine  embryos one and six embryos two remaining thoracic segments (figure 5b). We only managed to obtain two cuticles of embryos treated with RNAi against Ca-Kr. Both of them exhibit general A-P polarity, but no thoracic or abdominal segments are discernible (figure 5c). Similarly, severe defects were observed in the two cuticles we obtained from embryos treated with RNAi against Ca-knl: these embryos show two recognizable thoracic and one or two abdominal segments, albeit with severe dorsal defects, as well as an abnormal posterior terminal region (figure 5d). Cuticles of embryos treated with RNAi against Ca-tll (n ¼ 15) show a much less penetrant phenotype. In four individuals, the telson is missing, and two show a severe reduction of the number of abdominal segments (figure 5e). Only one specimen exhibited defects in the head and the thoracic region (not shown). We could detect no gap gene phenotypes or other obvious and consistent segmentation defects in embryos depleted for Ca-tal and Ca-gt (figure 5f,g). However, in 2 out of 39 of the gt and 5 out of 55 of the tal depleted cuticles we observe small hemilateral abnormalities (electronic supplementary material, figure S3 A, B, asterisks). We cannot rule out a weak effect of RNA depletion, but the cause of these abnormalities could also be mechanical or unspecific. We do not observe this type of effect in the other RNAi injected cuticles. The evidence from our cuticle preparations suggest that hb, Kr, kni/knl and tll have conserved roles as T1 T2   T1 T2   T1 T2   T1  T2  T3   T1  T2 T3 A1 A2 A3 A4 A5 A6 A7 A8   A7  A6  A5  A4  A3  A2  A1  T3  T2  T1  gap genes in C. albipunctata, while gt and tal are expressed in a gap-like manner (tal also being regulated by other gap genes) but do not play a classical gap-like role in trunk segment determination in this species.

Conclusion
We have characterized the homologue of tal/pri/mlpt in the nematoceran moth midge C. albipunctata. Similar to its homologues in other organisms [11,19], it produces a polycistronic primary transcript, which codes for several short peptides. We have shown that Ca-tal is expressed in a gap-gene-like manner in C. albipunctata, unlike in D. melanogaster where it initiates transcription in refined stripes during the late blastoderm stage [12]. We show that these early stages of expression are regulated by gap genes in C. albipunctata. Later expression patterns are more conserved between the two species. Despite its suggestive early embryonic expression pattern, Ca-tal cannot be classified as a segmentation gene. Our evidence reveals that Ca-tal is not regulating other segmentation genes, and does not cause a gap-like or any other segmentation phenotype upon knock-down by RNAi. The gap-like expression pattern of Ca-tal shows striking similarities to its homologue, the gap gene mlpt in the flour beetle T. castaneum. However, even this similarity may be superficial, as there are significant differences between the regulation of both homologues. The anterior domain of Ca-tal is repressed by Hb, while the posterior terminal domain is not affected in hb RNAi knock-downs (summarized in figure 6). In T. castaneum, the opposite is true: while the anterior domain of mlpt is not affected, the posterior domain forms late in if hb is depleted [19]. Furthermore, Ca-tal is repressed by knl (figure 6), while kni does not affect mlpt expression in T. castaneum [19]. In contrast, mlpt is activated by gt [19], while Ca-tal and gt show no genetic interaction. The only similarity between the two species is the role of Kr in tal/mlpt regulation: ectopic expression is seen upon Kr knock-down in the posterior of blastoderm embryos in C. albipunctata and T. castaneum. Based on the available evidence, it remains unclear whether the early gap-like expression pattern of tal/mlpt is an ancestral feature of segmentation patterning, or whether it has evolved convergently in beetles and nematoceran dipterans. Functional data from other basally branching dipteran lineages or suitable outgroups will be required to resolve this outstanding question.