Multiple roles of the polycistronic gene tarsaless/mille-pattes/polished-rice during embryogenesis of the kissing bug Rhodnius prolixus

Genes encoding small open-reading frames (smORFs) have been characterized as essential players of developmental processes. The smORF tarsaless/mille-pattes/polished-rice has been thoroughly investigated in holometabolous insects, such as the fruit fly Drosophila melanogaster and the red flour beetle Tribolium castaneum, while its function in hemimetabolous insects remains unknown. Thus, we analyzed the function of the tal/pri/mlpt ortholog in a hemimetabolous insect, the kissing bug Rhodnius prolixus (Rp). First, sequence analysis shows that Rp-tal/pri/mlpt polycistronic mRNA encodes two small peptides (11 to 14 amino acids) containing a LDPTG motif. Interestingly, a new hemipteran-specific conserved peptide of approximately 80 amino acids was also identified by in silico analysis. In silico docking analysis supports the high-affinity binding of the small LDPTG peptides to the transcription factor Shavenbaby. Rp-tal/pri/mlpt in situ hybridization and knockdown via RNA interference showed a conserved role of Rp-tal/pri/mlpt during embryogenesis, with a major role in the regulation of thoracic versus abdominal segmentation, leg development and head formation. Altogether, our study shows that tal/pri/mlpt segmentation role is conserved in the common ancestor of Paraneoptera and suggests that polycistronic genes might generate order specific smORFs.


Introduction:
A large number of essential genes required for biological processes have been discovered by genetic screenings in model organisms such as the fruit fly Drosophila melanogaster (e.g. [1]). While most loci important for developmental processes were identified in these original screenings, recent genetic and expression analyses in D.
melanogaster and in the beetle Tribolium castaneum showed that genes previously classified as putative noncoding RNAs encode functional small open reading frames (smORFs) or sORFs [2; 3]. smORFs, ORFs smaller than 100 amino acids, have been described as fundamental for several developmental processes of insects [4; 5; 6].
Although functional smORFs have been originally described in yeasts [7], gene prediction methods have, in general, discarded smORFs in genome-wide predictions (reviewed in [8]). Comparative genomic analysis of Drosophilid species showed an unexpected conservation of smORF containing genes, suggesting important biological roles for this new class of genes [9].
Later on, experimental analysis of conserved smORFs such as sarcolamban (scl), a conserved peptide involved in Ca 2+ uptake at the sarco-endoplasmic reticulum [10], and hemotin [11], a conserved phagocytosis regulator, provided further evidence that genes containing smORFs might constitute a reservoir of important players in metazoan genomes. In the past years, direct evidence of large-scale smORF translation has been obtained by ribosomal profiling of polysomal fractions in Drosophila, using the Poly Ribo-Seq technique [12]. This study was able to classify smORFs in two groups: "longer" smORFs of around 80 amino acids resembling canonical proteins, mostly containing transmembrane motifs, and shorter ("dwarf") smORFs. These 'dwarf' smORFs are in general shorter (around 20 amino-acid long), less conserved and mostly found in 5'-UTRs and non-coding RNAs.
While bioinformatic studies point to hundreds or thousands of genes containing putative smORFs, only a few functional studies have been performed. In insects, the function of the smORF founding member mlpt was only investigated in detail in holometabolous insects such as the D. melanogaster and T. castaneum [4; 5; 6] and more recently in basally branching Diptera [13]and Lepidoptera [14]. In T. castaneum mlpt acts as a gap gene during the process of embryonic segmentation, regulating Hox genes and thoracic versus abdominal specification; knockdown for mlpt leads to embryos with multiple legs, mille-pattes in French [6]. In D. melanogaster tal was identified in a spontaneous mutant with defects in the distal part of the legs, the tarsus [5].
Independently, in the same year, Pri peptides were shown to be required for the F-actin organization during trichome morphogenesis; embryos lacking pri show external cuticle defects, resembling polished rice [4]. More recently, Mlpt peptides were shown to be regulated by ecdysone, defining the onset of epidermal trichome development, through post-translational control of the Shavenbaby (Svb) transcription factor [15]. Altogether, Mlpt peptides display context-specific roles and interactions within different developmental processes.
While in D. melanogaster mlpt early embryonic gene expression is segmental displaying a pair-rule pattern, mutants do not display segmentation or homeotic alterations as reported in T. castaneum. D. melanogaster embryonic mutant phenotypes include broken trachea, loss of cephalopharyngeal skeleton, abnormal posterior spiracles and lack of denticle belts, structures of late embryonic mlpt expression [5].
To clarify whether the segmentation function of mlpt is ancestral, but has been lost in the lineage giving rise to D. melanogaster, or whether it is a recently arisen specialization of T. castaneum, it is important to study mlpt function in other insect groups. Since hemipterans, as hemimetabolous, comprise the sister group of holometabolous insects (reviewed in [16]), a functional characterization of mlpt in the emergent hemiptera model, the kissing bug Rhodnius prolixus would be important [17].
R. prolixus available genomic and transcriptomic resources [18; 19; 20; 21; 22], an established embryonic staging system [23] and the recent availability of in situ hybridization and RNA interference techniques are great advantages of this model system [17].
Here we report interesting conserved and new functional aspects of the prototypic smORF mlpt gene. First, sequence analysis identified a new hemiptera-specific peptide in the polycistronic mRNA of mlpt, a gene conserved throughout the Pancrustacean clade.
Second, molecular docking analysis indicates that the small peptide containing a LDPTG(L/Q/T)Y consensus motif interacts with the N-terminus of the transcription factor Svb as in D. melanogaster. Third, expression and functional analysis shows that the ortholog of mlpt acts during embryonic segmentation, regulating the transition between thoracic and abdominal identity, similarly to its role previously described in T.
castaneum. Fourth, a conserved role in tarsal patterning was also observed. Overall, segmentation and tarsal patterning are conserved roles of mlpt and provide evidence that generation of new peptides from smORFs might constitute an underestimated mechanism for the evolution of new genes.

Results
The recent description of several biologically important genes encoding smORFs among metazoans opened new avenues for molecular biology and functional genomics research. mlpt function has been largely investigated in holometabolous insects, while its function in hemimetabolous insects has not been fully investigated. In the current manuscript we describe a bioinformatic and functional analysis of mlpt function in the kissing bug R. prolixus. mlpt polycistronic gene and peptide distribution among arthropods Previous BLAST searches for genes encoding mlpt peptides against different arthropod genomes and transcriptomes provided evidence that this gene containing smORFs is restricted to insects and crustaceans [5]. The increase in arthropod genome sequences in the past years e.g. [24; 25; 26] allowed us to perform a complete search in the available arthropod genomes. Using a non-stringent BLAST approach (see methods for details) we identified mlpt orthologs in available insect genomes (Sup. Table 1). These orthologs encode two to several copies of a peptide of 11-32 amino acids, containing a LDPTG(L/Q/T)Y consensus motif (Figure 1 A -red boxes, Figure 1B). These peptides have been previously shown to mediate the switch of the svb transcription factor from a repressor to an activator [27] via the ubiquitin-conjugating complex, UbcD6-Ubr3, and proteasome recruitment [28]. Remarkably, hemiptera mlpt orthologs not only contain two smORFs encoding the LDPTG(L/Q/T)Y consensus motif ( Fig 1A-red boxes), but also a larger hemiptera-specific smORF of about 80 amino acids (Figure 1 A-green boxes and . This peptide appears to be smaller and less conserved in basally branching hemipteran species with circa 60 amino acids in the genome of the pea aphid Acyrthosiphon pisum and Pseudococcus longispinus and larger than 80 amino acids in triatomes such as Rhodnius prolixus and Triatoma pallidipennis. This new predicted peptide is potentially translated from a small CDS generated through post-transcriptional regulation, since it is interspersed by two introns in its respective gene in at least three hemipteran species. (Sup. Fig. 1). We named this new peptide as smHemiptera due to its restricted phylogenetic distribution. Interestingly, a polycistronic transcript containing all the aforementioned smORFs was identified in a digestive tract R. prolixus transcriptome ( [21] and our own observations).

Interaction mode prediction, hotspots and binding affinities of Shavenbaby-Mlpt complexes
Previous work in D. melanogaster identified the 31 N-terminal residues of the transcription factor (TF) svb as a Mlpt-dependent degradation signal, or degron [28]. In order to investigate the interaction modes between the peptides encoded by the polycistronic mRNA of mlpt and the transcription factor Shaven baby (Svb) from D. melanogaster (Dmel-Svb) and R. prolixus (Rpro-Svb) 3D structures were predicted and docking assays performed. Secondary structure analysis indicates that both Dmel-Svb and Rprol-Svb display mainly typical features of intrinsically disordered proteins, as previously suggested [28]. The disordered regions appear in large regions of both proteins, being interspersed by small alpha-helices and some very short beta-sheets ( Figure 2 A,B). Smaller peptides containing LDPTG(L/Q/T)Y motifs, Rprol-pptd1, Rprol-pptd2 and Dmel-pptd1 were considered disordered, as well as smHemiptera peptide ( Figure 1C).
In contrast, Dmel-pptd4, a peptide with 32 aa, and previously described as nontranslated in Drosophila cells [4; 5], showed two alpha-helices, intercalated by disordered regions (Figure 2 Table 2). According to the in silico data, the interaction between Dmel-pptd4 presented five salt bridges, seven hydrogen bonds and 127 non-bonded contacts. Its binding affinity parameters indicate a high affinity to the transcription factor Svb. The hotspots analysis, on amino acids located at the protein-protein complex interface and evaluated as the most critical for its maintenance, are listed on Supp. Table 2. The binding affinity change upon mutations indicate that, for the transcription factors (Dmel-Svb and Rprol-Svb), the destabilization of complexes tends to occur mainly through double mutations Supp. Fig. 1-2 In silico data did not detect high affinity interactions between smHemiptera and Rp-Svb N-terminal region, suggesting that this hemipteran specific peptide might show different molecular partners. Altogether, our molecular docking results corroborate previous experimental data [28] and suggest that similar amino acid residues in Svb and Mlpt are essential in both species for protein-peptide complex formation.

Rp-mlpt regulatory region shows putative binding sites for Ecdysone Receptor
Recently, mlpt expression was shown to be directly activated by the ecdysone pathway via direct binding of the nuclear receptor EcR to cis-regulatory sequences of mlpt [15]. We searched upstream, intronic and downstream sequences of D. melanogaster mlpt (Dm-mlpt), T. castaneum mlpt (Tc -mlpt) and R. prolixus mlpt (Rp-mlpt) for the occurrence of D. melanogaster Ecdysone Receptor binding sites ( Figure 3). As expected, one EcR binding site is observed in the regulatory region of Dm-mlpt; this region was previously shown to be directly bound by Ecdysone in ChIP-seq analysis of D.
melanogaster [15]. Searches in the T. castaneum mlpt locus with the same TF binding site showed four putative binding sites for EcR, upstream of the transcription start site (TSS) and one binding site several kilobases downstream, suggesting that a regulatory region responsive to ecdysone might reside upstream the TSS in this beetle. While the loci of D.
melanogaster and T. castaneum display similar genomic size, R. prolixus mlpt locus is much larger due to the large intronic regions which separate the three distinct coding regions (CDS) of the new hemipteran-specific peptide smHemiptera. Analysis of this locus in R. prolixus showed that six EcR binding sites occur particularly close to smHemiptera. This result suggests that ecdysone might also regulate mlpt expression during R. prolixus developmental transitions.

Rp-mlpt spatial expression pattern and relative expression during Rhodnius prolixus
embryogenesis mlpt embryonic expression was originally described in the beetle T. castaneum and has been characterized by a very dynamic zygotic patterning [6]. In D. melanogaster, mlpt expression was correlated with tissue folding, acting as a connection between patterning and morphogenesis [5]. Early expression of D. melanogaster mlpt starts as seven blastodermal stripes and a cluster of cells in the anterior part of the embryo, although segmentation is not affected in mlpt mutants. Later, after segmentation, mlpt is present in the trachea, posterior spiracles, pharynx, hindgut, and presumptive denticle belts. D. melanogaster mlpt mutants display reduced cuticular structures and ectopic expression of mlpt in the head induces extra skeleton components [5]. To investigate if levels of expression of mlpt also vary along embryonic development, we investigated its expression via Real-Time PCR. Small changes of Rpmlpt expression were observed between non-fertilized (0-6 hours NF) and early fertilized eggs (0-6 hours AEL), while a ten-fold increase in expression was detected by the analysis of 36-48 hours AEL eggs (Figure 5 A). Comparatively, expression of the transcription factor Rp-svb, a known interaction partner of mlpt in other organisms [27] increased its expression only 40% upon fertilization and decreased 55% at 36-48 hours AEL, when compared to non-fertilized eggs (0-6 hours NF) (Figure 5 B).

Rp-mlpt parental RNA interference is efficient and leads to a series of knockdown phenotypes
Since Rp-mlpt gene displays a complex expression patterning during embryogenesis, we sought to analyze its function by parental RNA interference (pRNAi) in R. prolixus, as previously described [23]. In this method, females are injected with double-stranded (dsRNA) synthesized against the gene of interest and phenotypic effects are evaluated in the offspring. Eggs collected from females injected with the control dsRNA (dsNeo) and the Rp-mlpt dsRNA showed a decrease up to 60-70% in Rp-mlpt expression in the latter dsRNA group, validating the knockdown (Figure 5 C). Increase in Rp-mlpt dsRNA injection concentration up to six micrograms per microliter (6µg/µl) did not further increase knockdown efficiency (data not shown). We also evaluated Rp-Notch expression, a receptor from a signaling pathway previously shown to interact with mlpt during leg formation [31], and embryonic segmentation in some insect species [32]. Rp-mlpt RNAi embryos shows that the distance between segmental stripes is reduced and that the distinction between thoracic and abdominal segments is less clear. Remarkably, in apparently stronger knockdown phenotypes, lack of the anterior-most segments of the head was also observed (Figure 10, asterisk). Altogether, Rp-hh expression analysis shows that Rp-mlpt is essential for proper thoracic and abdominal segmental distinction and for head formation.

Comparison of knockdown phenotypes of mlpt and other developmental genes
Previous functional analysis of mlpt in the T. castaneum and D. melanogaster showed that this gene is required for early patterning in the beetle but not in the fruit flies. suggesting that this leg corresponds to the L3 and that a duplication of L1 or L2 must have occurred. Interestingly, the expression of Ubx in the L3 does not occur as concentric ring as in the control, but rather in one side of the legs, suggesting that leg patterning was also affected (Figure 11). In addition, a slight expansion of Ubx towards posterior segments was observed in mlpt RNAi embryos ( Figure 11). Immunohistochemical analysis using a monoclonal antibody against the homeobox transcription factor ANTP showed a staining in the thoracic segments and the leg one and two (L1 and L2) in the controls. In contrast, segmental fusion and segmental duplication with corresponding changes in ANTP staining were observed in mlpt RNAi embryos. Altogether, these results provided evidence that mlpt is locally important for the distinction between thoracic and abdominal identity. Altogether, these results suggest that the Rp-mlpt role in R. prolixus is rather limited and mainly concentrated at the transition between the thoracic and abdominal region and that it has a minor role in head morphogenesis, while gap genes such as kr, gt, hb display a broader role in the segmental cascade of the hemiptera R. prolixus ( Figure 11). Previous studies and our own sequence analysis presented here showed that mlpt is not present in chelicerate and myriapod genomes, being restricted to Pancrustacea.

Discussion
Previous analyses identified specific roles for these smORF-encoded peptides Our docking and in silico mutation analysis identified the most important residues for the interaction of these small LDPTG peptides and the N-terminal region of the Svb protein, confirming previous experimental data in D. melanogaster [27; 28], and suggest a conserved interaction between small LDPTG peptides and Rp-Svb of R. prolixus.
The new hemipteran specific peptide identified here is approximately 35% longer in R. prolixus than in the most basally branching species analyzed, Pseudococcus longispinus (Figure 1 A, C). It has been recently proposed that smORFs could evolve to major ORFs via an "elongation" pattern [34]. Although smHemiptera is not a major ORF, size distribution of the smHemiptera peptide in hemiptera phylogeny corroborate this hypothesis. Interestingly, docking analyses of the larger hemiptera-specific predicted peptide (smHemiptera) were unable to detect specific molecular interactions between this peptide and the N-terminus of Svb, suggesting that this order-specific peptide might interact with other protein partners or other domains of Svb. Importantly, the pRNAi technique used in our study leads to the knockdown of the mature transcript and presumably of all predicted peptides encoded by the hemipteran mlpt gene. Cas9/CRISPR editing technology has been established in other non-model arthropod species [35] and its establishment in R. prolixus might help to unveil the small peptides' specific functions of mlpt in different developmental contexts.
Recent studies demonstrated that mlpt is required for Svb activation in adult tissues, particularly in nephric and intestinal D. melanogaster stem cells [36; 37]. Since mlpt is also expressed in R. prolixus digestive tract ( [21] and our own observations), it will also be interesting to address the functional role of this gene containing smORFs in other developmental contexts.

Conservation of mlpt expression among insects
In situ hybridization analysis of the Rp-mlpt gene during embryonic development shows a complex pattern of spatial expression, similarly to previously reported expression patterns in beetles, fruit flies and other holometabolous insects [5; 6; 13; 14; 33; 38]. Most differences in expression have been observed in early developmental stages. In D.
melanogaster, the expression occurs in seven blastodermal stripes, a pair-rule pattern, while in most other insect species, including the basally branching Diptera midge Clogmia albipunctata, the expression appears as a gap domain [13]. Thus, as observed for T. castaneum [6] and more recently in the hemiptera Oncopeltus fasciatus [33], Rpmlpt is first expressed in an anterior domain overlapping with the head and shortly after also in a posterior domain overlapping with the prospective thoracic segments (Figure 4).
Later on, during germ band elongation and dorsal closure, the dynamic expression of Rp-mlpt is observed in several tissues including the tips of the legs, the head and the antenna ( Figure 4). While expression in the head and in the legs have been reported in other species [5; 33], expression in the antennae at later stages has, to our knowledge, not been reported in other insect species.
Comparison of Rp-mlpt expression by RT-PCR between fertilized and nonfertilized eggs during the first hours of development (0-6 hours AEL) and 36-48 AEL shows ten times higher expression at later stages when compared to freshly laid eggs (0-6 hours AEL). In contrast, Rp-svb expression is higher during the initial stages of embryogenesis than in later stages, although its levels decrease by only 60% during these stages ( Figure 5). Altogether, Rp-mlpt displays changes in extensive relative expression among the analyzed developmental stages and is spatially very dynamic, while the relative expression of Rp-svb is less variable among the developmental stages. These results suggest that mlpt levels are mediated by transcriptional control, and that the translated Mlpt peptides might post-transcriptionally switch Svb from a repressor to an activator at 36-48 hours AEL, as previously reported for D. melanogaster [27; 28].
Recent data suggest that an ancient regulatory system involving an Mlpt/Ubr3/Svb module operates in several insect species and overlapping expression domains of mlpt and svb have been reported [33]. Finally, Rp-Notch expression is downregulated upon Rpmlpt RNAi ( Figure 5). In D. melanogaster, Notch and mlpt have been shown to genetically interact during tarsal joint development. Therein, the Notch signalling pathway activates mlpt expression in each presumptive joint region. A feedback loop involving Svb and the Notch-ligand Delta is also involved in mlpt regulation [31; 32]. Thus far, studies on the role of Notch during R. prolixus segmentation are lacking and data on other insects is contradictory [32; 39; 40]. Future studies in R. prolixus should address the role of the Notch pathway in segmentation and its relationship with the transcription factor Svb.

mlpt functions not as a classical gap gene in R. prolixus but rather participates in thoracic versus abdominal segmental identity
Three phenotypic classes were observed in mlpt knockdown embryos, mainly affecting segment formation (Figures 6 and 7). Comparison of the phenotypes of mlpt knockdown embryos with the phenotypes of transcription factor gap genes such as Kr, Hb and Gt ( Figure 11) showed clear phenotypic differences. mlpt knockdown embryos display localized phenotypic changes mainly at the transition between thoracic and abdominal segments, while knockdown of the aforementioned transcription factors shows larger effects and changes in molecular marker expression. In T. castaneum mlpt knockdown leads to extensive homeotic transformations and embryos show up to six leg pairs [6]. Recent data in other hemiptera showed that knockdown of three genes mlpt, Ubr3 and svb shows similar phenotypic effects as we describe here for R. prolixus such as posterior truncation, with the fusion/loss of thoracic segments, shortened legs and head appendages [33]. Phenotypes obtained by Rp-svb knockdown are similar to the Rp-mlpt RNAi embryos. Detailed analysis of the segment polarity gene Rp-hh showed that fusion takes place particularly at the thoracic regions and that abdominal segmentation and head formation is impaired in the strongest knockdown phenotypes (Figure 11). Analysis of the ventral midline gene Rp-single-minded (Rp-sim) showed that the ventral midline expression does not change upon Rp-mlpt knockdown, suggesting that dorsoventral patterning is not affected after mlpt RNAi (Sup. Figure 5).
Finally, several distal duplications of leg segments and leg malformations have been observed after Rp-mlpt RNAi (Figures 8 and 9). While several leg defects might be attributed to thoracic segment fusion, at least some examples of distal duplications are likely attributed to local effects of Rp-mlpt in the legs. We did not detect multiple rings of Rp-mlpt expression in R. prolixus; however, such rings were observed in Periplaneta but not in other hemipteran species [33; 41].

Evolutionary crossroads at mlpt regulation
Recent studies revealed that D. melanogaster mlpt is regulated by ecdysone through direct binding of the nuclear ecdysone receptor (EcR) to its cis-regulatory region [15]. Transcription mlpt and translation of small peptides containing LDPTG motifs are then able to switch the role of Svb from a repressor to an activator. Our molecular docking data provides evidence suggesting that the interaction between the small peptides and Svb might be conserved in R. prolixus. Whether ecdysone plays a role during R. prolixus embryonic development is unknown, but data from another hemipteran species, the milkweed bug Oncopeltus fasciatus, suggests that at least one early ecdysone-responsive gene, the transcription factor E75 is expressed as a pair-rule gene and generates similar phenotypes to the ones described here for mlpt with thoracic segmental fusion and lack of abdominal segments [42]. As described for the role of mlpt in segmentation ( [33] and here), E75 is also involved in the same process in O. fasciatus, while its role in segmentation was lost in the lineage giving rise to the Diptera D. melanogaster ( Figure   12). It is not clear if ecdysteroid titers change during R. prolixus embryogenesis and if ecdysone directly regulates mlpt expression during these stages. Interestingly, after Rpmlpt parental RNAi a few nymphs hatch and show defective leg hardening and darkening, suggesting that cuticle has been affected. Previous studies [42] and our data presented here suggest possible connections between the segmentation and molting gene networks, which should be addressed by future studies.

Material & Methods
Bioinformatic analyses -Mlpt peptide sequences from D. melanogaster and T. castaneum were used for BLAST searches [43] against available arthropod genomes and transcriptomes using relaxed parameters to maximize the chances to obtain genes encoding smORFs. The species used for the smORF identification are depicted in Sup. prolixus genome a strategy similar to [44] was designed. The D. melanogaster Ecdysone Receptor motif was obtained in the FlyFactorSurvey database [45] and used to scan upstream, intronic and downstream of mlpt genomic regions of D. melanogaster, T. castaneum and R. prolixus using FIMO [46].
Insect rearing, fixation and dissection -Insect rearing was performed as described by [72]. Approximately one week after blood-feeding eggs were collected daily and fixed in different stages of development. For fixation, up to 100 eggs were briefly washed with distilled water to remove the debris and then transferred to a 1.5 mL microtube containing 1 mL of distilled water. This microtube was maintained for 90 seconds in boiling water, and after this period the water was replaced by 1 mL of paraformaldehyde 12% (PBS) and fixed for one hour (6-8ºC). Then, the embryos were incubated with 1 mL of paraformaldehyde 4% containing 0,1% of Tween 20 under agitation (200rpm) for one hour at room temperature. The eggs were then washed repeatedly with PBST (PBS 1X, Tween 20 0.1%). For long-term storage, embryos were gradually transferred to ethanol 100% and then stored at -20ºC. For the dissections two forceps (Dumont No. 5) were used. The egg is hold with the help of one of them, while the other is used to pressure at the chorionic rim to remove the operculum. The chorionic rim is hold and the shell is opened transversely, leading to the embryonic release. For later stages, when segmentation is complete, at least stage 6 [23], the yolk is easily removed either with the help of a thin forceps or a glass needle, without damaging the embryo.

In situ hybridization, nuclear and antibody staining
Embryos stored in ethanol 100% at -20ºC were gradually transferred to PBST at room temperature. in situ hybridization and probe synthesis were performed as described by [73] including the proteinase K treatment. DAPI (4′,6-Diamidine-2′-phenylindole dihydrochloride, SIGMA) staining was performed as in [23]. All the images were acquired with the stereoscope Leica M205 and processed and analyzed with the software Leica Application Suite Advanced Fluorescence Version 0.4 (LAS AF v4 -Leica Microsystems). Images were assembled and the in situ hybridization staining (NBT/BCIP) was converted to a false fluorescence as described in [74].

RNA interference and real-time PCR
The RNA interference (RNAi) was performed similarly to [44] using a non-related dsRNA as a control (neomicin dsRNA-dsneo). For mlpt, two non-overlapping PCR fragments containing T7 promoter initiation sites at both ends were used as templates for   [75]. mlpt polycistronic mRNA organization. White boxes comprise the whole polycistronic mRNA containing the smORFs encoding peptides displayed in red, green and pink. B) Alignment of the previously described small peptide (red box in Figure 1A) containing an LDPTG domain [4; 5; 6], conserved in several arthropod species and whose number of paralogs in a single polycistronic gene varies among arthropod species. C) Alignment of a hemipteran specific peptide of about 80 amino acids (green box in Figure 1A. The pink smORF encoded peptide is only observed in triatomines and its alignment was omitted for simplicity.         shows an apparent segmental fusion at the thoracic region. Progressive stronger phenotypes show fusion also in the abdominal segments. In the strongest phenotype (right most) the head is affected (asterisk) and germ band elongation did not occur. Arrows highlight segmental fusions. Upper row -In control embryos at stage 6, pb is expressed in the third gnathal segment, corresponding to the prospective labium. After mlpt RNAi pb is either non-affected (right side) or an irregular shape (lef side). Duplications or lack of staining were not observed after mlpt RNAi. Kr RNAi embryos show duplication patterns, gt RNAi embryos show a smaller group of expressing cells and the lack of anterior gnathal segments. svb RNAi displays clear segmentation deviations from the wild-type pattern, but does not display changes in pb staining. Lower row -Ubx expression in control embryos. Expression is stronger at the first abdominal segment and in concentric rings of the third leg. After mlpt RNAi Ubx is still strongly expressed in the first abdominal segment and not in concentric rings of the third leg. Kr RNAi embryos show diffuse staining in the abdominal region. Strong phenotype of gt RNAi embryos lack abdominal Ubx expression, while this marker is expressed in two consecutive pairs of legs, suggesting duplication of the third leg (L3). Hb RNAi embryos display large changes in the fate map with posterior abdominal expression of Ubx appearing at anterior (gnathal) segments. No Ubx staining was observed in the legs.