smORF peptide family represents an overlooked reservoir of functional regulators during development

In order to find novel regulatory smORF peptides, we decided to identify candidates and to test their functionality by inducing their loss of function. We focused specifically on the Drosophila leg because, as an external organ, it facilitates phenotypic analyses and the identification of defects induced by loss of function of candidate genes. Furthermore, screening in the leg favors linking the type of defects to possible affected signaling pathways implicated for instance in proximo-distal axis patterning, tissue growth, joint formation or epidermal differentiation. To identify genes encoding putative smORF peptides, we performed differential expression analysis combined with a previously published smORF finding approach [18]. We thus generated transcriptomes of imaginal leg discs at two different stages of development, at wandering larval 3 stage (wL3) before pre-spiracle eversion, which indicates that the peak of ecdysone required for entry into metamorphosis has not yet occurred, and 2 hours APF (After Pupal Formation) (Fig 1A). Ecdysone signaling induces a transcriptional switch and leg evagination in the proximo-distal axis, possibly favoring our chances to find out regulatory smORF peptides. The bioinformatics analysis to identify smORFs is mainly based on the PhyloCSF method [19], which distinguishes coding and non-coding sequences based on substitution patterns in the whole genome alignment of 12 Drosophila species. This method allowed us to search for genes encoding for putative smORF peptides (Fig 1B) and to list 396 predicted ones (S1 Table), of which 103 are unannotated. Among them, prediction tools identify 162 smORF peptides with specific protein motifs, such as mitochondrial targeting sequence (MitoFates, DeepMito), peptide signal (SignalP 6.0) or transmembrane domain (TMHMM 2.0) (Fig 1C and S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. SmORF peptide family comprises a large number of developmental regulators.

(A) Schematic representation of the pipeline to identify putative genes encoding smORF peptides expressed in the leg disc. RNAseq was done at two developmental time points of the leg disc, at wandering L3 (wL3) and 2 hours After Pupal Formation (APF). Then bioinformatic analysis was performed to predict smORF peptides from the genes expressed in the leg. (B) Diagram representing the different types of putative smORF peptides found in the leg discs both at wL3 and 2h APF. uORF stands for upstream ORF (green), intORF for internal ORF (orange), dORF for downstream ORF (yellow). SmORF located in monocistronic RNA (pink), which are annotated (sCDS for short coding sequence) or non-coding (lncORF), encoded for either characterized smORF peptides (known functions), or uncharacterized ones (unknown functions). (C) Venn diagram showing predicted protein motif (signal peptides, mitochondria targeting sequence, transmembrane domain) found in 162 putative smORF peptides over the 396 identified. (D) Volcano plot showing log2 fold change values (x-axis) by–log10 corrected p-values (y-axis) for genes encoding for putative smORF peptides between larval and pupal stages. Note that pri/tal gene is the most differentially expressed gene. (E) Examples of tarsal phenotypes obtained after loss of function of gene encoding smORF peptides. The UAS-RNAi is expressed under the control of Dll^EM212-Gal4 at 29°C. Different defects can be observed, like shortening of the tarsus (compare length of the tarsus with the red line representing the length of the control), fusion of tarsal segments, incomplete joints (blue arrow-heads) and trichome defects. Scale bar = 200μm.

https://doi.org/10.1371/journal.pgen.1011004.g001

The majority of smORF (59%) are localized in a monocistronic RNA (216 genes), which are either annotated as coding (sCDS for short Coding Sequence) or non-coding (lncRNA and pseudogenes) [20]. Remaining smORF are found in canonical coding genes, localized in 5’UTR (uORF), 3’UTR (dORF) or within the main coding sequence (Fig 1B). Of the 216 putative smORF peptides, 96 have been characterized either through conservation among eukaryotes or through functional studies. Then, 120 putative smORF peptides remain with unknown function. Therefore, to figure out which of these smORF are functional and go beyond their theoretical identification, we induced their loss of function during leg development to test their role in vivo. Using both the available transgenic fly lines in the stock centers (Bloomington and VDRC) and newly generated fly lines, over these 120 genes, we were able to induce loss of function of 93 of them in the distal part of the leg using the Gal4/UAS system to drive either RNAi or gRNA (CRISPR/Cas9). We used the Distal-less-Gal4 driver (Dll^EM212), which is expressed specifically in the leg from the distal tibia to the claws, the tarsus, during Drosophila development [21]. Over the 93 tested genes, the depletion of 23 of them impaired tarsus formation. We observed different types of tarsal defects, like fusion of segments, incomplete joints, necrosis, tarsi reduced size, altered cuticle formation or trichome pattern (Figs 1E and S1). The diversity of phenotypes suggests that smORF peptides are implicated in different cellular processes. Notably, loss of function of a high proportion of tested genes (24%), which encode putative smORF peptides, induces phenotypes. Therefore, our functional screen highlights the smORF peptides as a reservoir of novel developmental and cellular regulators.

Interestingly, differential expression analysis at two developmental time points shows a remarkable switch in gene expression between larval and pupal stage (946 genes with log2FC>1 and 827 with log2FC<1). Among the genes encoding putative smORF peptides, the most differentially expressed gene is the pri/tal gene (log2FC = 4,77; Fig 1D and S1 Table). Pri peptides have been discovered for their role in tarsus formation [10], where they are known to control tarsal patterning and joint formation [22–24]. Pri peptides are required for the establishment of the transcriptional program controlling tarsal segmentation, but the underlying mechanisms are not known. Also, it has been proposed that Pri peptides control joint morphogenesis through Svb and Notch regulation during pupal stage [24]. As the molecular mechanisms of action of Pri peptides remain not well understood during larval and pupal leg development, we then investigated their functions in the light of our recent findings and analyzed the role of Ubr3 in this process [16].

Pri peptides play distinct roles at larval and pupal stages

As we observed a strong increase in pri expression between wL3 stage and 2h APF stages (Fig 1D), we analyzed pri mRNA localization by quantitative fluorescent in situ hybridization (smiFISH) [25]. As previously described [10], we observed pri mRNA at midL3 stage in the form of a ring-shaped pattern marking the presumptive territory of the tarsus, which stops at wL3 stage (except in the chordotonal organ) (Fig 2A). Then, at the onset of metamorphosis, pri is strongly reactivated in the whole leg disc and in the peripodial membrane (Fig 2A). The dynamic pattern of pri expression during leg development suggests different functions. To test this hypothesis, pri expression was specifically depleted in the tarsus at larval or/and pupal stages by using different genetic approaches (Fig 2B). To analyze the effect of pri depletion only during the larval stage, we used the tal¹ mutant [10], in which pri expression is specifically absent during the larval stage in the leg, but unaffected at the onset of metamorphosis (Fig 2C). Indeed, in our hands, depletion of pri with RNAi during the larval stage was not efficient enough to get rid of the larval function of pri. tal¹ allele affects the cis-regulatory genomic region controlling larval pri expression in the leg, named priI [26]. Indeed, ectopic expression of pri under the control of priI in tal¹ mutant background restores tarsus morphology (S2A Fig). Thus, the absence of pri specifically during the larval stage leads to the fusion of the tarsal segments and then to a shorter tarsus (Fig 2B). To specifically delete pri at pupal stage, we used the Gal4/UAS and the thermo-inducible systems (Dll>RNAi pri; tub-Gal80^ts). The absence of pri during pupal stage induces the loss of tissue integrity of the distal part of the leg (Fig 2B). Then, to perform pri depletion during both larval and pupal stages, we used the null allele mutant pri^S18 [10] to induce large pri^-/- clones in Minute context specifically in the tarsus, using the FRT/FLP system, in which the Flippase is expressed under the control of Dll. Continuous depletion of pri over larval and pupal stages accumulates both phenotypes and results in a shorter and dramatically altered tarsus (Fig 2B). Therefore, during leg development, Pri smORF peptides exhibit distinct functions as they are required for proximo-distal axis patterning in the leg disc and consequently tarsal segmentation, and then at pupal stage, they are essential to ensure tissue integrity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Pri peptides have distinct roles during leg development.

(A) Fluorescent in situ hybridization of pri mRNA at different stages of leg disc development, schematized above by the drawings: from left to right, midL3 for mid third instar larvae 3, wL3 for wandering third instar larvae 3, 0h APF (After Pupal Formation) corresponds to the start of metamorphosis, 2h APF. pri expression is transient and pulsatile in a typical ring-shaped pattern in the presumptive tarsal region in the midL3 disc and pri expression is abolished in the wL3 disc (except pri expression observed in the chordotonal organ, yellow arrow-head). At the metamorphosis, corresponding to 0h APF, pri is strongly expressed in the whole distal region. In 2h APF disc, pri is expressed in the epithelium of the disc and in the peripodial membrane. (B) Phenotypes of adult leg obtained after stage specific pri depletion, schematized by a red cross, during larval or/and pupal stages. pri depletion during larval stage, here obtained by using the tal¹ mutant, induces a strong reduction in tarsal size, fusion of the different tarsal segments and absence of tarsal joints. Loss of function specifically during pupal stage is induced by expressing UAS-RNAi pri under the control of Dll^EM212-Gal4 driver and tub-Gal80^ts at 29°C, and leads to tissue integrity loss. To deplete pri during larval and pupal stages, pri^-/- (tal^S18) clones in the Minute cellular context were induced specifically in the tarsus (Dll^EM212>Flippase). The resulting phenotype cumulates the defects described above, up to the tarsus disappearance. (C) pri in situ fluorescent hybridization (smiFISH) in control and in tal¹ leg discs. pri expression is absent in the tal¹ mutant specifically during larval stage (midL3, outlined by the yellow dashed-line) and reactivated at metamorphosis (2h APF). Note the persistence of fusion of the tarsal segments in tal¹ disc at 2h APF. All scale bars = 100μm.

https://doi.org/10.1371/journal.pgen.1011004.g002

Pri peptides are required for early steps of tarsal patterning

During larval stage, tarsal patterning is regulated by multiple actors, that define each segment composing the future tarsus. During the first and the second instar larvae, the morphogens Hedgehog (Hh), Decapentaplegic (Dpp) and Wingless (Wg) establish the anterior-posterior and dorso-ventral axis in the leg disc [27]. Consequently, the expression of Dll is activated in the leg disc, then defining the tarsal region during L3 stage [28]. Also, during early L3 stage, the EGFR signaling pathway is activated through the integration of the signal of Wg, Dpp and Dll at the center of the disc, known as the EGFR organizing center (EOC), and will govern the identity of the pretarsus. At mid-late L3 stage, a second wave of EGFR signaling (non-EOC), mostly dependent on the metalloprotease Rhomboid and the ligand Spitz, is activated in the tarsal region [29–31]. Thus, both EGFR and Dll subdivide the medial tarsal region and allow the expression of spineless (ss), which in turn induces rotund (rn), both TFs being necessary for subsequent tarsal patterning [22,32]. Furthermore, Notch signaling is required for patterning boundaries between segments, which prefigure joint formation [33]. Notably, rn is necessary for Notch pathway activation [34] (S2D Fig). During tarsal patterning, it was previously shown in tal¹ and tal^KG mutants, or in tal^S18 mutant clones, that pri was required for activating ss and rn transcription [10,22,23]. Here we showed in the tal¹ mutant, in which pri expression is specifically abrogated in leg disc at larval stage, that Dll protein is still present while Ss and Rn are absent (Figs 3A and S2B). We then investigated at which stage of the regulatory cascade pri was acting. We observed that Notch signaling pathway, which is activated from the larval stage, was absent in the presumptive region of the tarsus in tal¹ (Fig 3B), as confirmed by the absence of Deadpan (Dpn) protein and dysfusion-lacZ reporter line activity, both direct targets of the Notch pathway [35]. As EGFR is important for limiting Notch signaling at joint boundaries [36], we stained tal¹ L3 leg discs with anti-Phospho-ERK antibody, a marker of MAPK activity used as a read-out of active EGFR signaling pathway, and revealed its absence in the tarsal region (Fig 3C). Furthermore, we found that in tal¹ rhomboid mRNA was absent in tarsal region, showing that the second wave of EGFR activation is compromised (Fig 3D). Nevertheless, in the absence of pri, the initial EGFR wave is activated as indicated by the presence of the TF Clawless specific from the pretarsus and the formation of the claws (S2C Fig) [37]. Therefore, our data reveals that Pri peptides are required for rhomboid transcription, and consequently EGFR signaling, activation of the tarsal transcriptional program and Notch signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Pri peptides control tarsal patterning by activating transcriptional program and Notch and EGFR signaling pathways.

(A) Immunostaining of Dll, Ss and Rn transcription factors in control and in tal¹ mutant discs at L3 stage. Dll is present in both control and tal¹ distal region, whereas Ss and Rn proteins are absent in the tal¹ mutant leg discs. (B) Immunostainings of Deadpan (Dpn) and βGalactosidase (βGal) in a control and tal¹ in L3 leg disc. dys640-LacZ is an enhancer of dysfusion gene directly activated by Notch, showing the absence of Notch signaling in the tarsal region. (C) The anti-P-ERK signal, used as a read-out of EGFR signaling, stains at late L3 stage (lL3) the whole tarsal region in the control while being absent in the tal¹ mutant. The leg discs are outlined by the yellow dashed-line. In situ hybridization of rhomboid mRNA in control and in tal¹ leg discs. (D) The rhomboid mRNA is observed in the presumptive tarsal region in concentric ring pattern. In the tal¹ mutant, rhomboid expression is abrogated in the disc except in the chordotonal organ (black arrow). All scale bars = 100μm.

https://doi.org/10.1371/journal.pgen.1011004.g003

We conducted genetic epistasis analysis to determine the functional order of these genes in tarsal patterning. Interestingly, ectopic expression in the tarsus of tal¹ mutant of activated form of EGFR or the activated EGFR ligand sSpitz [38] failed to restore Notch and EGFR signaling, or transcription factors activation (S2E Fig). Ectopic expression of ss is not sufficient to activate rn, and ectopic expression of rn is not sufficient to activate Notch signaling (S2F Fig). None of these actors is able to replace pri function, suggesting that Pri peptides are acting at several steps in this molecular cascade. Furthermore, ectopic expression of pri in the wg domain, which is expressed in a sub-region of the leg during development [28], enables Notch signaling pathway and tarsal transcriptional program activation, as visualized respectively with Dpn and Rn immunostainings (Fig 4A). Also, these factors were reactivated beyond the area of wg expression domain, over a distance of several cell diameters (Fig 4B), which could be due to the cell non-autonomous properties of Pri peptides, as previously observed by several laboratories [10,11,39]. Furthermore, we generated pri^-/- clones in the Minute cellular context and observed that clone size greatly influences the localization of Dpn and Rn. Indeed, small clones show no defect in Dpn and Rn localization, probably because Pri peptides diffuse from neighboring wild-type cells and compensate for the absence of pri in these small clones, whereas in large clones encompassing most of the tarsus, Rn and Dpn pattern are strongly altered (Figs 4B, 4C and S2B). This genetic approach also illustrates the non-autonomous properties of Pri peptides.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Cell non-autonomous properties of Pri peptides synchronize transcriptional program and signaling pathways in tarsus.

(A) Rescue experiments have been conducted by expressing ectopically pri in tal¹ mutant background under the wingless-Gal4 driver (wg). Wg staining delimits the region where pri is expressed (outlined by the yellow dashed-line). Rn and Dpn are used as read-out of tarsal patterning and Notch signaling pathway activations. Rn and Dpn are normally patterned in the control disc, whereas they are absent in the tal¹ mutant. Ectopic expression of pri (wg>pri; tal¹) rescues partially the tal¹ phenotype since Rn and Dpn staining are restored in the wg region. Scale bar = 100μm. (B) Magnification on the wg region of the wg>pri; tal¹ leg disc. Note that Rn and Dpn are present at the same level beyond the wg region, suggesting that Pri peptides induce cell non-autonomously activation of tarsal patterning. Scale bar = 50μm. (C, D) Mosaic clones for pri^-/- (tal^S18) were induced in Minute cellular context with the expression of Flippase under the control of Dll driver. Generation of small clones in the tarsus does not affect Dpn patterning (C), whereas in large clones, Dpn patterning almost completely disappears (D). Scale bar = 100μm.

https://doi.org/10.1371/journal.pgen.1011004.g004

In conclusion, our data show that Pri peptides are required for the activation of signaling pathways and transcription factors that governs tarsal patterning. We propose that cell non-autonomous properties of Pri peptides could coordinate the activation of these actors within the tarsus to ensure harmonious tarsal development.

Pri peptides controls larval disc patterning in Svb/Ubr3 independent manner

To go further, we then investigated whether the roles of Pri peptides in leg development during larval stage were dependent on Svb and Ubr3, the partners identified for mediating Pri functions for trichome formation during epidermal differentiation [15,16] (Fig 5A). Both Ubr3 and svb are expressed during larval and pupal stages at comparable levels (S3A Fig). The svb mRNA and pri mRNA patterns during larval stage were visualized by fluorescent in situ hybridization (Fig 5B). While pri mRNA expression is transient at midL3 stage in the presumptive tarsal region and continuous throughout the disc from the onset of the pupal stage, svb mRNA is ubiquitously expressed at the same level throughout the leg disc during larval and 2h APF stages (Fig 5B). To go further and analyze the endogenous pattern of Svb protein in the leg disc, we generated a fly line in which the endogenous Svb was tagged to the GFP at the C-terminal position (KI svb::GFP, Fig 5C). Immunostainings against both GFP protein and 1S domain, which is specific from the Svb^REP [15], allow to visualize Svb^REP form (1S and GFP positive) and Svb^ACT form (only GFP positive) during leg disc development. We observed it is present as the repressor form in the larval stage and as the activator form in the pupal stage (Figs 5C, S3B and S3C). Strikingly, Svb, which is under its full-length repressive form at larval stage, is fully degraded at the timing of transient pri expression (S4A Fig). This complete degradation is dependent of pri because in tal¹ mutant, Svb protein persists (S4A Fig). To test whether Svb disappearance has a role in tarsal patterning, we ectopically expressed either Svb^REP, the full length form of Svb with 3 lysines mutated to inhibit its Pri-dependent processing [16], or the Svb^ACT, whose amino acid sequence corresponds exactly to the Pri-dependent processed form [15]. Their expression in larval leg disc has no detectable effect on Rn pattern (S4B Fig). In addition, expression of Svb^REP or Svb^ACT does not modify the Dpn pattern (S5B Fig), on the contrary of germinal isoforms of Svb, OvoA and OvoB, commonly previously used to mimic repressor and activator forms of Svb in somatic tissues (S5A and S5C Fig). These experiments reveal that Svb^REP and Svb^ACT are better suited to faithfully reproducing the effects of repressor and activator forms of Svb in somatic tissues. Therefore, these results suggest that the transient absence of Svb at midL3 appears to not have a role in tarsal patterning.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Svb is ubiquitously expressed in leg disc at larval and pupal stages.

(A) Drawing representing Svb Pri/Ubr3-dependent processing. svb is transcribed and translated as a long repressor form (Svb^REP), that contains the exon 1S. In embryonic epidermis, the presence of Pri peptides and Ubr3 induce the N-terminal domain degradation, leading to a shorter activator form of Svb (Svb^ACT). (B) svb expression was monitored along pri expression by fluorescent in situ hybridization (smiFISH). Fluorescent svb probes recognize only the 1S region of svb, showing that the somatic svb isoform is transcribed in imaginal discs. svb mRNA is expressed homogenously throughout the entire leg disc during larval and pupal stages and concomitantly expressed with pri at midL3 stage and at pupal stage. (C) Drawing representing the knock-in (KI) of GFP in the svb locus (svb::GFP). Anti-GFP (green) and anti-1S (purple) immunostainings on svb::GFP KI at larval stages (midL3 and wL3) and at pupal stages (0h and 2hAPF) show that Svb is ubiquitously localized in the leg disc and is in the Svb^REP form at larval stage and in its processed form Svb^ACT at the pupal stage, as confirmed by the absence of 1S signal. Scale bar = 100μm.

https://doi.org/10.1371/journal.pgen.1011004.g005

As Svb is ubiquitously localized in the leg disc at larval stage the majority of the time, we generated svb loss of function by generating mutant clones or inducing RNAi and stained leg discs to analyze the effects on tarsal patterning and Notch signaling. Persistence of Ss and Rn TFs or Dpn proteins in the absence of svb demonstrate that Svb is not required for their activation (Figs 6A, S6A, S6B and S6D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Pri peptides activate tarsal TF and EGFR and Notch signaling pathways in Svb/Ubr3 independent manner.

(A) Ss and Rn were stained in svb^-/- (svb^PL107) MARCM clones in L3 leg discs. Clones are in green and outlined in yellow in the close-up. In svb^-/- clones, both Ss and Rn proteins remain. (B) Ss and Rn were stained in Ubr3^-/- (Ubr3^B) MARCM clones in L3 leg discs. In the Ubr3^-/- clones, both Ss and Rn proteins remain, though the level of Ss is lower (green arrow-heads). (A,B) White square highlights the region displayed in the close-up showed on the right of the panel. Scale bar = 100μm, scale in the close-up = 10μm. (C) UAS-RNAi luciferase (ctrl) or UAS-RNAi Ubr3 were expressed under the control of Engrailed-Gal4 driver visualized in red (en>mRFP). The yellow line delimits the anterior and the posterior regions. Rn pattern remains unchanged when Ubr3 is depleted. Scale bar = 100μm.

https://doi.org/10.1371/journal.pgen.1011004.g006

We then examined the role of Ubr3 in tarsal patterning using the same genetic approaches, either by generating Ubr3 mutant clones or by inducing loss of function by RNAi. Notch pathway is not affected by loss of Ubr3, as visualized by the persistence of Dpn in RNAi- depleted region or in mutant clones (S6A and S6C Fig). Also, depletion of Ubr3 by RNAi does not modify Rn pattern (Fig 6C). Since the absence of Ubr3 leads to apoptosis [40], the slight variation on Rn protein levels might be the consequence of a deleterious cellular context. Additionally, Ss and Rn are still present in the Ubr3^-/- clones (Figs 6B and S6E), thus supporting that Ubr3 is not required for mediating pri function in the larval leg disc.

Our results reveal that Pri peptides functions in patterning during the larval stage are not mediated by Svb and are independent of Ubr3, revealing the existence of additional Pri molecular targets.

Pri peptides induce Svb processing in Ubr3 dependent manner at pupal stage

We observed that Svb full length form persists during larval stage and is processed at larval-pupal transition at the time of tarsus eversion (Fig 5C). Furthermore, we ectopically induced svb depletion by RNAi in the posterior part of the leg and we observed that Svb is processed homogenously in the pupal epithelium of the leg and that there is no persistence of the full-length repressor form (S3C Fig). As Pri peptides and Ubr3 induce Svb processing in epidermis, switching the Svb^REP to the Svb^ACT forms [16], we tested whether Svb processing at larval-pupal transition was dependent of Pri and Ubr3. We generated Ubr3^-/- clones in the leg disc and showed that Svb processing relies on Ubr3 as shown by the persistence of the 1S signal at pupal stage (Fig 7A, 7A’, 7B and 7B’). To test the role of pri, we generated pri^-/- clones in the Minute cellular context to obtain large clones and avoid rescue of mutant cells by neighboring wild-type cells, due to the cell non-autonomous property of the Pri peptides. We observed also the persistence of the 1S signal in pri^-/- mutant clones (Fig 7C and 7C’). Therefore, Pri peptides are required at pupal stage to induce Ubr3-dependent Svb processing. The function of Pri/Ubr3/Svb module is reiterated during development, specifically in the imaginal leg discs during the larval-pupal transition when pri expression is strongly reactivated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Pri and Ubr3 are required for Svb processing at the larval-pupal transition.

Anti-1S immunostaining in the pupal leg disc where mosaic clones for pri^-/- (tal^S18) and Ubr3^/- (Ubr3^B) were induced. The clones are indicated by the absence of GFP and contoured by the yellow dashed-line. The DNA is marked in blue, 1S in red and the GFP in green, white square highlights the region displayed in the close-up (A’-C’). (A, A’) In control clones, Svb is processed as 1S signal is lost. (B, B’) In Ubr3^-/- clones, 1S staining is remaining, showing that Svb^REP is not processed. (C, C’) pri^-/- mutant clones have been generated in Minute background with the Flippase under the control of Dll^EM212-Gal4 in order to get large clones (if pri^-/- clones are too small, they are behaving like control clones). The absence of GFP indicates that almost all the leg is clonal (yellow dashed-line) and positive for anti-1S signal. (A- C) Scale bar = 100μm. (A’- C’) scale bar = 20μm.

https://doi.org/10.1371/journal.pgen.1011004.g007

Pri peptides control cell survival during pupal stages

Our results showed that the Pri, Ubr3 and Svb are cooperating during pupal leg development. We then depleted by RNAi each gene specifically during pupal stage using the tub-Gal80^ts system to see how their absence affects pupal leg development. Interestingly, depletion of each gene induces distinct phenotypes in terms of severity, suggesting that Pri peptides possess additional developmental functions compared to Ubr3 and Svb. Indeed, RNAi depletion of pri leads to a more severe phenotype than the depletion of Ubr3, which in turn is more severe than svb depletion (Fig 8A). As the absence of pri during pupal stage induces the loss of tissue integrity, we tested whether cell death could be the cause of this phenotype. We thus stained leg disc with anti-Dcp-1, the cleaved form of the ortholog of human caspase-3, and we observed a significant increase in the number of apoptotic cells (Figs 8B and S7A), thus corroborating a role for Pri peptides in protecting cells from apoptosis. We ectopically express during pupal stage both RNAi pri and miRHG, a transgene which produces miRNAs against the pro-apoptotic genes rpr, hid and grim [41], to strongly inhibit the apoptosis. Despite a significant decrease in apoptosis (S7A Fig), miRHG rescues partially tarsus morphogenesis in the absence of pri (S7B Fig). Furthermore, to test if the functions of pri during pupal stage are mediated by Svb, we expressed Svb^ACT in the absence of pri. Interestingly, we observed that Svb^ACT rescues partially leg phenotype (S7B Fig), suggesting that functions of pri during pupal stage are partially mediated through the activation of Svb.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Pri peptides, Svb and Ubr3 protect cells from apoptosis at pupal stage.

(A) Depletion of pri, Ubr3 or svb in the tarsus is induced by RNAi specifically at pupal stage under the control of Dll^EM212-Gal4 driver and tub-Gal80^ts when larvae at wL3 stage are shifted to the restrictive temperature (29°C). Depletion of these genes alters leg morphogenesis to a different extend. Loss of pri induces abnormal joint formation (marked with green arrowheads), cuticle defects and loss of tissue integrity, which resembles to necrosis. Loss of function of Ubr3 leads to growth and cuticle defects, and abnormal joints. In the absence of svb, the segments of the tarsus are shorter, the trichome pattern is affected and the joint are not properly formed. (B) Depletion of pri, Ubr3 and svb were induced by RNAi at wL3 stage under the the control of Dll^EM212-Gal4 driver and tub-Gal80^ts. The Dll domain, stained by the presence of UAS-GFP, is outlined with the yellow dashed-line. Anti-Dcp-1 antibody stains apoptotic cells (in green) in pupal leg disc, that are present in the control leg disc (RNAi luciferase), specifically in the pretarsus, as often described. Depletion of pri, Ubr3 or svb induces an increase in the number of apoptotic cells. Scale bar = 100μm. (C) Co-immunoprecipitation (IP) of Ubr3::V5 and DIAP1::GFP with or without Pri peptides. DIAP1::GFP and Ubr3::V5 are co-expressed in S2 cells with or without pri, and then co-immunoprecipitated with anti-GFP (left panel) or anti-V5 (right panel) antibodies. In the absence of pri, DIAP1 and Ubr3 co-interact barely. With Pri peptides, this interaction strongly increases. (D) Model of Pri peptide functions during leg development: pri expression is spatio-temporally regulated during leg disc development. During L3 stage, Pri peptides activate Rotund and Spineless transcription factors, and EGFR and Notch signaling pathways. Therefore, Pri peptides coordinate transcriptional program and signaling pathways to ensure tarsal patterning. Then at larval-pupal transition, when pri is reactivated in the leg disc by ecdysone signaling, it induces Svb processing in a Ubr3 dependent manner. Pri expression is maintained during pupal stage in the leg disc. The module Pri/Ubr3/Svb is required for cell survival, but also morphogenesis and for maintaining epithelial integrity.

https://doi.org/10.1371/journal.pgen.1011004.g008

Then, we generated loss of function of Ubr3 and Svb to test also their role in cell survival during pupal stage. As shown before [40], the absence of Ubr3 induces increased apoptosis in pupal leg disc (Fig 8B). Also, svb depletion increases significantly Dcp-1 positive cells, indicating that Svb also plays a role in protecting cells from cell death (Figs 8B, S8A and S8B). We performed rescue experiments by co-expressing RNAi svb and the major anti-apoptotic factor DIAP1 or the viral caspase inhibitor p35 [42]. Although, expression of DIAP1 or p35 rescue significantly the apoptosis induced by the absence of svb (S8B Fig), the morphogenesis of the tarsus is partially restored (S8C Fig).

To go further, we investigated the role of the module Pri/Ubr3/Svb in the regulation of DIAP1. Since Svb was shown to protect cells from apoptosis via the regulation of DIAP1 transcription in digestive stem cells [13,14], we performed DIAP1 fluorescent in situ hybridization (smiFISH) in imaginal leg discs in the absence of svb. We did not observe change in the level of mRNA DIAP-1 during the time window encompassing larval-pupal transition, suggesting an alternative mechanism by which Svb protects cells from apoptosis (S8D Fig). Interestingly, Ubr3 was shown to protect cells from apoptosis in imaginal eye disc through its interaction and protection of DIAP1 protein [40], we thus tested the influence of Pri peptides on Ubr3/DIAP1 protein interaction in Drosophila S2 cells. Even though weak co-immunoprecipitation between Ubr3 and DIAP1 is observed without Pri peptides, the presence of Pri increases massively the interaction (Fig 8C). Our data suggests that Pri peptides and Ubr3 are cooperating for protecting DIAP1 from degradation and prevent cells from entering in apoptosis in vivo.

Altogether, our data reveals that the module Pri/Ubr3/Svb is protecting the leg from cell death and is necessary for morphogenesis and preserving tissue integrity throughout the development of the pupa. Moreover, rescue experiments suggest that Pri peptides, and also Svb, have additional roles beyond protecting cell from death throughout the pupal leg development.

The ability of Pri peptides to control multiple cellular events through the activation of distinct factors within the same tissue over time illustrates their pleiotropic functions in the temporal control of development (Fig 8D).

smORF peptides provide a pool of novel developmental actors

Bioinformatics analyses have identified hundreds of putative smORF peptides, which have been classified in function of their origin. We focused on genes encoding smORF peptides that have never been studied because they were recently annotated (sCDS) or classified as lncRNA or pseudogenes. Importantly, genetic tools, allowing us to deplete their function, are available for half of these genes in stock centers for the fly community, and their use allowed us to identify a significant number of potential candidates for controlling development. Interestingly, 40% of smORF peptides display a motif, that could be useful to address its biological function. More than 10% of smORF peptides appear to be addressed to mitochondria, suggesting a tendency of the smORF peptides to localize to this organelle compared to the whole proteome (6%) as previously observed [43,44]. Furthermore, the phenotypes obtained are diverse, affecting cell survival, segment fusion or tissue growth, suggesting that smORF peptides are involved in all cellular processes. Indeed, recent studies showed that they can exhibit multiple subcellular localization [45,46], with role in, for example, the regulation of calcium flux, the inhibition of protein activity [47], the antigen presentation [48] or the biogenesis of the respiratory chain [49]. Obviously, further studies are now needed to better understand the function of these smORFs. Interestingly, we found that half of these putative regulatory smORF peptides have orthologs in vertebrates (S1 Table), for instance, CG33169 (55AA) is encoding for the ortholog of human SMCO4 (59AA), a peptide of unknown function containing a transmembrane domain. Thus, Drosophila is a good model for identifying among the hundreds of existing smORF peptides new regulators of important cellular processes conserved in eukaryotes.

It is generally accepted that eukaryotic genes are monocistronic, i.e. they contain a single ORF. However, we found that a high proportion of predicted smORF peptides are located in the 5’UTR and 3’UTR, supporting the existence of polycistronic eukaryotic genes. Indeed, recent studies in Drosophila or vertebrates using mass spectrometry or ribosome profiling have shown that smORF peptides are indeed translated from the 5’ or 3’ UTR, even within the main ORF, demonstrating that polycistronic genes in eukaryotes are more widespread than expected [50–52]. This highlights the potential of RNAs to code for several proteins, giving the possibility of greatly increasing the eukaryotic proteome. The challenge now is to define criteria or experimental approaches to select, among the thousands of smORF peptides, those most likely to have important regulatory functions in development.

Finally, our approach to search for putative smORF regulatory peptides, based on bioinformatic analysis of RNAseq data, is thus handable to Drosophila or any other organism. The use of criteria such as differentially expressed genes, expression level, or specific physiological conditions, as in our case before and after the ecdysone peak, can restrict the analyses to a smaller pool of genes. This may also increase the chances of finding a regulatory smORF peptide with specific spatio-temporal expression, which could help identify function or potential interactors. Here, these criteria highlighted the pri/tal gene, already known to be crucial for leg development [10], thus validating our approach. As two peaks of pri expression occur during the development of the leg disc, we wondered whether they act with the same partners.

Pri peptides synchronize signaling pathways and transcriptional program to ensure tarsal patterning

At midL3 stage, Pri peptides were known to control ss and rn expression [22,23]. Besides a fundamental role in initiating the tarsal transcriptional program, we found that Pri peptides are also required to activate EGFR and Notch signaling pathways. We previously identified the transcription factor Svb and the E3 ubiquitin ligase Ubr3 as the molecular targets of Pri peptides during epidermal differentiation [15,16]. Here, we showed that the Pri/Svb/Ubr3 module is not mediating functions of Pri peptides for tarsal patterning during the larval stage.

Notch activation correlates with tarsal sub regionalization and segment emergence, which occurs in a Ubr3 and Svb independent manner. Our results suggest that pri is required for the second wave of EGFR activation in the tarsus, not in the initial EOC in the pretarsus, by regulating directly or indirectly the expression of rhomboid in concentric circles. Recently, it was shown in Drosophila embryonic tracheae that pri is also required for EGFR pathway activation in dorsal branches, thus supporting our data [53]. Nevertheless, we observed that ectopic activation of the EGFR pathway is insufficient to mediate Pri functions, while EGFR seems to be required for patterning at the same time as pri.

Ss requires pri for inducing rn expression in the leg disc [22]. Our experiments reveal that rn, as well as ss, was not able to restore tarsal patterning in the absence of pri and that rn is required for Notch signaling, thus suggesting that Pri peptides interfere several times in the regulatory cascade. Thus, we propose that Pri peptides are required concomitantly or reiteratively to activate key players, EGFR signaling, transcription factors Ss and Rn, and Notch signaling to synchronize the molecular events governing tarsal formation. Furthermore, cell non-autonomous properties of Pri peptides may be necessary to activate Pri targets at comparable levels within presumptive tarsal domain. It seems that Pri peptides are not acting like a gradient, but rather may be like a switch that either activates or does not activate its targets.

During the larval stage, Pri peptides are acting upstream of the signaling pathways and transcriptional cascade which govern tarsal patterning. Functions of Pri peptides are not mediated by Svb and Ubr3, thus suggesting that additional Pri peptides targets, direct molecular partners and/or indirect targets may exist. We speculate that the transcription factors Dll and Sp1 could interact with Pri peptides to mediate their functions as they control the same targets [8,31,35]. However, deciphering the nature of their interaction will be a long-term effort as the interdependence of the key players and cell non-autonomous properties of Pri peptides renders difficult in vivo genetic approaches and interpretation of the effects of their manipulation.

Pri/Svb/Ubr3 module ensures leg tissue integrity during pupal stage

Pri peptides molecular function in epidermal differentiation, specifically during trichome formation, is mediated by Svb transcription factor and E3 ubiquitin ligase Ubr3 [16]. Here we demonstrated that the Pri/Svb/Ubr3 module is reused during leg metamorphosis, triggered by ecdysone signaling.

As previously shown in digestive stem cells, this module is required for protecting stem cells from apoptosis [13,14]. Loss of function of one of these partners induces apoptosis during pupal stage and may be as a consequence, dramatic alteration of part of the tissue. As we showed that Pri peptides increase Ubr3/DIAP1 interaction in S2 cells, and that Ubr3 interacts with DIAP1 to protect cells from death [40], the couple Pri/Ubr3 could counteract apoptosis via promoting stabilization and protection of DIAP1 in vivo. However, inhibition of apoptosis or ectopic Svb^ACT expression at the pupal stage when pri is absent partially rescues leg morphogenesis, suggesting that Svb and Ubr3 mediate some of the functions of Pri peptides and that these peptides may have additional targets at the pupal stage in the leg. Even though Svb is also required for protecting cells from death, we did not find any effect of Svb on DIAP1 expression as previously described [14]. Morphological defects induced by svb loss of function are partially rescued when apoptosis is inhibited, suggesting that Svb may have additional functions beyond protecting cell from death. Indeed, Svb is also required for epidermal differentiation and trichome formation not only in the tarsus, but also in the tibia [54], and for joint formation [24]. However, during the first hours of pupal leg development, we did not observe a role for Svb on Notch signaling or joint patterning as previously described [24]. It might be due to the use of different genetic tools, like the svb::GFP knock-In line, which recapitulates the endogenous pattern of Svb, and the Svb^REP and Svb^ACT, which are the actual epithelial forms of Svb present in leg tissue, unlike the germline OvoA and OvoB forms used previously [24].

Interestingly, the module Pri/Ubr3/Svb is operating at different stages of development and in different tissues in Drosophila, for instance in embryonic epidermis and in pupal leg, in which pri expression is temporally regulated by the ecdysone signaling (Fig 8D) [26,39]. Moreover, the module Pri/Svb/Ubr3 is conserved among arthropods and regulates embryonic patterning [17], thus suggesting this module might be reiterated across arthropods development in different tissues and organs.

In conclusion, the plethora of molecular events regulated by Pri peptides during leg development is enabled by their pleiotropy. Indeed, they can simultaneously regulate different targets within the same tissue or even within the same cells. This pleiotropy is enhanced by their spatio-temporal transcriptional regulation, which relies on multiple enhancers [26]. For example, the pulsatile expression of pri in the imaginal leg disc depends on several enhancers in the larval and pupal stages, of which only the pupal enhancers are regulated by ecdysone [26]. We then propose that Pri peptides rhythm Drosophila development by coordinating multiple and distinct cellular processes in space and time.

Bioinformatic smORF peptides prediction in imaginal leg discs

Imaginal leg discs were dissected in cold PBS 1X on ice from wandering L3 larvae, before the ecdysone peak, visualized by pre-spiracle eversion, and 2 hours After Pupal Formation (APF). Total RNA was extracted with Trizol reagent (Ambion) according to the manufacturer’s protocol. Construction of RNA polyA+ bank and sequencing using paired-end 100bp reads were performed by IntegraGen. Kallisto [55] was used for pseudo alignement of reads to a reference combining the Ensembl 74 annotations and additional lincRNAs from modENCODE [56,57]. We then used Sleuth for differential expression analysis [58]. Small ORFs were predicted as in Mackowiak et al. [18] using the transcriptomes generated in this study. The data discussed in this article are available via the Gene Expression Omnibus (GEO) under accession number GSE225561.

Functional screen

Loss of function were induced by crossing Dll^EM212-Gal4; tub-Gal80^ts flies with lines expressing RNAi or gRNA and Cas9 under the control of UAS promoter at 29°C. Fly lines are available in Bloomington and VDRC stock centers, or were generated for this study. UAS-RNAi-white was used as the control. The lines giving a phenotype are listed in S1 Fig.

Fly stocks

The Drosophila lines used in this study are tal¹, tal^S18.1, UAS-RNAi pri [10], Ubr3^B [16], svb^PL107, UAS-OvoA, UAS-OvoB [59], PriI-Gal4 (generous gift from H Chanut-Delalande), UAS-pri [11], wingless-Gal4, dysfusion640-LacZ [35], UAS-DIAP1, UAS-miRHG [41], UAS-p35 [42]. MARCM clones were generated by using the following fly line: y, w, hs-FLP, tub-Gal80, FRT19A; UAS-mcd8-GFP; tub-Gal4/TM6b [14]. The Minute clones in tarsus were generated by using the following line: yw; Dll, UAS-FLP; FRT82B, Rps, Ubi-GFP/Cyo-TM6b.

The following lines were available from Bloomington and VDRC stock centers: Engrailed-Gal4, UAS-mRFP (BL30557), UAS-RNAi luciferase (BL31603) and UAS-RNAi white (BL28980), both used as controls, UAS-RNAi svb (v41584), UAS-RNAi Ubr3 (v22901, v106993).

The Knock-In of GFP protein at the C-terminal position of the Ovo/Shavenbaby protein in the endogenous locus was generated by CRISPR/Cas9 by InDroso compagny.

Hybridisation in situ

SmiFISH was performed as previously described [25] and FLAP-X sequence was used to generate pri fluorescent probes. Probes specific from the 1S region of svb mRNA were synthetized by Stellaris. Larvae were dissected and fixed in PFA 4% in 25 min at room temperature, then washed in PBT (PBS 1X/0,1% Triton100X) and permeabilized 20 min in PBT (PBS 1X/0,5% Triton100X). Samples were washed in the wash buffer (4M urea in SSC 2X) and incubated with hybridization mix (4 M Urea, 8μL de SSC20X, 40μL of Dextrane 20%, 3,5 μL of Vanadyl complex at 10 mM, 1,5 μL of competitor DNA, 2,5μL of smiFISH probe and 1,5 μL of water) at 37° overnight protected from light. Samples were rinsed in the wash buffer and in SSC2X. Then, samples were washed in PBT and leg discs were dissected and mounted in vectashield medium (Vector Laboratory).

For rhomboid in situ hybridization, probes (sense and the anti-sense) for rhomboid were synthesized according to standard procedures from LD06131 plasmid. Briefly, larvae were dissected in order to keep discs in PBS1X and fixed 20 min in PFA 4% at room temperature. Samples were washed in PBT (PBS1X/0,1% TritonX100), blocked 30 min in PBT (0,3% Triton X) and washed in PBT (0,1% Triton). Samples are permeabilized in Methanol/DMSO (90%/10%). Samples are rehydrated progressively, prehybridated 1 hour in hybridization buffer (50% Formamide, 4X SSC, torula RNA 1mg/ml, Heparine 0,05μg/ml, 2% Roche Blocking Reagent, 0,1% CHAPS, 50mM EDTA, 0,1% Tween20) and incubated overnight at 65°C with the denatured probe in the hybridization buffer. Samples are washed, rehydrated progressively, incubated with anti-DIG (Roche, 1/2000). Probe is then revealed with NBT/BCIP (Promega). Leg discs were dissected and mounted in a mix PBS/Glycerol.

Immunofluorescence

Larval and pupal imaginal discs were dissected in PBS1X and fixed in PFA 4% during 25 min at room temperature, then washed in PBS1X. Samples are blocked in PBS1X/BSA 0,3%/Triton 0,3% during 1 hour. Primary antibodies are incubated overnight at 4°C. Then, samples were washed and incubated with secondary antibodies for two hours at room temperature, then rinsed in PBS1X and mounted in Vectashield medium (Vector Laboratory).

Antibody against Rotund (Rn) was obtained by immunizing guinea-pigs with the Roe (Roughened eye) full length isoform encoded by the rotund gene and sharing its last 450 C-terminal residues (including five Zn fingers) with the Rn isoform [34]. Contrary to Rn, roe is not expressed and has no function in leg tissues [34]. GST-fused Roe was produced in E coli from a pGEX-Roe plasmid [60], purified through a glutathione column and used to immunize the guinea pigs. Crude serum was used at 1:500. The antibodies used in this study are: anti-Spineless, generously given by J Yuh Nung (1/1000); anti-Distal-less generously given by R Mann (1/500), anti-GFP (Mouse, Roche) (1/500); anti-GFP (Rabbit, Torrey Pines) (1/500); anti P-ERK (P-p44/42 MAPK, Cell Signaling Technology), (1/200); 1/50, anti-Wingless (DSHB 4D4-s), (1/50); anti-Delta (DSHB C594.9B), (1/500); anti-Dcp-1 (Cell Signaling Technology), (1/200); anti-1S, (1/1500) [16], anti-Deadpan (Abcam), (1/100), anti-βGalactosidase (Promega) 1/500. The secondary antibodies were coupled to Alexa Fluor 555, 488 or 647 (Invitrogen). The DNA was marked either with TO-PRO-3 Iodide (642/661) or DAPI (Thermofisher).

Co-immunoprecipitation and western blotting

S2 cells were transfected with pAc-V5::Ubr3, pAc-DIAP::GFP and pMT-pri. Pri expression was induced by CuSO₄ at 1mM for 2 hours. Co-immunoprecipitation and western blotting were done as previously described [16].

Image acquisition

Experiments with fluorescent markers were obtained using microscope Leica sp8. Experiments requiring white light like in situ hybridization or adult leg observation are acquired with microscope Nikon Eclipse 90i.