Peptides encoded by short ORFs control development and define a new eukaryotic gene family

Abstract

Despite recent advances in developmental biology, and the sequencing and annotation of genomes, key questions regarding the organisation of cells into embryos remain. One possibility is that uncharacterised genes having nonstandard coding arrangements and functions could provide some of the answers. Here we present the characterisation of tarsal-less (tal), a new type of noncanonical gene that had been previously classified as a putative noncoding RNA. We show that tal controls gene expression and tissue folding in Drosophila, thus acting as a link between patterning and morphogenesis. tal function is mediated by several 33-nucleotide-long open reading frames (ORFs), which are translated into 11-amino-acid-long peptides. These are the shortest functional ORFs described to date, and therefore tal defines two novel paradigms in eukaryotic coding genes: the existence of short, unprocessed peptides with key biological functions, and their arrangement in polycistronic messengers. Our discovery of tal-related short ORFs in other species defines an ancient and noncanonical gene family in metazoans that represents a new class of eukaryotic genes. Our results open a new avenue for the annotation and functional analysis of genes and sequenced genomes, in which thousands of short ORFs are still uncharacterised.

Figure 1. Characterisation of the tal Locus

(A) Genomic region 87F13–15 showing the location of tal and neighbouring genes. The boxed area around tal is magnified. The inverted triangles represent the insertion sites of P elements. The solid lines indicate the fragment deleted in each mutant, with the indetermination shown as dotted lines. KG1680 and tal¹ are regulatory alleles for the imaginal functions, S011041 is a hypomorph, and the deletions are nulls. (B–F) Male forelegs of different genotypes. In these panels, the tibia is labelled (Ti), the tarsal segments are numbered, and the arrow points to the sex comb. (B) The tibia and five tarsal segments can be observed in the wild type. (C) In the tal¹ mutant, the tarsal region is vestigial and unsegmented. (D) Similar phenotype in a tal-Gal4/tal^S68 leg. (E) tal-Gal4/tal^S68; UAS-tal shows a complete rescue of the phenotype. (F) In dpp-Gal4; UAS-tal ectopic expression of tal in the dorsal leg produces transformation of the distal tibia and fusion to tarsus 1, and ectopic sex comb in tarsi 1 and 2. These phenotypes are compatible with a transformation of tibial identity towards tarsus.

Figure 2. tal Regulates Tarsal Patterning

(A) Expression of the tal RNA in an 84-h leg disc in a ring in the presumptive tarsal region. (B) The expression pattern of the lacZ gene in the reporter line l(3)S011041 faithfully reproduces the expression of the RNA. The arrow points to the tarsal fold contained in the tal domain. (C) By 100 h, the tal RNA has disappeared from the developing tarsal primordium, although it remains in a dorsal chordotonal organ. (D) rn RNA expression in a third instar leg disc at 90 h AEL, in the presumptive tarsal region. (E) In a tal¹ mutant disc, rn expression is abolished. (F) In a dpp-Gal4; UAS-tal disc at 120 h AEL, the ectopic tal drives expression of rn, at a time when neither is normally expressed.

Figure 3. tal Has a Morphogenetic Function

Optical sections of the third instar leg imaginal discs. The discs are shown in a side view with dorsal up and distal to the right, and the tissue morphology is revealed by phalloidin-rhodamine (red) staining of the actin cytoskeleton and anti-β-integrin (green, yellow overlap) staining of basal membranes. The position of the tarsal fold (ventral side) is indicated with an arrowhead. (A–A'') Morphological changes in a wild-type leg disc. At 84 h, the tarsal fold starts to form as an apico-basal constriction of the epithelial cells. At 96 h, this constriction is followed by invagination of the cells. At 110 h, cells that originated in the tarsal fold form secondary folds that constitute the primordia of the tarsal segments. (B–B'') In a tal¹ mutant, the original tarsal constriction forms as in the wild type, but the tarsal fold never forms, and basal integrin staining remains stronger than in the wild type.

Figure 4. tal Is Required for Embryonic Development

(A–D) Expression of tal RNA throughout embryogenesis. (A) Expression of tal starts in seven blastodermal stripes and a cluster of cells in the anterior part of the embryo. (B) This expression refines to the tracheal precursors by the extended germ band stage. (C and D) Later, tal is present in the dorsal tracheal trunks (dt), posterior spiracles (ps), pharynx (ph), hindgut (hg), and presumptive denticle belts (db). (E) Dorsal tracheal trunks (dt) in a stage 16 wild-type embryo (dorsal view) revealed by the detection of the chitin binding protein. (F) Gaps in the dorsal tracheal trunks of a tal mutant. (G) Wild-type embryo cuticle: cephalopharyngeal skeleton (cps), ventral denticle belts (db), and posterior spiracles (ps). (H) In tal null mutants, these cuticular structures are missing or reduced. (I) Ectopic tal expression in the head produces extra cephalopharyngeal skeleton (ventral view; inset shows lateral view). (J) Wg protein distribution in the epidermis is normal in an extended germ band tal mutant embryo. (K) Expression of a shaven-baby reporter gene in ventral epidermis is not affected in a stage 17 tal mutant embryo. tal expression is not affected either in svb mutants (unpublished data). (L) Ventral view of the anterior-most segments of a stage 16 wild-type embryo. The denticle cells of the epidermis accumulate tubulin bundles prior to any denticle cuticle structures being observed. (M) In tal mutants, these tubulin bundles do not form.

Figure 5. The tal Transcript in Drosophila and Other Species

(A) LP10384 cDNA sequence with conceptual ORF translation; putative peptide identity is indicated on the right. Kozak consensi surrounding the start codons are underlined. Conserved domains in the type-A peptides are in bold type. (B) Graphic representation of tal and its homologs in other species, represented either by cDNAs (arrow ends) or genomic sequences (blunt ends). Type-A ORFs are represented by red boxes, and ORF-B by blue boxes. The tal gene family is at least 440 million years old and includes divergent orthologs and paralogs with different numbers of type-A ORFs. Note also that the gene duplication events in Bombyx and Lutzomia are independent. The ancestral gene had only two type-A ORFs, as shown by crustaceans and primitive insects.

Figure 6. Directed Mutagenesis and Translation of tal

(A) In these constructs, the coloured boxes indicate ORFs, and deletions are represented as empty segments. Engineered new ORFs are represented by bridged boxes. The UAS-tal construct contains the full cDNA and produces the complete rescue of the tal phenotypes and ectopic effects shown in Figures 1 and 4. Construct AB comprises one type-A ORF and one ORF-B, and produces the same functional effects. Construct delA has no type-A ORF and produces no effects. ATG-B forces translation of ORF-B, but still shows no effects. NoB has a mutation of the putative start codon of the ORF-B (empty box), thus preventing its translation, and produces the same functional effects as UAS-tal. delB has ORF-B deleted and is also fully functional. The 1A construct, which consists of the AB construct plus the deletion delB, contains only one type-A ORF and mimics the tal functional effects. In the construct 1A-FS, a single G was introduced after the start codon, causing a frameshift, which would result in the translation of a spurious 13-codon ORF (purple box). This construct is not functional. The Bm-wds construct contains one of the Bombyx tal full-length cDNAs and mimics the Drosophila UAS-tal results. (B) UAS-tal-GFP constructs tagging different ORFs, showing the in-frame insertion of the GFP coding sequence (green) (C) Peptides of expected size produced in vitro by Luciferase (61 kDa, control), 1A-GFP, 2A-GFP (28.4 kDa), and AA-GFP (33.1 kDa), but none by B-GFP (expected size, 31.6 kDa) The amount of protein produced seems to decay from 5′ to 3′ according to the ORF position, ORF 1A being the highest, and ORF AA the lowest. (D) UAS-tal-GFP constructs transfected into S2R+ cells. 1A-GFP, 2A-GFP, 3A-GFP, and AA-GFP (green) are detected, but not B-GFP. DAPI labels nuclei (blue), and nuclear DsRed transfected cells (red).