Sense and antisense iab-4 miRNAs are functional inhibitory RNAs that exhibit target selectivity. Shown are pouch regions of wing imaginal discs that carry the indicated tub-GFP sensors and express either UAS-DsRed-mir-iab-4 under ptc-Gal4 control (A,C) or UAS-DsRed-mir-iab-8 under dpp-Gal4 control (B,D). (A′,D′) Evidence for strong target inhibition is seen in the GFP-negative regions designated with asterisks; arrow in C′ indicates weak target inhibition.

Drosophila mir-307 and mammalian mir-338 produce antisense miRNAs. (A) Multiple alignment of the insect mir-307 locus (obtained from the University of California at Santa Cruz Genome Center, http://genome.ucsc.edu). Previously cloned miR-307 derives from the bottom strand (miR-307, “3p” arm). Both arms of the hairpin have been highly conserved during insect evolution (blue-shaded boxes), but miR-307 has diverged slightly in Apis (yellow, arrow). Below this is the DNA sequence of the mir-307 hairpin, color-coded for cloned sense (in red) and antisense (in green) miRNAs. (B) Multiple alignment of the vertebrate mir-338 locus. Previously identified miR-338 derives from the bottom strand. Both arms of the hairpin are strongly conserved in placental mammals. Below this is the DNA sequence of the mir-338 hairpin, color-coded for cloned sense (in red) and antisense (in green) miRNAs.

Comparison of miRNA precursors as sense and antisense transcripts. (A) Sense and antisense miRNA hairpins of known Drosophila miRNAs were assessed according to their free energy and structural parameters (Lai et al. 2003). The X-axis denotes each miRNA hairpin folded and scored on its sense strand (blue) and antisense strand (purple); the Y-axis denotes the score of the miRNA candidate. Most antisense strands are poorer candidates than their sense counterparts (e.g., mir-133); however, nearly one-third of them (gray box) achieve a score that is plausible as a miRNA candidate (e.g., mir-307). In a few cases, such as mir-4, the antisense strand is more compelling than the sense strand. The full analysis is available in Supplemental Figure S4. (B) Example of a miRNA locus that adopts comparable structures on both strands. (C) Example of a miRNA locus whose antisense structure is dissimilar to its sense counterpart.

Ectopic mir-iab-8 inhibits endogenous Ubx and abd-A proteins and induces homeotic phenotypes. (A–D) FLP-out clones expressing sense or antisense iab-4 miRNAs were assayed for their effect on Ubx or abd-A. A–D show DsRed/miRNA-expressing clones (in red), A′–D′ depict endogenous Hox proteins (in green), and A″–D″ display merged images; selected clone boundaries that overlap domains of Hox protein expression are marked with dotted lines. Ectopic mir-iab-4 reduces Ubx (A) but has no effect on abd-A (C), while ectopic mir-iab-8 nearly eliminates both Hox proteins (B,D). (E–G) Haltere-to-wing respecification induced by sense and antisense iab-4 miRNAs, characteristic of Ubx loss of function. (E) Wild-type thorax. The wing is marked with an asterisk and the haltere is marked with an arrow; inset shows a higher-magnification view of the haltere. (F) sd-Gal4/X; UAS-DsRed-mir-iab-4/+ animal exhibits mild haltere-to-wing transformation including extensive specification of anterior wing margin bristles (see inset); the endogenous wing has not yet inflated. (G) sd-Gal4/X; UAS-DsRed-mir-iab-8/+ pharate lethal animal exhibits a nearly complete transformation of haltere toward wing identity; the morphology of both wings is abnormal.

Ubx and abd-A are strong, direct targets of miR-iab-8-5p. (A,B) Alignments of the Ubx and abd-A 3′ UTRs across 12 Drosophilid genomes (obtained from the University of California at Santa Cruz Genome Browser, http://genome.ucsc.edu). The distal sites in abd-A are contained within a novel, extended isoform whose structure is supported by the indicated ESTs. (C,D) Canonical binding sites for sense and antisense iab-4 miRNAs in Ubx and abd-A. All sequences are from Drosophila melanogaster except for Ubx site #6; which is canonical only in nonmelanogaster group species (D. virilis [Dv] sequence is shown as a representative). Contiguous Watson-Crick seed-matches to miR-iab-4-5p are highlighted in green, while nucleotides that match the 5′ extension of miR-iab-8-5p are in red. The number of seed-matched nucleotides, t1A features (“A”), and depth of species conservation are tabulated to the right of each alignment. Seed-match data are highlighted green or red if predicted as a functional target of miR-iab-4-5p or miR-iab-8-5p, respectively; note that Ubx site #3 is a canonical binding site for both miRNAs. Additional conservation data are reported in Supplemental Figures S1–S3. (E,F) Examples of highly conserved miR-iab-8-5p/miR-iab-4-5p-binding sites in the 3′ UTRs of Ubx and abd-A. (G–J) Wing disc assays of the ability of sense and antisense iab-4 miRNAs to regulate Ubx or abd-A 3′ UTR sensors. Asterisks indicate strong regulation, while the arrow indicates weak regulation.

Sense and antisense transcription of the iab-4 locus in the BX-C produces mature miRNAs. (A) Map of the BX-C depicting the homeobox genes Ubx, abd-A, and Abd-B, and sense and antisense iab-4 hairpins; the Antp gene of the antennapedia complex (ANTP-C) is also shown. Targets with highly conserved “2–8” seed + t1A pairing to miR-iab-4-5p or miR-iab-8-5p are indicated; see Figure 3 and Supplemental Figures S1–S3 for details of target site pairing and conservation. Darker lines represent stronger regulatory relationships, as evidenced by multiple sites and experimental validation. Below this is the DNA sequence of the iab-4 hairpin region, with the sense-strand miRNAs shaded in green and antisense-strand miRNAs shaded in red. Alignment of mature miR-iab-4-5p and miR-iab-8-5p shows that the latter has a 2-nt “UU” extension at its 5′ end. (B) Proposed secondary structures of the mir-iab-4 and mir-iab-8 hairpins. (C–F) In situ hybridization demonstrates temporally and spatially distinct expression patterns of pri-mir-iab-4 and pri-mir-iab-8. (G) Northern analysis of RNA from 6- to 10-h embryos reveals processed pre-miRNA hairpins for mir-iab-4 and mir-iab-8, and mature miRNAs for iab-4-5p, iab4-3p, and iab-8-5p; iab-4-5p and iab-8-5p are predominant. Ethidium stains of 5S rRNA and Northern analysis of 2S rRNA species are shown as loading and transferring controls. (H) Tests of probe specificity. Titration series of synthetic mir-iab-4 and mir-iab-8 hairpin RNAs were analyzed using probes for all four of the iab-4 locus miRNAs. Both iab-4-5p and iab-4-3p probes recognize only the mir-iab-4 hairpin, while the iab-8-5p and iab-8-3p probes recognize only the mir-iab-8 hairpin.