Ciliary defects in dila mutant flies. (A) The number of flies climbing above a threshold in a bang test. Very few dila mutant flies show climbing ability compared with wild-type flies (OreR). ato mutant flies, which lack Ch organs, also fail this test. ***P<0.001; *0.01<P<0.05 (Student's t-test). (B) Schematic of a Ch scolopidium, showing the locations of transverse sections. Shown are ciliary rootlet (cr), proximal basal body (pb), distal basal body (db), transition zone (tz) and proximal cilium (pc). (C–K) Electron micrographs of transverse sections through antennal Ch scolopidia. (C) Wild-type scolopales (sc) contain cilia from two or three neurons. (D) Cilia in dila⁸¹ Ch organs are variably disrupted. Some cilia are absent in cross-section, indicating truncation (asterisk), whereas others have an abnormal axoneme structure. (E) Transverse section through a wild-type proximal cilium (pc) showing the 9+0 axoneme with dynein arms (indicated). (F) These are missing in the mutant. (G,H) dila mutant cilia showing disorganisation or reduction of microtubule doublets. (I) A dila mutant distal basal body (db) appears normal (transitional fibres indicated). (J) Wild-type putative transition zone (tz) showing electron-dense axoneme ring (possible fibrous appendages indicated). (K) Mutant transition zone. (L–O) Longitudinal sections through antennal Ch scolopidia. (L) Wild-type cilium base. (M,N) dila mutant showing grossly intact basal structures. (O) Grossly malformed cilia in a dila mutant.

Defects in sperm formation. (A) Progeny produced by single wild-type (OreR) and PNS-rescued dila mutant males when crossed to wild-type females. (B) Wild-type testis with organised and orientated sperm bundles. Nuclei are labelled with propidium iodide (PI, magenta) and flagella with β-tubulin (green). (C) A wild-type cyst of elongating spermatids, showing nuclei (PI, magenta) and their apical centriolar adjuncts are labelled with γ-tubulin (green). (D) In dila⁸¹ cysts of elongating spermatids, the nuclei are misorientated and disorganised with many bundles and centriolar adjuncts misplaced (arrows). (E) In other cases the cyst is disrupted with scattered nuclei that often lack γ-tubulin staining (arrow). (F) In wild-type cysts GFP–PACT labels the rod-shaped centriole. (G) In dila mutants GFP–PACT is irregularly localised. (H) β-Tubulin staining of flagella shows chaotic organisation of mutant sperm. (I–K) Transverse sections through elongating spermatids from dila mutant testis. (I) In many cases spermatids fail to individualise and are chaotically organised, and many show separation of the axonemes from their mitochondria (arrows). (J) A relatively well-formed cyst shows some individualisation failures (lack surrounding membranes; straight arrows). Axonemal polarity differences (curved arrows) are consistent with differing orientations of sperm within the cyst or looping of sperm flagella. (K) Higher magnification shows well-formed axonemal structures of both individualised (upper) and non-individualised (lower) sperm.

DILA-FLAG is associated with centrosomes in dividing Ch precursor cells. (A–A″) A dividing Ch sense organ precursor expressing DILA–FLAG (scaGal4, UAS-dila-FLAG). Some of the protein (green) colocalises with the pericentriolar marker, CP190 (magenta).

The dila gene and its expression. (A) Molecular map of the dila (CG1625) region from FlyBase and our RT-PCR studies. Indicated are the transcripts, the probe used for in situ hybridisation (blue line) and the P-element insertion site within the adjacent CG30001 gene, from which three excision deletion alleles were generated. Each deletion removes part of the 3′ end of the dila gene, including the coiled-coil motifs. (B–E) mRNA in situ hybridisation with dila probe. (B) At stage 11 dila is expressed first in Ch neural precursors. (C) Expression continues in both Ch and ES sensory neuron lineages at stage 14. (D) At stage 15 to 16 expression persists in the Ch and ES neurons. (E) In the distal adult testis dila mRNA is located in the heads of each individual cyst, where the nuclei of spermatids cluster.

Protein localisation in dila mutant cilia. (A,B) Wild-type embryonic Ch neurons stained with anti-HRP antibody, showing expression in the cilia (A) and subsequently in the scolopale (sc) lumen band (B). (C) In the dila⁸¹ mutant, anti-HRP antibody reactivity largely remains at the base of truncated cilia. (D,E) Expression of HRP and EYS in larval Ch neurons. (D) Wild-type. (E) dila mutant. (F) Schematic summary of EYS and HRP staining. IS, inner segment; OS, outer segment.

DILA–FLAG resides at the base of the cilium. (A) When expressed in larval Ch neurons, DILA–FLAG is localised at the basal body (bb) region, as indicated by anti-HRP antibody staining. The DILA–FLAG staining in the distal cilia (*) varies with the expression level and is presumed to be an artifact of protein overexpression. (B–B″) Sas-4 (magenta) reveals the distal and proximal basal bodies (db, pb) at the base of each cilium. DILA–FLAG staining overlaps with Sas-4 staining and is especially prominent in the cilium beyond the distal basal body, predicted to be the putative transition zone (tz). (C) When expressed at a low level, DILA–FLAG is confined to the putative transition zone. (D) Schematic of the cilium base showing the distribution of proteins.

Co-dependent localisation of DILA–FLAG and UNC–GFP. (A–F) Larval Ch neurons. (A) The UNC–GFP fusion protein (green) is detected in the region just above the distal basal body, as marked by Sas-4 expression (magenta). This is predicted to be the transition zone. (B) UNC–GFP in wild-type flies, co-stained with anti-HRP (magenta). (C) In dila⁸¹ mutants, UNC–GFP is absent. (D) In unc²⁴ mutants, DILA–FLAG staining is weak and diffuse, with little detectable in the basal body region (compare with Fig. 5B), except for some abnormal aggregations in one dendrite. By contrast, Sas-4 is still localised in dila (E) and unc (F) mutants. (G) The uncoordinated phenotype of the weak hypomorph dila²⁴⁴ is enhanced after mutation of one copy of unc. The phenotype was scored according to three classes of increasing severity of uncoordination (see Materials and Methods). The difference in distributions is significant according to a χ-squared test, P<0.001.