IFT88 maintains sensory function by localising signalling proteins along Drosophila cilia

(A) Schematic presentation of the components of the IFT complex found in humans, Chlamydomonas (C. reinhardtii) and Drosophila (D. melanogaster). IFT consists of three complexes: IFT-A, IFT-B and the BBsome. IFT-A and –B are further divided based on their genetic features and biochemical properties. For the IFT-A and –B proteins, the Chlamydomonas nomenclature is used. Proteins conserved and not conserved in Drosophila are encircled by full and dashed outlines, respectively. The scheme also shows where the three complexes interact and highlights some known binding partners (green boxes). This scheme is modified from Mourao et al (2016). (B) Ultrastructural organisation of the cilium of chordotonal neurons in the fly second antennal segment, as seen by electron micrographs. Here, two neurons project their cilia towards a cap cell. Each ciliary shaft is divided by the ciliary dilation into a proximal and a distal zone, which have distinct axonemal architectures, with only the proximal zone harbouring axonemal dynein arms (arrowhead). The base of the cilia is formed by distal and proximal basal bodies and the rootlet (Jana et al, 2016).

(A, C, D, E, F, G, H) Representation of the driver’s expression in the experimental settings used in (C, D, E, F, G, H). (B) Schematic representation showing that the hairpin (ID: JF03080) affects both DmIFT88 isoforms by targeting a common exon. (C) Bar plots (Mean ± S.D.) of real-time PCR quantification of DmIFT88 isoform levels. mRNA was extracted from the antennae of flies expressing a hairpin against either mCherry (negative control) or DmIFT88. The hairpin expression was driven using the tubulin promoter (Gal4^Tubulin). The total mRNA was extracted three times per hairpin and each RNA sample was measured in triplicates. The data are normalised using three different housekeeping genes (Su(TpI), TBP and elF-1A) to account for potential transcriptional changes of housekeeping genes as a consequence of the knockdown. The P-values are calculated using Mann-Whitney (One-tailed) test (*P-value < 0.05). (D) Box plots of maximum sound-evoked compound action potentials of chordotonal neurons in the second antennal segment of wild type flies (Canton S; negative control) and DmIFT88 knockdown flies expressing constitutively DmIFT88-IR (Gal4^Tubulin > UAS-DmIFT88 RNAi). The sound responses are virtually abolished from the antenna of DmIFT88 knockdown flies (****P-value < 0.0001 from Mann-Whitney test). Action potentials are measured from 4–6 d old flies. Median ± full range of variation (minimum to maximum). (E, F, G, H) The flies with knockdown on DmIFT88 using ubiquitously promoter (Gal4^Tubulin) do not grow chordotonal cilia. (E, G) Schematic representation and the representative EM images of WT chordotonal cilia in cross (E) and longitudinal section (G). (F, H) Representative images of cross-sections (F) and longitudinal-sections (H) of the chordotonal cilia in the adult flies ubiquitously (Gal4^Tubulin) expressing mCherry-IR (negative control) or DmIFT88-IR. Note that upon ubiquitous knockdown of DmIFT88, chordotonal neurons form ciliary transition zones (red arrows) but fail to extend cilia.

Source data are available for this figure.

(A, B, C, D) Representation of the drivers’ expression in the experimental settings used the following sections (B, C, D) of the figure as mentioned in the scheme. (B) Negative gravitaxis behaviour of flies expressing the corresponding RNAis under Gal4^Chat19b TubGal80^ts driver but grown at 18°C, where the Gal4 is inhibited. (C) Bar plots (Mean ± SD) of real-time qRT-PCR quantification of DmIFT88 isoform levels. mRNA was extracted from the antennae of 9 d old female flies expressing a hairpin against either mCherry (negative control) or DmIFT88. The RNAi expression was driven via the conditionally expressed promoter (Gal4^Chat19b TubGal80^ts) by placing newborn flies at 29°C. (D) Left: Representative images of chordotonal neurons in the second antennal segment of flies with mCherry or DmIFT88 knockdown for 15 d. Neurons are visualised by co-expressing GFP together with the respective hairpin. Long term knockdown of DmIFT88 alters ciliary axoneme bending (scheme in the Middle): cilia in DmIFT88 knockdown flies project out at a different angle from the dendrites than cilia in the control flies, as shown in the box plots (right). Data are obtained from antenna whole mounts to avoid artefacts from the sectioning process. Each box plot corresponds to the value of 50 angles from five different antennae. (E) Left: Representation of the driver’s expression in the experimental setting used on the right. Right: Days-dependent changes in climbing behaviour at 29°C in control (mCherry) and DmIFT88 RNAi flies under Gal4^Iav driver. In (B, C, E), P-values are calculated using Mann-Whitney test (ns - P-value ˃ 0.05, *P-value < 0.05, **P-value ≤ 0.01, ***P-value ≤ 0.001) in Prism. (D) P-values were calculated using the Welch Two Sample t test in Prism (GraphPad).

Source data are available for this figure.

(A, B, C, D, E, F) Representation of the drivers’ expression in the experimental settings used the following sections (B, C, D, E, F) of the figure as mentioned in the scheme. (B) Negative gravitaxis (climbing) behaviour of flies expressing the corresponding RNAis under Gal4^Tubulin and Gal4^Chat19b drivers. Climbing behaviour was measured from either 0–1 d (for flies expressing RNAis under Gal4^Tubulin) or 9 d (for flies expressing RNAis under Gal4^Chat19b) old flies with appropriate genotypes. Median ± full range of variation (minimum to maximum). Each box-plot corresponds to a total of minimum 50 flies measured in sets of 10 animals each. The P-value is calculated using Mann-Whitney test (***P-value < 0.001) in Prism. (C) Bar plots (Mean ± SD) of real-time PCR quantification of DmIFT88 (RC) levels in different experimental conditions. mRNA was extracted from the antennae of flies expressing a hairpin against either mCherry (negative control) or DmIFT88. The hairpin expression was driven using the tubulin (Gal4^Tubulin) and Chat promoter (Gal4^Chat19b). RNAs were extracted from either 0–1 d (for flies expressing RNAis under Gal4^Tubulin) or 9 d (for flies expressing RNAis under Gal4^Chat19b) old flies with appropriate genotypes. The total mRNA was extracted ≥3 times per hairpin and each RNA sample was measured in triplicates. The data are normalised using a housekeeping gene (elF-1A) to account for potential transcriptional changes of the housekeeping gene as a consequence of the knockdown or temperature shift. The P-values are calculated using Mann-Whitney (One-tailed) test (ns - P-value ˃ 0.05, *P-value < 0.05, ***P-value < 0.001) in Prism. (D, E, F) The flies with knockdown on DmIFT88 using cholinergic neuron-specific promoter (Gal4^Chat19b) grow chordotonal cilia and show no visible ultrastructure defects. (D, E, F) Representative EM images of cross-sections of the scolopidia (D), longitudinal-sections of the chordotonal cilia (E) and cross-sections of the axoneme of the proximal region of the cilia (F) in the adult flies tissue-specifically (Gal4^Chat19b) expressing mCherry-IR (negative control) or DmIFT88-IR. Number of samples are mentioned on respective images (Biological repeats n = 2).

Source data are available for this figure.

(A) PSI-BLAST search was performed against the fly genome using the mouse Gucy2e protein sequence. All six-particulate guanylyl cyclases encoded in the Drosophila genome were found in the search (Morton, 2004). The similarity of the protein sequences to each other is shown on the left side (maximum-likelihood phylogenetic tree). Two soluble guanylyl cyclases were included as outgroups. A behavioural screen was conducted to determine if any of the identified enzymes could be involved in climbing (or gravity sensing), and the results are shown on the right side. The genes were knocked down using RNAi; a hairpin against mCherry was used as a negative control. The RNAi was driven using the cholinergic-neuronal driver (Gal4^Chat19b), and flies were raised at 29°C to increase the potential phenotype, as Gal4 is sensitive to temperature. Only fly lines from the TRiP (Transgenic RNAi Project) library were considered (see the Materials and Methods section). When possible, 2–3 RNAi lines were used against the same gene. Each boxplot corresponds to 60 flies measured in groups of 10 flies each. Note that the RNAis against the two soluble GCs did not impair the adult negative-gravitaxis behaviour, which could indicate that only the particulate GCs are involved in gravity sensing. (B) CG34357 gene has three annotated transcripts in the fly and two of those differ only in the length of the 3′-untranslated regions (RD and RC). We focused on the two longer transcripts as the third transcript (RB) leads to a small truncated protein that only contains the extracellular part of the protein and thus in principle should not be able to bind to the intracellular DmIFT88 protein. The different functional domains are colour-coded (legends are described in (C)). (D) The exons targeted by the hairpins used in (D) are indicated in the scheme. (B, C) Protein model is drawn analogously to vertebrate particulate guanylyl cyclase, and functional domains are indicated using the same colours as in (B). The signal peptide is not shown in the protein model because it should be cleaved off according to signal peptide prediction (Nielsen, 2017). (D) Climbing assays were used to determine the role of CG34357 in negative-gravity sensing (a mCherry RNAi was used as a negative control, for a detail genotype of flies, see Table S5). All RNAis were driven using the Gal4^Chat19b, and the flies were raised at 29°C to increase the potential phenotype. Each box plot corresponds to a total of 60 flies measured in sets of 10 flies each. In (A, D), the P-values are calculated using Welch corrected (One-tailed) unpaired t test (ns - P-value ˃ 0.05, *P-value < 0.05, **P-value ≤ 0.01).

Source data are available for this figure.

(A, B) Schematic representation of the enhancer line, and the strategy used to determine CG34357 expression. (A) The schematic representation of the P-element used in this study containing a Gal4 gene which was randomly inserted after CG34357 exon 2 (Hayashi et al, 2002) is shown in (A). (B) The upper box in (B) contains a schematic representation of a P{GawB} element showing two P-element arms (5′ AND 3′) flanking Gal4 and mini-white genes. (B) The lower box in (B) schematically shows how this P-element was used to determine the CG34357 expression pattern: presumably the same enhancers that activate CG34357 also activate the Gal4 gene, which then drives the expression of the GFP reporter gene inserted at the 3′-end of UAS. (C) Readout of CG34357 expression by the use of the enhancer line driving UAS-mCD8::GFP in L3 larvae. We find CG34357 expression in chordotonal neurons (Ich5, dashed box) and in others, including non-ciliated class-I dendritic arborisation neurons, in the peripheral nervous system (arrowhead). (D) Stills of L3 larvae Ich5 dendrite tips live videos. DmGucy2d::GFP (left) and T1-DmGucy2d::GFP (right) are visible in the dendrite of lch5 neurons in wandering L3 larvae when expressed under the Gal4^Iav driver, with a very high accumulation at the ciliary base and very faint localisation in the axoneme when signal is pushed (black arrowheads). Although T1-DmGucy2d::GFP is also clearly visible at the ciliary dilation (empty arrowheads), which implies that is transported through the axoneme at least until the dilation, the movement of trains of neither DmGucy2d::GFP nor T1-DmGucy2d::GFP constructs (unlike GFP::DmIFT88) were visible during the duration of the timelapse videos (3 min).

(A, B, C, D) Representation of the drivers’ expression in the experimental settings used the following sections (B, C, D) of the figure as mentioned in the scheme. (B) Time-dependent changes in climbing behaviour at 18°C in control (mCherry) and DmGucy2d RNAi flies under Gal4^Chat19b driver, at which temperature the Gal4 protein is inhibited. Each box-plot corresponds to a total of 60 flies measured in sets of 10 animals each. The data are fitted using linear regression, where the area around the curve represents the 95% confidence interval. The P-values are calculated using Mann-Whitney test (ns - P-value ˃ 0.05, **P-value ≤ 0.01). (C) Days-dependent changes in climbing behaviour at 29°C in negative control (mCherry) and DmGucy2d RNAi flies under Gal4^Iav driver. The P-value is calculated using two-way ANOVA test (ns - P-value ˃ 0.05, *P-value < 0.05, **P-value ≤ 0.01, and ****P-value < 0.0001) in Prism. (D) Climbing gravitaxis behaviour at 29°C in negative control (mCherry) and a different DmGucy2d RNAi (IR2) fly line (#105185-KK; VDRC) under the Gal4^Chat19b driver. Each box-plot corresponds to a total of ≥ 60 flies measured in sets of 10 animals each. The P-values are calculated using Mann-Whitney test (ns - P-value ˃ 0.05, **P-value ≤ 0.01).

Source data are available for this figure.

(A, B, C, D) Representation of the drivers’ expression in the experimental settings used the following sections (B, C, D) of the figure as mentioned in the scheme. (B) Bar plots (Mean ± SD) of real-time PCR quantification of DmGucy2d isoform levels. mRNA was extracted from the antennae of female 9 d old flies expressing a hairpin against either mCherry (negative control) or DmGucy2d. The hairpin expression was driven using the conditionally expressed promoter (Gal4^Chat19b TubGal80^ts). The P-values are calculated using Mann-Whitney (One-tailed) test (*P-value ≤ 0.05). (C) Left: Representative images of chordotonal dendrite and protruding cilium in the second antennal segment of flies with mCherry or DmGucy2d knockdown for 15 d. Neurons are visualised by co-expressing GFP under Gal4^Chat19b together with the respective hairpin. Statistical analysis indicates that long term knockdown of DmGucy2d does not affect ciliary axoneme bending, as shown in the all-range box plots where the normalised average cilia bending angles are represented. N = 3 repeat experiments. Images and data were obtained from antenna whole mounts after tissue clearing to avoid artefacts from the sectioning process. (D) Left: Representative immunofluorescence images of adult chordotonal cilia (with Inactive [Iav] and NOMPC in the proximal and distal zone of the cilia in red and blue, respectively) from flies with control (mCherry) and DmGucy2d RNAis after 9 d post-knockdown induction. Right: all-range box plots of the normalised average mean Iav signal along the proximal part of the axoneme in flies in the two genotypes at 9- and 15-d post-knockdown induction (N = 2 and N = 3 repeated experiments, respectively). Scale bar: 1 μm. In (C, D), the P-values were calculated using Mann-Whitney test (ns - P-value ˃ 0.05) in Prism.

Source data are available for this figure.