Forkhead transcription factor Fd3F cooperates with Rfx to regulate a gene expression program for mechanosensory cilia specialization

Abstract

Cilia have evolved hugely diverse structures and functions to participate in a wide variety of developmental and physiological processes. Ciliary specialization requires differences in gene expression, but few transcription factors are known to regulate this, and their molecular function is unclear. Here, we show that the Drosophila Forkhead box (Fox) gene, fd3F, is required for specialization of the mechanosensory cilium of chordotonal (Ch) neurons. fd3F regulates genes for Ch-specific axonemal dyneins and TRPV ion channels, which are required for sensory transduction, and retrograde transport genes, which are required to differentiate their distinct motile and sensory ciliary zones. fd3F is reminiscent of vertebrate Foxj1, a motile cilia regulator, but fd3F regulates motility genes as part of a broader sensory regulation program. Fd3F cooperates with the pan-ciliary transcription factor, Rfx, to regulate its targets directly. This illuminates pathways involved in ciliary specialization and the molecular mechanism of transcription factors that regulate them.

Graphical abstract

Figure 1

fd3F Is Expressed Specifically in Developing Ch Neurons (A) Schematic arrangement of ciliated sensory neurons in an embryonic abdominal segment. Ch neurons, green; ES neurons, blue. (B) Schematic of the array of Ch neurons in the adult antenna that form Johnston's organ. (C) Schematic of a unit Ch organ from Johnston's organ, housing two Ch neurons. (D) fd3F mRNA in stage 14 embryos. (E) Fd3F protein localized in a subset of sensory neurons (marked by 22C10) corresponding to Ch neurons (stage 15 embryo). (F) Leg imaginal disc, Fd3F protein in Ch precursor cells of the femoral chordotonal organ (fco). Scale bars, 100 μm (D and F) and 20 μm (E).

Figure 2

Mutation of fd3F Results in Functionally Defective Ch Neurons (A and B) Cell bodies and dendrites of Ch neuron group (lch5) in third-instar larvae stained with anti-HRP. (C and D) Adult femoral Ch organ neurons showing sensory cilia (arrows) labeled with mCD8-GFP (elavGal4;UAS-mCD8-GFP).Scale bars, 5 μm. (E) Climbing assay showing percentage of adult flies climbing above a threshold. Genotypes are wild-type (w¹¹¹⁸), flies homozygous for the P element insert at fd3F (p{EP}), fd3F¹ heterozygotes (fd3F¹/FM6 balancer), and fd3F¹ homozygotes. Error bars are SD. ^∗∗∗p < 0.0001. (F) Shown on the left are relative CAP amplitudes recorded from the projections of Johnston's organ neurons in the antennal nerve (inset) as a function of the sound particle velocity. Shown on the right are corresponding absolute values of the maximum CAP amplitudes (n = 5 flies per strain). Error bars are SD. (G–L) Genes that show reduced expression in fd3F mutant embryos. Wild-type (left) and mutant (right) embryos showing mRNA expression of Ch-expressed genes (named at left) are shown. Expression is either virtually absent (e.g., Dhc93AB) or reduced (e.g., CG11253). Scale bars, 100 μm. See also Figure S1.

Figure 3

fd3F Regulates Genes Required for Ch Ciliary Motility (A) Shown on the left is a tone-evoked antennal displacement amplitude as a function of the sound particle velocity at the arista tip (inset, red arrow). Green lines indicate linearity, red lines nonlinearity. The orange arrows highlight the nonlinear sensitivity gain. Shown on the right is nonlinear sensitivity gain. A gain of one (hatched horizontal line) signals the absence of amplification. Error bars are SD. (B–G) mRNA expression of axonemal dynein genes in wild-type and fd3F mutant embryos. (B, D, and F) In wild-type embryos, these genes are Ch specific but rather weakly expressed (Ch cell transcriptome analysis supports this, Table S1). (C, E, and G) Expression is strongly reduced in fd3F mutant embryos. (H and I) TEM of Ch cilium cross-sections from adult antennae (Johnston's organ), showing the nine axonemal microtubule doublets. Axonemal dynein arms can be seen in wild-type (H, arrows), but not in fd3F mutant (I). Insets show this phenotype schematically. (J) Lower-magnification view of Ch units with the cilium in (I) ringed. The schematic shows the approximate location of the cross-sections. (K–N) Expression of genes required for axonemal dynein assembly or function shows Ch-specific (albeit weak) expression (K and M) that is strongly reduced in fd3F mutants (L and N). Scale bars, 100 μm. See also Table S1.

Figure 4

fd3F Regulates Genes Required for Ciliary Compartmentalization (A–D) mRNA expression of retrograde IFT-related genes in wild-type and fd3F mutant embryos. These genes show a Ch-enriched expression pattern of which the Ch neuron component (ringed and marked with arrows) is reduced in fd3F mutants. (E and F) The expression of an anterograde IFT gene, CG15161 (in a pattern consistent with expression in all ciliated sensory neurons), is unaffected in fd3F mutant embryos. Scale bars, 20 μm. (G and H) TEM cross-section of the tip of a Ch organ unit from adult antennal Johnston's organ. The Ch cilium tips (ci) are housed within the dendritic cap (c) (schematic shows the location of the cross-section). As shown in (G), the cilium tips are enlarged to fill the cap in the fd3F mutant antenna (H). (I and J) NompB-GFP fusion protein accumulates at the tips of Johnston's organ Ch neurons in fd3F mutant pupal antennae (cell bodies counterstained with 22C10). (K) In wild-type embryonic Ch neurons, anti-HRP marks the Ch cilia including the ciliary dilation (arrowheads). (L) The dilation is not apparent in Ch cilia from fd3F mutant embryos (arrowheads indicate the expected location of the dilation). (M–P) NompC protein is localized to the distal Ch cilia in wild-type neurons (M and O) but becomes dispersed along the cilium in fd3F mutant neurons (N and P). Lines indicate the approximate extent of the cilium. (Q) RempA-YFP protein is localized in the vicinity of the ciliary dilation in wild-type Ch cilia. (R) In fd3F mutant cilia, RempA-YFP protein becomes dispersed. (S) Schematic summary of protein localizations in wild-type and mutant Ch neuron cilia. Scale bars, 4 μm (I, J, and O–R) and 5 μm (K–N). See also Table S1.

Figure 5

Fd3F Directly Regulates nan and iav with Rfx (A) Schematics of the nan and iav genes showing locations of Fox motifs (F) and X boxes (X) in their upstream regions. The sequences surrounding the X boxes are shown, with asterisks representing identity in D. pseudoobscura. (B and C) Stage 16 embryos showing expression of GFP in Ch neurons for nan-GFP (B) and iav-GFP (C) reporter genes. (D and E) Expression of both reporter genes is lost in fd3F mutant embryos. (F and G) Expression of both reporter genes is abolished upon mutation of a single Fox motif (F1) in either upstream region. (H) Gel mobility shift assay with oligonucleotide probes (^∗) containing the nan-F1 or iav-F1 Fox motifs and purified Fd3F forkhead domain polypeptide (Fd3F^Fkh). The DNA-Fd3F^Fkh complexes are indicated. Addition of 100-fold excess of cold competitor oligonucleotide (nan-F1, iav-F1) reduces complex formation, but when the Fox motif is mutated (nan-F1m, iav-F1m), the cold competitor has no effect on complex formation. (I and J) Embryonic expression of iav in Ch neurons (I) is lost in Rfx⁴⁹ homozygous mutant embryo (J) (embryos stained in parallel). (K–N) Two abdominal segments from stage 16 embryos stained with anti-GFP (green) and 22C10 (magenta). Ch neuron expression of nan-GFP (K) and iav-GFP (M) reporter genes is abolished when the upstream X box is mutated (L and N). Scale bars, 100 μm (B–J) and 20 μm (K–M). See also Table S1.

Figure 6

Promoter Regions of Fd3F/Rfx Target Genes Contain a Paired Fox Motif/X Box Combination (A) Schematics of example gene regions (first exons and upstream regions only) with locations of X box/Fox motif pair. Below each is a screenshot from the UCSC genome browser (http://genome.ucsc.edu/) representing the sequence alignment of 12 Drosophila species and showing degree of identity as a histogram. X boxes and Fox motifs are outlined in blue and green, respectively. (B) Sequence logos (http://weblogo.threeplusone.com) summarizing the alignment of X boxes (left) and Fox motifs (right) from all identified fd3F target genes (see Table S2). The representations of information content (bits) emphasize the fact that the X boxes often comprise one strong half-site and a more degenerate one. (C–F) Embryos containing a Dhc93AB-GFP reporter gene. (C) Stage 16 embryo showing specific expression of Dhc93AB-GFP in Ch neurons. (D) Mutation of a proximal X box strongly reduces expression. (E) Mutation of Fox motif F1 reduces expression. (F) Mutation of Fox motif F2 has no discernable effect. Scale bars, 100 μm. See also Table S2.

Figure 7

Ectopic Activation of Target Genes by Fd3F Misexpression (A and B) Expression of Fd3F in embryos from control (UAS-fd3F, no driver) and misexpressing (scaGal4, UAS-fd3F) lines. (C–H) Expression of fd3F target genes. In (H), some of the ES neurons exhibiting ectopic expression of Tektin-A are indicated (cf. Figure 1A). (I and J) Expression of nan-GFP reporter gene. (K and L) Expression of Dhc93AB-GFP reporter gene. ES neurons exhibiting ectopic expression are labeled in (L). (M) Summary of target gene regulation by fd3F and Rfx is shown. In wild-type Ch neurons, Fd3F and Rfx cooperate to regulate a subset of Ch-specific and Ch-enriched genes required for ciliary specialization. In ES neurons, and in fd3F mutant Ch neurons, Rfx does not activate the Ch-specific genes but is able to activate Ch-enriched genes at a low level for “basal” retrograde transport. In Rfx mutants, Fd3F is unable to activate these genes alone. pan-ciliary genes are regulated by Rfx in Ch and ES cells. Note that in addition to these interactions, it is possible that ES neurons express their own ciliary specialization regulators, so that fd3F mutant Ch cilia do not “revert” to an ES cilium state. Scale bars, 100 μm (A–J) and 20 μm (K and L).