DYX1C1 is required for axonemal dynein assembly and ciliary motility

Abstract

DYX1C1 has been associated with dyslexia and neuronal migration in the developing neocortex. Unexpectedly, we found that deleting exons 2-4 of Dyx1c1 in mice caused a phenotype resembling primary ciliary dyskinesia (PCD), a disorder characterized by chronic airway disease, laterality defects and male infertility. This phenotype was confirmed independently in mice with a Dyx1c1 c.T2A start-codon mutation recovered from an N-ethyl-N-nitrosourea (ENU) mutagenesis screen. Morpholinos targeting dyx1c1 in zebrafish also caused laterality and ciliary motility defects. In humans, we identified recessive loss-of-function DYX1C1 mutations in 12 individuals with PCD. Ultrastructural and immunofluorescence analyses of DYX1C1-mutant motile cilia in mice and humans showed disruptions of outer and inner dynein arms (ODAs and IDAs, respectively). DYX1C1 localizes to the cytoplasm of respiratory epithelial cells, its interactome is enriched for molecular chaperones, and it interacts with the cytoplasmic ODA and IDA assembly factor DNAAF2 (KTU). Thus, we propose that DYX1C1 is a newly identified dynein axonemal assembly factor (DNAAF4).

Figure 1. Deficiency of Dyx1c1 in mouse causes phenotypes consistent with motile cilia defects

(a) Schematic of the strategy for producing the Dyx1c1 exon 2–4 deleted allele. A conditional knockout allele was inserted by homologous recombination and then exons 2–4 were deleted from the germline by breeding mice to an Hprt-Cre line of mice. (b) Left panel, PCR indicating the genotyping results from the wild-type (F1/R1, 437bp) and deleted alleles (F1/R2, 356bp). Right panels, Western blots show the presence of Dyx1c1 protein detected in brain and lung of wild-type (Dyx1c1^+/+), and heterozygous animals (Dyx1c1^Δ/+), but missing from homozygous mutants (Dyx1c1^Δ/Δ). (c) Evidence of hydrocephalus in Dyx1c1^Δ/Δ mice. Characteristic domed crown in Dyx1c1^Δ/Δ mice develops by P16 with the ventricles seen in coronal sections are greatly expanded in Dyx1c1^Δ/Δ relative to Dyx1c1^+/+ mice. Scale bar, 250mm (d) Situs inversus in Dyx1c1^Δ/Δ mice as seen in both the reversal of the milk filled stomach in neonates to the right side, and exposed viscera showing reversal of multiple organs including heart (He), stomach (St), and spleen (Sp). (e) Whole mount in situ hybridization analysis of Dyx1c1 in mouse embryos at E7.5. Dyx1c1 expression is restricted to the pit cells of the ventral node (upper panels: Dyx1c1 antisense probe, asterisk marks location of node; lower panels: Dyx1c1 sense control probe; upper and lower left panel: lateral view from the left; upper and lower right panels: frontal view). A, anterior; L, left; P, posterior; R, right; Scale bar, 500µm

Figure 2. Knock down of dyx1c1 in D. rerio

(a) Phenotypes of 48 hpf zebrafish embryos, with AUG-morpholinos (MO) injected between the one- and four-cell stage. Top, lateral images of wild-type and dyx1c1 morphant zebrafish embryos at 48 hpf. The morphants have a curly tail down phenotype and exhibit hydrocephalus (asterisk in enlarged image) as well as pronephric cysts (arrow in enlarged image). Middle, ventral views of the heart and bottom, dorsal views of the liver and pancreas in dyx1c1 morphant embryos at 48 hpf. The heart, liver and pancreas were visualized by in situ hybridizations for cmlc2, fkd2 and ins respectively. L, left; R, right; V, ventricle; Atr, atrium; Li, liver; P, pancreas. Scale bars, 500µm (b) Top, graph shows the defects in visceral asymmetry in 48 hpf dyx1c1 morphant embryos compared to uninjected controls. The laterality defects are associated with 2 ng of dyx1c1 AUG-MO. Bottom, graph showing alterations in the asymmetric gene expression of southpaw in the LPM of dyx1c1 morpholino injected embryos compared to uninjected controls at three developmental stages during 15–20 hpf.

Figure 3. Motile cilia are dysfunctional in Dyx1c1 mutant mice

(a) H&E stained sections of the cerebral ventricles show the presence of cilia on the surface of the ependymal cells in both Dyx1c1^+/+ and Dyx1c1^Δ/Δ mice. (b,c) Immunofluorescence analyses of lung sections stained for acetylated tubulin (green), and the outer dynein arm heavy chain Mdnah5 (b, red) and inner dynein arm light chain Dnali1 (c, red). Nuclei were stained with DAPI. In contrast to Dyx1c1^+/+ mice where Mdnah5 and Dnali1 co-localized with acetylated tubulin (yellow, upper panels in b and c), in Dyx1c1^Δ/Δ Mdnah5 and Dnali1 (lower panels in b and c) were completely absent from ciliary axonemes. (d) Flow of fluid (Indian ink) across the ependymal surface in brain ventricle cup preparations from a Dyx1c1^+/+ and Dyx1c1^Δ/Δ at mice P6. Directional flow was rapid across the surface of Dyx1c1^+/+ ependymal epithelia, while only non-directional passive diffusion was observed on ependymal surfaces in Dyx1c1^Δ/Δ. The starting point for the fluid is at the end of the pipette tip, seen bottom right. (e) TEM images of cross-sections through the trachea of wild-type and Dyx1c1^Δ/Δ. Abundant cilia were present in each, but cilia structures (red arrows) in the Dyx1c1^Δ/Δ trachea are surrounded by cellular debris and mucus. (f) TEM cross section images of ependymal and tracheal cilia in Dyx1c1^+/+ and Dyx1c1^Δ/Δ mice. The 9+2 microtubular structure was well preserved in Dyx1c1^Δ/Δ, except that the outer and inner dynein arms were lacking in tracheal cilia. The scale bars represent in (a,b,c,d) 10µm, in (e) 1µm (left panel) and 2µm (right panel) and in (f) 0.1 µm.

Figure 4. DYXC1 mutations in human PCD patients cause defective ODA and IDA assembly

(a) Schematic presentation of chromosome 15 and the genomic structure of DYX1C1. The positions of 8 of the identified mutations are indicated by black arrows, the position of the 3.5kb deletion is indicated by a rectangle. (b) Schematic showing the relative positions of seven DYX1C1 nonsense mutations identified in PCD patients and families in the DYX1C1 coding sequence. All mutations are clustered in the middle of DYX1C1 coding sequence and each mutation predicts to cause a premature stop prior to the tetricopeptide repeat domains (TPR) at the C-terminus of DYX1C1 (CS: p23-like C-terminal CHORD-SGT1 domain). (c) Transmission electron micrographs showing defects of outer and inner dynein arms in four PCD individuals with DYX1C1 mutations compared to a control without PCD. Rarely, outer dynein arms can be seen in cilia of the affected patient (OP-86 II2, red arrow). Scale bar, 0.2µm. (d) Respiratory epithelial cells from control and PCD patient OP-556 II2 were double-labeled with antibodies directed against acetylated tubulin (green) and DNAH5 (red). Both proteins colocalize (yellow) along the cilia in cells from the unaffected controls. In contrast, in patient cells, DNAH5 was absent from or severely reduced in ciliary axonemes (Supplementary Fig. 4). (e) Aberrant sublocalization pattern of the outer dynein arm heavy chain DNAH9 in cilia of respiratory epithelial cells from control and PCD patient F648 II1. Cells were double-labeled with antibodies directed against acetylated tubulin (green) and DNAH9 (red). Acetylated tubulin localizes to the entire length of the cilia, whereas DNAH9 localization is restricted to the distal part of the cilia. In contrast, in DYX1C1 mutant cells DNAH9 was completely absent from ciliary axonemes. (f) Respiratory epithelial cells from control and PCD patient OP-86 II2 and OP-359 II1 were double-labeled with antibodies directed against acetylated tubulin (green) and DNALI1 (red). Both proteins colocalize (yellow) to the ciliary axonemes in cells from an unaffected control, while DNALI1 was absent from the ciliary axonemes in DYX1C1 mutant cells. In d, e, and f nuclei are stained with Hoechst33342 (blue). Scale bar, 10µm.

Figure 5. DYX1C1 is localized in the cytoplasm of respiratory epithelial cells and interacts with DNAAF2/KTU

(a) Immunofluorescence analyses of mouse nasal epithelial cells stained for acetylated tubulin (green) and Dyx1c1 (red)m. In Dyx1c1^+/+ mice, Dyx1c1 localizes to the cytoplasm of the epithelial cells and partly to the basal bodies but is absent in the cilia. In the mutant, Dyx1c1 is absent from the cytoplasma. Nuclei were stained with Hoechst 33342 (blue). (b) Immunoblots performed with different lysate fractions (a, cytoplasmic and axonemal) demonstrate that DYX1C1 (right panels), as well as DNAAF2 (middle right panels), shows a strong signal in the cytoplasmic fraction but is almost absent in the axonemal fraction. LRRC48/DRC3 (middle left panels) was used as an axonemal control. Silver staining of the loaded lysates is shown on the left panels. (c) HEK293 lysates coexpressing myc-DYX1C1 and FLAG-DNAAF2 were immunoprecipitated with either rabbit control IgG or rabbit anti-DNAAF2 antibody. Western blotting with mouse anti-myc demonstrates that myc-DYX1C1 is efficiently immunoprecipitated by DNAAF2 (top panel), and Western blotting with mouse anti-FLAG confirms that FLAG-DNAAF2 is recovered in the immunoprecipitate (bottom panel) as compared to the control immunoprecipitation. Equal volumes (12 µl) of lysate and immunoprecipitate fractions were loaded on the same gel; lysate fractions represent 0.7 % of total lysate (1 ml volume) and immunoprecipitate fractions represent 1/15 lysis volume (33 µl resuspension). Magic Mark protein ladder (M) was used to estimate molecular weight of myc-DYX1C1 and FLAG-DNAAF2. The observed molecular weights of myc-DYX1C1 and FLAG-DNAAF2 are higher than the expected molecular weights of 48.5 and 91 kDa due to additional sequence from myc and FLAG epitope tags, respectively. (d) Yeast two-hybrid assay using BD-tagged DYX1C1 and AD-tagged DNAAF2 demonstrates a binary interaction between DYX1C1 and DNAAF2. Binary interactions were identified by yeast growth on media lacking adenine and histidine to select for HIS3 and ADE2 reporter gene activation (left panel). Interactions were additionally validated by evaluation of LacZ reporter gene activation (β-galactosidase colorimetric filter lift assay, right panel). Binding of BD-DYX1C1 and AD-DNAAF2 was validated by using the known interactors BD-USH2A_icd and AD-NINL_isoB as a positive control, and BD-USH2A_icd and AD-GAL4 as a negative control.