Dysfunction of the ciliary ARMC9/TOGARAM1 protein module causes Joubert syndrome

Abstract

Joubert syndrome (JBTS) is a recessive neurodevelopmental ciliopathy characterized by a pathognomonic hindbrain malformation. All known JBTS genes encode proteins involved in the structure or function of primary cilia, ubiquitous antenna-like organelles essential for cellular signal transduction. Here, we used the recently identified JBTS-associated protein armadillo repeat motif-containing 9 (ARMC9) in tandem-affinity purification and yeast 2-hybrid screens to identify a ciliary module whose dysfunction underlies JBTS. In addition to the known JBTS-associated proteins CEP104 and CSPP1, we identified coiled-coil domain containing 66 (CCDC66) and TOG array regulator of axonemal microtubules 1 (TOGARAM1) as ARMC9 interaction partners. We found that TOGARAM1 variants cause JBTS and disrupt TOGARAM1 interaction with ARMC9. Using a combination of protein interaction analyses, characterization of patient-derived fibroblasts, and analysis of CRISPR/Cas9-engineered zebrafish and hTERT-RPE1 cells, we demonstrated that dysfunction of ARMC9 or TOGARAM1 resulted in short cilia with decreased axonemal acetylation and polyglutamylation, but relatively intact transition zone function. Aberrant serum-induced ciliary resorption and cold-induced depolymerization in ARMC9 and TOGARAM1 patient cell lines suggest a role for this new JBTS-associated protein module in ciliary stability.

Keywords: Genetic diseases; Genetics; Proteomics.

Figure 1. ARMC9 associates with TOGARAM1 in a ciliary module.

(A) Schematic of full-length ARMC9 and fragments (frag) used as baits in Y2H bovine and human retinal cDNA library screens. The domains indicated are the predicted LisH (dark green), coiled-coil (CC) domain (light green), and the ARM-containing domain (purple). (B) Direct interaction analysis grid using full-length prey constructs. Selection of strains coexpressing bait and prey constructs was performed on quadruple-knockout medium (synthetic defined lacking leucine, tryptophan, histidine, and adenine [SD-LWHA]). The top row displays yeast colony growth when using fragment 1 of TOGARAM1 as prey. (C) β-Gal activity assay confirming the interactions. (D) Schematic of full-length TOGARAM1 and fragments used in Y2H bovine and human retinal cDNA screens. (E) TOGARAM1 screen results validated in a Y2H-directed interaction analysis on triple-knockout (SD-LWH) and quadruple-knockout (SD-LWHA) media. (F) Flag co-IP of 3xFlag-ARMC9, 3xFlag–TOGARAM1, 3xFlag-CCDC66, 3xFlag-CSPP1, and 3xFlag-CEP104 with 3xHA-ARMC9. 3xFlag-mRFP served as a negative control. Western blot (WB) analysis after Flag-tag purification indicated the presence of 3xHA-ARMC9, confirming the interactions. 3xFlag-mRFP showed no interaction with 3xHA-ARMC9. (G) ARMC9 interacted with TOGARAM1, as confirmed by TAP (dashed lines) and Y2H (solid lines) screens. Validation was subsequently performed using co-IP (dotted lines). (H) Silver stain gels of C-terminally and N-terminally Strep/Flag tandem affinity purification–tagged (SF-TAP–tagged) ARMC9 (left, large arrowhead, 80 kDa) and N-terminally SF-TAP–tagged TOGARAM1 (right, large arrowhead, 200 kDa) after protein purification. The small arrows indicate the expected protein bands of 2 TOGARAM1 isoforms (195.6 kDa and 189.4 kDa) in the ARMC9 TAP purification, and 2 endogenous ARMC9 isoforms (91.8 kDa and 75.7 kDa) in the TOGARAM1 TAP purification. FL, full-length.

Figure 2. TOGARAM1 variants cause JBTS.

(A) Pedigrees and segregation of TOGARAM1 variants. (B) Brain imaging features in individuals with TOGARAM1-related JBTS. MTS (arrowheads in left column, axial T2-weighted images) and elevated roof of the fourth ventricle (arrows in right column, sagittal T1-weighted [top 2] and T2-weighted [bottom] images). Much of the cerebellar tissue on the sagittal images (right panels) is hemisphere, based on axial and coronal views (not shown). (C) Multi-exon deletion in UW360. Primers flanking the predicted deletion amplify a 1064-bp product in the father (F) and the affected son (S) due to a 12,191-bp deletion, but not in the mother (M), because the predicted product was too large. Sanger sequencing of the breakpoint in gDNA (upper) and cDNA (lower) from the affected child confirmed the deletion of exons 4–7. Coding genomic schematic of Homo sapiens TOGARAM1. Transcript variant 1 is shown (NM_001308120.2; variant 2 NM_015091.4, not shown). (D) Protein schematic of TOGARAM1 with JBTS-associated variants indicated. TOG domains 1–4 are shown, with HEAT repeats indicated in gradients of blue.

Figure 3. Overexpression of TOGARAM1 affects ciliary length, and TOG2 domain variants reduce ARMC9 interaction.

(A) Images of untransfected control hTERT-RPE1 cells. The cilium is shown with the TZ marker RPGRIP1L (white) and the ciliary membrane marker ARL13B (green). Scale bar: 5 μm. (B–E) Transient mRFP-TOGARAM1 overexpression (red) in hTERT-RPE1 cells shown with the TZ marker RPGRIP1L (white) and the ciliary membrane marker ARL13B (green). (B) WT mRFP-TOGARAM1 (mRFP-TOGARAM1), (C) mRFP-TOGARAM1-Arg368Trp (mRFP-R368W), (D) mRFP-TOGARAM1-Leu375Pro (mRFP-L375P), and (E) mRFP-TOGARAM1-Arg1311Cys (mRFP-R1311C). Images are representative of more than 30 cilia assessed per condition over 3 experiments. Scale bars: 5 μm (B–E). (F) Quantification of cilium lengths with overexpression of WT and variant forms of mRFP-TOGARAM1 (untransfected n = 39, WT n = 36, Arg368Trp n = 32, Leu375Pro n = 35, Arg1311Cys n = 36). Box plot horizontal bars represent the median ± 95% CI. P > 0.05 (NS), ***P ≤ 0.001, and ****P ≤ 0.0001, by 1-way ANOVA with Tukey’s multiple comparisons test. No significant differences were found between cells overexpressing mRFP-TOGARAM1, mRFP-Arg368Trp, and mRFP-Leu375Pro. P = 0.0004 for untransfected versus Arg1311Cys. (G) Co-IP of HA-tagged ARMC9 and Myc-tagged TOGARAM1: WT and Myc-tagged TOGARAM1-Arg1311Cys interacted with ARMC9, whereas the TOGARAM1 variants Arg368Trp and Leu375Pro did not. (H) Y2H direct interaction analysis assay with ARMC9 and TOGARAM1: WT and TOGARAM1-Arg1311Cys interacted with ARMC9, whereas the TOGARAM1 variants Arg368Trp and Leu375Pro did not. 3-AT, 3-amino-1,2,4-triazole.

Figure 4. armc9- and togaram1-mutant zebrafish display ciliopathy-associated phenotypes.

(A–C) Larval phenotype demonstrating kidney cysts in armc9^–/– (B) and kidney cysts and a curved body shape in togaram1^–/– (C) larvae. Black boxes in A–C show ×3.5 magnification of the glomerulus region in the inset. Dashed lines highlight the kidney cysts in B and C. (D–F) Adult scoliosis phenotype in both mutants (E and F) compared with WT (D). (G–I) Immunofluorescence of the pronephric duct in 3-dpf larvae showing fewer cilia. White arrowheads in I point to the short remaining cilia in the togaram1 mutant. (J–L) Immunofluorescence of midbrain ventricles showed shortened cilia in 3-dpf armc9- and togaram1-mutant zebrafish larvae (K and L). (M–O) Immunofluorescence of 3-dpf zebrafish nose pits showed decreased cilia numbers in both mutants (N and O) compared with WT (M). (P–R) Scanning electron microscopy of 5-dpf zebrafish nose pits confirmed the reduced cilia numbers in armc9^–/– (Q) and togaram1^–/– (R) larvae. The controls were WT, +/+, or +/– siblings of –/–. Scale bars: 500 μm (A–C), 5 mm (D–F) and 10 μm (G–R).

Figure 5. ARMC9 and TOGARAM1 dysfunction results in short cilia.

(A) Immunoblot of endogenous ARMC9 in control and patient fibroblasts indicating trace amounts of ARMC9 isoforms 1 and 2 (92 and 75.5 kDa). β-Actin was used as a loading control. (B) ARMC9 schematic indicating JBTS-associated patient variants (green letters represent the variants found in the patients indicated in A). (C) Ciliary length in control and ARMC9 patient fibroblasts (control n = 1395, UW132-4 n = 699, UW132-3 n = 437, UW116-3 n = 656, and UW349-3 n = 353). Significance was assessed by 1-way ANOVA with Dunnett’s multiple testing correction. (D) Ciliation percentage in ARMC9 fibroblast lines (control n = 1723, UW132-4 n = 898, UW132-3 n = 584, UW116-3 n = 764, and UW349-3 n = 425). Results were not significant using a Kruskal-Wallis test. (E) Ciliary length in control and TOGARAM1 patient fibroblasts (yellow panel). P = 0.0003, by unpaired Student’s t test. hTERT-RPE1 cilia length in WT and TOGARAM1-mutant lines (purple panel) based on ARL13B staining. More than 100 cilia were pooled from 2 experiments (control n = 137, TOGARAM1-mutant line 1 n = 111, and TOGARAM1-mutant line 2 n = 178). P < 0.0001, by 1-way ANOVA with Dunnett’s multiple corrections test. (F) Ciliation percentage in TOGARAM1 patient fibroblasts (yellow panel: control n = 466 and UW360-3 n = 429 over 3 experiments). The results were not significant using a Mann-Whitney U test. The ciliation percentage in engineered TOGARAM1-mutant hTERT-RPE1 cells (purple panel: control n = 330, TOGARAM1-mutant line 1 n = 363, and TOGARAM1-mutant line 2 n = 357 over 3 experiments) Results were not significant using the Kruskal-Wallis test. White circles represent individual experiments from D and F. Box-and-whisker plots in C and E represent the median, with the 95% CI indicated by the notches. All ciliary length measurements were based on ARL13B staining. P > 0.05 (NS), **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

Figure 6. ARMC9 or TOGARAM1 dysfunction does not grossly affect the TZ.

(A) Normalized relative fluorescence intensity of ARL13B signal in human fibroblast cilia (yellow panel; data were pooled from 3 experiments; control [gray] n = 1089, ARMC9 [green] n = 582, and TOGARAM1 [blue] n = 126) and in 3-dpf zebrafish hindbrain cilia (pink panel; data were pooled from 4 experiments; 10 cilia were measured per larva; each data point represents 1 larva; control armc9 [gray] n = 42, armc9^–/– [green] n = 41, control togaram1 [gray] n = 45, togaram1^–/– [blue] n = 40). Bars represent the mean. Controls were WT, +/+, or +/– siblings of –/–. Statistical significance was assessed using a Student’s t test for both human fibroblast (Bonferroni-adjusted P < 0.025) and zebrafish (P < 0.05) experiments. **P ≤ 0.01 and ****P ≤ 0.0001. (B) Normalized relative fluorescence intensity of INPP5E signal in human fibroblast cilia (data were pooled from 3 experiments: control [gray] n = 620, ARMC9 [green] n = 248, TOGARAM1 [blue] n = 62). See Supplemental Figure 7 for ARL13B and INPP5E signal intensity across all ARMC9 fibroblast lines. Results were not significant using an unpaired Student’s t test. (C and D) Western blot analysis of ARL13B (C) and INPP5E (D) in ARMC9 UW132-4 patient fibroblasts. GIANTIN and β-actin served as loading controls, respectively. (E) Representative immunofluorescence signal for Arl13b (red) and polyglutamylated (green) in the 3-dpf zebrafish hindbrain cilia quantified in A. Scale bars: 10 μm. Original magnification, ×3.5 (insets). (F) Single hindbrain cilia stained with Arl13b (red) and Cc2d2a (green) in 3-dpf control, armc9^–/–, and togaram1^–/– zebrafish. Scale bars: 1 μm. (G) Representative immunofluorescence signal for RPGRIP1L (white) and ARL13B (red) in cilia from control and 2 TOGARAM1-mutant hTERT-RPE1 lines. Scale bars: 2 μm. (H) Representative immunofluorescence for RPGRIP1L (green) and ARL13B (red) in ARMC9 and TOGARAM1 patient fibroblasts. Percentages of cilia with robust RPGRIP1L puncta are indicated. Scale bars: 2 μm.

Figure 7. ARMC9- and TOGARAM1-mutant cilia display reduced tubulin PTMs in both patient fibroblasts and zebrafish ventricular cells.

(A and B) Immunofluorescence images and immunoblots of (A) acetylated and (B) polyglutamylated tubulin in ARMC9 patient fibroblasts versus control. In the immunoblots, GIANTIN and β-actin were used as loading controls. Scale bars: 3 μm. (C and D) Representative immunofluorescence images of 3-dpf zebrafish hindbrain cilia marked with Arl13b (red) and acetylated (green in C) or polyglutamylated (green in D) tubulin. Scale bars: 10 μm. Original magnification, ×3.5 (insets). Note that acetylated tubulin also marks axons in the developing brain, visible at the edges of the image in C. (E) Normalized relative fluorescence intensity for acetylated tubulin signal in human fibroblast cilia (yellow panel: control n = 1106, ARMC9 n = 532, and TOGARAM1 n = 131) and zebrafish hindbrain cilia (pink panel: pooled data from 2 experiments; 10 cilia measured per larva; each data point represents 1 larva; armc9 control [gray] n = 20, armc9^–/– [green] n = 21, togaram1 control [gray] n = 20, togaram1^–/– [blue] n = 20). (F) Normalized relative fluorescence intensity for polyglutamylated tubulin assessed in human fibroblast cilia (yellow panel: pooled from 3 experiments; control n = 602, ARMC9 n = 298, and TOGARAM1 n = 58) and zebrafish hindbrain cilia (pink panel: pooled data from 2 experiments; 10 cilia measured per larva; armc9 control [gray] n = 22, armc9^–/– [green] n = 20, togaram1 control [gray] n = 25, togaram1^–/– [blue] n = 20). Zebrafish controls were WT, +/+, or +/– siblings of –/–. In E and F, data points greater than 4 and less than or equal to 2 are not displayed but were included in the statistical analysis. For a complete graph of all data points and a graphical summary of all ARMC9 lines, see Supplemental Figure 8, A and B and C and D, respectively. Statistical significance (adjusted P < 0.025) was assessed using a Bonferroni-corrected Student’s t test for both fibroblast and zebrafish experiments. **P ≤ 0.01 and ****P ≤ 0.0001.

Figure 8. Fibroblasts from patients with JBTS exhibit abnormal axonemal stability.

(A) Cold-induced depolymerization assay schematic and ciliation percentages of treated cells normalized to nontreated controls. Statistical significance was assessed using a Bonferroni-corrected Kruskal-Wallis test, with P = 0.0003 and P = 0.02, respectively. White circles represent individual experiments. (B) Relative ciliation rates 2, 4, 6, and 8 hours after serum re-addition in human fibroblasts previously serum starved for 48 hours. At t0, t2, t4, t6, and t8 hours, respectively, the following numbers of cells were quantified: control 1, n = 455, n = 413, n = 350, n = 346, n = 395; control 2, n = 595, n = 431, n = 351, n = 368, n = 279; ARMC9 UW132-4, n = 218, n = 193, n = 229, n = 195, n = 189; and TOGARAM1 UW360-3, n = 496, n = 622, n = 513, n = 558, n = 492. Ciliation percentages were normalized to 100% at the time of serum re-addition, and percentages represent the amount of remaining cilia compared with t0. Error bars represent 95% CIs. See the “Statistics and reproducibility” section of Supplemental Methods for details on statistical testing for cilia stability assays.

Figure 9. Disruptions of the ARMC9-TOGARAM1 module affect ciliary length, axonemal PTMs, and stability.

(A) TOGARAM1 interacts with ARMC9 through its TOG2 domain. (B) Effects of TOGARAM1 overexpression (WT and with JBTS-associated variants) on ciliary length in TOGARAM1-mutant hTERT-RPE1 cells and consequences of JBTS-associated variants for the interaction with ARMC9. (C) Consequences of mutations in ARMC9 or TOGARAM1 for ciliary length and axonemal PTMs in patient fibroblast lines (black arrows) or zebrafish mutants (white arrows). TZ integrity despite ARMC9 or TOGARAM1 dysfunction is indicated with a green checkmark. The consequences of TOGARAM1 and ARMC9 mutations for ciliary stability in response to cold or serum re-addition in patient fibroblasts are indicated with black arrows. Yellow boxes represent pathogenic variants. Bold crosses indicate presumed loss-of-function mutations. del, deletion; fx, frameshift; LoF, loss of function; ZF, zebrafish; RPE1 mut, hTERT-RPE1 TOGARAM1–mutant lines. Protein domains: LisH, CC, ARM, TOG.