Characterisation of septin roles in ciliary biology and interpretation of CC2D2A variants using Caenorhabditis elegans

The sensory and signaling roles of primary cilia rely on compartmentalization mechanisms that establish the organelle’s protein and lipid composition. Septins, a family of cytoskeletal GTPases, are proposed to regulate a membrane diffusion barrier, or ‘gate’, at the ciliary transition zone (TZ) of mammalian cells. Here, we examine septin ciliary functions in vivo, within the sensory organs of Caenorhabditis elegans. Loss of one or both nematode septin genes (unc-59, unc-61) disrupts a small subset of cilia (eg. phasmid cilia), causing short and mispositioned axonemes due to truncated dendritic processes. Interestingly, septins do not appear to function at the ciliary TZ; mutants display normal TZ gating and localisations of the MKS and NPHP module gating complexes. Furthermore, double mutant analyses reveal a lack of septin gene interaction with the MKS & NPHP module pathways. Strikingly, via cell-specific rescue experiments, we find that UNC-61 regulates phasmid neuronal cilia via cell non-autonomous mechanisms, within the glial cells that form the sensory pore. We also find that endogenous UNC-61 localises to phasmid neuro-glia and glia-glia cell junctions, and provide evidence that septin gene loss disrupts junction integrity. Together, our data uncovers an unexpected function for glial septins in regulating a subset of sensory neuronal cilia. We propose that this cell-nonautonomous role involves septins functioning at cell junctions. The ciliary transition zone (TZ) helps establish the protein and lipid composition of primary cilia, and mutations in the conserved TZ protein CC2D2A cause Joubert syndrome. Here, CC2D2A function is examined in vivo by engineering eight conserved patient-like alleles into the C. elegans orthologue mks-6 using CRISPR–Cas9. Endogenous tagging enables quantitative assessment of MKS-6 localisation at the TZ, paired with structural (dye-filling) and behavioural (roaming) readouts. Of the tested alleles, two missense variants alter TZ recruitment: W1050R reduces MKS-6 signal by ~75% and phenocopies an mks-6 null with severe dye-filling and roaming defects; E1137V lowers TZ signal by ~20% yet produces no detectable structural or behavioural defects. A truncating nonsense allele (R612*) exhibits null-like structure and function phenotypes. The remaining missense alleles (W415S, V494F, D543H, P740S, Y1139N) are largely indistinguishable from wild type in these assays. Together, these findings provide a tractable framework for CC2D2A/mks-6 interpreting patient variants in-vivo.