A WDR35-dependent coatomer transports ciliary membrane proteins from the Golgi to the cilia

Intraflagellar transport (IFT) is a highly conserved mechanism for motor-driven transport of cargo within cilia, but how this cargo is selectively transported to cilia and across the diffusion barrier is unclear. WDR35/IFT121 is a component of the IFT-A complex best known for its role in ciliary retrograde transport. In the absence of WDR35, small mutant cilia form but fail to enrich in diverse classes of ciliary membrane proteins. In Wdr35 mouse mutants, the IFT-A peripheral components are degraded and core components accumulate at the transition zone. We reveal deep sequence homology and structural similarity of WDR35 and other IFT-As to the coatomer COPI proteins a and β′, and demonstrate an accumulation of ‘coat-less’ vesicles which fail to fuse with Wdr35 mutant cilia. Our data provides the first in situ evidence of a novel coatomer function for WDR35 likely with other IFT-A proteins in delivering ciliary membrane cargo from the Golgi necessary for cilia elongation.


Introduction 27
The primary cilium is a highly-specialized sensory organelle and signaling hub compartmentalized     (Keady et al., 2012). Thus, there is no defect in entry of IFT-B holocomplexes into cilia, 145 but their ability to exit from cilia is impaired in the absence of WDR35. 146 We next checked the stability of the IFT-A holocomplex by endogenous immunoprecipitation 147 of IFT-A core protein IFT-140 and its interactors. Whilst IFT140 immunoprecipitated all six compo-148 nents of the IFT-A complex from Wdr35 +∕+ embryo lysates, in Wdr35 −∕− samples both peripheral 149 components IFT139 and IFT43 were missing in MS ( Figure 3A) and immunoblot ( Figure 3B). More-150 over, core components were also significantly reduced in the purified mutant complex, suggesting 151 that WDR35 is critical for the stable IFT-A complex assembly. To distinguish defects in stability from 152 defects in assembly, we looked at total IFT-A component protein levels and we found total IFT139 153 and IFT43 levels were also undetectable on blots with lysates from both Wdr35 −∕− MEFs ( Figure 3C, 154 E) and embryos ( Figure 3D, F). This suggests WDR35 is not only critical for the formation of stable 155 IFT-A holocomplex but is also required for stability of its peripheral components. In contrast, the  163 These results suggest WDR35 might be a link between IFT-A core and periphery proteins, re- that the stability of IFT-A complex proteins is interdependent. 170 We next looked at the localization and levels of the IFT-A components by immunofluorescence.   Figure 6A). In contrast, homology searches with COPI and COPI 1/2, which 255 have HEAT/ARM repeats, did not yield any IFT components, as was the case with COPI , which 256 has TPR repeats but no WD40 domains. COPI and COPI 1/2, which have no identifiable repeat 257 domains, are most closely related to adaptor protein complex subunits AP2 and AP3. In summary, 258 using multiple rounds of sequence homology searches, we generated a broad set of putatively   Clusters of IFT and COPI subunits generated from the results of reciprocal sequence similarity searches with HHBlits using IFT144, IF140, IF122, and WDR35 as initial search queries, suggest a very close similarity between a subset of IFT proteins and the COPI ( and ') subunits. Clusters are color-coded according to protein structural motifs with TPR repeat proteins (blue) and dual WD40 repeat and TPR repeat-containing proteins (magenta). Lines between clusters indicate sequence-based proximity. (B) Structure prediction showed IFT144, IFT140, IFT122, and WDR35 to have close structural similarity to COPI complex proteins ( and '). 2.5Å X-ray structure of ' (PDB:3mkq) and IFT-A proteins are shown with N-terminal WD40 repeat (blue) and C-terminal TPR repeats (magenta). Sequence identity, similarity, and coverage between COPI -' and respective IFT-A proteins are shown in the table below. (C) The cartoon representation of COPI coat assembly subunits , , ', , , and . The and ' subunits have WD40 repeat (blue) and TPR repeat (magenta), and WD40 hydrophobic domain insertion in the Golgi membrane facilitates membrane curvature initiating COPI coat assembly.

WDR35 forms vesicular coats to traffic membrane proteins packed in vesicles from
306 the Golgi to cilia. 307 We further tested our hypothesis for the coatomer function of IFT-As by transmission electron         Rosenbaum, 2014). This strongly suggests that the coated vesicles around the control 339 cilia function to transport cargo from the Golgi to the cilia. In the absence of WDR35, non-coated 340 vesicles accumulate around the ciliary base marking a failure in this process in either the formation 341 and/or maintenance of this coat and subsequent fusion at the target ciliary membrane.

342
To further confirm our hypothesis that these Golgi-to-cilia vesicular coats are made of WDR35, 343 we performed correlative light and electron microscopy (CLEM) imaging in Wdr35 −∕− MEFs express-344 ing WDR35::EmGFP, which we had previously shown to completely rescue cilia phenotypes ( Figure   345 1A, B, Figure 9A). Expressing WDR35::EmGFP in Wdr35 −∕− ensures that every WDR35 particle was 346 labelled with EmGFP, minimizing competition with non-labeled species. Using Airyscan confocal 347 imaging of WDR35::EmGFP MEFs grown on grids for subsequent TEM, we saw WDR35 signal en-348 riched at the ciliary base of rescued mutant cilia. Moreover, we observed that this signal coincided 349 with the re-appearance of electron-dense vesicles now seen by TEM at the cilia base (Figure 9). 350 This is reminiscent of the localization reported for IFT140 in photoreceptor ciliary vesicles by EM

398
In contrast, IFT20 has been shown to localize to the Golgi (Follit et al., 2006; Noda et al., 2016). 399 Moreover, whilst there is in vitro evidence that BBSomes can cluster on liposomes, they do not 400 deform membranes, a key step in vesicle formation by coatomers (Jin et al., 2010). In contrast,  This raises a question as to why a protein like WDR35, which shares structural homology to  which acts as the energy source to drive fission and fusion of these events.

444
Recruitment, remodeling, and regulation of protein coats involve cycles of GTP hydrolysis, for 445 example ARF-1 regulates COPI coat (Dodonova et al., 2017). It is interesting to note that we and 446 others have been unable to purify IFT-A complex with any GTPases (Mukhopadhyay et al., 2010), 447 suggesting that any interaction is transient. This is even in conditions where we can purify endoge- is required for efficient trafficking into cilia. 481 We have demonstrated that an IFT-A-dependent coat for membrane vesicles derived from the 482 Golgi exists and is necessary for their fusion with the ciliary sheath, which is continuous with the 483 ciliary membrane. We also showed that this coat is necessary, to efficiently deliver cilia-destined sig-  Table 1. 508

509
Cells were trypsinized to a single-cell suspension and resuspended in 10 l Resuspension Buffer R 510 per 0.5x10 5 ∕ , ℎ (0.75 g/transfection) Wash Buffer-3 ( IP lysis buffer without any IGEPAL detergent). All wash buffer is aspirated, and dry 533 beads were stored at -80°C, or samples were sent immediately for mass spec.     The obtained model was processed later in Pymol software for structural analysis. solutions were pre-warmed to 37°C, and all steps were done at 37°C, to preserve the cytoskeleton.

631
(4) The fixation buffer was replaced with fresh fixation buffer and incubated for 4hrs at 4°C. (5)

632
After that, the sample was washed once in sample buffer and 2-3 times in distilled water, each for  To measure PCM1 intensity radially from the centrosomes, an average intensity projection of the z-