A novel RAB11-containing adaptor complex anchoring myosin-5 to secretory vesicles

Hyphal fungi grow rapidly by apical extension, providing a notorious example of polarized growth. The continuous supply of secretory vesicles necessary to meet the demands of the extending tip and the long intracellular distances existing between the tip and the basal septum, often localized > 100 µm away from the former, impose the need of efficient networks of intracellular traffic involving exquisite cooperation between microtubule- and actin-mediated transport. In Aspergillus nidulans kinesin-1 conveys secretory vesicles to the hyphal tip, where they are transferred to myosin-5, which focuses them at the growing apex, thereby determining cell shape. This relay mechanism and the central role played by myosin-5 in hyphal morphogenesis suggested that the mechanisms anchoring secretory vesicles to this motor should involve specific adaptor(s) ensuring the robustness of actomyosin-dependent transport. Secretory vesicles are charged with RAB11, a regulatory GTPase that determines the Golgi to post-Golgi identity transition. By using a combination of shotgun proteomics, GST-RAB pull-down assays, in vitro reconstitution experiments, targeted reverse genetics and multidimensional fluorescence microscopy with endogenously tagged proteins we show that RAB11, the master regulator of fungal exocytosis, mediates myosin-5 engagement both by contacting the motor and by recruiting UDS1, a homologue of an as yet uncharacterized Schizosaccharomyces protein ‘upregulated during mitosis’, which we demonstrate to be a novel RAB11 effector. Analytical ultracentrifugation determined that UDS1 is an elongated dimer and negative-stain electron microscopy showed that, in agreement, UDS1 is rod-shaped. UDS1 does not contact myosin-5 directly, but rather recruits the coiled-coil HMSV, which bridges RAB11/UDS1 to myosin-5. An HMSV-scaffolded complex containing UDS1 and myosin-5 is present in cells, and a RAB11-UDS1-HMSV complex can be reconstituted in vitro in a RAB nucleotide state-dependent manner. In the absence of UDS1/HMSV the steady state levels of myosin-5 at the apical vesicle supply center diminish markedly, such that microtubule-dependent transport spreading vesicles across the apical dome predominates over apex-focused actin-mediated transport. As a consequence, RAB11 and chitin-synthase B (a cargo of the RAB11 pathway) are not focused at the apex, being distributed instead across the apical dome. Therefore, the RAB11 effector UDS1/HMSV cooperates with the GTPase to adapt secretory vesicles to myosin-5, which is required for the apical targeting of RAB11 cargoes and thus for the normal morphology of the hyphae.

growth. The continuous supply of secretory vesicles necessary to meet the demands of 26 the extending tip and the long intracellular distances existing between the tip and the 27 basal septum, often localized > 100 µm away from the former, impose the need of 28 efficient networks of intracellular traffic involving exquisite cooperation between 29 microtubule-and actin-mediated transport. In Aspergillus nidulans kinesin-1 conveys 30 secretory vesicles to the hyphal tip, where they are transferred to myosin-5, which 31 focuses them at the growing apex, thereby determining cell shape. This relay mechanism 32 and the central role played by myosin-5 in hyphal morphogenesis suggested that the 33 mechanisms anchoring secretory vesicles to this motor should involve specific 34 adaptor(s) ensuring the robustness of actomyosin-dependent transport. 35 Secretory vesicles are charged with RAB11, a regulatory GTPase that determines 36 the Golgi to post-Golgi identity transition. By using a combination of shotgun proteomics, 37 GST-RAB pull-down assays, in vitro reconstitution experiments, targeted reverse 38 genetics and multidimensional fluorescence microscopy with endogenously tagged 39 proteins we show that RAB11, the master regulator of fungal exocytosis, mediates 40 myosin-5 engagement both by contacting the motor and by recruiting UDS1, a 41 homologue of an as yet uncharacterized Schizosaccharomyces protein 'upregulated 42 during mitosis', which we demonstrate to be a novel RAB11 effector. Analytical 43 ultracentrifugation determined that UDS1 is an elongated dimer and negative-stain 44 electron microscopy showed that, in agreement, UDS1 is rod-shaped. UDS1 does not 45 contact myosin-5 directly, but rather recruits the coiled-coil HMSV, which bridges 46 RAB11/UDS1 to myosin-5. An HMSV-scaffolded complex containing UDS1 and myosin-47  Wu et al., 2002). In the case of SVs these adaptors contain a RAB GTPase, be it 95

Myosin-5 is key for delivering RAB11 secretory vesicles to the hyphal tips 134
The efficiency of myosin-5 transport is reflected in the distribution of RAB11 SVs 135 accumulating in the tips before fusing with the PM. In the wild-type, these SVs gather at 136 the SPK/vesicle supply center. In myoE∆ cells completely lacking myosin-5 transport 137 SVs cannot be focused at the SPK, yet they still arrive at the tip by kinesin-1/microtubule-138 mediated transport (Pantazopoulou et al., 2014 (Figure 1B). 139 Consequently, RAB11 is delocalized from the SPK to a tip crescent that reflects the 140 steady-state distribution of the microtubules' plus-ends at the apical dome cortex (Figure  141 1A). This delocalization is paralleled by a conspicuous reduction of RAB11 in the tip 142 ( Figure 1B), strongly suggesting that myosin-5 is a major contributor to the transport of 143 RAB11. Consistent with a secretory defect, loss of myosin-5 results in abnormal hyphal 144 morphogenesis [ Figure 1B, note that exocytosis determines the shape of the cell wall 145 and markedly reduces growth ( Figure 1F) ( This could be tested directly because Sec4 is not essential in A. nidulans, despite of its 151 absence being nearly as debilitating as that of MyoE/myosin-5 ( Figure 1F). Indeed, the 152 amounts of RAB11 SVs accumulating in the tip were noticeably decreased in sec4∆ 153 hyphae ( Figure 1C). However, contrasting with myoE∆ mutants, sec4∆ mutants were still 154 able to gather RAB11 SVs at the SPK. In agreement, myosin-5/MyoE still concentrated 155 in the SPK in the absence of Sec4, albeit less efficiently as well ( Figure 1D and Movie 156 1). Therefore, these data establish that there must be another adaptor sharing with Sec4 157 the ability to engage SVs to myosin-5. Previous studies with fungal and metazoan cells 158  Figure 1E). In the wt, GFP-GTD, although partly cytosolic, 169 was present in SVs accumulating at the tip, indicating that the GTD is indeed sufficient 170 to localize to SVs in vivo. This recruitment of the GTD to SVs did not depend on 171 interaction with resident myosin-5 because in myoE∆ cells GFP-GTD localized to the 172 apical dome ( Figure 1C), recapitulating the distribution of RAB11 SVs ( Figure  173 1B,C)(Movie 2). We concluded that the MyoE GTD is sufficient to bind to SVs. 174 Remarkably, the MyoE GTD also concentrated at the apex of sec4∆ cells ( Figure 1C), 175 further confirming that myosin-5 transport of SVs is still operative without Sec4. 176

177
In the budding yeast, critical residue Tyr1415 is at the center of a Myo2 GTD patch that 178 binds the RABs linking the motor to SVs ( Figure 1F, schematics). Because the budding 179 yeast does not use microtubules to transport SVs, Y1415R substitution affecting a 180 residue crucial for the interaction between the RABs and the myosin-5 is lethal (Lipatova 181 et al., 2008). K1473 located on the opposite GTD surface to Y1415 ( Figure 1F) belongs 182 to a patch of residues that has been reported to bind the Sec15 subunit of the exocyst 183 weaken sec4∆ strains any further. In sharp contrast, the double sec4∆ myoE (Y1414R) 191 mutant combination was nearly lethal ( Figure 1F). As Y1414 is crucial for RAB binding, 192 these data provide strong genetic evidence that an exocytic RAB other than Sec4 is 193 capable of binding directly to the myosin-5 GTD. Da, matching the molecular weight of a dimer ( Figure 2D). Moreover, although the 245 flexibility observed at the level of individual particles precluded us from obtaining 2D 246 averages, individual EM images revealed a rod-shaped structure highly suggestive of a 247 highly elongated coiled-coiled dimer. Therefore, UDS1 is an elongated dimer, with an 248 approximate length of ~500 Å ( Figure 2E). 249 250 As the above RAB pull-down experiments using cell extracts do not rule out the 251 possibility that RAB11 and UDS1 interact by way of bridging protein(s), we used the His-252 tagged protein to repeat the GST-RAB pull-down assays. Figure 2F shows that purified 253 UDS1 behaves as the protein present in Aspergillus extracts, being pulled-down by 254 GTPgS-RAB11 but not by GDP-RAB11, nor by the inactive or active forms of RAB5b and 255 Sec4. In summary, UDS1 is a coiled-coil dimer that binds directly to the (GTP) active 256 form of RAB11. 257

Aspergillus UDS1 colocalizes with both myosin-5 and RAB11 SVs 258
In current models ( Figure 1A), RAB11 SVs arrive at the tip using kinesin-1 and are further 259 concentrated at the SPK by myosin-5. Figure 3A shows that in agreement with these 260 models RAB11 SVs fill a region at the apex that extends slightly beyond the SPK, as 261 defined by the strictly apical MyoE-GFP signal. In colocalization experiments with 262 RAB11, UDS1 behaves like MyoE, being restricted to the SPK, whereas RAB11 shows 263 a slightly broader distribution ( Figure 3B (hooking myosin to SVs). HMSV coprecipitated with UDS1-GFP as well, indicating that 290 these proteins also interact ( Figure 4A). In short, MyoE, UDS1 and HMSV appear to be 291 associates, and components of the RAB11 pathway 292 293 HMSV is a 994 residue-long protein whose 300 N-terminal residues are predicted to be 294 disordered, while the remaining ~700 residues have strong propensity to form coiled-295 coils ( Figure 4B). Like UDS1, HMSV localizes to the SPK, strictly colocalizing with MyoE 296 ( Figure 4C)(Movie 6). To determine the phenotypic consequences of removing UDS1 297 and HMSV we constructed null uds1∆ and hmsV∆ alleles. They are phenotypically 298 indistinguishable, resulting in a radial colony growth defect ( Figure 4D) and, at the 299 cellular level, in abnormally wide hyphae ( Figure 4E), both phenotypic features indicative 300 of defective exocytosis. Notably, the colony growth defect resulting from uds1∆ and 301 hmsV∆ was markedly weaker than that caused by myoE∆ ( Figure 4D). Double uds1∆ 302 hmsV∆ mutants behaved like the parental single mutants, consistent with the 303 corresponding products being components of a functional unit ( Figure S2). The fact that 304 both uds1∆ and hmsV∆ are hypostatic to myoE∆ ( Figure 4D  suggested instead that another factor might bridge HMSV to RAB11 (note that total cell 326 extracts ¾not purified proteins¾ were used as preys in this experiment). 327 328 HMSV scaffolds a myosin-5-containing heterotrimeric complex that binds to the 329 active RAB11 conformer 330 An appealing candidate to link HMSV indirectly to RAB11 was UDS1. Figure 5A shows 331 that GST-UDS1, but not the unrelated bait GST-GFP, pulled-down in vitro synthesized 332 HMSV-HA3. In contrast, neither GST bait pulled-down in vitro synthesized Uso1-HA3, 333 confirming specificity and establishing that purified UDS1 and HMSV interact. Therefore, 334 by interacting directly with both MyoE and UDS1, HMSV would act as scaffold of a 335 heterotrimeric complex that is recruited by RAB11 to SVs by contacting both UDS1 and 336 MyoE/myosin-5. 337 338 To test this model, we performed two sets of experiments. First, we demonstrated in vitro 339 that HMSV is recruited to active RAB11 only if UDS1 is present to bridge the interaction 340 ( Figure 5B). We performed GST-RAB pull-downs in the presence of bacterially 341 expressed UDS1, in vitro synthesized HMSV-HA3 or both. HMSV was recruited by 342 GTPgS-RAB11, but did so only when UDS1 was present in the reaction mix. Neither 343 conformation of RAB5b nor GDP-RAB11 pulled-down HMSV even when UDS1 was 344 present. We conclude that HMSV is an indirect effector of RAB11 that requires the 345 presence of UDS1 to be recruited to the GTPase. 346 347 Secondly, we demonstrated that the stable complex reconstructed in vitro, consisting of 348 MyoE, HMSV and UDS1, is present in cellular lysates and is scaffolded by HMSV. As 349 determined by anti-MyoE Western blotting of GFP-Trap immunoprecipitates of whole-350 cell extracts, MyoE strongly associates with UDS1-GFP and with HMSV-GFP, but not 351 with the unrelated bait Uso1-GFP ( Figure 6A). Indeed, the associations are so efficient 352 that co-immunoprecipitated MyoE could be visualized directly by silver-staining of SDS-353 PAGE gels ( Figure 6A, right). Despite HMSV appears to be the less abundant bait (anti-354 GFP western blot, Figure 6A, right), the interaction between MyoE and HMSV was 355 markedly more efficient than that between MyoE and UDS1, in agreement with the fact 356 that MyoE and UDS1 interact indirectly by way of HMSV. Consistently, the interaction 357 between MyoE and UDS1 was undetectable with hmsV∆ extracts (i.e. was completely 358 dependent on the presence of HMSV) ( Figure 6B), whereas than that between MyoE 359 and HMSV was completely independent of UDS1, taking place irrespectively of whether 360 wild-type or uds1∆ extracts were used ( Figure 6C). Lastly, the interaction between 361 UDS1-GFP and HMSV-HA3 was completely independent of MyoE ( Figure 6D Evidence that UDS1 and HMSV are a co-receptor assisting RAB11 to recruit 369 myosin-5 to SVs. 370 A diagnostic readout of myosin-5 transport is the focusing of SVs at the SPK. Consistent 371 with UDS1 and HMSV acting in a complex regulating actomyosin transport, both uds1∆ 372 and hmsV∆ affected RAB11 SVs similarly, reallocating them from the SPK to a crescent-373 shaped distribution in the apical dome ( Figure 7A). This effect was markedly less 374 prominent than that caused by myoE∆, which resulted in a broader crescent and, as 375 discussed above, in a marked reduction of the signal of SVs docked at the tip cortex 376 ( Figure 1B). Therefore, these data strongly indicate that myosin-5 transport is debilitated 377 in uds1∆ and hmsV∆ mutants, such that although this transport is not abolished, 378 MT/kinesin-1-mediated transport gains prominence, which results in targeting SVs to a 379 broader surface determined by the sites at which MTs' plus ends reach the apical dome. 380 Impairment of actomyosin transport in these mutants explains the partial exocytic deficit 381 that growth tests indicate ( Figure 4D).   indicating that the whole receptor can be split in two stable subcomplexes (Figure 7). 486 Therefore, both UDS1 and HMSV are necessary for the assembly of a receptor complex 487 whose absence results in debilitated F-actin-mediated transport, which is reflected in the 488 spreading of RAB11 SVs across the apical dome. Inefficient F-actin transport of RAB11 489 correlates with slower colony growth resulting from ablation of either co-adaptor, and 490 spreading of RAB11 SVs across the hyphal tip dome correlates with delocalization of its 491 cargo, ChsB, from the SPK. Of note, F-actin transport is not abolished without the co-492 adaptors (Figure 9), because RAB11 is able to bind MyoE directly, which makes the 493 phenotypic consequences of ablating UDS1 or HMSV less deleterious than those 494 resulting from removing MyoE, whose ablation leaves SV transport exclusively in the 495 hands of kinesin-1. In summary, we have identified a novel receptor complex required for the efficient 520 coupling of RAB11 SVs to the myosin-5 MyoE. Proof-of-concept that a motor-cargo 521 interface can be targeted by a small chemical has been recently provided (Randall et al., 522 2017). Although speculative at the moment, the possibility of interfering with fungal 523 growth by diminishing the efficiency of myosin-5 mediated transport is appealing. 524 525 Methods 526

Aspergillus techniques 527
Standard A. nidulans media were used for strain propagation and conidiospore 528 production. GFP and epitope-tagged alleles were introduced in the different genetic 529 backgrounds by meiotic recombination (Todd et al., 2007) and/or transformation (Tilburn 530 et al., 1983), which used recipient nkuA∆ strains deficient in the non-homologous end 531 joining pathway (Nayak et al., 2005). Complete strain genotypes are listed in Table S1. Large scale purification of proteins interacting with the GDP and GTPgS forms of RAB11-759 GST was carried out as described previously for GST-RAB11 (Pinar & Peñalva, 2017). 760 Bound proteins were loaded onto a 10% polyacrylamide gel, which was run until proteins 761 moved 1 cm into the gel. The protein mixture band was detected by colloidal Coomassie 762 staining, excised and processed for tryptic digestion and subsequent analysis by MS/MS 763 essentially as described (Pinar et al., 2019). For MS/MS analyses of GFP-tagged bait 764 associates, proteins were digested using the 'on-bead digest protocol for mass 765 spectrometry following immunoprecipitation with Nano-Traps' recommended by 766 Chromotek. In both cases mass spectra *.raw files were used to search the A. nidulans 767 FGSC A4 version_s10m02-r03_orf_trans_allMODI proteome database (8223 protein 768 entries) using Mascot search engine version 2.6 (Matrix Science). Peptides were filtered 769 using Percolator (Kall et al., 2007), with a q-value threshold set to 0.01. 770

Analytical ultracentrifugation 771 772
Sedimentation equilibrium analysis of UDS1-His was carried out in the Molecular 773 Interactions Facility of the Centro de Investigaciones Biológicas using an XL-A analytical 774 ultracentrifuge (Beckman-Coulter Inc.) equipped with a UV-VIS detector set at 237 nm. 775 Centrifugation was carried out in short (95 µl) columns at speeds ranging from 6000 to 776 9000 rpm, with a last high-speed (48,000 rpm) run to deplete the protein from the 777 meniscus and obtain the corresponding baseline offsets. Weight-average buoyant 778 molecular weights were determined by fitting, using HeteroAnalysis software (Cole,  779 2004), a single-species model to the experimental data (corrected for temperature and 780 solvent composition with SEDNTERP software (Laue, 1992)). 781 Negative staining electron microscopy. 782

783
Purified UDS1 was diluted to 0.2 µM in 150 mM NaCl, 25 mM HEPES pH 7.5 and 5% 784 glycerol, and stained with 2% (w/v) uranyl acetate. Specimens were examined under a 785 JEOL 1230 electron microscope equipped with a TVIPS CMOS 4kx4k camera and 786 operated at 100 kV. Data were collected at a nominal magnification of 40,000x, which 787 corresponds to 2.84 Å/pixel at the micrograph level. The length of 71 representative 788 particles selected from multiple micrographs was measured using ImageJ. 789  In the wild-type RAB11 is recruited to SVs during the Golgi-to-post-Golgi transition. 1131 RAB11 interacts with the GTD of MyoE and with the UDS1 dimer. UDS1 in turn connects 1132 active RAB11 to the HMSV scaffold (represented here as a monomer, but potentially 1133 being a dimer). HMSV bridges RAB11/UDS1 to the MyoE myosin-5. In the presence of 1134 the whole complex myosin-5 transport is most efficient (large arrow). In the absence of 1135 UDS1 or HMSV there is still myosin-5 transport due to the direct interaction between 1136 RAB11 and the MyoE GTD, albeit this transport is markedly less efficient (small arrows), 1137 such that the accumulation of SVs in the SPK is impaired and MT-dependent transport 1138 becomes more prominent, leading to the characteristic apical dome distribution of SVs 1139 in these mutants ( Figure 7A). Phenotypically the hmsV∆ uds1∆ double mutant is 1140 indistinguishable from either of the single mutant strains ( Figure S2). HMSV and UDS1 1141 might sustain efficient myosin-5 transport by reinforcing the interaction between RAB11 1142 and the motor or, alternatively, they might increase processivity of the motor or facilitate 1143 the switch from MTs to actin cables in the crowded cytoplasm of the hyphal tip. 1144 1145