The STRIPAK signaling complex regulates phosphorylation of GUL1, an RNA-binding protein that shuttles on endosomes

The striatin-interacting phosphatase and kinase (STRIPAK) multi-subunit signaling complex is highly conserved within eukaryotes. In fungi, STRIPAK controls multicellular development, morphogenesis, pathogenicity, and cell-cell recognition, while in humans, certain diseases are related to this signaling complex. To date, phosphorylation and dephosphorylation targets of STRIPAK are still widely unknown in microbial as well as animal systems. Here, we provide an extended global proteome and phosphoproteome study using the wild type as well as STRIPAK single and double deletion mutants from the filamentous fungus Sordaria macrospora. Notably, in the deletion mutants, we identified the differential phosphorylation of 129 proteins, of which 70 phosphorylation sites were previously unknown. Included in the list of STRIPAK targets are eight proteins with RNA recognition motifs (RRMs) including GUL1. Knockout mutants and complemented transformants clearly show that GUL1 affects hyphal growth and sexual development. To assess the role of GUL1 phosphorylation on fungal development, we constructed phospho-mimetic and -deficient mutants of GUL1 residues S180, S216, and S1343. While the S1343 mutants were indistinguishable from wildtype, phospho-deficiency of S180 and S216 resulted in a drastic reduction in hyphal growth and phospho-deficiency of S216 also affects sexual fertility. These results thus suggest that differential phosphorylation of GUL1 regulates developmental processes such as fruiting body maturation and hyphal morphogenesis. Moreover, genetic interaction studies provide strong evidence that GUL1 is not an integral subunit of STRIPAK. Finally, fluorescence microcopy revealed that GUL1 co-localizes with endosomal marker proteins and shuttles on endosomes. Here, we provide a new mechanistic model that explains how STRIPAK-dependent and - independent phosphorylation of GUL1 regulates sexual development and asexual growth. Author Summary In eukaryotes, the striatin-interacting phosphatase and kinase (STRIPAK) multi-subunit signaling complex controls a variety of developmental processes, and the lack of single STRIPAK subunits is associated with severe developmental defects and diseases. However, in humans, animals, as well as fungal microbes, the phosphorylation and dephosphorylation targets of STRIPAK are still largely unknown. The filamentous fungus Sordaria macrospora is a well-established model system used to study the function of STRIPAK, since a collection of STRIPAK mutants is experimentally accessible. We previously established an isobaric tag for relative and absolute quantification (iTRAQ)-based proteomic and phosphoproteomic analysis to identify targets of STRIPAK. Here, we investigate mutants that lack one or two STRIPAK subunits. Our analysis resulted in the identification of 129 putative phosphorylation targets of STRIPAK including GUL1, a homolog of the RNA-binding protein SSD1 from yeast. Using fluorescence microscopy, we demonstrate that GUL1 shuttles on endosomes. We also investigated deletion, phospho-mimetic, and -deletion mutants and revealed that GUL1 regulates sexual and asexual development in a phosphorylation-dependent manner. Collectively, our comprehensive genetic and cellular analysis provides new fundamental insights into the mechanism of how GUL1, as a STRIPAK target, controls multiple cellular functions.


Introduction
218 Finally, S1343 is C-terminally located in a highly conserved region. As described in the 219 Material and Methods section, the three triplets encoding S180, S216, and S1343 were 220 individually subjected to in vitro mutagenesis, resulting in substitution of the corresponding 221 serine triplets to either alanine (prevents phosphorylation) or glutamic acid triplets (mimics 222 phosphorylation because of the negative charge). After transformation of the gul1 deletion 223 strain with the mutated genes, we investigated three homokaryotic ascospore isolates of each, 224 the phospho-mimetic strains S180E, S216E, and S1343E, the phospho-deficient strains S216A 225 and S1343A, as well as three independent primary transformants S180A (see the Material and 226 Methods sections for construction). Western blot analysis using an anti-GFP antibody detected 227 the corresponding GUL1-GFP fusion proteins and thus confirmed the translational expression 228 of the mutated genes ( S4 Fig). All strains were phenotypically characterized concerning fruiting 229 body and ascospore formation as well as vegetative growth (Fig 3). Phospho-deficient and 230 phospho-mimetic strains S1343A and S1343E had similar characteristics compared to wild type 231 (Fig 3C, 3D, 3E, 3F). S180E and S180A were fully fertile generating mature fruiting bodies 232 and ascospores (Fig 3C and 3D); however, the number of perithecia per square centimeter in 233 S180A was considerably reduced by about 25 % compared to wild type (Fig 3E). Further, 234 S180A also showed a reduced growth rate comparable to Δgul1 (Fig 3F). An intriguing result 235 was obtained with S216. While S216E has a wild type phenotype, phospho-deficient strain 236 S216A is sterile and forms only protoperithecia. S216A also has a reduced growth rate 237 comparable to S180A and Δgul1. Thus, this phosphorylation site, which seems not to be 238 targeted by STRIPAK, regulates both sexual and hyphal development (Fig 3C, 3D). 239 Interestingly, none of the six phosphorylation mutants exhibits the severe hyphal swelling 240 phenotype observed in Δgul1 (Fig 4B). In conclusion, we hypothesize that the STRIPAK-241 dependent phosphorylation of S180 is a switch for hyphal growth, and to some extent, also 242 effects sexual development. In contrast, phosphorylation of S216 is STRIPAK independent, but 243 essential for the formation of mature fruiting bodies as well as hyphal growth. The 244 phosphorylation of S1343 seems to be not essential for sexual development and asexual growth. 245 246 GUL1 is not an integral subunit of the STRIPAK complex 247 As mentioned above, our previous affinity-purification MS analysis indicated that GUL1 248 interacts with the STRIPAK subunit PRO45, a homolog of mammalian SLMAP (24). Similarly, 249 SSD1 interacts in a two-hybrid analysis with the yeast protein FAR10, a homolog of PRO45 250 (37). In addition, this study considered a negative genetic interaction (GI) between far10 and 251 ssd1. Therefore, to determine whether GUL1 is an integral part of the STRIPAK complex or 252 only associated with it, we examined the GI by investigating the double mutant Δpro45Δgul1.
253 For this purpose, we compared the phenotype of the double mutant with the phenotype of the 254 corresponding single mutants by measuring the vegetative growth rates. Compared to wild type, 255 both single and double mutants showed reduced growth rates (Fig 5). We thus used this 256 phenotypical trait to calculate the GI of gul1 with pro45. It is assumed that the phenotype of a 257 double mutant is the result of the phenotype of both single mutants. Whereas a negative GI 258 denotes a reduced fitness of the double mutant compared to both single mutants, a positive GI 259 refers to a higher fitness than expected. Genes encoding for proteins of different pathways often 260 show a negative GI and those encoding for proteins of the same pathway mostly have a positive 261 GI (10, 38, 39). As control, we used the double mutant Δpro45Δpro11 and both single mutants 262 since both are known STRIPAK core subunits and show direct physical interaction (24). Thus, 263 both genes can be considered to have a positive GI. The absolute values of the vegetative growth 264 rates were calculated relative to wild type, with a value of 1 (S2 Table). The data of the single 265 mutants Δpro45, Δpro11, and Δgul1 are 0.494±0.03, 0.413±0.02, and 0.189±0.01, respectively. 266 The expected values were calculated as described previously (10) and are as follows: 267 Δpro45Δpro11, 0.204 and Δpro45Δgul1, 0.093 (see S2 Table). These expected values (light 268 blue bars in Fig 5) were compared to the experimentally obtained values. As expected, the 269 double mutant Δpro45Δpro11 showed no significant deviation of the experimental value from 270 the expected value, indicating the positive GI of pro11 and pro45, as expected. In contrast, the 271 experimentally obtained value for the double mutant ∆pro45∆gul1 was significantly lower than 272 the expected values (Fig 5). 278 revealed that GUL1 appeared within particle-like structures. These were evenly distributed 279 within the cytoplasm, and some appeared close to nuclei. This observation was further verified 280 when we investigated a strain that expresses both genes for gul1-gfp and h2a-mrfp (Fig 6). As 281 indicated by red arrows, GUL1 localizes close to nuclei, thereby suggesting a localization to 282 spindle pole bodies.
283 To address potential microtubule-dependent movement of GUL1 (18), we performed dynamic 284 live cell imaging (S1 movie). We asked whether the mutation of phosphorylation sites has an 285 effect on long distance movement of GUL1. Analyzing GUL1-GFP expressing strains revealed 286 extensive bidirectional movement of GUL1-GFP, which was most prominent in the vicinity of 287 growing hypha (Fig 7A). The velocity of processive particles was 2.4 µm/s (Fig 7B). We did 288 not observe significant differences analysing GUL1-GFP velocity in the phospho variants ( 290 Of note, the GUL1-GFP movement is reminiscent of endosomal shuttling in fungi (40, 41). 291 To address this point we studied strains expressing GFP-RAB5 and GFP-RAB7, which are 292 established markers for early and late endosomes. Interestingly, the bidirectional movement of 293 GUL1-GFP resembled the bidirectional shuttling of GFP-RAB5-positive endosomes (Fig 7A, 294 S2 movie). To address a potential role of the RBP GUL1 in endosomal mRNA transport we 295 studied co-localization of GUL1-DsRed with the poly(A) binding protein PAB1 296 (SMAC_03445) fused to GFP. Importantly, the latter was also identified in our differential 297 phosphorylation study (Table 1). We observed extensive co-localization in processively 298 moving units, suggesting that the RNA-binding protein GUL1 participates in endosomal 299 mRNA transport (Fig 7C-D; S4 movie). Importantly, this is the first evidence that this mode of 300 long-distance transport is also present in ascomycetes (42). Taken together, our fluorescent 301 microscopic investigation reveals that GUL1 acts close to nuclei and shuttles with PAB1 and 302 transport endosomes along microtubules.

304
Discussion 305 The STRIPAK multi-subunit complex is highly conserved within eukaryotes and the number 306 of reports is increasing that single subunits control a huge variety of developmental processes.
307 Despite the intense interest in STRIPAK, our current knowledge about dephosphorylation 308 targets is quite limited and our understanding of how STRIPAK regulates cell differentiation 309 remains basic. Thus, this study provides new fundamental insights into this research field. 310 We used a quantitative proteomic and phosphoproteomic analysis to identify targets of 311 STRIPAK in the model fungus Sordaria macrospora, for which a collection of STRIPAK 312 single and double mutants are available (43). Compared with our recent study (11), we have 313 now gone beyond this by identifying numerous novel STRIPAK dephosphorylation targets. In 314 detail, we identified five transcription factors, such as the GATA transcription factor PRO44, 315 which was shown to control fungal sexual fertility (44). In PRO44, we detected three 316 phosphorylation sites, two of which are differentially regulated in the double mutants. Notably, 317 another protein (SMAC_08582) shows similarity to serine/threonine kinase STK-57 in 318 N. crassa (45), and carries four phosphorylation sites of which three are differentially 319 phosphorylated in all STRIPAK mutants investigated in this study. Among these, S125 is also 320 differentially regulated in three single mutants of our recent investigation (11). Another 321 remarkable putative STRIPAK target is HAM5, the scaffold protein of the MAK-2 pathway 322 (25, 26), with 18 phosphorylation sites. Two sites seem to be differentially regulated in single 323 mutants, namely S506 in ∆pro11 and ∆pro22 as well as S1200 in ∆pro22 (11). Interestingly, 324 we also found the differential regulation of both sites in all three STRIPAK mutants. Our 325 investigation of two STRIPAK double mutants detected differentially phosphorylated proteins, 326 which seem to be unique in this experimental approach. For example, we detected the 327 serine/threonine kinase SMAC_00192, which has nine phosphorylation sites, with two (S782, 328 S788) that are differentially regulated. Intriguingly, we also identified numerous potential 329 RNA-binding proteins as targets of STRIPAK, thus suggesting extensive regulation of gene 330 expression by STRIPAK at the posttranscriptional level. Among the candidates were PAB1 331 (SMAC_03445), a poly(A)-binding protein that shuttles on endosomes (46), as well as GUL1, 332 a regulator of fungal morphogenesis (14, 17).

334 GUL1 is involved in different developmental processes
335 GUL1 is a highly conserved protein in yeast and filamentous fungi, but its cellular function is 336 currently only partly understood. Our analysis has now revealed an RNA-binding domain, a 337 nuclear localization signal, and a nuclear export signal -among others -in the primary 338 sequence of GUL1. These domains have led us to the conclusion that S. macrospora GUL1 is 339 an RNA-binding protein, as was previously shown by functional analysis in other filamentous 340 fungi and yeasts (16, 22, 47-49). In the human pathogenic yeast Candida albicans, SSD1, the 341 GUL1 homolog, was described as an mRNA-binding protein acting as a translational repressor 342 (49), and the GUL1 homologs in Magnaporthe oryzae and Aspergillus fumigatus were 343 described as cell wall biogenesis proteins (47, 48). In this study, we provide a comprehensive 344 overview of GUL1's possible roles, which are related to sexual development, hyphal 345 morphology, as well as vegetative growth. While the gul1 deletion strain shows a severe 346 reduction in fertility, the phospho-deficient GUL1 S216A variant displays a sterile phenotype and 347 both phospho-deficient variants, GUL1 S216A and GUL1 S180A , exhibit severely reduced 348 vegetative growth. Moreover, the sterile phenotype observed in GUL1 and STRIPAK mutants 349 suggests a further association between both, as was previously demonstrated with the 350 STRIPAK-associated GCK SmKIN3 (10). This association, however, is only fully functional 351 if the phosphorylation states of STRIPAK targets are tightly regulated. In essence, we provide 352 compelling evidence that the STRIPAK target GUL1 is extensively regulated at the level of 353 phosphorylation.

355 GUL1 is trafficking on endosomes
356 Fluorescence microcopy showed that GUL1 localizes not only to cytoplasmic punctae, but also 357 close to nuclei, thereby suggesting localization at the nuclear membrane. This hypothesis is 358 further supported by the interaction of GUL1 with the SLMAP homolog, PRO45, which 359 localizes to the nuclear membrane in wild type strains. However, lack of PRO11 or PRO22 is 360 known to prevent nuclear membrane localization of PRO45 (24), which in turn probably 361 reduces the level of dephosphorylation of GUL1. These observations are consistent with data 362 for the GUL1 homolog from yeast. In this case, nucleocytoplasmic shuttling of SSD1 is 363 essential for mRNA binding (21). 364 Our imaging data provide compelling evidence that in fungal cells GUL1is present on RAB5-365 positive transport endosomes, which shuttle along microtubules. Consistently, microtubule-366 dependent movement has been already described for GUL-1 from N. crassa (18, 50).
367 Endosomal mRNA transport is well-studied in the basidiomycete Ustilago maydis and key 368 components are the RNA-bindings proteins (RBPs) Rrm4, the poly(A)-binding protein PAB1 369 and the small glycine rich RRM protein Grp1 (40, 51). These RBPs form higher-order transport 370 mRNPs that contain cargo mRNAs encoding e.g. septins for endosomal assembly (51-53). 401 Our functional investigation of phospho-deficient and mimetic mutants also demonstrates that 402 phosphorylation of GUL1 at S180 and S216 is critical for vegetative growth. S180 from GUL1 403 corresponds to the phosphorylation site S164 in SSD1, while the sites corresponding to GUL1 404 S216 and S1343 are not predicted as phosphorylation sites in the yeast protein. Moreover, the 405 phosphorylation of GUL1 seems to be dependent on different signaling complexes, as proposed 406 in our new model depicted in Fig 8. While S180 has a conserved recognition site for the NDR 407 kinase, namely COT1, S216 is most probably phosphorylated by a casein kinase. From our 408 phosphoproteome data, it therefore follows that S180 is dephosphorylated by STRIPAK, while 409 a yet unknown phosphatase acts on S216. COT1, which was intensively investigated in 410 N. crassa, is part of the MOR complex, and is regulated by the upstream GCK POD6. All 411 components of the MOR complex are crucial for the polar organization of the actin 412 cytoskeleton, and hence, fungal morphology (9, 16, 17). In N. crassa, gul-1 deletion is able to 413 partially suppress the phenotype of cot-1, and thus; is a dominant modifier of the NDR kinase 414 COT-1, the homolog of the yeast kinase Cbk1p (14, 16, 17). 558 Using specific primers (S5 Table), we generated four plasmids, containing phospho-mimetic 559 and phospho-deficient mutations (S9 Fig). After DNA-mediated transformation of the 560 abovementioned plasmids into ∆gul1, we obtained homokaryotic single spore isolates of 561 phospho-mimetic strains S180E, S216E, and S1343E and of the phospho-deficient strains 562 S216A and S1343A. However, we failed in generating homokaryotic isolates of the phospho-563 deficient strain S180A. In total, we investigated 340 ascospores from two independent primary 564 transformants. From 105 germinated ascospores, none showed resistance against 565 nourseothricin, indicating that the ascospores do not carry the gul1-complementation vector.
566 This result strongly suggests that the phospho-deficient mutation S180A is lethal, and only 846 Tables and Figures: 847 856 ∆pp2Ac1∆pro22, and ∆pro11∆pro22 compared to the wild type identified ten phosphorylation sites in 857 GUL1. Two out of ten phosphorylation sites are differentially phosphorylated in all three STRIPAK 858 mutants. For each phosphorylation site of GUL1, log2 ratio of reporter ion intensity in deletion strain 859 and wild type relative to the respective standard deviation is given. Bold numbers indicate an 860 upregulation of the phosphorylation site compared to the wild type. Regular numbers indicate no 861 regulation of the phosphorylation site compared to the wild type. The phosphorylation sites marked in 862 red were further analysed in this study (see also Fig 2). Standard deviations: ∆pro11: 0.62; 863 ∆pp2Ac1∆pro22: 0.67; ∆pro11∆pro22: 0.57. In our previous study, we found seven phosphorylation 864 sites (S1 Table.).