The Pleiades cluster of fungal effector genes inhibit host defenses

Summary Biotrophic plant pathogens secrete effector proteins to manipulate the host physiology. Effectors suppress defenses and induce an environment favorable to disease development. Sequence-based prediction of effector function is difficulted by their rapid evolution rate. In the maize pathogen Ustilago maydis, effector-coding genes frequently organize in clusters. Here we describe the functional characterization of the pleiades, a cluster of ten symplastic effectors. Deletion of the pleiades leads to strongly impaired virulence and accumulation of reactive oxygen species (ROS) in infected tissue. Eight of the Pleiades suppress the production of ROS upon perception of pathogen associated molecular patterns (PAMPs). Although genetically redundant, the Pleiades target different host components. The paralogs Taygeta1 and Merope1 suppress ROS production in either the cytoplasm or nucleus, respectively. Merope1 targets and promotes the autoubiquitination activity of RFI2, a conserved family of E3 ligases that regulates the production of PAMP-triggered ROS burst and influences flowering time in plants.


Introduction 40
Plants are constantly colonized by a myriad of microbial organisms, yet for the most part, they 41 remain asymptomatic. In order to interact with their host plants, pathogenic microbes evolved 42 molecules known as effectors. These secreted molecules (proteins, RNA and small 43 metabolites) manipulate host physiology and development to suppress immune responses and 44 create an environment that promotes the pathogen's proliferation. Following secretion, effector 45 molecules can either remain in the space between the plant cells, the apoplast (apoplastic 46 effectors) or be translocated into the host cytosol (symplastic effectors) (Sánchez-Vallet et al., 47 2018; . As a result of being secreted molecules, effectors are exposed 48 to the host immune system and can induce resistance responses following their recognition. 49 Hence, the evolution of effector coding genes is governed by two major processes, evasion of The pleiades, a cluster of secreted proteins that contribute to virulence 138 The U. maydis pleiades cluster (Cluster 10A) encodes ten proteins, which, beyond their 139 predicted secretion signals, lack any sequence similarity to known protein domains. 140 Additionally, no cysteine residues (which are frequent in apoplastic effectors) are encoded 141 outside of the predicted secretion signals. The pleiades contain three gene families, A 142 (UMAG_03745, UMAG_03746, UMAG_03747 and UMAG_03750) B (UMAG_03748, 143 UMAG_03749) and C (UMAG_03752, UMAG_037453), based on protein sequence similarity 144 (25% or more). Two genes (UMAG_03744 and UMAG_03751) encode for proteins without 145 homology to any other protein in the cluster (Fig1 a). While none of the pleiades show paralogs 146 outside of the cluster, orthologs of all three gene families are well conserved across the 147 sequenced smut fungi and display high synteny between U. maydis, Sporisorium reilianum, 148 and S. scitamineum, Thus, the gene cluster is conserved among these species (Table S1). showed that they code for proteins without secretion signals (Table S2). We verified the 155 secretion of two Pleiades by monitoring the localization of mCherry fusions of Tay1 156 (UMAG_03752) and Mer1 (UMAG_03753) proteins expressed by U. maydis during biotrophic 157 growth in maize. Three to four days post infection (dpi), U. maydis expressing Mer1SP-158 mCherry-Mer1 or Tay1SP-mCherry-Tay1 showed localization of the mCherry signal in the 159 edges and tips of the hyphae (Fig 1b, Fig supplement S1c) and at the host cell to cell crossings 160 (Fig 1b, arrowheads). On the other hand, U. maydis expressing mCherry-Mer1 (without its 161 predicted secretion signal) showed a diffuse localization of the mCherry signal throughout the 162 whole hyphae (Fig 1b). We next used the U. maydis AB33 strain to express Mer1 and Tay1 in 163 axenic culture. AB33 filaments in vitro in response to nitrate, mimicking to some degree 164 developmental changes induced during host colonization (Brachmann, Weinzierl, Kämper, & 165 Kahmann, 2001). Using the strong, constitutive otef promoter we found that full length Mer1-166 3xHA and Tay1-3xHA accumulated in both, the cell pellet and culture supernatant fractions, 167 whereas the non-secreted protein Actin was only detectable in the cell pellet fraction (Fig 1c,  168 Fig supplement S1d). When using the tay1 promoter (which is ten times stronger than the mer1 169 promoter (Lanver et al., 2018), we could not detect the expression of these proteins in vitro 170 (Fig 1d). Taken together, the presented data shows that Mer1 and Tay1 are soluble proteins 171 secreted by U. maydis into the biotrophic interphase upon host colonization. 172 Deletion of the whole pleiades cluster has been shown to impair virulence of U. maydis 173 (Kamper et al., 2006). Therefore, we generated deletions strains in the solopathogenic SG200 174 U. maydis background to dissect the specific virulence contribution of individual gene families 175 within the pleiades. We generated deletions of the entire gene cluster, individual gene families 176 or all the genes in the cluster except atl1 (formerly ten1, UMAG_03744), since the latter was 177 previously reported to contribute to virulence (Erchinger, 2017). Deletion of the whole cluster 178 had the strongest effect on virulence. Plants infected with this strain predominantly showed 179 mild disease symptoms like chlorosis and small galls (Fig 1d). Simultaneous deletion of the 180 gene families A, B, C together with plo1 also showed a considerable reduction in virulence, 181 although not as strong as in the whole cluster deletion. Finally, deletion of family C showed a 182 mild defect in virulence, which could be complemented ectopically by either tay1 or mer1. 183 Deletion of family A or B alone did not show any measurable effect on virulence (Fig 1d). 184 Altogether our experiments show that the pleiades contribute to virulence additively and that 185 atl1 has the greatest impact. 186

318
The paralogs Tay1 and Mer1 target different cellular compartments 319 To clarify how Tay1 and Mer1 suppress ROS production, we analyzed their sub-cellular 320 localization. We constructed mCherry fusions of Tay1 and Mer1 and expressed them in maize 321 epidermal cells by biolistic bombardment. GFP-nuclear localization signal (NLS) was co-322 transformed and used as a nuclear marker. Confocal microscopy showed that mCherry-323 Tay128-398 localized primarily to the cytoplasm and almost no mCherry signal was detected in 324 the nucleus (Fig 3a). On the other hand, mCherry-Mer123-341 was in the nucleus as well as the 325 in the cytoplasm (Fig 3a). These results lead us to hypothesize that these paralogs may target 326 different cellular compartments in order to suppress ROS production. We tested this 327 hypothesis using a mis-localization approach. We fused Tay1 and Mer1 to either an NLS, NES 328 (nuclear export signal) or Myr (myristoyl lipid anchor signal, targeting proteins to the 329 cytoplasmic side of the plasma membrane) and expressed these proteins in N. benthamiana. 330 Myc-tag fusions of Tay128-398 and Mer123-341 were used as controls since this tag is not expected 331 to affect cellular localization. Plants were treated with flg22 and ROS production was monitored 332 over time as described above. Localizing each paralog to different cellular compartments had 333 opposite effects. Tay1 showed an increased ROS-burst inhibition upon forced cytoplasmic 334 localization (NES and Myr fussions). Fusing Tay128-398 to the NLS signal did not affect its 335 activity (Fig 3b). In contrast, Mer1 showed decreased ROS-burst inhibition upon forced 336 cytoplasmic localization (NES and Myr fussions). Fusing Mer123-341 to the NLS signal did not 337 affect its inhibitory activity (Fig 3c). Hence, by integrating the maize microscopy studies with 338 the N. benthamiana mis-localization data, we propose that Tay1 acts in the host cytoplasm 339 whereas Mer1 acts in the host nucleus.  Fig 4b). Additionally, HA-YFP was not co-precipitated in the presence of 384 Myc-Mer123-341 (Fig 4b). These results show that the interaction between Mer1 and RFI2 385 homologs is specific. 386 Since we established earlier that Mer1 targets the host nucleus (Fig 3), we tested whether 387 RFI2 homologs also localize to this cellular compartment. As before, we co-bombarded 388 fluorescently labeled proteins into the maize epidermis and verified their localization by 389 confocal microscopy. GFP-ZmSRFI2A co-localized with mCherry-Mer123-341 in the nucleus and 390 cytoplasm. GFP-ZmSRFI2A and mCherry-Mer123-341 co-localized in the nucleus and to a lesser 391 extent, in the cytosol. (Fig 4c). Co-expression of GFP-ZmSRFI2A, GFP-AtSRFI2A or GFP-392 AtSRFI2B with mCherry-Mer123-341 in the epidermis of N. benthamiana showed similar results, 393 although the nuclear localization of the E3s was more pronounced in this system (Fig  394   supplement S4c). Taken together, our results show that Mer1 specifically interacts with 395 members of the RFI2 E3 family and that this interaction likely happens in the nucleus. The 396 interaction with maize, N. benthamiana, and Arabidopsis homologs also confirms our earlier 397 assumption that the Pleiades' targets are conserved across monocots and dicots.   Myc-Mer123-341 showed a reduced PAMP-triggered ROS burst compared to wild-type (Col-0) 431 plants (Fig 5a). Interestingly, this phenotype occurred even at low expression levels of Mer1, 432 indicating that the mechanism of ROS-burst suppression is requiring only minor effector 433 amounts (Fig supplement S5c). The rfi2A and rfi2B knock out plants also showed a reduced 434 ROS burst, similar to that of Myc-Mer123-341 plants. Finally, the double knocks out rfi2A/rfi2B 435 did not show a further ROS burst reduction upon flg22 treatment compared to the single knock 436 outs (Fig 5b). This data, together with evidence from the literature, indicate that RFI2 family 437 proteins are the targets of Mer1. They are necessary for ROS production in response to 438

PAMPs and this role is conserved across monocots and dicots. 439
Arabidopsis plants expressing Mer123-341 also showed an early flowering phenotype compared 440 to their Col-0 background when grown under long day conditions (Fig supplement S5a, S5b). 441 In contrast, rfi2A and rfi2B plants did not flower, while the rfi2A/rfi2B double knockout flowered 442 slightly late (Fig supplement S5b). 443 Since expression of Mer123-341 phenocopied the rfi2 knockouts, we hypothesized that Mer1 444 must have an inhibitory effect on the E3s. To test this hypothesis, we produced and purified 445 Myc-Mer123-341, MBP-HA-AtRFI2A and MBP-Strep-AtRFI2B from Escherichia coli and assayed 446 the effect of Mer1 on the ubiquitination activity of the E3s in vitro. Both E3s showed a moderate 447 auto-ubiquitination activity, that was strongly enhanced by addition of Mer1 (Fig 5c). In the 448 case of AtRFI2A, we could detect the higher molecular weight autoubiquitination products of 449 the E3 by western blot directly with α-HA and α-Ubiquitin antibodies. In the case of AtRFI2B, 450 we could only detect the ubiquitination products with the α-Ubiquitin antibody. As negative 451 controls, we performed reactions lacking either E1 or E3 enzymes, which showed no 452 ubiquitination products.     The analysis of the U. maydis genome revealed that a significant number of putative effector 507 genes, whose expression is induced upon host infection, are physically clustered in the 508 genome (Kamper et al., 2006). Similar to prokaryotic operons, gene clusters are commonly 509 found in diverse fungi to co-regulate functionally connected genes (Howlett,Idnurm,& 510 Heitman, 2007; Keller & Hohn, 1997; Rokas, Wisecaver, & Lind, 2018). As no functional 511 characterization of an U. maydis effector cluster has been reported, a common role for these 512 clusters during biotrophy was based on assumptions (Kamper et al., 2006). Here, by using an 513 approach analogous to those used for fungal metabolic gene clusters (Howlett et al., 2007), 514 although relying on heterologous gene expression rather than knock outs, we demonstrate that 515 eight clustered effectors share the ability to suppress PAMP-triggered immunity independent 516 of their sequence relationship. Therefore, for the pleiades, their position in the genome is a 517  However, these two accessions are notoriously different with regards to their flowering 603 behavior (Passardi et al., 2007), which might indicate the presence of an additional factor that 604 needs to be inactivated in Col-0 in order to reveal the early flowering phenotype. Accordingly, 605 this factor could be affected by Mer1. Nonetheless, an effector that dampens immunity while 606 simultaneously promoting flowering would be a great advantage for smuts, which usually 607 sporulate only in the host floral tissues. This is in line with the concept that effectors frequently 608 target regulatory nodes to shift the balance from immunity to growth and development (Uhse 609 & Djamei, 2018). 610 In conclusion, we have shown that Pleiades, a heterogeneous group of proteins, whose genes 611 cluster in the genome of U. maydis, share the ability to suppress PAMP triggered immunity. a benthamiana (Figure 2b) were generated by Gateway Cloning (Katzen, 2007). All other 635 plasmids were generated by the GreenGate system (Lampropoulos et al., 2013). The modules 636 used were either amplified by PCR or obtained from the published system. Additionally, we 637 generated two GreenGate destination vectors. pECGG, is based on a pET backbone and was 638 used for expression of proteins in E. coli. pADGG, is based on a pGAD backbone and was 639 used as prey vector for yeast two hybrid assays. 640 U. maydis knock out strains were generated by homologous recombination with PCR-derived 641 constructs (Kämper, 2004). For complementations and protein expression, strains were 642 generated by insertion of p123 derivatives into the ip locus (Loubradou,Brachmann,643 Feldbrugge, . Transformants were verified by southern blot and/or PCR. All 644 plasmids and strains used in this study can be found in Supplementary materials. 645

Maize infection assays 646
Pathogenicity assays and disease symptom scoring were performed as described by (Kamper 647 et al., 2006). Briefly, U. maydis SG200 and its derivatives were cultured in liquid YepsLight 648

Confocal microscopy 664
Confocal microscopy was performed with a Zeiss LSM 700 confocal microscope. GFP was 665 excited at 488 nm using an argon laser. Fluorescence emission was collected between 500-666 540 nm. mCherry was exited at 561nm and emission was collected between 578-648 nm. 667 Images were processed with ZEN blue 2.3 lite. 668 Protein secretion in U. maydis 669 Detection of secreted proteins from fungal cultures was performed as described in (Djamei et  Invitrogen) (WGA-AF488 10µg/ml, Tween20 0.02% in PBS) by applying vacuum 3 times. 694 Visualization of the stained samples was performed by direct observation with a widefield 695 microscope equipped with Apotome2 (Axio Imager.Z2 sCMOS camera, Axiocam colour 696 camera and Apotome2). DAB was visualized with a color camera or by Phase contrast. Chitin, 697 labeled with wheatgerm agglutinin AlexaFluor488, was visualized by Apotome2 structured 698 illumination with a 480/40nm excitation filter and 525/50nm emission filters. Images were 699 processed with ZEN blue 2.3 lite. The experiments were repeated at least 3 times. 700 Yeast transformation and two-hybrid assays 701 All yeast protocols were done according to the Yeast Protocols Handbook (Clontech, 702 Mountainview, CA) with minor modifications. Strain AH109 was transformed with bait vectors 703 (pGBKT7 and derivatives) and strain Y187 was transformed with prey vectors (pAD, pADGG 704 and derivatives) by the LiAc/PEG method. All positive clones were verified for the presence 705 of the corresponding plasmid by DNA extraction with 20mM NaOH and PCR (Supplementary  706   materials). 707 The Y2H screen was performed by mating the strain carrying pGBKT7-Mer123-341 (bait) against 708 a yeast library carrying cDNA from maize tissue infected with U. maydis in the pAD vector 709 (Farfsing, 2004). Positive clones were selected in high stringency media (SD -leu, -trp, -ade, -710 his). 711 For one to one matings AH109 pGBKT7-Mer123-341 or AH109 pGBKT7 were mated against 712 Y187 pADGG-ZmRFI2A, Y187 pADGG-ZmRFI2B, Y187 pADGG-ZmRFI2T, Y187 pADGG-713 ZmRFI2E, Y187 pADGG-AtRFI2A, Y187 pADGG-AtRFI2B, Y187 pADGG-NbRFI2A or Y187